Datasets:

medarc

/

clinical_pile_v1_minhash_deduped

like 1

nagd(moo4)2_nanocrystals_with_diverse_morphologies:_controlled_synthesis,_growth_mechanism,_photolum

## Abstract: Pure tetragonal phase, uniform and well-crystallized sodium gadolinium molybdate (NaGd(MoO 4 ) 2 ) nanocrystals with diverse morphologies, e.g. nanocylinders, nanocubes and square nanoplates have been selectively synthesized via oleic acid-mediated hydrothermal method. The phase, structure, morphology and composition of the as-synthesized products are studied. Contents of both sodium molybdate and oleic acid of the precursor solutions are found to affect the morphologies of the products significantly, and oleic acid plays a key role in the morphology-controlled synthesis of NaGd(MoO 4 ) 2 nanocrystals with diverse morphologies. Growth mechanism of NaGd(MoO 4 ) 2 nanocrystals is proposed based on timedependent morphology evolution and X-ray diffraction analysis. Morphology-dependent down-shifting photoluminescence properties of NaGd(MoO 4 ) 2 : Eu 3+ nanocrystals, and upconversion photoluminescence properties of NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ and Yb 3+ /Tm 3+ nanoplates are investigated in detail. Charge transfer band in the down-shifting excitation spectra shows a slight blue-shift, and the luminescence intensities and lifetimes of Eu 3+ are decreased gradually with the morphology of the nanocrystals varying from nanocubes to thin square nanoplates. Upconversion energy transfer mechanisms of NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ , Yb 3+ /Tm 3+ nanoplates are proposed based on the energy level scheme and power dependence of upconversion emissions. Thermometric properties of NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ nanoplates are investigated, and the maximum sensitivity is determined to be 0.01333 K −1 at 285 K. Nowadays, lanthanide-doped nanocrystals, especially upconversion nanocrystals, have become the current focus of intensive researches due to their unique photoluminescence properties and consequently numerous applications, such as bio-imaging and bio-probe, photodynamic/chemo-therapy, drug delivery, temperature sensing, solar cells, optoelectronics and photocatalysis . Compared with the luminescence from conventional organic dyes or quantum dots, lanthanide luminescence from nanocrystals exhibits many advantages, including high photostability (high resistance to optical blinking and photobleaching), large Stokes/anti-Stokes shifts, sharp emission bandwidths, abundant emission channels, long excited-state lifetime, low cytotoxicity, and low synthesis expenditure . All these merits endow lanthanide-doped nanocrystals with efficient luminescence, high detection sensitivity and signal-to-noise ratio, and ease of use in aforementioned applications. The morphology (size and shape) of nanocrystals will affect their physicochemical properties, and the synthesis of nanocrystals with tunable morphologies is particularly significant for the applications in biological and biomedical fields . So morphology-controlled synthesis of nanocrystals has attracted much attention from researchers. Double alkaline rare-earth molybdates ARe(MoO 4 ) 2 (A = alkali metal cation, Re = trivalent rare-earth metal cation) have been demonstrated to be promising candidates as luminescent host materials for numerous applications, due to their favorable chemical and physical stability, large lanthanide admittance, and relatively low phonon energy . Many researches have been devoted to the synthesis and luminescence properties of lanthanide-doped molybdate microcrystals or phosphors . Nevertheless, the synthesis or luminescence properties of double alkaline rare-earth molybdate nanocrystals are rarely reported, which result from the faster crystallization and growth rate and difficulty in controlling the growth process of double molybdates 30,31 . Bipyramid-like NaLa(MoO 4 ) 2 : Eu 3+ nanocrystals were synthesized hydrothermally using oleic acid/oleylamine as surfactant 32 . NaLa(MoO 4 ) 2 : Eu 3+ , Eu 3+ /Tb 3+ shuttle-like nanorods composed of nanoparticles were prepared hydrothermally using ethylene glycol as ligand and their luminescent properties were discussed 33,34 . However, most of these works focused on either nanocrystals with only a single morphology or poor-crystallized composite nanoparticles, and controlled synthesis of double molybdates nanocrystals with diverse morphologies has not been reported so far. Solution-based wet chemical methods, which allow a fine control of size, shape and chemical homogeneity of the products by fine tuning of experimental conditions, are universally employed to synthesize nanocrystals 35,36 . Some organic additives with functional groups or long hydrocarbon chains (e.g. oleic acid, citrate acid, oleylamine, ethylenediamine tetraacetic acid, and cetyltrimethyl ammonium bromide) can act as complexing agents and shape modifier by adjusting the growth rate of different facets under hydrothermal conditions 37,38 . In this work, we present a novel template-free morphology-controlled hydrothermal synthesis of NaGd(MoO 4 ) 2 nanocrystals. Pure tetragonal phase, uniform and well-crystallized NaGd(MoO 4 ) 2 nanocrystals with several distinct morphologies, including nanocubes and square nanoplates, can be selectively synthesized by a modified hydrothermal method using oleic acid as complexing agent. The morphology of the synthesized NaGd(MoO 4 ) 2 nanocrystals can be controlled by simply tuning the contents of oleic acid in the precursor solution. Effects of oleic acid and sodium molybdate (Na 2 MoO 4 ) on the formation of NaGd(MoO 4 ) 2 nanocrystals and growth mechanism of NaGd(MoO 4 ) 2 nanoplates are discussed. Meanwhile, morphology-dependent down-shifting photoluminescence properties of NaGd(MoO 4 ) 2 : Eu 3+ nanocrystals, upconversion photoluminescence properties of Yb 3+ /Er 3+ and Yb 3+ /Tm 3+ square nanoplates, and thermometric properties of NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ square nanoplates are investigated in detail. ## Results and Discussion Crystal structures, morphologies and compositions. NaGd(MoO 4 ) 2 nanocrystals with specific uniform morphologies are synthesized hydrothermally at 180 °C for 12 h with different contents of oleic acid (0.25, 0.75, 1.25 and 1.5 ml) and 10 mmol Na 2 MoO 4 in the initial precursor solutions. Figure 1a shows the XRD patterns of NaGd(MoO 4 ) 2 nanocrystals samples with four typical morphologies, including nanocubes (pattern i) and square nanoplates with different aspect ratios (patterns ii-iv), in the 2 theta range of 10-80°. All the diffraction peaks can be readily indexed to a tetragonal phase NaGd(MoO 25-0828). The average crystallite sizes calculated using Scherrer's formula from the broadening of the diffraction peak (112) in the four patterns are 58, 51, 48 and 49 nm, respectively. What's more, compared with the standard diffraction data, quite intense (004) diffraction peak is found in the pattern iii of Fig. 1a, indicating a preferentially oriented crystallization might exist along the (001) planes of NaGd(MoO 4 ) 2 nanocrystal. Due to the morphological characters of the samples and discrepancy in sample preparation procedure for the XRD measurements, the enhancement of (004) peaks in patterns ii and iv is not evident. The SEM images of the NaGd(MoO 4 ) 2 nanocrystals with four typical morphologies were shown in Fig. 1b-e. Obviously, all the samples exhibit uniform, regular and well-crystallized nanocrystals. In Fig. 1b, the sample consists of monodisperse and uniform nanocubes with side length of ~150 nm. In Fig. 1c-e, the samples are composed of monodisperse and uniform square nanoplates. The thicknesses are about 85, 70, 50 nm and the side lengths are about 250, 400 and 500 nm for the samples in Fig. 1c-e, respectively. The chemical composition of the NaGd(MoO 4 ) 2 nanocrystals with the morphology of square nanoplates (shown in Fig. 1d) was analyzed by the EDS spectrum shown in Fig. 1f. The sample is confirmed to be composed of Na, Gd, Mo and O. The measured atomic ratio of Na, Gd, Mo and O is close to the stoichiometric proportion of NaGd(MoO 4 ) 2 . C peak and excessive proportion of O come from a little oleic acid adsorbed on the surface of the sample. The Si and Pt peaks arise from silicon substrate and conductive coating. More details about the morphological and structural features were further investigated by employing TEM, HRTEM and SAED. The morphologies of the samples shown in the TEM images (Fig. 2a-d) are consistent with that in SEM images. The HRTEM images taken at the edge of the nanocube/nanoplates (Fig. 2e-h) reveal perfect crystalline surfaces. The interplanar distances between adjacent lattice fringes of the four samples are all about 0.26 nm, which correspond to the d spacing of (200) or (020) planes of tetragonal NaGd(MoO 4 ) 2 structure. These lattice fringes indicate that the nanoplates grow along and directions, namely (001) planes, which agrees well with the speculation from the XRD analysis. Taking the morphology of the nanocrystals (square plate) into account, it is also inferred that the normal direction of the upper surface of the square nanoplates is the zone axis ( orientation). Thus the upper surface of the square nanoplates belongs to (001) planes of tetragonal NaGd(MoO 4 ) 2. The SAED patterns of all the samples show highly ordered sharp spots, which indicate the single crystalline nature of the samples. The diffraction dots are indexed to (200) and (020) planes of tetragonal NaGd(MoO 4 ) 2 . FTIR analysis was performed to investigate the surface properties of the samples. Supplementary Fig. S1 presents the FTIR spectra of the NaGd(MoO 4 ) 2 nanocrystals with four typical morphologies (Fig. 1b-e). The spectra are similar in shape. A broad band at about 3399 cm −1 corresponds to the O-H stretching vibrations is observed, arising from surface-adsorbed ambient water. Small peaks at about 2927 and 2856 cm −1 are attributed to the stretching vibration of -CH 2 in skeletal chain of oleic acid. The peaks at about 1635 and 1460 cm −1 are ascribed to the vibrations of the C = O groups of oleic acid 39 . The strong absorption bands at 796 and 704 cm −1 are assigned to the F 2 (ν 3 ) antisymmetric stretch and the peak at 434 cm −1 is ascribed to F 2 (ν 4 ) bending mode vibrations related to the O-Mo-O stretching vibrations in the MoO 4 tetrahedron 40 . These results show the existence of residual complexing ligand on the surface of the samples. Formation of the NaGd(MoO 4 ) 2 nanocrystals. Effects of Na 2 MoO 4 on the morphology of the NaGd(MoO 4 ) 2 nanocrystals. Generally speaking, the molar ratio of the starting source reagents in the precursor solutions would affect the morphology and/or phase of the products in the hydrothermal synthesis procedure. To investigate the effects of the Na 2 MoO 4 on the morphology of the synthesized NaGd(MoO 4 ) 2 nanocrystals, NaGd(MoO 4 ) 2 samples are synthesized with different amounts of Na 2 MoO 4 varying from 2 to 12 mmol, and the fixed amount of oleic acid (1.25 ml) in the precursor solution. Supplementary Figs S2 and S3 present the XRD patterns and SEM images of the as-synthesized samples. From the XRD patterns, it is observed that all the samples are pure tetragonal phase NaGd(MoO 4 ) 2 (ICDD No. 25-0828). Similarly, intense diffraction peak (004) is also observed in pattern c of Supplementary Fig. S2. As can be seen from the SEM images, the samples are comprised of nanoparticles and bipyramids when a small amount of Na 2 MoO 4 is added. With the increasing amount of Na 2 MoO 4 , the samples evolve toward square nanoplates. And both the side length and thickness of the square nanoplates are reduced with the increase of Na 2 MoO 4 content (Supplementary Fig. S3c-e). The sample exhibits irregular nanoflakes with nearly round shape at further increasing Na 2 MoO 4 content (12 mmol, Supplementary Fig. S3f). According to Bravais-Friedel-Donnay-Harker theory, high-index facets with high surface free energy have a large growth rate and will not be expressed in the equilibrium morphology of the resulting crystal 41 . Based on crystal structure models of tetragonal NaGd(MoO 4 ) 2 shown in Supplementary Fig. S4, packing density of Gd 3+ /Na + for some low-index facets are calculated to be 0.0364 −2 for {001} facets, 0.0332 −2 for {010}/{100} facets and 0.0227 −2 for {101} facets. The packing densities of Gd 3+ /Na + on the {001} facets are higher than that on other facets. Na 2 MoO 4 will ionize and provide (MoO 4 ) 2− anions in the hydrothermal solution. When excess (MoO 4 ) 2− exist in the solution, it will be preferentially adsorbed on {001} facets of tetragonal NaGd(MoO 4 ) 2 nanocrystal nuclei due to the strong electrostatic interaction between (MoO 4 ) 2− and Gd 3+ /Na + 42 . This preferential adsorption of (MoO 4 ) 2− on {001} facets will reduce the growth rate along directions and cause preferentially oriented crystallization along the (001) plane in the NaGd(MoO 4 ) 2 nanocrystals. So the samples exhibit thinner square nanoplates at higher Na 2 MoO 4 contents. A slightly lower adsorption energy of (MoO 4 ) 2− for the other facets than that for {001} facets will also cause the adsorption of (MoO 4 ) 2− on the other non-{001} facets, but the restriction of the growth rate is weaker than that for {001} facets. Thus the side length is slightly reduced with increasing Na 2 MoO 4 content. More excess amounts of Na 2 MoO 4 will provide more (MoO 4 ) 2− anions, and break the equilibrium of growth dynamics for the nanocrystals and make the products more irregular. Effects of oleic acid on the formation of NaGd(MoO 4 ) 2 nanocrystal. As is previously discussed, appropriate amount of Na 2 MoO 4 favor the formation of the synthesized NaGd(MoO 4 ) 2 nanocrystals with regular morphology. So the amount of Na 2 MoO 4 is further fixed at 10 mmol and NaGd(MoO 4 ) 2 nanocrystal samples synthesized with different contents of oleic acid (0-1.75 ml) in the initial precursor solution. As can be seen from the XRD patterns shown in Supplementary Fig. S5, all the diffraction peaks of the as-synthesized samples can be easily indexed as pure tetragonal phase NaGd(MoO 4 ) 2 (ICDD No. 25-0828). No peak from impurities or other phases can be found in these patterns, indicating high purity of the samples. A relatively intense diffraction peak (004) is observed in pattern f (Supplementary Fig. S5). Figure 3 shows the SEM images of the synthesized NaGd(MoO 4 ) 2 nanocrystal samples. When no oleic acid is added, the products exhibit barrel-like nanocylinders with the height of ~300 nm and diameter of ~200 nm (Fig. 3a). Whereas when oleic acid was added into the reaction system, new morphologies appear. Nanocubes were obtained when 0.25 ml oleic acid was added (Fig. 3b). When the content of oleic acid is increased to 0.5 ml, the morphology of the products varies from nanocubes to square nanoplates (Fig. 3c). With the increase of oleic acid content ranging from 0.5 to 1.5 ml, the morphologies of the products are all square nanoplates, whereas the thickness of the nanoplates decreases and the side length increases gradually (Fig. 3c-g). The square nanoplates become increasingly thinner and flatter with the increase of oleic acid contents, and seem to be squashed gradually. When 1.75 ml oleic acid is added, the products become irregular thinner nanoflakes with near-circular shape and thickness less than 50 nm (Fig. 3h). It is noted that oleic acid acts as complexing agent and shape modifier, and plays a key role in the morphology-controlled synthesis of NaGd(MoO 4 ) 2 nanocrystals. When lanthanide nitrates solution is added into the mixed solution of oleic acid and ethanol, lanthanide oleate complexes ((RCOO) 3 Ln) are formed through strong coordination interaction and ion exchange process. The oleate complexes could control the concentration of free Ln 3+ in solution and thus help to control the growth of the nanocrystals in a dynamical view 43 . In addition, the oleic acid in the hydrothermal solution will limit the growth rate of specific planes of the nanocrystals through interactions with lanthanide ions 44 . In our case, oleate ions are supposed to be selectively adsorbed on the {001} facets of square NaGd(MoO 4 ) 2 nanocrystals. The adsorbed oleate ions will reduce the reactivity of the {001} facets and limit the growth rate along the direction (perpendicular to the {001} planes) of NaGd(MoO 4 ) 2 nanocrystals. Therefore, the more amount oleic acid is added, the more (001) and (001) facets are expressed in the eventual equilibrium morphologies of nanocrystals, which will result in the formation of NaGd(MoO 4 ) 2 square nanoplates. When excess amounts of oleic acid is added (1.75 ml or more in our case), the growth kinetics will be changed. A more reduced crystal growth rate along the directions means a relatively faster growth rate along (001) planes (i.e. along and directions). A faster and faster growth rate along (001) planes will make the growth behavior out of kinetic control and lead to the formation of irregular near-circular nanoflakes. Therefore, the morphology of the as-synthesized NaGd(MoO 4 ) 2 nanocrystals evolves in the sequence of nanocylinders, nanocubes, square nanoplates and irregular nanoflakes with the increase in the oleic acid content. Growth mechanism of NaGd(MoO 4 ) 2 nanocrystals. It is hard to observe the crystallization process in the hydrothermal apparatus directly and the growth mechanism of hydrothermally synthesized nanocrystals is generally inferred from the morphology observation and XRD analysis of the products obtained at different reaction time intervals. Taking NaGd(MoO 4 ) 2 square nanoplates as an example, time-dependent morphology evolution and XRD analysis are carried out to disclose the growth mechanism of NaGd(MoO 4 ) 2 nanocrystals. The SEM images and XRD patterns of the products obtained for different reaction time (0, 0.5, 1, 3 and 6 h) with 1.25 ml oleic acid and 10 mmol Na 2 MoO 4 in the initial precursor solutions are presented in Supplementary Fig. S6. Amorphous poor-crystalline precursors were formed in the initial stage before hydrothermal reaction, which can be confirmed by the corresponding SEM images and XRD patterns of the products obtained at 0 h. When the reaction time is prolonged to 0.5 h, some small particles appear in the SEM image (Supplementary Fig. S6b) and a small peak emerges in the XRD pattern, which means crystal nuclei are gradually formed as the hydrothermal reaction proceed. Square nanoplates are found when the reaction time is prolonged to 1 h and some small ones are also found in the SEM image (Supplementary Fig. S6c). Meanwhile the diffraction peaks in the XRD patterns of the products become sharper and stronger, and the peaks fit well with the pure tetragonal phase NaGd(MoO 4 ) 2 . This indicates that the nuclei grow bigger first to form rudiments of NaGd(MoO 4 ) 2 nanocrystals with the morphology of square nanoplates under hydrothermal conditions in the presence of oleic acid. Then some of the rudimental crystals grow even bigger. By and large, the nanocrystals in this stage are not well developed, since some square nanoplates with smaller size are always observed in the SEM image (Supplementary Fig. S6c). With further increasing reaction time, the rudimental NaGd(MoO 4 ) 2 square plates grow bigger and bigger, and the smaller ones disappear gradually in the meantime (Supplementary Fig. S6c-e). This can be deemed as the ripening process of the NaGd(MoO 4 ) 2 nanocrystals. In this stage, large square nanoplates develop even bigger at the expense of smaller ones, driven by the thermodynamic minimization of the surface energies of the nanocrystals. This phenomenon is often observed in the synthesis process of nanocrystals, and universally known as Ostwald ripening 45 . Eventually, uniform and well-crystallized NaGd(MoO 4 ) 2 nanocrystals with regular morphology are formed at the end of the ripening process. A possible growth mechanism is proposed based on the morphology evolution as follows. First the precursor is converted to NaGd(MoO 4 ) 2 nuclei in the nucleation stage under hydrothermal conditions. Subsequently, the NaGd(MoO 4 ) 2 nuclei grow to rudimental NaGd(MoO 4 ) 2 nanocrystals, followed by the Ostwald ripening process until well-crystallized NaGd(MoO 4 ) 2 nanocrystals are formed. In brief, NaGd(MoO 4 ) 2 nanocrystals are formed through "Nucleation → Crystallization → Ostwald ripening" growth process. From the above analysis, it can be seen that the amounts of Na 2 MoO 4 and oleic acid have significant effects on the formation of the NaGd(MoO 4 ) 2 nanocrystals, and the NaGd(MoO 4 ) 2 nanocrystals are formed step by step with increasing reaction time under hydrothermal conditions in the growth process. The effects of Na 2 MoO 4 and oleic acid on the formation of NaGd(MoO 4 ) 2 nanocrystals with diverse morphologies, and the growth mechanism of NaGd(MoO 4 ) 2 square nanoplates is summarized and illustrated schematically in Fig. 4. Photoluminescence and thermometric properties of NaGd(MoO 4 ) 2 : nanocrystals upon lanthanide (Eu 3+ , Yb 3+ /Er 3+ and Yb 3+ /Tm 3+ ) doping. Morphology-dependent down-shifting photoluminescence of NaGd(MoO 4 ) 2 : Eu 3+ nanocrystals. NaGd(MoO 4 ) 2 : 5% Eu 3+ nanocrystals with the morphologies of nanocubes and square nanoplates with different thicknesses (corresponding to the morphologies shown in Fig. 3b-g) are synthesized, and their morphology-dependent down-shifting photoluminescence properties are investigated in detail. Figure 5a shows the photoluminescence excitation spectra of NaGd(MoO 4 ) 2 : 5% Eu 3+ nanocrystals monitoring at 616 nm. Broad and intense excitation bands lie in the range from 235 to 350 nm, which are referred to charge transfer (C-T) absorption corresponding to the electron transfer from 2p orbit of O 2− to 5d orbit of Mo 6+ within the MoO 4 2− group in the host molybdate. The full widths at half-maximum (FWHM) of the C-T bands are all about 55 nm. These broad and intense C-T excitation bands indicate that the dopant Eu 3+ in NaGd(MoO 4 ) 2 nanocrystals can be excited efficiently by ultraviolet light radiation around 280 nm. Sharp peaks centered at 362, 395 and 465 nm are observed for all spectra shown in Fig. 5a. These peaks are attributed to the characteristic 4f → 4f transitions ( 7 F 0 → 5 D 4 , 7 F 0 → 5 L 6 and 7 F 0 → 5 D 2 ) of Eu 3+ . The asymmetric C-T band can be fitted by two Gaussian peaks, and the fitting curves for spectrum (i) and (vi) in Fig. 5a are depicted in Supplementary Figs S7 and S8. A blue-shift of more than 2 nm in the C-T bands from spectrum (i) to (vi) is observed in the fitting curves. The nanocrystals with a smaller thickness will have a larger energy gap due to quantum confinement effect. The charge transfer band, which is related to the bandgap of NaGd(MoO 4 ) 2 host, is thus shifted towards the higher energy side. On the other hand, the blue-shift of the C-T bands indicate that the bonding energy between the central Mo 6+ and the ligand O 2− becomes stronger with the morphology varying from nanocubes to thin nanoplates. The intensities of C-T bands and other intrinsic peaks of Eu 3+ in excitation spectrum decrease gradually, which result from luminescent quenching effect. Figure 5b shows the photoluminescence emission spectra of NaGd(MoO 4 ) 2 : 5% Eu 3+ nanocrystals under excitation of C-T band at 280 nm. The spectra for different morphologies are similar in shape, in which four emission peaks at 592, 616, 655, and 703 nm are associated with 5 D 0 → 7 F J (J = 1, 2, 3, 4) transitions of Eu 3+ . The sharp and intense red emission lines at 616 nm ( 5 D 0 → 7 F 2 transition) suggest that the Eu 3+ dopant ions occupy the sites without inversion symmetry in the host NaGd(MoO 4 ) 2 nanocrystals. What's more, the emission intensity decreases slightly from spectrum (i) to (vi) due to the decreased radiative transition probability caused by surface quenching effect. Compared with the nanocubes, the thin nanoplates are smaller in size and have a larger surface-to-volume ratio. The energy of activator (Eu 3+ ) may be trapped by surface defects, ligands and other quenchers which leads to enhanced surface quenching effect 46 . The luminescence dynamics of NaGd(MoO 4 ) 2 : 5% Eu 3+ nanocrystals at 616 nm for different morphologies are investigated, as shown in Fig. 5c. All the curves show single-exponential decay and can be well-fitted by a single-exponential function I = I 0 exp(− t/τ), where I 0 is the luminescence intensity at t = 0, τ is the lifetime. The lifetimes of Eu 3+ are determined to be 0.672, 0.620, 0.603, 0.592, 0.586 and 0.571 ms for different morphologies from nanocubes to thin square nanoplates, respectively. The lifetime of an excited state depends on the depopulation (radiative or nonradiative transitions) probability of electrons from this excited state. Due to the surface quenching effect, the energy in the upper excited state of Er 3+ easily migrates to the surface and is trapped by the surface defects, ligands or other quenchers, which increases the nonradiative transition probability and therefore reduces the lifetime of the excited state. So the thin square nanoplates with the smallest size and the largest surface-to-volume ratio have the lowest lifetime. Upconversion photoluminescence of NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ and Yb 3+ /Tm 3+ thin square nanoplates. Upconversion luminescence properties of NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ , Yb 3+ /Tm 3+ thin nanoplates are investigated. Figure 6a presents the upconversion luminescence spectra of NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ nanoplates with fixed Yb 3+ concentration (10%) and different Er 3+ concentrations (0.5%, 1% and 2%) under 980 nm excitation. Intense green luminescence peaks centered at 530 and 553 nm, and red emission peaks at 657 and 670 nm are observed, which are both characteristic intra-configurational 4f → 4f transitions of Er 3+ . The two green emission peaks at 530 and 553 nm are ascribed to the transitions 2 H 11/2 → 4 I 15/2 and 4 S 3/2 → 4 I 15/2 of Er 3+ , respectively. The red emission peaks correspond to the transition 4 F 9/2 → 4 I 15/2 of Er 3+ . Both the green and red upconversion emissions are split into several subpeaks due to Stark splitting of the upper energy levels. The integral emission intensities of the three samples are depicted in the inset of Fig. 6a. The sample with Er 3+ doping concentration of 1% possesses the highest integral emission intensity, indicating the optimal Er 3+ doping concentration is 1%. At higher Er 3+ doping concentration the upconversion luminescence becomes less efficient, owing to the concentration quenching effect and the cross relaxation between Er 3+ . To ascertain the upconversion energy transfer mechanism, investigation of power dependence of upconversion emissions is performed. The double logarithmic plots of green and red upconversion emission intensities versus pump powers for the NaGd(MoO 4 ) 2 : 10% Yb 3+ /1% Er 3+ nanoplates are depicted in Fig. 6b, together with the linear fitting curves. Generally, the number of photons involved in the upconversion process may be inferred from the slopes of the plots 47 . The slope value of the fitting curves for green and red upconversion emissions is 1.96 and 1.86 respectively, revealing two photons are involved in both green and red upconversion processes. The proposed upconversion mechanism based on the energy level scheme and power dependence of upconversion luminescence is schematically shown in Fig. 6c. Yb 3+ ions act as sensitizer to absorb energy of 980 nm excitation light and transfer it to activator Er 3+ ions. Electrons in the ground state ( 4 I 15/2 ) of Er 3+ can be excited to 4 I 11/2 state through energy transfer (ET) process from Yb 3+ , and subsequently excited to 4 F 7/2 state through energy transfer upconversion (ETU) process. The states 2 H 11/2 and 4 S 3/2 can be populated by means of nonradiative multiphonon relaxation (MPR) form the state 4 F 7/2 . Radiative transitions from 2 H 11/2 / 4 S 3/2 to the ground state of Er 3+ generate green upconversion emission. For red upconversion emission, there are two ways to populate the upper excited state 4 F 9/2 : MPR process from state 4 S 3/2 and ETU process from state 4 I 13/2 . Then red upconversion emission can be expected by radiative transition from the populated state 4 F 9/2 to the ground state. Figure 7a presents the upconversion luminescence spectra of NaGd(MoO 4 ) 2 : Yb 3+ /Tm 3+ nanoplates with fixed Yb 3+ concentration (10%) and different Tm 3+ concentrations (0.5%, 1% and 2%) under 980 nm excitation. Intense emission peaks at 796 nm in the infrared wave range corresponding to the transition 3 H 4 → 3 H 6 of Tm 3+ are observed in all the three spectra. All the three samples exhibit nearly pure near-infrared upconversion luminescence. The upconversion emission intensity of the NaGd(MoO 4 ) 2 : Yb 3+ /Tm 3+ samples decreases with increasing Tm 3+ concentration, which is caused by the concentration quenching effect. The sample doped with 10% Yb 3+ /0.5% Tm 3+ possesses the most intense emission intensity. Some relatively weak emission peaks can also be observed in the visible region of the magnified spectra for the samples doped with 0.5% and 1% Tm 3+ (shown in inset of Fig. 7a). Blue emission peaks at 477 nm and red emission peaks at 649 nm are ascribed to transitions 1 G 4 → 3 H 6 and 1 G 4 → 3 F 4 of Tm 3+ . Since the emission peaks at 796 nm locate in the invisible wave range, the upconversion luminescence appears blue in color to the naked eye. Figure 7b depicts the double logarithmic plots of near-infrared upconversion emission intensities versus pump powers for the NaGd(MoO 4 ) 2 : 10% Yb 3+ /0.5% Tm 3+ nanoplates. The slope value of the fitting curves is 1.93 for near-infrared upconversion emissions and 2.30 for near-infrared emission, suggesting that the near-infrared upconversion luminescence belongs to two-photon process, while the blue one involve three-photon absorption. Figure 7c shows the proposed upconversion mechanism and the energy level scheme of NaGd(MoO 4 ) 2 : Yb 3+ /Tm 3+ . There are energy mismatches between the transitions within Yb 3+ ( 2 F 5/2 → 2 F 7/2 ) and Tm 3+ ( 3 H 6 → 3 H 5 , 3 F 4 → 3 F 2 , and 3 H 4 → 1 G 4 ). So the energy transfer between Yb 3+ and Tm 3+ needs the assistance of phonons of the host. State 3 H 5 is populated from 3 H 6 by phonon assisted ET process. Electrons in state 3 H 5 can relax to state 3 F 4 through the MPR process. The phonon assisted ETU process will populate the state 3 F 2, 3 from state 3 F 4 . Electrons in the state 3 F 2, 3 can relax to state 3 H 4 , from which electrons relax radiatively to the ground state generating dominated near-infrared upconversion emission (796 nm). State 1 G 4 might be populated by another phonon assisted ETU process from 3 H 4 to 1 G 4 . Electrons in 1 G 4 state can decay radiatively to either state 3 F 4 or state 3 H 6 , which will cause the blue (477 nm) or red (649 nm) upconversion emissions. For the ET and ETU processes 3 H 6 → 3 H 5 , 3 F 4 → 3 F 2 , and 3 H 4 → 1 G 4 , the number of phonons needed in a similar host NaGd(WO 4 ) 2 is about 2, 3 and 5, respectively 48 . The more the number of phonons is, the lower the probability of energy transfer is. Therefore, it is speculated that the ETU process 3 H 4 → 1 G 4 hardly occurs and the electrons in the 3 H 4 state are likely to relax radiatively to the ground state rather than to be excited to 1 G 4 state through phonon assisted ETU process in our case. This explains why the infrared upconversion luminescence is quite intense compared with the visible upconversion luminescence. Thermometric properties of NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ thin square nanoplates. Since the emission intensity ratio from two thermally coupled energy levels of lanthanide ions is sensitive to ambient temperature, optical thermometry can be realized in Ln 3+ doped upconversion nanocrystals based on the temperature-dependent upconversion luminescence 49 . Adjacent thermally coupled energy levels 2 H 11/2 and 4 S 3/2 of Er 3+ follow a Boltzmann-type population distribution, and are employed to investigate thermometric properties of NaGd(MoO 4 ) 2 : 10% Yb 3+ /1% Er 3+ nanoplate crystals. Figure 8a shows the temperature-dependent upconversion luminescence spectra (normalized to 1 at the maximum emission value) from 85 to 285 K under excitation of 980 nm. The upconversion luminescence intensity ratio R (I 525nm /I 550nm ) of the two green emission bands increases evidently with increasing ambient temperature. The intensity ratio R can be expressed as: , where g, σ, ω are degeneracy, emission cross-section and angular frequency of radiative transitions from the 2 H 11/2 and 4 S 3/2 levels to the ground level 4 I 15/2 ), Δ Ε is the energy gap between the 2 H 11/2 and 4 S 3/2 levels, k B is the Boltzmann constant, and T is the absolute temperature. The dependence of upconversion luminescence intensity ratio R with the temperature is plotted in Fig. 8b. The intensity ratio R varies from 0.0006 to 0.9752 with the temperature increasing from 85 to 285 K. The fitting curve is also presented according to the above equation, which matches well with the experimental data. Δ Ε can be further calculated to be 777.45 cm −1 from the fitting results, which is very close to the experimental energy gap between the two levels. The thermometric sensitivity S is defined as the rate of change of R(T) as follows, The calculated sensitivity is plotted as a function of absolute temperature in Fig. 8c. The thermometric sensitivity of NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ nanocrystals increases with increasing temperature in the temperature range of measurement. Compared with the square plate NaGd(MoO ) 2 : 10% Yb 3+ /1% Er 3+ microcrystals synthesized as previously reported by us (temperature-dependent upconversion luminescence and thermometric sensitivity are shown in Supplementary Figs S9 and S10), the NaGd(MoO 4 ) 2 : 10% Yb 3+ /1% Er 3+ nanocrystals possess a more sensitive thermometric property. The maximum value of sensitivity of the NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ nanocrystals is 0.01333 K −1 at 285 K, which is higher than reported values of many other Yb 3+ /Er 3+ doped materials in a similar temperature range (around 300 K) 23, . From the fitted equations for luminescence intensity ratio (R)-Temperature curves for nanocrystals and microcrystals (shown in Fig. 8b and Supplementary Fig. S10a), it is found that, compared with that for microcrystals, both Δ E and the proportionality constant C increase in Equation ( 1 increases, which leads to the enhanced sensitivity as defined in Equation ( 2). This is why nanocrystals have a more sensitive response to temperature, compared with microcrystals. ## Conclusion Pure tetragonal phase, uniform and well-crystallized NaGd(MoO 4 ) 2 nanocrystals with diverse regular morphologies can be selectively synthesized via oleic acid-mediated hydrothermal synthesis method by simply tuning the contents of oleic acid in the precursor solution. (MoO 4 ) 2− ions will be preferentially adsorbed on the {001} facets of tetragonal NaGd(MoO 4 ) 2 , which have a higher packing density of Gd 3+ /Na + ions (0.0364 −2 ). Thus, appropriate amount of Na 2 MoO 4 in the precursor solution favor the formation of NaGd(MoO 4 ) 2 nanocrystals with regular morphology. Since oleic acid in the hydrothermal solution helps to control the growth rate of the nanocrystals, especially along directions of tetragonal NaGd(MoO 4 ) 2 , the amount of oleic acid plays a key role in the morphology-controlled synthesis of NaGd(MoO 4 ) 2 nanocrystals. The morphology of the as-synthesized NaGd(MoO 4 ) 2 nanocrystals evolves in the sequence from nanocylinders, nanocubes, square nanoplates to irregular nanoflakes with increasing oleic acid content. Time-dependent morphology evolution and XRD analysis of the products suggest that the growth of NaGd(MoO 4 ) 2 nanocrystals is governed by a "Nucleation → Crystall ization → Ostwald ripening" growth mechanism. Investigation of down-shifting photoluminescence properties confirm that lanthanide dopant in NaGd(MoO 4 ) 2 host nanocrystals can be excited efficiently by broad band ultraviolet light through charge transfer absorption (around 280 nm). Due to quantum confinement effect and stronger bonding energy between the Mo 6+ and ligand O 2− , the charge transfer band has a slight blue-shift, and the intensity is decreased with the morphology of the nanocrystals varying from nanocubes to thin nanoplates. As a result of surface quenching effect, both the down-shifting emission intensity and lifetime of the Eu 3+ doped nanocrystals decrease gradually from nanocubes to thin square nanoplates. On the basis of energy level scheme and pump power dependence of upconversion emissions, the mechanisms for upconversion photoluminescence of NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ , Yb 3+ /Tm 3+ nanocrystals are proposed. Two-photon process accounts for both the visible (green and red) upconversion emissions of NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ nanocrystals and the near-infrared upconversion emission of NaGd(MoO 4 ) 2 : Yb 3+ /Tm 3+ nanocrystals. While the blue upconversion emission of NaGd(MoO 4 ) 2 : Yb 3+ /Tm 3+ nanocrystals involves three-photon absorption. Lower probability of phonon assisted ETU process 3 H 4 → 1 G 4 of Tm 3+ lead to nearly pure near-infrared upconversion luminescence of NaGd(MoO 4 ) 2 : Yb 3+ /Tm 3+ nanocrystals. NaGd(MoO 4 ) 2 : Yb 3+ /Er 3+ nanocrystals exhibit excellent thermometric properties with a relatively high sensitivity (0.01333 K −1 at 285 K). NaGd(MoO 4 ) 2 nanocrystals have a more sensitive response to temperature compared with microcrystals. Investigations of photoluminescence and thermometric properties manifest that NaGd(MoO 4 ) 2 nanocrystals are promising candidates for luminescent hosts in luminescent imaging, temperature sensing, color display and other tremendous down-shifting/upconversion applications. Synthesis of NaGd(MoO 4 ) 2 and NaGd(MoO 4 ) 2 : Eu 3+ , Yb 3+ /Er 3+ and Yb 3+ /Tm 3+ nanocrystals. A predetermined amount of oleic acid was added into 10 ml ethanol. After vigorous stirring for 30 min, 0.5 ml Gd(NO 3 ) 3 solution (0.5 mmol) and a certain amount of Na 2 MoO 4 solution were added into the above solution under continuous stirring. After additional agitation for 1 h, the as-obtained translucent precursor solution (total volume 41 ml) was transferred into a 60 ml Teflon-lined stainless steel autoclave, which was then sealed and maintained at 180 °C for 12 h. The final precipitate products were collected by centrifugation, washed several times with deionized water and ethanol, and dried at 50 °C for 5 h in air. ## Chemicals. NaGd(MoO 4 ) 2 : Eu 3+ , Yb 3+ /Er 3+ , Yb 3+ /Tm 3+ nanocrystals were synthesized following a similar procedure except for introducing the proper amount of corresponding lanthanide nitrates to the precursor solution as described above. Characterization. Powder X-ray diffraction (XRD) was performed on a Rigaku Smartlab diffractometer with Cu Kα radiation at a scanning rate of 10° min −1 . Scanning electron microscope (SEM, FEI Quanta 400F) and transmission electron microscope (TEM, FEI Tecnai G2 F30) were employed for the observation of the morphology. Energy dispersive X-ray spectroscopy (EDS) data were obtained using the SEM equipped with the energy dispersive X-ray spectrometer. TEM images, high-resolution TEM (HRTEM) images and selected-area electron diffraction (SAED) patterns were performed at an accelerating voltage of 300 kV. Fourier transform infrared (FTIR) spectra were obtained in transmission mode on a Bruker Equinox 55 FTIR spectrometer with the samples sandwiched between two KBr plates. Photoluminescence excitation and emission spectra were recorded on an Edinburgh FLSP920 spectrometer equipped with a 980 nm diode laser, a 450 W continuous xenon lamp and a 60 W microsecond flash lamp as excitation sources and a R928 red-sensitive photomultiplier tube as detector. The samples were annealed at 500 °C for 1 h prior to upconversion luminescence measurements. All the measurements were performed at room temperature except for the thermometric upconversion photoluminescence.

chemsum

{"title": "NaGd(MoO4)2 nanocrystals with diverse morphologies: controlled synthesis, growth mechanism, photoluminescence and thermometric properties", "journal": "Scientific Reports - Nature"}

total_synthesis_and_chemoproteomics_connect_curcusone_diterpenes_with_oncogenic_protein_brat1

## Abstract: Natural products are an indispensable source of lifesaving medicine, but natural product-based drug discovery often suffers from scarce natural supply and unknown mode of action. The study and development of anticancer curcusone diterpenes fall into such a dilemma. Meanwhile, many biologicallyvalidated disease targets are considered "undruggable" due to the lack of enzymatic activity and/or predicted small molecule binding sites. The oncogenic BRCA1-associated ATM activator 1 (BRAT1) belongs to such an "undruggable" category. Here, we report our synthetic and chemoproteomics studies of the curcusone diterpenes that culminate in an efficient total synthesis and the identification of BRAT1 as a cellular target. We demonstrate for the first time that BRAT1 can be inhibited by a small molecule (curcusone D), resulting in impaired DNA damage response, reduced cancer cell migration, potentiated activity of the DNA damaging drug etoposide, and other phenotypes similar to BRAT1 knockdown. ## 2 Natural products have been valuable sources and inspirations of lifesaving drug molecules 1 . Their accumulated evolutionary wisdom together with their structural novelty and diversity makes them unparalleled for novel therapeutic development. However, their natural scarcity, structural complexity, and unknown mode of action often hamper their further development in the drug discovery pipeline. Total synthesis is frequently employed to solve the material supply and chemical probe synthesis for comprehensive biological profiling and target identification 2 . Meanwhile, many biologically validated disease targets are considered as "undruggable" from a chemical standpoint due to the lack of enzymatic activity and/or small molecule binding sites 3 . The BRCA1-associated ATM activator 1 (BRAT1) protein has been validated as an oncogenic protein involved in various cancers but belongs to the "undruggable" category with no known small molecule inhibitors to date. Herein, we report a collaborative effort in the total synthesis and chemoproteomics profiling of curcusone natural products which reveals BRAT1 as a key cellular target and validated curcusone D as the first BRAT1 inhibitor. The curcusone diterpenes (Fig. 1A) were isolated from Jatropha curcas, a widely used ingredient in traditional remedies for a variety of ailments including cancer. Structurally, they share a characteristic [6-7-5] tricyclic skeleton with the daphnane and tigliane diterpenes 4 . Curcusones A-D (1a-d), isolated by Clardy and co-workers in 1986, were unambiguously identified as two epimeric pairs at the C2 position 5 . Since then, around thirty curcusone molecules have been isolated including curcusones F-J 6 , which lack the dienone moiety in the seven-membered ring. Structurally rearranged analogs like spirocurcasone (3) and dimeric analogs such as dimericursone A (2a) and dimericursone B (2b) were discovered recently 7,8 . Among them, 1a-1d exhibited low micromolar IC50 values against a broad spectrum of human cancer cell lines 6 . However, their mode of action remained unknown and no total syntheses of 1a-1d were reported prior to this study. While the closely related daphnane and tigliane diterpenes have attracted a significant amount of synthetic interest , the curcusone molecules have surprisingly received little attention despite their therapeutic potential. In 2017, we reported the first total syntheses of the putative structures of 1i and 1j in 21 steps (Fig. 1B), ultimately leading to the conclusion that the originally proposed structures of both 1i and 1j were incorrect 14 . Our synthesis involves a gold-catalyzed tandem furan formation and furan-allene [4+3] cycloaddition to build the 5,7-fused ring system with an oxa bridge and a Diels-Alder reaction to construct the 6-membered ring. In 2019, Stoltz and co-workers reported their studies toward synthesizing 1a-1d (Fig. 1C) . Their approach features a divinylcyclopropane-cycloheptadiene rearrangement to forge the 7-membered ring and reached advanced intermediate 14 after 12 steps from 8. Total Synthesis and Probe Synthesis. Our ongoing interest in natural products that can covalently modify cellular proteins 18 prompted us to continue pursuing the total synthesis and target identification of 1a-1d with an electrophilic cycloheptadienone moiety. This unique structural feature could allow them to form a covalent bond with nucleophilic residues of certain cellular proteins 19 . Previous cytotoxicity studies found that reduction and/or oxidation of the C6-C7 double bond greatly reduced their anticancer activity 6 . As such, an approach allowing variation of the C6 and C7 substituents would be highly desirable. We envisioned 15 as an advanced intermediate (Fig. 1D). α-Halogenation followed by two methylation reactions would lead to 1a and 1b, which could be oxidized to 1c and 1d via a-hydroxylation. A ring closing metathesis (RCM) or an intramolecular aldol condensation was planned to form the 7-membered dienone Our synthesis started with preparing 23 (Fig. 2), a known compound synthesized from 8 in three steps -extended silyl enol ether formation, vinylogous Mukaiyama aldol reaction, and NaBH4 reduction 20 . We combined the first two steps into a one-pot reaction; crude 22 was then subjected to NaBH4 reduction directly to produce multi-decagram scale of 23 in one batch. We next needed to prepare 24 for the Claisen rearrangement. NaH-promoted addition-elimination between 23 and 20b afforded 24, albeit in low yield (35%). We then used a Mitsunobu reaction between 23 and 20a to prepare 24, but the hydrazine byproduct derived from diethyl azodicarboxylate could not be separated from 24. The recently reported redox-neutral organocatalytic Mitsunobu conditions were also explored but failed to provide 24 21 . Fortunately, the hydrazine byproduct could be tolerated in the Claisen rearrangement. After 24 was heated at 140-150 °C in DMF for 18 h, the Claisen rearrangement did occur, but the rearranged product 25 further cyclized to provide tricyclic compound 26 (X-ray, Fig. S2) as a single diastereomer in 48% yield from 23. Without getting 25 to prepare triflate 18, we decided to continue with 26 and explore the hidden cyclopentane-1,3-dione symmetry to synthesize 17. We started with investigating 1,2-addition of lithiated ethyl vinyl ether (27) to 26 theorizing that a global hydrolysis would release the methyl ketone and the aldehyde at once to form 17 for the aldol condensation. This 1,2-addition turned out to be nontrivial. When two equiv. of 27 was used, only less than 10% yield of 28 was obtained. Owing to their oxophilicity, cerium chloride and lanthanum chloride have been used to promote 1,2-additions 22 . Unfortunately, both failed in our case. Eventually, the 1,2-addition was improved by increasing the amount of 27 to 10 equiv. 23 , and 28 was prepared in 57% yield. 28 was then subjected to hydrolysis upon the treatment with p-toluenesulfonic acid and 17 was obtained in 51% yield from 26. Meanwhile, we were delight to observe the formation of 15 in the same reaction, albeit in very poor yield (<5%). We were encouraged to achieve a global hydrolysis/aldol condensation cascade to synthesize 15 from 28 in one step and identified FeCl3 24 in combination with TMSCl as the optimal conditions. When crude 28 was treated with a premixed FeCl3 (0.2 M in 2-methyltetrahydrofuran) and TMSCl in toluene at room temperature, global hydrolysis occurred to generate 17 in situ, which further underwent FeCl3-promoted intramolecular aldol condensation to afford 15 in 39% yield from 26. With the [6-7-5] tricyclic carbon skeleton quickly assembled in only six steps, we next needed to introduce the two methyl groups. Johnson iodination converted 15 to iodoenone 31, which was surprisingly unstable. Therefore, after a quick workup, crude 31 was immediately subjected to the next Stille cross coupling with tetramethylstannane to provide 32 in 45% yield over two steps. Finally, α-methylation of enone 32 at C2 position delivered a 1:1 mixture of separable (-)-1a and (-)-1b in 41% yield (88% brsm; 9 steps total). α-Hydroxylation of 1a with KHDMS and MoOPH gave separable (-)-1c and (-)-1d in 63% yield (d.r. 1:1; 84% brsm; 10 steps total). Additionally, in order to obtain analogs for biological activity comparison, we converted (-)-1b to (+)-3 and (-)-33 (a synthetic derivative named as pyracurcasone) by following a reported one-step procedure 7 . The 1 H, 13 C NMR, and other analytic data of our synthetic samples matched well with the reported ones, which also conclude that the absolute configuration of 1a-1d assigned by Clardy et al. in 1986 is opposite of the actual ones. We then set out to synthesize 2a from 1a-1d via a biomimetic dimerization. The proposed biosynthesis of 2a consists of a sequence of oxidative dehydrogenation of 1a/1b or dehydration of 1c/1d to form a reactive cyclopentadienone intermediate followed by Diels-Alder dimerization and cheletropic extrusion of carbon monoxide 8 . From there, 2a could be converted to 2b via another oxidative dehydrogenation and double bond isomerization. We started with 1c and 1d. After an unfruitful attempt to synthesize 2a by heating them directly at elevated temperatures, a 1:1 mixture of them was first converted to their mesylates (34). After extensive exploration, we identified that using triethylamine as a base in 1,4dichlorobenzene at 150 °C produced (-)-2a in 18% yield over two steps, which provides a direct evidence to support the proposed biosynthetic pathway. Under our conditions, the formation of 2b was not observed. To elucidate the anticancer mechanism and identify potential cellular targets of curcusones, an alkyne-tagged probe molecule 37 was designed for chemoproteomics studies. Since the dienone is likely protein-reactive and is critical for the observed activity, we decided to minimize structural perturbation of this part and used the tertiary alcohol as a handle to link with a terminal alkyne. 37 was synthesized in 59% yield from (−)-1d via a DCC-promoted coupling with 36. Cytotoxicity and Target Identification. We evaluated the cytotoxicity of curcusones and their analogs in breast cancer MCF-7 cells using the WST-1 assay (Fig. 3A). Synthetic 1a-1d, natural 1b and 1d, and intermediates 15 and 32 exhibited micromolar EC50 values against MCF-7 cells with 1d being the most potent curcusone. Importantly, the cytotoxicity values for synthetic 1b and 1d were virtually identical to the values of their naturally isolated counterparts. Analog 33 showed slightly better antiproliferation activity indicating the feasibility of finely tuning the cycloheptadienone moiety to improve potency, but 2a was not active even at 100 µM. Likely due to the full confluency of the tested MCF-7 cells, the EC50 values we obtained were about one order of magnitude higher than previously reported (1.6-3.1 µM EC50 values for 1a-1d) 6 . Gratifyingly, 37 retained similar anticancer properties of 1d, thus warranting its use in competitive chemoproteomic studies. We then identified the cellular targets of curcusones by competitive chemoproteomics using probe 37. MCF-7 cells were treated with 1d for 4 hours followed by lysis, treatment with 37, CuAAC with biotin azide, enrichment, digestion, and LC-MS/MS analysis using label-free quantification (Fig. 3B and Table S1). The best competed target was BRAT1, which acts as a master regulator of the DNA damage response (DDR) and DNA repair by binding to BRCA1 and by activating DDR kinases such as ATM and PRKDC (DNA-PKcs) following DNA damage . Knockdown of BRAT1 increased the constitutive level of apoptosis in human osteosarcoma cells 25 and decreased cancer cell proliferation and tumorigenicity in vitro and in mouse tumor xenografts 27 . BRAT1 is also an unfavorable prognostic marker in kidney and liver cancers 28 . Therefore, targeting BRAT1 is a promising strategy for cancer treatment. 9 We next characterized the physical interaction between 1d and BRAT1. We overexpressed FLAG-BRAT1 in HEK-293T cells and performed a thermal shift assay by treating lysates with 1d, heating as indicated, and probing the remaining soluble FLAG-BRAT1 by Western blotting (Fig. 3C). We observed 10 thermal destabilization of 1d-treated BRAT1 indicating a direct interaction. To validate endogenous BRAT1 as a target of 1d in live cells, we employed a competitive pulldown experiment. MCF-7 cells were treated with 1d for 4 hours before lysis, treatment with probe 37, CuAAC with biotin azide, streptavidin enrichment, elution, and Western blot visualization. Indeed, native BRAT1 was enriched by 37 and was competed by 1d (Fig. 3D). Additionally, 1d competed the enrichment of BRAT1 from cervical cancer HeLa and triple negative breast cancer MDA-MB-231 cells, thus validating native BRAT1 as a cellular target of 1d across these cell lines. In situ treatment of 1d in live HeLa cells competed BRAT1 enrichment by 37 at low micromolar concentrations (EC50 = 2.7 µM; Fig. 3E). Assuming that 1d binds to BRAT1 irreversibly, we determined the binding constants Ki (3.5 µM) and kinact (0.0079 min -1 ; Fig. 3F). These results demonstrate that 1d is the first small-molecule binder of BRAT1. BRAT1 Modulation. To determine whether 1d inhibits BRAT1 in cells, we generated stable BRAT1 KD HeLa cells via shRNA retroviral transduction (Fig. S1A). We then compared the protein expression profiles of BRAT1 KD cells versus 1d-treated cells (3 µM, 24 h) by global proteomics analysis (Fig. 4A-C and Table S2). Among 3347 quantified proteins in compound-treated cells, we found only 36 up-and 42 down-regulated proteins. Importantly, 31 of the 78 dysregulated proteins were also dysregulated in BRAT1 KD cells, thus indicating that 1d functionally inhibits BRAT1 in cells. Notably, several wellknown cancer migration and progression drivers were downregulated (Fig. 4B), including TRIM47 which mediates cancer migration 29 , the bona fide oncoprotein and potential biomarker WBP2 30 , and frequently highly amplified oncogene FNDC3B 31 . None of these proteins have previously been functionally linked to BRAT1. We then investigated the effect of 1d treatment and BRAT1 KD on cancer cell migration in WT and BRAT1 KD HeLa cells, as well as WT MCF-7 and MDA-MB-231 cells (Fig. 4D-G). As expected, BRAT1 knockdown greatly diminished migration of HeLa cells, and treatment with 1d at 1 µM concentration also reduced migration of all cell lines by ~4-fold. Our global proteomics experiment also revealed several commonly downregulated key DNA repair proteins (Fig. 4C) such as (i) POLD1 which synthesizes DNA during repair 32 , (ii) USP47 which facilitates base-excision repair 33 , (iii) FANCI which mediates the repair of DNA double strand breaks and interstrand crosslinks 34 , and (iv) BRCC3 which stabilizes the accumulation of BRCA1 at DNA breaks 35 . These proteins have not been previously linked to BRAT1 either. Most notably, 1d treatment (24 hours) significantly downregulated the actual physical target, BRAT1 (ratio 0.18), as confirmed by Western blotting (Fig. S1B). Collectively, these findings demonstrate the importance of BRAT1 as a master regulator of the DDR and that 1d inhibits BRAT1 in cells. We then investigated whether 1d would potentiate the DNA damaging effect of the clinical drug and topoisomerase inhibitor etoposide via BRAT1 inhibition. WT or BRAT1 KD HeLa cells were treated with DMSO, etoposide, 1d, or etoposide and 1d combined. Subsequent DNA damage was then measured by fluorescence microscopy using γH2AX staining (Fig. 4H, I and Fig. S1C). Treatment with 1d (3 µM) or KD of BRAT1 alone did not increase γH2AX signal. However, co-treatment of 1d with etoposide led to a 2-fold increase. Similarly, etoposide treatment significantly increased γH2AX signal in BRAT1 KD cells, recapitulating the 1d/etoposide co-treatment results. Importantly, 1d treatment did not increase γH2AX signal in etoposide-treated BRAT1 KD cells, confirming that the 1d-etoposide synergism is linked to BRAT1 inactivation. Furthermore, co-treatment of 1d with etoposide also increased cytotoxicity in HeLa, MCF-7, and MDA-MB-231 cells (Fig. 4J). Likewise, there was increased cell death in BRAT1 KD HeLa cells following etoposide treatment relative to WT cells (Fig. 4K). Altogether, these results demonstrate that targeting BRAT1 with 1d is a promising anticancer strategy for chemosensitization to DNA damaging drugs. In summary, we completed the first asymmetric total synthesis and target identification of the curcusone natural products. Our convergent synthesis builds upon a cheap and abundant chiral pool molecule (8) and features a thermal -sigmatropic rearrangement and an FeCl3-promoted global hydrolysis/aldol condensation cascade to rapidly construct the critical cycloheptadienone core. This efficient synthetic route yielded 1a and 1b in 9 steps, 1c and 1d in 10 steps, and 2a in 12 steps from (S)-(−)-8. The successful synthesis of 2a from 1c/1d experimentally supports the proposed Diels-Alder dimerization and cheletropic extrusion biosynthesis. By performing chemoproteomics with the alkyne probe 37, we identified the previously "undruggable" oncogenic protein BRAT1 as a key cellular target of 1d. Furthermore, 1d inhibits BRAT1 in cancer cells, thereby reducing cancer cell migration, increasing susceptibility to DNA damage, and inducing chemosensitization to the approved drug etoposide. To our knowledge, 1d is the first known small-molecule inhibitor of BRAT1, a master regulator of the DDR and DNA repair. Many promising clinical trials are underway targeting DDR proteins such as PARP, ATR, ATM, CHK, and DNA-PK as monotherapies or in combination with other treatments . Olaparib, a PARP inhibitor, was approved by FDA in 2014 as a monotherapy to treat germline BRCA1/2-mutant ovarian cancer 36 . Further structure-activity optimizations of the curcusones may thus yield novel BRTAT1 inhibitors as potential lead medicines for monotherapies or combination therapies. ## Experimental Procedures and Data For detailed experimental procedures and compound characterization data, see the Supplementary Information.

chemsum

{"title": "Total Synthesis and Chemoproteomics Connect Curcusone Diterpenes with Oncogenic Protein BRAT1", "journal": "ChemRxiv"}

computational_screening_of_all_stoichiometric_inorganic_materials

## Abstract: The compositional space for inorganic materials remains vastly unexplored. Walsh and colleagues have designed procedures that use well-established chemical knowledge to quantify the number of possible multi-component compounds. They show how chemical filters can be applied to quickly and effectively narrow down the number of results and focus on those with target functionality. ## INTRODUCTION Currently, over 184,000 entries in the Inorganic Crystal Structure Database (ICSD) involve 9,141 structure types; 1 66,814 of these materials have also been subject to quantum mechanical calculations, and information on their basic electronic structures and thermodynamics is included in the Materials Project 2 (powered by the PYMATGEN infrastructure 3 ). The configurational phase space for new materials is immense, but blind exploration of the periodic table is a daunting task. Fortunately, over a century of research in the physical sciences has provided us with myriad rules for assessing the feasibility of a given stoichiometry and the likelihood of particular crystal arrangements. Examples of chemical phenomenology include the radius ratio rules 4 and Pettifor maps 5 for structure prediction, as well as electronegativity and chemical hardness for predicting reactivity. 6 Pauling's rules 7 provide predictive power for ionic or heteropolar crystals. A wealth of knowledge exists for understanding the physical properties of tetrahedral semiconductors. 8 Recent examples of searches for new materials that draw from existing chemical knowledge include 18-electron ABX compounds, 9 hyperferroelectric superlattices, 10 and organic-inorganic perovskites. 11,12 The reliability and predictive power of atomistic material simulations is increasing. 13,14 Many approximations are being removed as high-performance supercomputers reach petaflop scale. This includes more accurate quantum ## The Bigger Picture The discovery of functional materials is critical for technological advancements that will play a role in addressing global challenges, ranging from catalysis for sanitation, semiconductors for harvesting solar energy, and biomimetic materials for health. There is a concerted global effort to reduce the time it takes to realize new materials via databases, highthroughput screening, informatics, and mapping out the ''materials genome.'' Here, we show how the compositional space for stoichiometric, inorganic materials can be quantified by simple rules and how the vast space can be explored quickly and cheaply with the use of key chemical concepts and element properties in the search for candidate materials with target properties. We exemplify the application of this approach by identifying a chalcohalide material with potential for watersplitting applications and carrying out a comprehensive search for new compositions that could adopt the widely studied perovskite crystal structure. mechanical treatment of electron-electron interactions in the solid state, 15 as well as more realistic models of chemical disorder. 16 However, because of the computational cost, high-throughput screening with first-principle techniques is usually limited to hundreds or thousands of materials-a small fraction of the overall phase space. We report a procedure for navigating the materials landscape with low computational effort, and it can be achieved with simple chemical descriptors. We first explore the magnitude of the task at hand by enumerating combinations of elements and ions for binary, ternary, and quaternary compositions. We demonstrate that chemical constraints can narrow the search space drastically. Examples of how deeper insights can be gained are illustrated for electronic (photoelectrodes for water splitting) and structured (perovskite-type) materials. The procedure can be used to comfortably explore the vast compositional space or as the first step in a multi-stage high-throughput screening process. Instead of being a roadblock to achieving new functionality, the combinatorial explosion for multi-component compounds provides fertile ground for the discovery of innovative materials. ## Elemental Combinations To begin, one can map chemical space by enumerating the ways in which the constituent elements of the periodic table can combine. If we restrict ourselves to the first 103 elements (to the end of the actinide series), the combinations (i.e., C 103 n ) for two, three, and four components are 5,253, 176,851, and 4,421,275, respectively. For five components, the combinations exceed 87 million. Physically, the situation is more complex. Elements can combine in different ratios, leading to variation in material stoichiometry, e.g., the binary combinations AB, AB 2 , A 2 B 3 , and A 3 B 4 . Given elements can also adopt multiple oxidation states, each with a unique chemical behavior, e.g., Sn(II)O, Sn(IV)O 2 , and Sn(II)Sn(IV)O 3 . For our enumeration of feasible compounds, we next consider the accessible oxidation states of each element in stoichiometry up to quaternary A w B x C y D z , where the integers w, x, y, and z % 8. This definition includes, for example, common ternary pyrochlore oxides (A 2 B 2 O 7 ) and quaternary double perovskites (A 2 BCO 6 ). Using the most common oxidation states extends the first 103 elements of the periodic table to 403 unique ions. The number of combinations is now drastically increased, as shown in Table 1, such that four-component candidate materials exceed 10 12 . In order to reduce this composition space, we can introduce selection rules (filters) from chemical theory. We note that the estimations discussed here represent a lower limit on the number of accessible materials. We consider regular inorganic compounds and exclude, for example, non-stoichiometry, organic systems, hybrid organic-inorganic materials, electrides, and intermetallics, for which additional considerations are required for predicting viability. Chemical Filters Rule 1: Charge Neutrality Ions tend to combine into charge-neutral aggregates. The same thinking applies to both ionic solids and more covalently bonded semiconductors. Any periodic solid must be charge neutral; otherwise, it would have an infinite electrostatic potential. Balancing oxidation states and fulfilling the valence octet rule are equivalent, e.g., III-V semiconductors, such as GaAs, can be represented as Ga 3+ As 3 . Our implementation is inspired by the work of Pamplin 20 and Goodman 21 on the subject of multi-component semiconductors. A charge-neutrality constraint significantly reduces the total number of candidate materials. The rule states that the formal charges (q) of the components sum to 0, i.e., wq A + xq B + yq C + zq D = 0. (Equation 1) Charge neutrality contracts the compositional space by at least an order of magnitude for binaries, ternaries, and quaternaries (Table 1). Rule 2: Electronegativity Balance Further to assuming that all charge-neutral combinations of oxidation states are accessible, we can implement a second constraint based on the electronegativity of the component elements. The empirical electronegativity (c) scale represents the ''attraction'' of a particular atom for electrons. For a stable compound, the relation c cation < c anion should be obeyed, i.e., the most electronegative element carries the most negative charge. Here, we employ the Pauling electronegativity scale, which reduces the allowed compositions by a factor of between 4 and 10 for the different numbers of components (Table 1). It is also instructive to consider existing materials databases (the ICSD and Materials Project). For binary compounds, we find fewer combinations from our estimates as implemented in SMACT (Semiconducting Materials from Analogy and Chemical Theory) than from the ICSD (Figure 1), which can largely be attributed to our exclusion of intermetallics and polymorphs. In the Materials Project, multiple entries for a single crystal structure and chemical composition are removed, and the number of compositions are in close agreement. For ternaries and quaternaries, the compositions passing both charge and electronegativity tests continue to rise exponentially, whereas the number in existing databases remains relatively constant. The increased complexity of ternary and quaternary systems means that their synthesis, characterization, and reporting are more challenging than for binary systems. Nevertheless, the large differences between the numbers of potential and reported materials suggest that wide areas of unexplored compositional space may contain stable and useful materials. a q, charge neutralitiy; X, electronegativity balance. The numbers reported in this section are vast, and using modern electronic-structure techniques to perform quantitative screening for application is unimaginable. Exploration of the hitherto neglected compositional space will require further guidelines. In the following sections, we demonstrate how additional descriptors can be applied for identifing materials for specific applications. ## Compositional Descriptors Several useful properties can be estimated from knowledge of the chemical composition alone, and here we explore the application of some of these approaches. ## Descriptor 1: Electronic Chemical Potential The concept of atomic electronegativity has been successfully extended to solids, where the geometric mean becomes the single-value descriptor, i.e., This descriptor represents a mid-gap energy between the filled (valence band) and empty (conduction band) electronic states. This corresponds to the electronic chemical potential (Fermi level) at 0 K. 22 Butler and Ginley 23 found a linear correlation between the solid electronegativity and the electrochemical flat-band potentials for a range of semiconductors. This was subsequently extended to a wider dataset including metal oxides, chalcogenides, and halides. 24 The method provides a rapid procedure for the estimation of absolute electron energies for multi-component materials. It is now commonly used in the computational screening of new materials for electrochemical applications. Descriptor 2: Electronic Structure Many tight-binding model Hamiltonians exist for semiconductors and dielectrics. 8 One recent approach is based on the atomic solid-state energy (SSE) scale, 29 which provides information on valence and conduction bands on the basis of the frontier orbitals of the constituent ions. Whereas the Mulliken definition of electronegativity is an average of the ionization potential (IP) and electron affinity (EA) of an atom, the SSE reflects the IP of an anion (filled electronic states) and EA of a cation (empty electronic states). The energies of 40 elements were fitted from a test set of 69 closed-shell binary inorganic semiconductors, 29 which has recently been extended to 94 elements. 30 According to the tabulated SSE scale, the band gap (E g ) can be estimated as For multi-component systems, the limiting values (cation with highest EA and anion with lowest IP) are used. The SSE has a root-mean-square deviation of 0.66 eV against the measured band gaps of 35 ternary semiconductors (see Table S1). This simple method allows for rapid screening of band gaps and absolute bandedge alignment. Both methods (Equations 2 and 3) have been implemented for arbitrary compositions on the basis of tabulated atomic data in the SMACT package. Because no crystal-structure information is included at this level, the results are qualitative, and the models do not distinguish, for example, between polymorphs. ## Electronic Structure: Photoelectrodes We now use the compositional space and chemical descriptors defined above to search for potential materials for solar fuel generation via photoelectrochemical water splitting. The properties that are required for viable photoelectrodes include (1) a band gap in the visible range of the electromagnetic spectrum to absorb a significant fraction of sunlight and (2) upper valence and lower conduction bands bridging the water oxidation and reduction potentials, enabling the redox reaction. We set an optimal band-gap range between 1.5 and 2.5 eV. Although the free energy for water dissociation is 1.2 eV, the combination of loss mechanisms found in practical devices could require a band gap as large as 2.2 eV. 31,32 Metal oxides combine many attractive properties for water splitting (e.g., stability and cost), but they usually have band gaps too large to absorb a significant fraction of sunlight. The formation of multi-anion compounds offers a route to modifying the electronic structure. We consider ternary metal chalcohalides (i.e., A x B y C z ), with B = [O,S,Se,Te] and C = [F,Cl,Br,I]. We restrict the A cations to those with an SSE higher than the water reduction potential (approximately 4.5 V in relation to the vacuum at pH = 0). The conditions of charge neutrality and electronegativity are used for performing an initial screening that yields 52,094 combinations. With the additional band-gap criterion, the combinations are reduced to 7,676, and the pool of cations is reduced from 25 to 7 with A = [B,Ti,V,Zn,Ga,Cd,Sn]. We further rule out any boron-containing combinations at this stage, because these are known to form discrete molecular units (e.g., BClSe). Finally, we screen compositions according to the environmental sustainability of the elements. We use the Herfindahl-Hirschman Index (HHI R ), which has been developed in the context of thermoelectric applications, for elemental reserves. 33 This index includes factors such as geopolitical influence over materials supply and price. The HHI R for a given composition can be obtained as the weighted average over the constituent elements. At this stage, because stoichiometry is variable, we consider the mean HHI R for each A 1 B 1 C 1 combination. The band-edge positions of the 20 candidates with the smallest HHI R values are presented in Figure 2. The HHI R has the effect of eliminating all combinations containing Ga, Te, and Br (although relatively abundant, most of the world's Br is produced from the Dead Sea, making it geopolitically sensitive, as reflected in a high HHI R ). There are no entries in the ICSD for most of the candidates that we identified; however, reports can be found for Cd 2 O 6 I 2 , Sn 2 SI 2 , and Zn 6 S 5 Cl 2 . Both Cd 2 O 6 I 2 and Sn 2 SI 2 feature in the Materials Project and have band gaps of 3.3 and 1.6 eV, respectively, calculated within density functional theory (DFT). These compare with the SSE band gaps of 2.5 and 2.0 eV. The third compound, Zn 6 S 5 Cl 2 , is reported to have an optical gap of 2.7 eV 36 , which compares with the SSE band gap of 2.4 eV. Only one oxygen-containing compound survived the band-gap screening criterion; the values for metal oxyhalides are generally too large. For O y I z , the iodide forms the upper valence band (low binding energy of I 5p), whereas it is the oxide (O 2p) for other halides. However, the sensitivity of the oxide ion to its crystal environment is well documented, 27,37 and consequently its SSE carries the greatest uncertainty. 29 This is one aspect where knowledge of the local structure (electrostatic potential) could significantly improve the accuracy of the results. We must connect composition to crystal structure in order to make more accurate property predictions. Global optimization of crystal structures from first principles is a formidable task, although great progress is being made in this area. 38 We instead adopt an approach based on analogy with known structures through chemical substitutions, as developed by Hautier et al. 39 It uses data-mined probability functions, as implemented in the Materials Project. To demonstrate the translation from composition to material, we performed crystalstructure mining for the four combinations with the lowest HHI R . The 88 predicted structures were then subjected to a full DFT lattice optimization procedure and ranked by total internal energy. Finally, accurate band gaps were predicted for the lowest-energy structures by hybrid DFT (HSE06 electron exchange and correlation 40,41 ). The compound Sn 5 S 4 Cl 2 has an indirect band gap of 1.6 eV and a direct gap of 1.8 eV, which lies within the target range. The band gaps of the other three lowest-energy compounds were calculated to be between 3.0 and 3.4 eV. Full details of the workflow (Figure S1) and band gaps (Table S2) can be found in the Supplemental Information. The newly identified compound, Sn(II) 5 S 4 Cl 2 , adopts a structure formed from two distinct Sn centered polyhedra: (1) a distorted octahedron with equatorial S and apical Cl ions and (2) a distorted tetrahedron with 4 S ions and a stereochemically active Sn lone pair (Figure S2). The polyhedra form interlocking chains in three dimensions. The electronic density of states reveals an upper valence band composed of hybridized Sn s -Cl p orbitals; such Sn s-based valence bands are considered promising indicators for hole mobility. 42 The lower conduction band is composed mainly of overlapping Sn p orbitals. The chemical structure and bonding characteristics suggest that this material should have favorable carrier transport, crucial for optoelectronic applications. Crystal Structure: Perovskites One of the most successful approaches to discovering new materials is structural analogy. The concept is to take a crystal structure with a known chemistry and to replace elements within the structure to tune the physical properties. In the simplest case, this involves direct isovalent substitution, e.g., Zn(II)S / Cd(II)S. Structural analogy can be extended to aliovalent cross-substitution (also termed cation mutation), e.g., Zn(II)S / Cu(I)Ga(III)S 2 . A systematic methodology was outlined more than 40 years ago in a paper by Pamplin 20 for enumerating charge-neutral tetrahedral semiconductors. The challenge of going beyond tetrahedral semiconductors is predicting crystal structure. The radius of ions within a lattice has a long history as a geometric descriptor of structural stability. A key example is the application of radius ratio rules by Goldschmidt 43 to predict the propensity of a ternary ABC 3 combination to form the perovskite structure: where t is the tolerance factor and r is the ionic radius. Values of t > 1 imply a relatively large A site favoring a hexagonal structure, 0.9 < t < 1 predicts a cubic structure, and 0.7 < t < 0.9 means that the A site is small, preferring an orthorhombic structure. For t < 0.7, other (non-perovskite) structures are predicted. These rules have recently been extended to describe structure-property relationships in hybrid organic-inorganic perovskites. 11,12 In this section, we apply our screening procedure to include knowledge of the crystal structure and estimate the size of the perovskite materials space. We start by enumerating the elemental combinations. We then reduce the set by requiring an octahedral coordination environment for the B site, as contained in the Shannon dataset, 44 and require a combination of oxidation states that are charge neutral. This list is then assessed in terms of t, as defined by the Shannon ionic radii. 44 We consider single-anion compositions based on C = [O,S,Se,F,Cl,Br,I]. The chargeneutrality and octahedral B-site constraints reduce the 176,851 elemental combinations to 41,725. The tolerance-factor constraint, 0.7 < t < 1.0, further reduces this to 26,567. For potential applications in the energy sector, we can consider candidates with HHI R smaller than that of CdTe (a commercial thin-film photovoltaic material), resulting in a final population of 13,415. For each anion, an orthorhombic perovskite structure is the most common prediction, and hexagonal is the most rare (Figure 3). The fraction of cubic perovskite structures remains roughly constant within the respective halide and oxide or chalcogenide series; however, it is more dominant for the halides. The presence of Br or I makes a material less sustainable (higher HHI R ); otherwise, there is little to differentiate the anions. Far more oxide and chalcogenide perovskites are predicted than halides. The higher anion charge allows for three distinct cation combinations (I-V, II-IV, and III-III), whereas halides have only I-II. In addition, a greater radius compatibility is found for the group VI anions. We find that the number of plausible perovskite structures increases with the anion radius; however, the lower crustal abundance for heavier elements reduces the number that meet the sustainability criterion. A search of the Materials Project over the same anion space reveals 920 materials, a small fraction of those predicted from SMACT (26,567). The search includes all standard perovskite space groups. 45 For oxide perovskites, 8.26% of those identified from SMACT are found in the Materials Project; for sulfides, this falls to 0.45% and to 0.12% for selenides. To some extent, the greater number of oxide perovskites discovered reflects the greater research activity in this field; however, synthesis of chalogenide perovskites has been reported, and there is interest in these materials for technological applications. 49,50 Of the ABC 3 materials reported in the Materials Project, 48% of oxides, 35% of sulfides, and 20% of selenides are in perovskite space groups. Why are there so few chalcogenide perovskites? The tolerance-factor arguments that work well for metal oxides may not hold for chalcogenide perovskites. Oxygen forms more ionic compounds because of its higher electronegativity and lower polarizability than those of S, Se, and Te. When covalent bonding becomes prevalent, it is known to result in deviations from tolerance-factor behavior. An example is the case of NaSbO 3 , for which t = 0.92 is commensurate with the formation of cubic perovskite but which forms the non-perovskite ilmenite structure. Goodenough and Kafalas 51 explained this deviation as a result of strong s bonding between Sb and O. This procedure demonstrates the power of searching through materials on the basis of structural analogy. Only a small fraction of possible perovskite materials have been synthesized. Although some may not represent thermodynamic ground states, they could be accessible through kinetic control of crystal growth or templated on a substrate. Many interesting chalcogenide perovskites are waiting to be discovered. The final pool of 13,415 feasible compositions is within the grasp of explicit computation by quantum mechanical methods, albeit as part of an ambitious project. Indeed, high-throughput screening of 5,400 multi-anion cubic perovskite structures via DFT has been reported 25,52 and revealed 32 promising new materials for watersplitting applications. ## Conclusions We have demonstrated the utility of chemical theory in quantifying the magnitude of the compositional space for multi-component inorganic materials. Even after the application of chemical filters, the space for four-component materials exceeds 10 10 combinations. We further estimate that the five-component space exceeds 10 13 combinations. There are many applications in which materials with even higher-order compositions have been developed, e.g., in high-temperature superconductors, where six to seven component materials are common. The number of potential materials is not infinite, but it is immense. The scale of the combinatorial explosion emphasizes the need for effective material-design procedures that employ existing chemical and physical knowledge in a targeted manner. Stochastic sampling of this chemical space is unlikely to be effective in yielding materials with specific functionality. We have presented a procedure that uses simple descriptors to support materials exploration, discovery, and design. All element counts and plots presented in this paper were created with custom codes based on SMACT and written in the Python 2.7 programming language. Elemental data are collated from multiple sources (see Table 2) and made HHI elemental Herfindahl-Hirschman Index calculated from geological and geopolitical data 33 Ionization potential NIST Atomic Spectra Database 59 Pauling electronegativity updated values of electronegativity on Pauling's scale were compiled in the CRC Handbook; 53 for elements 95 (Am) and above, Pauling's recommended value of 1.3 was used; 60 the value for krypton (3.0) was derived from the bond energy of KrF 2 and reported in a scientific paper 61 SSE ''solid-state energy'' model of semiconductors and dielectrics 29,30 SSE (Pauling) extended estimates of solid-state energy from the correlation between known values and Pauling electronegativity 30 Where possible, values recommended by the National Institute of Standards and Technology (NIST) were used. ## EXPERIMENTAL PROCEDURES Chem 1, 617-627, October 13, 2016 625 algorithmically accessible in a unified object-orientated interface. Example routines that check element and oxidation-state combinations against the conditions of charge neutrality and electronegativity are provided. Scripts that generate the results and plots reported in this paper are available with the SMACT codes. A number of tutorials working through the combinatorial explosion are provided at https://github.com/WMD-group/SMACT_practical. The codes, collectively named Semiconducting Materials by Analogy and Chemical Theory, are inspired by the pen-and-paper procedure reported by Pamplin in 1964. 20

chemsum

{"title": "Computational Screening of All Stoichiometric Inorganic Materials", "journal": "Chem Cell"}

interactions_between_shape-persistent_macromolecules_as_probed_by_afm

## Abstract: Water-soluble shape-persistent cyclodextrin (CD) polymers with amino-functionalized end groups were prepared starting from diacetylene-modified cyclodextrin monomers by a combined Glaser coupling/click chemistry approach. Structural perfection of the neutral CD polymers and inclusion complex formation with ditopic and monotopic guest molecules were proven by MALDI-TOF and UV-vis measurements. Small-angle neutron and X-ray (SANS/SAXS) scattering experiments confirm the stiffness of the polymer chains with an apparent contour length of about 130 Å. Surface modification of planar silicon wafers as well as AFM tips was realized by covalent bound formation between the terminal amino groups of the CD polymer and a reactive isothiocyanate-silane monolayer. Atomic force measurements of CD polymer decorated surfaces show enhanced supramolecular interaction energies which can be attributed to multiple inclusion complexes based on the rigidity of the polymer backbone and the regular configuration of the CD moieties. Depending on the geometrical configuration of attachment anisotropic adhesion characteristics of the polymer system can be distinguished between a peeling and a shearing mechanism. ## Introduction Shape-persistence is an important key feature in self-organisation strategies of supramolecular building blocks resulting in high structural perfection of the obtained molecular assemblies , such as shape persistent macrocycles, cage compounds or rotaxanes . Especially shape-persistent polymers are of significant scientific interest as their defined structural characteristics offer various applications as sensor materials, biomimetic filaments or organic electronics . Furthermore, compared to polymers with flexible chains, shape persistent macromolecules with high structural rigidity are able to form stable aggregates based on multiple supramolecular interactions, which can be detected and quantified without the presence of side effects, such as self-passivation or coiling processes. Dendrimers, nanoparticles and shape-persistent polymers had been previously discussed as scaffolds for the design of multiple ligands of high affinity . Nevertheless, well-defined model systems in which the influence of rigidity and regularity on cooperativity of binding was systematically investigated have not been reported so far. Rigid linear polymers have been considered as suitable scaffolds for the design of supramolecular systems showing multiple interactions. A high rigidity of the macromolecule is maintained by rigid, linear repeat units, such as trans-ethenylene, ethynylene, or p-phenylene moieties. The observed persistence lengths of polyconjugated polymers ranged from 6 to 16 nm, depending on the side groups and the method of determination . Among many supramolecular interactions, such as hydrogen bonding, π-π-interactions or hydrophobic host-guest interactions , the interactions of cyclodextrins (CDs) with hydrophobic guest molecules are of special interest, since CDs are readily available bio-based materials and interactions take place under physiological conditions . CDs are ideal candidates for the investigation of multivalent interactions as they combine high affinities with a versatile integrability in macromolecular systems . CDs have already been employed for the construction of supramolecular polymers , supramolecular hydrogels , molecular printboards or multivalent interfaces with tunable chemical and physical properties. Herein, for the first time, we present studies concerning the synthesis of shape-persistent CD polymers to investigate multivalent binding with ditopic guest molecules on the molecular level (Figure 1). The ditopic guest (shown in red colour) should act as a connector between opposing CD moieties. Only a few examples of shape-persistent CD polymers have been reported so far, including CD-modified conjugated oligomers and polymers composed of rigid phenylene ethynylene (PPE) structure units which are able to form self-inclusion complexes with tunable electrochemical properties . The synthesis of PPE, in which two β-CD rings were attached to every second phenylene group, was described by Ogoshi et al. using a Sonogashira-Hagiwara coupling. We preferred a poly-phenylene-butadiynylene backbone, synthesized by a Glaser-Eglington coupling, since the repeating unit is long enough (l = 0.944 nm) to allow the connection of one CD moiety at each phenylene unit. Based on the stiffness of the polymer chain self-passivation of CD polymer modified surfaces is reduced to a minimum. Furthermore, the ethynyl end groups are easily functionalized by click chemistry. Isothermal titration calorimetry (ITC), fluorescence spectroscopy, quartz crystal microbalance (QCM), surface plasmon resonance (SPR) and atomic force microscopy (AFM) have been employed to quantify the strength of the multivalent interactions . Because binding affinities can be very high for multivalent supramolecular systems, the constituents are commonly used in low equilibrium concentrations. Since AFM even allows the investigation of single molecules, such as DNA or molecular self-assembling based on "Dip-Pen" nanolithography , it was chosen as the most reliable technique to probe highly cooperative recognition processes. The investigation of cooperativity of multiple host-guest interactions using AFM has been reported by several groups . Huskens and co-workers measured the supramolecular interactions between a β-CD-modified planar surface and mono-, diand trivalent adamantane guest molecules attached to an AFM tip and found enhancement factors up to 2, depending on the force loading rate . We have previously explored the adhesion characteristics of dense CD layers on an AFM tip and a planar silicon surface connected by various ditopic linker molecules. In this system we were able to switch adhesion and friction by applying external stimuli onto the responsive ditopic linkers . In contrast to previous work our molecular toolkit, based on ditopic connector molecules, allows the independent determination of unspecific interactions between CD polymers at tip and planar surface as well as the specific interactions to ditopic connector molecules. In the following, we describe the first example of multivalent interaction of ditopic guest molecules with shape-persistent CD polymers covalently attached to an AFM tip and a planar surface. Nano force measurements between CD and CD polymer, CD polymer and CD, and CD and CD at the tip and the planar surface, respectively, exerted by the adamantane ditopic connector molecules were systematically investigated. All four configurations are schematically depicted in Figure 2. ## Synthesis of the shape-persistent CD polymer Our synthetic approach for the preparation of modified poly(phenylene butadiynylene)s bearing one CD molecule per repeat unit started from 2,5-dibromo-4-methylbenzoic acid (2) , which was esterified to 3 with tert-butanol catalyzed by H 2 SO 4 (Scheme 1). The TMS-protected diacetylene derivative 4 was prepared by Sonogashira reaction of 3 with trimethylsilylacetylene. Subsequent deprotection of the TMS groups using tetra-n-butylammonium fluoride and saponification of the tert-butyl ester with trifluoroacetic acid resulted in the corresponding benzoic acid 6. The latter was coupled to 6-monoamino-6-deoxy-β-CD using N,N'-dicyclohexylcarbodiimide (DCC) and 1-hydroxybenzotriazole (HOBt) applying a procedure known for terephthalic acid . The resulting product, monomer 7, was easily isolated due to its low solubility in water which was attributed to self-inclusion between hydrophobic phenyl moieties and β-CD rings leading to daisy chains . The polymerization of 7 was performed through Glaser coupling in pyridine catalyzed by Cu(I)/Cu(II). After removal of low molecular weight material by ultrafiltration polymer 8 was isolated as a light orange solid in 91% yield. Polyrotaxane formation, which might prevent the accessibility of the CD-moieties located on the polymer backbone, was avoided by the presence of pyridine as a non-polar solvent. Both NOESY NMR experiments and circular dichroism (results not shown) do not indicate any significant interaction of the CDs and the aromatic backbone. Compared to monomer 7, peak broadening and the disappearance of the 1 H NMR signals of the acetylene protons at 4.54 and 4.36 ppm indicate the formation of polymer 8. The presence of the conjugated backbone was confirmed by UV-vis and fluorescence measurements in water. Compared to 7, a characteristic bathochromic shift could be observed both in the absorption and emission spectra of polymer 8 (Figure 3) showing the presence of the extended polyconjugated π-system. Quantitative information about the molecular weight distribution of 8 was obtained by MALDI-TOF measurements using an ionic liquid matrix (HABA/TMG 2 ) . A representative MALDI spectrum, shown in Figure 4, exhibits a wide range of broad signals starting from the signal of the dimer at m/z 2,621.33 Da detected as [M + Na] + and ending at the 38mer at m/z 48,196.23 Da for a S/N ratio ≥3, with an average 1297.4 mass units shift corresponding to one additional repeating unit. Among each discrete envelope, one to three supplementary ions, have been detected with a constant 165.2 mass unit shift, revealing the presence of small quantities of the repeat unit originating from unmodified benzoic acid derivative 6, e.g., at 2,621.33 and 2,786.52 Da (Figure 4). The MS analysis reveals the high structural perfection of the polymer 8 where at most one CD entity per polymer molecule is missing. Integration of the relative distribution of the most intense ions of each population allowed to estimate both the number average molecular weight, M n , and the mass average molecular weight, M w , of 8,765.77 Da, and 22,023.56 Da, respectively. These values result in a polydispersity index PDI = M w /M n of 2.59 typical for normal distributions. From the value of M w an average contour length L = 17 nm of the macromolecule was calculated. A more detailed analysis of the MS data is provided in Supporting Information File 1. ## SANS and SAXS measurements of the CD polymer Structural characteristics of the CD polymer 8 have been investigated by small-angle neutron and X-ray scattering experiments (SANS/SAXS). SANS data (KWS-1, JCNS at Heinz Maier-Leibnitz Zentrum ) for a polymer concentration range from 0.005 to 0.03 g/cm 3 are presented in Figure 5. SANS intensities are normalized to polymer concentration and therefore scattering intensities depend on polymer chain mass (or mass of chain aggregates), square of scattering contrast, conformation of polymer chain, and interaction between the Table 1: Structural parameters of polymer 8 (apparent radius of gyration, scattering at zero angle, radius of gyration of polymer cross-section, scattering at zero angle of polymer cross-section, apparent contour length obtained from the ratio between I(0) and I CS (0), and calculated apparent mass of polymer 8, obtained from the length of monomer unit chains (aggregates). There are only minor differences in scattering for concentrations up to 0.02 g/cm 3 indicating no significant aggregation between polymer chains with increasing concentration which would lead to highly ordered polymer species. The decrease of scattering intensity for the highest concentration of 0.03 g/cm 3 can be attributed to interaction of polymer chains. The SAXS curve measured at 0.03 g/mL shows a similar shape as the neutron data (Supporting Information File 1, Figure S1). The low-q range of scattering data has been analyzed with a Debye function. The apparent radius of gyration R g,app and the scattering at "zero angle", I(0), were obtained by fitting the scattering data for q < 0.02 −1 : ( where x= q 2 R g,app 2 . The scattering intensity is given by (2) where the apparent molar weight, M app , is connected with the real molar weight, M, via a structure factor S(0) (interaction among polymer chains) as M × S(0) = M app and Δρ m is the difference in neutron scattering length density between polymer and solvent normalized to the density of polymer. The local structure of the polymer cylindrical cross-section was extracted by applying indirect Fourier transformation (IFT) to the experimental data from the high-q range. Detailed information applying this method is presented in Supporting Information File 1. The resulting parameters for the concentration dependence of I(0), scattering at "zero angle" of a cylindrical crosssection of polymer I CS (0), radius of gyration R g,app , and radius of gyration of a cylindrical cross-section R g,CS are presented in The flexibility of chains of polymer 8 was determined by means of a Holtzer plot . Detailed information and the corresponding data are presented in Supporting Information File 1. The absence of a characteristic inflection point, where the scattering intensity changes from q −1 as for rigid cylinder to q −2 (or to q −5/3 when self-avoidance is important) as for flexible chains, indicates that polymer chains are short and rigid, i.e., that the persistence length is of the same order as the contour length of the polymer. The SAXS data has been analyzed by models representing the expected shape of polymers. It was assumed that there is no interaction between aggregates, which means that the scattering intensities depend only on the size and shape of the aggregates . Details are shown in Supporting Information File 1. The scattering data could be described (Figure 5 above and Figure S1 in Supporting Information File 1) by a population of rigid cylinders of length 110 ± 5 and radius of cross-section of 12 ± 2 . Neglecting the interaction between polymer chains in the model leads to the slightly lower length values. ## Complexation of monotopic and ditopic guests In contrast to monomer 7, polymer 8 was soluble in water up to a concentration of 0.15 mM (based on the repeating unit). This allows the investigation of the complexation of ditopic and monotopic guests, 9 and 10, respectively. The solubility of the host polymer 8 as a function of the concentration of both guests 9 and 10 (Scheme 2) was determined by UV−vis spectroscopy using the extinction coefficient ε of 8 (14,800 M −1 cm −1 ) at 425 nm. A more detailed description of the solubility measurements is presented in Supporting Information File 1. Addition of hydrophilic guest 10 caused an increase in solubility of host polymer 8 in water (Figure 6). The surprisingly steep initial slope of the phase solubility diagram, m = 1.4 (repeating unit/guest) could be well represented by a model where every second CD moiety has to be complexed by the hydrophilic guest to significantly improve the solubility in water. Binding constants of about 40,000 M −1 , which were in the same range as literature values for the incorporation of adamantane derivatives into β-CD, were obtained using ITC measurements considering a two-step sequential complexation with guest 10. Further information is provided in Supporting Information File 1. Incomplete complexation with cationic guest molecules is indicated by a significant lower binding constant of 670 M −1 for the second binding complexation step, which is strongly inhibited as a result of the electronic repulsion of charged guest molecules in close proximity to each other. In contrast, a pronounced reduction of the solubility of CD polymer 8 was observed in the presence of ditopic guest 9, which was attributed to the interconnection of polymer chains through the complexation of the ditopic guest. The very low concentration of connector 9 necessary for the almost complete precipitation of the host polymer 8 can be explained by the high integrability of the host-guest system based on the shape-persistence of the polyconjugated polymer backbone of 8. cyanate groups, which smoothly react with amines forming stable thiourea links . Monolayers of β-CD or β-CDpolymer were obtained by attachment of monoamino β-CD or amino-modified CD polymer 12, synthesized from polymer 8 (Scheme 3) through Cu(I)-catalyzed azide-alkyne cycloaddition (CuAAC) with the triethylene glycol linker 11 (N 3 -TEG-NH 2 ) which had been prepared in a five-step procedure . ## Probing multivalent interactions by AFM The adhesive forces of 12, due to supramolecular interactions with ditopic guest 9, between a planar silicon surface and an AFM tip both modified with the CD polymer 12 or 6-monoamino-6-deoxy-β-CD were systematically investigated by AFM. While adhesion was very weak in pure water, significant adhesion took place over a wide range of distances in a 10 μM solution of ditopic guest 9 (Figure 7a-d). For comparison, we also investigated the adhesion forces between CD and 12, 12 and CD, and CD and CD at the tip and the planar surface, respectively, caused by the adamantane connector 9. Adhesive forces were recorded as function of the tip-surface distance upon retracting of the tip from the surface for all four configurations. The pull-off force required to detach the tip from the surface in the presence of connector molecules was of the order of 500 pN for the CD-CD configuration and about 1 nN for all configurations involving CD polymers (12). These values are significantly higher than the pull-off forces of about 250 pN measured in control experiments for all configurations. The graphical summary in Figure 7a suggests that the pull-off forces for the 12-12 configuration are slightly higher than for the 12-CD and for the CD-12 configuration. While the pull-off force is similar, the overall appearance of the force curves differs for the three polymer configurations. The interaction distance varies significantly for the different configurations. The CD-CD configuration has the shortest and the polymer-polymer configuration the longest range of interactions. The interaction range can be quantified by the tip-surface distance at which the last rupture occurs, referred to as maximum rupture length. The histograms of the maximum rupture length for all four configurations are presented in Figure 7. For the CD-CD configuration, the most probable maximum rupture length of 5 nm corresponds to the combined height of the monolayers on tip and surface, each of about 2.5 nm. The typical rupture length for the CD-12 configuration is 10 nm, while it is 29 nm for the 12-CD configuration. The difference in maximum rupture length indicates a difference in the detachment mechanism. In the CD-12 configuration, the polymers bind to the sloped facets of the asperity of the AFM tip. Upon pulling, the polymers are sheared from the tip apex by rupturing all bonds simultaneously leading to one large rupture peak at a small tip-surface distance. For the 12-CD configuration, a force plateau observed in the force-distance curve in Figure 7c reveals the peeling of a polymer chain from the CD-coated surface resulting in a rupture length similar to the length of the polymer chains. For the 12-12 configuration, many additional small detachment events lead to a broadening of the pull-off curve and reveal the rupture of bonds for tip-surface distances as large as 110 nm in Figure 7d. The broad distribution of rupture length, which extends to roughly the double of that of the 12-CD configura- tion, indicates that individual long polymer chains interlock, explaining also the characteristic stretching events in the forcedistance curve. The most probable maximum rupture length for the 12-12 system is 38 nm, which is double of the average polymer length of 17 nm predicted from the MALDI-TOF and SANS/SAXS results. The agreement confirms the picture that the maximum rupture length reflects the final detachment of supramolecular bonds at the end of stretched polymer chains attached to AFM tip and surface. The higher sensitivity of our AFM set up compared to the MALDI-TOF instrument allowed us to even detect single rup- ture events at a distance up to 250 nm, which proved that some individual chains had a length of at least 125 nm. Compared to MALDI-TOF measurements in which the small number of high molecular weight polymer chains are hardly detectable, AFM experiments overemphasize the few longest polymer chains probing the interactions of the regularly spaced CDs in CD polymer 12 and ditopic connector molecules. Due to this observation AFM is an excellent detection tool for analysing cooperative effects in ordered supramolecular systems. The differences between the four configurations of functionalization can be further quantified by integration of the force curves, resulting in the work of separation which has been employed before as a suitable parameter for the quantification of polymer detachment . In line with the characteristic shape of the example force curves, the work of separation increased significantly in the order CD-CD, CD-12, 12-CD and 12-12 configuration (Figure 8). The relative increase in the work of adhesion from control experiments to measurements of the specific interactions caused by the connector molecule 9 was even higher than the respective increase in pull-off force due to the very short range of the non-specific adhesive interactions. The significant difference in the interaction range and thus in the work of separation between CD-12 and 12-CD configuration can be explained by the asymmetry between curved tip and flat surface and the resulting difference in the detachment mechanism. Polymers attached to the surface bind to the side faces of the tip with its nanometer-scale apex radius. Upon retraction, the force acts along the polymer and shears the polymer off the tip, with all bonds rupturing more or less simultaneously. In contrast, polymers attached to the tip bind to the flat surface such that upon retraction the polymer is peeled from the surface by the orthogonal force, one bond breaking after another. The different detachment scenarios are depicted in the schematic drawings in Figure 2. The shearing configuration (CD-12) leads to simultaneous rupture of all bonds, while the peeling configuration (12-CD) involves bending of the polymer and consecutive rupture. The strongest adhesion is offered by the supramolecular interlocking of polymers attached to tip and surface. Supramolecular interconnection between two CD polymer 12 molecules through the ditopic guest 9 is expected to be superior to the one between CD polymer 12 and CD because of the higher regularity of the CD spacing at the polymer compared to the spacing within the CD monolayer. We conclude that the regularity of the CD polymer 12 allows to establish a much higher number of supramolecular bonds with the connector 9 giving rise to about a fivefold enhancement of the work of separation. Many force curves exhibit a well-defined last rupture event. A representative example is shown in Figure 9a, where the force drops from around 63 pN to zero at a distance of 110 nm. The distribution of rupture forces for the last rupture events, shown in Figure 9b, has a clear maximum at 63 pN, determined by a Gaussian fit to the distribution, and a weak second maximum at about double this value. We conclude that 63 ± 10 pN is the rupture force for a single bond between our supramolecular polymers 12 established by the ditopic guest 9. The value agrees with rupture force measured for adamantane-CD complexes with CD molecules in the surface layers when the stiffness of the AFM cantilever is taken into account . Force curves like those shown in Figure 7 can be repeated on the same spot of one sample many times with very similar results. The repeatability confirms the reversibility of the underlying interactions. It is difficult to estimate the number of supramolecular bonds contributing to pull-off forces of 1 nN in to the one for the CD-CD system previously described . This spike force may be enhanced by the multivalency effects discussed above, but its strength indicates that more than one polymer molecule might be involved. ## Conclusion In conclusion, regular water-soluble shape-persistent CD polymers based on poly(phenylene butadiynylene) were prepared by a straightforward Glaser coupling/click chemistry approach, which can be attached to planar silicon surfaces as well as AFM tips. Structural perfection of the resulted polymers was con-firmed by MALDI-TOF measurements revealing the presence of high molecular weight materials with up to 38 repeat units. High integrability of the scaffold was proven by UV-vis supported solubility measurements upon addition of ditopic adamantane connectors. Small-angle neutron scattering and X-ray experiments reveal the presence of stiff cylindrical polymer chains with contour lengths of about 13-16 nm, which corresponds to the values obtained by MALDI and AFM measurements. Hard substrates with the shape-persistent polymers and interconnected by ditopic guest molecules require about five times higher separation energies than those functionalized with conventional CD monolayers. This significant enhancement of adhesion can be attributed to a strong cooperative effect favored by the rigidity of the polymer backbone and the regular spacing of the CD moieties. The range of adhesive interactions could be extended from 5 to 38 nm, which will also allow the interconnection of surfaces with higher roughness. The stiff polymers exhibit a clear contrast between shearing and peeling mechanisms, depending on the geometrical configuration of attachment. The distribution of the maximum rupture lengths in the force microscopy experiments confirms the molecular weight distribution of the CD polymers estimated by MALDI-TOF and the average contour length determined by SANS/SAXS. In addition, force microscopy experiments emphasize the longest polymer chains and their maximum length.

chemsum

{"title": "Interactions between shape-persistent macromolecules as probed by AFM", "journal": "Beilstein"}

nickel_catalyzed_electrochemical_c(sp_2_)−c(sp_3_)_cross-coupling_reactions

## Abstract: HIGHLIGHTS Electrochemical C(sp 2 )−C(sp 3 ) coupling reactions are developed using bench stable, inexpensive substrates and Ni catalysts;  The electrochemical cross-coupling exhibits broad substrate scope and good yields;  The electrochemical cross-coupling are practical in making pharmaceutical candidates; Reaction scalability was demonstrated using flow cell synthesis. with a broad scope, good yields, and practical applications, which expands the synthetic toolbox to forge carbon-carbon bonds.Summary: Nickel (Ni) catalyzed carbon-carbon (C−C) cross-coupling reactions haves been considerably developed in last decades and has demonstrated unique reactivities compared to palladium. However, existing Ni catalyzed cross-coupling reactions, despite success in organic synthesis, are still subject to the use of air-sensitive nucleophiles (i.e. Grignard and organozinc reagents), or catalysts (i.e. Ni 0 pre-catalysts), significantly limiting their academic and industrial adoption. Herein, we report that, through electrochemical voltammetry screening and optimization, redox neutral C(sp 2 )-C(sp 3 ) cross-coupling reactions can be accomplished in an undivided cell configuration using bench-stable aryl halide or β-bromostyrene (electrophiles) and benzylic trifluoroborate (nucleophiles) reactants, non-precious, bench-stable catalysts consisting of NiCl2•glyme pre-catalyst and polypyridine ligands under ambient conditions. The broad reaction scope and good yields of the Ni-catalyzed electrochemical coupling reactions were confirmed by 50 examples of aryl/β-styrenyl chloride/bromide and benzylic trifluoroborates. Its potential applications were demonstrated by electrosynthesis and late-stage functionalization of pharmaceuticals, and natural amino acid modification. Furthermore, to testify practical industrial adoption, three electrochemical C−C cross-coupling reactions were demonstrated at gram-scale in a flow-cell electrolyzer. An array of chemical and electrochemical studies mechanistically indicates that the studied electrochemical C−C cross-coupling reactions proceed through an unconventional radical trans-metalation mechanism. The presented Ni-catalyzed electrochemicalC(sp 2 )-C(sp 3 ) cross-coupling paradigm is highly productive, easily operative, and atomically economic, and is expected to find wide-spread applications in organic synthesis. The Bigger picture: Electrosynthesis has recently been recognized an enabling technology for organic synthesis. In principle, substrates or catalysts in electrosynthesis can be selectively anodically or catholically activated to participate desired reaction sequences in an electrolyzer. The attractive synthetic merits of electrosynthesis include migrating the use of reactive (sometimes even dangerous) oxidants and reductants, enabling the access of highly reactive catalytic intermediates which are not easily handled in traditional thermal reactions, and thus representing a green, atomically economical synthetic strategy. Most of reported electrosynthesis methodologies were based on an anodic or cathodic process. However, paired redox neutral electrosynthesis merging simultaneous anodic and cathodic processes remains challenging. Herein, we report a redox neutral Ni-catalyzed electrochemical C(sp 2 )-C(sp 3 ) cross-coupling paradigm ## Introduction In the past half-century, transition metal catalyzed carbon-carbon (C−C) cross-coupling reactions have gained significance advances regarding reaction scopes, selectivity, and catalytic mechanisms, and achieved tremendous success in organic synthesis of pharmaceutical molecules, agrochemicals, and organic materials. Catalyzed C−C cross-coupling reactions have historically been dominated by Pd-based catalysts. In addition to replacing the expensive, precious Pd metal, Ni metal is characteristic of more negative 2+/0 and 1+/0 redox potential than Pd 2+/0 to enable unique oxidative addition reactivities in activating C-X (X = Cl and Br) bonds 6 and has found increasing importance in C-C cross coupling reactions. However, Ni catalyzed C−C crosscoupling reactions are still limited by a number of well-known synthetic limitations. Ni-based Kumda, Negish, and Suzuki, and reductive couplings are practically hampered by the use of either strong nucleophiles, sacrificed reductants, or sensitive Ni 0 pre-catalysts, e.g. widely used Ni(COD)2 (where COD is 1,5-cyclooctodiene) and typically require rigid reaction conditions using inert atmosphere glovebox or Schenk-line techniques. It remains a long-standing challenge to develop Ni-catalyzed cross-coupling reactions using bench stable chemicals and easy-handling conditions for widespread academic and industrial adoption. 8 Efforts have been made to develop well-defined air stable Ni II and Ni 0 pre-catalysts and encapsulated Ni 0 pre-catalysts. 11 However, these practices are still limited by the pre-formation of Ni pre-catalysts under rigid air-free conditions and the need of special stabilization ligands for most of them. On the other side, literature has witnessed the powerful applications of electrochemistry in organic synthesis. By precisely controlling redox potential in an electrolyzer cell, substrates or catalysts can be selectively anodically or catholically activated to participate desired reaction sequences. Thereby, electrosynthesis not only migrates the use of reactive (even dangerous) oxidants and reductants and enables the access of highly reactive catalytic intermediates which are not easily handled in traditional thermal reactions, representing a green, atomically economical synthetic strategy. In spite of being known for many decades, until very recently electrosynthesis has aroused recurred attention and is believed to impart profound impacts on organic syntheis. For instance, anodic reactions including alcohol oxidation, 16 C−H functionalization, alkene functionalization, cyclization, and C−O 29 and C-N couplings, and cathodic reactions including arene or alkene hydrogenation, and arylboronic acid hydroxylation 34 were demonstrated with good selectivity and yields. Ni-catalyzed cathodic reductive C−C homocouplings were first reported by Jennings and co-workers in 1976. Ni-catalyzed reductive cross-electrophile C−C couplings was pioneered by Jutand, Perchion and coworkers and have been recently advanced by several groups, representing an attractive technology for C-C formation without using strong, reactive reductants as in traditional thermal reactions. However, Nicatalyzed redox-neutral cross couplings remain very rare, 43 in which anodic oxidation of an nucleophile and cathodic reduction of an electrophile are coupled to forge the C-C bond formation while no sacrificed stoichiometric electron donor is required. It is also worth noting that more than 97.5% of ca. 900 electrosynthesis methodologies reported between 2000 and 2017 were based on an anodic or cathodic process. 13 The development of paired redox neutral electrosynthesis has been very challenging as merging an anodic redox reaction and a cathodic redox reaction is often plagued by side reactions of reactive intermediate in each redox reaction. 13 For example, homo- ## Results and discussion Instead of randomly testing combinations of nucleophiles, electrophiles, and catalysts, we first set out to identify individual anodic and cathodic SET half-cell reactions for the proposed full-cell C−C coupling reactions using the electrochemical cyclic voltammetry (CV) method. For the cathodic half-cell reaction, we aimed to explore the SET reduction of Ni II -based catalysts to activate aryl and vinyl halide electrophiles by the Ni III/I(II/0) redox cycle to achieve R−Ni III(II) −X intermediate, which is mechanistically accessible in traditional Ni-based thermal couplings. 6 For the anodic half-reaction, nucleophiles including carboxylic acid 12 and organic trifluoroborate 44 are well documented as carbon radical precursors (Rʹ• in Figure 1) upon SET oxidation. It is noted that organic trifluoroborates have been used as versatile radical precursors for metal photoredox catalytic coupling reactions with aryl halides by Molander and coworkers. 45 Electrochemical screening of the proposed half-cell reactions was conducted through the cyclic voltammetry (CV) method using a three-electrode system. As shown in Figure 2B(i) (gray curve), in the presence of 3 equivalents 2,2'-bipyridine (2,2'-bpy) ligand, NiCl2•glyme displayed a reversible redox signal at E1/2 = -1.49 V (vs. Fc +/0 ), which corresponding to the Ni II/I redox couple. Then, 10 equivalents of organic halides (R−X) were added to the electrolyte and CV curves were collected again. Among tested organic halides, C(sp 2 ) precursors (arly halide and alkenyl bromide) or C(sp) precursors (alkynyl bromide) could be activated by the Ni I intermediate while C(sp 3 ) precursors were inactive. For example, when methyl 4-bromobenzoate was added (green trace in precursors (alkynyl bromide) and C(sp 3 ) sources (potassium benzyltrifluoroborate, phenylacetate, LG We then optimized the NiCl2•glyme/polypyridine catalyst system using cyclic voltammetry with methyl 4-bromobenzoate as a model electrophile. As shown in Figure 2C(i), seven different polypyridine ligands including 4,4'-di-tert-butyl-2,2'-bipyridyl (dtbbpy), 6,6'-dimethyl-2,2'bipyridyl (dmbpy), 2,2'-bpy, dimethyl 2,2'-bipyridine-4,4'-dicarboxylate (dmcbpy), 1,10phenanthroline (1,10-Phen), 2,2'-biquinoline (biq), and terpyridine were screened to identify the most suitable ligand for the Ni-catalyst. Among all the ligands, dtbbpy prompted the strongest current intensity increase (green curve), indicating that Ni I (dtbbpy) + is the most reactive species to oxidative addition of the C-Br bond of methyl 4-bromobenzoate. Besides 2,2'-bpy and dtbbpy, 1,10-Phen also aroused strong current response (purple curve) and thus can also be a suitable ligand. Terpyridine ligand displayed the lowest current response under the same conditions (Figure S5). We further investigated the effect of Ni/ligand ratio on the reactivity of the Ni-catalyst. The CV curves of NiCl2•glyme with addition of various ratio of dtbbpy ligand showed continuous change (Figure S5). In the absence of the dtbbpy ligand, no reversible redox signal was observed. When 1 -3 equivalents of dtbbpy ligand was added, there were two set of quasi-reversible redox signals. Further increase the ligand ratio to 5 equivalents, the redox signals overlapped to one set of fully reversible redox signal. It indicates that there is an equilibrium for Ni II complexes in the solution: Ni II ↔ Ni II (dtbbpy) ↔ Ni II (dtbbpy)2 ↔ Ni II (dtbbpy)3, which is consistent with a previous UV-Vis study. 30 In the presence of methyl 4-bromobenzoate substrate, the addition of 1. S1). It was found that the yield for 1 was further improved to 93% using K2CO3 additive. The essentiality of NiCl2•glyme catalyst, dtbbpy ligand, and electrolysis was determined by control experiments (Table S1, SI). In addition, both reaction selectivity and rate were largely affected by current intensity. Lower selectivity was obtained under a higher or lower current intensity (64% under 1.0 mA, 77% under 5.0 mA current). Under 1.0 mA current electrolysis, the reaction was significantly decelerated as a reaction time of 48 h needs to fully convert the substrate. Other solvents, such as THF, MeCN, CH2Cl2, MeOH, and DMSO were not effective to this reaction (only 0 -15% yield was observed, Table S2, SI). Moreover, similar as under thermal reaction conditions, 16 the reactivity and selectivity of this reaction is highly sensitive to the ligand structure (Table S3, SI). In particular, dtbbpy and 2,2'-bpy ligands exhibited the best efficiencies with isolated yields of 93% and 87%, respectively. 1,10-Phen and tridentate terpyridine (tpy) ligands gave moderate yields of 67% and 73%, respectively. It is noteworthy that the best selectivity between cross-coupling product 1 and homo-coupling product 1' was obtained by using the dtbbpy (95:5) and tpy (96:4) ligands, which tend to suppress the homo-coupling of strong electrophiles. However, other ligands (dmbpy, dmcbpy, and biq) were not effective. Moreover, no product was observed when a bidentate bis-phosphine ligand, 1,2-bis(diphenylphosphino)benzene (dppb) was used (Table S3). It was observed that the dppb ligand underwent oxidation near to the oxidation potential of benzyltrifluoroborate, which could destabilize the corresponding Ni catalyst (Figures S6 and S7). After establishing optimal reaction conditions for yield and selectivity for the Ni-catalyzed electrochemical C(sp 2 )-C(sp 3 ) cross-coupling, we next tested the reaction scope on both aryl halide and benzylic trifluoroborate using the most efficient Ni/dtbbpy catalyst. As shown in Figure 3, a wide range of aryl chlorides including both electron-rich and electron-deficient arenes were suitable to this Ni-catalyzed electrosynthesis system (1 to 5). The electron-deficient aryl chlorides (1 to 3, 74% to 86% yield) delivered better yield than the electron-rich ones (4 and 5, 46% and 31% yield). It is probably due to the low activity of electron-rich aryl chloride substrates with the Ni I intermediate. Aryl bromides displayed better efficiencies than the corresponding aryl chlorides, as 1 to 5 were isolated in 77% to 93% yield by using aryl bromide substrates. The reaction exhibited comparable efficiency upon scale-up, for example, 89% yield was obtained on a 2.5 mmol scale reaction of 1 (0.5 g). The substituent position of aryl bromide displayed a moderate effect to the reaction efficiency, as the para-, meta-and ortho-substituted methyl bromobenzoate delivered 93%, 71%, and 89% yield (1, 6 and 7), respectively. Aryl bromides with functional groups as diverse as ester (1, 6, 7, and 10), ketone (2), fluoride (3), methoxy group (9 and 10), amide (14 and 15), aldehyde (11), nitrile (12) and alkenyl (19) were effective in this reaction. Substrates possessing strongly electron-donating substituents such as t Bu and methoxy groups could also provide moderate to good yield (72% for 8 and 53% for 9). When 4-bromo-phenol was used as the electrophile as a control experiment for entry 9, no cross-coupling product was observed, which is attributed to the oxidation of the substrate itself at a less positive potential than the borate nucleophile. The observation emphasizes the protection of oxidization susceptible functional groups under the investigated electrochemical conditions. It is interesting that for the substrates possessing strong electron-withdrawing substituents such as aldehyde, acetyl, and cyano groups, best results were obtained by using 2,2'-bpy ligand (83% and 91% yield for 2 from chloride and bromide, respectively, 82% yield for 11, and 74% yield for 12). Furthermore, in the case of 4bromo(trifluoromethyl)benzene, homo-coupling product, 4,4'-bis-(trifluoromethyl)biphenyl (13'), was obtained as the only product when using dtbbpy and 2'2-bpy ligands. Interestingly, 21% yield of cross-coupling product 13 was obtained by using the tpy ligand, implying the Ni/tpy ligand combination is more compatible with electron deficient electrophiles to suppress homo-coupling. In addition to examine the substituent positions and functional groups of the aryl halide substrates, we also investigated the tolerance of this electrosynthesis system to common protecting groups which are widely used in organic synthesis, such as amide, tert-butyloxycarbonyl (Boc), benzyl ether (BnO), and acetal. All of these protecting groups were well tolerated, as evidenced by good isolation yield of 14 to 18 (67% to 86% yield). The π-conjugation extended aryl bromide substrates including 4-bromophenylethene, 3-bromofluorene, and 2-bromonaphthalene also smoothly proceeded this cross-coupling reaction with moderate to good yield (19 to 21, 43% to 84% yield). Moreover, a variety of aryl bromides consisting of nitrogen-containing heterocyclic groups including 6-bromoquinoline, 6-bromoisoquinoline, and Boc protected 6bromotetrahydroisoquinoline, and 5-bromoindole, which are prevalent building blocks in bioactive molecules, delivered moderate to good yield (22 to 25, 52% to 81% yield). The substrate scope of benzylic trifluoroborate salts was also investigated. As shown in Figure 3, both electron-rich and electron-deficient benzylic trifluoroborates were approved efficient carbon radial precursors in this cross-coupling reaction (26 to 31, 74% to 95% yield). Functional groups, including esters, methoxy group, and trifluoromethyl group were tolerant to this Nicatalyzed electrosynthesis. The substituent positions displayed negligible effects to the reaction efficiency, as comparable yield was obtained for the para-, meta-, and ortho-substituted benzylic trifluoroborates (26 to 28, 77% to 82% yield). In the presence of two strong electron-donating methoxy (MeO-) groups, the highest yield, 95%, was gained for 29, which is interpreted as the favorable oxidation kinetics of the corresponding trifluoroborate substrate. The π-conjugation extended naphthalen-2-ylmethyl trifluoroborate is also highly productive in this electrochemical C(sp 2 )-C(sp 3 ) cross-coupling reaction, as 72% yield was obtained for 32. Beside the benzylic trifluoroborates, (benzyloxy)methyl trifluoroborate also manifested reasonable reactivity in this reaction with a yield of 47% (33). In the CV screening studies (Figure 2B), some other substrates also showed reactivity in the anodic half-reaction. For example, β-bromostyrene and methyl 3-bromopropiolate showed reactivity in the anodic half-cell reaction (Figure S4, SI). β-bromostyrene was briefly examined as an electrophile for the Heck-type like C(sp 2 )-C(sp 3 ) cross-coupling. In the reaction of βbromostyrene and potassium trifluoro(4-(methoxycarbonyl)benzyl)borate, 48% yield of product 34 and 47% yield of the homo-coupling product 34' were obtained by using dtbbpy as ligand. According to entry 13, the tpy ligand exhibited the better selectivity to suppress the homo-coupling product. Then the coupling reaction using β-bromostyrene was optimized with the typ ligand. The improved yield and selectivity for the cross-coupling product 34 were obtained in the presence of the tpy ligand (83% yield, 90% selectivity) (Table S4, SI). As shown in Figure 3, both electronrich and electron-deficient benzylic trifluoroborates were efficient in this cross-coupling reaction (34 to 40, 63% to 92% yield). Functional groups including esters, methoxy group, and benzodioxol group were tolerant in this Ni-catalyzed electrochemical reaction. The π-conjugation extended naphthalen-2-ylmethyl trifluoroborate also provided good reactivity in this reaction, as 67% yield was obtained for 41. However, other anodic nucleophiles (3-bromopropiolate, phenylacetic acid, potassium pivalate, and potassium phenyltrifluoroborate) didn't provide satisfactory results (see Figure S9 and the SI for more discussions). To demonstrate potential applications of this Ni-catalyzed electrochemical C(sp 2 )-C(sp 3 ) cross-coupling methodology, we first exploited the synthesis of pharmaceutical molecules containing the diphenylmethane structural component. Beclobrate analog (42, a hypolipidemic candidate 46 ) and Bifemelane (43, an antidepressant candidate 47 ) were synthesized with 74% and 56% overall yield, respectively (Figure 4A and 4B). We further utilized this methodology in latestage functionalization of pharmaceuticals which is a popular way for fast discovery of new drag candidates. Fenofibrate is a pharmaceutical molecule of the fibrate class and used to treat abnormal blood lipid levels. 48 As shown in Figure 4C, Fenofibrate was successfully converted to a series of brand-new compounds (44 to 48, 41% to 86% yield) in up to 2.5 mmol (0.93 g) scale form a regular vial electrolyzer cell. Another new Clofibrate derivative (a lipid-lowering agent) was synthesized using this electrochemical approach (49, 63% yield) (Figure 4D). In addition, the electrochemical C−C cross-coupling reaction was also effective in modification of brominated natural amino acids, e.g. phenylalanine, (Figure 4E) (50, 83% yield). To further testify the potential industrial adoption of the present electrochemical cross-coupling reaction, flow cell synthesis (Figure 4F and 4G) was demonstrated with compounds 1, 29 and 48 with a reaction scale greater than 2.0 g. It should be noted reaction solutions were only flushed with nitrogen gas in the flow cell synthesis without using rigid glovebox or Schlenk-line techniques. Under the flow-cell condition, all three compounds were obtained with good to excellent yields (86% for 1 at 3.0 g scale, 92% for 29 at 2.0 g scale, and 84% for 48 at 3.0 g scale). To gain mechanistic understandings of this Ni-catalyzed electrochemical C(sp 2 )-C(sp 3 ) cross-coupling reaction, a radical-trapping experiment was conducted for the anodic half-reaction. As shown in Figure 5A, controlled potential electrolysis (at 1.2 V, vs. Fc +/0 ) of the potassium trifluoro(4-(methoxycarbonyl)benzyl)borate and 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) in a divided-cell produced radical coupling product 51 with 86% isolated yield, which confirms the formation of 4-(methoxycarbonyl)benzyl free radical in the anodic oxidation process. In addition, plots of overpotential over the logarithm of kinetic current and the corresponding fitted Tafel plots were constructed to determine charge transfer rate constants (k 0 ) of potassium benzyltrifluoroborate and phenylacetic acid in the presence of 2.5 equiv Cs2CO3 in the anodic oxidation process (Figure 5B and see the SI for detail). k 0 of potassium benzyltrifluoroborate and cesium phenylacetate were calculated as 5.56 x 10 -5 cm/s and 1.39 x 10 -5 cm/s, respectively. The higher charge transfer rate constant of potassium benzyltrifluoroborate indicates faster electrochemical reactivity to generate carbon radicals than cesium phenylacetate, which is consistent with the better efficiency of potassium benzyltrifluoroborate in the cross-coupling reaction than cesium phenylacetate (30% yield). It is believed that the quick formation of the Electrochemical studies were conducted to gain additional mechanistic insights for the cathodic process. As shown in Figure 5B, methyl 4-bromobenzoate substrate displayed irreversible redox signal with onset potential at -2.05 V (vs Fc +/0 ) and dtbbpy ligand delivered reversible redox signal with E1/2 = -2.70 V (vs. Fc +/0 ), respectively. The mixture of NiCl2•glyme and 1.5 equiv dtbbpy ligand exhibits three redox peaks at E1/2 = -1.74 V, -2.44 V, and -2.70 V (vs. Fc +/0 ), which corresponding to Ni II/I , Ni I/0 redox couples, and the free ligand. When the methyl 4-bromobenzoate substrate was added, significant increase of reductive current and disappearance of the return peak was observed for the Ni II/I redox couples. It indicates that the Ni I is the reactive species for the oxidative addition of aryl halide. In addition, CV curves of the reaction mixture displayed -1.60 V and 0.33 V (vs. Fc +/0 ) onset potentials for cathodic and anodic half-reactions, respectively (Figure S10). The potential of cathode was retained between -1.7 and -1.9 V (vs. Fc +/0 ) during the reaction (Figure S11), and the observation further indicates that the Ni I/0 redox couple is not involved in the cathodic process. Based on the chemical and electrochemical studies, a possible reaction mechanism for this Ni-catalyzed electrochemical C(sp 2 )-C(sp 3 ) cross-coupling is proposed and illustrated in Figure 6. 45 which relies on an iridium photocatalyst to activate trifluoroborates and regenerate a Ni 0 catalyst, the key mechanistic difference of the present electrochemical coupling is that both the reactive carbon radical and Ni I intermediate are generated electrochemically. Without using the expensive iridium photocatalyst and the reactive Ni 0 catalyst, the present electrochemical cross coupling is more affordable, scalable, and practical. Molander's photocatalytic cross coupling is also capable of using alkyl trifluoroborate nucleophiles as radical precurors. 49 Nevertheless, this electrochemical cross-coupling protocol is not effective to handle reactive alkyl radicals. Through optimization of reaction conditions and catalysts, it is likely to expand the scope of nucleophiles to alkyl and phenyl trifluoroborates, and even carboxylic acids for electrochemical cross coupling. ## Conclusions In summary, a Ni-catalyzed electrochemical cross-coupling methodology was developed to forge the C(sp 2 )-C(sp 3 ) bond with broad substrate scope, excellent functional group tolerance, selectivity, and good yields. In addition, the cyclic voltammetry proved an effective and efficient way for the discovery, optimization, and mechanistic understanding of anodic and cathodic halfreactions and can be used as a go-to method for developing other useful electrosynthesis methodologies. Compared to traditional thermal Ni catalyzed cross-coupling reactions, the present electrochemical approach is advantageous as all reactants and catalysts are bench stable without using reactive oxidants/reductants and complex inert atmosphere techniques. As exemplified in gram-scale synthesis in the flow-cell synthesis and the late-stage functionalization of pharmaceuticals, this electrochemical C−C coupling methodology is expected to be widely applied to the construction of C(sp 2 )-C(sp 3 ) bonds in developing pharmaceutical molecules, agrochemicals, and organic materials. The Ni-catalyzed electrochemical C-C cross coupling reactions can be further advanced for broader substrates and extended to other types of coupling reactions. Moreover, the present new C-C bond formation paradigm (and also extended reactions) can offer rich opportunities to pursue fundamental mechanistic studies and thus lead to the discovery of new catalytic knowledge at the interface of synthetic chemistry and electrochemistry.

chemsum

{"title": "Nickel Catalyzed Electrochemical C(sp 2 )\u2212C(sp 3 ) Cross-Coupling Reactions", "journal": "ChemRxiv"}

observations_of_tetrel_bonding_between_sp³-carbon_and_thf

## Abstract: We report the direct observation of tetrel bonding interactions between sp 3 -carbons of the supramolecular synthon 3,3-dimethyl-tetracyanocyclopropane (1) and tetrahydrofuran in the gas and crystalline phase. The intermolecular contact is established via s-holes and is driven mainly by electrostatic forces. The complex manifests distinct binding geometries when captured in the crystalline phase and in the gas phase. We elucidate these binding trends using complementary gas phase quantum chemical calculations and find a total binding energy of À11.2 kcal mol À1 for the adduct. Our observations pave the way for novel strategies to engineer sp 3 -C centred non-covalent bonding schemes for supramolecular chemistry. ## Introduction Non-covalent interactions are key forces that drive phenomena such as host-guest chemistry, molecular aggregation, crystallization and protein folding. 1,2 In recent years, important intermolecular interactions like hydrogen and halogen bonding 1,3-7 have been contextualized as s-hole interactions. A s-hole can be seen as a Lewis acidic site along the vector of a covalent bond, the location of which coincides with the s* orbital of that bond. The extreme outcome of a s-hole interaction can be the breaking and/or making of a s bond, such as in the formation of I 3 from molecular I 2 and I . 11,12 A similar rationale can be applied to so-called 'p-hole interactions' involving electron-defcient aromatic rings, 13,14 or polarized double bonds with related covalent bond-forming chemistry such as in aldol-type reactions. In principle, sand p-hole interactions should be available with all the non-metallic elements of the periodic table. This includes carbon; an element of central importance to life and ubiquitous presence in synthetic chemistry. One might thus wonder to what extend carbon can be exploited as locus of Lewis acidity to establish 'tetrel-bonding interactions' (in analogy to halogen-and chalcogen-bonds). 18 Such interactions are well-known for sp 2 -hybridized C-atoms in carbonyls and have recently been reported for the sphybridized C-atoms of (coordinated) acetonitrile, 27 carbon monoxide 28 and carbon dioxide. Non-covalent interactions with sp 3 -hybridized carbon atoms are implicated in the advent of canonical S N 2 nucleophilic displacement reactions 12, and can persist with methyl groups in crystal structures. However, a supramolecular synthon to predictably generate directional tetrel-bonding interactions centred on sp 3 -C has not yet been experimentally disclosed. We envisaged that 1,1,2,2tetracyanocyclopropane (TCCP) derivatives could fulfl this role. 39,40 These rings are synthetically viable and contain a sterically accessible electrophilic site located roughly on the two sp 3 C-atoms in the (NC) 2 C-C(CN) 2 fragment. This is exem-plifed by the molecular electrostatic potential (MEP) map of 3,3-dimethyl-TCCP (1) shown in Fig. 1. The calculated s-hole potential of +44 kcal mol 1 lies in-between the s-holes of water (+55 kcal mol 1 ) and ammonia (+35 kcal mol 1 ), which are prototypical s-hole (i.e. hydrogen bond) donors. The Lewis acidic site of 1 should thus be able to form a tetrel bonding interaction with an electron-rich partner such as the lone pair electron cloud on tetrahydrofuran (THF, estimated at 40 kcal mol 1 ). Here we report on the verifcation of this hypothesis by synthesizing 1 and showing thatas anticipated -1 binds to THF via intermolecular sp 3 -C/O interactions, both in the crystalline state and in the gas phase. ## Results and discussion Cyclopropane 1 was readily prepared in a one-pot cascade reaction from acetone, malononitrile and molecular bromine (Scheme 1). Presumably, cyclization to 1 proceeds from an intermediate formed by the nucleophilic attack of in situ generated [BrC(CN) 2 ] on the Knoevenagel condensation product of acetone and malononitrile. 42 The yield of our procedure (83%) is higher than obtained by previously reported methods (max. 72%). 47 All literature procedures with a yield in excess of 50% 42,43, (maximum 72%) 47 use a two-step approach starting from an activated malononitrile derivative 42,43, and/or use electrochemical synthesis. 47,48 Single crystals suitable for X-ray diffraction measurements (see ESI † for details) were obtained by slow evaporation of a solution of 1 in THF. The molecular model of [1/THF] resulting from the diffraction study is shown in Fig. 2a. All the intramolecular distances and angles within this structure can be considered as normal (not shown). 49 The plane running through the O-and C-atoms of the THF molecule is roughly coplanar with the cyclopropane ring plane in 1 (: plane-plane ¼ 8.2 ). Interestingly, the oxygen atom of the THF molecule is directed towards C1/C3/C4 of the cyclopropane ring in 1, with very short intermolecular distances, in particular sp 3 -C1/O1 of 3.007 (C3/C4 /O1 ¼ 3.1 , not shown). This is 0.213 within the van der Waals radii of O (1.52 ) and C (1.70 ) and thus consistent with a bonding interaction. 12,27,39,40,50 Further stacking of [1/THF] in the crystal is aided by weak N1/N2/ C1/C3/C4 interactions (max. 0.067 van der Waals overlap, see Fig. S3 †). ‡ A DFT optimization at the B3LYP 51,52 -D3(BJ) 53 /def2-TZVP 54,55 level of theory of the atomic coordinates found in the crystal structure converged at a nearly identical structure (see Fig. S5 †). The interaction energy (DE) was computed to be 10.1 kcal mol 1 . This is much larger than interactions of dimethyl ether halogen bonded to I-C 6 F 5 (5.6 kcal mol 1 ) or hydrogen bonded to water (6.7 kcal mol 1 ) at this same level of theory. 37,56 Interestingly, the [1/THF] structure shown in Fig. 2b was found to be 1.1 kcal mol 1 more stable, representing the true energetic minimum with DE ¼ 11.2 kcal mol 1 (see also Fig. S6 †). The structure is similar to the crystal structure but with the THF oriented almost perpendicular to the cyclopropane plane, with : plane-plane ¼ 83.8 . The distances between the THF-O and the two sp 3 (NC) 2 C-C(CN) 2 atoms display up to 0.297 van der Waals overlap, which is 0.084 more than observed in the crystal structure. This difference likely originates from the lack of any other interactions in the idealized gas phase computation versus various other potential weak interactions within the crystal of [1/THF]. Rotational spectroscopy is the technique to experimentally discriminate between the two relative orientations of [1/THF] (Fig. 2) that are so close in energy in the gas phase calculations (1.1 kcal mol 1 ). Thus, we conducted chirped pulse Fourier transform microwave (CP-FTMW) spectroscopy 57,58 to assign the geometry of [1/THF] in the gas phase (see ESI † for 1 are spectroscopic parameters extracted from this experiment together with predicted values based on DFT calculations of [1/THF] with 'coplanar' or 'perpendicular' THF orientations. The experimental rotational constants (in particular B and C) provide a conclusive assignment of the [1/THF] complex in the 'perpendicular' orientation, which is also the DFT-energetic minimum (righthand side of Fig. 2). ## details). Shown in Table To date we were unable to quantify tetrel bonding interactions with 1 in solution, but we did observe a very large and unusual solvent dependency for the 1 H and 13 C NMR resonances of 1 (detailed in Fig. S7 and Table S4 †). For example, the methyl protons of 1, which are g to CN, span a range of 1.39 ppm passing from benzene through toluene, acetonitrile, methanol and chloroform, to acetone. In comparison, the ethoxy methyl protons in ethyl acetate and diethyl ether vary by just 0.34 and 0.12 ppm, despite being closer to functional groups. 64 These results seem to suggest strong and geometrically specifc interactions between 1 and most solvent molecules. Based on these preliminary observations, we anticipate that future studies will demonstrate that tetrel bonding interactions with tetracyanocyclopropane derivatives also persist in solution. To gain more insight into the physical origins of the [1/ THF] adduct, the 'perpendicular' structure was subjected to a Morokuma-Ziegler inspired energy decomposition, 37, an 'atoms-in-molecules', 62 and a non-covalent interaction analysis. 63 The energy decomposition analysis revealed that the interaction is mainly electrostatic in origin (52.7%) followed by dispersion (30.7%) and orbital interactions (16.8%). Interestingly, orbital mixing occurred between the HOMO of THF and the LUMO of 1 (3.86 kcal mol 1 stabilization) and between the HOMO1 of THF and the HOMO of 1 (4.80 kcal mol 1 stabilization, see Fig. S8 † for details). The 'atoms-in-molecules' analysis of [1/THF] shown in Fig. 3a reveals several bond critical points (bcp's) between the N-atoms of 1 and several CH hydrogens of THF, indicating very weak hydrogen bonding interactions (r z 0.005 a.u.). The densest bcp of r ¼ 0.0115 a.u. is present between the THF O-atom and one of the sp 3 (NC) 2 C-C(CN) 2 atoms (highlighted in yellow). § In line with these results, the NCI plot shown in Fig. 3b clearly reveals that there are two sp 3 -C/O interactions that are mainly electrostatic in origin (blue), and that the C-H/N interactions are mainly dispersive (yellow/green). For comparison purposes, a cyclopentane adduct was calculated after in silico O / CH 2 mutation and geometry optimization of structure [1/THF]. This resulted in the structurally similar [1$$$cyclopentane] adduct shown in Fig. 3c (see ). The adduct is also much less stable with DE ¼ 4.0 kcal mol 1 , which is mainly driven by dispersion (59.4%) followed by electrostatic interactions (22.2%). The NCI analysis of this adduct depicted in Fig. 3d clearly shows that this adduct is only held by dispersive C-H/N interactions (green). ## Summary and concluding remarks In summary, it was shown that 1 can form [1/THF] complexes in the crystalline state and in the gas phase with a calculated interaction energy of up to 11.2 kcal mol 1 . These complexes are held together by strong polar interactions between the de facto Lewis acidic site in between the sp 3 -hybridized C-atoms of 1 and the Lewis basic THF-O. These results demonstrate that tetrel-bonding interactions with sp 3 -carbon centres can indeed be used to engineer supramolecular complexes, thus paving the way for their exploration in other molecular disciplines, e.g. supramolecular chemistry, crystal engineering and medicine. ## Conflicts of interest The authors declare no conflict of interest.

chemsum

{"title": "Observations of tetrel bonding between sp³-carbon and THF", "journal": "Royal Society of Chemistry (RSC)"}

modeling_the_influence_of_correlated_molecular_disorder_on_the_dynamics_of_excitons_in_organic_molec

## Abstract: In this Letter, we investigate the role of correlated molecular disorder on the dynamics of excitons in oligothiophene-based organic semiconductors. We simulate exciton dynamics using the Frenkel exciton model and we derive parameters for this model so that they reflect the specific characteristics of all-atom molecular systems. By systematically modifying the parameters of the Frenkel exciton model we isolate the influence of spatial and temporal molecular correlations on the dynamics of excitons in these systems. We find that the molecular fluctuations inherent to these systems exhibit long-lived memory effects, but that these effects do not significantly influence the dynamic properties of excitons. We also find that excitons can be sensitive to the molecular-scale spatial correlations, and that this sensitivity grows with the amount of energetic disorder within the material. We conclude that control over spatial correlations can mitigate the negative influence of disorder on exciton transport. The optoelectronic properties of organic molecular semiconductors are known to depend sensitively on how molecules are arranged within the material. The inability to exploit this dependence for improving material performance is a problem that hinders the development of electronic applications that incorporate these materials, such as organic photovoltaic (OPV) and light-emitting (OLED) devices. This problem originates, in part, from a lack of theoretical methods that can reliably predict the excited-state electronic properties of materials with disordered or irregular microscopic structure. Many efforts to address this problem have thus been focused on developing ways to accurately include the effects of disorder in traditional theoretical models. 8, In this letter, we extend these efforts with a theoretical approach designed to reveal the specific effects of correlated molecular disorder on electronic energy transport in organic molecular semiconductors. Our approach combines classical molecular dynamics (MD) simulations, semiempirical electronic structure calculations, and the Frenkel exciton model, in order to identify the characteristics of molecular correlations in these materials and isolate their influence on the microscopic dynamics of electronic excitations. Using this approach, we find that energy transport in these materials can be very sensitive to the presence of spatial correlations, and that this sensitivity depends on the width of the distribution of excitation energies. We illustrate that fluctuations in these energies play a fundamental role in driving the dynamics of electronic excitations, but we also highlight that these dynamics are insensitive to the temporal correlations that arise due to regular nuclear vibrations. These findings suggest that it may be necessary to consider the interplay between correlated and uncorrelated disorder when applying molecular design principles to the development of organic molecular semiconductors. Energy transport in organic conjugated systems is determined primarily by the microscopic dynamics of Coulombically bound excited electron-hole pairs, known as excitons. In condensed phase systems, intermolecular electronic coupling can drive excitons to delocalize across many individual molecules. The properties of these delocalized excitons depend on the strengths of electronic couplings and on how they are distributed in space and time. The characteristics of a material's molecular structure that are most relevant to the dynamics of excitons are thus encoded in these electronic coupling distributions. Unfortunately, these distributions are difficult to compute because they require evaluation of the exited state electronic structure and because they tend to vary significantly with small changes in nuclear configuration and therefore must be computed separately for each pair of molecules in the system. The computation of electronic properties is not the only challenge associated with modeling exciton dynamics in disordered systems. Accurate modeling also requires the ability to address system sizes large enough to fully accommodate delocalized excitons, and a sampling scheme capable of capturing the effects that arise due to local variations in molecular structure. If excitons delocalize over more than a few molecules, then standard electronic structure methods, such as those based on density functional theory, are generally too computationally expensive to fully meet these requirements. As an alternative, site-based phenomenological models provide a simple and efficient platform for computing the static and dynamic properties of delocalized excitons in extended heterogeneous systems. Here, we simulate the properties of delocalized excitons using the Frenkel exciton model 21,22 in which the excited electronic properties of a N-molecule system are expressed in terms of the Frenkel Hamiltonian, where |i represents a state with an exciton localized on molecule i, i denotes the energy of that state, and V ij denotes the intermolecular electronic coupling between states |i and |j . Because this simple model lacks explicit chemical detail, all aspects of a system's molecular structure must be described implicitly, in terms of model parameters. For this reason, the effects of molecular disorder are often incorporated into this model by dressing static model parameters with additional random components. The characteristics of these random components are often assumed to be Gaussian distributed, spatially uncorrelated, and described by simple dynamics that are either temporally uncorrelated (e.g., white noise) or arising from weak coupling to a bath of harmonic oscillators. 18, Although these assumptions introduce disorder in a well defined and easily controllable manner, it is not obvious to what extent the molecular structure that they imply is physically realistic. In our approach, the parameters of the Frenkel Hamiltonian are assigned non-randomly based on the microscopic structure of configurations generated with all-atom MD simulation (see supplementary information for more details). We consider the effect of thermal fluctuations on the statistics of these model parameters, and we quantify the spatial and temporal correlations that are associated with these statistics. By determining the separate influences of these correlations on the dynamics of excitons we can critically assess the assumptions that are commonly applied when parameterizing the Frenkel exciton model for organic molecular semiconductors. We present results for room temperature (T = 300K) condensed phase systems that are made up entirely of assembled sexithiophene (T6) molecules. We choose T6-based materials specifically because they have been well studied both experimentally and theoretically. We consider systems with two different characteristic morphologies: a monolayer film of 150 longitudinally aligned molecules and an amorphous bulk of 343 T6 In Fig. 1 we illustrate how differences in molecular structure influence the characteristics of site energetic disorder. Fig. 1(b) shows the spatial distribution of site energies (i.e., i in Eq. 1) as derived by applying our parameterization method to a single MD configuration of the monolayer film. This plot illustrates that the monolayer film includes significant spatial disorder, because the molecules are not arranged on a regular lattice, and significant energetic disorder, because molecules exhibit a wide range of excitation energies. Fig. 1(c) contains a plot of P ( ), the probability for a molecule in a given system to have a site energy . We observe that P ( ) is broad and asymmetric for both the monolayer and the amorphous system, which have a standard deviations of 0.10eV and 0.18eV, respectively. The distribution of the monolayer system is both red-shifted and narrowed relative to that of the amorphous system. These differences reflect the influence of favorable π−stacking interactions between neighboring molecules in the film, which simultaneously limit configurational variations and promotes conjugation through molecular planarization. Molecular packing effects in condensed phase systems can lead to the emergence of spatial correlations in molecular and electronic structure. These correlations are often neglected when assigning parameters in the Frenkel exciton model. Our approach to assigning these parameters preserves these correlations, which enables them to be quantified. We define the spatial correlation function for site energies as, where • • • denotes an average over all available configurations of a given system, ¯ is the average site energy for molecules in the system, r ij is the center-of-mass separation between the molecules associated with sites i and j, and δ(x) is the Dirac delta function. As this plot illustrates, closely spaced molecules tend to have correlated excitation energies. We observe that these correlations die off rapidly with distance, not extending much beyond distances of about 0.5nm (i.e., roughly the nearest neighbor distance), but are slightly longer ranged in the monolayer system. Despite the lack of long-range spatial correlations, it has been found that excitons readily delocalize in both of these systems. 20 The time dependence of the model parameters in Eq. 1 are generated to reflect the evolution of a given system from MD simulation. A representative trace of (t) for a single site of the model monolayer film, as plotted in the inset of Fig. 2(a), reveals that the excitation energy of individual molecules can exhibit significant fluctuations. The temporal correlations in these model parameters can be characterized by computing the time correlation function for site energies, and for intermolecular electronic couplings, where the averages implied by • • • include data generated for a system with a given molecular structure and the overbars indicate time average. As illustrated in Fig. 2, both C (t) and C V (t) exhibit non-trivial forms, including distinct short and long time decay profiles along with a remarkably long lived oscillatory feature. For both correlation functions we attribute the initial fast decay (i.e., τ ∼ 100fs) to the dephasing effect from nuclear ballistic motion, and the slower decay (i.e., τ ∼ 500fs) to ring-ring torsional dynamics. We attribute the long lived oscillations, with periods of approximately 20fs and 100fs, to the C-H and C-C bond stretching vibrations, respectively. These correlations are significantly more complicated than what is usually assumed in applications of the Frenkel model to organic molecular semiconductors. 18,23,24,26 Some previous efforts to develop more realistic descriptions of the phonon spectral density have combined MD simulation with excited state electronic structure calculation, however, due to computational cost these efforts have been limited to relatively small system sizes and short time scales. To investigate how the molecular correlations in these materials influence the dynamics of excitons we carry out simulations using the time-dependent Hamiltonian in Eq. 1. Specifically, we solve the time-dependent Schrodinger equation to obtain the exciton wavefunction, where T is the time-ordering operator and c i (t) is the wavefunction coefficient in the molecular site basis. We then isolate the influence of specific correlations by modifying the Hamiltonian, replacing correlated parameters with uncorrelated random noise. By using the time dependent Frenkel Hamiltonian derived from MD simulations we are able to captures the effect of nuclear motion on the properties of the exciton, however, this approach to dynamics omits the feedback of the exciton on the dynamics of the nuclei. Omitting this feedback yields a significant gain in computational efficiency, however, the resulting electronic dynamics are thus prevented from properly thermalizing. Since the transient effects associated with exciton thermalization are most pronounced on timescales that are longer than we simulate here, 35,36 we expect the errors associated with the omission of this feedback to be small and have no influence on the nature of our conclusions. This issue is discussed more thoroughly in the Supporting Information. We quantify the dynamic properties of excitons in terms of their mean-squared displacements (MSD), which we compute from the solution to Eq. 5 using the formula, where r i denotes the center-of-mass position of the molecule associated with site i. We compute the MSD by averaging over a nonequilibrium ensemble of trajectories that are each initialized with the exciton localized on a single site. Similarly, the delocalization of excitons can be quantified in terms of the inverse participation ratio, as presented in the Supporting Information. amorphous bulk under different model conditions. In both panels the fully mapped reference condition is represented with a solid blue line, the static condition is represented with a solid red line, the condition with no spatial correlations is represented by a solid yellow line, and the condition with no temporal correlations is represented by a dashed magenta line. The black dashed lines denote MSD between a localized and fully delocalized exciton. We compare the exciton dynamics generated under four different sets of model conditions. The first condition is the reference condition, in which the parameters of H(t) are mapped directly from the results of MD simulations following the method in Ref. 20. The second condition is a static condition, designed to evaluate the effect of molecular fluctuations on the dynamics of excitons. For the static condition all model parameters are time independent, mapped from a single randomly drawn configuration from the MD simulations. The third and forth set of model conditions modify those of the reference condition by eliminating either spatial or temporal correlations in the model parameters, respectively. Under the third set of model conditions, spatial correlations are eliminated by assigning i (0) for each site randomly from the distributions in Fig. 1(c), but the time-dependent component δ i (t) = i (t) − i (0) for each site remains identical to that of the fully mapped reference Hamiltonian. With this set of conditions V ij (t) is left unmodified from the reference Hamiltonian. Under the forth set of model conditions, temporal correlations are eliminated from the time-dependent components δ i (t) and δV ij (t) = V ij (t) − V ij (0) by assigning both as Gaussian white noise with the same variance as the reference Hamiltonian. The MSD computed for each of the four conditions are plotted in Fig. 3. We find that under the reference condition, excitons in both the monolayer and the amorphous systems exhibit similar dynamics, with ∼ 100fs of rapid diffusion followed by a plateauing as excitons reach the system boundaries. We observe that for the static conditions the initial exciton dynamics are similar but they taper off early (i.e., t ≥ 50fs) due to the onset of Andersen localization. 37 These results thus indicate that while molecular fluctuations are central to sustaining the excitons dynamics in these systems, their local mobility is primarily determined by the distribution of electronic couplings. Fig. 3 also illustrates that exciton dynamics in these materials are not sensitive to the non-Markovian time correlations exhibited by C (t) and C V (t) in Fig. 2. Specifically, the model conditions without time correlations reveal that the MSD for excitons is essentially unaffected when the dynamics of the fully mapped reference Hamiltonian are replaced with Gaussian white noise. Notably, these results are consistent with a similar finding that non-Markovian effect in biological light harvesting systems do not play a significant role in the exciton dynamics at room temperature. 38 To expand upon this point, we compare the statistics of exciton displacements across many individual trajectories. In Fig. 4 we plot the probability distribution, P (D 2 ), where is the squared displacement of an exciton computed from a single trajectory. This figure illustrates that the statistics of exciton displacements are similar for both systems under model conditions with or without time correlations. The MSD for excitons generated under model conditions without spatial correlations, as plotted in Fig. 3, reveals that the influence of spatial correlations on exciton dynamics is system dependent. Removing spatial variations from the reference model has minimal effect in the monolayer film, however, in the amorphous system the absence of spatial correlations results in a reduction of exciton diffusivity. Because the spatial correlations for each system are very similar (see Fig. 1(d)), the origin of this system dependence must involve a different aspect of molecular structure. Based on the difference in P ( ) between the two systems (e.g., Fig. 1)c, we hypothesize that the sensitivity of exciton dynamics to spatial correlations is mediated by the amplitude of energetic disorder in the system. We test this hypothesis by artificially amplifying the disorder in the i 's in the model amorphous film with and without the inclusion of spatial correlations. As illustrated in Fig. 5, if the amplitude of disorder in this system is systematically increased, then the overall exciton diffusivity decreases (as expected) and the difference between the MSDs with and without spatial correlations grows. In other words, the effect of spatial correlations on exciton dynamics grows more pronounced as the microscopic structure of a system is made more disordered. As Fig. 5 summarizes, the importance of spatial correlations on exciton dynamics depends on the amount of energetic disorder in the system. This observation implies that current rule-of-thumb design criteria may be insufficient for guiding ongoing material development efforts. In addition, this implies that control over spatial correlations can possibly mitigate the negative influence of disorder on exciton transport properties. We have also found that exciton dynamics in these systems are insensitive to the details of temporal correlations, however, this insensitivity may not necessarily persist in all organic molecular semiconductors. Simple phenomenological models, such as we have utilized here, provide a convenient and efficient framework for exploring the interplay between molecular structure and opto- electronic material properties. The continued development and application of these modeling approaches is thus important for advancing our understanding of these systems. A useful measure to characterize the extent of delocalization of exciton wavefunction is the inverse participation ratio (IPR) to the finite-size effect. The final value of IPR for the amorphous bulk is higher because of its larger system size compared to the monolayer film. We also explore IPRs generated under three other different conditions in Fig. 1: calculations with static Hamiltonian (solid red lines), calculations with no spatial correlation in the static disorder (solid orange lines), and calculations with no time correlation in the timedependent fluctuations (dashed magenta lines). In general, the qualitative behaviors under each condition are very similar to those found in the analysis of MSDs in the main text. IPRs generated from the same Hamiltonian but without the time-dependent components (i.e. δV ij (t) = 0 and δ i (t) = 0) results in a significantly reduced IPR due to the Anderson localization effect. Fig. 1 also shows that temporal correlation does not play an important role in exciton transport as the IPR generated from the Hamiltonian without temporal correlation (dashed magenta lines) is nearly identical to the IPR generated from the molecular simulations (solid blue lines). The distributions of IPRs at t = 100f s sampled from 100 initial configurations are plotted in FIG. 2, which again shows that the IPR distributions generated from molecular simulations and from Hamiltonian with white noise to be similar, confirming the minimal effect of temporal correlation. Similar to the findings in the main text, the effect of spatial correlation on IPRs is system dependent: it has little effect on the monolayer film, but reduces the IPR in the disordered film (dashed lines in FIG. 1). FIG. 3 shows the IPR in disordered film with its static disorder artificially increased, and our results demonstrate that the effect of spatial correlation on IPR increases with the magnitude of static disorder. ## Parametrization of Frenkel Hamiltonian We follow a recently developed method for mapping the structure of a N -molecule configuration onto the parameters of a corresponding N × N Frenkel Hamiltonian matrix. In this method, molecular configurations are generated using classical molecular dynamics (MD) simulations and each individual configuration is translated into Frenkel model parameters based on the analysis of N single-molecule excited state electronic structure calculations. Specifically, for a given configuration we perform a single electronic structure calculation on each individual molecule, treating all other molecules as an effective dielectric medium. Electronic structure is computed using a semiempirical Pariser-Parr-Pople (PPP) Hamiltonian with excited state properties computed at the level of configuration interactions singles. Intermolecular couplings are evaluated by computing the diabatic coupling between the locally excited molecule pairs through transition densities. A complete description of this method, including information about classical force fields, electronic structure methods, and benchmarking against higher level theories can be found in Ref. . ## Exciton Dynamics Without Excited State Forces We have chosen to simulate exciton dynamics with a method that omits the effects of excited state nuclear forces. In our method, the dynamics of the classical subsystem evolve on the potential energy surface of the electronic ground state, and are thus unaffected by the state of the exciton. In the absence of feedback between the electronic and nuclear degrees of freedom,the composite system cannot properly thermalize. This simply means that excitons will not relax into an equilibrium energic state, but rather will exhibit behavior associated with the high-temperature limit. Including the effects of excited state forces in our model is straightforward yet very computationally expensive so we omit them. We justify this omission by recognizing that this approximate method for treating dynamics is accurate in the short time limit. The timesacle for energetic relaxation of excitons is expected to be governed by two timescale. First, is the timescale associated with the molecule reogranization time, i.e., the time for excited molecules to relax on the excited state potential energy surface. This timescale is on the order of 100fs, but for delocalized excitons involves relatively small changes in excitation energy. Second, is the timescale associated with exciton migration on a disordered energetic landsacape. This timescale is determined by the mobility of excitons and the characteristics of the hetereogeneous energetic landscape. In these materials we expect the timescale to be on the order of picoseconds. Ishizaki, A.; Fleming, G. R. New J. Phys. 2010, 12, 055004. Shi, L.; Willard, A. P. J. Chem. Phys. 2018, 149, 094110.

chemsum

{"title": "Modeling the Influence of Correlated Molecular Disorder on the Dynamics of Excitons in Organic Molecular Semiconductors", "journal": "ChemRxiv"}

uranyl_speciation_in_the_presence_of_specific_ion_gradients_at_the_electrolyte/organic_interface

## Abstract: Uranyl (UO 2+2 ) speciation at the liquid/liquid interface is an essential aspect of the mechanism that underlies its extraction as part of spent nuclear fuel reprocessing schemes and environmental remediation of contaminated legacy waste sites. Of particular importance is a detailed perspective of how changing ion concentrations at the liquid interface alter the distribution of hydrated uranyl ion and its interactions with complexing electrolyte counterions relative to the bulk aqueous solution. In this work, classical molecular dynamics simulations have examined uranyl in bulk LiNO 3(aq) and in the presence of a hexane interface. UO 2+ 2 is observed to have both direct coordination with NO − 3 and outer-sphere interactions via solvent-separated ion-pairing (SSIP), whereas the interaction of Li + with NO − 3 (if it occurs) is predominantly as a contact ion-pair (CIP). The variability of uranyl interactions with nitrate is hypothesized to prevent dehydration of uranyl at the interface, and as such the cation concentration is unperturbed in the interfacial region. However, Li + loses waters of solvation when it is present in the interfacial region, an unfavorable process that causes a Li + depletion region. Although significant perturbations to ion-ion interactions, solvation, and solvation dynamics are observed in the interfacial region, importantly, this does not change the association constants of uranyl with nitrate. Thus, the experimental association constants, in combination with knowledge of the interfacial ion concentrations, can be used to predict the distribution of interfacial uranyl nitrate complexes. The enhanced concentration of uranyl dinitrate at the interface, caused by excess adsorbed NO − 3 , is highly relevant to extractant ligand design principles as such nitrate complexes are the reactants in ligand complexation and extraction events. ## Introduction Chemical separation and purification of uranium, notably from aqueous solutions, is essential to various environmental and industrial applications. 4 The highly stable U(VI) exists in the dioxo form, UO 2+ 2 , and can exhibit complicated speciation via complexation by solute anions including nitrate or carbonate. Nitric acid solutions are the most relevant to uranyl separations within the nuclear fuel cycle. Within solvent extraction processes that include Plutonium Uranium Reduction EXtraction (PUREX) 4,5 and Group ActiNide EXtraction (GANEX), 2 hydrated UO 2 (H 2 O) 2+ n and uranyl nitrate complexes 6,7 are the reacting species with extracting ligands. The complexation reaction is presumed to occur at the aqueous/organic phase boundary, and thus the speciation of the metal ions at the interface is of significant importance. 8,9 Although it is well-known that there may exist significant concentration gradients of solutes near the liquid/liquid interface, how this perturbs the speciation of metal ions and their complexes relative to the bulk aqueous phase has not been the topic of significant study. An additional complication is that the heterogeneous environment of the liquid/liquid interface 14,15 may lead to a broad ensemble of local chemical environments that have the potential to shift energetic preferences. In bulk nitric acid it is well-known that UO 2+ 2 is on average pentacoordinate and associates with nitrate anions to form UO 2 (H 2 O) m (NO 3 ) n (2 -n)+ where n + m = 5. 16 However the association is weak, as has been measured by a number of experimental methods (UV-Vis, 17,18 IR/ Raman 19 NMR, 20,21 EXAFS and microcalorimetry 25 ). In the case of the mononitrate complex (n = 1) the reported association constant (K 1 ) varies from 0.05 -0.70, while the second association constant (K 2 ) for the formation of UO 2 (NO 3 ) 2 is generally agreed to be much lower at 0.02 -0.05 at 298 K, depending upon the experimental methodology. Density functional theory (DFT) studies have examined uranyl coordination and nitrate binding modes, and identified bidentate (η 2 ) nitrate to be significantly more stable than monodentate (η 1 ) in the gas phase. In contrast, the free energy difference between η 2 and η 1 in solution is predicted to significantly decrease, such that an approximate equal population of both coordination modes should be observed in the aqueous phase. 19,26,27 Despite the efficacy of DFT studies of isolated uranyl nitrate complexes, 21,27,29,30 such methods are not able to study the speciation, solvation, and complex ion-ion interactions that occur in bulk electrolytes near industrially relevant conditions, let alone their interfaces. Molecular dynamics simulations have emerged as a powerful tool to study multi-component solutions and their interfaces, providing a molecular scale understanding of the complex correlations amongst local solution environments and dynamic equilibria between different chemical species. Yet the applicability of MD simulation is constrained by the fidelity of the potentials that define intra-and inter-molecular interactions. Several non-polarizable pairwise additive potentials have been developed for uranyl cation, and parameterized for aqueous and nitrate containing solutions under dilute conditions. Unfortunately, as we demonstrate, at appreciable NO − 3 concentrations those models significantly over-predict the degree of uranyl nitrate association and lead to long-range correlations of uranyl nitrate complexes at modest ionic strength. This work begins by optimizing the interaction terms between UO 2+ 2 and NO − 3 using a electrostatic continuum-type correction (ECC). The optimized force fields reproduce the experimentally-determined uranyl nitrate association constants and associated speciation over a wide range of uranyl and nitrate concentrations within bulk LiNO 3(aq) . The electrolyte/hexane interfacial region is then examined at high ionic strength, where significant perturbations to ion hydration across solvation shells, ionion interactions, and the heterogeneity of the interface, all have the potential to alter the association constants of UO 2+ 2 and NO − 3 relative to the bulk. Changes to the association constants would significantly complicate prediction of uranyl speciation, and thus reactivity, at the interface. Within the interfacial region, MD simulations predict ion-specific interfacial adsorption that leads to the formation of weak ion double layering, and generates distinct ion concentration gradients approaching the interface. The changing hydrogen bond network and ion gradients significantly influence the water dynamics and organization, while having modest impact upon nitrate fluctuations in the primary coordination sphere of uranyl. Interestingly, a large affect of electrolyte concentration lies within the the timescales associated with species in the first coordination sphere of UO 2+ 2 , as well as its residence within the interfacial region. The timescales of solvent exchange and the residence time of UO ## Simulation Configurations and Protocols All atom molecular dynamics simulations were performed using the GROMACS 2016.2 software package 34 to study uranyl nitrate speciation in bulk LiNO 3(aq) and LiNO 3(aq) /hexane under varying electrolyte concentrations. Initial system configurations were generated using Packmol, 35 Bulk simulations were performed with a series of concentrations presented in Table 1. These include a 0.05 M and 0.25 M UO 2+ 2 with background electrolyte LiNO 3 from 1 to 5 M so as to compare experimental studies by Suleimenov et al. 36 of uranyl in nitric acid solutions. To compare with prior data reported by Ye et. al. 37 , additional simulations of 0.25 M UO 2+ 2 with HNO 3 were also performed. All electrolyte/hexane simulations were performed at 0.25 M UO 2+ 2 and 0.5 M NO − 3 as a function of LiNO 3 concentration from 1 -5 M (Table 1). Non-bonded interactions were modeled using Lennard-Jones and coulombic interactions. Lorentz-Berthelot mixing rules were used for obtaining combinations of σ and parameters. The UO 2+ 2 and Li + ions were modeled using Wipff et al. 38,39 and Joung et al. 40 force fields respectively, while the NO 3 force fields are derived from from Ye et al. 41 and Benay and Wipff 32 . The UO 2+ 2 force field reproduces the experimentally observed hydration of 5 in the first solvation shell in bulk water, 8,38,42 whereas the Li + potential was parameterized to reproduce the experimental hydration free energies and ion hydration in the aqueous phase. 40,43 The all-atom General Amber Force Field (GAFF) 44 were implemented for n-hexane, with modified Lennard-Jones parameters to reproduce the experimental density and enthalpy of vaporization as developed by Vo et al. The UO 2+ 2 , Li + , and NO − 3 atom charges were then scaled from 100% to 75% in 5% increments using ECC, 48 which is an indirect correction to account for solvent driven polarization effects on hydrated ions. 49 In this manner, the coulombic interaction were tuned to reproduce the experimentally determined equilibrium constants 18,36 for different uranyl nitrate species and ensure coordination numbers and nitrate denticity that are consistent with experimental studies and prior ab-initio studies. 26,28 As described in the Results and Discussion, the ECC of 90% was observed to best reproduce the experimental uranyl nitrate association constants under ionic strengths similar to Suleimenov et al. 36 and provide reasonable coordination environments. It was employed for all subsequent molecular dynamics simulations. The TIP3P water model 50 was used for the bulk and electrolyte/hexane systems. Optimised force field parameters are given in the Supplementary Information, Table S1. All systems were first equilibrated in the isobaric-isothermal NPT ensemble for 40 ns using the Nose-Hoover thermostat 51 and Parrinello-Rahman barostat, 52 followed by isochoricisothermal NVT ensemble for 10 ns. The simulations were performed at 298 K using periodic boundary conditions with a leap frog verlet integrator 53 using a time step ∆t of 2 fs. PME (Particle-Mesh Ewald) summation 54 was used for long range electrostatic interactions. After equilibration, 30 ns production runs were performed in the NVT ensemble and used for data analyses. Sampling frequencies of the production run include 25 fs & 3 ps dump times depending upon the property of interest. ## Data Analysis The focus of this work is to understand the variations in uranyl speciation that result from significant changes to ion concentration and changes to solution structure at the interfaces of electrolytes with non-polar media, relative to the bulk electrolyte phase. Toward this end, the macroscopic interfacial properties (interfacial tension and width) were examined alongside analyses that reveal the local structure-including the coordination environments, solvation structure, and molecular speciation. The dynamic properties of molecular interactions, obtained from the relevant time correlation functions, are also reported. Statistical errors were determined using standard deviation of the calculated quantity over the length of the sampled trajectory. Interfacial Tension. The Kirkwood and Buff 55 pressure tensor method was used to calculate the interfacial tension, γ, 56 as an integral over the z dimension as where L z is the box length, N int is the number of interfaces (N int = 2 in Figure S1) and P zz , P yy , and P xx are the diagonal components of the pressure tensor. Local Structure and Speciation. Atom pair distribution functions were first used to examine inter-atomic distance correlations. These were compared to prior experimental and simulation data during force field validation. The composition of the primary coordination sphere about UO 2+ 2 , and the solvation environments about NO 3 and Li + were determined from networks of inter-molecular interactions using the ChemNetworks software package. 57 Distance geometric criterion were employed to define edges between nodes represented by UO 2+ 2 , H 2 O, NO 3 and Li + . These criterion were based upon the first minimum of the associated atom pair correlation functions of interest (including as shown in Figure S7). Geometric criterion are listed in Table S2. The distribution of denticities of nitrate complexation to uranyl was determined from the edge Dynamic Properties. It is of interest to understand how the presence of the liquid/liquid interface may alter the dynamic behavior of water of solvation and ion-ion interactions. The time of interaction of H 2 O as well as the NO 3 in the primary coordination sphere of UO 2+ 2 were calculated based upon a geometric cutoff r min , that defines the interaction. The probability P(t) associated with the interaction at time t and t+ ∆t is and the respective residence time 58 (τ ) is, where N (t, ∆t) is the continuous time duration of molecule/ion in the solvation shell or primary coordination sphere about the reference molecule/ion. 59 Nitrate ions are observed to have ∼ 10 × faster dynamic exchange between the primary coordination sphere of uranyl and the second solvation shell, relative to water. Fast dynamic properties have been previously been reported to be sensitive to the geometric cutoff employed to define primary and secondary regions about a solute. To investigate the sensitivity of the nitrate residence time about UO 2+ 2 , the dynamic correction procedure of Ozkanlar et al. 58 was employed to remove the transient breaking and formation of the interaction caused by the U Within the correction procedure, a tolerance of 1 ps with average persistence value of 7 ps was used. The computed residence times of nitrate in the uranyl solvation shell without correction was found to be 10 ps, whereas the correction procedure yielded a very similar value of 12 ps. ## Interfacial Slab and Identification of Truly Interfacial Molecules Analysis. To identify variations in speciation and solution structure in the interfacial region, two separate analyses were performed. First, a slab of the solution in the interfacial region was analyzed by taking a 10 increment in the z direction, consisting of 5 on either side of the Gibbs dividing surface of the water, defined as the z -axis position where the density of H 2 O is half of its value in bulk. The speciation, ion concentrations, residence times, and other properties were calculated in each slab and then compared to the analogous metrics of species present the instantaneous surface of the water. The Identification of Truly Interfacial Molecules (ITIM) algorithm 60,61 was employed to define the instantaneous surface of water and ions directly in contact with the organic phase, and for the comparison of the speciation and dynamic properties of the ions in the slabs vs. the instantaneous surface. The density of molecules in the instantaneous surface is fitted to a Gaussian function to obtain the position along z of the mean µ 0 of the distribution. The µ 0 is then used as a reference point (µ = 0) in the interface to define interfacial crest regions (Figure S2) (where the molecular density in the z direction negative to µ (5 )) and the trough regions (in the positive direction relative to µ). 3 Results and Discussion presence of a significant amount of tri-nitrato complex generally not observed in experimental estimations. 37 When nitrate is bound, there is a ∼65% η 2 coordination whereas prior analysis of the relative energetics of η 2 vs. η 1 in solution indicated no significant thermodynamic preference. 19,26,27 Finally, extended organizations are observed in the form of loosely bound intact uranyl nitrate species that appear to be correlated with the presence of UO 2 (NO 3 ) 2− 4 and UO 2 (NO 3 ) 3− 5 as these species have bridging and electrostatic interactions with one another via H 2 O, Li + , and NO 3 -, as observed in the U-U RDF (Figure S3). Although this was noted within the simulation literature, 37,41,62 there lacks strong experimental evidence for such long-range correlations. In combination, these data preclude simple calculation of the equilibrium constants K 1 and K 2 because of the complex equilibria with higher-nitrate containing species and water or nitrate bridged multinuclear U-containing configurations. Indeed, using the experimentally measured K 1 and K 2 values of 0.12 and 0.04, 36 it would be predicted that ∼25% of all uranyl species would exist as the mononitrate, and only ∼8% as To address these issues the electrostatic continuum correction methodology was em- ployed. This approach scales all ion charges and herein it is optimized to reproduce the experimentally observed equilibrium constants for the formation of the mono-and di-nitrato uranyl complexes. As cross-validation, the ligand denticity and solution organization as a function of LiNO 3 concentration was examined across all ECC values. As the speciation of uranyl-nitrate were examined (Figure 1A), the systematic scaling from 100% was observed to decrease the likelihood of highly coordinated uranyl ions by nitrate, effectively removing the equilibria of the UO as well as the loosely organized aggregated species (as demonstrated Figure S3). At a charge scaling of 90% of the original value, the probabilities of over-coordinated uranyl species decreased significantly and loosely bound uranyl nitrate aggregates dissipated (Figure S3). 64 Similarly, the strongly hydrated Li + ions maintain an average hydration number of ∼4.3 at 1 M [LiNO 3 ], using a distance cutoff r min of 3.0 in good agreement with the bulk aqueous phase. 65,66 Fitting the equilibrium constants K 1 and K 2 to the simulation data, using the equations S7 and S9 in the Supplementary Information, yields values of 0.12 and 0.04, respectively, which are well-within the range of experimental observation from spectroscopic measurements and are closest to the values of K 1 = 0.15 ± 0.04 at 6.25 M ionic strength in a solution of NaNO 3 and HClO 4 , and K 1 = 0.11 in LiNO 3(aq) . 18,19,36 In the system with 3 M LiNO 3 , the η 2 and η 1 modes of uranyl nitrate coordination is observed at 52% and 48%, respectively (Figure 1B), which is in good agreement with prior ab-initio simulations and experimental studies that predicted nearly equal favorability of the two binding nodes in the solution phase. 19,26,27 Charge scaling greater than 10% leads to a significant weakening of the uranyl-nitrate interactions, and minimal concentration of any uranyl nitrate complexes and instead solvent separated ion-pair interactions, as demonstrated by the RDF in Figure S4. In combination, these data indicate that the 90% ECC provides the best representation of uranyl-nitrate interactions in LiNO 3(aq) across a range of concentrations. Using this optimized potential, the experimental K 1 and K 2 are well-reproduced, the ratio of mono-and bidentate NO 3 binding modes are in agreement with ab initio 27 and experimental predictions, 19 and the solution structure as a whole is consistent with experimental observation. [LiNO S4). Ion solvation exhibits important dynamic properties, characterized by the exchange between first and second solvation shells on the ps to ns timescale. 67 These phenomena are intimately related to complexation reactions that occur via solvent dissociation pathways. 68,69 The residence times of solvating H 2 O about uranyl are generally observed to be high, in the range 40-775 ps depending upon the simulation and experimental techniques, and solution conditions. Consider that NMR which has a distance dependent signal perturbation. 21,67,70 Although the distance at which NMR begins to measure the dynamic exchange of H 2 O is not necessarily known, the computational residence time is generally defined as being strictly between the first and second solvation shells. Strong ion-dipole interactions of UO 2+ 2 with water, 71 and polarization across solvation shells is largely responsible for the long residence times. 72 However, this may be perturbed by long-range competitive interactions with background electrolytes, changes to overall solution-phase dynamic properties, 73 ## Characteristics of UO 2+ 2 at the Electrolyte/Hexane Interface The macroscopic and microscopic behavior of liquid/liquid interfaces are deeply intertwined. The interfacial tension (γ) increases nearly linearly with LiNO 3 concentration (Table S 3), in a manner consistent with the ion concentration at the electrolyte/hexane interface (Figure 4). The slope corresponding to the change in interfacial tension as a function of electrolyte concentration has been proposed to be a more accurate indicator of ion-specific effects 75,76 than an individual γ value at a specific concentration. 77 The average dγ/dm of 1.69 ± 0.48 mN/mM (mili Newton per meter Molar) is in agreement with the experimental value of 1.23 ± 0.12 mN/mM for the analogous LiNO 3(aq) /vapor system. 77 We now consider the more in-depth molecular scale interfacial chemistry, with an emphasis upon understanding the concentration dependent speciation of uranyl in the interface relative to the bulk. Ion Concentration Gradients. It is well-known that ions that reside at the interface perturb molecular-scale behavior as they introduce competitive interactions within an already altered environment relative to the bulk solution. Background electrolytes have been shown to influence metal-ligand chelation and speciation, as well as the rate determining steps in reaction kinetics. These may in turn influence mass-transfer kinetics across liquid interfaces. 68 A standard method to understand the ion concentration approaching the interface is to plot the charge density profiles (shown in Figure 4A). The charge densities as a function of concentration for UO 2+ 2 , NO 3 and Li + are plotted relative to the position of the mean of the water densities (µ 0 ) present in the truly interfacial water layer. The figure shows a sharp negative peak between 0 and -5 and a positive peak between 0 and 5 . Collectively, this indicates a weak electric double layer structuring at the liquid/liquid interface. To further understand the distribution of ions in various interfacial regions (the crest and trough) we plotted the number density distribution of ions in the truly interfacial layer (Figure 4B-D) and present the percent distribution of all ions in the truly interfacial layer (layer 1) and subjacent layers in Table 2. The distribution of ions along µ axis reveals that both UO 2+ 2 and NO − 3 predominantly reside in the trough region (positive µ) whereas the number density of Li + is distributed evenly in the crest and trough regions of the truly interfacial layer. From the percentage of ions in the interfacial layer (Table 2) it is apparent that very little Li + exists at the instantaneous surface, although there is significant population within the interfacial region as demonstrated by the density profile. These data are consistent with recent X-ray photo-electron spectroscopy interpretations of the prevalence of Li + in lithium iodide solutions at the electrolyte/vapor interface. 78 Prior work has demonstrated that Li + sheds solvating H 2 O within the instantaneous surface, which disfavors residence therein. 61 When combined with observation of relatively consistent NO − 3 concentration in the instantaneous surface and subjacent layers, the negative charge density in the region directly contacting the organic phase is presumed to be an outcome of Li + cation depletion in the instantaneous surface rather than anionic excess. not lose any H 2 O of solvation within the interfacial region. Nitrate is observed to interact with uranyl in both primary and secondary solvation shells at the electrolyte/hexane interface, as shown in Figure 6C. Indeed, the SSIP form predominates. The concentration of SSIP uranyl-nitrate species is ∼5.5 times that of the complexed mononitrate species at 1 M and ∼6.5 times at 5 M compared in both the bulk and at liquid/liquid interface. The increase in nitrate solvation is observed to be nearly linear in both primary and secondary solvation shells, however the growth of SSIP interactions is significantly more steep. Given the changes to the solvation properties of nitrate and uranyl in the interfacial region, and the introduction of competitive interactions at the interface, it is reasonable to question whether the nitrate association constants for uranyl would vary in the interfacial region relative to the bulk. The fractions of uranyl-nitrate complexes within the interfacial region of electrolyte/hexane (shown in Figure 2) reveal that the varying interfacial organization and ion concentrations do not perturb the uranyl-nitrate association constants relative to the bulk region. Within the interface, the mol fraction of UO 2 (NO 3 ) + complex increases linearly from 0.17 ± 0.02 at 1 M to 0.25 ± 0.08 at 5 M [LiNO 3 ]. Even though the fraction of UO 2 (NO 3 ) 2 complexes are less than UO 2 (NO 3 ) + complexes, it also increases from 0.048 ± 0.004 at 1 M to 0.098 ± 0.003 at 5 M [LiNO 3 ]. The K 1 value of 0.13 and K 2 value of 0.06 are obtained by fitting to equations S7 and S9, respectively. A similar coordination behavior is observed in terms of percent denticity changes for all binding modes from 1 M -5 M at the electrolyte/hexane interface compared to bulk. Solvation Dynamics. Although the solvent exchange rate is typically considered a rate limiting process for metal-ligand complexation, when this reaction further depends upon residence in the interfacial region, then both the rate of solvent exchange and the rate of migration in and out of the interface, becomes highly important. Within the interface, the solvent exchange dynamics are significantly faster than in the bulk. At 1 M LiNO 3 , the residence time of water is only a fifth of that in the bulk (75 ps vs. 450 ps). As such, ## Conclusions We present optimized force fields for the interaction of UO 2+ 2 in LiNO 3(aq) that maintains accurate association constants for the formation of uranyl nitrate complexes over a 1 -5 M electrolyte concentration regime. Under these conditions, the organization and dynamics of the bulk electrolyte solution was investigated. Subsequent biphasic simulation of the electrolyte/hexane system reveal several interesting features of the interfacial region. As anticipated based upon prior work, significant ion concentration gradients are observed for both Li + and NO − 3 , where depletion is observed for the former and excess is observed for the latter. Interestingly, the concentration of uranyl at the interface is the same as in the bulk, presumably because significant populations of solvent separated ion pairs with nitrate prevent uranyl dehydration therein. Second, the timescales of solvent exchange about uranyl are comparable to the residence time of the cation in the interfacial region. Thus, either of these processes may become the rate limiting step for interfacially mediated complexation reactions with extracting ligands. Both processes are also significantly slowed-down as [LiNO 3 ] is increased, nearly doubling over the 1 -5 M regime. Finally, it is shown that despite significant changes to the interfacial organization and dynamics, the uranyl nitrate association constants are unperturbed. Therefore, the knowledge of ion concentration at the interface can be used to predict the changes to uranyl nitrate speciation and thus, the reacting species with extracting ligands like tributyl phosphate as part of the mechanism of solvent extraction.

chemsum

{"title": "Uranyl Speciation in the Presence of Specific Ion Gradients at the Electrolyte/Organic Interface", "journal": "ChemRxiv"}

synthesis_of_multilamellar_walls_vesicles_(mlwv)_polyelectrolyte_surfactant_complexes_(pescs)_from_p

## Abstract: Multilamellar wall vesicles (MLWV) are an interest class of polyelectrolyte-surfactant complexes (PESCs) for the wide applications ranging from house-care to biomedical products.If MLWV are generally obtained by a polyelectrolyte-driven vesicle agglutination under pseudoequilibrium conditions, the resulting phase is often a mixture of more than one structure. In this work, we show that MLWV can be massively and reproductively prepared from a recentlydeveloped method involving a pH-stimulated phase transition from a complex coacervate phase (Co). We employ a biobased pH-sensitive microbial glucolipid biosurfactant in the presence of a natural, or synthetic, polyamine (chitosan, poly-L-Lysine, polyethylene imine, polyallylamine). In situ small angle X-ray scattering (SAXS) and cryogenic transmission electron microscopy (cryo-TEM) show a systematic isostructural and isodimensional transition from the Co to the MLWV phase, while optical microscopy under polarized light experiments and cryo-TEM reveal a massive, virtually quantitative, presence of MLWV. Finally, the multilamellar wall structure is not perturbed by filtration and sonication, two typical methods employed to control size distribution in vesicles. In summary, this work highlights a new, robust, non-equilibrium phase-change method to develop biobased multilamellar wall vesicles, promising soft colloids with applications in the field of personal care, cosmetics and pharmaceutics among many others. ## Introduction Polyelectrolytes and surfactants may assemble into complex structures known as polyelectrolyte-surfactant complexes (PESCs). When these compounds are oppositely charged, their self-assembly process is mainly driven by electrostatic interactions and it results in the formation of aggregates, which have a broad range of applications in biological materials, drug delivery, surface modifications, 9 colloid stabilization 10 and flocculation 11 and consumer health-care products. The rich mesoscopical and structural organisation of surfactants combined with the electrostatic interactions with polyelectrolytes give rise to a wide range of structures and phases. Many works reported cubic or hexagonal mesophases 15,16 but also a number of micellar-based structures: pearl-necklace morphologies, 2,18,19 interpenetrated polyelectrolytewormlike/cylindrical micelles network, 18, spheroidal clusters composed of densely packed micelles held by the polyelectrolyte, the latter known as complex coacervates when they form a liquid-liquid phase separation. 18,23,24 Very interesting PESCs structures are formed when the surfactant forms low curvature vesicular morphologies. It is in fact generally admitted that modifying vesicles by the addition of polyelectrolytes is an interesting, cheap and simple approach to obtain nanocapsules, 22 which are good candidates to be used as versatile delivery systems, 18,22 like gene delivery, 1,21,25,26 or as MRI contrast agents. 27 One of the first PESCs vesicular systems has been reported more than 20 years ago in DNA-CTAB (cetytrimethylammonium bromide) systems, which were the precursors of a number of carriers for gene transfection and often referred to as lipoplexes, when cationic lipids replace surfactants in DNA complexation. 28,29 If the term lipoplexe supposes the use of nucleic acids as complexing agents, similar structures, often addressed to as onion-like structures 30 or multilamellar vesicles, 13 were observed using both lipids and surfactants complexed by a wide range of polyelectrolytes. However, multilamellar, or onionlike, vesicles are rather characterized by single-wall membranes concentrically distributed from the outer to the inner core of the vesicle. Lipoplexes, on the contrary, are vesicular objects with a large lumen and a dense multilamellar wall. For this reason, in this work we employ the name multilamellar wall vesicles (MLWV). The mechanism of formation of MLWV was addressed by several authors, but a common agreement is not achieved, yet. Several works propose that the lipid:polyelectrolyte ratio controls the fusion of single-wall vesicles MLWV, 18,28, while others rather observe vesicular agglutination under similar conditions. In fact, a general consensus has not been found and a multiphasic system including agglutinated vesicles and MLWV are actually observed. 38 The question whether or not MLWV, and PESCs in general, are equilibrium structures and how they DOI: 10.26434/chemrxiv.12058929 are formed is still open, especially when they are prepared under non-equilibrium conditions. 18 To the best of our knowledge, the only works exploring a stimuli-induced approach in the synthesis of MLWV in particular, and PESCs in general, were proposed by Chiappisi et al.. 20,39 However, the pH variation in these work was still performed under pseudo-equilibrium conditions with equilibration times ranging from 2 to 15 days for each pH value. In a recent work, we have explored a Co-to-MLWV phase transition under nonequilibrium conditions using a continuous variation in pH, 40 as illustrated by Figure 1. We could show that in the presence of G-C18:1, an acidic microbial glycolipid biosurfactant, 41,42 and poly-L-lysine (PLL), a cationic polyelectrolyte (PEC), the pH-stimulated micelle-to-vesicle phase transition of the lipid drives a continuous, isostructural and isodimensional, transition between complex coacervates and multilamellar wall vesicles. In the present work, we generalize the method of preparing MLWV through a phase transition approach performed under non-equilibrium conditions and we show its performance in comparison to the more accepted method of vesicular agglutination. We show that this method can be applied to a broader set of polyelectrolytes and we explore in more detail the structure and size control of MLWV. ## Chemicals In this work we use microbial glycolipids G-C18:1, made of a single β-D-glucose hydrophilic headgroup and a C18 fatty acid tail (monounsaturation in position 9,10). From alkaline to acidic pH, the former undergoes a micelle-to-vesicle phase transition. 41 The syntheses of glucolipid G-C18:1 is described in Ref 43 and 42 , where the typical 1 H NMR spectra and HPLC chromatograms are given. The compound used in this work have a molecular purity of more than 95%. The polyelectrolytes used in this work are chitosan, obtained from the deacetylation of chitin from crusteans' shells, poly-L-lysine, widely used in biomedical field, and polyethylenimine. Chitosan oligosaccharide lactate (CHL) (Mw ≈ 5 KDa, pKa ~6.5) 44 with a deacetylation degree >90%, poly-L-lysine (PLL) hydrobromide (Mw ≈ 1-5 KDa, pKa ~10-10.5), 45 polyallyllamine hydrochloride (PAH) (Mw ≈ 1-5 KDa, pKa ~9.5), 45 polyethylenimine (PEI) hydrochloride (linear, Mw≈ 4 KDa, pKa ~8) 46 and gelatin (type A, from porcine skin, Mw ≈ 50-100 KDa, isoelectric point 7-9) are purchased from Sigma-Aldrich. All other chemicals are of reagent grade and are used without further purification. ## Preparation of stock solutions G-C18:1 (C= 5 mg.mL -1 , C= 20 mg . mL -1 ), CHL (C= 2 mg . mL -1 ), PLL (C= 5 mg . mL -1 , C= 20 mg . mL -1 ), PEI (C= 5 mg . mL -1 ), PAH (C= 2 mg . mL -1 ) and gelatin (C= 5 mg . mL -1 ) stock solutions are prepared by dispersing the appropriate amount of each compound in the corresponding amount of Milli-Q-grade water. The solutions are stirred at room temperature (T= 23 ± 2 °C) and the final pH is increased to 11 by adding a few μL of NaOH (C= 0.5 M or C= 1 M). ## Preparation of samples Samples are prepared by mixing appropriate volume ratios of G-C18:1 stock solutions at pH 11 and cationic polyelectrolyte (PEC) stock solutions, as defined in Table 1. The final total volume is generally set to V= 1 mL or V= 2 mL, the solution pH is about 11 and the final concentrations are given in Table 1. The pH of the mixed lipid-PEC solution is eventually decreased by the addition of 1-10 µL of a HCl solution at C= 0.5 M or C= 1 M. The rate at which pH is changed is generally not controlled although it is in the order of several µL . min -1 . Differently than in other systems, 47,48 we did not observe unexpected effects on the PESC structure to justify a tight control over the pH change rate. employed using an energy of E= 12 keV and a fixed sample-to-detector (Eiger X 4M) distance of 1.995 m. For all experiments: the q-range is calibrated to be contained between ~5. 10 -3 < q/ -1 < ~4.5 . 10 -1 ; raw data collected on the 2D detector are integrated azimuthally using the inhouse software provided at the beamline and so to obtain the typical scattered intensity I(q) profile, with q being the wavevector (𝑞 = 4𝜋 sin 𝜃 𝜆 , where 2θ is the scattering angle and λ is the wavelength). Defectuous pixels and beam stop shadow are systematically masked before azimuthal integration. Absolute intensity units are determined by measuring the scattering signal of water (Iq=0= 0.0163 cm -1 ). The same sample experimental setup is employed on both beamlines: the sample solution (V= 1 mL) with the lipid and PEC at their final concentration and pH ~11 is contained in an external beaker under stirring. The solution is continuously flushed through a 1 mm glass capillary using an external peristaltic pump. The pH of the solution in the beaker is changed using an interfaced push syringe, injecting microliter amounts of a 0.5 M HCl solution. pH is measured using a micro electrode (Mettler-Toledo) and the value of pH is monitored live and manually recorded from the control room via a network camera pointing at the pH-meter located next to the beaker in the experimental hutch. Considering the fast pH change kinetics, the error on the pH value is ± 0.5. ## Polarized Light Microscopy (PLM) PLM experiments are performed with a transmission Zeiss AxioImager A2 POL optical microscope. A drop of the given sample solution is deposited on a slide covered with a cover slip. The microscope is equipped with a polarized light source, crossed polarizers and an AxioCam CCD camera. ## Cryogenic transmission electron microscopy (cryo-TEM) Cryo-TEM experiments are carried out on an FEI Tecnai 120 twin microscope operated at 120 kV and equipped with a Gatan Orius CCD numeric camera. The sample holder is a Gatan Cryoholder (Gatan 626DH, Gatan). Digital Micrograph software is used for image acquisition. Cryofixation is done using a homemade cryofixation device. The solutions are deposited on a glow-discharged holey carbon coated TEM copper grid (Quantifoil R2/2, Germany). Excess solution is removed and the grid is immediately plunged into liquid ethane at -180°C before transferring them into liquid nitrogen. All grids are kept at liquid nitrogen temperature throughout all experimentation. Images were analyzed using Fiji software, available free of charge at the developer's website. 49 ## Results In recent publications, 40,50 we have explored the complex coacervation between microbial glycolipids and cationic polyelectrolytes (PEC). For this reason, this aspect is only briefly shown in here. Cryo-TEM images presented in Figure 2 show the structure of PECcomplexed G-C18:1 complex coacervates above pH 7. Irrespective of the selected PEC, all systems show spheroidal colloids of variable size in the 100 nm range. One can identify two types of structures, both typical of complex coacervates: 23,24,50,51 dense aggregated structures, shown in Figure 2a,c and very similar to what was found by us 40,50 and others, 23 are attributed to dehydrated, densely-packed, micelles tightly interacting with the polyelectrolyte; a biphasic medium composed of spheroidal, poorly-contrasted, colloids embedded in a textured medium describe hydrated structures of less defined composition, probably describing an intermediate of coacervation step. The latter were also reported by us 40,50 and others. 51,52 In all cases, the complex coacervate phase (Co) is a PESC forming in the micellar region of the surfactant's phase diagram and having the specificity of a liquid-liquid phase separation, 18,24 compared to other supramicellar PESCs undergoing a solid-liquid phase separation. 18 The difference between dense and poorly-contrasted structures is PEC-independent and it is more related to the stage of coacervation. At an early stage, colloids with a relatively low electron density form and coexist with a rich micellar phase. Free micelles progressively interact with residual polymer chains. At a later, entropy-driven (dehydration and counterion release), 53 stage of coacervation, droplets with a higher electron density massively form. Unfortunately, neither the texture of the particles nor their internal structure can be easily controlled as they strongly depend on the type of PEC, its stiffness, charge density, stage of coacervation and even kinetics. For these reasons, isolating a specific structure in a Co phase can be challenging and we have ourselves found coexisting dense and poorly-contrasted structures, 40 thus preventing any reasonable structure-composition generalization concerning the images presented in Figure 2. At pH below 7, vesicular structures with multilamellar walls (MLWV phase) are observed by cryo-TEM for all PEC samples (Figure 3). These structures are closely-related to a lipoplexe-type phase rather than to an onion-like phase, whereas the latter is composed of concentric single-wall vesicles, while the former keeps a free lumen and a thick multilamellar wall. 28 Figure 3 also shows a strong packing of the multilamellar walls as well as a strong interconnection between adjacent vesicular objects, in agreement with lipoplexes and other multilamellar wall vesicles reported in the literature. 22 The walls are constituted of alternating sandwiched layers composed of tightly packed polyelectrolyte chains and interdigitated layers of G-C18:1. 40 d-spacing can be directly estimated from cryo-TEM images (Figure 3e,f) and we find a set of values of d= 33.7 ± 4.95 for the PLL system and d= 31.6 ± 3.00 , 25.3 ± 4.60 and 41.1 ± 0.30 respectively for CHL, PAH and PEI systems. Within the error, these values are compatible with interdigitated G-C18:1 layers, of which the thickness can be estimated to be around 30 by applying the Tanford relationship, 54 but also close to what is classically recorded for lipoplexes. 21,22,32 One should note an interesting feature on Figure 3d: the multilamellar walls of the PECSs involving PEI appear less tightly packed and more disordered DOI: 10.26434/chemrxiv.12058929 9 than for other PESCs. This effect may be a consequence of the freezing protocol, although all samples have been frozen in the same way, or related to the specific use of PEI. Another hypothesis, which does not exclude the previous one, could be that the local disorder results from electrostatically induced undulations of the membrane, as already reported on lamellar DNA-lipid complexes. 55 Cryo-TEM images recorded on the Co (Figure 2) and MLWV (Figure 3) phases show that the Co-to-MLWV transition is a general property of G-C18:1 PESCs: it strictly depends on the lipid phase behavior, while the polyelectrolyte only guarantees the cohesion between the lipid membranes. We highlighted elsewhere 40 4a), a broad correlation peak is observed at about q= 0.17 -1 for all lipid:PLL ratios, where the peak can be more pronounced either with concentration (B profile) or lipid:PLL ratio (A profile). SAXS profiles B and C were previously assigned to complex coacervates, and more details on the structure of the Co phase can be found in Ref. 40 . In similar systems, the slope at low q was shown to be indicative of the shape of the PESC; 39 here, the slope is below -3. If such values are typical of fractal interfaces, 56,57 we cannot unfortunately draw any conclusion on the structure of the complex coacervates, most likely because the Co phase in these systems is heterogeneous. 40 Below pH 7 (Figure 4b), a sharp diffraction peak and its first harmonics are visible respectively around q1= 0.17 -1 and q2= 0.34 -1 , characteristic of the (100) and ( 200) reflections of a lamellar order in the walls, described previously and shown in Figure 3. The dspacing of 37 is in agreement with the ones deduced from cryo-TEM (Figure 3e,f). Similar results are obtained at different lipid:PLL ratios (Figure 4c,d) but also for other PEC. All pH-resolved in situ contour plots in Figure 4 show three common features: 1) the Co-to-MLWV transition between pH 8 and 7, where q1 and q2 refer to the first and second order peaks of the lamellar wall; 2) a low-q shift of q1 and q2 when pH decreases to about 4.5, indicating a swelling of the lamellar period, and 3) a loss of the signal between about pH 4.5 and pH 3.5, below which a constant peak at higher q-values (generally around q= 0.2 -1 ) appears. These phenomena were quantitatively described in more detail in Ref. 40 and will only be summarized hereafter. When fully deprotonated at basic pH, G-C18:1 is in a high curvature, micellar, environment (Co phase) at basic pH. This state, represented by the drawing superimposed on Figure 4d, is proven by both cryo-TEM and the broad correlation peak at about q0= 0.17 -1 . After crossing the transition pH range between 8 and 7, the number of negative charges decreases and G-C18:1 is entrapped in a low-curvature, interdigitated layer, environment. The continuity between q0 and q1 strongly suggest an isostructural and isodimensional transition between the micelle and membrane configutations, without any loss of the interaction with the polyelectrolyte. This is also sketched on Figure 4d. When the pH is decreased further, the COOH content increases and thus the membrane charge density decreases. The interlamellar distance consequently increases due to the repulsive pressure exerted by the charged polyelectrolyte, which undergoes hydration and increase internal electrostatic repulsion. 2,59,60 When hydrogenation of carboxylate groups reach a certain extent, attractive interaction with PLL can no longer hold the membranes together and MLWV then lose their long-range lamellar order, which results in their complete disruption and the concomitant expulsion of PLL. Below pH 3, this mechanism leads to the precipitation of a polyelectrolyte-free lamellar, L, phase, which is also observed for PEC-free G-C18:1 solutions. A closer look at the experiments in Figure 3 indicates two additional features. The pH stability domain of the MLWV phase seems to vary with the lipid:PLL ratio. Comparison of Figure 3c and Figure 3d, respectively recorded at lipid:PLL= 1:1 and 1:2 reveal that the q1 peak of the MLWV phase is observed between pH 8 and 7. At the 1:2 ratio the MLWV phase starts at about pH 8 while at the 1:1 ratio the MLWV phase is only visible at pH is below 7. At higher concentrations (C= 10 mg . mL -1 ), but still for a 1:1 ratio, the stability frontier seems to be shifted at pH of about 7.5. 40 Although we do not have enough data to draw a general trend, it is wellknown that the lipid:polyelectrolyte ratio reflects the negative:positive charge ratio and for this reason it has a direct impact on the electroneutrality, thus affecting a number of structural features of PESCs: the wall thickness of the multilamellar structure, 20,61 the PESC morphology and colloidal stability. 18 For instance, order is noticeably improved when the charge ratio approaches 1:1, 62 and micelle-polyelectrolyte complex coacervation can be favoured or not. 63 This ratio is particularly crucial to control the properties of the lipoplexes and thus their applications: lipid/DNA ratio was reported to influence both the formation of lipoplexes and the release of DNA 64 and gene transfer activity. 65 Many authors have shown that the lipid:polyelectrolyte ratio actually controls the formation of MLWV structures 18,28, over agglutinated single-wall vesicles, but in fact it is more likely that a general consensus has not been found, yet, and reality often consists in a mixtures of MLWV and agglutinated DOI: 10.26434/chemrxiv.12058929 12 vesicles, 38 although many authors do not specify it. One of the reasons that could explain such discrepancy is the parallel influence of several other parameters like the charge density on both the lipid membrane and in the polyelectrolytes, the rigidity of the latter, the bending energy of the lipid membrane, the ionic strength and so on. 14,18 In the present case, it is important to note that: 1) G-C18:1 forms a stable MLWV phase with all PEC tested in this work and of different origin (biobased vs. synthetic) and rigidity. 2) Multilamellar wall vesicles are stable in the neutral pH range, which can be a good opportunity for applications in the biomedical field, for instance. An interesting remark concerns the long-range order inside the vesicular multilamellar walls. The width of the lamellar peak around q ~0.2 -1 is more than ten times larger for the CHL (Figure 4e, Δq ~3.10 -2 -1 ) than the PLL (Figure 4c,d, Δq ~2.10 -3 -1 ) system, either suggesting an average smaller size of the lamellar domains or a poorer lamellar order in the case of the MLWV obtained from CHL. The reason behind such difference could be the bulkiness and stiffness of CHL with respect to PLL, 31 but one should recall from Figure 2 and related discussion that [G-C18:1 + CHL] solutions do not form an extensive Co phase. We have already made the hypothesis that the Co phase is necessary to form the MLWV phase, 40 and we will reinforce this assumption in the next part of this work. The data collected so far show that G-C18:1 interacts with all polyelectrolytes tested in this work and that its micelle-to-vesicle phase transition drives the Co-to-MLWV transition. As one could reasonably expect, the strong electrostatic interactions between the positively-charged PEC and negatively-charged G-C18:1 drives the PEC formation across the entire pH range. To test the solidity of the PESCs synthesis using G-C18:1 and polycations, we employ gelatin as a possible alternative polyelectrolyte and which could be interesting to prepare biobased PESCs. We use a commercial (Aldrich) source of gelatin type A, a natural protein of isoelectric point between 7.0-9.0, below which the charge becomes positive. Figure 5 shows pH-resolved in situ contour plots of gelatin and [G-C18:1 + gelatin] samples. The control gelatin sample in Figure 5a shows no specific contribution across the entire pH range between 0.1 < q / -1 < 0.4. Interestingly, the [G-C18:1 + gelatin] sample presented in Figure 5b does not show any signal either in the same pH and q range, except for the systematic signal of the lamellar, L, phase of G-C18:1 below pH 4. 40,41 Despite an expected positive charge density of gelatin, the in situ SAXS experiment shows no sign of the Co phase above pH 7, indicating that the charge density is probably too low to interact with negatively-charged G-C18:1 micelles. Although somewhat unexpected because interactions with negatively-charged sodium dodecyl sulfate micelles across a wide compositional and pH range were reported in other studies, 66 this result is not a surprise. What it is more interesting from a mechanistic point of view is the lack of the MLWV phase below pH 7. Given its isoelectric point, type A gelatin is positively charged below pH 7 and it is then expected to interact with G-C18:1 negative membranes. In this work we have used a broad set of polyelectrolytes, of which the different chemical nature let us explore various aspects of their interactions with G-C18:1. If the nature of the polyelectrolyte (stiffness, charge density, …) is known to strongly affect the morphology and structure of PESCs, 14,31 in this work we show that: 1) when the Co and MLWV phases are formed, the structure of the corresponding colloidal structures is very similar, whichever the polyelectrolyte used, even if local phenomena like swelling or long-range order may vary from one polyelectrolyte to another. 2) The Co and MLWV phases are only obtained with polyelectrolytes with a net positive charge, that is polycations. 3) The MLWV phase is always preceded by the Co phase, which seems to be a necessary condition to drive the isostructural and isodimensional Co-to-MLWV transition. This phenomenon does not occur when gelatin is employed and where the MLWV phase is not observed. On the contrary, the MLWV phase is obtained for the CHL system, despite the fact that we do not have a proof by SAXS of the Co phase. To this regards, we must outline that the SAXS signal for the [G-C18:1 + CHL] system at basic pH is dominated by the precipitated CHL phase, which we think to be in major amount but not the only phase. Cryo-TEM shows the presence of an unknown fraction of complex coacervates, which we believe to be source of the MLWV phase at pH below 7. We also believe DOI: 10.26434/chemrxiv.12058929 14 that the higher disorder of the MLWV phase in the [G-C18:1 + CHL] system (broader first order diffraction peak compared to the PLL-derived MLWV in Figure 4e) could be attributed to the smaller fraction of the initial Co phase. In other words, the presence of a less ordered MLWV phase in the CHL system could then the indirect proof that probably a small fraction of the Co phase forms in the CHL system. ## Quantitativity and size control If the synthesis of PESCs involving vesicles and polyelectrolytes, and eventually forming MLWV, has long been addressed in the literature, 36,67,68 very few studies, if none, address the issue of quantitativity in relationship to the mechanism of formation. In particular, the synthesis of MLWV from a continuous isostructural phase transition from a coacervate phase has not been addressed before, because MLWV are generally obtained by mixing vesicles and polyelectrolytes in solution. 18,28, 36 If some authors state that the formation of MLWV is driven by the lipid:polyelectrolyte ratio, other authors show that a mix of agglutinated vesicles and MLWV are actually obtained. 37,38 Other procedures could probably be followed to increase this control when working with pre-formed vesicles, such as the insertion of the polymer into the hydrophobic vesicle bilayer, which was reported in the case of polycations bearing pendant hydrophobic groups. 36,69 However, it was found that such interaction could be accompanied by lateral lipid segregation, highly accelerated transmembrane migration of lipid molecules (polycation-induced flip-flop), incorporation of adsorbed polycations into vesicular membrane as well as aggregation and disruption of vesicles. 69 To evaluate the amount of MLWV with respect to agglutinated vesicles, we compare the sample obtained by continuous Co-to-MLWV phase transition with a sample obtained by the more classical approach consisting in mixing G-C18:1 single-wall vesicles and polyelectrolyte, the main one employed in the literature of MLWV. If SAXS can prove the presence of a multilamellar structure, it cannot be easily employed to quantify and discriminate between the two structures. For this reason, instead of SAXS, we evaluate the content of MLWV between the two methods of preparation by combining cryo-TEM with optical microscopy using crossed polarizers. If cryo-TEM can differentiate between agglutination and MLWV, its high magnification is poorly compatible with good statistics, unless a large number of images are recorded. On the contrary, optical microscopy using cross polarizers is the ideal technique to differentiate, on the hundreds of micron scale, between MLWV and agglutinated vesicles: multilamellar structures (but not single-wall vesicles) show a characteristic maltese cross DOI: 10.26434/chemrxiv.12058929 pattern 70 under crossed polarizers, found both in concentric lamellar emulsions 71 and in spherical lamellar structures. 72 Cryo-TEM of a samples obtained from a Co-to-MLWV phase transition was shown in Figure 3 and, as already commented above, they show a massive presence of vesicular structures having multilamellar walls, as also confirmed by the corresponding SAXS data presented in Figure 4. Figure 6 shows two representative microscopy images of a typical sample prepared with the same approach; images are collected under white (a,d) and polarized light with polarizers at 0°-90° (b,e) and 45°-135° (c,f). The system is characterized by a large number of vesicles highly heterogeneous in size but all below ~10 μm. Under polarized light and crossed polarizers the entire material displays a typical maltese cross colocalized with each vesicle. Despite the aggregation of the vesicles, also observed with cryo-TEM, maltese crosses are welldefined and nicely separated and each identifying single multilamellar wall vesicles. The entire material displays such a characteristic birefringency, strongly suggesting a quantitative presence of MLWV. The experiment consisting in mixing acidic solutions (pH 3.8) of pre-formed G-C18:1 single-wall vesicles and PLL is shown in Figure 7. A preliminary investigation by optical microscopy results in a different behavior and distribution of signal with respect to the sample obtained through the Co-to-MLWV phase transition. Figure 7a shows representative images of a sample being constituted of aggregated objects, each of size below 1 μm, expected for G-C18:1 vesicles. 42 The corresponding images recorded using crossed polarizers (Figure 7b,d) show a broad, undefined, birefringency associated to the aggregates with little, if no, content of maltese crosses. The featureless, generalized, birefringency signal suggests that MLWV are either not formed or they form in small amounts, in good agreement with the data presented by others. 37,38 This assumption is confirmed by cryo-TEM images recorded on the same system and showing a mixture of structures including agglutinated vesicles but also "cabbage-like" and multilamellar structures (Figure 7e-f). The massive presence of MLWV structures obtained through the phase transition process compare to the mixture of structure obtained from a direct mixing of preformed vesiclespolyelectrolyte solutions confirms the crucial role of the complex coacervates in the formation of MLWV: coacervation seems to be a requirement to the extensive formation of vesicular structures with multilamellar walls. 40 This is also in agreement with the data obtained from the [G-C18:1 + gelatin] system presented in Figure 5 and prepared using the pH variation approach. Also in that case, the absence of a complex coacervate phase had as a consequence the absence of the MLWV phase. An additional piece of evidence comes from the CHL system, in which the limited amount of the Co phase generates a more disordered MLWV phase. Combination of the data obtained with gelatin and employing the in situ pH variation with the data obtained by mixing vesicle and polyelectrolyte solutions at a given pH demonstrates the importance of the precursor Co phase during the phase change method in order to obtain a massive presence of MLWV structures. If the Co-to-MLWV phase transition is able to quantitatively produce MLWV, its main drawback is the poor control over their size distribution, as shown both by TEM and optical microscopy. To improve this point, we employed filtration (Figure 8a-c) and sonication (Figure 8d-f), these methods being known to efficiently control vesicles size distribution, 73 but unclear whether or not they have any deleterious impact on the MLWV structure. According to the cryo-TEM data in Figure 8a-c, filtration (pore size, φ= 450 nm) promotes the stabilization of colloidally-stable spherical MLWV, of which the diameter seems to be contained between 50 nm and about 300 nm, in agreement with the filter pore size. Concerning the effect of sonication, Figure 8d-f also shows a large number of spherical, un-aggregated, MLWV colloids, although the diameter appears to be bigger of several hundred nanometers if compared to the filtered sample. The cryo-TEM results are confirmed by intensity-filtered DLS experiments, presented in Figure 8g. The as-prepared sample (black curve) shows a MLWV distribution centered at 716 nm, while the filtered sample shows a distribution centered at 460 nm. To better evaluate the impact of sonication, we tested the influence of sonication time and according to DLS data (Figure 8g) we find that at t= 30' the size distribution is centered at higher diameter values and it is even broader than the as-prepared sample. Applying the same sonication conditions, but over a longer period of time (t= 1' or t= 1'30''), it is possible to reduce the MLWV diameter even if the size distribution is broader than the filtration approach, in agreement with the cryo-TEM data. These experiments show that control of the size distribution of MLWV is possible using standard methods employed in liposome science without perturbing the multilamellar wall structure. Finally, when the membrane reaches neutrality, polymeric repulsion becomes strong enough to disassemble the lamellae. The polyelectrolyte will most likely be entirely solvated and at sufficiently low pH (< 3) the G-C18:1 precipitates in the form of a lamellar phase, possibly free of the polyelectrolyte, a behavior characteristic of the control lipid solution at the same pH. We employ four polyelectrolyte, synthetic and natural and with different characteristic of rigidity and charge density (chitosan, poly-L-Lysine, polyethylene imine, polyallylamine); however, the nature of the polyelectrolyte does not seem to be a relevant parameter concerning the fate of the transition, as otherwise found for most PESCs. This may be explained by the strong proximity between the lipid and the polyelectrolyte throughout the isostructural Co-to-MLWV transition. If the method described in this work does not allow a tight control over the size distribution of MLWV, we also find that the multilamellar wall structure is stable against filtration and sonication, two common methods employed to control the size of vesicles. Last but not least, we show that if we employ the classical approach consisting in mixing pre-formed vesicles with a cationic polyelectrolyte solution at a given pH, we find a much broader structural diversity, including agglutinated single-wall vesicles, multilamellar but also cabbage-like structures, in agreement with previous literature studies.

chemsum

{"title": "Synthesis of Multilamellar Walls Vesicles (MLWV) Polyelectrolyte Surfactant Complexes (PESCs) from pH-Stimulated Phase Transition Using Microbial Biosurfactants", "journal": "ChemRxiv"}

visualizing_3d_molecular_structures_using_an_augmented_reality_app

## Abstract: We present a simple procedure to make an augmented reality app to visualize any 3D chemical model.The molecular structure may be based on data from crystallographic data or from computer modelling. This guide is made in such a way, that no programming skills are needed and the procedure uses free software and is a way to visualize 3D structures that are normally difficult to comprehend in the 2D space of paper. The process can be applied to make 3D representation of any 2D object, and we envisage the app to be useful when visualizing simple stereochemical problems, when presenting a complex 3D structure on a poster presentation or even in audio-visual presentations. The method works for all molecules including small molecules, supramolecular structures, MOFs and biomacromolecules. ## INTRODUCTION Conveying information about three-dimensional (3D) structures in two-dimensional (2D) space, such as on paper or a screen can be difficult. Augmented reality (AR) provides an opportunity to visualize 2D structures in 3D. Software to make simple AR apps is becoming common and ranges of free software now exist to make customized apps. AR has transformed visualization in computer games and films, but the technique is distinctly under-used in (chemical) science. 1 In chemical science the challenge of visualizing in 3D exists at several levels ranging from teaching of stereo chemistry problems at freshman university level to visualizing complex molecular structures at the forefront of chemical research. Visualization can be especially challenging since molecules are getting larger and more complex and span 3D. An elegant way to visualize molecules in 3D is to 3D print the desired structure, and protocols of how to do this starting from molecular structures have recently been described. 2 To describe the geometry and symmetry of complex molecules chemists are often forced to draw molecules in simplified or schematic ways and thus neglect information. One example of a highly complex molecule that is difficult to display in 2D is the Molecular Borromean Rings prepared by Stoddart and co-workers (Figure 1a). In their structural representation some atoms and labelling of atoms are omitted to simplifying the structure. 3 In the simplified 2D image bonds and atoms are overlapping and thus still make it difficult to visualize the geometry and symmetry. Many chemists even draw the same molecule twice in different formats in the same paper to better explain the connections of the different elements and its geometry. This is illustrated with the porphyrin nano-ball by Anderson (Figure 1b) and the supramolecular complex between biotin uril and the iodide anion (Figure 1d). It can be challenging to come up with a new synthetic route for complicated molecules such as Paclitaxel (Figure 1c) because it is hard to visualize how sterically congested regions effects each other. 6 For these types of problems described above, a simple way to visualize molecules in 3D would be beneficial. to view these structures in AR. The app can be found via the QR code or the link in Figure 2. a) The Borromean rings are shown in two different ways to emphasize the geometry and symmetry of it. b) The porphyrin nanoball is presented in two different ways to highlight the geometry and the connection of the different elements in the nanoball. c) Paclitaxel is complex molecule with many stereo centers. A 3D view of it helps to appreciate the complexity and may aid retrosynthetic analysis. d) A topview and sideview of the crystal structure of the biotin uril. Download the app by following the link below (Figure 2) and see how the AR works. In this contribution we describe how to make a simple AR app for a mobile device (e.g. phones and tablets) to visualize molecules in 3D. It is important to emphasize that this guide is made in such a way, that no programming skills are needed, and that only free software is used. When the app is made it is free to transfer the app via an USB cable from the computer to the mobile devices. However, if you want to publish the app in Google Play it requires a one-time payment to Google Play of 25 $. This guide must be followed tightly, because a series of programs are needed. When the AR app is made and transferred to a mobile device, a camera opens and it recognises an image. The image may be on a poster, in a book or on a screen. The recognition leads to a 3D model (of one or more molecules of your Journal 5/18/21 Page 4 of 14 own choice) is opening as a part of the real world through your mobile device. When all the software is installed correctly, the app is simple to make and can be used at a poster session or in classroom. Download an example of the AR app (follow the link or QR code below) on your android mobile device and see how it works and what it looks like (in the real augmented world). Once the app is downloaded and opened, then point the camera at Figure 1 and 2 and a 3D model of the molecules will appear in AR. Link to the AR app: https://play.google.com/store/apps/details?id=com.UniCPH.Android.MoleculAR P Figure 2. A 2D image of Biotin uril. Download the app from the link below or the QR code and point the mobile device to see the structure in 3D through the camera. https://play.google.com/store/apps/details?id=com.UniCPH.Android.MoleculAR ## THE HOW TO GUIDE: STEP 1: MAKING A 3D MODEL OF THE MOLECULE If you don't have a specific molecule or just want to make the app to visualize a common molecule, then skip this step and go to "Step 2: Working with Jmol". There are many programs that allow you to draw molecules in 3D. The most realistic 3D models for the given molecule are either from a crystal structure or a high level optimization calculation by using software such as "Gaussian". It is also possible to draw the molecule in chemical drawing software, such as "Chem3D" in the "ChemDraw" software package. When the geometry of your molecule is satisfying, then save it at the desktop as a "mol"-file. Only the geometry of the molecule is needed to be correct at this state. The format of your molecule is added in the next step. If you want all the nitrogen atoms to be blue, then write "color nitrogen blue" or if you want the bonds to be black then write "color bonds black" etc. When To publish an app to Google play a developer account is needed, this requires a one-time payment of 25 $ (2019). Find the website by searching "Google play developer console", pay for the account and log in. A new app is created by clicking the "Create application" button and filling in the name and default language, finish by clicking "Create" and the "Store listing" menu opens. In this menu fill in the following information: 1. Short description, 2. Full description, 3. Upload an image to create the icon for the app, this image must have a resolution of 512x512. 4. Upload at least two screen shots. 5. Upload a "feature graphic" this image must have a resolution of 1024x500. The left hand navigation menu contains the two mandatory steps "Content rating" and "App content" both of these are questionaires pertaining to the nature of the content, complete these by filling in the required information. Now navigate to the "Pricing & distribution" and perform the following steps. 1. Choose "FREE" app. 2. Select the countries that you want the app to be avaliable in. 3. Verify that you app complies with content guidelines and US export laws by checking the boxes in the "Consent" section. Finalise by pressing "save draft". To publish the app navigate back to the App eleases section, press the "Edit release" button in the "Production track" section. Finalise by clicking the review button. Google play will now review that all the necesary information have been provide and the app complies with registration. This can take up to seven days. ## NOTE : PERFORMANCE IMPROVEMENTS If very large structures or slower android devices are used performance of the app can be a little sluggish. It is not withing the scope of this paper to describe the full set of rendering optimizations one can do in Unity but the following two hints should get you a good portion of the way. Journal 5/18/21 Page 13 of 14 ## Remove excess strutures: When Jmol is used to generate the obj-file, an addition sphere is created for each atom, even though these spheres is located at the same spot as the original atom sphere unity takes up processing power calculating positions of these. To remove these extra sphere open the unity project again. In the Hierarchy panel expand the hierachy until the structure is expanded, right click at the structure and select "Unpack prefab". Now scroll down to the children named "SprereXXX" where XXX is a integer, select them all and press "delete" this should not affect the 3D structure of the molecule but signigicantly increase the performance. ## CONCLUSION We have described how to make a simple augmented reality app to visualize any 3D chemical model using free softwares. The method works for all molecules including small molecules, supramolecular structures, MOFs and biomacromolecules.

chemsum

{"title": "Visualizing 3D molecular structures using an augmented reality app", "journal": "ChemRxiv"}

a_theorized_new_class_of_polyhedral_hydrocarbons_of_molecular_formula_cnhn_and_their_bottom-up_scaff

## Abstract: We address the use of Euler's theorem and topological algorithms to design 18 polyhedral hydrocarbons of general formula C n H n that exist up to 28 vertexes containing four-and six-membered rings only; compounds we call "nuggets". Subsequently, we evaluated their energies to verify the likelihood of their chemical existence. Among these compounds, 13 are novel systems, of which 3 exhibit chirality. Further, the ability of all nuggets to perform fusion reactions either through their square faces, or through their hexagonal faces was evaluated. Indeed, they are potentially able to form bottom-up derived molecular hyperstructures with great potential for several applications. By considering these fusion abilities, the growth of the nuggets into 1D, 2D, and 3D-scaffolds was studied. The results indicate that nugget 24a (C 24 H 24 ) is predicted to be capable of carrying out fusion reactions. From nugget 24a , we then designed 1D, 2D, and 3D-scaffolds that are predicted to be formed by favorable fusion reactions. Finally, a 3D-scaffold generated from nugget 24a exhibited potential to be employed as a voxel with a chemical structure remarkably similar to that of MOF ZIF-8. And, such a voxel, could in principle be employed to generate any 3D sculpture with nugget 24a as its level of finest granularity.On a very thought-provoking article in New Scientist, entitled "Why think up new molecules?", Prof. Roald Hoffman presented reasons to justify why he thinks that this is a worthwhile venture 1 . Speculative, inventive and somewhat risky predictions to either confront or make an exquisite use of a theory, are, by their very nature, scientific endeavors. As Prof. Roald Hoffman concludes, "The predictor leaves the safety of known molecules and properties for the unknown. He or she takes a risk. And, in a way, flirts-in a game of interest and synthesis-with the experimentalist. " 1 . In this article, we do indeed take this path and present a new subclass of hydrocarbons we call nuggets.Polyhedral hydrocarbons of general formula C n H n comprise a class of organic compounds that can exhibit unique properties, such as: tensioned bonds in rings that may be formed by three, four or more carbon atoms 2 ; energy storage capability 3 ; high density 3 ; aromaticity or antiaromaticity 4 ; magnetism 5 ; and symmetry such as the ones exhibited by platonic solids and regular prisms 5 . However, due to their sometimes strongly stressed bonds, syntheses of polyhedral hydrocarbons are hardly easy. In this sense, Eaton et al. 6 reported a synthetic strategy for the polyhedral hydrocarbon cubane (C 8 H 8 ), which is a tetraprism system. Further, Katz et al., synthesized the C 6 H 6 compound, which is a triprism system 7,8 . In particular, this compound exhibits a more tensioned structure than cubane 7,8 . In addition, the C 10 H 10 polyhedral hydrocarbon was also synthesized 8,9 .From a structural perspective, the bond angles of polyhedral hydrocarbons, that are either platonic or prismanes, are of smaller values (60°-90°), when compared with the most common bond angles of carbon atoms (109.5°). These small bond angles introduce a structural tension, which tends to energetically destabilize the system.An interesting aspect of the polyhedral hydrocarbon cubane is its ability to store a large quantity of energy 10 . Based on the cubane synthesis, a set of derivatives was prepared that presented potential to be applied to materials science due to their cube fusion abilities. Examples of the cubane derivatives are the octamethylcubane 11 and octacyclopropylcubane compounds 12 . In addition, Moran et al. evaluated the viability of carbon and hydrogen www.nature.com/scientificreports/ formed cages with ions, in which these systems have the potential to be applied in magnetic resonance, acting as contrast agents, with semiconductive and ferromagnetic properties 13 . Cubane derivatives can also be employed as additives, for example, in fuel, due to their tensioned structures 14 . In addition, 4-methyl-cuban-1-amine and 4-methyl-cuban-1-methylamine compounds exhibited antiviral biological activity 15 . Finally, if synthesized in larger amounts, heptanitrocubane would perhaps be one of the most effective non-nuclear explosives possible 16 . Poater et al. studied several structural and energy aspects of a class of packed carbon nanoneedles, that were conceptualized by stacking up units of 4, 6, and 8 carbons with potential applications to nanomedicine by acting as drug carriers through nonpolar biologic media 17 . The ability of the polyhedral hydrocarbons to be structurally fused was further examined by Katin et al. 18 The authors studied a material based on polyprismanes and concluded that these systems are similar to the carbon nanotube 18 . In addition, the interactions of the orbitals between the parallel rings of these materials seem to be the main component associated with the stability of the systems 19 . Karpushenkava et al. 20 , studied both structural and vibrational properties of a set of polyhedral hydrocarbons of the C n H n cage class in gas phase. The authors concluded that when the energy associated with the cage tension is either negative or slightly positive, the corresponding compounds could be synthesized. An unique exception was verified for a triprism compound with a cage energy of + 55.2 kJ mol −1 (ref 20 ). Wang et al., reported three stable isomers of the type C 24 H 24 . In their article, G3(MP2) calculations revealed that the optimized geometries of these systems have a positive value for Δ f H 21 . These geometries are unstable when compared to their fullerene isomers. In addition, one of the structures formed with Si has the potential to be a semiconductor material and, by replacing the CH groups with nitrogen atoms, high-energy density materials can be prepared 21 . On the other hand, DFT methods were also employed by Shamov et al. 22 to predict both structural and energy properties of a set of C n H n compounds, with n being 12, 16, 20, and 24. Both C 12 H 12 and C 20 H 20 compounds were synthesized, and the energetic properties indicated that C 16 H 16 and C 24 H 24 could be prepared. In this sense, Ohno et al., investigated both dimers and trimers of the regular prisms, with 6, 10, 12, 14, 16, 18 and 20 faces, connected by cubane-shaped bridges 23 . Their results also revealed that these compounds are able to be formed in organic reactions at low temperatures. Moreover, due to the metastable nature of the regular prismatic compounds, they could be potentially employed, for example, in energy storage 23 . In this article, we employ Euler's theorem to deduce polyhedra containing four-and six-membered rings that exist up to 28 vertexes, that we call "nuggets". We then evaluate their energetics in order to conjecture the likelihood of their existence. Finally, because all nuggets can be fused together in several manners, either through their square faces, or through their hexagonal faces, we investigated the fusion abilities of this set of nuggets to investigate the perspectives for their growth into 1D, 2D, and 3D-scaffolds. ## Results and discussion The nuggets structural possibilities from Euler's theorem. Our intention was to design hydrocarbon polyhedra that could be potentially stable. Although there are polyhedral hydrocarbons of the type C n H n with triangular faces, such as the tetrahedron 24 and the triprism 25,26 , as well as ones with pentagonal faces, such as the dodecahedron and the pentaprism 9,26 , we decided to restrict our work to polyhedra whose faces are polygons with an even number of vertices. Such systems can have alternating double bonds, thus potentially displaying energy stabilization due to electronic delocalization. Let us first consider polygonal hydrocarbons of formula C n H n . The smallest polygon with this formula is triangular C 3 H 3 . However, C 3 H 3 is a radical system. The same happens with C 5 H 5 , as shown in Fig. 1. Actually, all neutral polygonal C n H n hydrocarbons with n being an odd number must be radical systems. On the other hand, when n is an even number with n ≥ 4, the C n H n polygonal hydrocarbons are neutral systems, with cyclobutadiene, C 4 H 4 , and benzene, C 6 H 6 , displaying planar structures and thus being the most important members of this class. But, when n is equal to or larger than 8, the compounds become non-planar 27 . Figure 2 shows images of these polygonal compounds up to n = 10. Because we intend to grow the polyhedra into 1D, 2D, and 3D-scaffolds by fusing together their polygonal faces, we will restrict the polyhedra in this work to those with square and hexagonal faces only, since it would be very difficult, if not impossible, to fuse together two significantly non-planar and twisted faces. In these polyhedral compounds, each carbon atom must be bound to a single hydrogen atom as well as to three other carbon atoms as well. Euler's theorem 28 defines a relation between the numbers of faces, edges and vertices for any simple polyhedron: the polyhedra of our interest. Simple polyhedra are topologically equivalent to a sphere, that is, these systems are polyhedra that have no central cavities as "donuts". Therefore, if inflated, in the limit, these systems would become spheres. There are two possibilities for a hydrogen atom bonded to a carbon atom in a carbon polyhedron: either it is located inside or outside the polyhedron. If the hydrogen atoms appear in the interior of the polyhedron, steric effects would be very significant due to the congestion between other hydrogen or carbon atoms, especially for the smaller polyhedra. Moreover, if all hydrogen atoms always point inwards, at least one hydrogen atom would have an HCC angle less than 90°, which is not reasonable from the point of view of chemical bonds. Therefore, to be chemically realistic in applying Euler's theorem, we will focus on carbon polyhedra with the hydrogen atoms of the CH bonds always pointing outwards. Euler's theorem for simple polyhedra relates the number of faces (F), edges (E), and vertices (V) by the formula: where V is the number of vertices, E is the number of edges and F is the number of faces. (1) where F 4 is the number of square faces, and F 6 is the number of hexagonal faces. Of course, each square face of the polyhedron delimits four edges, and each hexagonal face delimits six. However, if the edges are counted from each polyhedral face, they would be counted twice, since each and every edge of the polyhedron is shared by exactly two faces. Accordingly, the relation between the number of edges, E, and the number of square and hexagonal faces of such a polyhedron is given by the following equation: For our polyhedra, the number of vertices is represented by the union of three edges. That is, each carbon atom is chemically bonded to exactly three other carbon atoms, i.e. V = V 3 ; the fourth bond being to a hydrogen atom. And each edge is bounded by two distinct end points: the vertices. Therefore, the relation between the number of edges and the number of vertices is given by: From Euler's formula, Eqs. (1), and (4): From Eqs. (2), (3), and (5), we obtain: (2) www.nature.com/scientificreports/ By simplifying the term 6F 6 on both sides of Eq. ( 7), we finally obtain that F 4 = 6. This reveals that any simple polyhedron that has only square and hexagonal faces must always have 6 square faces for an arbitrary number of hexagonal faces, except one. This exception is because Euler's formula is a necessary, but not sufficient condition for a polyhedron to exist. As can be intuited from Fig. 3 a configuration of one hexagon and six squares cannot be possibly closed into a polyhedron without forming at least a second hexagonal face. Consequently, the number of hexagonal faces must be either 0 (for the cube), or equal or greater than 2 for a constant number of six square faces. Design and computational details. The software Blink 29 developed by the research group of one of us (SL) was used to generate a set of unique nuggets from the cube up to the three different solids with 28 vertexes, all with 6 square faces and up to 10 hexagonal faces. From graph encoded manifold and UNIVs data 30 the Blink software is capable of generating several representations of graphs, only formed by faces with even numbers of vertices-squares, hexagons, octagons, etc. In this article, the Blink software was employed to map the topologically different and possible shapes of up to n = 28 vertices. Among all possibilities generated, we selected, according to chemical criteria, a subclass that we call nuggets that is composed of those that have structural forms containing six squares and an arbitrary number of hexagons, either equal to zero, or greater than or equal to two, generating a set of three-dimensional representations of the nuggets. From this class, we selected the first 18 that led to chemically different structures 29,30 . Hence, we generated a set of all different such polyhedra, starting with the cube, C 8 H 8 , up to those containing 28 vertices, of empirical formula C 28 H 28 , a number which we found to be reasonable to explore from a chemical point of view. Figure 4 shows the chemical structures of all 18 nuggets obtained, identified by the number of vertices, that is of carbon atoms, which is identical to the number of hydrogen atoms, and an additional letter in case there are more than one such nuggets for a given number of vertices. Being fully aware that predicting the properties of unusual molecules is risky, in order to calculate structural, vibrational and energy properties of the set of 18 nuggets, we needed to choose a quantum chemical model chemistry that would be at the same time both accurate enough and workable, given the size of the systems that we want to study, to be able to make educated inferences on the prospects of their chemical realities. We thus chose the ωB97XD functional by Chai and Head-Gordon because of its inclusion of a version of empirical Grimme's D2 dispersion as well as long-range correction with superior results 31 , together with the 6-31G* basis set of Petersson et al. 32 , for ease of computation of the larger hyperstructures formed by the molecular building blocks. Accordingly, all geometries of the designed nuggets, as well as the more complex 1D, 2D, and 3D systems were fully optimized by ωB97XD/6-31G* calculations via both Spartan'14 33 and Gaussian09 34 softwares. All structures have been characterized to be minima with frequency calculations. Nuggets exhibiting polyhedral chirality. From Fig. 4, nugget 24b , nugget 26b , and nugget 28b exhibit polyhedral chiral properties, as can be seen, in an illustrative manner, in Fig. 5, below, where we represent their respective pair of enantiomers. www.nature.com/scientificreports/ Nuggets as voxels. Voxels are the three-dimensional (3D) equivalents of pixels. Analogously to pixels, which can be used to generate any 2D images by juxtaposition, voxels can be likewise used to generate any 3D sculptures. Voxels can be virtual as in computer 3D graphics or real as in 3D printers. For a carbon polyhedron to be able to efficiently function as a voxel, it should possess the important property of 3D space-filling. That property being satisfied, they could in principle perhaps function as solid controllable building blocks that could be used to assemble any arbitrary 3D structures by juxtaposition. Of all nuggets that we studied, in only three of them, the carbon atoms define space-filling polyhedra that could function as chemical voxels: nugget 8 (cubane), nugget 12 (hexaprismane or -prismane) and nugget 24a (a truncated octahedron hydrocarbon). Let us first consider nugget 8 (cubane), of point group O h . Cubane's chemical stability with respect to selfdecomposition in the absence of any other reagents is something that can be inferred from its corresponding calculated energy change of reaction. Accordingly, let us consider the possibility of a nugget 8 , cubane, molecule dissociating into either 2 molecules of cyclobutadiene (C 8 H 8 → 2C 4 H 4 ), or into 4 molecules of ethyne (C 8 H 8 → 4C 2 H 2 ), Fig. 6. The ΔE ωB97XD/6-31G* values for these reactions are equal to + 368.8 kJ mol −1 and 551.2 kJ mol −1 ; large values that prevent such dissociation from occurring despite cubane's highly tensioned cubic structure. These ΔE ωB97XD/6-31G* values indicate that these entropy-favored self-decompositions, are unlikely to occur spontaneously. These www.nature.com/scientificreports/ findings are consistent with the fact that, as previously mentioned, cubane (nugget 8 ) has already been prepared 6 . Further, cubane growth in three dimensions is predicted to be a stable allotrope of carbon. Actually, a carbon allotrope with this 3D-structure could be very well used as an energy storage compound and would probably exhibit a larger mass density when compared with all other allotropes of carbon, including diamond. Let us now examine the case of nugget 12 , the hexaprismane, which has the structure of a prism with two parallel hexagonal faces linked through six square faces (Fig. 4). Hexaprismane can be thought of as a face-to-face dimer of benzene. The calculated energy of dissociation of nugget 12 into two benzene molecules (C 12 H 12 → 2C 6 H 6 ) Fig. 7a, yields a ΔE ωB97XD/6-31G* = − 389.8 kJ mol −1 , indicating that, in this case, the spontaneous chemical selfdecomposition of hexaprismane is predicted to be highly likely to occur. As a reinforcement to this affirmation, the thermal cycloaddition of two benzene molecules [6 + 6] is symmetry forbidden 35 . Indeed, so far and despite many attempts, nugget 12 , C 12 H 12 , the hexaprismane, has never been synthesized. These facts point further in the direction that the growth of nugget 12 to three dimensions would quickly spontaneously transform such a hypothetical solid into superimposed layers of graphene, such as graphite. Recently, a vertical stacking of graphene has been evolved into materials with highly tunable electronic properties and unique functionalities: the van der Waals heterostructures (vdWHs) 36 . So, for all practical purposes, it is very unlikely that the hexaprismane hydrocarbon nugget 12 could ever be of practical use as a chemical voxel. Nevertheless, the geometric concept of an hexaprismane polyhedron as a chemical voxel has recently been realized by the synthesis of isoreticular pillar layered metal organic frameworks exhibiting properties such as catalytic activity 37 . Two other self-dissociation reactions that could be thought of for the hexaprismane nugget 12 would be: (i) self-dissociation 7b,c, respectively. These two large positive calculated values reveal, as expected, that the self-decomposition of hexaprismane nugget 12 into two benzene molecules is the one most likely to occur spontaneously. The third and last carbon voxel is nugget 24a , which has the geometric form of a truncated octahedron: a space-filling Archimedean solid displaying many geometric properties, nugget 24a is a hydrocarbon, not the C24 fullerene which presents the same carbon structure 38 , which is geometrically equivalent to both the B 12 N 12 Fullerene reported by Matxain et al. 39 as well as to ZIF-8, a very stable and largely researched metal-organic framework, MOF 40 . Due to its high symmetry, and much less strained chemical bonds than either cubane or hexaprismane, nugget 24a is a possibility to be considered as a carbon voxel. Let us now proceed by first examining its three possible forms of self-decomposition of nugget 24a : (a) into 4 benzene molecules, with a ΔE ωB97XD/6-31G* value of − 154.3 kJ; (b) into 6 cyclobutadiene molecules, with a ΔE ωB97XD/6-31G* value of + 2311.6 kJ; and (c) into 12 acetylene molecules, with a ΔE ωB97XD/6-31G* value of + 2858.9 kJ, Fig. 8a-c, respectively. These results indicate that nugget 24a , although possibly unstable with respect to a self-decomposition into 4 benzene molecules, can be expanded as voxel into a 3D solid that would constitute an allotrope form of carbon. By being constituted by carbon atoms only, and noncoplanar vicinal six-membered rings, it cannot be split into benzene molecules or into graphene layers that would benefit from electron delocalization for stabilization. The geometric arrangement of the carbon-only hexagons in a such a perfectly packed 3D solid, placing each and every carbon atom in a condition of equilibrium of forces, would most certainly prevent its dismantling. Its infinite 3D expansion leads to a carbon-only solid compound which would constitute an allotrope of carbon. So much so that a sample has been found and properly characterized as a natural, super-hard, and transparent crystalline polymorph of carbon from the Popigai impact crater in Russia, formed because of a natural shockwave event 41 , and established to be consistent with such structure 42 . Stability of the nuggets. Now, we turn our attention to the structural stabilities of the non-voxel nuggets. Due to their molecular formula, their self-dissociation into ring compounds is a bit more complex, necessarily being at least into a mixture of benzene and cyclobutadiene, according to where n = 6p + 4q, with n, p, and q being integers. Further, there can be multiple combinations of p and q integer numbers that solve this expression for a given integer value of n. However, due to their geometric shapes, it is not always possible for these nuggets to be disassembled into combinations of benzene and cyclobutadiene molecules according to any stoichiometrically possible pair of values of p and q. Indeed, some of these disconnections could www.nature.com/scientificreports/ be shape forbidden. Finally, self-dissociations could also happen into ethyne molecules according to C n H n → (n/2) C 2 H 2 , a reaction that would always be possible since n is necessarily an even number and there are no geometric restrictions for any edges to be detached from the polyhedra. Table 1 shows ωB97XD/6-31G* calculated energies of reaction for all possible shape-allowed self-dissociations of all studied nuggets. From Table 1, complete dissociations into ethyne molecules are unlikely to happen for all nuggets, the same happening for self-dissociations producing any number of cyclobutadiene molecules. Thus, we can divide the nuggets into two groups, according to their energies of self-dissociation reaction ΔE ωB97XD/6-31G* . The first group of nuggets is comprised by the ones with at least one of the calculated ΔE values being negative: nugget 12 (hexaprismane), nugget 18 , and all nuggets 24 (including the truncated octahedron, nugget 24a ). These are the nuggets that may perhaps be less stable. The second group of potentially more stable nuggets comprises nuggets 8, 14, 16, 20 (a,b,c), 22, 26 (a,b,c) and 28 (a,b,c). This group includes nugget 28b which exhibits polyhedral chirality. As far as we know, so far, none of them have been reported in the literature, not even as a theoretical possibility. These results reveal that most of the designed nuggets are seemingly energetically stable and, probably, not easily capable of self-dissociation into simpler organic compounds. On the other hand, the nuggets of formula C 20 H 20 , C 24 H 24 , C 26 H 26 , and C 28 H 28 possess structural isomers. Table 2 shows the energy of isomerization for all energetically favorable possibilities between these isomers. From Table 2, the most stable isomers for each of the molecular formulas are nugget 20c , nugget 24b , nugget 26a , and nugget 28a . However, transformation of one of the isomers into the other, involves fracturing a relatively rigid polyhedron through rearrangements of chemical bonds, thus rendering this type of transformation not likely. Vibrational frequencies. We now turn to examine the rigidity of the carbon scaffolds of the nuggets, that is, how they would vary from being hard and inflexible to soft and malleable as the number of vertices (carbon atoms) increases. We regard rigidity as a desirable property in a constrained geometry polyhedral compound, contributing to its structural stability and to other properties such as less susceptibility to thermal relaxation of excited states. Accordingly, in this work, we use the lowest calculated vibrational frequency of each nugget as a measure of its rigidity, the larger this frequency, the more rigid the compound. Indeed, the lowest frequency vibration, generally corresponds to a collective movement of all atoms of the molecule, fluttering in a synchronized manner along the corresponding normal coordinate. Table 3 shows frequency values for the lowest vibrational modes for each of the 18 nuggets, after geometry optimization, from ωB97XD/6-31G* density functional theory, DFT, calculations. For comparison purposes, Table 3 also shows the lowest vibrational frequency of other compounds, where one can see that, as expected, cyclic compounds are generally more rigid than linear ones. Further, the presence of double bonds certainly increases rigidity in otherwise similar compounds. Let us first consider the case of nugget 8 (cubane, C 8 H 8 ), which can be regarded as having been formed by two piled up cyclobutadienes. Cubane (ν ωB97XD/6-31G* = 628 cm −1 ) is more rigid than a cyclobutadiene (ν ωB97XD/6-31G* = 547 cm −1 ), indicating a sturdier structure. On the contrary, nugget 12 (ν ωB97XD/6-31G* = 394 cm −1 ), the -prismane, which can be regarded as having been formed by two piled up benzene molecules, is actually more flexible than benzene, which has a ν ωB97XD/6-31G* value of 414 cm −1 . In general, it can be argued that the sturdier the structure, the more difficult it is for it to get disassembled. Accordingly, as previously discussed, nugget 12 would probably easily self-dismantle into two benzene molecules. If we consider all other nuggets, from nugget 14 to nugget 28c , one of them, nugget 24a stands out as being the most rigid, having a very large lowest ν ωB97XD/6-31G* of 372 cm −1 . Nugget 24a is certainly special, displaying a very symmetric structure. This points to a molecular structure with much more balanced forces in each atom than those of the other nuggets. This reinforces the possibility of its 3D expansion, as discussed above, as likely being a very stable carbon allotrope that will probably be found to exhibit unique physical properties. All other nuggets display rigidities that are seemingly large enough to guarantee their structural stabilities. As one would expect, the more prolate ones (the "c" ones) are less rigid than the more spherical ones (the "a" ones). Naturally, as the number of carbon atoms in their structures increases, the nuggets tend to become less and less rigid. Nevertheless, their rigidities are, of course, still larger by a large difference than those displayed by the n-alkanes, and even by the cyclic alkanes with the same number of carbon atoms. All of this points to the direction that they could all be synthesized, as the synthetically challenging cubane indeed has been 6 . As rigid as they are, the nuggets can then be fused together to form even larger structures, generating an assortment of shapes and forms that can bring about regular and irregular solids, porous structures, etc., with many potential applications to materials science. To examine such possibilities, let us now turn to their energetic properties of fusing. ## Energetics of nugget-nugget face-fusion reactions. To be able to design novel 1D, 2D, and 3D-scaffolds from the set of nuggets considered in this article, let us now study the ability of these systems to perform face-fusion reactions. Because the nuggets present both square and hexagonal faces, their growths must occur via the fusion reactions of either two square or two hexagonal faces. However, not all these face-fusions may take place because some of the faces of these nuggets, mostly the hexagonal faces, are not exactly flat surfaces, but slightly skew polygons, whose vertices are not all coplanar. In such cases, for a fusion to occur, a requirement of spatial complementarity may not always be possible because the hexagonal faces tend to be all concave. On the other hand, square faces in these polyhedra are almost all invariably planar. Therefore, face-fusion reactions are generally predicted to occur more frequently through square faces, rather than via the usually more skewed hexagonal faces. www.nature.com/scientificreports/ all leading to a huge number of possibilities. Table 4 shows the energies of reactions, one for each type of fusion (whenever possible) that displayed the least ωB97XD/6-31G* energy values of reaction for each pair of identical nuggets. Results on Table 4 indicate that while there are 18 square face fusions, the number of hexagonal face fusions possible is only 5. The values of energy of hexagonal face-fusion reactions range from − 185.5 kJ for nugget 24a to 638.8 kJ to nugget 12 , with the same numbers for square face fusion reactions ranging from − 80.2 kJ, for nugget 26b , to + 427.4 kJ for nugget 8 , cubane. Although the larger the nugget, the more likely it is to display negative face-fusion energies of reaction, we notice an exception to this rule: among the 18 nuggets designed in this article, two identical molecules of the carbon voxel nugget 24a are predicted to perform hexagonal face-fusion reactions with the largest negative value of ΔE ωB97XD/6-31G* = − 185.0 kJ. Therefore, of all nuggets studied, nugget 24a www.nature.com/scientificreports/ is predicted to exhibit the largest aptitude to be applied to growth as 1D, 2D, and 3D-scaffolds, especially when one considers its voxel characteristics. ## Growth of nuggets into patterns. Upon face-fusion reactions, nuggets can grow into either regular or irregular structures. Let us first consider possible fused compounds displaying structures with regular patterns. The simplest of these patterns are tessellations: covering of the space with nuggets, without overlaps or gaps. Tessellations can occur in one, two or three dimensions, and are the result of face-fusion reactions of a nugget, or of a combination of nuggets, made up by their translations, rotations or reflections. The carbon voxels, nugget 8 , nugget 12 and nugget 24a would be natural candidates. However, as explained above, only nugget 24a would make such a chemically feasible tile for this purpose. Let us therefore turn to consider the growth of nugget 24a in 1 dimension. The idealized self-fusion reaction of two of them via one of its all-equivalent hexagonal faces, 2C 24 H 24 → C 42 H 36 + C 6 H 12 , ΔE ωB97XD/6-31G* is − 185.0 kJ, where C 6 H 12 refers to cyclohexane leads to a generator of the simplest 1D scaffold extension. Figure 9 shows its optimized geometry together with the released cyclohexane for easier visualization. Next, to evaluate the ability of nugget 24a in generating 2D-scaffolds, the following idealized fusion reaction was now considered: C 24 H 24 + C 42 H 36 → C 58 H 46 + C 8 H 14 , see Fig. 10 (left), where C 8 H 14 is (1R,6S)-bicyclo[4.2.0] octane, Fig. 10 (right) and whose predicted energy of reaction is − 85.7 kJ. Due to its 2D-structural arrangement its stability is substantially more accentuated when compared with the formation of the essentially linear C 60 H 48 1D compound obtained by fusing together the 1d-generator compound in Fig. 9 with another nugget 24a . This is because now a larger quantity of viable fusion reactions was carried out. Finally, let us evaluate the ability of nugget 24a in generating 3D-scaffolds. The following idealized fusion reaction was considered: C 58 H 46 + C 24 H 24 → C 71 H 52 + C 11 H 18 , see Fig. 11, where C 11 H 18 stands for (1s,1aS,4ar,7aR)nonahydro-1H-cyclobuta[de]naphthalene. The infinite 3D expansion of this polyhedron will lead to a carbon-only compound that would constitute an allotrope of carbon 42 . A solid model image of a piece of this allotrope can be seen in Fig. 12 below. It is noteworthy that, by acting as a space filling carbon voxel in this manner, at least in principle, nugget 24a could be employed to generate any 3D sculpture with itself as its finest granularity level. Another seemingly rigid allotrope of carbon can also be made from nugget 24a in the form of a regular skew apeirohedron. Similarly, but not exactly like the one advanced by Zhou et al. 43 , this will be formed by joining the carbon voxels nugget 24a through hexagonal pyramidal bridges linking hexagonal faces of one to square faces of others, in a manner so that each external square face of the hexagonal prismatic bridge shares an edge with a square face of one of the polyhedra while its opposite edge is shared with a hexagonal face of the other. Figure 13 exemplifies such a hexagonal prismatic bridge between two nuggets 24a . In this case, the idealized chemical 4), these bridged connections of hexagonal faces are more energetically favorable than connections via square faces. Therefore, the regular skew apeirohedron can then be formed by linking together, in this manner, each nugget 24a by 4 of its 8 hexagonal faces according to Fig. 14 below 44 . This putative allotrope of carbon, adding to previous exotic carbon allotropes 45 , would be very stable and rigid. Its density, however, would be evidently smaller than that of the space filling allotrope shown in Fig. 12. The presence of zeolite-like nanoporous cavities inside its structure could be a singular feature, that could perhaps prove to be the origin of many emerging and interesting properties. Other types of polyhedral solids, with larger cavities, can also be conceptualized, such as the one made, this time by nugget 16 , via square face-fusions, and whose projection in one plane reveals a semiregular or Archimedean tessellation, that can be grown indefinitely Fig. 15. Such a compound, if ever obtained, would also likely behave as a load resisting skeleton due to its symmetric nature. Furthermore, this structure could also be grown www.nature.com/scientificreports/ in 3D leading to lengthy tubular cavities that could prove eventually useful. Structures such as these, with large cavities in the middle, suggest applications to materials science as catalysts, porous powders, etc. Many more combinations can be conceptualized by connecting the nuggets. Figure 16 shows a helix compound made by fusion of nugget 28b via two of its quasi-planar hexagonal faces. Such a compound, whose form resembles a twisted rope, would exhibit helicity, a form of chirality. Besides, these regular and aesthetically appealing structures, several other large structures can be conceived by binding together several of the nuggets, leading to a myriad of hydrocarbon structures that would extend far beyond what is being here presented. The geometric possibilities of molecular structures that could in principle be formed based on these nuggets are truly vast: "symmetries, spirals, trees, waves, foams, tessellations, meanders, cracks, and stripes with fractal dimensions" 46 . ## Conclusions Euler's theorem and topological strategies were employed in order to theoretically design a set of 18 hydrocarbon nuggets of general formula C n H n containing four-and six-membered rings, that exist up to 28 vertexes. From Euler's theorem we demonstrated that all such polyhedra must contain exactly six four-membered rings, for an arbitrary number of six-membered rings equal or greater than two. Among these 18 nuggets, 13 are novel systems, with 3 of them exhibiting polyhedral chirality. We also showed that, with the exception of hexaprismane, which is predicted to easily self-dissociate into two benzene molecules, and therefore unlikely to be synthesizable; and also with the exception of nugget 18 , which is presumably expected to dissociate into three benzene molecules, all other nuggets are likely to be relatively stable and not self-dissociate or degrade. Subsequently, vibrational properties revealed that the designed nuggets are sufficiently rigid. In this sense, the nuggets with 28 carbons are predicted to exhibit a structural rigidity, in average about 100 times greater than that of the linear alkane n-octacosane C 28 H 58 . We also explored the expansions of these nuggets into larger structures by face-fusion reactions involving mainly hexagonal and sometimes square faces. Nugget 24a , the carbon voxel, resembles the most a fullerene (6 and 5-membered rings, however) in terms of the spherical shape, and possesses a chemical structure similar to the MOF ZIF-8. Due to its energetically favorable face-fusion reactions, Nugget 24a is deemed to be the most suitable one to have a large potential to be applied to growth as 1D, 2D, and 3D-scaffolds. Accordingly, any 3D sculpture could be generated with nugget 24a www.nature.com/scientificreports/ at its finest granularity level if sufficient synthetic control is one day discovered; or perhaps by carving from the innovative carbon allotrope presented in Fig. 11. In conclusion, as mentioned in the previous section, the nuggets could be in principle expanded into all sorts of forms: "symmetries, spirals, trees, waves, foams, tessellations, meanders, cracks, and stripes of fractal dimensions" 46 . Their scaffolds may be decorated with strategically placed substituents as quantized perturbations, to promote attractive forces between them for a potential use in molecular tectonics. Perhaps they can form designer hyperstructures made layer by layer in a precisely chosen sequence where electronic or even exotic phenomena, typically requiring exceptionally low temperatures, can be explored. In summary, these are structures that should www.nature.com/scientificreports/ be considered as possibilities and of interest to researchers from all areas of carbonaceous nanomaterials (e.g., fullerene, nanotube, graphene, etc.). Finally, we also present the perspective of novel carbon allotropes, both space filled, as well as with cavities, hinting at interesting properties if synthesized or found as it appears to be the case with the natural, super-hard, and transparent crystalline polymorph of carbon from the Popigai impact crater in Russia, formed because of a natural shockwave event 41,42 . Received: 14 December 2020; Accepted: 12 February 2021

chemsum

{"title": "A theorized new class of polyhedral hydrocarbons of molecular formula CnHn and their bottom-up scaffold expansions into hyperstructures", "journal": "Scientific Reports - Nature"}

chemical_synthesis_and_immunological_evaluation_of_new_generation_multivalent_anticancer_vaccines_ba

## Abstract: Tumor associated carbohydrate antigens (TACAs), such as the Tn antigen, have emerged as key targets for the development of synthetic anticancer vaccines. However, the induction of potent and functional immune responses has been challenging and, in most cases, unsuccessful. Herein, we report the design, synthesis and immunological evaluation in mice of Tn-based vaccine candidates with multivalent presentation of the Tn antigen (up to 16 copies), both in its native serine-linked display (Tn-Ser) and as an oxime-linked Tn analogue (Tn-oxime). The high valent vaccine prototypes were synthesized through a late-stage convergent assembly (Tn-Ser construct) and a versatile divergent strategy (Tn-oxime analogue), using chemoselective click-type chemistry. The hexadecavalent Tn-oxime construct induced robust, Tn-specific humoral and CD4 + /CD8 + cellular responses, with antibodies able to bind the Tn antigen on the MCF7 cancer cell surface. The superior synthetic accessibility and immunological properties of this fully-synthetic vaccine prototype makes it a compelling candidate for further advancement towards safe and effective synthetic anticancer vaccines. ## Introduction Cancer immunotherapy approaches based on synthetic vaccines containing tumor-associated carbohydrate antigens (TACAs) represent a topic of high interest. Anticancer vaccines have been conceived as innovative therapeutic agents for inclusion in multi-therapy settings in order to elicit antitumor immunity in patients with prior surgical resection of a primary tumor. This approach has the potential of preventing, or at least prolonging the time to recurrence at the metastatic sites, while exhibiting reduced toxicity compared to cytotoxic treatments such as radio-and chemotherapy. 4 Active immunotherapy educates the immune system of a patient to recognize tumor-associated antigens displayed on the vaccine construct and to trigger an immune response selectively directed towards malignant cells. 5,6 Extraordinary advances in Immunology and Synthetic Chemistry have led to the development of subunit vaccines based on tumor-associated glycans for the generation of more effective carbohydrate-directed immune responses. The administration of carbohydrate antigens alone in a vaccine formulation can only weakly activate B-cells, resulting in production of low-affinity IgM antibodies and short-living plasma cells. This is because, apart from a few exceptions, 11 carbohydrates are T-independent antigens that are not able to associate with MHC molecules and activate helper T-cells (T H ) to provide the required signals for B-cell stimulation. This, in turn, results in impaired antibody isotype-switching (to highaffinity IgG antibodies) and defcient production of both longliving plasma cells and memory B-cells. 12,13 As a consequence, conjugate-vaccine design relies on the hapten-carrier effect, whereby helper T-cell epitopes are included in the form of a protein carrier (semi-synthetic vaccines) or as discrete T Hepitopes (fully-synthetic vaccines), either directly connected to B-cell epitopes through linker moieties or grafted onto nonimmunogenic scaffolds or nanoparticles. Aberrantly glycosylated versions of the protein mucin-1 (MUC1) are overexpressed in most human epithelial cancers, 20,21 making this glycoprotein a target of high interest for both diagnostic and immunotherapeutic applications. 22 Cancer-related MUC1 tandem repeats display truncated carbohydrate moieties at O-glycosylation sites, exposing glycan epitopes such as the Tn antigen that are normally hidden in mucins expressed in non-transformed cells. The Tn antigen consists of an N-acetyl-D-galactosamine (GalNAc) unit a-O-linked to the serine (Ser) or threonine (Thr) of a peptide backbone, 23,24 and is overexpressed in approximately 90% of breast carcinomas, 25 and in 70-90% of colon, bladder, cervix, ovary, stomach and prostate cancers. As such, it represents an excellent target for vaccine development, and we and other research groups have focused efforts on the design of anticancer vaccine candidates based on the Tn antigen. The inherent low immunogenicity of carbohydrate antigens is even more critical in TACAs, as they are self-derived antigens and therefore not prone to being recognized as foreign molecules by the immune system. 39,40 Thus, a fundamental challenge for TACAbased vaccines involves the ability to produce functional, isotype-switched, IgG antibodies that are able to recognize native antigens expressed on cancer cells and selectively promote their clearance. Advances in the chemical-immunology feld have provided evidence that through an accurate structural design, it is possible to develop effective TACA-based anticancer vaccines with promising preclinical outcomes. 10, To overcome important limitations associated with protein-hapten conjugates, fully synthetic vaccine approaches are being developed that enable the assembly of structurally defned and easily characterizable constructs via modular chemical strategies. 15, In this context, we report herein the design, synthesis and immunological evaluation of unprecedented fully synthetic vaccines with high Tn-antigen valency that are able to elicit robust and functional immune responses in mice. ## Design and synthesis Since 2005, some of us have pioneered the use of cyclic Regioselectively Addressable Functionalized Templates (RAFTs) as designed scaffolds for the presentation of B-and T-cell epitopes in synthetic vaccine prototypes. 18,56,57 The RAFT core consists of a cyclic decapeptide featuring two Pro-Gly residues as b-turn inducers, which stabilize its conformation in solution and provide a relatively rigid structure where lysines' side chains point towards two spatially opposite directions, thus defning an upper and a lower domain. 58 In these early prototypes, four units of the Tn antigen moiety were grafted onto these scaffolds by using oxime ligation, a versatile chemoselective reaction that enabled the straightforward preparation of tetravalent structures by coupling deprotected aminooxy-glycan moieties without the need for activating agents. 59,60 While tetravalent vaccine constructs were immunogenic, 56,59, we reasoned that higher carbohydrate valency would enhance recognition properties due to the multivalent effect, leading to increased B-cell receptor (BCR) clustering, which represents an early and key step that impacts downstream immune signaling, including B-T cell communication. 67,68 Therefore, to go one step further and improve the immunological properties and potency of early prototypes, we developed a new set of fully synthetic Tn-based vaccine candidates displaying a 4-fold increase in B-cell epitope ratio as well as additional T cell epitopes for more efficient immune responses (Fig. 1a). Thus, multivalency-driven recognition of carbohydrate antigens by BCRs followed by receptor-mediated internalization of the vaccine construct incorporating a CD4 + epitope would allow antigen-presenting B cells to load the T H epitope onto MHC-II molecules on the cell surface. Direct B-T cell contact via T-cell receptor (TCR) on CD4 + T lymphocytes would provide bidirectional signaling, ultimately leading to B-cell proliferation and differentiation (Fig. 1b). Initially, we generated a small series of multivalent glycodendrimers as B-cell epitope carriers (Scheme 1, compounds 2, 5-8) to investigate whether a higher order multivalency could be benefcial to antibody binding in an anti-Tn mAb interaction assay (Fig. 2). 74 We frst synthesized Tn-based glycodendrimers 2 and 5, in which GalNAc units are a-O-linked to the Ser hydroxyl groups (referred to as "Tn-Ser" along the text), as in the native Tn antigen. Starting from intermediate 1 (see the ESI, Scheme S1 †), bearing four protected Tn-Ser residues, global Oacetyl and Fmoc removal gave tetravalent Tn-Ser glycodendrimer 2 (Scheme 1a). Meanwhile, functionalization of the free amino group of the lower lysine side chain in 1 with Bocaminooxyacetic acid N-hydroxysuccinimide ester (Boc-Aoa-NHS), 75 followed by base-mediated deprotection of the four GalNAc-Ser residues afforded intermediate 3. Deprotection of the Boc-aminooxy group on 3 and subsequent oxime ligation with core scaffold 4, 72 which displays four a-oxo-aldehyde groups, 76 provided hexadecavalent Tn-Ser glycodendrimer 5 in four steps from intermediate 1 following a convergent strategy involving unprotected moieties (see the ESI, Scheme S4 †). Conversely, the Tn-analogue glycodendrimer 8, in which the GalNAc units are attached via oxime linkages (referred to as "Tn-oxime" along the text), was synthesized in a divergent fashion by grafting the aminooxy-GalNAc moiety onto a hexadecavalent, a-oxo-aldehyde functionalized scaffold as the last step in the route. 73 With our set of compounds in hand featuring hexadecavalent Tn-Ser/oxime structures (Scheme 1, compounds 5 and 8, respectively) and their tetravalent analogues (Scheme 1, compounds 2 and 6 (ref. 61), respectively), we then performed direct interaction assays to evaluate their ability to be recognized by anti-Tn antibodies (anti-Tn mAb clone 9A7). 74 Tetravalent construct bearing GlcNAc-oxime residues (Scheme 1, compound 7 (ref. 73)) was used as a negative control (Fig. 2). While hexadecavalent Tn-Ser compound 5 exhibited the strongest binding, the interaction curve of the Tn-oxime glycodendrimer 8 indicated that this multivalent system can serve as an effective analogue of the native Tn antigen. In contrast, tetravalent Tn-containing compounds 2 and 6 showed absorbance values close to the baseline, comparable to those of negative control 7. These initial results prompted us to pursuit the synthesis of the complete vaccine structures based on selected glycodendrimers 5 and 8 for immunogenicity studies in mice. In addition to the hexadecavalent vaccine prototypes, the corresponding tetravalent vaccine constructs derived from 2 and 6 were also synthesized to evaluate the impact of antigen valency in the elicited immune response. To complete the design of our synthetic vaccine prototypes, we incorporated T helper CD4 + and CD8 + epitopes from ovalbumin (OVA) into the previous glycodendrimers to generate potent immune responses with strong and long-lasting production of IgG antibodies against the T-cell independent Tn carbohydrate antigen. OVA 323-339 CD4 + T helper and OVA 257-264 CD8 + T cell epitopes were synthesized "in-line" incorporating a cysteine residue at the C-terminus for further chemoselective conjugation to the core cyclopeptide scaffold (Scheme 2). The synthesis started with protected peptide sequence 9, which was obtained using standard Fmoc-based automated solid phase peptide synthesis (SPPS) on a Rink amide resin. Treatment with a TFA/TIS/H 2 O (96 : 2 : 2) cocktail resulted in the removal of all acid-labile side-chain protecting groups, with concomitant cleavage from the resin, affording peptide 10 in a 41% overall yield. We frst focused our efforts on the synthesis of the "nativelike" Tn-Ser vaccine prototypes based on glycodendrimers 2 and 5. Starting from building block 1 (see Scheme 1), which displays protected Tn-Ser peripheral residues and a free amino group of the lysine of the scaffold, coupling with the NHS ester of Boc-[S-(3-nitro-2-pyridinesulfenyl)]-cysteine (Boc-Cys(NPys)-NHS) in DMF provided intermediate 11 (Scheme 3a). 81,82 Taking advantage of the thiol activating nature of the NPys group, Boc removal (1 : 1 TFA/CH 2 Cl 2 ) from the cysteine residue, followed by disulfde bridge formation with cysteine-containing peptide protected Tn-Ser peripheral glycodendrimer S5 to the a-oxoaldehyde-bearing central scaffold 4 through oxime linkages (see the ESI, Scheme S6 †), was functionalized at its lower domain with OVA peptide 10 via disulfde bridge formation by using the three-step procedure described above (Scheme 3b). The resulting Tn-Ser hexadecavalent construct 15, however, was found to be unstable to the various deprotection conditions used to remove the O-acetyl and Fmoc moieties (see the ESI, Scheme S8 †), which presumably affected the internal oxime linkages, failing to afford the fully-deprotected vaccine candidate 16. To address this problem, we designed an alternative strategy towards a modifed vaccine construct (19) in which the internal oxime linkages were replaced with 1,4-triazoles (Scheme 4). Since this modifcation involves the "internal" part of the fnal glycosylated structure and not the external B-cell epitope display, the binding ability of the construct should not be affected. In contrast to the synthetic route towards 16, the new strategy allows for a more convergent assembly via late-stage CuAAC and also enables O-acetyl and Fmoc removal from the Tn-Ser moiety at an earlier stage in the synthesis. In the event, CuAAC reaction between a slight excess of fully-deprotected peripheral glycodendrimer 17 (see the ESI, Scheme S9 †) equipped with an alkyne handle on the lower domain and OVAfunctionalized, azide-bearing core scaffold 18 (see the ESI, Scheme S10 †) was carried out in a degassed DMF/PBS mixture using our previously reported protocol in the presence of With Tn-Ser vaccines 13 and 19 in hand, we next directed our efforts towards the synthesis of the Tn-oxime vaccine candidates. Unlike for the Tn-Ser constructs, the presence of a single free amino acid (Lys on the lower domain of the scaffold) in glycodendrimers 6 and 8 enabled the use of a versatile divergent strategy involving late-stage chemoselective functionalization at this position followed by installation of the T cell OVA epitopes in the last step of the route. Thus, starting from glycodendrimer 6 (see Scheme 1b), vaccine construct 20 was synthesized in 53% yield over three steps via Boc-Cys(Npys) installation, and disulfde bridge formation with OVA peptide 10 (Scheme 5). Following this three-step sequence, vaccine candidate 21 was obtained analogously from hexadecavalent glycodendrimer 8, although its poor solubility in DMF required the use of a DMF/ PBS mixture (1 : 1, pH 7.4) to achieve the coupling of the Boc-Cys(NPys) residue with a satisfying 41% yield after RP-HPLC purifcation (Scheme 5). ## Immunological evaluation With the synthetic vaccine candidates in hand, we next evaluated their ability to elicit immune responses. Groups of fve C57BL/6 mice were immunized subcutaneously three times every two weeks with each vaccine construct (50 mg dose) (Group A: hexadecavalent Tn-oxime 21; Group B: hexadecavalent Tn-Ser 19; Group C: tetravalent Tn-oxime 20; Group D: tetravalent Tn-Ser 13) in combination with the saponin QS-21 as an adjuvant 84 (20 mg). In addition, a group of control mice (Group E) were immunized with compound 21 without adjuvant (Fig. 3). Three weeks after the last immunization, blood was collected for serological analysis and the mice were sacrifced to assess cellular immunity. Notably, no toxic side effects (e.g. local inflammation, systemic reactions, mouse weight loss or death) were observed over the course of the immunizations (data not shown), indicating the non-toxicity of the synthetic vaccine constructs. First, we evaluated the ability of the constructs to generate antibody responses and of the antisera to bind the Tnantigen in different presentation modes [native Tn-Ser residues (5, 2) and unnatural Tn-oxime analogues (8, 6), as well as in higher (5, 8) and lower (2, 6) valency, see Scheme 1]. Microtiter plates were coated with the corresponding Tn glycodendrimers lacking the OVA peptide and the total anti-Tn IgG levels in blood sera were detected by ELISA. Group A mice, immunized with hexadecavalent oxime-linked Tn construct 21 in combination with QS-21, exhibited the highest IgG levels against its glycodendrimer counterpart 8 (Fig. 4a). Moreover, this group was the only one in which all fve mice were able to generate humoral responses. In contrast, Group B (hexadecavalent Tn-Ser 19 + QS-21) showed variable but lower IgG levels, while Groups C (20 + QS-21) and D (13 + QS-21) in which mice were immunized with the corresponding tetravalent constructs, exhibited IgG levels similar to the no adjuvant control (Group E). Interestingly, IgG antibodies produced by Group A mice (hexadecavalent Tnoxime 21 + QS-21) were also able to recognize hexadecavalent Tn-Ser glycodendrimer 5 (Fig. 4b), showing therefore no clear preference for the native or unnatural Tn presentation on the scaffold. However, antisera from these mice (Group A) were less efficient in binding the Tn antigen presented through lowvalency glycodendrimers Tn-oxime 6 and Tn-Ser 2 (Fig. 4c and d). Therefore, increased Tn antigen valency was found to be crucial for high-affinity binding of the IgG antibodies to the Tn antigen construct. On the other hand, while Group B mice showed antisera (IgG antibodies) able to bind both high-valency glycodendrimers 8 and 5 (Fig. 4a and b), their levels were considerably lower than those exhibited by Group A, and were not able to bind tetravalent glycodendrimers 6 and 2 (Fig. 4c and d). In these assays, tetravalent compounds 20 and 13 (Groups C and D, respectively) elicited IgG antibody levels similar to the no adjuvant Group E, highlighting the importance of the increased multivalency of vaccine candidates 21 and 19 to generate potent humoral responses. In addition to the presence of the OVA CD4 + T helper epitope, co-administration of QS-21 as an adjuvant was found to be essential for antibody classswitching and elicitation of IgG antibodies, with IgM antibody levels being negligible at the last time point (see the ESI, Fig. S45 †). 85 Antibody titration using oxime-linked glycodendrimer 8 for coating was carried out for the two groups showing the highest OD values, i.e. Group A (mice immunized with hexadecavalent Tn-oxime 21 + QS-21) and B (mice vaccinated with hexadecavalent Tn-Ser 19 + QS-21) (Fig. 5). Antibody subtyping of the anti-8 IgG isotypes revealed that Group A mice showed not only high total IgG titers but also elevated levels of the IgG1, IgG2b and IgG2c antibody subtypes. Interestingly, the IgG2c antibody titers in this group were as high as those observed for IgG1 antibodies, suggesting that 21 in combination with QS-21 elicits a balanced Th1/Th2 immune response (Fig. 5e). The IgG2c subtype is of particular interest since it is associated with potent antitumor effect such as complement-and antibody-dependent cell toxicity in mice. 86 In contrast, IgG antibody titers from Group B mice were not signifcantly higher than those of the no adjuvant control Group E, and subtyping of these antibodies revealed a bias towards the IgG2b subclass. These results indicate that hexadecavalent vaccine construct 21 bearing oxime-linked Tn antigen analogue is more efficient than its native Tn-Ser counterpart 19 in generating potent humoral responses, and was therefore selected for further immunological studies. We then evaluated the ability of the antisera to recognize the Tn antigen in a native context and analyzed the binding of the vaccine-induced serum antibodies to a human cancer cell line (MCF7) expressing the natural Tn antigen by using fluorescence microscopy. Notably, IgG antibodies elicited by immunization with compound 21 plus QS-21 (Group A mice) were able to specifcally bind Tn-expressing MCF7 cells. Moreover, indirect immunofluorescence images showed high signal coming from sera from Group A mice, whereas antisera from mice immunized with 21 without QS-21 (Group E) showed no Tn-specifc IgG binding (Fig. 6a). Antibody binding of sera from mice Group A showed broad membrane surface localization (Fig. 6b). In contrast, although compound 21 alone (Group E) was able to elicit IgG antibodies to a small extent (Fig. 5e), these antibodies were not able to specifcally recognize natural Tn antigen expressed in this tumor cell. These results highlight that antibodies elicited by the Tn-oxime analogue vaccine construct 21 coadministered with QS-21 are functional and able to recognize the native Tn antigen expressed on the cancer cell surface, confrming the tumor specifcity of the generated antibody responses. Next, we also investigated the cellular immune responses elicited by Tn-oxime construct 21 by testing its ability to specifcally activate T cells. At the time of sacrifce (three weeks after the last immunization, day 49), whole splenocytes were harvested and assayed for T cell restimulation in the presence of full-length OVA protein. After 48 h stimulation, splenocytes were analyzed for activation markers using flow cytometry by assessing the percentage of CD4 + CD44 high (ref. 87 and 88) and CD8 + CD107a + (ref. 89) in the restimulated splenocyte pools. Notably, vaccine construct 21 in combination with QS-21 activated CD4 + as well as CD8 + T cells after stimulation with the specifc antigen, confrming the immunogenicity of compound 21 for both B-and T-cells (Fig. 7). Therefore, construct 21 stands out as a potent and safe synthetic Tn analogue vaccine candidate that, coadministered with QS-21, is able to induce robust humoral and cellular immune responses without toxic side effects. ## Conclusions In summary, we report the design, synthesis and immunological evaluation of fully synthetic hexadecavalent Tn-based vaccine candidates with a 4-fold increased carbohydrate epitope ratio compared to earlier generation vaccines. These Tn-based constructs were designed as higher-valency B celltargeting modules to address the need for an effective BCR clustering to boost B cell activation and antibody production. The synthesis of the hexadecavalent Tn-Ser construct resulted especially challenging due to late-stage global deprotection issues that resulted in cleavage of the internal oxime linkages and product decomposition. Instead, synthetic access to an alternative Tn-Ser construct (19, Scheme 4) was achieved by replacing the four internal oxime linkers with triazole groups, enabling a convergent assembly of the modules using CuAAC click chemistry. To circumvent the challenges associated to the native Tn-Ser vaccine candidates, as a more practical chemical solution we generated more synthetically accessible vaccine constructs based on Tn analogues displaying the Tn antigen via oxime linkages. Thus, we developed a divergent and streamlined strategy whereby convenient functionalization of the hexadecavalent Tn-oxime glycodendrimer 8 was carried out in a three-step late-stage sequence to give the fnal vaccine structure 21 (Scheme 5). In mouse immunization studies with QS-21 as an adjuvant, hexadecavalent Tn-oxime and Tn-Ser constructs 21 and 19 elicited higher IgG antibody responses than their tetravalent analogues 20 and 13, confrming the high-valency beneft of our new-generation vaccines. Between the hexadecavalent constructs, Tn-oxime vaccine candidate 21 elicited the highest IgG antibody titers, consistently higher than those generated by Tn-Ser construct 19, and were also the most efficient in binding the Tn antigen in both presentation modes, i.e. oxime-linked Tn (8) and Tn-Ser (5). Moreover, IgG subtyping of the vaccine-induced antibodies showed different patterns for Group A and Group B sera. Immunization with 21 (plus QS-21, Group A) elicited high IgG1 and IgG2c isotype titers, whereas vaccination with 19 (plus QS-21, Group B) revealed a bias towards the IgG2b subtype (Fig. 5). Notably, antibodies generated by Group A mice were also able to bind Tn-expressing MCF7 cancer cells as assessed by fluorescence microscopy (Fig. 6), confrming the tumoral specifcity of the produced antibodies. Furthermore, vaccine construct 21 was also able to activate CD4 + and CD8 + cellular responses, as assessed by flow cytometry analysis of OVA-restimulated splenocytes (Fig. 7). In conclusion, we have developed a synthetically accessible newgeneration vaccine candidate (21) based on a multivalent oxime-linked Tn-antigen analogue that showed no toxicity and was able to generate potent and functional humoral and cellular immune responses. The superiority of this fully synthetic Tnbased vaccine construct 21, in terms of both its synthetic accessibility and immunological properties, makes it a promising candidate for further development towards cancer immunotherapy applications. ## Ethical statement Animals were cared for and handled in compliance with the Guidelines for Accommodation and Care of Animals (European Convention for the Protection of Vertebrate Animals Used for Experimental and Other Scientifc Purposes) and internal guidelines. Mice were housed in standard cages and fed on a standard diet ad libitum. All the experimental procedures were approved by the appropriate local authorities. The CIC bioGUNE animal facility is fully accredited by AAALAC International. ## Conflicts of interest There are no conflicts to declare.

chemsum

{"title": "Chemical synthesis and immunological evaluation of new generation multivalent anticancer vaccines based on a Tn antigen analogue", "journal": "Royal Society of Chemistry (RSC)"}

anomalous_p-backbonding_in_complexes_between_b(sir3)3_and_n2:_catalytic_activation_and_the_breaking_

## Abstract: Chemical transformations of molecular nitrogen (N2), including the nitrogen reduction reaction (NRR), are difficult to catalyze because of the weak Lewis basicity of N2. In this study, it was found that Lewis acids of the types B(SiR3)3 and B(GeR3)3 bind N2 and CO with anomalously short and strong B-N or B-C bonds. B(SiH3)3•N2 has a B-N bond length of 1.48 Å and a complexation enthalpy of -15.9 kcal/mol at the M06-2X/jun-cc-pVTZ level. The selective binding enhancement of N2 and CO is due to p-backbonding from Lewis acid to Lewis base, as demonstrated by orbital analysis and density difference plots. The p-backbonding is found to be a consequence of constructive orbital interactions between the diffuse and highly polarizable B-Si and B-Ge bond regions and the p-regions of N2. This interaction is strengthened by electron donating substituents on Si or Ge. The p-backbonding interaction is predicted to activate N2 for chemical transformation and reduction, as it decreases the electron density and increases the N-N bond length. The binding of N2 and CO by the B(SiR3)3 and B(GeR3)3 types of Lewis acids also has a strong s-bond contribution. The relatively high sbond strength is connected to the high positive surface electrostatic potential [VS(r)] above the B atom at the pyramidal binding conformation. Electron withdrawing substituents increase the potential and the s-bond strength, but favor the binding of regular Lewis acids, such as NH3 and F -, more strongly than binding of N2 and CO. Molecules of the types B(SiR3)3 and B(GeR3)3 are chemically labile and difficult to synthesize. Heterogenous catalysts with the wanted B(Si-)3 or B(Ge-)3 bonding motif may be prepared by B-doping of nanostructured silicon or germanium compounds. B-doped silicene show promising properties as catalyst for the electrochemical NRR. ## Introduction Nitrogen in its elemental formal is highly inert due to the very strong chemical bond of the nitrogen molecule. The high bond energy of the nitrogen triple bond in connection with the relatively modest bond energies of single and double bonds involving nitrogen transforms to high kinetic barriers for utilizing molecular nitrogen in chemical transformations. An important example is the Haber-Bosch process for converting nitrogen and hydrogen gas to ammonia, which relies on high temperatures and pressures in combination with transition-metal catalysis. Nature also depend on transition-metal catalysis for nitrogen reduction, but the nitrogenase enzymes manage the task at ambient conditions due to a more advanced catalytic machinery. 1,2 The development of efficient catalysts for nitrogen reduction has been hindered by the weak Lewis basicity of N2. Traditional Lewis acids, such as the boron trihalides, do not form donor-acceptor complexes with N2. Very strong Lewis acids, such as B(CF3)3, bind N2 but are too reactive to be useful. To avoid catalyst inhibition, chemical activation of N2 requires Lewis acids that preferentially binds N2 with binding energies that are stronger or at least on par with the binding energies for other Lewis bases that may be present. In the terminology of theoretical catalysis, N2 binding needs to break scaling relations for binding energies. In particular for electrochemical nitrogen reduction, the binding of N2 has to be competitive with the binding and reduction of the proton to avoid inhibition of the catalyst. 3 A number of transition metal compounds have been developed that can bind and catalytically activate N2. Their function has largely been attributed to the presence of low lying d-orbitals that allow for the concurrent acceptance of electron density from the N2 sorbital and p-backdonation towards the N2 p* orbital. The CO molecule is isoelectronic to N2 but a stronger Lewis acid and its catalytical activation follows a similar protocol. Main group chemistry has been less successful in nitrogen activation, but recent studies have demonstrated fixation and reduction of N2 by some novel hypo-valent borylene compounds. 11,12 These are argued to work by a similar mechanism as the transition metal catalysts, with N2 s-donation into an empty sp 2 -hybrid orbital and pi-backdonation from a fully occupied p-orbital on boron. However, it should be noted that N2 is bound to one borylene unit B at each end and thereby a delocalized p-system similar to that of a conjugated hydrocarbon is formed. In an attempt to characterize the N2 binding properties of trivalent boron Lewis acids, we observed an unexpected behavior upon replacing carbon for silicon as the atom bonded to boron. Whereas B(CH3)3 does not bind N2, B(SiH3)3 forms a complex with a very short and strong B-N bond. In Fig. 1 , we show the geometries and complexation enthalpies of the complexes of B(SiH3)3 with N2, NH3 and CO, together with the same properties for the corresponding complexes of B(CF3)3. All values have been obtained at the DFT M06-2X/juncc-pVTZ level of theory, and for the B(SiH3)3 complexes we also compare with coupled cluster calculations with values in italics, i.e. CCSD/6-31G+(d´,p) for geometries and CCSD(T)/juncc-pVTZ for energies. Beginning with the B(SiH3)•N2 complex, we find a very short B-N bond of 1.48 (1.53 ). This is even shorter than the sum of the covalent single bond radii, which amounts to 1.56 . 13 The formation of the complex results in slight increases in the B-Si and N-N bond lengths by 0.014 (0.003) and 0.012 (0.015) , respectively. The complexation enthalpy is -15.9 (-12.4) kcal/mol, which is congruent with a strong B-N bond, but a somewhat higher value than could have been anticipated based upon the bond length. Comparing the B(SiH3)3•N2 complex to the B(SiH3)3•NH3 complex, we find a considerably longer B-N bond of 1.64 (1.65 ) but a stronger interaction with a complexation enthalpy of -35.2 (-34.1) kcal/mol in the latter complex. Considering that NH3 is a much stronger Lewis base than N2, a much larger difference in the complexation enthalpy could have been anticipated. The formation of the B(SiH3)•CO complex results in similar changes to the geometry as the formation of the B(SiH3)3•N2 complex, but the B-N bond of the former is slightly longer than the B-C bond of 1.46 in the latter. The complexation enthalpy of -45.4 kcal/mol is much lower than for the N2 complex, and despite CO being a significantly weaker Lewis base than NH3 the CO-complex is stronger than the B(SiH3)3•NH3 complex. We continue with comparing the binding geometries and complexation enthalpies of the B(SiH3)3 complexes with those of the B(CF3)3 complexes. It should first be noted that B(CF3)3 is a very strong and chemically labile Lewis acid that only has been detected as a transient intermediate from thermal dissociation of B(CF3)3•CO. 14 The complexation enthalpy of the B(CF3)3•N2 complex is only slightly higher compared to the B(SiH3)•N2 complex, i.e. -14.7 vs ΔH Cmpl -15.9 kcal/mol, but the B-N bond is much longer in the former, i.e. 1.62 vs 1.48 . However, B(CF3)3 forms a very strong complex with NH3 with a complexation enthalpy of -61.2 kcal/mol, which is almost twice the strength of the interaction in the B(SiH3)•NH3 complex. On the other hand, the difference in B-N bond length compared to B(SiH3)3•NH3 is relatively small with a bond length of 1.60 in B(CF3)3•NH3. In contrast to B(SiH3)3, B(CF3)3 also forms a much stronger complex with NH3 than with CO, and the bonding in B(CF3)3•CO is significantly weaker than in the B(SiH3)•CO complex, although the B(CF3)3•CO bond is strong with a complexation enthalpy of -30.1 kcal/mol. Summarizing the geometrical and energetics data of Fig. 1, it is indicated that B(CF3)3 binds all three Lewis bases with a similar mechanism and the variation in complexation enthalpy agrees with their relative Lewis basicities. B(SiH3)3 seems to bind NH3 following a related mechanism, whereas the binding of N2 and CO invokes an additional component to the binding that results in much shorter intramolecular bond lengths (B-N or B-C) and enhanced binding strengths. The physical origin of the enhanced binding of N2 and CO is nontrivial to deduce, but it is reasonable to anticipate a connection to the p-backbonding mechanism prevalent in transition metal compounds that activates N2 and CO, and which has been indicated in the N2 activating hypo-valent borylene compounds. In this study we attempt to investigate this hypothesis in greater detail but we also take a comprehensive perspective in analyzing the Lewis acid-base interactions for this type of compounds. Furthermore, we investigate the potential for optimizing the selectivity for N2 binding and activation by means of chemical derivatization. Finally, we seek to identify nanostructured materials that are synthetically accessible and invoke the necessary chemical functionalities for use as heterogenous catalysts. ## Methods and theoretical procedures Structures of molecules and molecular complexes have been optimized at the M06-2X/jun-cc-pVTZ level of Kohn-Sham density functional theory. The M06-2X density functional is highly accurate for main-group chemistry, including non-covalent interactions, and it is parameterized to include short and mid-range London dispersion interactions. 15 This functional has been shown produce highly accurate energetics for classical Lewis adducts, as well as frustrated Lewis pairs, with an average deviation of 0.6 kcal/mol relative complete basis set extrapolated DLPNO-CCSD(T) energies. 16 The jun-cc-pVTZ basis set is the cc-pVTZ basis set augmented with diffuse s, p, d functions on non-hydrogen atoms. 17 The geometries and energies obtained with M06-2X have been compared with coupled cluster calculations at the CCSD(T)/jun-cc-pVTZ//CCSD/6-31G+(d´,p) level of theory for selected systems. To characterize the capacity of the Lewis acids to supply electrons for p-backdonation, the average local ionization energy [Ī(r)] was computed on molecular surfaces defined by the 0.001 au electron density contour. The Ī(r) is rigorously defined by eq. 1 within generalized Kohn-Sham density functional theory. 18,19 𝐼 ̅ (𝐫) = − ) 𝜀 + 𝜌 + (𝐫) 𝜌(𝐫) -./. ## +01 where ei is the energy of orbital i, ri(r) is the density of the orbital, and r(r) is the total electron density. According to Janak's theorem, 20 the negative values of the orbital energies can be considered approximations to the ionization energies, and Ī(r) can be interpreted as the average energy needed to ionize an electron at a point r in the space of a molecule or atom. Surface Ī(r) [ĪS(r)] has been demonstrated to be an effective tool for predicting local reactivity for electrophilic processes, such as electrophilic aromatic substitution reactions. 19 Minima in ĪS(r) [ĪS,min] reflect the positions most likely to donate electrons and thus most susceptible for electrophilic attack. To characterize the p-holes of the Lewis acids and their capacities to participate in electrostatic interactions with Lewis bases, the surface electrostatic potential was computed at the same isodensity contour as in the ĪS(r) computations. The V(r) is defined by, where ZA is the charge on nucleus A located at RA, and ρ(r) is the electron density function. V(r) is a physical observable, and qV(r) corresponds to the electrostatic interaction energy for a positive or negative point charge (q) at different positions (r) in space. Surface maxima in V(r) [VS,max] has been demonstrated to be an efficient indicator of the most active positions for nucleophilic noncovalent interactions, such as halogen and hydrogen bond donating sites, 19 but also for characterizing interaction sites at Lewis acids, such as BCl3 and BH3. 21 All DFT and ab initio computations have been performed using the Gaussian16 suite of programs. 22 The computations of ĪS(r) and VS(r) have been performed using the Hs95 program 19 of Tore Brinck and the ELF computations using the topchem2 program. 23 ## Lewis acidities To improve the understanding of the chemical mechanism and chemical requirements for selective binding of N2 and CO, we have listed computed properties for a series of boron-based Lewis acids and their complexes with N2, CO, NH3 and Fin Table 1. The complexation enthalpy of Fis included as a fluoride ion affinity has been used a general descriptor for assessing Lewis acidity. 24 For comparison, we have also included some related Lewis acids with Al or Ga instead of B as the coordinating atom. First of all, we note that the suspected p-backbonding behavior that we found in complexes of B(SiH3)3 with N2 and CO seems to be confined to compounds where a central boron atom is bonded to Si or Ge. Only in the B(SiR3)3 and the B(GeR3)3 compounds do we observe the short intramolecular B-N or B-C bonds and the enhanced binding strengths for N2 and CO when compared to binding strengths for NH3 and F -. Substituting B for Al, or Si for C, H or Cl increases the bond length and complexation enthalpies. In particular, B(CH3)3 or BCl3 do not form a stable complex with N2. .8 a Surface electrostatic potential maximum (VS,max) above the B or A atom. b VS,max of the Lewis acid in the geometry of the complex with N2. c Surface minimum of the average local ionization energy (ĪS,min) in the region above the B-Si bond. Geometry of Lewis acid from its complex with N2. d Intermolecular B-N (or Al-N) bond distance for the complex with N2. e Forms no stable complex with N2. f VS,max or ĪS,min for the geometry of the Lewis acid in its complex with NH3. g No ĪS,min above the B-H bond. ĪS(r) value above the B-H bond. ## Orbital analysis of p-backbonding It is not obvious why a compound of the type B(SiR3)3 or B(GeR3)3 should participate in pbackbonding as there are no occupied lone-pair p-orbitals on B in these compounds that can interact with the p*-orbitals of N2 or CO. However, an analysis of the occupied orbitals of B(SiH3) in a distorted pyramidal geometry corresponding to that in B(SiH3)3•N2, shows that the two degenerate orbitals 18 and 19 as well as the degenerate HOMOs have the proper symmetry for the orbitals to interact with the p and p* orbitals of N2 or CO. The first type of orbitals (18,19) has very little density on the B and thus by itself will overlap only weakly with the ptype orbitals. On the other the orbital energy of 18 and 19 (-16.5 eV) is similar to the p orbitals (-14.8 eV) and should favor a constructive interaction. The second type of orbitals, the HOMOs (27,28), has a shape and extension that should enable a good overlap with the p and p* orbitals. On the other hand, the HOMO energy (-8.8 eV) is intermediate between the p energy (-14.8 eV) and the p* energy (0.8 eV), and thus we do not expect a simple interaction with either of these. Overall, the interaction of the two sets of degenerate orbitals of B(SiH3)3 with the p and p* orbitals of N2 will generate four set of degenerate orbitals, eight in total, where the six lowest energy orbitals are likely to be occupied. Investigating the occupied orbitals of B(SiH3)3•N2, we indeed find three sets of degenerate orbitals that are occupied and consistent with such an interaction, as shown in Fig. 2 The first two of these orbitals, i.e. number 22 and 23 with an energy of -16.9 eV, are strongly p bonding between B and N. These orbitals have large contributions from 18 or 19 on B(SiH3)3 and the p orbitals on N2, but also a smaller contribution from one of the HOMOs that enhances the B-N p bonding. The orbital energy (-16.9 eV) is slightly lower than orbital energy (-16.5 eV) of 18 and 19. The second type of orbital (25, 26) have contributions from the same orbitals as in 22 and 23, but is clearly non-bonding between B and N and the orbital energy (-15.9 eV) is slightly higher. Thus, orbitals 22,23 together with 25,26 will result in a significant p bonding contribution to the B-N interaction. Fig. 2 The top row to the left shows the orbitals of B(SiH3)3 that have the correct symmetry for interaction with the p and p* orbitals of N2. The bottom row shows the orbitals of B(SiH3)3•N2 that are formed due to these interactions. Note that orbitals 22,23 contribute strongly and 34,35 weakly to the p-bonding of the complex. Orbitals 25,26 are non-bonding in this respect. The third set of orbitals is the HOMOs, i.e. number 34 and 35 with an energy of -8.8 eV. These orbitals have a very similar shape and energy as the HOMOs of bare B(SiH3)3, but the orbitals of the complex has an additional contribution that can be expressed as a linear combination of the N2 p and p* orbitals, or more exactly as a very small in phase contribution from p(p) on N(1) in the B-N(1)-N(2) sequence and a somewhat larger out of phase p(p) on N(2). At first glance, it may seem the HOMOs of B(SiH3)3•N2 are nonbonding with respect to B-N but a closer inspection shows that they have a slight bonding character due to the shape of the contributing HOMO of B(SiH3)3, which extends over the B-N bond region, and the sign (in phase) of the p(p) on N(1). The change in sign of the wavefunction between N atoms also means that the orbital weakens the N-N p bond and the larger contribution from p(p) on N(2) shifts electron density towards N(2). The fourth set of degenerate orbitals is the LUMOs of B(SiH3)•N2, they are essentially identical in shape and energy to the p* orbitals of N2 and only have a minor contribution from the B(SiH3) orbitals. These orbitals may be important for photochemical activation of B(SiH3)•N2 or could become partly occupied upon reduction of the complex. Summarizing our findings we note that the B-N p-bonding characters of orbitals 22,23,34 and 35 are strengthened by a reduction of the B-N distance thereby explaining the very short B-N distance in B(SiH3)•N2. For comparison, we have analyzed the occupied orbitals of the B(CF3)3•N2 complex, and in this complex there is no orbital that has a significant p-bonding character between the B and N. As indicated by the orbitals of Fig. 2 the p-bonding interaction in B(SiH3)3•N2 is different from the classical picture of p-backbonding as a donation of electrons from occupied d-orbitals or p-orbitals into the antibonding p*-orbitals of N2. However, as discussed by Pettersson and Nilsson, the classical picture is simplified and it is not representative for the pbackbonding of transition metal surfaces with N2 or CO. 25 Instead the orbitals responsible for the p bonding to the ligand on those surfaces resemble the orbitals of Fig. 2 that provides the p bonding in B(SiH3)3•N2, i.e. it is orbitals similar to 22 and 23 of B(SiH3)3•N2. It is important to remember that all orbitals have to be orthogonal; this restricts their potential shapes and symmetries, and the classical picture of p-backbonding is not consistent with the orthogonality requirement. Caution should be taken when interpreting the bonding between certain atoms in a molecule based on a few orbitals, as the canonical orbitals typically are complex and delocalized; it is the combination of all occupied orbitals that gives the total electron density and determines the bonding in the molecule. To estimate the capacity for p-bonding interaction with the N2 porbitals, we argue that the surface average local ionization energy [ĪS(r)] is a better descriptor than the energy of any individual orbital of the Lewis acid, e.g. the HOMO, as ĪS(r) can be defined as a functional of the total electron density and is invariant to orbital rotation. The positions with the lowest ĪS(r), the ĪS,min, are the position from which electrons are most easily removed or donated, and the values of the ĪS,min are indicative of the average electron binding energy at those positions. As shown in Fig. 3, there is a ring shaped region of low ĪS(r) above the B-Si bond region in B(Si(CH3)3)3 where there is a p-bonding interaction with N2 in B(Si(CH3)3)3•N2. Similar ĪS,min regions are found in all the compounds of the types B(SiR3)3 and B(GeR3)3 that form short and strong p-type bonds with N2 and CO, as well as in B(CH3)3, which forms a p-type bond with CO but not N2. The ĪS,min value increases with the substituent on B in the order, Ge(CH3)3 > Si(CH3)3 > GeH3 > Si(OH)3 > SiH3 ≈ Si(SiH3)3 ≫SiF3 which seems to reflect the p-bond donating capacity, as the B-N bond length increases in approximately the same order, with B(Ge(CH3)3)3•N2 having the shortest bond (1.46 ) and the B(SiF3)3•N2 the longest bond (1.51 ). Inductive donors, such as CH3, are expected to donate electrons into the B-Si (or B-Ge) bond and strengthen the p-bond with N2, whereas inductive acceptors, particularly F, withdraw electron from the B-Si or B-Ge bond and weaken the B-N p-bond. It is important to remember that there is both a s and p contribution to the bonding of N2 and CO, and that inductive acceptors strengthens the s-bond. Thus, there is no simple correlation between the strength of the p-bond, as indicated by the B-N bond length or ĪS,min, and the complexation enthalpy. When it comes to B(CH3)3, it does not bind N2 and binds CO only weakly; this is partly a consequence of a weaker s-bond than in B(SiH3)3, but primarily due to a much weaker p-bond as indicated by a higher ĪS,min, 12.89 eV as compared to 11.53 eV for B(SiH3)3. BH3 binds N2, but weakly, due to a much stronger s-bond, as indicated by highly positive VS,max (vide infra), and a relatively low ĪS(r) (13.0 eV) in the p-bonding region of the Lewis acid. ## Density difference maps To better understand the bonding interactions of B(SiH3)3, we have computed the density difference (DD) maps for the complexes with N2, NH3 and F -, see Fig. 4. For comparison, we have also included the DD for the B(CF3)3•N2 complex. The overall shapes of the DD are all rather similar, with a buildup of electron density above and below the B nucleus with a shape that is intermediate between a p and a sp 3 orbital. Thus, in all the complexes there is an accumulation of electron density in the B-N bond region; for the DD of B(SiH3)3•N2, the upper part is wider than in the other complexes consistent with a partial B-N p-bond. On the other hand, the donut shaped depletion of electron density above the B-Si bond region should not be seen as the result of donation of electron density into the p-bond, as the corresponding depletion is slightly bigger for B(SiH3)3•NH3 and much bigger for B(SiH3)3•F -. Instead we interpret this depletion as the results of a polarization of electron density from the B-Si bond region towards the region below the B resulting from the interaction with the lone pair of N2. The depletion in B(SiH3)3•Fis bigger and more diffuse, as Fcarries a full negative charge and the distribution of negative charge is not as localized as in the lone pairs of N2 and NH3. An important observation from the DD of B(SiH3)3•N2 is that there is a density depletion in the N-N bond region; this indicates that the B(SiR3)3 compounds not only binds N2 strongly, but also weakens the N-N bond. The weakening of the N-N bond together with the buildup of p density at the outer nitrogen (N(2)) activates the molecule for chemical transformation. Comparing the DDs of the B(SiH3)3•N2 and B(CF3)3•N2 and reveals interesting information about the difference in bonding and reductive activation between these complexes. The overall pictures are rather similar, but there is much smaller depletion in the B-C bond region of the latter compared to that in the B-Si region of B(SiH3)3•N2; this difference is consistent with the B-Si bond density being more diffuse and polarizable and with the electron withdrawing effect of the CF3 group. There is also a considerable difference between the N2 pregions of the two complexes. In both complexes there is a buildup of p-density at N(1) ,the N closest to the B, which can be viewed as the result of a polarization of N2 p-density due to the high positive electrostatic potential on B. In B(SiH3)3•N2, there is additionally a buildup of electron density at the N(2), and we interpret it as the result of the p-bonding character of the interaction and the contribution to the density from the HOMOs, which have a significant p(p) contribution at N(2) (see Fig 2). As already indicated, this build up may be important for the catalytic activation of N2. ## Electrostatic contribution to s-bonding Recent studies have shown that even strong donor-acceptor interactions, such as halogen and hydrogen bonds involving charged soft Lewis bases (e.g. Br -) , that traditionally are considered to have a significant charge transfer contribution often can be characterized and quantified by only considering electrostatics and polarization. 26,27 Here we will analyze the variation in sbond strength among the different Lewis acids and bases and argue that an electrostatic model can provide at least a semi-quantitative agreement. Beginning with the B-Si compounds it can be noted that their high Lewis acidities to a significant extent can be traced to a high surface electrostatic potential [VS(r)] at the B, i.e. a high VS,max value, see Fig. 3. The high VS,max is not surprising considering that the B-Si Lewis bases are electron deficient with an empty p-orbital as the LUMO. The value of the VS,max becomes further increased when the Lewis acid is distorted to the pyramidal geometry present in the complexes. In fact, the VS,max value in many cases exceed 50 kcal/mol, which is much higher than the values typically found in neutral molecules except for the acidic hydrogens of strong hydrogen bond donors. The Lewis acidity does not follow the VS,max value strictly; as an example N2 binds stronger to B(Si(CH3)3)3 than to B(SiH3)3 despite B(Si(CH3)3)3 having a lower VS,max. This behavior can be traced to the combination of a stronger p-bond and a stronger polarization in the N2 interaction with B(Si(CH3)3)3, i.e. the importance of polarization in B(Si(CH3)3)3•N2 is enhanced because of the very short B-N distance due to the p-bond interaction and the higher polarizability of CH3 compared to H. The introduction of strongly electron-withdrawing substituents, such as CF3, SiF3 and CN, substantially increases the VS,max value, and thus enhances the Lewis acidity. However, this effect is more important for the interactions with NH3 and Fthan for interactions with N2 and CO, as the electron-withdrawing substituents weaken the p-bond interaction. This is particularly evident for B(CN)3, which is one of the strongest binders of NH3 and F -, but does not form stable complexes with N2 and CO. The effect may be enhanced by the resonance withdrawing capacity of CN as CF3 and SiF3 are inductive electron acceptors. To obtain a Lewis acid that preferentially binds N2 and CO, substituents that donates electron density into the B-Si or B-Ge bond are preferred as these promote the formation of a p-bond with N2 and CO. Somewhat surprisingly, we find that the detrimental effect of electron withdrawing substituents is bigger for CO binding compared to N2 binding. Intuitively, this is surprising considering that CO by all measures have a stronger lone pair and thus should have a stronger electrostatic interaction with B. However, the p-bonding interaction seems to be more important for CO compared to N2 and since the electron withdrawing substituents reduce pbonding, this effect takes precedence in the complexes with CO. It is also interesting to note that there are enhanced binding strengths of NH3 to B(SiR3)3 and B(GeR3)3 when compared to traditional boron based Lewis acids, e.g. B(CH3)3, BCl3 and BH3. This is consistent with previous reports that also NH3 can participate in p-backbonding interactions. 28 In comparison, the fluoride affinity follows the VS,max of the Lewis acid more closely indicating an interaction dominated by electrostatics. However, due to small size and negative charge of F -, polarization plays an integral role and explains why B(CH3)3 and BH3 binds Fmore weakly compared to the other Lewis acids. Because of the anomalously large polarization effect and the generally very high binding strength, it can be argued that fluoride affinity is not a representative scale of general Lewis acidity. We have also compared the boron based Lewis acids to some aluminum and gallium based Lewis acids with similar structures. Despite featuring very high VS,max at Al or Ga, these Lewis acids bind N2 and CO only weakly while being relatively intermediate binders of NH3 and strong binders of F -. The strongest Lewis acid of the Al-compounds is Al(SiF3)3, and it has the highest VS,max of all the non-charged Lewis acids that are investigated in this study. Accordingly it has the lowest Fcomplexation enthalpy of the neutral Lewis acids, whereas the N2 complexation enthalpy is relatively modest at -13.0 kcal/mol. We note that AlCl3, in contrast to BCl3, binds N2 with a negative complexation enthalpy, but that the binding strength is reduced going to GaCl3. In this context, it should be noted that all the Al and Ga based Lewis acids remain nearly planar around the central coordinating Al or Ga atom after coordination to N2 and that the bonding distance is larger than the sum of the covalent radii. ## Covalent character of s-bonding In contrast to the p-interactions, which has been analyzed in terms of orbital interactions, the s-bond contribution to the interactions of the boron based Lewis acids bases has so far been rationalized only in terms of electrostatics and polarization. This analysis has provided a mean for explaining the variations in the complexation enthalpy with respect to Lewis bases and the substituents on the Lewis acid. In this context, it is interesting to note that ELF-analysis, which has been shown to be a stringent tool for distinguishing between physical and covalent bonding, shows the B-N bond to be non-covalent. 29 However, as pointed out by several researchers, it is more appropriate to consider a gradual scale between covalent and non-covalent bonding. 21, Politzer et al. argue that the bonding in BCl3•NH3 is of significant coordinative covalent character based on the strength of the interaction, the relatively short B-N bond length, and the pyramidal structure of the BCl3 in the complex. Following similar arguments we suggest that all the stable BR3•N2 and BR3•CO complexes have some covalent character to the intermolecular s-bonding, but that the covalent character in many cases is relatively weak considering the significant p-contribution to the binding and the relatively low binding strength. In contrast, the Al and Ga based Lewis acids form complexes with N2 that have relatively long Al-N or Ga-N bonds and tetragonal structure. This indicates non-significant p-bonding and a covalent character of the s-bonding that is much lower than for the B compounds. Interestingly, we find that even the carbocation C(CH3)3 + binds N2 weekly with a complexation enthalpy of -2.2 kcal/mol and with a B-N distance as long as 2.97 . This observation confirms that the N2 binding of the boron based Lewis acids cannot be rationalized by electrostatic considerations alone. ## Realizing the B(Si-)3 and B(Ge-)3 bonding motifs The B(SiR3)3 and B(GeR3)3 compounds are promising candidates for N2 and CO activation because of their strong and selective binding of these Lewis bases. In addition, the p bonding mechanism is likely to provide catalytic activation for chemical transformation, including reduction. However, none of the Lewis acids in this category in Table 1 has yet been synthesized. To our knowledge, B(SiPh3)3 is the only molecule of this type that has been prepared. 34 We have made some preliminary calculations on this molecule and found it to bind N2 relatively weakly with a complexation enthalpy of around -9 kcal/mol. The poor binding seems to be the consequence of a combination of electronic and steric crowding, and additionally there may be a kinetic barrier for binding due to a steric shielding of the B atom by the phenyl groups in the free Lewis acid. We hypothesize that it may be easier to prepare heterogenous catalysts with the favorable B(Si-)3 or B(Ge-)3 bonding motif. Solid silicon and germanium have been prepared in the forms of crystals, 2-D materials and nanoparticles. This type of materials is commonly doped with boron to obtain semi-conductors. In particular for nanoparticles, it has been shown that the boron atoms accumulate at the surface, and a similar behavior is expected for larger particles and crystals. 35,36 B-doped silicon nanoparticles have also been shown to be resistant against oxidation in air, which is an import property if they are to be used as electrocatalyst. In Fig. 5 we show the DFT-PBE optimized structure of boron substituted and hydrogenated silicene, which is the silicon analog of graphene. As seen from the figure, this material has the advantage that the B(Si-)3 unit has a pyramidal geometry already before binding N2, and thus is preorganized to bind N2. The B-doped H-silicene is a relatively strong N2 binder and has a similar B-N bond length and N2 binding energy as B(SiH3)3 when computed using the PBE functional. Assuming that the error due to the functional is similar for both compounds, we predict a N2 binding enthalpy close to -16 kcal/mol for the B-doped H-silicene. Preliminary calculations of the distal pathway for electroreduction to ammonia, indicate that the first reductive step is rate-determining with a limiting potential close to 1.5 V. This is an encouraging result, but shows that further structural and chemical optimization will be needed to afford selective and efficient catalysts. ## Conclusions Lewis acids of the types B(SiR3)3 and B(GeR3)3 are found to bind N2 and CO with anomalously short and strong B-N or B-C bonds. The very short B-N bond in the complexes with N2 is particularly remarkable considering that N2 is a very weak Lewis acid. This selective binding enhancement is attributed to p-backbonding based on an analysis of the occupied orbitals in the complexes with N2, and an analysis of the density differences associated with the formation the complexes. However, the classical picture of p-backbonding as a donation into the unoccupied p*-orbitals of N2 is found to be a simplification. The p-bonding is a consequence of constructive orbital interactions between the diffuse and highly polarizable B-Si and B-Ge bond regions and the p-regions of N2. The B-Si and B-Ge bond regions are characterized by a ring shaped region of low ĪS(r). The value of the ĪS,min in these regions reflects the p-bond strength in the complexes, as ĪS,min follows the order of the B-N bond length, i.e. the shorter the bond, the lower the ĪS,min. The p-backbonding interaction is expected to activate the N2 unit for chemical transformation and reduction, as it decreases the electron density and increases the length of the N-N bond. The binding of N2 and CO by the B(SiR3)3 and B(GeR3)3 Lewis acids also has a strong s-bond contribution. The relatively high s-bond strength is connected to the high positive surface electrostatic potential [VS(r)] above the B atom, the boron VS,max. The magnitude of the VS,max is further increased when the B-Si coordination becomes pyramidal upon interaction. Introduction of electron withdrawing R-substituents increases the VS,max value and thereby the s-bond strength, but also leads to a higher ĪS,min and reduced p-backbonding strength. Thus, such substituents increase the general Lewis basicity, but will favor the binding of regular Lewis acid such as NH3 and Fmore strongly than binding N2 and O2. Another observation is that the boron based Lewis acids in contrast to Al-based Lewis acids generally have a significant covalent contribution to the s-bonding, which is indicated by intermolecular B-X bonds that are significantly shorter than the sum of the van der Waals radii and pyramidal geometries around the central B atom in the complexes. ## N2 reduction reaction (NRR) Our computational results for the B(SiR3)3 and B(GeR3)3 Lewis acids indicate that these types of molecules have the potential to catalyze nitrogen reduction reactions. Unfortunately, they are highly reactive and difficult to synthesize. It may be easier to prepare heterogenous catalysts with the wanted B(Si-)3 or B(Ge-)3 bonding motif. Boron doped crystals, 2-D materials and nanoparticles may be prepared by regular synthesis techniques used for preparation of semiconductor materials. We have shown that such materials will have the B(Si-)3 unit in a favorable bonding geometry for N2 ligation. Preliminary calculations of electrochemical reduction of boron-substituted and hydrogenated silicene indicate potential for efficient catalysis but shows that further studies and optimization of chemical composition and nanostructure are needed.

chemsum

{"title": "Anomalous p-backbonding in Complexes between B(SiR3)3 and N2: Catalytic Activation and the Breaking of Scaling Relations", "journal": "ChemRxiv"}

boronic_acids_for_functionalisation_of_commercial_multi-layer_graphitic_material_as_an_alternative_t

## Abstract: A novel radical-based functionalisation strategy for the synthesis of functionalised commercially obtained plasma-synthesised multi-layer graphitic material (MLG) is presented herein. 4-(trifluoromethyl)phenyl boronic acid was utilised as a source of 4-(trifluoromethyl)phenyl radicals to covalently graft upon the graphitic surface of MLG. Such a methodology provides a convenient and safer route towards aryl radical generation, serving as a potential alternative to hazardous diazonium salt precusors. The structure and morphology of the functionalised MLG ( Ar f-MLG) has been characterised using XPS, Raman, TGA, XRD, SEM, TEM and BET techniques. The XPS quantitative data and Raman spectra provide evidence of successful covalent attachment of 4-(trifluoromethyl)phenyl groups to MLG. ## Introduction Graphene describes an allotropic form of carbon consisting of a single layer hexagonal array of sp 2 hybridised carbon atoms. The combination of stacked graphene planes makes up the well-known structure of graphite. Whilst graphite has been known for several hundred years, graphene itself was not isolated until 2004. 1 Graphene possesses excellent mechanical, electrical, optical, thermal and biocompatible properties. A whole host of potential applications for these materials has been touted by both academia and industry. 2,3 Although research on this material is still in its early stages of development, plenty of applications have been proposed which harness and exploit the aforementioned properties. Notwithstanding all of the aforementioned properties of these materials, there are some significant challenges which hinder the practical application of graphene materials on any large or industrial scale. 4 Graphene is a challenging material to work with due to its limited dispersibility within most solvents. Additionally, characterisation challenges are very common. 5 Pristine graphene is only available in small laboratory scale quantities whilst larger scale industrially derived graphene contains many defects, oxygen functionalisation and are typically obtained as multi-layer graphitic material (MLG). 6 Furthermore it is typically obtained with a broad distribution of sizes and morphologies. Issues such as processability can be addressed via functionalisation. Functionalisation can also alter the properties of the material in a number of ways. For example, addition of new functional groups can enhance solubility and/or generate new composite materials with specifically designed properties. 7,8 The incorporation of additional organic or inorganic moieties to the surface of such materials also creates steric repulsion between the sheets and stacks, thus providing an energy barrier against undesired clumping. 9 Furthermore, covalent functionalisation can also alter the graphitic structure by introducing a band gap, thereby allowing the material to switch to lower conductance states. 10 It is thus imperative to obtain new, safe routes to introduce covalent functionality to new materials. 7,8 Over the 16 years since graphene was first isolated, a vast amount of research has been devoted to the development of methods towards covalently functionalised graphene. Many of these methodologies utilise hazardous conditions, such as diazonium salts. From a synthetic point of view, diazonium salts provide an excellent and straightforward source of radicals from which the only by-product is nitrogen gas. 19 This offers a clean methodology for attaching functional groups to the material. Such radicals combine with radicals present on the surface to form new covalent bonds as demonstrated by a number of groups. Their application, however, does have some undesirable synthetic conditions which makes it challenging for large scale industrial production. Many diazonium salts decompose violently and can be sensitive to friction and shock. 28 As a result, access to radicals through alternative precursors is desirable for the purpose of large-scale functionalisation of graphene and graphitic material. Through an industrial collaboration, we embarked on an investigation to explore alternative starting materials which allow access to aryl radicals required for covalent functionalisation. From an industrial perspective, plasma-synthesised multilayer graphitic material 29,30 is more accessible and as such there is significant value in finding synthetic methodologies for its derivatisation. As such we aimed to explore these more accessible materials for functionalisation as opposed to graphene. Herein, we wish to report a preliminary account of these investigations. We show that 4-(trifluoromethyl)phenyl radicals, accessed via oxidative conditions of 4-(trifluoromethyl)phenyl boronic acid, can be used to covalently functionalise the outer surfaces of plasma-synthesised MLG, thus forming 4-(trifluoromethyl)phenyl functionalised MLG ( Ar f-MLG). We also explore the impact of carrying out this functionalisation by examining the changes in relation to the original material. The functionalised material was characterised using XPS, Raman, TGA, XRD, SEM, TEM, and BET spectroscopic and analytical methods. ## Synthesis of Ar f-MLG As a means of exploring alternatives to generate aryl radicals in situ for the purposes of covalently functionalising MLG, we looked at aryl boronic acids as potential sources. Aryl boronic acids are well known for their applications in cross coupling reactions, leading to the formation of C-C bonds. It has previously been demonstrated that aryl radicals can be generated from aryl boronic acids under various oxidative conditions, as an alternative strategy for forming C-C bonds. For example, oxidants and combinations of oxidants such as Mn(OAc) 3 , 31 AgNO 3 / K 2 S 2 O 8 , 32,33 have all been employed for these purposes. Baran recently reported the application of a mixture of silver nitrate and potassium persulfate for the formation of C-C bonds. 32,33 It has been postulated that the persulfate is reduced by the silver to form both [SO 4 ] 2 and [SO 4 ] , which in turn react with the aryl boronic acid to generate aryl radicals (Scheme 1). 32 The silver(II) is then reduced back to silver(I) as shown. The transformations proceed within a biphasic aqueous/DCM solvent system. We utilised the Baran protocol for generating aryl radicals in the presence of MLG as a means of functionalising this material. For the purposes of aiding the subsequent characterisation, we chose 4-(trifluoromethyl)phenyl boronic acid as the aryl radical source. This precursor was chosen since it was reported to be most the efficient of those tested (due to the electron withdrawing nature of the CF 3 group) 32 and it contained fluorine which could be readily observed by X-ray photoelectron spectroscopy (XPS). Prior to the reaction with the graphene material, a control reaction was undertaken. In this control reaction, 4-(trifluoromethyl)phenyl boronic acid was reacted with silver nitrate and potassium persulfate in a 1 : 1 mixture of water and DCM (Scheme 2). We found that 44 h stirring at room temperature was sufficient for complete conversion of the starting material. Two products were separated from the DCM layer following work up. These products were identified as the homo-coupled product 4,4 0 -bis(trifluoromethyl)-1,1 0 -biphenyl (A) and 4,4 0 -bis(trifluoromethyl)-1,1 0 -biphenyl ether (B) by NMR spectroscopy and mass spectrometry, in an approximate 1 : 1 ratio. The former product (A) was expected, whilst the oxygen-bridged compound (B) was not initially anticipated. We believe that this product originates from the radical species reacting with water to form phenol, which then undergoes a C-O coupling reaction to form the ether product. It was important to confirm the identify of these products since they can also form intermolecular interactions with the surface of the material, which can confuse later interpretation and its characterisation. With evidence of radical generation, we proceeded to repeat the same conditions in the presence of MLG with the aim of functionalising this material with the aryl groups (to form Ar f-MLG) with the expectation that both 4,4 0 -bis(trifluoromethyl)-1,1 0 -biphenyl and 4,4 0 -bis(trifluoromethyl)-1,1 0 -biphenyl ether organic side products would also be present within the reaction mixture (Scheme 3). After 44 h, the newly formed Ar f-MLG was isolated from the reaction mixture. This involved purification steps to ensure that organic and inorganic by-products were mostly removed from aryl functionalised MLG ( Ar f-MLG). The reaction mixture was centrifuged and the resulting powder was subjected to repeated dispersion/centrifugation cycles using large volumes of water, acetonitrile and DCM. The remaining Scheme 1 Reported mechanism for the generation of aryl radicals using the approach by Baran and co-workers. 32 Scheme 2 Reaction of 4-(trifluoromethyl)phenyl boronic acid with silver nitrate and potassium persulfate in a biphasic water and DCM reaction. solid material was then placed under high vacuum (10 6 bar) for 168 h. ## Analysis and characterisation of Ar f-MLG The new functionalised material, Ar f-MLG, was analysed and characterised via a range of techniques in comparison with the original MLG sample. This allowed us to examine the degree of functionalisation and its impact on the graphitic structure. The results acquired from these techniques are outlined below. ## X-ray photoelectron spectroscopy Both original and functionalised samples were examined by XPS, the results of which are presented in Fig. 1-4 and Table 1. Additional details and spectra are provided in the ESI, † in Fig. S1. We initially characterised MLG and subsequently Ar f-MLG to identify and compare surface characteristics, as well as the changes to its elemental composition. The data confirms changes to the elemental composition of the material upon functionalisation (Table 1 and Fig. 1). Most notable is the incorporation of fluorine into Ar f-MLG with a relative atomic concentration of 3.5% (at%) for the F 1s orbital. Deconvolution of the high-resolution XPS spectrum of C 1s spectrum (Fig. 2a) shows that MLG consists of seven environments for the carbon atoms. These were as follows with the binding energies within brackets: C sp 2 (284.5 eV), p-p* (290.9 and 294.0 eV), CQO (288.3 eV), C-O (286.6 eV), C sp 2 (284.8 eV) and O-CQO (289.6 eV). A breakdown of the elemental compositions of these components is shown in Table 1. Carbons represent 94.9 at% on the total elemental composition at the surface as expected from graphitic material. The corresponding C 1s spectra for Ar f-MLG (Fig. 2b), consists of nine components corresponding to C sp 2 (284.5 eV), p-p* (290.9 and 294.0 eV), CQO (287.7 eV), C-O (286.4 eV), C sp 2 (284.8 eV), O-CQO (289.0 eV), C-F x (284.9 eV) and CF 3 (292.7 eV). Whilst both show graphitic character, Ar f-MLG shows additional C-F functionality, originating from the reaction of 4-(trifluoromethyl)phenyl radicals with the graphitic material. Overall, there is a reduction in the percentage carbon in this sample. This is consistent with functionalisation. Furthermore, there is a large increase in the concentration of sp 3 hybridised carbon centres, with respect to sp 2 hybridised carbons, from 8.8 at% up to 15.8 at%. Some of this increase has be attributed to oxidation of the MLG sample as a result of the oxidising conditions of the reaction (vide infra). As outlined below however, the level of increase in the oxygen content does not account for such a large increase in sp 3 hybridised centres. These observations are therefore consistent with some degree of the covalent attachment of the aryl functional groups to the conjugated sp 2 network to form Ar f-MLG. The O 1s spectrum for MLG (Fig. 3a) consists of five components CQO (531.7 eV), O 1s C-O-C (533.1 eV) and O 1s satellite structures (535.1 eV, 536.9 and 538.9 eV). This confirms the presence of a significant degree of oxygen functionality already within the starting material which originates from its plasma processing. The presence of these oxygen functionalities in MLG corresponds to 4.8 at%. The O 1s spectra for Ar f-MLG looks similar to MLG (Fig. 3b), displaying the same components corresponding to CQO (531.7 eV), C-O-C (533.1 eV) and O 1s satellite structures (535.1, 536.9 and 538.9 eV), with similar respective ratios. Upon functionalisation however, the oxygen content almost doubled, increasing from 4.8 to 8.5 at%. This increase in oxygen content is attributed to the oxidising reagent which is required for radical generation. In addition to generating the 4-(trifluoromethyl)phenyl radical it also increased the level of oxidation of the graphene surface to some degree. In order to confirm this, we carried out a control reaction under the same conditions with only K 2 S 2 O 8 and a sample of MLG. We refer to this new material as c Ox-MLG. Its surface elemental composition as determined by XPS is also presented in Table 1 for comparison. In this case, a large increase in the oxygen concentration was indeed observed in the sample up to 10.7 at%. Even though the percentage of sp 3 carbon centres in Ar f-MLG is higher than c Ox-MLG, the increase in oxygen content is much lower (cf. 8.5 at% vs. 10.7 at%). This is consistent with the fact that the additional sp 3 centres originate from aryl group functionalisation. Furthermore, an increase in oxygen content may also be explained by the fact that a portion of 4-(trifluoromethyl)phenoxide [OC 6 H 4 (CF 3 )] groups could also be covalently bonded to MLG in addition to C 6 H 4 (CF 3 ). The ether species (B) was of course found in the control reaction in the absence of MLG (where both A and B were formed). Fig. 4 highlights the XPS spectra at higher binding energies, indicating the presence of other functional groups. Upon functionalisation, a peak at 687.9 eV (Fig. 4a) confirms the presence of fluorine in the Ar f-MLG material. This is not present within the starting material (see Fig. 1a and Table 1). This binding energy is consistent with the presence of the 4-(trifluoromethyl)phenyl functional group. 36 Alongside this evidence of incorporation of this functional group are changes to the elemental compositions of carbon and oxygen. The carbon content reduced to 86.5 at%, as expected. Furthermore, the scenario of physisorption of the starting material (trifluoromethylphenyl boronic acid) on to the MLG surface can be ruled out due to the absence of any boron content in the XPS data. The material was washed multiple times and the corresponding filtrates were monitored by 19 F and 1 H NMR spectroscopy to ensure that the material was free of any unreacted or residual organic species. The presence of silver (3d orbitals) was observed in the spectrum for Ar f-MLG at 367.0 and 372.8 eV. The former Ag 3d 5/2 peak was further deconvoluted revealing signals at 367.9 eV and 366.9 eV (Fig. 4b). These resemble silver salts in the form of AgCl/Ag and Ag 2 SO 4 , respectively. 37,38 This is likely to be due to silver salts retained within the material. These proved rather challenging to remove and they appeared to be trapped within the material despite multiple washings steps. Furthermore, both Ar f-MLG and c Ox-MLG materials revealed a small percentage of sulfur incorporation as a result of the oxidising agent. Again, these proved challenging to remove completely. Some nitrogen moieties are present within both MLG and Ar f-MLG at relatively consistent compositions (see Fig. S4, ESI † and Table 1). The retention of these impurities within Ar f-MLG results from the inherent nature of this plasmasynthesied graphitic material where the silver, for example, can get trapped within pores and defects in the material (vide infra). ## Determining the decomposition characteristics by TGA Thermogravimetric analysis (TGA) enables us to examine how the materials decompose, which can be very useful in confirming the nature and degree of functionalisation within the material. Accordingly, the thermal stability of both MLG and Ar f-MLG was investigated using TGA measurements at temperatures up to 700 1C. A comparison of the results is depicted in Fig. 5. In both cases, a gradual mass loss is observed with distinct or well-defined mass loss regions. Pristine graphite has been found to show little decomposition until around 600 1C within an O 2 atmosphere and up to 1000 1C within a nitrogen atmosphere. Accordingly, mass loss within both samples at these lower temperatures can be attributed to the decomposition of covalently and non-covalently bonded functionality over a gradual period. 39,40 Initial mass loss (up to 100 1C), is attributed to the removal of water from the surface. Decomposition from this point is then assigned to the removal of covalently bonded oxygen functionality. In the case of graphene oxide, it has been found that the decomposition of covalently bound oxygen functionality takes place above 150 1C. 41 Ar f-MLG shows a lower thermally stability than MLG, decomposing at a faster rate. This is assigned to the increased covalent functionality it possesses, including more oxygen functionality and 4-(trifluoromethyl)phenyl moieties. The presence of silver salts on the surface of Ar f-MLG may also open up other pathways leading to a more rapid mass loss from this material. ## Examining the graphitic structure by Raman spectroscopy Raman spectroscopy was used to examine the graphitic structure and extent of defects within MLG and Ar f-MLG, using a wavelength of 514 nm. Raman spectra for the functionalised and unfunctionalised material are depicted in Fig. 6. The spectra for the control material, c Ox-MLG (K 2 S 2 O 8 only) is presented in Fig. S3 (ESI †). A comparison of the spectra for MLG and Ar f-MLG revealed some changes to the structure of the two materials. Both contained the characteristic peaks for graphitic materials, most notably the G band which appeared at 1576.6 cm 1 for MLG and 1580.3 cm 1 for Ar f-MLG. This doubly degenerate E 2g band confirms the sp 2 carbon network within the materials. 10,42,43 The presence of strong D bands indicates the low crystallinity of the materials and also a high degree of sp 3 carbon centres. This is expected for plasma-synthesised MLG's. The level of defects is revealed by the D bands, which occur at 1344.6 cm 1 (for MLG) and 1350.9 cm 1 (for Ar f-MLG). The D band represents the breathing modes with A 1g symmetry involving phonons near the K zone boundaries. 44 The 2D bands at 2704.0 cm 1 (MLG) and 2708.0 cm 1 ( Ar f-MLG) correspond to the second order symmetry allowed overtone of the D band. Within both spectra, the 2D bands are broad and heavily upshifted in respect to that of single layer graphene, suggesting the presence of multiple layered stacks. This is further confirmed by examining the relative intensity of the G band compared to the 2D band, thus providing I 2D /I G ratios. 45 I 2D /I G ratios lower than one are indicative of multi-layered structures. 46 The ratios for MLG and Ar f-MLG are 0.44 and 0.47, respectively. Additional defect-induced bands, which are more intense in the spectrum for Ar f-MLG, are observed at 1618 cm 1 and 2930 cm 1 . 47,48 The higher level of defects within the MLG material most likely originates from the plasma processing during the synthesis. This can lead to an increased number of covalently attached oxygen functionalities at the outer surfaces of the sheets. Addition of these oxygen groups to the sp 2 3 sites, and thus, this confirms an increase in the covalent functionalisation in comparison to the original MLG material. This increase is likely to originate from the desired 4-(trifluoromethyl)phenyl incorporation and additional oxygen functionality. As highlighted above, the XPS analysis indeed confirmed a significant increase in oxygen functionality which was attributed to the oxidising agent K 2 S 2 O 8 . In order to confirm this, the Raman spectrum of c Ox-MLG was also examined. This exhibited similar changes to those found in Ar f-MLG, however, to a lesser extent with respect to the starting MLG. For example, the I D /I G and I 2D /I G ratios were found to be 0.56 and 0.40 (cf. with the values 0.65 and 0.47 for Ar f-MLG). This is consistent with the incorporation of 4-(trifluoromethyl)phenyl groups onto the MLG material alongside some increase in the oxidation level of the material. It should be noted that the level of functionalisation is estimated to be lower than some diazonium salt methodologies and thus further optimisation of this new strategy is needed. Nevertheless, both the Raman spectroscopic and XPS data are both consistent with functionalisation of MLG with trifluoromethylphenyl groups. ## Analysis of interlayer spacing using X-ray diffraction data In order to gain more detailed information on the interlayer spacing and orientation of the planes within MLG and Ar f-MLG, an X-ray diffraction (XRD) investigation was carried out. The XRD spectra for both MLG and Ar f-MLG are shown in Fig. 7. The spectra show that both materials contain hexagonal ABAB stacking (2H) and rhombohedral ABCA (3R) stacking. This is consistent with that found in commercially available graphene and graphitic-based materials. 49 Strong diffraction peaks are present within both spectra at 26.61, corresponding to graphitic 2H (002) and 3R (003) planes, with an interlaying spacing of 3.35 . Analysis of the line shape of this signal suggests that for the majority of the materials, the number of graphene layers within both MLG and Ar f-MLG are in the region of 58 to 73 layers (see ESI † for further details). As such, the material could therefore also be described as graphite nanostructures. 50 For MLG, two smaller intensity lines are also observed at 42.591 and 44.561, which correspond to the 2H, (100) and (101) stacking planes. Two further small intensity lines are observed at 43.441 and 46.201, which correspond to the 3R, ( 101) and (012) stacking planes. The presence of this four-lined pattern is more clearly seen in the expanded section of Fig. 7. This provides evidence of the 3R and 2H phases within the structure. For these, a ratio of 49 : 51% (3R : 2H) was determined (see ESI † for details). These are consistent with that observed within various related materials. 49,51 Again, for MLG two additional lines at 54.731 and 77.621 are observed which correspond to graphite 2H (004) and ( 110) planes. Many additional diffraction lines are present within the spectrum of the Ar f-MLG sample, suggesting other chemical species have been incorporated within the structure. The characteristic four-line pattern also appears to be present within this spectrum albeit the 2H (101) and 3R (012) lines are obscured by some new lines. This confirms that there is no significant change to the interlayer spacing upon functionalisation. As highlighted with asterisks (*), the majority of these correspond to the diffraction of the AgCl, Ag and Ag 2 SO 4 impurities incorporated into the material (PDF card numbers: 00-006-0480, 52 01-073-6977 53 and 01-074-1739 54 ). The presence of these silver compounds is in agreement with XPS data presented above. There are no observable lines at 2y angles lower that 26.61 This suggests that largest interlayer spacing value between planes corresponds to the graphitic planes. Therefore, it is most likely that functionalisation has taken place at the outer surfaces of the material. Functionalisation within the internal MLG structure would, of course, result in an increase in the interlayer spacing. The lines corresponding to the 2H and 3R arrangements in both MLG and Ar f-MLG show only minor differences confirming the functionalisation has only a small impact on the crystallinity and interlayer spacing of the MLG structure. Again, this is consistent with functionalisation at the outer positions only. ## Analysis of surface morphology using SEM and TEM Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were used to investigate the surface morphology and lateral dimensions of the graphitic sheets. Four SEM images were taken of a representative example of each material at low and high magnifications (Fig. 8 and 9). The images reveal that both materials consist of multi particulate material as agglomerates which range in diameter between 10-30 mm for MLG and o10 mm in size for Ar f-MLG. The quantity of agglomerates within MLG exceeds that of Ar f-MLG, suggesting that the functionalisation process has led these loosely bound aggregates to unravel. At lower magnifications MLG and Ar f-MLG display a powder like appearance. At increased magnifications, it can be seen that these agglomerates are irregularly arranged consisting of flakes comprised of mostly multi-layered stacks, physically aggregated by p-p interactions. These stacks adopt wavy topography and curled edges. The non-uniform nature of the morphology of these materials is most evident in the higher resolution images. These defects are likely to result from the plasma ablation process during synthesis. The particle size distribution is due to the method of processing of this material. Under the conditions utilised to record the images, it was found that the samples underwent charging. In order to obtain more enhanced images, gold coating was added using a sputter coater. The gold coating can be seen with a scaly appearance, as exemplified in Fig. 10. Additional SEM images of Ar f-MLG are provided in Fig. 11. These highlight some of the features of the materials related to their porosity. As can be seen in the image on the left, there is a macro sized pore (circled) which contains some additional material. We believe that this could be some smaller flakes of Ar f-MLG which become trapped within the in-plane pore, and for this reason are difficult to remove during the purification process. The presence of material which could be trapped silver nanomaterials can also be seen on an image in Fig. S7 (see ESI †). The image on the right of Fig. 11 highlights the slit like channels which are found within the spacing between subsequent stacks. The implications of these features on the surface area and porosity of the MLG and Ar f-MLG materials are discussed in the following section. Selected TEM images of MLG and Ar f-MLG are presented in Fig. 12. These also shown that both materials consist of multiple platelets with significant clumping. These multi-layered structures can be observed by differences in the contrast. Darker areas within the TEM images represent more dense areas within increased number of layers, whilst light areas represent those of less layers. It can be observed by TEM images that the stacks of layers exhibit lateral dimensions between 100-500 nm in diameter. As with the SEM images, they also indicate material which contains a broad range of morphologies across the surface of both materials. When comparing the SEM and TEM images for Ar f-MLG with those of MLG, it can be concluded that there are no visual changes of major significance in the morphology upon functionalisation. Nevertheless, the orientation of stacks in Ar f-MLG appear to be less agglomerated than those within MLG. This causes the individual stacks to appear much clearer within the SEM images. TEM images also reveal that the flakes remain very similar in size and distribution. ## Surface area and porosity analysis Nitrogen adsorption-desorption measurements were used to determine the surface area and pore size distribution of MLG and Ar f-MLG, utilising Branauer-Emmett-Teller (BET) analysis and Barret-Joyner-Halenda (BJH) analysis. 55 The results of these investigations are presented in Fig. 13. It was found that MLG and Ar f-MLG both exhibited Type IIb N 2 adsorptiondesorption isotherms typical of materials composed of platelike particles (Fig. 13a). 56 A Type H3 hysteresis loop is present in each isotherm, indicating the occurrence of capillary condensation within pores of B4 nm in size. 57 Hysteresis of this type is usually associated with plate-like aggregates or adsorbents containing slit-like pores. MLG and Ar f-MLG were found to exhibit BET surface areas corresponding to 380.22 and 192.13 m 2 g 1 , respectively (Table 2). A significant contribution of this area originates from pores between 1.7-300 nm. These pores are mostly present within the space between neighbouring stacks which correspond to 75.5% (MLG) and 87.4% Ar f-MLG of the total surface area. The pore size distribution profile was examined by plotting the pore size distribution as a function of incremental pore volume (Fig. 13b) and incremental pore area using the adsorption branch of the isotherm (Fig. 13c). Overall, the curves reveal similar patterns for both MLG and Ar f-MLG in each plot, suggesting a minor impact to the overall structure upon functionalisation. In parallel, the average pore size for MLG and Ar f-MLG were 9.53 nm and 11.70 nm, respectively. The main difference, however, is a decrease in the area and volume of pores affecting mostly pores with diameters of less than B75 nm. This is an indication that functionalisation causes a significant reduction in the quantity of pores within the mesoporous region. This finding coincides with the significant decrease in BET surface area (almost half of MLG). We suggest that the functionalised sheets re-orientate themselves to adopt curled, scroll-like edges due to enhanced interactions between the functionalised stacks. With such a situation, many of the slit-like pores and channels between neighbouring stacks would become inaccessible to the nitrogen adsorbent molecules. In addition, upon functionalisation, we also observe blockage of pores with smaller aggregates (as seen in Fig. 11). This is can be clearly seen in the SEM image in Fig. 11a which shows a macropore with a width of 700 nm in addition to Fig. 11b where stacks of sheets can be observed. 58 ## General remarks All solvents and reagents were purchased from commercial suppliers and used with no further purification. The MLG material was synthesised and provided by Perpetuus Carbon Technologies. Synthesis of MLG. Natural flake plasma processing of natural graphite was carried out using a custom-made multi-electrode dielectric barrier discharge (DBD) plasma reactor as described elsewhere. 29 Control reaction: reactivity of radicals generated from 4-(trifluoromethyl)phenyl boronic acid in the absence of MLG. This reaction was carried out using similar conditions to that of previous works. 32 The reagent 4-(trifluoromethyl)phenyl boronic acid (0.092 g, 4.83 10 4 mol) was dissolved in a solvent mixture (40 mL) consisting of water and DCM (1 : 1), followed by the addition of K 2 S 2 O 8 (0.266 g, 9.85 10 4 mol) and AgNO 3 (0.027 g, 1.60 10 4 mol). The mixture was allowed to stir for 44 h at room temperature. The biphasic mixture was filtered and the DCM layer was separated from the filtrate using a separating funnel. The DCM layer was then evaporated to dryness to give a yellow oil consisting of two compounds: 4,4 0 -bis(trifluoromethyl)biphenyl (A) and bis((trifluoromethyl)diphenyl)ether (B). Characterisation of A: Control reaction: MLG treated with potassium persulfate ( c Ox-MLG). MLG (0.160 g) and K 2 S 2 O 8 (0.570 g, 2.120 10 3 mol) were dispersed in a solvent mixture (20 mL) consisting of water and DCM (1 : 1). The mixture was allowed to stir for 44 h at room temperature. The reaction mixture was then centrifuged and the resultant c Ox-MLG solid was separated and washed repeatedly with water, acetonitrile, and DCM through several dispersion/centrifugation cycles. The resultant material was dried in vacuum for 1 week. ## Characterisation methods X-Ray photoelectron spectroscopy (XPS) analysis was performed using a Kratos Axis Ultra-DLD photoelectron spectrometer with Thermal gravimetric analysis (TGA) was carried out using a PerkinElmer TGA 4000 instrument. The samples were heated from room temperature up to 900 1C (5 1C min 1 ) under a nitrogen atmosphere (50 mL min 1 ). Raman spectroscopy was performed using a Renishaw inVia confocal Raman microscope equipped with an Ar + visible green laser with an emission wavelength of 514 nm. Spectra were collected in a reflective mode by a high sensitive charge couple device (CCD) detector. Powder X-ray diffraction (XRD) patterns were collected using a Panalytical X'Pert diffractometer with a Cu anode irradiation (l = 1.541 ) operating at 40 kV and 40 mA. Phase identification was performed by matching experimental patterns against entries in the ICDD standard database. Scanning electron microscopy (SEM) images were obtained using a Zeiss Supra 35VP (FEG) SEM instrument. The samples were gold-coated using a sputtering coater to enhance the resolution of the images. Transmission microscopy (TEM) images were obtained using a Jeol 2100 field emission gun (FEG) TEM with a 200 kV power source. The surface area and porosity characteristics of the materials were analysed using a Micromeritics ASAP 2020 physisorption analyser. Samples were degassed under 0.667 Pa for 720 minutes at 150 1C with a heating rate of 10 min 1 . The surface area and pore size distribution were measured at 77 K using Brunauer Emmett Teller (BET) and Barrett Joyner Halenda (BJH) cumulative pore volume methods, respectively. 1 H and 19 F NMR spectra were performed on a Bruker 400 MHz Ascendt 400, which operated at 400 MHz for 1 H nuclei and 376.6 MHz for 19 F nuclei. Chemical shifts are reported in parts per million (ppm). NMR spectra were obtained in CD 3 CN solvent (1.93 ppm) and internal reference for 19 F. Mass spectrometry (MS) was carried out on compounds A and B using a Thermoscientific ISQ Single quad with direct insertion probe and the identity of the compounds were confirmed for the preinstalled library of compounds. ## Conclusions In this article, we provide a preliminary account outlining the successful covalent functionalisation of MLG with 4-(trifluoromethyl)phenyl radicals. This has been achieved by using 4-(trifluoromethyl)phenyl boronic acid as a radical source utilising Baran's protocol. The newly formed material, Ar f-MLG, was found to be decorated at a number of positions at the outer surface as confirmed by a number of spectroscopic and analytical techniques. At this stage of the development, some challenges associated with this functionalisation methodology and the nature of the plasma-synthesised multi-layer graphitic material have been identified. The attachment of 4-(trifluoromethyl)phenyl moieties is accompanied by an increase in oxygen functionality around the outer surfaces of the MLG stacks, as a result of the oxidising conditions. Furthermore, the task of removal of entrapped impurities, particularly silver in this case, will need to be addressed. Nevertheless, early indications suggest that this approach could provide access to aryl radicals in a costeffective and safer alternative to hazardous diazonium salts. As a result, this methodology could offer a novel safer approach to synthesise functionalised MLG materials on a larger scale with potential to be developed industrially. We are currently investigating ways in which this methodology can be optimised for practical application and exploring other derivatives. This may assist further processing and provide enhanced interaction with other materials. The development of new functionalisation strategies on commercially derived graphitic materials, of course, becomes increasingly important applications across materials science. ## Conflicts of interest There are no conflicts to declare.

chemsum

{"title": "Boronic acids for functionalisation of commercial multi-layer graphitic material as an alternative to diazonium salts", "journal": "Royal Society of Chemistry (RSC)"}

dna_microviscosity_converts_ruthenium_polypyridyl_complexes_to_effective_photosensitizers

## Abstract: A unique radiative decay engineering strategy using DNA microviscosity for the generation of ruthenium polypyridyl complex (RPCs) mediated singlet oxygen for selective damage of DNA and killing cancer cells is reported. This investigation also demonstarte the effect of light-driven RPCs on bacterial growth arrest, through DNA nick, and differential localization in cancer and non-cancer cells. Moreover, upon binding with DNA, RPCs experience high local microviscosity, which causes significant enhancement of the excited state lifetime and thus generates singlet oxygen. The visible-light-triggered singlet-oxygen efficiently produce nick in DNA and inhibits bacterial growth. RPCs also localize inside the nucleus of the cancer cell and in the vicinity of the nuclear membrane of non-cancerous cells, confirmed by live-cell confocal microscopy. The results provide a facile platform for the novel antibiotic intended discovery combined with cancer therapy. ## Introduction: In recent years, ruthenium-based mononuclear complexes have been extensively investigated for their versatile chemical tunability, exciting photophysical and photochemical properties and long-lived electronically excited state. The Ruthenium polypyridyl complexes offer access to facile chemical modification which facilitates a range of diverse ligands to co-ordinate, imparting a distinct functionality. Notably, the valency of this transition metal can be fundamentally saturated and hence, exhibits an inertness towards nonspecific targets in the cellular milieu. These complexes have been the choice of molecule as a metallodrug in pharmaceutics for chemotherapeutic applications because it can be tuned for better DNA interaction. Besides, drug discovery and its specific delivery coupled with ruthenium complexes have proved to be an efficient route for targeting various genomic diseases like cancer. Also, these complexes are of great interest in the field of tumor detection and have drawn attention in the development of new generation drugs with in vitro as well as in vivo applications. Analogous to this, both ruthenium complexes and nanoparticles have also been explored for the antibacterial properties and light-induced activities. Although, metal complex induced DNA cleavage and cancer cell cytotoxicity are reported, they suffer from a serious problem of UV light excitation or very low absorbance in visible region. Moreover, they induce apoptosis in cancer cell via DNA damage, however, required prolong treatment time (24 h) with ruthenium complex, which surrogates several unknown hits and hence remains incomprehensible. Even though, the pre-existing therapies have achieved remarkable success; their use has always confined by various side-effects due to non-specific targeting. Photodynamic therapy seems to be a proficient approach as it is devoid of harmful effects with targeted delivery. Herein, we have extensively investigated the spectroscopic properties, single crystal XRD structure, mode of DNA binding and their efficacy as antibacterial and anticancer agents of four RPCs (C1, C2, C3, and C4; Scheme 1a, 1b, Table S1). The synthesis of RPCs has reported earlier. The synthesized RPCs generate ROS via the radiative decay engineering mediated by local DNA microviscosity. Ultimately, this leads to light-induced DNA cleavage, and eventually, results in nicked circular DNA (Scheme 1c). The visible light-mediated activities of these RPCs lead to interference with DNA functionality in vivo and hence cell lysis. Extensive investigation of this class of molecules can be useful for novel broad-spectrum antibiotic and cancer therapy. ## Result and Discussion: The absorption and emission maxima of C4 were measured to be 430 nm and 630 nm, respectively, with considerable overlap of emission and excitation spectra (Figure 1a). Moreover, we titrated the plasmid DNA with different concentrations of C4 and exposed to circularly polarised light. We found a significant change in emission maxima of DNA for the corresponding increasing concentration of C4 (Figure 1b) confirming their interaction. Thus, the luminescence intensity of RPCs upon addition of increasing concentrations of plasmid DNA was also measured. The plot exhibits an increasing trend in luminescence intensity, which further implies the DNA and RPCs interaction (Figure 1c). With this preliminary observation, we investigated the effect of Ethidium bromide (EtBr), a known DNA intercalator, addition to the mixture of C4 and DNA solution. Interestingly, with the gradual addition of EtBr, the luminescence intensity of C4 decreases (Figure 1d). Such decrease illustrates the displacement of C4 from DNA by EtBr due to the competitive binding. Hence, it confirmed the intercalation of C4 on DNA. These studies have been done for all the mentioned RPCs, and the results for C1, C2, and C3 are combined in SI (Figures S1-4). Furthermore, the measured luminescence lifetime RPCs shows a notable enhancement(ca.10 times) upon DNA binding (Figure 2a). Supported by the above results, EtBr displacement causes reduction in their luminescence lifetime (Figure 2b). The lifetime plot shows a regular trend against DNA and EtBr concentration (Figures 2c-d). This enhanced lifetime reveals the increased time upon intercalation of RPCs with DNA, resulting in the nick generation. However, the lifetime of metal complexes is not limited to DNA interaction, rather local viscosity of medium also plays the most important role in the considerable lifetime enhancement. To validate this, we performed experiments using Ethylene Glycol (EG) as a source of viscous medium. The concentrations of EG were varied from 0-100% in water. The results show a high degree of lifetime dependence on viscosity. Clearly, the lifetime for C2 and C4 enhanced upto 10 folds for the final 100% EG concentration (Figures 3a-b). The distribution of lifetime data follows a linear trend against EG concentrations (inset Figures 3ab). The results for C3 are provided in SI (Figure S3c). Further, we checked the binding of RPCs with DNA concerning to major and minor groove binding (Figure 3c). We found that the lifetime of C4 drastically restored to original value upon addition of excess Hoechst, wellknown DNA minor groove binder, to the C4-DNA solution. However, we do not observe any effect of spermine, which is a DNA major-groove binder. Hence, conforming RPCs minorgroove binding. Our previous report on structurally similar RPCs, suggests the generation of ROS by RPCs. Our hypothesis claims the ROS mediated DNA nick generation as a mode of metal complex action. These phenomena successively lead to a series of events when studied in vivo. For this, we monitored the fluorescence of 9,10-Anthracenediyl-bis(methylene)dimalonic acid (ADBM-DMA) upon addition of RPCs and visible light irradiation. As expected, the luminescence intensity gradually decreases with time due to the endoperoxide formation and gets saturated in a time-frame of 15 minutes (Figure 3d). This confirms the generation of singlet-oxygen by the RPCs. However, there is an insignificant effect of visible light on the luminescence intensity of ADBM-DMA without RPCs (Figure S5). To understand the role of excited state lifetime on the singlet oxygen generation, we performed luminescence quenching experiment of C1-2 and C4 with molecular oxygen in H 2 O and D 2 O. As expected, we observed a significant fraction of 3 MLCT state is quenched by molecular oxygen (Figure S6, Table 2). For further clarification, we investigated the effect of designed RPCs induced DNA cleavage activity by the generation of nick in supercoiled (SC) DNA upon photo-irradiation. The ROS generated by metal complexes, upon visible-light irradiation, produces nicked circular (NC) and hence relaxes the DNA. The time-frame of photo-irradiation for optimum nick generation in supercoiled DNA was determined to be 1 h (Figure 4a). Also, RPCs were studied tocheck the extent of their activity on DNA. It is imperative to mention that under the same photoirradiation conditions and molecular concentration; C4 produced the maximum nick (Figure 4a-b). Moreover, effect of different concentrations of the same RPC was also verified for the calculation of optimal as well as maximum nick generation. The concentration of 10 µM (for C4) was found to be sufficient for the conversion of supercoiled DNA into the nicked circular (Figure 4c). However, there were no effects on supercoiling of DNA upon incubation with C4 under the dark conditions which further confirms the significance of photo-irradiation (Figure 4c Motivated by these findings, we extended our investigation in cellulo systems. Keeping the idea of ROS generation and DNA photocleavage, we explored the effect of these RPCs against the gram-negative bacteria, E. coli. These RPCs were incubated with freshly growing bacterial culture, at logarithmic phase, followed by 1h visible-light irradiation. The treated cultures were grown further under the optimal condition, and the optical density (OD) of the bacterial culture was measured after 4.5 hours. Particularly, C4 was found to comprise of most effective bacterial inhibition activity, as indicatedby arrest of OD with the progression of time (Figure 4d). In contrast, there is no growth arrest under the same experimental conditions except the dark instead of photoirradiation (Figure S19). To understand better the optimal concentration of RPCs for bacterial growth inhibition, we titrated different concentrations of RPCs. For C4, 50 µM concentration was found to be most effective against bacterial growth, as it showed significant inhibition. Further, we investigated the growth inhibition of different gram-positive and negative bacteria and found to have a similar effect, although the extent of growth arrest is different (Figure S20). For better insight, we captured the SEM images of bacteria with and without treatment of C4 under dark and light conditions. Interestingly, the bacteria treated with C4 and photo-irradiation have disrupted cell-membrane and distorted morphology (Figure 5c). However, the controls retain normal cell membranes and morphology (Figure 5a-b). Next, we investigated the effect of RPCs in the mammalian cells. We have performed the cellviability assay for both normal and cancer cell lines. The concentration used in this investigation was much lower in comparison to the measured IC 50 value (Figure S21). Interestingly, we found that the C4 mainly localizes to the nucleus of the skin cancer cell line, B16F10, after a short duration of incubation (Figure 5d-f). This bolstered our hypothesis of binding and generating the nick in the DNA and after that inducing the cytotoxic effects. We captured the time-lapse images of B16F10 cells after treatment with C4 and found that it leads to the cell lysis (see movie S1). On the contrary, C4 is mainly distributed in the cytoplasm and near the nuclear membrane of the normal cell lines, CHO (Figure S22). This phenomenon is attributed to the fact that there is perturbed permeability of the cell membranes of cancer cells than normal cells due to differential lipid peroxidation. We further plotted the luminescence intensity distribution of RPCs inside the cells. The intensity profile illustrates localization of RPCs in the inner region of the nucleus for B16F10 (Figure 5g-h) and outside the nuclear membrane for CHO cells (Figure 5i-j). Moreover, the distribution of C4 was further verified in other cell lines with different ROIs of B16F10 and BHK21 (Figure S23-24). ## Conclusions: In summary, we have reported ruthenium-based metal complexes exhibiting versatile photodynamic characteristics. The luminescence lifetime of these compounds increases drastically as they experience enhanced local micro-viscosity upon DNA binding. RPCs generate singlet oxygen due to enhanced lifetime, which further produces visible lightdependent nick in the supercoiled DNA. Moreover, RPCs show antibacterial activities and get localized in the nucleus of cancer cells. Furthermore, the visible light-dependence emission property is appealing as the mode of activation is non-harmful under the physiological context and can be utilized for targeted cancer therapy and antibiotics. Type: -Ecotron OD measurement: -Synergy H1 (BioTek), Version-Gen5 2.07 Multi-mode plate reader ## UV-Visible and fluorescence measurements Shimadzu UV-1800 dual-beam spectrophotometer was used for UV-Vis. measurements and Horiba jobin yvon fluorolog used for fluorescence measurements. ## Circular Dichroism and time-resolved measurements Circular dichroism spectra were recorded on a Jasco J-815 circular-dichrograph using 10 mm quartz cuvette containing 1 ml solution. CD spectra were measured in continuous mode with standard sensitivity (100 mdeg), 0.1 nm data pitch, scanning speed 200 nm/min, response 1.0 sec using TAE buffer (pH 8.5) containing 25 mM MgCl 2 . CD spectra of 5.0 nM DNA were recorded in absence and in presence of increasing concentration of metal complexes. ## Scanning Electron Microscopy The images were captured in Carl Zeiss ultra-plus Scanning Electron Microscope using 5-20 KV HT and 10,000 times magnification. ## Confocal Microscopy The LASER of 488 for nm was used for the RPCs excitation and images were captured using Olympus Fluoview FV 3000 confocal microscope. ## XRD Single crystal X-ray structural studies were performed on a Bruker D8 venture instrument. ## Methods: 2.1 UV-Visible and fluorescence measurements The UV-Vis and fluorescence properties of the Ruthenium polypyridyl complexes (RPCs) were measured in water. The gradual increase in fluorescence intensity of metal complexes with the addition of varying concentration of DNA was monitored. Further, the intercalation of metal complex with DNA was confirmed by titrating the varying concentration of Ethidium bromide (EtBr) with the DNA and metal complex in bound state. ## Time-resolved measurement The instrument response function (IRF) was measured before and after fluorescence lifetime measurement using a dilute suspension of Ludox (from Sigma) colloidal silica. The emission polarizer was positioned at magic angle (54.7º) polarization with respect to excitation polarizer. Exponential fitting function was employed by iterative deconvolution method using supplied software DAS v6.2. The quality of the fitted data was judged from the reduced chi-squared value (χ 2 ), calculated using the IBH software provided with the instrument. ## Singlet oxygen detection experiment The singlet oxygen detection was done using 9,10-Anthracenediyl-bis(methylene)dimalonic acid (ADBM-DMA). 2 µM ADBM-DMA was added to the metal complexes and photoirradiated followed by monitoring of fluorescence emission. As a control, in the simultaneous experiment, the fluorescence emission of only ADBM-DMA after photo-irradiation was monitored to see the effect of light on the photophysical property of ADBM-DMA. ## Viscosity dependent lifetime measurement The lifetime dependence of metal complexes (5 µM) on viscosity of solution was monitored. The lifetime was measured with the varying concentrations of Ethylene Glycol (EG). The range of solvents varies from 100% TAE buffer (pH 8.5) to 100% EG. The change in lifetime was plotted against viscosity. ## DNA photocleavage assays Plasmid DNA was purified from bacterial JM109 cells using FavorPrep Plasmid Extraction Mini Kit. The concentration of DNA was measured at 260 nm and diluted to final concentration of ~50 ng/µl with distilled water. 150 µl of DNA samples from the stock was distributed in different wells of 96 well-plate. Then metal complexes were added in varying concentration of 1 µM, 2 µM, 5 µM and 10 µM followed by photo-irradiation for 1 hour. Photo-irradiated DNA samples were then run into 1% EtBr-agarose gel at 70V for 25 minutes. The bands were then visualized and intensities calculated using syngene tool software. The experiments were done thrice for each metal complex independently. ## Scanning Electron Microscopy C4 (final concentration = 50 µM) was added to the freshly grown aliquots of 1 ml E. coli culture (OD= 0.3± 0.03). It was photo-irradiated for 1 hour. In one of the aliquots, metal complex was added and kept in dark as control for photo-irradiation. The cells were pellet down @ 11000 rpm for 1 minute and washed thrice with distilled water. The final volume was made upto 1 ml with distilled water. 5 µl of the samples was taken for SEM imaging. ## Cell culture: The B16F10, CHO and BHK21 cells were grown with regular supplementation of DMEM+10% FBS medium at 37°C and 5% CO 2 . For microscopy, the cells were seeded on glass bottom dishes. The C4 was added to cells and incubated for 15 minutes. Thereafter, the cells were washed with PBS followed by imaging.

chemsum

{"title": "DNA Microviscosity Converts Ruthenium Polypyridyl Complexes to Effective Photosensitizers", "journal": "ChemRxiv"}

cs₂co₃-promoted_reaction_of_tertiary_bromopropargylic_alcohols_and_phenols_in_

## Abstract: The reaction of bromopropargylic alcohols with phenols in the presence of Cs 2 CO 3 /DMF affords α-phenoxy-α'-hydroxyketones (1:1 adducts) and α,α-diphenoxyketones (1:2 adducts) in up to 92% and 24% yields, respectively. Both products are formed via ring opening of the same intermediates, 1,3-dioxolan-2-ones, generated in situ from bromopropargylic alcohols and Cs 2 CO 3 . ## Introduction Due to the relative stability, ease of handling and the presence of reactive sites, bromoacetylenes are widely applied in synthetic organic chemistry. They are known to be involved in various transformations including homo-and cross-coupling , addition , cycloaddition and other reactions. Of particular synthetic value is the addition to the triple bond of bromoacetylenes to provide vinyl adducts, which can undergo numerous transformations. For example, bromoacetylenes were demonstrated to add imidazoles, imidazolines , and benzimidazoles to give vinyl bromides. Sulfonamides reacted with bromoacetylenes to deliver N-bromovinyl-p-toluenesulfonamides that under Heck reaction conditions afforded N-(p-toluenesulfonyl)pyrroles . The CsF-promoted nucleophilic addi-tion of isocyanides to bromoacetylenes furnished the functionalized bromovinyl amides followed by Pd-catalyzed formation of 5-iminopyrrolone . Sequential nucleophilic addition/intramolecular cyclization of amidine with bromoacetylenes led to imidazoles . Also, M 2 CO 3 -catalyzed (M = K or Cs) addition of phenols to bromoacetylenes produced bromovinyl phenyl ethers, which were converted into 4H-chromen-4-ones, benzo[b]furans, etc. . The latter reaction attracted our attention and prompted us to explore the interaction of phenols and bromopropargylic alcohols under the reported conditions. The bromopropargylic alcohols are readily available from acetylenic alcohols and hypobromite or N-bromosuccinimide . The presence of the hydroxy group expands the synthetic ## Entry Alkali metal carbonate (equiv) potential of these bromoacetylenes. Thus, we have recently demonstrated a highly selective hydration/acylation of tertiary bromopropargylic alcohols with carboxylic acids promoted by alkali metal carbonates . The reaction proceeds via the ringopening of 1,3-dioxolan-2-one intermediates formed with hydroxy and alkynyl groups of bromopropargylic alcohol and alkali metal carbonate. In the light of the above, it was unclear, in which direction would proceed the reaction of bromopropargylic alcohols and phenols. In the present paper, we report on the results of these studies. ## Results and Discussion Initially, bromopropargylic alcohol 1a and phenol (2a) were chosen as the model substrates for our investigation (Table 1). Completion of the reaction was monitored by IR and 1 H NMR spectroscopy by the disappearance of the bands at 2196-2212 cm -1 (-C≡C-Br) and signals of the bromopropargylic alcohol 1a, respectively. Under the conditions previously used for the addition of phenols to bromoacetylenes (K 2 CO 3 or Cs 2 CO 3 , DMF, 110 °C), the reaction turned out to be non-selective: along with the expected bromovinyl phenyl ether 3a (3-9%) and phenoxyhydroxyketone 4a (25-39%), diphenoxyketone 5a was isolated in 9-24% yield (Table 1, entries 1-3). At 50-55 °C, the reaction slowed down and became more selective (Table 1, entries 4 and 5). With Cs 2 CO 3 (1 equiv) at 50-55 °C, the reaction proceeded for 3 h, the yield of the phenoxyhydroxyketone 4a increased up to 55% and 5-phenoxymethylene-1,3-dioxolan-2-one 7, one of the probable intermediates, was isolated in 5% preparative yield (Table 1, entry 4), whereas the use of 2 equiv of Cs 2 CO 3 led to slightly more selective reaction (Table 1, entry 5). Further lowering the temperature reduces the selectivity toward phenoxyketone 4a. At room temperature, the full conversion of bromopropargylic alcohol 1a took 15 h and yields of phenoxyketones 4a and 5a decreased (Table 1, entry 7). In the presence of K 2 CO 3 (1 equiv) at 50-55 °C, the same reaction was completed for 8 h, the yields and selectivity being not improved (Table 1, entry 10). In these cases, 5-phenoxymethylene-1,3-dioxolan-2-one 7 was also isolated in 6-9% preparative yield. Hydrocarbonates CsHCO 3 and KHCO 3 were also tested in the reaction, which gave 5-bromomethylene-1,3-dioxolan-2-one 6a as a major product in 29-36% yield ( 1, entries . The efforts to increase the yield of diphenoxyketone 5a using 2 equivalents of phenol (2a) in the reaction with bromopropargylic alcohol 1a (Table 1, entries 8 and 9) failed. Employing the reaction conditions similar to those given in entries 4 and 6 (Table 1), we examined the substrate scope of the process relative to other phenols (Scheme 1). It was found that the electronic character of the substituents and the steric hindrance affected the reaction outcome. Bromopropargylic alcohol 1c having a tert-butyl group reacted with phenol (2a) in DMF for 3 h to give phenoxyhydroxyketone 4l in only 34% yield, 5-bromomethylene-1,3-dioxolan-2one 6b (5%) being isolated (Scheme 3). In DMF/H 2 O (3 h), the conversion of 1c was incomplete (50%) and phenoxyhydroxyketone 4l was obtained in 39% yield. So, the steric hindrances of the bulky groups noticeably affect the reaction. The reaction of secondary and primary bromopropargylic alcohols (4-bromobut-3-yn-2-ol and 3-bromoprop-2-yn-1-ol) and phenol (2a) with 1 equiv of Cs 2 CO 3 , DMF, 50-55 °C, for 3 h did not gave any products, the competitive polymerization of bromopropargylic alcohols 1 being predominant. Finally, chloroacetylenic alcohol was involved in the reaction with phenol (2a, 1 equiv Cs 2 CO 3 , DMF, 50-55 °C, 3 h) to afford the corresponding product 4a in 29% isolated yield (Scheme 4). We tested aniline and 2-naphthylamine as nucleophiles (DMF, 50-55 °C) in the reaction of bromopropargylic alcohol 1a (Scheme 5). But such a protocol turned out to be ineffective providing no desired products. Several control experiments were performed to gain insight into the reaction mechanism (Scheme 6). When the reaction of 5-bromomethylene-1,3-dioxolan-2-one 6a and phenol (2a) was carried out with KOH, the conversion of the starting 6a was 55% and crude product contained phenoxyketone 4a, diphenoxyketone 5a and 5-phenoxymethylene-1,3-dioxolan-2-one 7. Using 2 equivalents of phenol (2a) in the reaction of 5-bromomethylene-1,3-dioxolan-2-one 6a (Cs 2 CO 3 , DMF, 110 °C, 20 min) gave phenoxyketone 4a and diphenoxyketone 5a in 40 and 16% yields, correspondingly. These results confirm that compound 6a is the main intermediate to form phenoxyketones. Next, we carried out the experiment using CO 2 gas with DBU as a base. In comparison with reactions without CO 2 (Table 1, entries 17 and 18), bromopropargylic alcohol 1a with free CO 2 gas in the presence of 100 mol % of DBU and phenol (2a) (DMF, 50-55 °C, 3 h) afforded phenoxyketone 4a, 5-bromomethylene-1,3-dioxolan-2-one 6a and 5-phenoxymethylene-1,3-dioxolan-2-one 7 in 27, 4 and 19% yields, respectively. This result suggest that Cs 2 CO 3 acts as a source of CO 2 for the formation of 5-bromomethylene-1,3dioxolan-2-one 6a. Obviously, the formation of phenoxyhydroxyketone 4 proceeds via 1,3-dioxolan-2-one 6 generated from bromopropargylic alcohol 1 and Cs 2 CO 3 . Then, Br-substitution/hydration of 6 and the release of CO 2 give product 4 (Scheme 7). Apparently, diphenoxyketone 5 results from decarboxylative conversion of 1,3-dioxolan-2-one 7 leading to intermediate A, nucleophilic attack of phenolate at the less sterically hindered carbon of the above zwitterion A and subsequent protonation of anion B (Scheme 8). Based on these plausible mechanisms for the formation of phenoxyketones, it can be assumed that a decline of the Cs + concentration after Cs 2 CO 3 convertion to CsBr (because of the very poor solubility of CsBr in DMF) has an influence on the rate of diphenoxyketone formation. In addition, the suppression of the di(nitrophenoxy)ketone formation can be due to the lower basicity of a reaction mixture since nitrophenols 2d,e are more acidic than phenols 2a-c,f-i (pK a values: 9.99 phenol (2a), 9.40 α-naphthol (2b), 9.57 β-naphthol (2c), 7.18 p-nitrophenol (2d), 7.23 o-nitrophenol (2e), 10.28 p-cresol (2f), 10.27 p-methoxyphenol (2g), 9.36 p-bromophenol (2h), 10.19 eugenol (2i)). Addition of water to the reaction mixture also reduces the pH of the medium and simultaneously increases the concentration of hydroxide ions, therefore, diphenoxyketones 5 were not produced and dihydroxyketones 8 were formed as side products in these cases. Among the approaches to produce α-phenoxyketones, the most common methodologies are base-catalyzed alkylation of the corresponding phenols with halo- and mesyl ketones (Scheme 9), the preparation of which are not always selective and high-yielded. The ring opening of ArOCH 2 -epox-Scheme 10: α-Ketol rearrangement of phenoxyketones 4a and 4f. ides , the SmI 2 -catalyzed reductive coupling of acid halides with ketones and acetolyses of α-phenoxy-αdiazoketones were also employed. Recently, F. P. Cossío et al. have described a method for the preparation of benzo ## Conclusion We have shown that the main direction of the reaction of bromopropargylic alcohols and phenols in Cs 2 CO 3 /DMF is the hydration/phenoxylation of bromopropargylic alcohols to afford phenoxyketones. This step-economical process takes place under mild reaction conditions using simple readily available starting materials. The synthesized phenoxyketones are of interest as valuable building blocks for the production of other important molecules (e.g., amino alcohols, diols, etc.) and potential pharmaceuticals. α-Hydroxyketones are structural subunits of natural products and compounds possessing immunosuppressant , antidepressant , amyloid-β protein production inhibitory , urease inhibitory , farnesyl transferase inhibitory (kurasoin A and B) , antitumor and antibacterial (doxorubicin, olivomycin A, chromomycin A 3 , carminomycin I, epothilones) activities. ## Experimental General information ## ORCID ® iDs Olesya A. Shemyakina -https://orcid.org/0000-0001-7371-3982

chemsum

{"title": "Cs₂CO₃-Promoted reaction of tertiary bromopropargylic alcohols and phenols in DMF: a novel approach to \u03b1-phenoxyketones", "journal": "Beilstein"}

the_following_article_has_been_submitted_to_j._chem_phys_(2021)_communication:_effect_of_oxidation_o

## Abstract: Excited state lifetimes of neutral titanium oxide clusters (TinO2n-x, n < 10, x < 4) were measured using a sequence of 400 nm pump and 800 nm probe femtosecond laser pulses. Despite large differences in electronic properties between the closed shell stoichiometric TinO2n clusters and the suboxide TinO2n-x (x = 1-3) clusters, the transient responses for all clusters contain a fast response of 35 fs followed by a sub-picosecond excited state lifetime. In this non-scalable size regime, subtle changes in the sub-ps lifetimes are attributed to variations in the coordination of Ti atoms and localization of charge carriers following UV photoexcitation. In general, clusters exhibit longer lifetimes with increased size and also with addition of O atoms. This suggests that removal of O atoms develops stronger Ti-Ti interactions as the system transitions from a semiconducting character into a fast metallic electronic relaxation mechanism. ## INTRODUCTION Bulk titania (TiO2) materials are the subject of numerous experimental and theoretical investigations 1,2 due to their wide application in water splitting, white pigments, 6 and photocatalysis. 7 The most important aspect of photoactivity is the production of charge carriers with sufficient lifetimes to participate in chemical reactions. Absorption of a photon exceeding the optical gap results in the creation of an exciton, or electron-hole pair, that can recombine through non-radiative processes such as internal conversion or relax through electron scattering. In strongly correlated materials, such as titania, the charge carriers are coupled to lattice vibrations to form polarons, which serve to trap mobile carriers by reducing their mobility and photoconversion yields. Polaron formation is affected by the local geometry of the material, where despite identical chemical compositions, recombination is two orders of magnitude faster in the rutile phase than the anatase phase of TiO2. 8,9 TiO2 nanoparticles follow two relaxation channels, 10 with holes and electrons relaxing separately over ns to μs dependent upon the particle size and structure. However, the factors affecting these separate lifetimes and how to control these mechanisms remains poorly understood. A major limitation for bulk titania is the large bandgap which limits electron transport. Defect engineering has become a major focus for titanium oxides, where oxygen deficient materials contain a smaller bandgap, increasingly delocalized density of states (DoS) 14,15 and therefore can utilize more of the sun's visible spectrum. Suboxides of titania, referred to as TinO2n-x (x > 0), are easily produced from bulk TiO2 materials, are non-toxic, 19 possess unique optical properties, 18,20 and may have increased catalytic activity over their stoichiometric counterpart. 17,21 In particular, Magnéli phase titanium oxides (TinO2n-1, n = 4-9) display enhanced electrical conductivity 22 and increased stability over stoichiometric titania. 23 The overall reactivity of bulk titanium oxides is thought to be heavily dependent on O vacancies and associated Ti 3+ sites, 24,25 yet a precise understanding of their influence on the behavior of excitons, polarons, and free charge carriers in titania is needed. Clusters serve as atomically precise models that enable fundamental studies on the factors that affect charge carrier recombination in bulk materials. Similar to the band structure and photoabsorption in bulk titania, stoichiometric (TinO2n) clusters are closed shell systems with low energy photoexcitation, described as an electron transfer from the O 2p-orbitals to the Ti 3d-orbitals or ligand to metal charge transfer (LMCT). Our measurements on the excited state lifetimes of sub-nanometer TinO2n (n < 10) clusters found their large optical gap acts as a relaxation bottleneck, making the properties of the S1 state strongly correlated with the measured lifetimes. 26 The variation in sub-picosecond relaxation dynamics revealed the local cluster geometry and charge carrier separation to be important in adjusting excited state lifetimes. 26 Similarly, suboxide clusters act as models for the distortions caused by O vacancies on the surfaces of titania materials. However, the electronic properties of suboxide clusters are quite different due to the presence of partially filled 3d-orbitals, dramatically changing the DoS below the photoexcitation energy and are therefore expected to possess different excited state dynamics. Here, fs pump-probe spectroscopy is employed to measure the excited state lifetimes of suboxide TinO2n-x clusters (n < 10, x < 4). This study supplies a foundation for understanding molecular-scale titania, which will ultimately lead to the production of materials with improved reactivity. ## EXPERIMENTAL METHODS The experimental methods and instrumental setup were described previously. 26,27 Briefly, neutral clusters were produced through ablation of a pure 0.25" diameter Ti rod by the second harmonic of a pulsed Nd:YAG laser in the presence of a seeded He gas pulse (1% O2). The ablation plume was confined to a 1 x 60 mm collision region and reduced to a molecular beam diameter of 2 mm by a charged skimmer, deflecting all but the neutral clusters. Neutral clusters were ionized by a sequence of 35 fs laser pulses from a Ti:Sapphire laser and analyzed using a home-built Wiley-McLaren 28 type time-of-flight mass spectrometer. The ionized clusters separated in arrival time due to their m/z ratio within a fieldfree region and were subsequently recorded using a microchannel plate (MCP) detector. The 400 nm (3.1 eV) pulse was used to excite (pump) the clusters to an intermediate state and the 800 nm (1.55 eV) probe pulse was sent through a programmed delay-stage before recombining with the 400 nm beam for ionization. A delay to the 800 nm laser pulse was scanned by 10 fs steps and the change of intensity with delay was recorded. An average of 200 shots per time step as the probe was delayed from -1.6 -6.8 ps. The pump-probe instrumental response function (IRF) of 35 fs was measured using non-resonant excitation of Ar gas. Mass spectra were recorded using 400 nm pump and 800 nm probe pulses of 9.9 x 10 14 W/cm 2 and 3.1 x 10 15 W/cm 2 intensity, respectively. All transient signals were fit using a combination of two Gaussian functions convoluted with an exponential decay to account for the relaxation lifetimes. 27,29 3. RESULTS AND DISCUSSION ## TITANIUM OXIDE CLUSTER DISTRIBUTION Ionization of the neutral titanium oxide molecular beam by the 400 nm pump and 800 nm probe lasers at temporal overlap produced a mass spectrum of Ti2O to (TiO2)11 (Fig. 1). The primary clusters recorded follow the series of TinO2n-x (x = 0-3) and grow through addition of TiO2 units. The cluster distribution is consistent with previous studies, 30 with the highest intensity peaks generally composed of TinO2n-1 or TinO2n-2, and is in agreement with their stabilities. 15,31 Neutral stoichiometric (TinO2n) clusters have previously only been recorded experimentally using single photon (VUV) ionization thought to be void of fragmentation, 32 supporting that fragmentation is not significant in our experiment. Thus, the formation of O deficient clusters is due to kinetic limitations during the growth of clusters in the ablation plasma. Sub-nm titanium oxide clusters form hollow cage-like geometries 33,34 that are different from the bulk lattice structure. The cluster geometries and electronic structures were calculated using time-dependent density functional theory and CAM-B3LYP functional as described in detail in other publications. 26,31 The two terminal O atoms in TinO2n clusters have the lowest binding energy, making the geometry of TinO2n-1 clusters similar to TinO2n but lacking one dangling O atom. The lack of one dangling O atom causes the HOMO to shift from an O 2p-orbital onto a Ti 3d-orbital making both the HOMO and LUMO primarily localized on Ti atoms. 31,35 Therefore, photoexcitation in suboxides is a d-d transition and occurs with a much smaller optical gap. 35 The TinO2n- ## SIZE EFFECT ON SUBOXIDE CLUSTER LIFETIMES Surprisingly, despite the large changes to the electronic and structural characteristics of the clusters as they gain and lose O atoms, 26,31 their excited state dynamics remain roughly consistent. The transient signals for all clusters contain a fast (35 fs) response and a sub-ps relaxation lifetime (τ). No longlived states are recorded, and the ratios (κ) of fast/sub-ps fitting coefficients are similar (Table 1). These similarities suggest that the sub-nm scale is perhaps the most important feature, and that relaxation is efficient in these sub-nm clusters due to the restricted proximity of their diameter. The fast component of the transient signal is attributed to a rapid relaxation of a nonresonant excited state. The remaining sub-ps transient ion signal is proportional to the neutral cluster's intermediate excited state population as it relaxes to lower energy. Pumpprobe transients of the TinO2n-1 (n < 10) cluster series is presented in Figure 2. Transient signals of the remaining suboxide clusters are presented in the Supplemental Information (Fig. S1 and Fig. S2). Clusters represent a size regime of non-scalable properties, where every atom impacts the collective electronic and structural properties. Therefore, subtle differences in the excited state lifetimes highlight the variation of cluster geometries and electronic structures on dynamics and excited state lifetimes. Although there is variation in the lifetimes of all cluster series, each TinO2n-x series (x = 0-3) exhibits a gradual increase in excited state lifetime with the addition of TiO2 units (Table 1), due to a combination of increased charge carrier separation and overall increase in bonding coordination. Despite the similarities, each cluster series exhibits a unique trend as they grow in size. A near linear increase in lifetime occurs with size in TinO2n-1 clusters up to n = 7, with the exception of Ti6O11 (Fig. 3b). All clusters in this series (except Ti6O11) contain a mirror plane to stabilize the reduced Ti atoms. 31 Thus, due to a lack of symmetry, the tri-coordinated Ti site of Ti6O11 may retain additional d-electrons that facilitate a faster decay, deviating from the trend. The TinO2n-2 clusters are more compact, given the absence of any terminal O atoms, and universally have shorter lifetimes. The TinO2n-2 cluster series of n ≥ 4 shows an oscillatory nature (Fig. 3c), where oddnumbered clusters have a longer lifetime over even-numbered clusters due to a higher localization of charge carriers. The lifetimes of the TinO2n-3 series also alternate with increasing cluster size and is most pronounced for the smallest cluster sizes (Fig. 3d). Both oxygen-deficient series exhibit the opposite behavior from the TinO2n series, where even-numbered clusters exhibited longer lifetimes than odd-numbered clusters. 26 ## OXIDATION EFFECT ON SUBOXIDE CLUSTER LIFETIME The oxidation states of the Ti atoms are commonly assumed to involve complete electron transfer, where each O atom removes 2 electrons from the Ti atoms. Removal of each O atom from the stoichiometric cluster returns two d-electrons to the Ti atoms. Therefore, the suboxides contain many delocalized d-electrons that should influence the metallicity and consequently the relaxation dynamics of the cluster. Unfortunately, measurements of metallic behavior, such as conductivity, are not possible for clusters of just a few atoms. Another indicator for metallicity is short excited state lifetimes and has been well established for small metal clusters. Metallic and nonmetallic properties can be identified by the different relaxation behaviors of optically excited states. Strong interactions between delocalized valence d-electrons causes relaxation in metallic species on the fs timescale via Auger-like electron-electron scattering, whereas a weak coupling between electronic excitation and nuclear motion facilitates long (picosecond or longer) lifetimes of electron-hole characteristics in non-metallic semiconductors. Metallic scattering processes dominate if there are many delocalized electrons and a sufficiently high DoS, such as is the case even in small clusters. Thus, the relaxation by electron scattering processes results in many electrons occupying lowlying excited states, similar to the bulk. However, internal conversion is an alternative possible pathway of relaxation and cannot be ruled out as contributing here. Small molecules can exhibit excited state lifetimes on the order of 10s of fs, particularly when there is passage through a conical intersection between two potential energy surfaces. Despite the well-known role of dangling O atoms in facilitating energy relaxation through conical intersections, 44 clusters void of terminal O atoms exhibit similar lifetimes to those with them. This suggests that internal conversion is not driving the relaxation. Further, the clusters are sufficiently small such that electrostatic interactions between the hole and electron are efficient for relaxation. The transient signals of neutral Ti2O4-x (x < 4) clusters reveal an increase in excited state lifetimes with oxidation (Fig. 4). Here, each O atom changes the oxidation state of the Ti atoms linearly, from a formal oxidation state of +3 (Ti2O3) and +2 (Ti2O2) which decreases the lifetime by 15% and 26% from the stoichiometric cluster, respectively. Geometries of Ti2O4-x (x < 4) clusters are well established 31,45 (Fig. 4). Ti2O4 is the least rigid cluster (containing two terminal O) and therefore should be the easiest to traverse a conical intersection since internal conversion is less effective in rigid clusters. Yet, it contains the longest lifetime of the series. In contrast, the more rigid ring structure of Ti2O2 contains no dangling O atoms and has a faster relaxation, suggesting that internal conversion is not the dominant relaxation mechanism occurring here. Although the number of relaxation pathways decrease with the removal of O atoms, the bond distances shorten 31 and d-orbital occupancy increases, resulting in a faster relaxation. This influence of O content on lifetime aligns with a metallic to insulator transition occurring with oxidation. Thus, our data supports that relaxation in clusters occurs via Auger-like electron scattering processes similar to bulk metals. The transient response in O deficient Ti3O6-x (x < 4) clusters is different than the other cluster series (Fig. 5). The fully oxidized cluster, Ti3O6, has the shortest excited state lifetime in the Ti3On series, while Ti3O3 shows a longer lifetime. However, the Ti3O3 transient contains an unreliable sub-ps component with high experimental noise. Therefore, the high variation in κ is not expected to indicate any valuable scientific trend. Ti3O6 is unique among the stoichiometric clusters in that it contains a tri-coordinated Ti atom and tri-coordinated O atom, which are not present in other stoichiometric (n < 6) structures. 15,26,35 The tri-coordinated Ti atom sites are common in clusters exhibiting suppressed lifetimes. Ti3O5 exhibits a slightly longer lifetime, even though it is further undercoordinated, due to presence of partially filled d orbitals which are delocalized across two Ti atoms. In general, the delocalization of the d orbitals is correlated with extended lifetimes in suboxides. Similar to the Ti2O4-x clusters, Ti4O8-x (x < 4) clusters also highlight an increased lifetime with oxidation (Fig. 3a), supporting a metallic to semiconducting transition. Ti4O8-x (x < 4) clusters have 3D structures with the Ti atoms forming a tetrahedron core and O atoms bridging the Ti atoms or as terminal groups for Ti4O7 and Ti4O8. The Td symmetry of Ti4O6 lacks terminal O atoms and also contains an increased number of d-electrons which manifest in a large decrease in lifetime. Ti4O5 is similar to Ti4O6 but contains one less bridging O atom that reduces the Ti-O coordination and decreases the Ti-Ti bond length, resulting in a slightly shorter lifetime. Although there is minimal change in geometry in Ti4O8-x (x < 4) clusters, a higher d-orbital occupancy occurs with decreased O atoms from Ti4O8, 35 resulting in a linear decrease in the excited state lifetime. Lifetimes of Ti5O10-x (x < 4) clusters do not change significantly with O (Fig. 3a). The similar lifetimes are due to a similar Cs symmetry and align with the size transition between local and global excitations, where charge carrier delocalization no longer fills the cluster diameter. Further, Ti5O10 exhibits a slightly reduced lifetime in the stoichiometric series, bringing the dynamics closer to Ti5O9. Interestingly, several stoichiometric clusters exhibit shortened lifetimes that deviate from the proposed metallic trend. The TinO2n clusters, where n = (3,5,6), contain a tri-coordinated Ti atom instead of the fully tetra-coordinated Ti atoms of the other clusters, and therefore may retain d-electrons that reduce their lifetimes. Such incomplete electron transfer leading to an atypical 3+ oxidation state is proposed for clusters as small as TiO2. 46 Larger clusters (TinO2n-x, n = 6-9) follow similar trends in excited state behavior (Fig. 3a). Generally, the TinO2n-1 and TinO2n clusters show longer lifetimes, and exhibit similar lifetimes due to related structures and possibly incomplete electron transfer, leading to retention of d-electrons on the stoichiometric cluster. 15 Further, in TinO2n clusters, the excited state avoids localization on the Ti atoms with terminal Ti-O bonds, 26 which ensures that the excited states behave similarly in the various O deficient clusters and accounts for the minimum influence of lifetime with oxidation. Clusters without terminal O atoms (TinO2n-2 and TinO2n-3) show shorter lifetimes and increased d-electron occupancy, indicating that the delectron scattering is a dominant mechanism affecting dynamics and that bridging O atoms have a minor effect on excited state lifetimes. A particular outlier to the described trends is Ti7O13, which exhibits a significantly longer lifetime than Ti7O14 (Fig. 3a). This switched behavior is attributed to its unique structural features, where removal of an O atom from Ti7O14 drives a significant compression of the local Ti-O bonds and creates a new bond forming a tetra-coordinated Ti site adjacent to a tetracoordinated O atom. This high coordination site may account for its exceptionally long lifetime, in opposition to tricoordinated Ti sites facilitating fast relaxation. Further, Ti7O13 exhibits the largest separation of charge carriers and electron delocalization, supporting that this delocalization of is correlated with lifetime. Clusters generally exhibit longer lifetimes with higher oxidation and shorter lifetimes upon removal of O atoms. Although the optical gap of suboxides decreases by ~3 eV from the stoichiometric cluster, 15 it does not have a significant influence on the excited state lifetime. This suggests that removal of O atoms develops metallic Ti-Ti bonds of lower coordination, causing the system to transition into a fast scattering-type electronic relaxation mechanism. This is consistent with the idea that as the clusters become more metallic, the lifetimes decrease. Excited state lifetimes are modified by electron-hole interactions which are influenced by Ti bond coordination and cluster size. These results suggest that enhanced excited state lifetimes in bulk titania materials may be achieved through manufacturing structures similar to the Ti7O13 cluster that contain delocalized d-electrons and higher Ti coordination. ## CONCLUSION The low-lying excited state lifetimes of neutral TinO2n-x (n = 1-9 and x < 4) clusters were measured using fs pump-probe spectroscopy, and trends in their transient signals related to the size and oxidation are presented. An oscillation in lifetimes as clusters grow in size is attributed to structural differences between the clusters that control charge localization and polaron-like formation. The signal returns to baseline for all clusters, suggesting that relaxation is efficient for these sub-nm materials. We show that the level of coordination increases with cluster size, related to a longer lifetime. The excited state lifetimes of titanium oxide clusters change with oxidation, which affect the Ti coordination and charge carrier localization. The lifetimes show a behavior consistent with a metallic to semiconducting transition with oxidation and related removal of d-electrons from the system. The fundamental information provided herein leads to a deeper understanding of the factors affecting O vacancies in bulk-scale titania materials and will assist in the production of future catalysts of increased reactivity. ## Supporting Information See supplemental material for experimental measurements and fitting of larger TinO2n-x (n = 4-9, x < 4) clusters. ## Corresponding Author *Scott.Sayres@asu.edu

chemsum

{"title": "The following article has been submitted to J. Chem Phys (2021) Communication: Effect of Oxidation on Excited State Dynamics of Neutral TinO2n-x (n<10, x<4) Clusters", "journal": "ChemRxiv"}

breaching_the_axial_limits_in_ln(iii)_single-ion_magnets_using_external_electric_field

## Abstract: Single-Molecule Magnets have potential applications in several nano-technology applications including in high-dense information storage devices and realization of this potential application lies in enhancing the barrier height for magnetization reversal (Ueff).Recent literature examples suggest that the maximum values that one can obtain using a ligand field are already accomplished. Here we have explored using a combination of DFT and ab initio CASSCF calculations, the way to enhance the barrier height using an oriented external electric field for top three Single-ion Magnets ([Dy(Py) 2) and [Dy(Cp Me3 )Cl] ( 3)). For the first time our study reveals that, for apt molecules, if appropriate direction and value of electric fields are chosen, the barrier height could be enhanced twice that of the limit set by the ligand field. This novel non-chemical-fine tuning approach to modulate the magnetic anisotropy is expected to yield new generation SIMs. There is a great interest in the area of single-molecule magnets (SMMs) as they are reported to have potential applications in information storage devices, cryogenic refrigeration, quantum computing and spintronics devices to name a few. 1 SMMs containing Lanthanide(III) ions are gaining interest in recent years as they possess huge barrier height for magnetization reversal (Ueff) and at the same time possess record high blocking temperatures (TB). While there are various classes of molecules exhibit blocking temperatures in the range of 4-15 K, 2 higher blocking temperatures are found for organometallic Dy(III) single-ion magnets (SIMs) containing substituted cyclopentadienyl ligands (TB in the range of 48 K to 80 K). 3 It is well-known that the shape of the electron density of the ground state mJ levels of the Lanthanide ion is critical in dictating their magnetic properties. The Ln(III) ions are classified as those possessing oblate density require strong axial with no/weak equatorial ligation and those with prolate density demands strong equatorial ligand with weak/no axial ligation. Synthetic chemists have been utilizing these ideas to develop novel molecules with attractive Ueff and TB values. 4 While most of the molecules possessing very high-blocking temperature also possess substantial Ueff values, often the TB is only a fraction of the Ueff values reported. While establishing the relationship between the Ueff and TB and mechanism beyond the singleion relaxation has gained attention, 5 it is also equally important to realize large Ueff values to move forward. To obtain a larger Ueff value for lanthanide complexes, various chemical fine-tuning methods such as (i)designer ligands that control the ligand field around the Ln(III) ion in an anticipated fashion, 6 (ii) maintaining symmetry around the metal centre, 2, 7 (iii) incorporating diamagnetic elements in the cluster aggregation to enhance the axiality 8 or (iv) incorporate transition metal or radicals to induce exchange interaction as a way to suppress tunneling have been explored. 4,9 With numerous Dy(III) mononuclear complexes reported in the literature, it has been stated that the axial limit that controls the overall Ueff value has been reached. 2a While increasing TB value being the focus at present, other avenues to enhance the Ueff values have not been explored. As chemical fine-tuning of the ligand field has already reached its potential, we aim to search an alternative route to enhance the Ueff values in Ln(III) SIMs. In this context, using various computational tools, here we set out to explore the role of an applied electric field in the magnetization reversal of Ln(III) SIMs. Recent examples in this area where electric field has been utilized to modulate the magnetic properties offers strong motivation for this work. 10 To enumerate the effected of an oriented external electric field (OEEF) on Lanthanide SIMs, we have chosen three examples [Dy(Py)5 11 (2) and [Dy(Cp Me3 )Cl] 3a (3) complexes. All three complexes are characterized well and among the best-known SIMs in their family. Particularly complex 1 found to exhibit a Ueff value of 1815 K with a blocking temperature of 14 K while complex 2 found to have Ueff value of 63 K with a TB of 3 K. Complex 3, on the other hand, do not exhibit any out-of-phase signals and hence is not a Single-ion magnet. 3c Computing the magnetic anisotropy of the Ln(III) SIMs in the presence of electric field has not been attempted, and there are multiple challenges present to account for such effects. The application of oriented electric fields is expected to distort the geometry and capturing this effect is crucial in understanding the magnetic anisotropy. As Ln(III) SIMs are known to be extremely sensitive to small structural changes, static OEEF on Xray structure unlikely to reveal the real scenario. As structure optimization with the ab initio CASSCF calculations is not practical at present, here we have chosen a combination of methodology wherein DFT calculations in the presence of electric field were performed to obtain reasonable structures. These structures were then subject to ab initio CASSCF/RASSI-SO/SINGLE_ANISO calculations in the presence of same electric field to capture both the structural distortion and also the electric field effect on the magnetic anisotropy (see computational details for more information). Ab initio calculations were performed on the crystal structures of complexes (or models derived from the X-ray structure) of 1 -3 in the absence of any external perturbation (see Table S1-S3 in ESI). Complex 1 and 2 are well-known examples, exhibiting strong axiality in the estimated gz values with the computed barrier height of 1183 cm -1 and 181 cm -1 , respectively (relaxation via 4 th excited Kramers doublet). 2a, 12 . As geometries of 1 and 2 are relaxed in the presence of an electric field, it is imperative to understand how the optimized geometry in the gas phase fair to the X-ray structure. The optimized geometry of complexes (1opt and 2opt) reveal elongation of all the bonds within the molecule as intermolecular interactions in the crystal lattice are removed. The axial Dy-O(1) bond length increases from 2.110 in the X-ray structure to 2.142 in 1opt, and the average Dy-N bond also increases ~0.05 in 1opt geometry (see Table 1). Similar elongation has been witnessed in Er-N/Cl bond lengths in complex 2. The CASSCF calculations on 1opt and 2opt yield Ucal value of 1118 cm -1 and 144 cm -1 , respectively, assuming relaxation via 4 th excited state (See Fig. 1) and these values are marginally smaller than the X-ray geometry due to relatively weaker axial ligand field (LF) in optimized geometries (see Table S4-S5). In the next step, we have attempted to optimize the geometry in the presence of oriented external electric field (OEEF) starting from 0.004 au (atomic unit and equivalent to 0.2 V/). 10c, 13 The electric field applied here varies from 0.004 au to 0.026 au and lies within the limit of ionization energies, bond dissociation energies and is accessible for most of the STM tips. While the electric field induced spectroscopic techniques uses smaller field, organic reactions performed using OEFF are comparable to the electric field utilized here. Applying the electric field along the +z-axis in 1opt (See Fig. 1a and Fig. S1 in ESI), elongates the Dy-O(1) bond and at the same time shortens the Dy-O(2) bond and therefore breaks the pseudo D5h symmetry of the molecule. We have performed ab initio CASSCF calculations on this optimized geometry 4z 1opt (here superscript denotes the amount of OEEF applied x 10 -3 au along +z direction) in the presence of electric field (EF) wherein a reduction in the barrier height was witnessed. This is due to the fact that Dy-O(1) elongation cause the weakening of the axial LF and hence reduces the axial anisotropy for the oblate Dy(III) ion. Although there is a simultaneous shortening of Dy-O(2) is noticed, 4z 1opt geometry reveal that elongation is greater than the shortening (see Fig. S1) and hence this is not fully symmetric leading to a smaller Ucal value of 1108 cm -1 for 4z 1opt. In the next step, we increase the OEEF value in a stepwise manner to 0.012 au and see clearly that increase in the electric field increase the Dy-O(1) bond further and at the same time shortens the Dy-O(2) bond albeit asymmetrically. This led to a further reduction in the barrier height with a value of 1040 cm -1 noted for 12z 1opt structure (see Table S6 and S9-S11 in ESI). This reduction in barrier height can be rationalized by analyzing Lo-Prop charges at the spin-free ground state. By increasing OEEF, the LoProp charge on O(1) gradually decreases while it increases in O(2) (see Table S8 and S16). Perceiving this effect, we switched the OEEF direction along the x/y direction of complex 1opt (see Fig. S1 in ESI). The optimized structure of 4x 1opt (here superscript denoted the amount of OEEF applied x 10 -3 au along +x direction). Here Dy-N(1) bond length found to increase sharply from 2.62 to 2.80 vis a vis 1opt vs 12x 1opt (see Table 1) geometry and at the same time two of the Dy-N (along the -x direction) found to shorten asymmetrically. Also, the effect of applying OEEF along Dy-N(1) direction can be seen in a substantial decrease in the LoProp charge of N(1) atom while the charges on the oxygen atoms remain unaltered (see Table S8 in ESI). As three Dy-N bonds are significantly elongated at 12x 1opt geometry, one could expect a large barrier height, however, ab initio calculations reveal a contrary with a barrier height diminishing with an increase in OEEF value yielding a Ucal value of 939 cm -1 for 12x 1opt which relaxes via 3 rd excited KDs. (see Table S6 and S12-S14 in ESI). This is due to alteration of Dy-N distances that are accompanied by a variation in the O-Dy-O angle, which is reduced to 157 in 12x 1opt from 178 in 1opt geometry (see Table 1). Thus, the application of the electric field along the perpendicular or gx-direction worsens the barrier height for complex 1. To prove that the reduction is solely due to the O-Dy-O bending, we have performed one additional set of calculation on 12x 1opt geometry wherein the O-Dy-O angle is fictitiously set at 178 and this structure yield a barrier height of 1162 cm -1 (See Fig. S2 and Table S15 in ESI). This value estimated is ~50 cm -1 higher compared to optimized geometry offering a possibility, however small, to enhance Ucal value in 1 using an applied electric field. Furthermore, increasing the OEEF at 0.016 au results in dissociation of the Dy-N bond, which sets the electric field limit at x/y direction of the molecule. To further understand how the structure alteration occurs due to OEEF, it is important to understand the nature of dipoles and their behavior in the applied electric field conditions. Application of OEEF expected to polarize a non-polar bond and enhance the ionic character of a polar bond. 13 For a Ln-L bond, the application of OEEF will stretch it further if the dipolar field creates opposite dipole with respect to the Ln-L dipole and will help to shorten it, if the dipolar field is in the same direction as the Ln-L dipole (see Fig. 2b). Therefore, the molecule has to be chosen in such a way that an increase in the Ln-L bond length will enhance the magnetic anisotropy and will subsequently increase barrier height (Ueff). Applying the OEEF along an equatorial Ln-L bond in oblate ions such as Dy 3+ or along an axial Ln-L bond in prolate such as Er 3+ or Yb 3+ thus likely to increase the Ueff value beyond the X-ray structure reported values. However, if the OEEF is applied along the opposite directions, this is expected to decrease the Ueff value further. Based on the knowledge gained, we intuitively expand the study to a prolate Er(III) ion using complex 2. We have narrow down to this example for two reasons (i) to choose a well-studied prolate Er(III) SIM with a significant barrier height (ii) to choose an Er(III) SIM with a strong equatorial ligand and a weak axial ligand along only one direction as this would be expected to facilitate the enhancement of the Ucal value upon application of OEEF. Upon applying the OEEF along the Er-Cl direction (gz axis, see Fig. 2), with the same step size as before, the Er-Cl bond length found to increase significantly (see Fig. S3 in ESI and Table 1) reaching a maximum value of 3.04 at 0.026 au EZ ( 26z 2Opt). The application of OEEF beyond this value found to cleave Er-Cl bond suggesting possible ionization/decomposition limit. Additionally, the {N3Er} out-of-plane pyramidal shift (parameter  see Fig. 2 and S3 in ESI) also found to alter upon application of OEEF. As OEEF is applied along the Er-Cl bond, this bond elongates and also pushes the Er 3+ ion down and hence decreases the  value. The  value decreases from 0.5 at the 2opt to 0.3 at 26z 2opt. If the OEEF is applied along the -z direction (Cl-Er direction), this tends to enhance the pyramidalization (see Fig. S3 in ESI) and thus,  value increases to 0.62 at 26-z 2opt. Theoretical studies performed earlier on complex 2 reveal that this is an important parameter that enhances the barrier height. 15 Application of OEEF along the gz in 2 (.i.e along Er-Cl bond) axis enhances the Ucal 163 cm -1 at 4z 2opt to a remarkable 317 cm -1 at 26z 2opt. This estimate is one of the highest obtained for any Er(III) SIMs. 16 Computed QTM (and TA-QTM) values reveal a smooth decrease of these values from 2.2 at 4z 2opt to 1.3 at 26z 2opt. (see Table S17-S24 in ESI). Also, a smooth linear increase of the negative B2 0 parameter was observed for complex 2 under the applied electric field range along the +z direction (see Fig. S4 and Table S27 in ESI). If OEEF applied on the reverse direction on complex 2, i.e., along the -z direction, a reverse trend was visible with a gradual decrease in Ucal value. As expected, here the Er(III)-Cl bond length decreases and a  value were noticed upon applying an electric field in the -z direction. The Ucal value decreases from 131 cm -1 at 4-z 2opt to a much smaller value of 52 cm -1 (via 3 rd excites state) at 24-z 2opt structure (see Table S25-S27 at ESI). Further, the Ucal vanishes to zero at 26-z 2opt with a notable ground state QTM. We have also plotted the β-electron density of the Er(III) under the applied electric field conditions, and this nicely reflects the changes observed (see Fig. S5 for plot corresponding to 26-z 2opt, 2opt and 26z 2opt). After achieving such a large Ucal value for complex 2, we extended the study further to another Dy(III) example namely [Dy(Cp Me3 )2Cl] (complex 3) (Cp Me3 = trimethylcyclopentadienyl) (see Figure S6 top) which is a model complex derived from X-ray structure of the famous precursor [Dy(Cp ttt )2Cl]. 3a Our calculations on the optimized structure reveals a very small Ucal value of 144 cm -1 relaxing via first excited state due to high QTM being operation due to the coordination of -Cl along the equatorial direction (see Table S28 in ESI). To quench this QTM, we have applied the OEEF along the Dy-Cl bond direction and this leads to weakening of Dy-Cl bond and gradually the Ucal value increases from 160 cm -1 at 4z 3opt to 519 cm -1 at 22z 3opt structure (see Table S28 and Figure S7). The Dy-Cl bond length increases from 2.59 at 4z 3opt to 2.94 at 22z 3opt. As the Dy-Cl bond increases with the electric field, a simultaneous increase of Cp-Dy-Cp angle was observed. Application of the electric field beyond 0.022 au results in rupture of the Dy-Cl bond. Thus, the Ucal value increases by three times more than that of the optimized geometry. To this end, here we have explored the possibility of finetuning the barrier height for magnetization reversal using oriented external electric fields in Ln(III) SIMs. Enhancement in Ucal value twice that of the X-ray structures offers a viable nonchemical fine-tuning way to enhance the barrier height beyond the limit set by the ligand fields. This novel approach expected to trigger substantial interests to obtain new generation SIMs unveiling its potential applications.

chemsum

{"title": "Breaching the Axial Limits in Ln(III) Single-Ion Magnets Using External Electric Field", "journal": "ChemRxiv"}

presence_of_amorphous_carbon_nanoparticles_in_food_caramels

## Abstract: We report the finding of the presence of carbon nanoparticles (CNPs) in different carbohydrate based food caramels, viz. bread, jaggery, sugar caramel, corn flakes and biscuits, where the preparation involves heating of the starting material. The CNPs were amorphous in nature; the particles were spherical having sizes in the range of 4-30 nm, depending upon the source of extraction. The results also indicated that particles formed at higher temperature were smaller than those formed at lower temperature. Excitation tuneable photoluminescence was observed for all the samples with quantum yield (QY) 1.2, 0.55 and 0.63%, for CNPs from bread, jaggery and sugar caramels respectively. The present discovery suggests potential usefulness of CNPs for various biological applications, as the sources of extraction are regular food items, some of which have been consumed by humans for centuries, and thus they can be considered as safe. T he use as well as presence of nanoparticles (NPs) in food is a hotly debated area, owing to their short and long term effects on human health and the environment . The promise of targeted and/or sustained release of drug, food colourants and flavours, while incorporated with NP, makes the pursuit of understanding of their functioning and fate a worthy exercise . Although, substantial development in the engineering of consumable NPs and their effects in vitro and in vivo have taken place 8 ; few biodegradable NPs have entered clinical trials and have been marketed . While NP formulations for topical applications are accepted by majority of population 11 , the idea of the consumption of these particles, either for curing a disease or for having nutritional or flavouring benefit, creates an alarm for the public. The reason behind this seems to be their potential effect on human health following consumption, which has received little attention; and the lack of awareness, which has raised concerns regarding the safety of nanomaterials in biological and food applications . A way around this problem, could originate out of direct derivation of nanomaterials from food products, especially from traditional food items. These materials could be considered safe for biological applications when there is no known toxicity and thus may possibly alleviate the misapprehension that all NPs are toxic. History of nanotechnology is replete with examples of use of nanomaterials dating back to millennia . The dye used in colouring hair to black, during the Greco-Roman period, is now known to have been consisted of PbS nanocrystals (NCs) 12 . Romans exhibited their mastery in technology in the Lycurgus cup by harnessing the optical properties of gold (Au) NPs 13 . The extraordinary mechanical strength and a sharp cutting edge in Damascus sabre have recently been attributed to the presence of carbon nanotubes (CNTs) and cementite nanowires 14 . In all the cases mentioned above, while the technology based on nanomaterials were known to different civilizations, the nanoscale nature of their functional constituents have only been revealed recently. The 'nano' dimensionality is not only confined to engineered materials or technology; nature also creates NPs or nanostructures which are present as functional components in an organism; either in the form of enzymes which catalyze most of the biological reactions or as ribosomes which act as the sites for protein synthesis. It was the invention of sophisticated microscopic and analytical techniques which has led to the discovery of these nanostructures. In this regard, the idea of searching for nanomaterials within regular food items cannot be inexplicit. This motivated us to search for NPs in food items, which can potentially be used for biological application, where the concern of the origin and toxicity of the nanomaterials can easily be waived off. Herein, we report the presence of carbon nanoparticles (CNPs) in regular carbohydrate based food caramels, such as in bread, jaggery, corn flakes and biscuits. The CNPs have been found to be present in those samples, where the preparation of food mainly involves heating of the starting ingredients in absence of water, leading to formation of caramels. Arguably; this discovery revealed that human consumption of nanomaterials in the form of food caramels has its history possibly from the period when human for the first time started eating bread. Carbon dots (C-dots), which are CNPs below 10 nm are emerging as viable alternatives to semiconductor quantum dots (Qdots) owing to their important photoluminescent properties and lack of any known cytotoxicity . The wavelength-tuneable emission properties have made them promising candidates as new 'nanolights' 17 . The optimism has led to increased recent interests in developing methods for their syntheses, involving both top-down and bottomup approaches 17 . Incidentally, there are also efforts to understand and tune their optical properties, based on the surface functional groups . We have recently observed that caramelization of poly (ethylene glycol) under microwave irradiation constitutes formation of biocompatible C-dots 15 . This prompted us to analyse the components of different commercial and homemade caramel containing food items for the presence of CNPs. Amazingly, we found that the light to dark brown coloured caramels present in carbohydrate containing foods such as bread, jaggery, corn flakes and biscuit consist of amorphous CNPs, which were similar to those obtained from caramel produced upon heating of commercial sugar. The discovery of the prominent presence of CNPs in regular food items provides simpler and safer sources i.e. daily food items, where the extracted CNPs can be directly used for biological applications. Further, our observations of the hydrophilicity of the surface functional groups of the CNPs, containing carboxylic and alcoholic groups which provide easier alternatives for their conjugation with different therapeutics, could further make them preferred fluorescent candidates for biological applications . ## Results The CNPs isolated from the outer brown part of the bread bun and the caramels obtained from commercially purchased sugar (following caramelization) and jaggery were analysed by UV-vis and fluorescence spectroscopy. The UV-vis spectrum of each of the dispersions consisted of a peak (marked with asterisk) and a shoulder (marked with arrow) between 240 nm and 400 nm and is shown in Figure 1. In addition, there is the presence of a strong background till 540 nm. The peaks and the background extinction are known to occur for CNPs and they are consistent with the literature reports 17 . The exact assignment of the peaks is still not known and hence the difference among the individual samples could not be explained. CNPs were also extracted from commercially procured biscuits and corn flakes, the spectroscopic data for which are shown in Figure S1 (supplementary information). Photographs of the original samples (bread, jaggery and sugar) which served as the preparatory ingredients are shown in Figures 2a, 2b and 2c. The photoluminescence spectra corresponding to the above dispersions (and other samples) are shown in Figures 2d-2i (and Figure S1). The pictures of the dispersions in the presence of white light and UV light are represented in Figures 2j-2o. Under normal white light the dispersions have the characteristic caramel colour, whereas, under UV light (l ex 5365 nm) it showed blue luminescence. All of the dispersions exhibited excitation dependent emission spectra as shown in Figures 2g-2i, which were similar to C-dots reported previously 17 . It was also observed that with increase in the wavelength of excitation from 325 to 375 nm the luminescence intensity increased, the maximum emission intensity being observed for the excitation wavelength of 375 nm, whereas, further increase of the excitation wavelength resulted in the decrease of emission intensity. Additionally, along with decrease in the fluorescence intensity with increasing excitation wavelength the emission maxima showed red-shift, displaying the property of excitation tuneable emission. The excitation dependent emission is an intrinsic property of CNPs, which has been widely reported by several research groups, including us 15,17, . The quantum yields (QYs) of the CNPs obtained from different food sources were calculated using quinine sulphate as the standard 25 . At an excitation wavelength of 365 nm, the QYs of the CNPs are summarized in Table 1. The results indicated that these samples had QY typical of C-dots, which is on the order of 1%; with the highest being observed for samples from bread (1.2%) and that from jaggery had the lowest value (0.55%). Transmission electron microscopy (TEM) images of the samples obtained from the dispersions of different caramel sources (bread, jaggery and sugar) are represented in Figures 3a-3c, which showed the presence of uniform spherical NPs. The particle distributions calculated from the images are shown in Figures 3d-3f. The average particle sizes as calculated from the TEM images for samples from bread, jaggery and sugar caramel, were determined to be 27.5 6 6.1, 20.3 6 7.5, 4.3 6 1.5 nm respectively. Similarly, samples from corn flakes and biscuits indicated the presence of NPs having sizes of 10.5 6 2.8 nm and 3.9 6 1.3 nm respectively (Figure S1, supplementary information). The results clearly indicated that NPs were present in the dispersions extracted from bread, jaggery, caramel of sugar and other materials. While the sample from bread had the highest particle size, the particles from sugar caramel produced at 200uC had the lowest size and the particle sizes of the sample from jaggery were in between the two. In addition, caramels from sugar, produced by heating at 180uC for 10 min, had particles of size 25.8 6 12.4 nm (Figure S2, supplementary information). Thermogravimetric analysis of sugar indicated decomposition starting at below 200uC with steady decrease in weight till 350uC (Figure S3, supplementary information). Loss of weight signifies the dehydration process of carbohydrate or formation of CO 2 . Thus, the NPs could possibly be produced at a temperature even lower than 200uC. Samples from bread, jaggery and caramel showed broad X-ray diffraction (XRD) peak at about 2h 5 18u (Figure S4, supplementary information), with no clear signature for crystalline nature of any of the samples. The above results indicated that NPs present in the caramels of bread, jaggery, corn flakes, biscuits and sugar possibly consisted of amorphous carbon. Further, 13 C NMR (nuclear magnetic resonance) spectra of samples from bread, jaggery and caramel (Figures S5, S6 Further, in order to probe the extent of cytotoxicity of the extracted CNPs, we performed XTT based cell viability assay at varying concentrations of CNPs. The plot of percentage viability of cells to that of varying concentration of CNPs (0.05 mg/mL to 2.0 mg/mL) is shown in Figure S11. As is clear from the figure, no cytotoxicity was apparent even at the highest concentration of CNPs (2.0 mg/mL) used. In addition, one way ANOVA showed that the differences in the mean percentage viabilities of cells at different concentrations of CNPs extracted from jaggery (F5 0.652, P5 0.689) and bread (F 5 1.152, P 5 0.384) were not statistically significant. ## Discussion The UV-visible and fluorescence spectra of the dispersion of CNPs extracted from different food sources displayed features similar to those of C-dots synthesized chemically and thereby suggesting the presence of CNPs in the samples. The fundamental mechanism of photoluminescence of CNPs is still a major question; however, it is thought that the presence of different surface trap sites could be one of the factors for the luminescence. 17,22 The origin of fluorescence from the obtained dispersion could only be attributed to the presence of CNPs because the analysis of the starting material for preparation of bread did not show any significant fluorescence (Figure S8, supplementary information). Sugar is known to be a nonfluorescent material, but the caramel prepared upon heating sugar showed the emergence of excitation tuneable luminescence, further confirming the formation of CNPs. Additionally, it was observed that the heating temperature for preparing the caramel had significant effect on the size of NPs formed. Caramels prepared at 180uC and 200uC had the sizes of 25.8 6 12.4 and 4.3 6 1.5nm respectively. This indicated that smaller particles were possibly formed at the higher temperature. In other words, the larger particle sizes of NPs obtained from bread and jaggery could be due to their low heating temperatures, whereas, the smaller particles sizes of NPs obtained from sugar caramel, corn flakes and biscuits could be due to higher heating temperatures. It may also be mentioned here that there could be other factors, such as the rate and duration of heating and chemical constituents of the samples, determining the sizes of the produced CNPs. The possibility of the formation of CNPs while preparing and analysing the sample in electron microscopy can be ruled out because when drop cast sample from sugar solution was observed in TEM no such particle formation was detected, even under the exposure of a 200 kV electron beam for several minutes. The images obtained at different time of irradiation, of the sample from sugar solution, in the electron beam of TEM are shown in Figure S9 (supplementary information). It is worth noting that similar extraction process was also used for determining the presence of CNPs, if any, in the interior white part of bread. TEM analysis revealed the presence of inhomogeneous particles (Figure S10a, supplementary information) which could be due to the suboptimal temperature in the inner zone. The fluorescence intensity of this dispersion was significantly low compared to that obtained from the brown part of the bread (Figure S10b, supplementary information). The size of the CNPs produced varied from sample to sample, indicating the possibility of heating temperature as the primary factor determining their sizes. However, it was interesting to observe that for all samples the particles produced were nearly uniform and spherical. To have an idea of the amount of CNPs which can be extracted from a food source we analysed the amount of particles obtained from 1 g each of jaggery and the brown layer of bread. It was observed that about 3 and 2 % w/w of CNPs, in the respective samples were present. The calculation is based on the amount of the starting ingredient taken for the isolation of CNPs and the sample recovered after purification. However, the amount recovered from these materials cannot solely be attributed to CNPs as polymeric layer will always remain surrounding these particles. Isolation of nude CNPs without the polymeric layer has not been possible in the present condition; even then it can give an approximate value about the fraction of particles extracted. Amorphous nature of NPs present in all the samples is demonstrated by the results of powder XRD data (Figure S4, supplementary information) as no peaks of crystalline origin was detected. The NMR studies revealed that the CNPs were coated with hydrophilic carbohydrate units. No peaks corresponding to the aromatic region was observed, which again supported the luminescence to be originating from the CNPs present in the dispersion. In summary, our current work revealed the presence of CNPs in carbohydrate based regular food caramels from bread, jaggery, corn flakes and biscuit. The excitation wavelength dependent emission characteristic of the CNPs from food caramels were similar to those generated from sugar; however, the particle sizes varied indicating temperature -dependent formation of CNPs of different sizes. NMR spectroscopy revealed that the CNPs were coated with carbohydrate units. It is interesting to note that for centuries these caramels containing CNPs have been consumed by human beings with no known toxicity and thus it can be considered to have no or minimum risk on human health and may be used as a safe nanomaterial. Our finding of the presence of fluorescent CNPs in food caramels may also help their use in tracking and imaging conjugated biomolecules and drugs in vivo, without being imperilled. ## Methods Preparation and extraction of CNPs. Bread buns were purchased from the local market (Homa Bread, Guwahati, India) and analysed to check the presence of CNPs within it. The top brown layer of bread was carefully excised and 1 g of it was dissolved in 20 mL methanol by sonicating it at 35 kHz in a bath sonicator (Elmasonic TI-H-5 Elma, Germany) for 10 min. Following sonication, the volume of the methanol was reduced to 3 mL in a rotary evaporator before further purification. Jaggery (prepared from sugarcane juice) purchased from the market was heated following a traditional procedure which is as follows. Initially, jaggery (say 50 g) was mixed with water (about 10-15 mL) to make it sufficiently moist. The entire amount was then placed on a hot pan, which was being heated in the medium flame of a commercial gas stove. The mixture was constantly stirred using a kitchen spatula. In about 5 min, when the colour of the mixture turned dark brown, the entire amount was transferred to a pan containing a thin layer of oil and brought to room temperature. The oil layer helped in preventing aggregation of the mass and also in spreading the content over the pan. The jaggery caramel, which was ready for use then, was dissolved in methanol and allowed to stand for a few minutes to remove larger impurities. The sedimented particles from both the samples were removed by filtration. Centrifugation of the supernatant at 5000 rpm was performed further to remove impurities of smaller size. The yellow coloured supernatant thus obtained was further purified by column chromatography (using silica 60-120 mesh) with methanol:dichloromethane (2:3) mixture and finally dialysis (using 1 KDa membrane) was carried out to remove salts and other ions, if any. Similar procedures were followed for the extraction of CNPs from commercially purchased cornflakes and biscuits. Caramel was also prepared in laboratory by heating commercially available sugar. Sugar (Daurala Sugar Works, India) was taken in a glass vial and heated in an oil bath at 200uC for 10 min till the solid turned brown. The brown colored sticky mass was cooled down to room temperature and was then dissolved in methanol, followed by purification using column chromatography. The sample was then concentrated by evaporating the solvent in rotary-evaporator and then it was dissolved in water and finally dialysed before further analysis. Characterization of CNPs. The extracted CNPs were characterized using TEM (JEOL 2100 UHR-TEM), UV-vis (Perkin Elmer Lambda 25) and fluorescence spectrophotometers (FluoroMax-4, Jobin Yvon). The TEM analysis was performed at an accelerating voltage of 80 kV, unless otherwise mentioned, and the sample was prepared by drop casting 5 mL of the respective sample on a 300 mesh carbon coated copper grids and subsequent air drying before analysis. 13 C NMR (100 MHz) of the dispersion was carried out in a Varian 400 MHz FT-NMR using D 2 O as the solvent. Thermogravimetric analysis (TGA) was performed in Q600 SDT Simultaneous DSC-TGA heat flow analyzer and powder XRD study was done using a Brucker D8 Advanced X-ray diffraction measurement system, with Cu Ka source (l 51.54 A ˚). The quantum yield (QY) was calculated using quinine sulphate in 0.10 M H 2 SO 4 solution as a standard, at an excitation wavelength of 365 nm, and the absorbance was kept below 0.15. The QY of the samples were determined according to equation 1. Where, QY is the quantum yield, m is the slope of the plot of integrated fluorescence intensity vs absorbance and g is the refractive index of the solvent. For the aqueous solutions g S /g R 51. The subscript R refers to the reference fluorophore i.e. quinine sulphate solution and S for the sample. The values obtained are given in Table S1, supplementary information. In vitro cytotoxicity assays. HeLa cells were obtained from National Center for Cell Sciences (NCCS), Pune, India and were cultured in Dulbecco's modified Eagle's medium supplemented with penicillin (50 units mL 21 ), streptomycin (50 mg mL 21 ), and 10% (v/v) fetal bovine serum. Cells were maintained in 5% CO 2 humidified incubator at 37uC. The cells were seeded in a 96 well culture plates at a density of 5000 cells/well and were allowed to grow overnight. The CNPs were then added into the wells in a concentration range of 50 ng/mL to 2 mg/mL and incubated in a humidified incubator for 24 h at 37uC and 5% CO 2 . XTT (Bis(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide) (Sigma-Aldrich, USA) based cell viability assay was carried out according to the manufacturer's protocol, to determine the percentage of viable cells. The assay is based on the metabolic activity of the cells to reduce the tetrazolium salt XTT to orange coloured compounds of formazan and the intensity of the dye is proportional to the number of metabolic active cells. The percentage cell viability of the untreated cells (control) was taken as 100%. All measurements were collected in triplicate and the values are expressed as mean 6 standard error (SE). Statistical analysis for ANOVA was performed using Sigma Plot.

chemsum

{"title": "Presence of Amorphous Carbon Nanoparticles in Food Caramels", "journal": "Scientific Reports - Nature"}

fumarate-based_metal-organic_frameworks_as_a_new_platform_for_highly_selective_removal_of_fluoride_f

## Abstract: Adsorption and removal of fluoride from brick tea is very important but challenging. In this work, two fumarate-based metal-organic frameworks (MOFs) were synthesized for the selective removal of fluoride from brick tea infusion. MOFs were examined for adsorption time, effect of dose, and uptake capacity at different initial concentrations and temperatures. Remarkably, over 80% fluoride removal was achieved by MOF-801 within 5 min at room temperature, while no significant adsorption occurred for the catechins and caffeine in the brick tea infusion. Further, with the use of the Langmuir equation, the maximum fluoride uptake capacity for the nontoxic calcium fumarate (CaFu) MOF was calculated to be as high as 166.11 mg g −1 at 373 K. As observed from FTIR, EDX and XPS results, hydroxyl group in MOFs were substituted by fluoride. This work demonstrates that the novel fumarate-based MOFs are promising materials for the selective removal of fluoride from brick tea infusion. Tea is a popular and healthy beverage due to that drinking tea has numerous health benefits, such as lower cardiovascular risk, reduce body fat, as well as decrease the risk of tumors 1,2 . However, Tea plants can accumulate and store a large amount of fluoride in mature leaves by adsorbing it from the soil and air without toxicity symptoms 3 . An abundance of fluoride can be released during tea infusion and the bioavailability of the released fluoride is nearly 100% to consumers, owing to the soluble fluoride ions from tea are easily adsorbed through the gastrointestinal track 4 . As we all known that fluoride is an essential element to mammals and a moderate amount of fluoride helps bone development whereas excessive intake of fluoride can lead to various diseases such as dental and skeletal fluorosis 5,6 . The fluoride content in tea is thought to be safe and could contribute to human health when the concentration of fluoride at low levels of 100-300 mg kg −1 , but it contained in some special tea (i.e., brick tea) is usually extremely high with levels up to 600 mg kg −1 7 . Brick tea based fluorosis is mainly found in the northwestern of China, such as Qinghai, Tibet, Inner Mongolia, and Sinkiang, where most of minorities are habitual consumer of brick tea with high level of fluoride 8 . To date, brick tea type fluorosis is still considered as a severe health problem in some parts of China, due to it is impossible to change these minorities brick tea habitual consumption. Hence, it is necessary to develop suitable methods to strictly control and remove the fluoride from tea. It is well known that adsorption method is one of the most applicable methods due to its low cost and simple operation 9,10 . Recently, Zhao et al. reported on a plant polyphenol-Ce hybrid adsorbent for the effectively adsorption of fluoride during the period of tea plants growing 7 . However, for these plant polyphenol-Ce-based adsorbents, it was difficult to separate the adsorbents from the soil and their fluoride adsorption capacity is still low, making them difficult to widespread industrial use. Therefore, a simple strategy to develop high capacity and selective fluoride adsorbents is highly desired. As a new class of crystalline inorganic-organic porous hybrid materials, metal-organic frameworks (MOFs) have emerged as one kind of promising material to develop novel adsorbents 11,12 . Due to the exceptional internal surface area, tailored structure, tunable pore architectures combined with diverse framework functionalities, MOFs experienced fast development and have displayed a vast range of promising applications such as catalysis 13 , gas storage and separation 14 , drug delivery 15 , as well as adsorption and removal of hazardous materials 16 . For the adsorption of contaminant-related applications, MOFs have been widely exploited for the selective adsorption and removal of toxic dyes 17 , pharmaceuticals 18 , nitrogen compounds 19 , sulfur compounds 20 , and heavy metal ions 21 , because the pore size and shape of MOFs can be easily controlled to facilitate the uptake of targeted guest molecules. Moreover, in the recent years, a few pioneering studies demonstrating the promise of MOFs in the removal of fluoride from water have also been reported. For example, Liu et al. first reported on the fabrication of MIL-96(Al) for the selective defluoridation of drinking water 22 . Later, Lin and co-workers reported on an amine-functionalized zirconium MOF (named UiO-66-NH 2 ) used as enhanced adsorbents for fluoride removal 23 . Further, De groups reported a promising aluminium fumarate MOF (AlFu) adsorbent for the removal of fluoride from groundwater 24 . These MOFs-based porous adsorbents exhibit fast and excellent adsorption abilities for the fluoride removal from water system. However, to date, employment of MOFs as adsorption materials for the selective removal of fluoride from tea system has not been reported. Herein, we report two fumarate-based MOFs (i.e., MOF-801 and CaFu) for the highly selective adsorption of fluoride from brick tea infusion. As a subfamily of porous MOFs materials, a series of zirconium(IV)-based MOFs (Zr-MOFs) have been developed since 2008 due to their inherent thermally and chemical stability 25 . MOF-801 is the smallest member of Zr-MOFs with a formula of Zr 6 O 4 (OH) 4 (fumarate) 6 and fcu topology 26 . MOF-801 possesses hydrophilic adsorption sites since plenty of hydroxyl groups are involved in the nodes of Zr(IV). Given the presence of zirconium-bound hydroxyl groups in the nodes of framework which are expected to facilitate the adsorption of fluoride via the anion exchange behavior. Impressively, the prepared MOF-801 exhibits excellent adsorption performance and stability toward fluoride removal from brick tea infusion. Furthermore, a homologous nontoxic calcium fumarate (CaFu) MOF was also synthesized and employed to efficient remove fluoride from tea infusion. These fumarate-based MOFs not only show high efficiency and selectivity towards the fluoride removal, but also experimentally simple and easy to handle for the uptake of fluoride by using tea bag model. To the best of our knowledge, this is the first example of MOF-based adsorbents for the selective adsorption of fluoride from brick tea infusion. ## Results and Discussion Synthesis and characterization of the fumarate-based MOFs. MOF-801 is a typical microporous Zr-based MOF consisting of Zr 6 nodes bridged by fumarate linker to give the three dimensional (3D) structure (Fig. 1a) 27 . Fumaric acid was chosen as the organic linker in this work because it is an important biologically occurring molecule and could be used as a typical food additive 28 . MOF-801 has two crystallographically independent tetrahedral cage sizes of 5.6 and 4.8 and one octahedral cage size of 7.4 26 . The phase purity of the as-synthesized product was checked by PXRD. For comparison, the pattern of the simulated from the crystallographic data of MOF-801 is also shown. As shown in Fig. 2a, all of the experimental PXRD pattern diffraction peaks are well matched with the MOF-801 standard literature values 26 . No miscellaneous PXRD peaks were observed, indicating that the as-synthesized MOF is undoubtedly MOF-801. The XPS was employed to identity the chemical compositions, especially the valence state of the elements of the MOF. The XPS full spectrum, as presented in Fig. 2c, confirms that the existence of C, O, and Zr in the MOF-801 sample. The high-resolution of Zr 3d spectrum is shown in Fig. 2d, the binding energy peaks located at around 182.89 and 185. 21 eV can be ascribed to 3d 5/2 and 3d 3/2 of Zr(IV), which is similar with the reported Zr-based MOFs 29 . Therefore, the results suggest that these peaks only belong to Zr(IV) in the MOF-801 framework. The N 2 adsorption-desorption isotherm was also implemented to measure the surface area of MOF-801 and the result is displayed in Fig. 2b. The N 2 adsorption-desorption isotherm of MOF-801 exhibits typical type-I behavior (Fig. 2b), which is related to microporous material. The Brunauer-Emmett-Teller (BET) surface area is 755 m 2 g −1 , and the calculated pore volume is 0.44 cm 3 g −1 . The morphology of the sample was then characterized by SEM and TEM. These SEM images reveal that the obtained MOF-801 nanoparticles (NPs) are of spherical shape with a uniform size and good dispersity (Fig. 3a and b). The morphology of these NPs were further identified by TEM (Fig. 3c and d). One can see that the MOF-801 nanospheres exhibit a narrow size distribution, which are in good agreement with the SEM observation. Preliminary statistics based on the product shown in Fig. 3 indicates that the average size of the MOF-801 nanospheres is 150 nm. ## Adsorption behavior of fumarate-based MOFs for fluoride removal from brick tea infusion. Studies regarding the removal of fluoride from water system have shown that porous MOFs are good adsorbents for the adsorption of fluoride 30 . However, to date, investigations used MOFs to adsorb fluoride from tea infusion are still scarce up. Therefore, in this study, MOF-801 adsorbent was checked for the adsorption of fluoride from brick tea infusion. The pH of the solution plays a key factor in the fluoride removal, which influences the surface charge of the adsorbents. In order to evaluate the influence of pH on the adsorption of fluoride, 40 mg of MOF-801 was suspended in 25 mL of brick tea infusion with the initial fluoride concentration of 8 mg L −1 at various pH (from 2 to 8). As can be seen from Fig. 4a that the adsorption is higher at a lower pH and drops drastically after pH 5. This can be attributed to the fact that at pH 5.5, the MOF NPs become neutral in charge and at a higher pH, the adsorbent NPs become negatively charged as shown in Fig. 4b. In acidic pH conditions, the MOF-801 is positively charged facilitating the adsorption of fluoride. At pH 6, the fluoride adsorption was reduced, which due to the competitive adsorption of OH − in the brick tea infusion. Since we are concerned about drinking tea and the pH value of the actual prepared brick tea infusion is 5.4, we did not adjust the pH during the following experiment. To deepen the understanding of the equilibration time for maximum uptake of fluoride and to have a better understanding of adsorption kinetics, the adsorption of fluoride on MOF-801 from brick tea infusion was investigated as a function of contact time. As shown in Fig. 4c, 40 mg of MOF-801 is used as adsorbents to capture 25 mL brick tea infusion with the initial fluoride concentration of 8 mg L −1 . Accordingly, the fluoride adsorption capacity increased rapidly during the first 2 min and gradually attained the adsorption equilibrium only in 5 min. In can be found that MOF-801 can remove nearly 80% of the fluoride ions present in the respective brick tea infusion within 5 min. Tea has been well studied for its health benefits on human because tea leaves contain large amounts of catechins, including (−)-epicatechin gallate (ECG), (+)-catechin (C), (−)-epigallocatechin gallate (EGCG), (−)-epicatechin (EC), (+)-gallocatechin gallate (GCG), and (−)-epigallocatechin (EGC), which are typical powerful antioxidants 2 . As can be seen from Fig. 4d, no significant loss of catechins and only minor loss of caffeine (Caf) can be observed within 30 min, indicating that the MOF-801 adsorbents possess excellent selective adsorption and removal of fluoride from brick tea infusion. The residual zirconium concentration of the brick tea infusion was determined with ICP. Significantly, there was no residual zirconium ion can be detected in the brick tea infusion. The results indicate that MOF-801 will be a promising candidate to adsorption of fluoride from brick tea infusion. Further, the adsorption kinetic data was modeled by the pseudo-second-order kinetic equation (Equation S1) 31 . As can be seen from Fig. S1, the parameter values of this kinetic model can be calculated with the plot of t/q t versus t, and the obtained correlation coefficient is 0.9999, revealing that the selective removal of fluoride from brick tea infusion onto the MOF-801 frameworks follows this kinetic model very well. The kinetic rate constant (k 2 ) and the equilibrium adsorption capacity (q e ) values of MOF-801 under this brick tea infusion condition were determined to be 0.69 g mg −1 min −1 and 3.91 mg g −1 , respectively. The adsorption capacity is important for the application of adsorbents. The amount of fluoride adsorbed per gram of MOF-801 was investigated by exposing the MOF to the brick tea infusion under a wide range of fluoride concentrations. In order to make sufficient time for removal to occur, tea infusion were tested after 60 min of exposure. It was found that the adsorption capacities remarkably increased as the fluoride initial concentration increased in the brick tea infusion (Fig. 5a), suggesting the favorable selective adsorption of fluoride by MOF-801 at high concentrations. Further, to quantitatively predict the adsorption capacity of MOF-801, we employed the Langmuir model to fit the adsorption isotherm data and high correlation coefficients can be obtained from Fig. S2. In the case of Langmuir model, the adsorption process of adsorbent is occurred as a mono-layer over the adsorbent surface 23 . When the adsorption sites of MOF-801 are occupied by the fluoride in the tea infusion, then these occupied sites cannot be used for the adsorption anymore. Thus, the maximum adsorption capacity of MOF-801 for fluoride can be calculated using the Langmuir equation (Equation S2) 23 . Figure S2 shows the plots of C e /q e versus C e over MOF-801 for fluoride removal at different temperatures, and the maximum adsorption capacity values (q m ) can be determined from the slops. According to Equation S2, the q m of MOF-801 for fluoride in the brick tea infusion is 32.13 mg g −1 at 298 K (Table 1). This value is superior to that of the many conventional fluoride adsorbents , and even comparable to some of the novel MOF based adsorbents which were used in the simple water system 22,23,36 . Additionally, as displayed in Fig. 5a, the fluoride adsorption capacity over MOF-801 increases with increasing the temperature, and the q m becomes 38.60 and 45.72 mg g −1 at 308 and 318 K, respectively (Table 1). This trend shows the positive effect of MOF-801 on the adsorption isotherm at higher temperatures, suggesting that the removal of fluoride of MOF-801 in tea infusion could be endothermic reaction . To gain a better understanding of the thermodynamic feasibility and the adsorption process, three basic thermodynamic parameters like Gibbs free energy change (ΔG) (Equation S3), entropy change (ΔS) (Equation S4), and enthalpy change (ΔH) (Equation S4) were calculated from the removal of fluoride on MOF-801 by using standard methods 22 . As shown in Table 1, the obtained ΔG values of MOF-801 are -1.99, -2.25, and -2.49 kJ mol −1 at 298, 308, and 318 K, respectively. The negative values of ΔG at all temperatures reveal that the adsorption process of fluoride to MOF-801 can be spontaneous adsorption within this temperature range. Significantly, ΔG value becomes more negative when the temperature is increased, suggesting that the adsorption process is more favorable at high temperatures. The Van't Hoff plot, constructed according to Equation S4, gave straight line which is shown in Fig. 5b, and the values of ΔS and ΔH can be determined from the intercept and slop of the plot. It can be seen from Table 1 that ΔH and ΔS were determined to be 4.54 kJ mol −1 and 21.97 J mol −1 K −1 . As we know that the value of ΔH range 2.1-20.9 kJ mol −1 is regarded as physical adsorption while range 20.9-418 kJ mol −1 can be corresponded to chemical adsorption 40 . Thus, the ΔH value in this work implies that the fluoride adsorption process occurred by MOF-801 due to the physical adsorption. The positive ΔH value (Table 1) indicated that the adsorption of fluoride over MOF-801 was an endothermic process, which was in accord with the increasing adsorption capacity associated with increasing adsorption temperature (Fig. 5a). The endothermic process may be due to a stronger interaction between pre-adsorbed water and the MOF than the interaction between fluoride and the MOF. At the same time, the obtained positive value of ΔS (21.97 J mol −1 K −1 ) further confirmed that the increase of randomness at the solid adsorbent/tea infusion solution interface during the fluoride adsorption reaction over MOF-801. This phenomenon can be attributed to the released water molecules at the interface is greater than the adsorbed fluoride ions by the MOF-801 adsorbents 41 . Therefore, the driving force of fluoride adsorption (negative ΔG) on MOF-801 is due to an entropy effect (positive ΔS) rather than an enthalpy change (positive ΔH). The reusability is one of the important issues for the practical application of adsorbents. A crucial problem in the use of adsorbents is that they suffer from the low efficiency of separation. The tea bag model described here is designed to overcome these challenges. In a simple and easy to use design, a tea bag containing MOF-801 NPs were prepared and dipped in fluoride contaminated brick tea infusion as shown in Fig. S3b. After adsorption, the tea bag was first removed from the brick tea infusion, washed with diluted NaOH (0.01 M) and water, and then dried at 70 °C. After this treatment, the tea bag containing MOF-801 NPs were used again for other consecutive runs under the same adsorption conditions for 1 h. As shown in Fig. S3a, no significant loss in the adsorption efficiency of fluoride from brick tea infusion can be observed in the subsequent five consecutive cycles, indicating that the MOF-801 adsorbents possess excellent long-term adsorption stability and could be reused for multiple rounds. Since the adsorbent dose is an important factor for the control of fluoride removal efficiency, the parameter of MOFs dose were tested and the results are presented in Fig. 6. In an easy and simple method for practical application, the effects of MOFs dose were carried out by exposing MOFs and brick tea leaves to deionized water directly with boiling at 373 K for 30 min. It was obvious that the final adsorbed percentage of fluoride increased with the amount of MOF-801 (Fig. 6a). As the amount of MOF-801 increased from 0.4 to 2.0 g L −1 , the efficiency of fluoride uptake gradually increased from 18% to 70%. The higher fluoride adsorption efficiency at the higher MOF-801 dose was due to the more active sites of MOFs available present in the tea infusion. Significantly, the losses of the catechins and caffeine were all lower than 5% (Fig. 6c), suggesting that the MOF-801 adsorbents could highly selective adsorption of fluoride from brick tea infusion. When the amount of MOF-801 increased to 4.0 g L −1 , the efficiency of fluoride removal went up to 92%. However, the losses of the catechins and caffeine were increased to around 20% in this conditions. This comparison implies that the MOFs dose is a key factor for selective adsorption of fluoride from brick tea system. Although a higher dose of MOFs is beneficial to removal of fluoride, an over MOFs dose must be avoided due to the increase of catechins and caffeine loss at higher dose. Therefore, it is obviously that the best MOF-801 dose range for the selective removal of fluoride from brick tea infusion is below 2.0 g L −1 . One may criticize the fact that the Zr-based MOFs adsorbents were not bio-compatible for practical applications, nevertheless, homologous nontoxic calcium fumarate (CaFu) MOF was also synthesized and tested to obtain a good performance in the field of fluoride removal from tea infusion. The structure of the as-synthesized CaFu was characterized by PXRD (Fig. S4) and SEM (Fig. S5). It is clearly seen that the CaFu material consists of irregular shape particles with the size around 4.5 μm. Similarly, a high CaFu dose exhibited a high fluoride adsorption from the tea infusion and the percentage of fluoride removal increased to 37% with 2.0 g L −1 of CaFu (Fig. 6b). Although this value is lower than that of above obtained MOF-801, it is still superior to that of Tea-Al biosorbent which we reported recently 42 . In our previous work, we have reported the synthesis of aluminum oxide decorated tea waste based biosorbent (e.g., Tea-Al), which is promising for the fluoride removal from the brick rea infusion 42 . However, a critical drawback of Tea-Al is non-selective fluoride removal from brick tea infusion. The fumarate-based MOFs adsorbents described here are designed to overcome this challenge. As shown in Fig. 6d, no significant losses of the catechins and caffeine were observed with the dose of CaFu below 2.0 g L −1 . Furthermore, the initial fluoride concentration-dependent removal capacity was also obtained to investigate the adsorption isotherm of fluoride on CaFu adsorbent. 30 mg of CaFu and 0.5 g of brick tea were mixed with 25 mL of 8-512 mg L −1 fluoride solution. The adsorption isotherms of CaFu adsorbent were obtained after boiling in tea infusion for 30 min at 373 K. Figure 7 shows that the adsorption capacity of CaFu also increased as the initial concentration of fluoride increased in the tea infusion. As displayed in the inset of Fig. 7, the isotherm data fit the Langmuir model well, and the correlation coefficient is 0.9886. Remarkably, the maximum adsorption capacity of CaFu for fluoride in the brick tea infusion is 166.11 mg g −1 at 373 K. To date, there have been only two pioneering studies on the fluoride removal from tea infusions (e.g., Tea-Al 42 and Fe 3 O 4 /Al 2 O 3 -PUF 43 ). The maximum adsorption capacity value of CaFu is the highest value ever reported for fluoride removal from the brick tea infusion system 42,43 . The present work may provide potential of synthesis of such nontoxic MOFs-based adsorbents for application in fluoride removal from brick tea. To shed light on the mechanism of fluoride adsorption on MOFs, FT-IR (Fig. 8a), EDX (Fig. S6), and XPS spectra (Fig. 8b-d) were used to characterized CaFu before and after adsorption of fluoride. Prior to adsorption, the IR spectrum of CaFu contains two strong bands around 1594 and 1405 cm −1 corresponding to the -O-C-Ogroup, suggesting that the Fu species is coordinated to the Ca atoms. The sharp band of CaFu around 3465 cm −1 and a small band around 2745 cm −1 are assigned to the stretching of Ca nodes terminal -OH group and the hydrogen-bonding between the -OH in the Ca nodes and aqua, respectively 44 . After adsorption of fluoride, the stretching of -OH group at 3465 cm −1 is remarkably diminished and the -OH stretch of the hydrogen-bonding based at 2745 cm −1 disappeared completely (Fig. 8a). Furthermore, the FT-IR spectra were also used to characterize MOF-801 before and after fluoride adsorption (shown in Fig. S7). The FT-IR spectra results are similar with what was observed on the CaFu before and after adsorption of fluoride systems. Based on these observations, we propose a simple mechanism displayed in Fig. 1b for fluoride removal over MOFs: first, fluoride ions can be adsorbed onto the porous fumarate-based MOFs via interactions between the fluoride ions and the activity metal center in the framework. There are abundant of hydroxyl groups around the nodes of MOFs. Then the fluoride replaces hydroxyl group on the metal-node in the structure of MOFs through the anion exchange behavior. (Fig. 8c), which is similar with that of Ca 2+ ions 45 . For the sample of CaFu after fluoride adsorption, apart from those binding energy peaks belonging to pure CaFu frameworks, singles of F 1 s appeared at 684.8 eV can be detected (Fig. 8b,d). In particular, it is worth to mention that Ca 2p 3/2 and Ca 2p 1/2 of the sample of CaFu after fluoride adsorption were shifted to 348.3 and 351.9 eV (Fig. 8c), demonstrating that the bonding environment of Ca nodes was changed after adsorption of fluoride. These results are good consistent with the IR observation, and further confirming that the adsorption reaction depended on the fluoride and the Ca-node coordinatively unsaturated centers of CaFu. ## Conclusion In summary, two fumarate-based MOFs have been synthesized and used in the highly selective removal of fluoride from brick tea infusion. The adsorption capacity of MOF-801 for fluoride from the tea infusion was 32.13 mg g −1 at 298 K. Besides, the adsorption capacity of CaFu was 166.11 mg g −1 at 373 K. Furthermore, the two fumarate-based MOFs showed a highly selective fluoride adsorption from the tea infusion and no significant losses of the catechins and caffeine were observed with the dose of MOFs below 2.0 g L −1 . FTIR and XPS results point to the key importance of numbers of node-based coordinatively unsaturated adsorption sites for the effective fluoride adsorption to occur. Present study suggests that these fumarate-based MOFs have great potentially useful for the fluoride adsorption from brick tea leaves. The structure characterization of the samples were collected by the powder X-ray diffraction (PXRD) patterns with Cu target from 5 to 50°. Scanning electron microscope (SEM) and transmission electron microscope (TEM) images were performed by a Hitachi S-4800 and JEOL JEM 2100 at 200 kV, respectively. X-ray photoelectron spectroscopy (XPS) were performed on the Catalysis and Surface Science Endstation of National Synchrotron Radiation Laboratory (NSRL). Nitrogen adsorption-desorption isotherms were obtanied on a micromeritics TriStar II 3020 adsorption analyzer at 77 K. Fourier transform infrared spectrometer (FTIR) were recoreded on a Nicolette is 50 FTIR spectrometer. The fluoride concentration was measured by a fluoride ion selective electrode (9609 BNWP). Catechins and caffeine concentrations were determined by the high performance liquid chromatography (HPLC, Waters 2695) with a 2489 ultraviolet (UV)-visible detector. ## Synthesis of the MOF-801. The MOF-801 was prepared according to a recently published with some modifications 25 . 1.6 g of ZrOCl 2 ⋅8H 2 O and 0.58 g of fumaric acid were dissolved in 27 mL DMF-formic acid (v/v = 20:7) mixed solution. After dissolved thoroughly, the clear solution was put into an autoclave for crystallization at 130 °C for 6 h. After reaction, the obtained products were harvested by centrifugation, washed with DMF and ethanol, and then dried overnight at 100 °C under vacuum. Synthesis of the CaFu. 0.6964 g of fumaric aicd and 0.9491 g of Ca(CH 3 COO) 2 were dissolved in distilled water (30 mL). After dissolved thoroughly, the clear solution was put into an autoclave for crystallization at 65 °C for 16 h. After cooling to room temperature, the products were obtained by centrifugation, washed with ethanol, and then dried overnight at 100 °C under vacuum. Fluoride adsorption kinetic screening. Firstly, the initial fluoride stock brick tea infusion was prepared by dispersing 0.5 g of brick tea into 25 mL of deionized water (1:50 g mL −1 ) with boiling for 30 min at 373 K. After that, the mixture solution was filtered, and the initial brick tea infusion was obtained. The fluoride concentration in the brick tea infusion was measured to be 8 mg L −1 by a fluoride ion selective electrode. The 10000 mg L −1 of fluoride solution was prepared by dissolving NaF in deionized water. Kinetic experiments were performed by exposing 40 mg of MOF-801 to 25 mL of brick tea infusion with the initial fluoride concentration of 8 mg L −1 in a 50 mL polypropylene centrifuge tube. The mixture solutions were placed in a vapour-bathing constant temperature oscillator at 298 K under a speed of 250 rpm. Then the solutions were filtered after a certain of adsorption time, both initial and the remaining fluoride ion, catechins and caffeine concentrations were determined with fluoride ion selective electrode and HPLC, respectively. Fluoride adsorption isotherm and thermodynamic. The maximum adsorption capacity of MOF-801 was performed by exposing 40 mg of MOF-801 to 25 mL of the brick tea infusion in a 50 mL polypropylene centrifuge tube with fluoride concentrations of 8, 16, 32, 64, 128, 256 mg L −1 . The solutions were placed in a vapour-bathing constant temperature oscillator at 298 K under a speed of 250 rpm for 60 min. Then the adsorbents were separated by filtration, and the remaining fluoride concentrations were measured with fluoride ion selective electrode. To further get thermodynamic parameters (ΔG, ΔS, ΔH) of MOF-801, the adsorption was also performed at 308 K and 318 K. Experimental procedure for reusability tests. For the reusability of the MOF-801 for the fluoride removal from brick tea infusion, easy to use tea bag containing 40 mg of MOF-801 NPs were prepared. The tea bag containing MOF-801 was dipped in brick tea infusion in a 50 mL polypropylene centrifuge tube with fluoride concentrations of 8 mg L −1 . At the end of the adsorption, the tea bag was removed from the brick tea infusion and the adsorbent was washed with NaOH solution (0.01 M, 5 mL × 3). After sonication for 30 min, the tea bag containing adsorbents was collected, washed with distilled water three times, and then re-dipped in the brick tea infusion (25 mL, 8 mg L −1 ) for the next cycle. To test the adsorption potential of the regenerated MOF-801 adsorbent, five cycles of regeneration studies were carried out. ## Effect of dose. The effects of MOFs dose were carried out by exposing 10-100 mg (0.4-4 g L −1 ) of MOFs (MOF-801 or CaFu) and 0.5 g of brick tea to 25 mL of deionized water. The mixture solutions were then boiled at 373 K for 30 min. Then the solutions were filtered, and the remaining fluoride ion, catechins and caffeine concentrations were determined with fluoride ion selective electrode and HPLC, respectively. CaFu maximum uptake per gram. The CaFu adsorption isotherm experiments were determined by exposing 30 mg of CaFu and 0.5 g of brick tea to 25 mL of deionized water in a polypropylene centrifuge tube with initial fluoride concentrations of 8-512 mg L −1 . These solutions were then boiled at 373 K for 30 min. Then the solutions were filtered, and the remaining fluoride ion concentrations were determined with fluoride ion selective electrode.

chemsum

{"title": "Fumarate-based metal-organic frameworks as a new platform for highly selective removal of fluoride from brick tea", "journal": "Scientific Reports - Nature"}

room-temperature_chemical_synthesis_of_c_2

## Abstract: Diatomic carbon (C 2 ) exists in carbon vapour, comets, the stellar atmosphere, and interstellar matter, but although it was discovered in 1857, 1 it has proved frustratingly difficult to characterize (Figure 1), since C 2 gas occurs/exists only at extremely high temperatures (above 3500°C). 2 Since 1930, several experimental methods to generate C 2 have been developed by using extremely high energy processes, such as electric carbon arc and multiple photon excitation, 3,4 and the C 2 species obtained were reported to exhibit singlet dicarbene (double bond) and/or triplet biradical (triple bond) behavior. 5-7 In contrast, recent theoretical simulations suggest that C 2 in the ground state should have a singlet biradical (quadruple bond) character. 8,9 Here, we present a straightforward room-temperature/pressure synthesis of C 2 in a flask. We show that C 2 generated under these conditions behaves exclusively as a singlet biradical, as predicted by theory. We also show that spontaneous, solvent-free reaction of in situ-generated C 2 under an argon atmosphere results in the formation of graphene, carbon nanotubes (CNTs) and fullerene (C 60 ) at room temperature. This is not only the first chemical synthesis of nanocarbons at ordinary temperature and pressure, but also provides experimental evidence that C 2 may serve as a key intermediate of various sp 2 -carbon allotropes. Diatomic carbon (C 2 ) is historically an elusive chemical species. Considerable efforts have been made to generate/capture C 2 experimentally and to measure its physicochemical properties. The first successful example of artificial generation of C 2 , which was confirmed spectroscopically, involved the use of an electric carbon arc under high vacuum conditions. 10 Subsequent chemical trapping studies pioneered by Skell indicated that C 2 behaves as a mixture of singlet dicarbene (double bond) and triplet biradical (triple bond) states in a ratio of 7:3 to 8:2 (Supplementary Fig. S1A). 3,6 Multiple photon dissociation of two-carbon small molecules (acetylene, ethylene, tetrabromoethylene, etc.) by infrared or UV irradiation in the gas phase was also developed to generate C 2 , but this photo-generated C 2 also exhibited several electronic states. 4 Recently, other approaches for the isolation of C 2 have been reported, using potent electron-donating ligands to stabilize C 2 by means of dative interactions (L:→C 2 ←:L), but such stabilized complexes no longer retain the original character of C 2 (Supplementary Fig. S1B). Instead, theoretical/computational simulation has been applied recently, and the results indicated that C 2 has a quadruple bond with a singlet biradical character in the ground state (Fig. 1). These various theoretical and experimental findings have sparked extensive debate on the molecular bond order and electronic state of C 2 in the scientific literature, probably because of the lack of a method for the synthesis of ground-state C 2 . For the present work, we focused on hypervalent iodane chemistry, aiming to utilize the phenyl-λ 3 -iodanyl moiety as a hyper-leaving group (ca. 10 6 times greater leaving ability than triflate (-OSO 2 CF 3 ), a so-called super-leaving group). 15 We designed [β-(trimethylsilyl)ethynyl](phenyl)-λ 3 -iodane 1a, 16 in the expectation that it would generate C 2 upon desilylation of 1a with fluoride ion to form anionic ethynyl-λ 3 -iodane 11, followed by facile reductive elimination of iodobenzene (Fig. 2A). Gratifyingly, exposure of 1a to 1.2 equivalents of tetra-n-butylammonium fluoride (Bu 4 NF) in dichloromethane resulted in smooth decomposition at -30 °C with the formation of acetylene and iodobenzene, indicating the generation of C 2 ! However, all attempts to capture C 2 with a range of ketones and olefins, such as acetone (3), 1,3,5,7-cyclooctatetraene (4), styrene (7), and 1,3,5-cycloheptatriene, failed, though they smoothly reacted with arc-generated C 2 on an argon matrix at -196 °C (Supplementary Fig. S1). 7,17 These findings immediately suggested that the putative C 2 synthesized here at -30 °C has a significantly different character from C 2 generated under high-energy conditions (Supplementary Fig. S2). Taking account of the fact that quantum-chemical calculations suggest a relatively stable singlet biradical C 2 with quadruple bonding in the ground state, we next examined an excellent hydrogen donor. 9,10-Dihydroanthracene (12) has very weak C-H bonds (bond dissociation energy of 12: 76.3 kcal mol -1 vs CH 2 Cl 2 : 97.3 kcal mol -1 ) 18,19 that might effectively trap the putative singlet biradical C 2 . When 12 was added to the reaction mixture, anthracene (13) was obtained accompanied with the quantitative formation of acetylene (Fig. 2A), which clearly suggests that the generation of C 2 and subsequent hydrogen abstraction from 12 gave acetylene. The formation of acetylene was confirmed by Raman spectroscopy after AgNO 3 trapping, and the amount of acetylene was estimated by the quantitative analysis of Ag 2 C 2 thus generated. These results strongly support the relatively stable (singlet) biradical nature of our C 2 , in accordance with the theoretical calculations. Thus, we turned our attention to the galvinoxyl free (stable) radical 14 in order to trap C 2 directly. To our delight, O-ethynyl ether 15 was obtained in 14% yield, accompanied with the formation of acetylene (84%) (Fig. 2B). The structure of 15 was fully characterized by 1 H/ 13 C NMR spectra: an upfield-shifted acetylenic proton was seen at 1.78 ppm in the 1 H NMR, as well as considerably separated 13 C NMR chemical shifts of two acetylenic carbons (C α : 90.4 ppm, C β : 30.0 ppm), clearly indicating the presence of an ethynyl ether unit. 20 In solution, di-galvinoxyl alkyne 16 was undetectable or barely detectable even when excess amounts of 14 were used, though 15 was obtained as almost the sole product in all cases. On the other hand, when we performed the trapping reaction in the presence of 2 equivalents of 14 under solvent-free conditions, 16 was clearly observed by atmospheric pressure chemical ionization (APCI) mass (MS) spectrometry, although in very small quantity (Supplementary Fig. S3). 21 These findings are consistent with the valence bond model of a singlet biradical species, according to which the energy barrier of the second hydrogen abstraction is lower by approximately 10 kcal/mol compared with the first hydrogen abstraction, which has to overcome the bonding energy of the singlet biradical. 22 It should be noted that the O-phenylated product was not formed at all, excluding alternative single electron transfer (SET) pathways, such as those via ethynyl(phenyl)-λ 2 -iodanyl radical (Supplementary Fig. S4). 23 In order to obtain more direct information about the generation of C 2 "gas", we designed a connected-flask, solvent-free experiment (Fig. 2C): a solvent-free chemical synthesis of C 2 using 1a with 3 equivalents of CsF was carried out in one of a pair of connected flasks (Flask A), and 3 equivalents of 14 was placed in the other flask (Flask B). The reaction mixture in Flask A was vigorously stirred at room temperature for 72 hours under argon. As the reaction proceeds in Flask A, generated C 2 gas should pass from Flask A to Flask B. Indeed, the color of 14 in Flask B gradually changed from deep purple to deep brown as the reaction progressed. After 72 hours, the formation of 15 and 16 was confirmed by APCI-MS analysis of the residue in Flask B. We then performed a 13 C-labeling experiment using 1b- 13 C β , which was synthesized from H 3 13 C-I in 8 steps. 24 Treatment of 1b- 13 C β (99% 13 C) with Bu 4 NF in the presence of 14 in CH 2 Cl 2 gave a mixture of 15-13 C α and 15-13 C β , suggesting that C 2 is generated before the O-ethynyl bond-forming reaction with 14 (Fig. 2D). The observed O-13 C/ 12 C selectivity (71:29) may be related to very fast radical pairing between C 2 and 14 prior to ejection of iodobenzene from the solvent cage. 25 We also carried out 13 C-labeling experiments using 1b- 13 C β in solvents of different viscosities. The observed O-13 C/ 12 C selectivity decreased as the viscosity decreased, and the regioselectivity was almost lost (52:48) under solvent-free conditions. Similarly, the O-13 C/ 12 C selectivity was 51:49 in the connected-flask experiment. All these findings rule out stepwise addition/elimination mechanisms (Supplementary Fig. S5). Given that C 2 generated at room temperature or below behaves exclusively as a A Reaction of 1a with Bu 4 NF in the presence of 9,10-dihydroanthracene (12) B Reaction of 1a with Bu 4 NF in the presence of galvinoxyl free radical ( 14) D 13 C-Labeling experiment using 1b- 13 singlet biradical, as theoretically predicted for the ground state, we examined whether this ground-state C 2 would serve as a molecular element for the formation of various carbon allotropes. Today, sp 2 -carbon allotropes such as graphene, carbon nanotubes (CNT) and fullerenes, in which sp 2 -carbon takes the form of a planar sheet, tube, ellipsoid, or hollow sphere, are at the heart of nanotechnology. 26 But, in contrast with the rapid growth of their practical applications, the mechanisms of their formation remain unclear. Various models and theories for the growth of sp 2 -carbon allotropes have been proposed, most of which include the addition/insertion of C 2 into a growing carbon cluster as a key step. 27,28 However, this idea lacks experimental verification. To investigate this issue, we examined the solvent-free reaction of the present singlet biradical C 2 in order to avoid hydrogen quenching. Notably, simple grinding of CsF and 1.5 equivalents of 1a in a mortar & pestle at ambient temperature for 10 min under an argon atmosphere resulted in the formation of a dark-brown solid containing various sp 2 -carbon allotropes, as determined by resonance Raman spectroscopy (Supplementary Fig. S6), matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry (MS) (Fig. 3A) and electrospray ionization (ESI) MS (Supplementary Fig. S7). Careful examination of the Raman spectra and high-resolution transmission electron micrograph (HRTEM) images indicated that high-quality graphite with few defects and an interlayer distance of 0.33 nm (Fig. 4A-C) and amorphous carbon had been mostly synthesized (Supplementary Fig S6A), together with very small amounts (<0.0001%) of C 60 and CNTs (ca. 0.5-7 nm in diameter, Fig. 4D and Supplementary Fig. S6B, S7, S8, and S9A), and we did not detect larger fullerenes, such as C 70 , C 76 , C 78 , and C 84 . This specificity may reflect the ambient temperature/pressure condition, as the electric carbon arc method generally affords a fearsome mixture of sp 2 -carbon allotropes. By using 1b- 13 C β , we further confirmed that C 60 is synthesized from C 2 . Grinding of 1b- 13 C β with CsF under the same reaction conditions as above afforded C 60 - 13 C 30 , which was detected by means of MALDI-TOF and ESI MS, while non-labeled C 60 was not detected at all (Fig. 3B, Fig. S9B). The formation of this unique fullerene is solid evidence for the role of C 2 , as its occurrence probability in nature is extremely small (0.01) 30 . 1 Swan W. On the prismatic spectra of the flames of compounds of carbon and

chemsum

{"title": "Room-temperature chemical synthesis of C 2", "journal": "ChemRxiv"}

actinic_wavelength_action_spectroscopy_of_the_io_-reaction_intermediate

## Abstract: Iodinate anions are important in the chemistry of the atmosphere where they are implicated in ozone depletion and particle formation. The atmospheric chemistry of iodine is a complex overlay of neutral-neutral, ion-neutral and photochemical processes, where many of the reactions and intermediates remain poorly characterised. This study targets the visible spectroscopy and photostability of the gas-phase hypoiodite anion (IO − ), the initial product of the I − + O3 reaction, by mass spectrometry equipped with resonance-enhanced photodissociation and total ion-loss action spectroscopies. It is shown that IO − undergoes photodissociation to I − + O ( 3 P) over 637 -459 nm (15700 -21800 cm −1 ) due to excitation to the bound first singlet excited state. Electron photodetachment competes with photodissociation above the electron detachment threshold of IO − at 521 nm (19200 cm −1 ) with peaks corresponding to resonant autodetachment involving the singlet excited state and the ground state of neutral IO possibly mediated by a dipole-bound state. ## Introduction In the atmosphere, iodide (I -) and its oxides, hypoiodite (IO -) and iodite (IO2 -) are relevant due to their reactivity towards ozone while iodate (IO3 -) anions, alongside its conjugate acid (HNO3) , have been implicated in particle formation . Both Iand IOx -(x =1, 2, 3) are detected in ground-based atmospheric gas measurements where IO3concentrations exhibit a diurnal response, with lowest values detected during daytime, suggesting a link to photochemistry and night-time halogen chemistry 5 . In the lower stratosphere, both Iand IO3ions are present in detectable quantities and their fates in the presence of persistent ozone concentrations are under investigation 6 . The relevance of iodine oxoacids (both neutral and anions) as aerosol nucleation agents is apparent 4 but details on their photostability under actinic radiation is lacking. In the gas phase, Ireacts with O3 in a stepwise process first producing IO -, with subsequent reaction steps with O3 to form IO2 -+ O2 and ultimately IO3 -+ O2 . The overall rate limiting step in the formation of IO3is the first step, I -+ O3 , while the subsequent steps involving IOand IO2occur with high efficiency (91% and 84% of the collision rate, respectively). Interestingly, the slow reversible reactions of IOand IO2 -(but not IO3 -) with dioxygen have also been studied with the latter case regenerating IO -10 . Although the reaction kinetics and nucleation processes of these anion intermediates are becoming better characterised, there remains a lack of information on the photochemistry of these atmospheric anions, including the simplest iodine oxide anion, IO -. The electron binding energies of the iodine oxides have been measured and suggest that IO -(eBE = 2.3805 eV, 520 nm) , IO2 -(eBE = 2.575 eV, 481.5 nm) 12,14 but not IO3 -(eBE = 4.70 eV, 263.8 nm) 14 may undergo photodetachment within the actinic window presenting a pathway to the corresponding neutral radicals. Recently, we demonstrated that photodissociation of the iodo-oxides anions, IO1-2occurs within the visible wavelength range (at 500 nm) and thus that these pathways are competitive with photodetachment at wavelengths relevant to the Earth's atmosphere. To explore this assertion, we present the first comprehensive investigation of the simplest iodine oxide, IO -, using a combination of resonance enhanced photodissociation and ion-loss spectroscopy and demonstrate that IOundergoes both photodissociation to I -+ O( 3 P) together with electron photodetachment to the neutral IO • radical under visible wavelength irradiation. ## Experimental Results Visible light absorption by IOwas investigated using tunable-laser irradiation of m/z selected IOions confined within a room temperature linear quadrupole ion trap 15 . Action spectra were measured by monitoring both the generation of Iphotoproduct ions and the decrease in total ion count resulting from electron detachment. The thermodynamic limits for these pathways, and other plausible ones, are shown in Figure 1 with the four lowestenergy destruction pathways for IOranked by their relative energy (as determined from NIST data) . Assuming single-photon absorption, over the visible range, only I -( 1 S) + O ( 3 P) photodissociation with an onset at 13729 cm -1 (728 nm) and IO + ephotodetachment with an onset at 19200 cm -1 (521 nm) are accessible. Notably the I -( 1 S) + O ( 3 P) product pathway, the only photodissociation channel accessible in the visible wavelength range, generates O ( 3 P), which can react with molecular oxygen to form ozone as per the Chapman ozone-oxygen cycle 20 . Figure 1: Schematic of the thermodynamic limits for the three lowest energy dissociation pathways of IO − (from NIST data) . The electron photodetachment energy is also included 13 . All energies are relative to the IOelectronic ground state energy. A representative single wavelength photodissociation mass spectrum of IO -(m/z 143) obtained at 19231 cm -1 (520 nm) is shown in Figure S1. The sole product ion observed is m/z 127, assigned as Iand assumed to form with O ( 3 P). Photofragment Oions (m/z 16) are not expected given insufficient photon energy (cf. Figure 1) but such ions lie below the low m/z cut-off of the instrument and, consequently, we cannot rule out their formation from multiphoton processes. A plot of Iphotoproduct yield as a function of laser power at 520 nm (Figure S1 inset) displays a linear trend, consistent with a single photon photodissociation process. The photodepletion spectrum of the IOsignal (Figure 2A, obtained by plotting the difference in the total IOion count with the laser on and the laser off) exhibits a structured band system spanning the 15700 -22000 cm -1 range (636.9 -454.5 nm). The resonance-enhanced photodissociation (REPD) spectrum (Figure 2B) appears similar to the IOphotodepletion spectrum but with better signal to noise and exhibits a broad structured band centred at 19000 cm -1 with an onset at 15700 cm -1 , which tails off to baseline at 21800 cm -1 . A series of peaks spaced by ~230 cm −1 is present within this broad band which, as will be shown, arise from a vibrational progression in an electronically excited state of IO -. Peak locations are listed in Table S1. The presence of this extended vibrational progression implies that the anion excited-state is bound (at odds with some previous studies 21 but in accord with more recent reports 13 ) and therefore suggests that the photodissociation of IOis mediated by electronic state curve crossing. It is notable that peaks expected to appear at 19410 and 20060 cm −1 appear to be supressed relative to neighbouring peaks and this will be discussed later. The total ion-loss spectrum (Figure 2C), which presumably results from electron photodetachment from IO -, is obtained by plotting the difference between the photodepletion and the photodissociation spectra, and is measured with an ion-trap storage time of 500 ms (allowing four laser pulses to irradiate the ion-packet per MS cycle). Since the Iphotoproduct is transparent in the visible region, the multi-shot experiment does not suffer additional complications arising from electron photodetachment from product ions. Over the spectrum, the total ion-loss signal rises sharply at photon energies greater than the EA of IO -(2.3805 eV, 19198 cm -1 ) 13 . The discernible peaks in the ion-loss spectrum are attributed to resonant autodetachment and two of these peaks align with the supressed peaks in the REPD spectrum. At the higher energy end of the broad band, the signal does not return to zero (unlike the REPD spectrum) presumably due to electron photodetachment into the continuum. The weak signal between 18100-19200 cm -1 below the detachment threshold is likely due to absorption by hot ions. Overall, the S/N of Figure 2B is superior to Figures 2A and 2C as the detection of photoproduct ions is essentially a background free measurement. The normalised ratio of photodissociation signal to the total ion-loss signal can be represented as a bias spectrum, which is shown in Figure 2D. At the photon energy corresponding to the EA, the two processes have equal yields, with photodissociation dominating below the EA. On the high energy side of the EA, electron loss dominates although the bias returns to zero in a few instances between the resonance peaks of the photodetachment. At photon energies greater than 20 000 cm -1 the photodepletion is dominated by electron loss. The same analysis was performed with shorter (Figure S3) and longer (Figure S4) ion trap storage times, allowing for either 1 or 9 laser pulses, respectively, to irradiate the ions before they are scanned out of the ion trap. For the longer (9 pulse) case, the S/N of the total ion-loss spectrum is significantly improved with at least five peaks clearly apparent (labelled I-V). Under these conditions however, these spectra may be affected by saturation, this is particularly obvious for the REPD spectrum. The sharp features labelled I-V in Figure 2 generally align with peaks in the total ion-loss spectrum, and this will be discussed below. (Spectra corresponding to storage times of 220 ms (1 laser pulse) and 1000 ms (9 laser pulses) are included in the supporting information: Figures S3 and S4, respectively). ## Computational Results The calculated potential energy curves for the first seventeen electronic states of IOassociated with the three lowest energy dissociation limits are plotted in Figure S5. Of these molecular states, five are within the energy range of the experiment (<22000 cm -1 ) and were subjected to higher level (icMRCI+Q/aug-cc-pwCV5Z-PP) treatment. These results are presented in Figure 3 with the calculated neutral ( 2 P3/2) potential energy curve also shown. The two spin-orbit states for the ground state of neutral IO are split by 2093(5) cm -1 such that the higher spin-orbit state ( 2 P1/2) resides at 21293 cm -1 relative to the IOground state (based on combining the experimental EA with computed spin-orbit values) 13 , and thus this state is at the higher energy edge of the experimental range and is probably not relevant to the experiments discussed here. The ground state of the anion (X < l a t e x i t s h a 1 _ b a s e 6 4 = " q 5 2 X a G 4 L O e r B f r 3 f q Y t i 5 Y x c w e + A P r 8 w f h K J c q < / l a t e x i t > N m 6 t x 6 t F + t 1 X p q z F j 3 7 4 A e s t 0 / r / p q S < / l a t e x i t > Bond length () o H e w G y p Q P + E P 9 v h F T T + u B 5 5 h J j 0 J P / / a G 4 l 9 e I w J 3 t x U L P 4 y A + + z z I T e S G A I 8 ## Photodepletion and photodissociation We now return to discuss the photodepletion and REPD spectra (Figure 2A and B) considering now the calculated potential energy curves and molecular parameters. Calculated molecular constants for the 1 1 P excited state of IO − state are Te = 15040 cm −1 with an equilibrium bond length re = 2.34 (see SI for more information). Using spline fits to the icMRCI single-point energies, the Duo program 23 provided vibrational constants: we = 289.4 cm -1 and wece = 1.72 cm -1 . The experimentally derived ground electronic state parameters for IO − , reported by Gilles et al., are re = 1.929 , we = 581 cm -1 and wece = 4.37 cm − 1 11,12 . The FC factors based on these parameters were calculated using the PGOPHER program 24 for the 1 1 P -X 1 S + transition and the results are plotted in Figure S6A. To better match the simulated intensity profile with the experimental photodepletion spectrum the Te value was raised from 15040 cm −1 to 16440 cm −1 (Figure S6B) and in this case the most intense transition corresponds to v' = 10. Alternatively, extending the excited state bond-length to 2.45 similarly shifts the FC simulation (Figure S6C). In this case, the most intense transition corresponds to v' = 16. In both cases the origin transition is predicted to be very weak, making it difficult to observe and assign. Ultimately, the correct vibrational numbering for the vibronic transitions requires an accurate value for Te and because this calculation has an uncertainly on the order of ±1500 cm −1 , confident vibronic assignments are not possible. The simulation should be ultimately assessed against a direct absorption spectrum. An alternate strategy could be to probe these vibronic transitions with raregas tagging pre-dissociation spectroscopy. These spectra are also likely to be perturbed by mixing of states, as will be discuss in the next section, further complicating the analysis. Nevertheless, the similarities between the measured and simulated spectra suggest the calculations provide useful insights into the excited state probed in the experiments. ## Total ion-loss spectrum Contained within total ion-loss spectrum in Figure 2(C) are five peaks labelled I -V. The two strongest, I and III, are located at 19410 cm -1 and 20020 cm -1 and spaced by 610 cm -1 . The next peak, labelled V, is centred at 20710 cm -1 and spaced 690 cm -1 from III. The positions I, III, and V align with supressed peaks in the REPD spectrum (Figure 2B), as indicated with the vertical dotted lines. The two smaller peaks labelled II (19810 cm -1 ) and IV (20450 cm -1 ) are spaced by 640 cm -1 . These two peaks also align with peaks in the REPD spectrum but the corresponding REPD peaks look comparatively sharper as if the higher energy side of the peak is missing. The spacing of ca. 650 cm -1 in both sets is close to the known vibrational frequency for the ground state of the IO neutral: 682 cm -113 . Figure 4 shows an expanded section of the key FC region from Figure 3 along with the REPD and total ion-loss spectra (from Figures S3 and S4) showing the position of peaks I-V and their relation to the known vibrational levels of the neutral IO electronic ground state. The neutral IO vibrational levels are located using known experimental values 13 . Peaks I-V do not align exactly with the vibrational energy levels of the X of the DBS and the anion 1 1 P valence state. Higher DBSs associated with IO X 2 P3/2 excited vibrational states can decay through a vibrational autodetachment process. Within the diabatic formulism, the DBSs and valence 1 1 P IOstates interact, leading to mixed vibronic states with both DBS and VS character. The situation is illustrated in Figure S7. In this region of the spectrum, the vibrational energy level spacing for the 1 1 P state of IOis around a third that of neutral IO X 2 P3/2 state. From Figure 4 it appears that vibrational levels associated with the 1 1 P state IOlie ≈100 cm −1 below and above the three lowest X 2 P3/2 neutral vibrational levels. Neighbouring zero-order states associated with the 1 1 P state IOvibrational levels (shown in blue Figure S7) and the DBSs (shown in red Figure S7) presumably interact leading to mixed states having valence and DBS character. Relative rates for dissociation and electron detachment will depend on the respective rates for autodetachment and dissociation from the zeroorder DBS and valence state and the mixing coefficients of each state. It is likely that transitions to the zero-order valence states are significantly stronger than those to the zero-order DBSs such that the intensity of the transition will depend on the coefficient for the valence state in the mixed state. The fact that strong transitions give rise to detachment (peaks I and III) suggests that, when energetically allowed, the rate of electron detachment from the zero-order DBSs is more rapid than dissociation from the zero-order valence states. It is possible that the lower rotational levels of IO can support dipole bound states whereas the higher rotational levels cannot. This means that lower rotational states of some vibrational levels may detach whereas higher levels dissociate. There are hints that this happens in some of the asymmetric photodissociation peak shapes (e.g., I and II). A recent photoelectron study of IOusing high-resolution velocity map imaging by Wang et al. 13 reported electron photodetachment (ePD) peaks at 19432 cm -1 , 20104 cm -1 , and 20721 cm −1 . These peaks align with peaks I, III, and V measured here. Peak I was attributed to the origin transition to the electronic excited 1 1 P state (230 cm −1 above the EA). Our experimental and computational results show that peak I is not the origin transition and that vibrational levels associated with the 1 1 P state reside below the EA. Also, Wang et al. located an ePD peak at 19829 cm -1 and surmised that it may be associated with an excited triplet anion state. This peak aligns with II and we conclude that it is a transition to a vibrational level associated with the 1 with peaks in the total ion-loss spectrum attributable to electron loss via autodetachment facilitated by mixing of first singlet excited state (1 1 P) of IOand DBSs associated with vibrational levels of neutral IO in its ground electronic state (X 2 P3/2). These results provide further considerations for the formation and fate of IOin the atmosphere. That is, while electron photodetachment from IO -(above the EA) feeds into accepted (IO • ) radical-driven ozone depletion pathways 30 , the observation of IOphotofragmentation yielding I -+ O ( 3 P) represents a possible pathway for ozone regeneration via the Chapman cycle 20 . The wavelength dependence of these competing photochemistries may play a role in the reported diurnal behaviour of IOin the atmosphere 5 . Future investigation of the relative branching fraction between photo-detachment and -dissociation will be central to atmospheric models of iodooxide anions and flow-on implications for ozone concentrations and particle formation. ## Experimental Experiments were performed using a modified linear quadrupole ion trap mass spectrometer (Thermo Fisher Scientific LTQ XL) coupled with a tuneable, nanosecond pulsed, OPO laser (GWU-Lasertechnik flexiScan) pumped by a Spectra-Physics QuantaRay INDI, which is explained in detail elsewhere . IOwas created by first dissolving potassium iodate (KIO3) in HPLC grade methanol (>99.7%) and subjected to electrospray ionisation to generate IO2via source fragmentation. IO2was then isolated in the ion-trap and dissociated via collision induced dissociation (CID) into IOthat was isolated and interrogated via the laser in an MS 3 experiment. Photodissociation experiments involved isolating and storing IOallowing irradiation with laser pulses, which was timed with a mechanical shutter synchronised with each individual isolation cycle. Photodissociation action spectra were recorded by a laser ON/OFF acquisition procedure. The ion signal with the laser off was subtracted from the ion signal with the laser on, to give the photoinduced signal, which was normalised by the total ion count when the laser is off. The accumulated raw data was processed via an in-house python script. Photodissociation action spectra were constructed by plotting the peak area of the only detected photoproduct, I -(127 m/z), and normalised to the total ion count, against the photon energy. These experiments were performed using three different ion-trap isolation cycles: 220 ms, 500 ms, and 1000 ms which correspond to one, four, and nine laser shots, respectively. Different isolation times were used to increase the signal-to-noise ratios of the photodepletion spectra. Spectra with isolation periods of 500 ms are reported in the main text while spectra with isolation periods of 220 and 1000 ms are included in the supporting information. ## Theoretical Using the MOLPRO 2019.2 program 34 potential energy curves (PEC) were constructed by scanning the IObondlength with the internally contracted multireference configuration interaction method with Davidson correction (icMRCI+Q) . The state-averaged complete active space self-consistent field (CASSCF) method was deployed to generate the wavefunction used in the icMRCI+Q calculation. CASSCF calculations were carried out for the seventeen states correlating with the three lowest energy dissociation limits 22 , which are I - . With this the seventeen states are labelled as follows: 1 3 Sand 1 3 P for the I -( 1 Sg) + O( 3 Pg) channel, X 1 S + , 1 1 S + , 1 1 S -, 1 1 P, 2 1 P, 1 1 Δ, 1 3 S + , 2 3 S + , 2 3 S -, 2 3 P, 3 3 P, 1 3 Δ for the I( 2 Pu) + O -( 2 Pu) limit, and 2 1 S + , 3 1 P, 2 1 Δ for the I -( 1 Sg) + O( 1 Dg) limit 22 . The X 1 S + , 1 1 P, 1 3 S -, and 1 3 P states are necessary to understand our results and thus for these four states the icMRCI+Q method was deployed to yield accurate energies. The X 2 P3/2 ground state of IO was also treated similarly. For both CASSCF and icMRCI+Q methods the oxygen atom was treated with the aug-cc-pwCV5Z basis set and for the iodine atom the pseudopotential approximation (-PP) extension was used . The C2v symmetry point group was used with the active space consisting of all 14 valence electrons distributed among the 8 valence molecular orbitals (4a1, 2b1, 2b2). Spectroscopic constants of the 1 1 P state of IOwere calculated using the Duo program 23 by fitting a spline to the icMRCI+Q PEC points with vibrational energies of the 1 1 P excited state determined using an RKR procedure. Experimental values from Gilles et al. were used for the X 1 S + state 11,12 .

chemsum

{"title": "Actinic Wavelength Action Spectroscopy of the IO -Reaction Intermediate", "journal": "ChemRxiv"}

porous_organic_polymers_as_fire‐resistant_additives_and_precursors_for_hyperporous_carbon_towards_ox

## Abstract: Cyclotriphosphazene (CP) based porous organic polymers (POPs) have been designed and prepared. The introduction of CP into the porous skeleton endowed special thermal stability and outstanding flame retardancy to prepared polymers. The nonflammable level of PNK-CMP fabricated via the condensation of 2,2'-(1,4-phenylene)diacetonitrile (DAN) and hexakis(4acetylphenoxy)cyclotriphosphazene (HACTP) through Knoevenagel reaction, in vertical burning tests reached V-2 class (UL-94) and the limiting oxygen index (LOI) reached 20.8 %. When used as additive, PNK-CMP could suppress the dissolving out of PEPA effectively, reducing environment pollution and improv-ing the flame retardant efficiency. The POP and PEPA co-added PU (m POP %: m PEPA % = 5.0 %: 5.0 %) could not be ignited under simulated real-scale fire conditions. The nonflammable level of POP/PEPA/PU in vertical burning tests (UL-94) reached V-0 class with a LOI as high as 23.2 %. The smoke emission could also be suppressed, thus reducing the potential for flame spread and fire hazards. Furthermore, carbonization of PNK-CMP under the activation of KOH yield a hyperporous carbon (PNKA-800) with ultrahigh BET surface area (3001 m 2 g À 1 ) and ultramicropore size showing excellent ORR activity in alkaline conditions. ## Introduction Porous organic polymers are a burgeoning family of sustainable materials utilizing natural, abundant and renewable precursors. These materials have gained increased attention in recent years. Different to other porous materials, POPs are constructed by pure organic units via covalent bond through various synthetic methods and reactions. These inherent features, such as simple synthetic routes, wellcontrolled porosity, pre-designable structure and functionality make POPs applicable to various fields, including gas uptake and separation, energy and environment, organic photovoltaic, catalytic and other important areas. In many cases, the combination of functional monomers and porous properties endowed outstanding performance to the targeted POPs excelling the pure monomers and overcoming the drawbacks existed in the monomers. Inspired by these results, we anticipated to achieve a series of fireresistant POPs via the introduction of a flame retardant monomer (cyclotriphosphazene). By changing the starting composition of building units and reaction types, two novel CP-based POPs (conjugated PNK-CMP and PNS-CMP) were prepared. Compared with PNS-CMP, PNK-CMP exhibits higher thermal stability and used as the additive or co-additive to the commercial materials studying the flame retardancy performance. Furthermore, the multiple heteroatoms doping structure make these materials ideal precursors for the preparation of porous carbons with controllable element composition which are the most popular materials applied in the renewable energy and environmental related fields, e. g. rechargeable batteries, metal-air batteries, supercapacitors (SCs) and many other new technologies. ## Results and Discussion As depicted in Figure 1, PNS-CMP and PNK-CMP were prepared according to the previous reported protocols. Briefly, PNS-CMP was fabricated via the self-polymerization of hexaphenoxycyclotriphosphazene (HPCTP) according to Scholl reaction under the catalytic of AlCl 3 in chloroform. PNK-CMP is synthesized via the copolymerization of 2,2'-(1,4phenylene)diacetonitrile (DAN) and HACTP referred to Knoevenagel reaction under the catalytic of sodium methylate in methyl alcohol/THF mixtures. The detailed procedures are given in the electronic supporting information (ESI). The targeted porous carbons denoted as PNSA-800 and PNKA-800 were fabricated via simple carbonization of prepared CMPs at 800 °C under the activation of KOH in a mass ratio of M polymer : M KOH = 1 : 2. The control samples, i. e., PNS-800 and PNK-800 were prepared under the identical conditions but without the activation of KOH. And the detail was given the electronic supporting information. Figure S1 exhibits the Fourier transform infrared (FT-IR) spectroscopy of prepared samples. Characteristic vibrations bands located at 1220 and 1420 cm 1 belonging to the CP ring could be clearly observed for both polymers. Furthermore, feature peaks ascribed to the stretching vibration of newly formed C=C bonds (1596 cm 1 ) and the inherent C � N bonds (2218 cm 1 ) could also be detected from the FTIR of the PNK-CMP. And most important of all, the stretch vibration bands attributed to the carbonyl group around 1690 cm 1 is almost disappeared, further validating the successful built-up of the porous skeletons. Figure S2a presents the solid state 13 C NMR of as-synthesized polymers, from which strong carbon signals attributed to the phenyl units distributed from 70 to 145 ppm could be found for both polymers. Meanwhile, characteristic peaks of cyano group appeared at 117.3 ppm could be observed from the 13 C NMR of CPK-CMP. The solid-state 31 P NMR spectroscopy of PNK-CMP were shown in the Figure S2b, typical signals ranged from 0 to 25 ppm assigned to the CP ring could be detected. The corresponding elemental analysis evidenced the coexistence of N, O, C and H for both prepared polymers. Low temperature N 2 uptake measurements were performed to evaluate the porous properties of prepared materials. As shown in Figure 2a, almost vertical uptake could be observed in the low-pressure region (P/P 0 < 0.01), indicating the existence of micropore for all these prepared materials. With the increasing of pressure, continue increase could be found for all these materials again. However, except the activated samples, obvious hysteresis loop in the medium pressure range and fast-adsorption in the pressure beyond 0.95 could be clearly detected, suggesting significant mesoporosity and the widely existence of macropore for other materials. Meanwhile, the calculated BET surface area for PNK-CMP was only 138 m 2 g 1 , but it increased to 513 and 3001 m 2 g 1 for PNK-800 and PNKA-800, respectively. Similar to PNK-CMP, the BET surface areas of PNS-CMP were changed from 173 m 2 g 1 to 479 (PNS-800) and 1812 m 2 g 1 (PNSA-800). Besides, both the micro and meso pore volume was also increased, demonstrating large amount of smaller pores were generated during carbonization. All these demonstrate the activation of KOH could further improve the pore properties of POPs, significantly enhancing the surface areas. And that strategy is transferable across other POPs, especially the cyano containing materials to minimize the pronounced swelling effect in the porous adsorbed species. Figure 2b presents the pore size distributions curves of prepared samples. PNK-CMP shows a main peak at 2.18 nm with secondary peaks at 1.60, 2.74 and 3.79 nm, respectively. PNS-CMP exhibits a main peak at 2.10 nm with small peaks at 1.34, 2.48 and 3.89 nm, respectively, indicating hierarchical pore structure for prepared polymers. However, the pore size is centered at micropore ranges for the carbonized samples, e. g., PNKA-800 are located at 0.57 nm and PNSA-800 are situated at 1.43 nm, implying the substantially increased micropore. And the detail about the porosity was listed in Table S1. Field emission scanning electron microscope (FE-SEM) and TEM were performed to investigate the microstructure of as-synthesized samples. Figure 3a and Figure 3d showed the SEM of prepared polymers, from which, bulk stacked by aggregated spherical nanoparticles could be observed for both polymers. As presented in Figure 3g, 3j and Figure S3, the morphology could be maintained at a great extent after the activation. Widely distributed pore, ascribed to the inherent skeleton structure or the space generated by the pile up of particles, could be detected from the TEM given in Figure 3b-c, 3e-f, 3h-i, 3k-l, and Figure S3 by the light and shade contrast. Like in previous report, the powder XRD patterns validate amorphous structure of prepared polymers (Figure S4a). And still, no clear peaks could be observed from the XRD of annealing samples (Figure S4b), indicative the amorphous features of prepared porous carbon. TGA were performed to examine the thermal stability of prepared polymers. According to Figure S5, owing to the high polarity of prepared polymers, dramatically weight loss attributed to the evaporation and desorption of adsorbed water (16.3 % for PNK-CMP and 6.8 % for PNS-CMP) could be observed at the low temperature range. Even at the temperature of 800 °C, the remaining carbon is beyond 55 wt % for PNK-CMP (vs 32 % of PNS-CMP), suggesting the super thermal stability of CP-based CMPs. Raman spectra of pyrolysis samples all displayed two intensive peaks around 1347 and 1586 cm 1 , assigned to the D band and G band, respectively (Figure S6). All these samples present high graphitic degree with a high intensity ratio of G band to D band (I D /I G ). For example, the intensity ratio of PNKA-800 is 0.98, and it reaches 0.95 for PNSA-800. Figure 4 and Figure S7 displayed the X-ray photoelectron spectroscopy (XPS) of PNKA-800 and PNSA-800, respectively. The XPS survey spectrum of PNKA-800 (Figure 4a), shows obvious signals belonging to the C, N, O, and P, validating the existence of afore mentioned element in prepared carbon samples. Figure 4b shows the high-resolution C1s spectra, a dominant peak at 284.8 eV, combined with two small peaks distributed at 285.9 eV and 287.6 eV could be observed. As presented in the Figure 4c, the N 1s spectra could be deconvoluted into three peaks located at 398.8, 399.8, and 401.1 eV, corresponding to the pyridine, pyrrole, and graphitic N species, respectively. High resolution P 2p peaks could be divided into two peaks, situated at 133.2 and 133.9 eV (Figure 4e), attributing to P C bond and P O bond, respectively. In accordance with the previous results, two obvious peaks at 530.9 and 532.4 eV, assignable to the C O and C P bond, could be obviously observed. The special features of PNKA-800, especially the ample N, P content, stimulate us to study electrocatalytic property towards oxygen reduction in alkaline conditions (O 2 -saturated 0.1 M KOH solution). Figure 5a presents the LSV curves of PNKA-800 and Pt/C at 1600 rpm. PNKA-800 shows a onset potential (E onset ) of 0.935 V as well as a half-wave potential (E 1/ Furthermore, PNKA-800 exhibits a limited current density of 4.75 mA cm 2 at 0.2 V (vs. 5.51 mA cm 2 at 0.2 V). Figure S6 displayed the CV curves of PNKA-800. In contrast to the featureless curve in Ar-saturated solution, obvious cathodic peak could be clearly detected in the O 2 saturated solution. Figure 6b exhibited the polarization curves of PNKA-800 measured from 400 rpm to 2500 rpm, and the insert part is the corresponding Koutecky-Levich (K L) plots calculated according the Equation S1. The electron-transfer number (n) obtained from the slopes of K L plots are 3.75 at the potential of 0.2 V, suggesting a four-electron pathway for ORR. And that could also be validated by the RRDE measurements shown in in Figure 5c and 5d. The calculated electron transfer number according to the Equation S3 and S4 are ranged from 3.51 to 3.96, close to that of the commercial Pt/ C with a low H 2 O 2 yield below 17.1 % in the region from 0.2 to 0.8 V and consistent well with the result obtained from the K L plots, further determining a high selectivity towards the four-electron reduction pathway of oxygen. For the real application of as-synthesized sample, an excellent cycle stability is desired. Hence, i-t test was conducted at static potential of 0.6 V (vs. RHE). Figure 5e presented the time dependent current density curves of PNKA-800. After a continuous running of 20000 s, only a negligible loss of 3.1 % of the initial current was detected for the synthesized sample, much better than that of commercial Pt/C catalyst (37.2 %), indicative PNKA-800 possessing superior durability. Figure 5f shows the polarization curves measured before and after the methanol-crossover test, almost identical curves were found with a negative shift of 10 mV in E 1/2 (vs 37 mV for Pt/C given in Figure S8). All these results evidenced PNKA-800 can act as a remarkable oxygen reduction electrocatalyst in alkaline solution applied in the metal-air battery. For the unique structure, cyclotriphosphazene (CP) and CP-derived materials are widely investigated as the flame retardant additive, lowering smoke emission and heat release rate. Pentaerythritol octahydrogen tetraphosphate (PEPA) is a commercial flame retardant commonly used as additive in thermoplastic polyurethane elastomer (TPU), improving the flame retardant efficiency. A minimum addition of 10 % can play a better flame retardant effect. But PEPA could migrate out from the TPU which are harmful to the environment. The similarity of structure and elemental composition for the PEPA and POP, make PEPA easily get into the hole of PNK-CMP. Inspired by this, we anticipated to inhibit the releasing of PEPA from the TTU via introduction of porous CMP. It is the first time that the porous PNK-CMP was applied as additive or co-additive with PEPA to commercially available TPU improving the flame retardant efficiency. As shown in Figure 6, CMP/TPU was fabricated via simple physical processing in different mass ratios of POP to TPU. One could observe clearly that the color of composites darkens with the increasing of mass ratio. While, CMP/PEPA/ TPU composites was prepared stepwise. PEPA was initially absorbed in the porous skeleton of CMP, then the composites was mixed with the TPU (mPOP% : mPEPA% = 5.0 % : 5.0 %). To evaluate the flammability, vertical burning tests (UL-94) and limiting oxygen index (LOI) measurements were conducted. In UL-94 tests, the TPU/POP/PEPA composites possessed the highest level (V-0), similar to the TPU/PEPA, but higher than that of TPU/CMP composites (V-2). As shown in Table 1 and Video S1, S2, even directly exposed to an igniter for 10 s, the TPU/POP/PEPA sample could not be ignited. When the igniter was removed, the flame was extinguished instantly, and no obvious flame was observed on the surface of samples. After that test, compared with the TPU/POP composites, TPU/POP/PEPA maintained the original shape better, suggesting certain flame retardancy. Furthermore, as delivered in Figure 5c and 5d, large amount of nonflammable gas (N 2 ) was generated during the combustion process, released by the decomposition of composites, which are further transformed into protective carbon-layer coated on the material surface (Figure 5e). Meanwhile, the lOI value reaches 25.2 % for the TPU/POP/PEPA composites higher than the TPU/POP composites (23.0 %). The POP and PEPA coadded TPU (m POP % : m PEPA % : m TPU % = 5.0 % : 5.0 % : 90 %) could not be ignited under simulated real-scale fire conditions. And as observed form Figure 5c, 5d and 5e, a large amount of nonflammable gas was emitted from materials preventing the burning, after combustion carbon-layer was formed which to prevent further transfer heat to composites. This process conforms to the mechanism of intumescent flame retardant. Furthermore, when used as additive, PNK-CMP could suppress the dissolving out of commercial PEPA effectively, reducing environment pollution and improving the flame retardant efficiency. In contrast, TPU polymer insulation materials are easily ignited. Furthermore, the fire retardancy could also be validated by the TGA given in the Figure S9. Compared with the pure polymer, the thermal stability of composites enhanced greatly. And the ternary complex (m POP % : m PEPA % : m TPU % = 5.0 % : 5.0 % : 90 %) presents a TG similar to the binary complex (m PEPA % : m TPU % = 10 % : 90 %). ## Conclusions In summary, cyclotriphosphazene based CMPs were welldesigned and facilely prepared. The special elemental composition and special thermal stability endowed these polymers ideal candidates for multiple atoms doped porous carbon applicable for the energy and environment-related fields. Under the activation of KOH, hyperporous carbon with ultrahigh BET surface area of 3001 m 2 g 1 and ultra uniform pore size distribution were obtained which could further be used as the carbon-based catalysts for ORR with excellent performance. The introduction of CP into the porous skeletons give outstanding flame retardancy to prepared polymers. And the porous features could suppress the dissolving out of commercial PEPA effectively, reducing the pollution triggered by traditional flame retardant. For example, the nonflammable level of POP and PEPA co-added TPU (m POP % : m PEPA % = 5.0 % : 5.0 %) reached V-0 class with a LOI as high as 23.2 % that could not be ignited under simulated real-scale fire conditions.

chemsum

{"title": "Porous Organic Polymers as Fire\u2010Resistant Additives and Precursors for Hyperporous Carbon towards Oxygen Reduction Reactions", "journal": "Chemistry Open"}

exofit_trial_at_the_atacama_desert_(chile):_raman_detection_of_biomarkers_by_representative_prototyp

## Abstract: In this work, the analytical research performed by the Raman Laser Spectrometer (RLS) team during the ExoFiT trial is presented. During this test, an emulator of the Rosalind Franklin rover was remotely operated at the Atacama Desert in a Mars-like sequence of scientific operations that ended with the collection and the analysis of two drilled cores. The in-situ Raman characterization of the samples was performed through a portable technology demonstrator of RLS (RAD1 system). The results were later complemented in the laboratory using a bench top RLS operation simulator and a X-Ray diffractometer (XRD). By simulating the operational and analytical constraints of the ExoMars mission, the two RLS representative instruments effectively disclosed the mineralogical composition of the drilled cores (k-feldspar, plagioclase, quartz, muscovite and rutile as main components), reaching the detection of minor phases (e.g., additional phyllosilicate and calcite) whose concentration was below the detection limit of XRD. Furthermore, Raman systems detected many organic functional groups (-C≡N, -NH 2 and C-(NO 2 )), suggesting the presence of nitrogenfixing microorganisms in the samples. The Raman detection of organic material in the subsurface of a Martian analogue site presenting representative environmental conditions (high UV radiation, extreme aridity), supports the idea that the RLS could play a key role in the fulfilment of the ExoMars main mission objective: to search for signs of life on Mars.Led by ESA with the collaboration of Roscosmos, the ExoMars 2022 rover mission will pursue the detection of signs of present or past life on Mars 1,2 . To achieve this goal, the designed payload of the Rosalind Franklin rover will employ a set of panoramic instruments (PANCAM 3 and ISEM 4 ) to explore the surrounding environment, thus providing crucial data to be used in the navigation of the rover and in the identification of areas of high scientific interest. A ground-penetrating RADAR (WISDOM 5 ) and a passive neutron spectrometer (ADRON-RM 6 ) will investigate the subsurface, helping in the selection of potential drilling places. The ExoMars Drill Unit 7 (hosting the MA_MISS visible and near infrared spectrometer 8 ) will collect geologic samples down to a depth of 2 m, thus accessing material that have been sheltered from UV Radiation and further alteration processes. CLUPI 9 will provide textural information from the sampled materials through the collection of high-resolution images, while the sample preparation and distribution system (SPDS) will crush the materials and deliver the powders to the analytical laboratory of the rover 10 . Here, the visible/near-infrared spectrometer (MicrOmega 11 ) and the Raman Laser Spectrometer (RLS 12 ) will perform coordinated analyses 13 to identify the mineralogical composition of the samples and to reveal the potential presence of biomarkers. Spectroscopic results will be used to select the optimal scientific targets to be delivered to MOMA (Mars Organic Molecule Analyzer system), that will extract and analyse the organic molecules potentially preserved within the mineralogical matrix 14 . Apart from the technical and engineering challenges that meant the development of the mentioned instruments, the success of the mission also relies on the complex coordination work required for their remote control and synergic management Recognizing the need for training the ExoMars teams and enhancing collaboration practices between instrument working groups, ESA organized the ExoMars-like Field Testing (ExoFiT) trials 15 , the second of which was carried out at the Atacama Desert (Chile) in February 2019. In addition to presenting a Martian-like desertic landscape, the presence of extremophile microorganisms populating Atacama's subsurface made this the ideal location to test the ability of the rover's payload to detect biomarkers in this kind of environments 16 . During the trial, an emulator of the Rosalind Franklin rover (Charlie) was used to perform a complex sequence of scientific and engineering operations (from descending the landing platform to collecting drill cores) following the ExoMars Reference Surface Mission (RSM) 17 . During the mission simulation, the LCC team (Local Control Centre, located at the Atacama Desert, near the ESA Paranal Observatory) manoeuvred the rover and managed the acquisition and upload of the collected data. From 11,000 km of distance, the RCC team (Remote Control Centre, located at the European Centre for Space Applications and Telecommunications, UK) simulated the operations on Mars, planning the different activities for the next sol by only relying on the data returned by the rover 18 . As part of the LCC team, science and engineering roles were covered by personnel from the University of Valladolid (UVa) and the National Institute for Aerospace Technology (INTA), who carried out the Raman characterization of the subsoil cores drilled by the rover. The Raman characterization was done using two spectrometers. A first mineralogical evaluation of the samples was performed using the RAD1 system (RAman Demonstrator 1), which is a portable RLS technology demonstrator assembled by the RLS team to carry out in-situ analyses in terrestrial analogue sites 19 . The RAD1 spectrometer has similar range of analysis (70-4200 cm −1 ), laser wavelength (532 nm) and power output (7 mW on the sample), spot of analysis (≈ 50 µm) and spectral resolution (6-10 cm −1 ) to the RLS, providing spectra qualitatively comparable to those soon gathered on Mars. Afterwards, more detailed spectroscopic analyses were carried out in the laboratory by means of the RLS ExoMars Simulator, which characteristics have been described elsewhere 20 . As detailed in previous works, this is the optimal instrument to predict the potential scientific outcome of the RLS flying model 21,22 . Indeed, in addition to the RLS-like optical spectral characteristics (as the RAD1), the spectrometer is coupled to a replicate of the ExoMars/ SPDS, and integrates the same algorithms developed for the RLS to perform the automatic multi-point analysis of Martian samples (e.g. Signal to Noise Ratio optimization, florescence quenching and acquisition parameters selection 23 ). Raman spectra from the ExoFiT exercise were obtained under the same operational constraints of the rover, and were finally compared to XRD data, being this the reference instrument for the mineralogical study of geological samples. Recognizing the scientific and logistic value of this mission simulation, the present work aims to (1) summarize the preliminary analytical results obtained by the RLS team from the study of Atacama Desert samples, (2) evaluate advantages and disadvantages provided by the use of the RLS representative prototypes in ExoMarsrelated studies, and (3) extrapolate valuable information about the potential role the RLS could play in the fulfilment of the ExoMars mission objectives. ## Atacama desert (Chile). The Atacama Desert is a high plain covering an area of more than 100,000 km 2 between northern Chile and southern Peru. The hyper arid climate of this region, persisting unchanged for the last 10 million years, is due to the concurrence of the foehn effect (triggered by the Andean Mountains), the Humboldt current and high-pressure atmospheric conditions (caused by Pacific anticyclones) 24 . The ExoFiT trial was carried out in the region of Antofagasta, about 11 km west of the ESO Paranal observatory (altitude of 2200 m). According to previous studies, three kinds of rocks dominate the mineralogy of this area: granodiorites (white-pink colour) and andesite (dark green) are composed of plagioclase and quartz in different concentration ratio, while gabbros (dark-gray color) contain feldspar and amphibole/pyroxene minerals 25 . In addition to these primary minerals, alteration products such as phyllosilicates and oxides (e.g. hematite) can be found in the area together with evaporites (nitrates, sulphates and chlorides). Based on the data collected at the ESO Paranal observatory, this area presents high temperature oscillations (from − 8 to + 25 ºC), extremely low humidity values (5-20%) and an average annual rainfall below 10 mm 26 . In addition to the mentioned parameters, the extremely high levels of surface ultraviolet (UV) irradiance (> 1100 W/ m 2 ) 27 , make this area the perfect terrestrial analogue site to investigate the suitability of microbial life in extreme environments, similar to those that can be found on Mars and other planets 16 . Despite the harsh environmental conditions, extremophile microorganisms populate the subsurface of Atacama by relying on metabolic mechanisms that may have analogies with those that could be adopted in the shallow subsurface of Mars . In light of the forthcoming deployment of Raman spectrometers on Mars (beside the RLS, Sherloc 31 and SuperCam 32,33 instruments onboard the NASA/Mars 2020 rover also need to be mentioned), Vitek et al. published several works using Atacama rocks and soil samples to assess the capability of this technique to detect biomarkers, gathering encouraging results . ## Rover activity and samples collection. As can be seen in Fig. 1, the Martian-like landscape of the area selected by the LCC team presents a reddish desertic pavement made of gravel, boulders and interspersed sand patches. The area also features small clay deposits and salt crusts, being these units of great astrobiological interest. Besides site selection, the LCC team took care of manoeuvring the rover and operating the whole set of ExoMars instruments accordingly to the commands received from mission control. As mission coordinator, the RCC made an assessment of the landing site, planning the descent from the landing platform and the driv-ing through a safe route to reach an area of scientific interest. The subsurface stratigraphy of the selected site was then analysed by WISDOM (scan grid of approximately 5 * 5 m). Based on subsurface radar results, RCC selected the optimal drilling site. After drilling, the extracted soil core was imaged by CLUPI, separated in two samples (upper part UP and lower part LP) and sent for Raman analysis. During the ExoFiT trial, two experiment cycles were conducted, giving a total of two cores and four subsamples in total. Since the granulometry of the sample affects the quality of Raman results 37 core samples were crushed and sieved to obtain a grain size distribution resembling the one prepared by the ExoMars/SPDS. The resulting samples were then placed into a replicate of the ExoMars sample holder and analysed by Raman. Instruments. For the in-situ characterization of drill cores, Raman analyses were performed directly at the analogue site and by following the time constraint imposed by the mission simulation. To do so, the RAD1 spectrometer was used. Assembled by the ERICA group, this portable instrument is composed of a commercial excitation laser source of 532 nm, a high resolution Thermo Electrically (TE) Cooled CCD Array spectrometer (2168 × 512 pixels) and a high line density diffraction grating (1800 lines per mm). The instrument was optically harnessed by optical fibers to a microscope with a 50 × objective, reproducing the analytical footprint of RLS (50 µm). The time constrains applicable to these analyses during ExoFiT test limited the time per sample to 1-1.5 h, a timeframe shorter than the nominal ExoMars rover operations. Complementary analyses were done at the laboratory using the RLS ExoMars Simulator, a system with similar spectroscopic performance to RAD1, but incorporating the automatic operation capabilities of RLS. The instrument includes a continuous green excitation laser (532 nm), a high resolution TE Cooled CCD Array spectrometer and an optical head with a long working distance objective of 50x. The instrument is coupled to three axis micrometric positioning system with a refillable container (emulating the ExoMars sample holder) that allows the definition of analysis rasters on the sample. Software-wise, the RLS ExoMars Simulator implements the same algorithms developed for the RLS 23 , allowing the automatic analysis of the samples auto adjusting the acquisition parameters. For both spectrometers, spectra acquisition was performed through a custom developed software based on LabVIEW 2013 (National Instruments), while the IDAT/SpectPro software was used for data processing and interpretation 38 . Knowing that the quantum efficiency of CCD detectors varies with the wavelength, the intensity of all spectra was corrected by following the method presented by Sanz Arranz et al. 39 Besides Raman analyses, the mineralogical characterization of powdered materials was complemented by XRD data. For this purpose, a laboratory Discover D8 XRD (Bruker) was used. The diffractometer is composed of a Cu X-ray excitation source (wavelength 1.54 ) and a LynxEye detector. Fine-powdered rocks (granulometry ≤ 150 µm) were analysed by setting a scan range between 5 and 70° 2θ, a step increment in 2θ of 0.01 and a count time of 0.5 s per step. The collected diffractograms were interpreted using the BRUKER DIFFRAC.EVA software. ## Results RAD 1. Drilled cores were analysed in-situ by using the portable RAD1 spectrometer. For this purpose, powdered samples were placed in a replicate of the ExoMars sample holder and, after flattening, a raster of measurements was performed by moving the X positioner at regular intervals of ≈ 300 µm. For each spot of analysis, the acquisition parameters were optimized manually. It must be noted that in-situ analyses were hampered by meteorological conditions, since the strong wind blowing during the trial produced vibrations to the spectrometer, compromising the acquisition of many Raman spectra. For this reason, a very limited number of spectra per sample (between 4 and 6) could be collected within the time constraints imposed by the mission simulation. Despite the limited amount of Raman data, different mineral phases were successfully detected. Starting from ADC2 drill core, quartz (SiO 2 , 142, 204 and 464 cm −1 , Fig. 2a) was detected in both UP (upper part) and LP (lower On the other hand, Raman analysis of sample ACD2-LP displayed peaks at 276, 475 and 514 cm −1 , revealing the presence of feldspar (Fig. 2c). However, the mineral phase within this group was not clearly identified due to the low Signal to Noise Ratio (SNR) of the obtained spectra. Besides feldspar, the same spectra displayed a broad band at 3600 cm −1 (vibration of -OH group), together with an additional minor signal at 695 cm −1 . According to the work published by Wang et al., 2015 42 , these signals are characteristic of phyllosilicates within the mica subgroup (probably muscovite). In-situ Raman analysis of ADC1 core were quite inconsistent since the two subsamples were highly fluorescent, which is a side-effect from electronic excitation that increases the background signal in the spectra, masking mineral Raman bands. Although the long exposure of the spot to the excitation laser helps quenching the florescence, this operation could not be performed due to the abovementioned vibrations induced by the wind. Despite this limitation, feldspar minerals were effectively detected in both LP and UP subsamples. ## RLS ExoMars simulator. After the automatic adjustment of the spectra acquisition parameter to the constraints established for the ExoFiT trial, the number of Raman spectra automatically collected from each sample www.nature.com/scientificreports/ with this instrument varied between 9 and 12, which is below the minimum number of analysis per core that are expected to be carried out on Mars (20). This can be explained by the fact that the time dedicated to the in-situ study of drilled cores was narrowed due to logistic reasons (two RSM measurement cycles needed to be compressed within a time frame of 10 sols). However, additional spots were analysed to reach a total of 39 spectra per sample, being this the maximum number of analysis to be nominally performed on regular operations on Mars. The results described below are based on the complete set of Raman data gathered from each sample, although the summary provided in Table 1 allows one to distinguish the minerals detected within the ExoFiT-constrained time frame (black cross) from those additionally detected using nominal ExoMars mission parameters (red cross). Starting from the ADC2 core, both UP and LP subsamples showed Raman features from quartz (main peak at 464 cm −1 and secondary signals at 124, 202, 263, 354, 805 and 1159 cm −1 , Fig. 3a), anatase (TiO 2 , main peaks at 142, 394, 510 and 634 cm −1 , Fig. 3b) and feldspar. By comparing the vibrational profile of feldspar spectra, different mineral phases were identified. For example, the positions of the peaks detected in the spectrum shown in Fig. 3c (main signals at 478 and 508, together with minor peaks at 167, 286, 409, 566, 763, 809 and 1100 cm −1 ) matched perfectly with the Raman features from albite 43 , being this mineral the Na-rich end member of the plagioclase subgroup (NaAlSi 3 O 8 ). As displayed in Fig. 3d, further spectra matched with anorthite reference spectrum (peaks at 150, 276, 401, 473, 515, 756, 799 and 1120 cm −1 , confirming the additional presence of K-feldspars in both subsamples. Albite and anorthite spectra were found to be often associated with additional peaks at 264, 407, 702 and 3628 cm −1 , which are consistent with the muscovite reference spectrum (KAl 2 (Si 3 Al) O 10 (OH) 2 , Fig. 3e). In addition to the mentioned mineral phases, calcium carbonate was additionally detected (main peaks at 149, 275, 709 and 1085 cm −1 , Fig. 3f) in both subsamples. Raman results from ADC2-UP showed a higher and more complex mineralogical heterogeneity of this subsample when compared to AC2-LP. As shown in Fig. 3g, additional Raman peaks were found at 220 and 670 cm −1 , matching the characteristic signals of amphibole minerals (probably actinolite, Ca 2 (Mg,Fe) 5 Si 8 O 22 (OH) 2 ). As can be seen in Fig. 3h, the detection of clear peaks at 411, 490, 620, 1008 and 1135 cm −1 revealed the presence of gypsum (Ca 2 SO 4 ) as additional evaporitic mineral. One of the spectra gathered from the upper part of the core displayed two weak bands in the spectral range between 3600 and 3700 cm −1 . In detail, the peak at 3630 cm −1 matches the mentioned -OH vibration from muscovite, while the signal at 3695 cm −1 could be associated with additional clay minerals such as kaolinite, serpentine or chlorite 42 . In addition to those, a broad band at 3425 cm −1 was found associated with two phyllosilicate spectra. When compared to the Raman emission of organic functional groups described elsewhere 40 , the detected signal matches the characteristic position of the in-phase bending mode of aromatic amines (-NH2, Fig. 4c). Raman spectra from the laboratory analyses of both drill cores, ADC1 and ADC2, showed a similar mineral composition. Indeed, quartz, anatase, plagioclase and K-feldspar were found to be the main mineral components of both UP and LP subsamples, while calcite and actinolite were exclusively detected in the upper part of the core. Besides the detection of mineral phases, the RLS ExoMars Simulator could detect Raman features from potential biomarkers, as displayed in Fig. 4a. A doublet at 2190 and 2250 cm −1 was clearly identified in both UP and LP samples that, according to the results presented in previous works, could correspond to the vibration mode of different functional groups containing nitrogen 40 . Similarly, the peak observed around 1445 cm −1 can be either attributed to the symmetric bending of -CH2 or the asymmetric bending of -CH3. Further potential organic peaks were detected on sample UP, as shown in Fig. 4a, with features appearing at 2800 and 2850 cm −1 that fall within the C-H stretching region (2800-3100 cm −1 ) 44 . Concerning sample LP, additional vibrational features from organic functional groups are shown in Fig. 4b,d, with peaks detected in the range between 1300 and 1700 cm −1 . More specifically, the three signals at 1340, 1380 and 1530 cm −1 can be assigned to the C-(NO2) functional group (both symmetric and asymmetric stretching), while the peak at 1644 cm −1 can be related to the stretching mode of C=N. As shown in Fig. 6, diffractograms from ADC1 subsamples displayed wider and less intense peaks. Considering that XRD analyses were run under the same measurement conditions, including the amount of powdered material, it can be deduced that ADC1 subsamples could contain some secondary phase of low crystallinity. By comparing the position of the detected peaks with XRD reference pattern, both subsamples are mainly composed of quartz with minor amounts of plagioclase and muscovite. In the case of LP sample, hematite signals were detected (33.15 and 35.68 2θ), while UP displayed minor amounts of amphibole. The overall results gathered from the use of both spectroscopic and diffractometric systems are provided in Table 1. ADC1 and ADC2 drill cores are characterized by a complex mixture of organic and inorganic compounds. Summarized in Table 1, the mineralogical characterization obtained from the combined use of in-situ and laboratory Raman spectrometers is in good agreement with XRD data. Beside confirming the identification of the main mineral phases (quartz, feldspar and muscovite), the two RLS representative prototypes also detected additional minor compounds, whose concentration was often below the detection limit of XRD (anatase, calcite, amphibole and additional phyllosilicate, depending on the sample). Having in mind the forthcoming ExoMars mission, this result is extremely relevant as it demonstrates that the analytical strategy based on the multipoint Raman analysis of powdered samples could effectively help disclosing the composition of complex mineralogical mixtures. The two Raman spectrometers, operating under the same operational constraints of the RLS instrument, were able to detect phyllosilicate minerals, which are one of the main scientific targets defined for the ExoMars mission. Indeed, it is well known that phyllosilicates are capable of hosting microorganisms and accumulating biomarkers within their crystalline structure, thus potentially playing a key role in the preservation of life traces on Mars 45 . In fact, the large phyllosilicate deposits detected from orbit at Oxia Planum 46,47 were one of the main drivers in its selection as the landing site for the Rosalind Franklin rover. The results obtained from the phyllosilicate-bearing samples agree with this thesis, as Raman spectra often presented features corresponding to different organics functional groups. Even though the Raman-based detection of organics in Atacama Desert samples was already achieved in previous works 34 , the great astrobiological relevance of the present research is based on the fact that (1) spectra were collected by Raman systems engineered to mimic the quality of RLS, and (2) drilling sites were remotely selected by the RCC team, who was operating the mission simulation from 11,000 km of distance having no more inputs than the data returned from the rover. The functional groups detected by Raman (including -C≡N, -NH2 and C-(NO2)) are compatible with the presence of nitrogen-fixing microorganisms in the drilled samples. Again, this result fits with previous works presented by Maza et al. 2019, who revealed the presence of six potential nitrogen fixers in the subsurface of the Atacama Desert 48 . Furthermore, it must be noted that LP samples returned the higher number of biomarkers spectra, suggesting that the microbial activity in the subsoil (below 15-20 cm of depth) is higher than in the surface. This gradient in microbial activity may be due to the fact that more favourable conditions for life proliferation can be found at higher depths (e.g., higher water content and lower exposure to UV radiation). Knowing that the analysis of subsoil samples is the core strategy for the ExoMars mission to detect traces of life on Mars, the Raman results here described are extremely promising as they confirm its efficacy. Evaluation of RLS representative prototypes. By participating to the ExoFiT trial, the RLS team could evaluate advantages and disadvantages provided by the use of the RLS representative prototypes in ExoMarsrelated studies. Even though there are numerous studies evaluating the capabilities of the RLS ExoMars Simulator, this is the first work presenting Raman data gathered from the portable RAD1 system. For this reason, comparing the results of the two instruments could help evaluating the real scientific capabilities of the RAD1. As shown in Table 1, the main mineralogical phases were correctly identified in RAD1 datasets, which results were in perfect agreement with spectra provided by the RLS ExoMars Simulator. However, when evaluating phases in minor proportions, some additional compounds could be detected by the laboratory setup. One of the main reasons is the fluorescence background, being more intense in spectra obtained with RAD1 at the LCC than it is in the laboratory ones. In the case of ADC2 subsamples, those containing a higher concentration of low crystalline phases (according to XRD), the fluorescence background covered almost completely the Raman vibrational features in the spectra. This difference can be justified by the use of different analytical approaches. In the laboratory, spectra fluorescence was minimized by automatically performing laser-induced quenching on each spot of analysis (by using the same algorithm that implements RLS). However, this procedure was not feasible for in-situ analyses due to the mentioned time constraints and the stability problems of the spectrometer (triggered by adverse meteorological conditions). It should be also noted that, despite the additional time required by florescence quenching, the number of spectra collected by the RLS ExoMars Simulator within the time constraints of the ExoFiT Trial was higher than those gathered by the RAD1 (manually operated). This result highlights that a more efficient characterization could be achieved through analysis automation. Learning from the Atacama trial experience, the RLS team is planning to couple the RAD1 system to a portable XYZ positioner as well as to implement its software with RLS algorithms for automatic multi-point analysis of samples. These improvements will allow to optimize data collection and to ensure a better simulation of the automatic operating mode of the RLS, thus increasing the scientific relevance of in-situ Raman studies of terrestrial analogue sites. ## Considerations for the ExoMars mission. Using the data provided by the rover, the controlling team at the RCC the rover was capable of analysing the surrounding environments and identifying areas of scientific interest (PanCam and ISEM), investigating the textural features of the surface (CLUPI), determining the stratigraphy of the subsoil (WISDOM), extract drill cores (ExoMars drill emulator) and analysing their composition (RLS representative prototypes). Strictly focusing on Raman operations (the logistical and engineering challenges faced during the trial will be presented in a specific work), focusing on Raman operations (the logistical and engineering challenges faced during the trial will be presented in a specific work), the time frame dedicated to the spectroscopic analysis of drilled cores was found to be too narrow to achieve the number of spectra established for the nominal operation of RLS on Mars (between 20 and 39). As summarized in Table 1, the ability to detect minor or trace compounds of great scientific relevance (in this case of study, phyllosilicates and organics functional group) increases with the number of analysed spots per sample. In this sense, the RLS ExoMars Simulator missed the identification of the organic functional groups detected by RAD1 in sample ADC2-UP, this particular case evidences that more than 39 spectra per sample could be sometimes needed. Knowing the RLS will work in combination with MicrOmega, the additional information provided by the IR spectral images could be used to plan more targeted Raman analysis during real operations (for the ExoFiT trial the analysed spots were randomly selected), thus increasing the chances of detecting organics on Mars. However, if a scientifically interesting sample is collected from the Martian subsoil, the chances of detecting potential biomarkers could be increased by performing more than one cycle of spectroscopic analysis. Indeed, this procedure could help optimizing the use of the 32 single-use ovens equipped by MOMA to run GCMS analysis, thus enhancing the possibilities to fulfil the main objective of the mission. ## Conclusions During the Atacama ExoFiT test a complex series of operations, starting with the descent of the rover from the landing platform and ending with the extraction and analysis of drilled cores, were successfully carried out. Focusing on the analytical characterization of subsoil samples, the RLS representative prototypes demonstrated the key role that Raman spectroscopy could play in the fulfilment of the ExoMars mission objectives. By simulating the operational constraints of the RLS, the instruments used in this exercise disclosed the complex mineralogical composition of the samples, providing results qualitatively comparable to those obtained by a laboratory XRD system. In addition to the inorganic matrix, Raman spectrometers also detected several additional signals that could be assigned to biomarkers. In preparation of the upcoming ExoMars mission, this result confirms the capabilities of Raman spectroscopy, which was able to detect extremophilic microorganisms potentially colonizing the subsurface of Martian-like environments. Similar results on Mars would help in the selection of geological samples to be analysed by MOMA. In spite of the promising results, the comparison between RLS ExoMars Simulator and RAD1 data from sample ADC2-UP suggests that the nominal number of spots per sample the RLS will be nominally analyse on Mars (between 20 and 39) may not be sufficient to ensure the detection of trace compounds potentially present in the sample. This is why the ExoMars mission foresees an unprecedented cooperative approach (combined science), by which the instruments of the analytical laboratory will be able www.nature.com/scientificreports/ to analyse the same spot of the samples. More specifically, this capability will allow RLS to dedicate part of its operation to the analysis of sample spots previously identified by MicrOmega as regions of interest. Nevertheless, if during operations a sample of Martian subsoil reveals itself to be of high scientific interest, the possibility of running an additional cycle of combined MicrOmega-RLS analysis should be considered. Attending to the lessons learnt from the Atacama ExoFiT test, and recognizing the value of mission simulations in preparation for the ExoMars mission, the RLS team is planning to perform improvements (both hardware and software) of the portable RLS representative prototype, aiming to increase the scientific relevance of in-situ Raman studies of terrestrial analogue sites. Received: 3 September 2020; Accepted: 30 December 2020

chemsum

{"title": "ExoFiT trial at the Atacama Desert (Chile): Raman detection of biomarkers by representative prototypes of the ExoMars/Raman Laser Spectrometer", "journal": "Scientific Reports - Nature"}

induced_effects_of_advanced_oxidation_processes

## Abstract: Hazardous organic wastes from industrial, military, and commercial activities represent one of the greatest challenges to human beings. Advanced oxidation processes (AOPs) are alternatives to the degradation of those organic wastes. However, the knowledge about the exact mechanisms of AOPs is still incomplete. Here we report a phenomenon in the AOPs: induced effects, which is a common property of combustion reaction. Through analysis EDTA oxidation processes by Fenton and UV-Fenton system, the results indicate that, just like combustion, AOPs are typical induction reactions. One most compelling example is that pre-feeding easily oxidizable organic matter can promote the oxidation of refractory organic compound when it was treated by AOPs. Connecting AOPs to combustion, it is possible to achieve some helpful enlightenment from combustion to analyze, predict and understand AOPs. In addition, we assume that maybe other oxidation reactions also have induced effects, such as corrosion, aging and passivation. Muchmore research is necessary to reveal the possibilities of induced effects in those fields. Results and discussionFenton's reagent (mixture of hydrogen peroxide and ferrous iron) is one of the most effective AOPs, which was developed in the 1890s by Henry John Horstman Fenton as an analytical reagent 13 . Fenton's reaction is based on the catalyzed decomposition of H 2 O 2 by ferrous iron to produce reactive ?OH 14 . It has been found effective in treating various industrial wastewater components including a wide variety of landfill leachate 15 , pesticides 16 and surfactants 17 , as well as many other substances. Oxidations of ethylenediaminetetraacetic acid (EDTA) by ## H azardous organic wastes from industrial, military, and commercial activities represent one of the greatest challenges to human beings 1,2 . Advanced oxidation processes (AOPs) are alternatives to the degradation of those organic wastes . In 1987, Glaze et al. 3 defined AOPs as ''near ambient temperature and pressure water treatment processes which involve the generation of hydroxyl radicals in sufficient quantity to effect water purification''. The hydroxyl radical (?OH) is a powerful, highly reactive chemical oxidant, which reacts very quickly with organic compounds. Recently, other studies have suggested that, besides ?OH, AOPs can also generate other oxidizing species, such as sulfate radicals 6,7 . It is generally believed that, depending upon the nature of the organic species, they are oxidized by radicals mainly through hydrogen abstraction or addition itself to double bonds of the contaminant 8 . AOPs are important tools for environmental technology, so they have to be placed on more sound scientific and engineering basis. However, the knowledge about the exact mechanisms of AOPs is still incomplete 9,10 . The most difficult problem is how to chooce or design the most efficient AOPs system for the given pollutant. So, the reaction mechanisms, efficiency improvement, and their mathematical modeling will be the key subjects of the future research. Some investigators called AOPs as ''cold combustion'' 11,12 . Just because, similar with combustion, AOPs is a kind of oxidation process which could oxidize and mineralize the organic matter under mild conditions. But we tried to discover more similarities between them and achieve some enlightenment which is helpful to analyze, predict, understand and improve the oxidation efficiency of AOPs. Here we report a similar phenomenon exist in the AOPs: induced effects, which is a general property of combustion reaction. In ancient times, human ancestors drilled wood to make fire. As a Chinese saying goes, ''a single spark can start a prairie fire''. Both of them mean that combustion can be induced by a small fire. Our experimental results indicated that AOPs may also have induced effects. The following experimental findings directly or indirectly reflect the induced effects of AOPs from different angles. Fenton's reagent were carried out in our experiments, and the results are shown in Fig. 1. The results indicated that the reaction rate is very slow in the early stage of the reaction, only about 22% COD of EDTA have been oxidized in the first 25 min. Meanwhile, the solution temperature also raised very slowly, from 32uC raised to 42uC in the first 25 min. However, after this early stage, a sudden violent oxidation reaction has occurred, and the temperature raised very quickly, reached 80uC in the 10 minutes. About 45% of EDTA oxidation have achieved in this 10 minutes. According to usually employed hydroxyl radical theory, it implies that a large number of ?OH produced at this stage. However, it is difficult to understand why a long waiting time is needed before producing those large number of ?OH. We believe that this is a induction reaction which just like the burning process of a pile of wood from hardly ignition to strong burning, and finally burned out. The oxidation efficiency of Fenton's reaction could be strongly accelerated by adding UV radiation 18,19 . Although 254 nm UV radiation could penetrate only a very short distance into the mixed solution of H 2 O 2 and EDTA (Fig. 2). The result shows that addition of UV radiation could cut the waiting time of occurring violent oxidation reaction from 25 to 15 minutes (Fig. 1a). This phenomenon can also be explained by induced effects. Firstly, UV radiation could quickly initiate the oxidation reaction in the range where UV radiation could penetrate. Then, already happened reactions induce the whole reaction. In fact, researchers have developed a variety of methods to induce advanced oxidation reaction in recent decades, such as optical 20 , electrical 21 , ultrasonic 22 and microwave 23 . Inspired by the induced effects, we tried to feed easily oxidizable organic matter to promote the oxidation of refractory organic compound in the solution. The experiment results agree well with our notions. In the EDTA oxidation experiments using Fenton process, the method of pre-feeding potassium oxalate in the solution, similar with UV, could also cut the waiting time of occurring violent oxidation reaction. The reason for choosing the potassium oxalate as an example is that it will not add COD of the solution, and then not affect the calculation of COD removal efficiency. According to the result, the addition of 11 mM potassium oxalate could cut the waiting time from 25 to 20 minutes when 50 mM EDTA was treated by Fenton process (Fig. 1). This means that the pre-oxidation of potassium oxalate could induce the oxidation of EDTA. It will be easily understood if we connect it with a very common practice: oil usually used to quickly induce a log of wood combustion. Using easily biodegraded organic matter, such as sucrose and starch 24,25 , to increase the biodegradation is a common method in the biological process. But this strategy is rarely used in the physicochemical method. To be sure, using easily oxidizable organic matter to promote the efficiency of AOPs has important practical value for the degredation of refractory organic compounds. tert-Butanol, which is a strong radical scavenger, was adopted as the indicator for the hydroxyl radical type reaction. As shown in Fig. 1, the addition of tert-butanol markedly reduced the AOPs efficiency, indicating that the ?OH was the main active species in the process. The results suggested that these procedures (UV, K 2 C 2 O 4 ) actually enhanced the hydroxyl radical production 26,27 . In addition, it is well known that oxygen will not be consumed if there is no fuel in a combustion reaction. Similarly, when study the reaction of degradation organic compounds by UV/H 2 O 2 process, we discovered that H 2 O 2 will consumed very slowly if there is no organic compounds in the solution (Fig. 3). This is another phenomenon to reflect the similar of AOPs with combustion. Based on this experiment result, even though the exact mechanism involved are not known, it is understandable if we assume that organic compounds play an important role for ?OH production along with oxidant, catalyst, UV or other possible factors in the AOPs. From the above experimental results and discussions, we hypothesize that AOPs may have induced effects, which is similar to combustion. Although the mechanism is unclear, the findings from those studies raise new thinking in the relationship of AOPs and combustion, especially about induced effects. For example, combustion reaction is induced by the heat, but what induced the reaction of AOPs? We hope others will take up the challenge to begin addressing this important issue. It not only helps us to understand the mechanism of AOPs in more depth but also help us to find new methods to improve its efficiency from the angle of induced effects. In addition, it will raise a bold ideas: maybe all oxidation reactions have induced effects, such as corrosion, aging and passivation. Muchmore research is necessary to reveal the possibilities of induced effects in those fields.

chemsum

{"title": "Induced effects of advanced oxidation processes", "journal": "Scientific Reports - Nature"}

direct_quantification_of_nanoparticle_surface_hydrophobicity

## Abstract: Hydrophobicity is an important parameter for the risk assessment of chemicals, but standardised quantitative methods for the determination of hydrophobicity cannot be applied to nanomaterials. Here we describe a method for the direct quantification of the surface energy and hydrophobicity of nanomaterials. The quantification is obtained by comparing the nanomaterial binding affinity to two or more engineered collectors, i.e. surfaces with tuned hydrophobicity. In order to validate the concept, the method is applied to a set of nanoparticles with varying degrees of hydrophobicity. The technique described represents an alternative to the use of other methods such as hydrophobic interaction chromatography or water-octanol partition, which provide only qualitative values of hydrophobicity. ## E ngineered nanomaterials (NMs) are widely used in many consumer and healthcare products, as well as novel nanomedicines. 1 To enable a quick and reliable NMs risk-benefit assessment, there is an urgent need for robust, standardised and reliable methods for their characterisation and toxicity screening. For this purpose, the understanding of the behaviour and fate of these NMs when in contact with biological systems is important. Physical and chemical properties such as size, surface chemistry and surface charge have been identified as essential parameters to determine, because they affect the NMs' mode of action in a given environment (water, buffer, biological fluid, etc.) through different surface molecular interactions. In particular, hydrophobicity is considered as an important property since it has a critical role in various biological processes such as protein adsorption, 2 interaction with biological membranes, 3 cellular uptake, 4,5 immune response, 6 and haemolytic effect. 7 It is recognised that hydrophobicity is a key property to be controlled for nanomedicine applications since it has a direct influence on the stability and bio-distribution of nanovectors. 3,6,8 Only a few methods are currently available for characterisating the hydrophobicity of NMs 9 -for example surface adsorption assays, 10 NMs relative affinity for reference phases, and hydrophobic interaction chromatography. 11 None of those methods enables, for all NM types, a full characterisation and quantification of the NMs hydrophobicity and they all involve expensive and time-consuming analytical techniques. The development of a fast and reliable general method for NMs hydrophobicity characterisation is therefore of great interest. We described in a previous publication 12 a new method for the separation of NMs according to their hydrophobicity based on a set of collectors composed of fluorinated hydrophobic surfaces whose surface energy components can be modified and finetuned by the layer-by-layer (LbL) deposition of polyelectrolytes. A proof-of-concept study of the method was carried out with hydrophobic and hydrophilic polystyrene nanoparticles. The experimental results were qualitatively supported by the eXtended Derjaguin Landau Verwey Overbeek (XDLVO) theory, 13 enabling the assessment of the different forces in play. Here we extend this method to quantitatively determine the surface energy components of the NMs by measuring their binding affinity to the collectors' surfaces, via analysis of their adsorption kinetics. The adsorption kinetics is calculated by measuring the number of nanoparticles binding to the different collector as a function of time, by Dark-Field microscopy. The method is particularly suitable for nanoparticles with large light scattering capability (for example noble metals with typical size >50 nm), but it is also applicable to other materials such as SiO 2 with typical size >200 nm. On the other hand, by using a special dark-field condenser 14 or other single-particle microscopy techniques 15 it is in principle possible to detect NMs of any material and down to sizes of the order of 20 nm. The surface energy potential acting between each NM and the collector is then calculated using the XDLVO theory. The energy barrier potential between the NMs and the collector represents the potential energy at which the NMs are repelled by the collector surface. 16 This energy barrier potential is inversely proportional to the binding affinity: the larger the energy barrier potential, the lower the affinity. In some conditions, the NMs may be completely repelled by the collector due to very high energy barrier potential. Electrostatic forces are mainly responsible for the formation of the energy barrier; hence for an accurate calculation, the Z-potential for both the surfaces and the NMs should be rigorously measured. The NM's surface energy and in particular the degree of hydrophobicity is then calculated by comparing the NM binding affinity of two or more collectors. The experiments were performed with Au and SiO 2 nanoparticles (NPs) with different degrees of hydrophobicity to test the validity of the approach. An advantage of this method is that it can cover all the surface energy range with only one set of collectors, and a quantitative determination of the surface energy is possible thanks to several measurements on different collectors. ## Results Surface characterisation of collectors. The first step of the method is to determine the binding affinity of the AuNPs with the different collector surfaces. Each collector prepared as described above is characterised by distinct surface energy components that control the binding affinity of the nanomaterial. First, the binding kinetics of AuNPs was measured on each collector, in a buffer solution (Phosphate Buffer, PB 10 mM, pH = 7). The buffer composition has been chosen in order to partially neutralize the charges present on both the surface and the AuNPs, which would otherwise lead to a long-range repulsion. The principle of the method is described in Fig. 1. The selectivity and specificity of the AuNPs binding to the surfaces strongly depend on characteristic of the interaction forces such as interaction strength as a function of distance and attractiveness and repulsiveness. The collector and nanoparticles characterisation, the study of the binding affinity of the AuNPs on the different collectors, calculation of the acting potential between the AuNPs and the collector surfaces, and extraction of the surface energy component of the AuNPs from the adsorption rates of the AuNPs on the collectors surfaces are presented below. In order to fabricate the collectors with different surface properties, a standard Si wafer has been coated first with a plasma deposited layer of polytetrafluoroethylene PTFE (hydrophobic) and then with several layers of polyelectrolytes (PE) (poly (diallyldimethylammonium chloride, PDDA and polysodium 4styrene sulphonate, PSS) for giving the surface a more hydrophilic character. The subsequent adhesion of PEL layer also permits the modification of the surface free energy components of the collectors. Five surfaces denominated T0 (PTFE), T1, T2, T3 and T4 are thus produced with the numerical component of the name corresponding to the number of PEL layers deposited. According to the Owens-Fowkes-Wendt (OFW) theory, 17,18 the total surface energy of a solid is the sum of the dispersive component (taking into account the non-polar interactions), called γ LW (Lifschitz-van der Waals component) and of a polar component γ AB (acid-base component). Solid materials with low γ AB are called "hydrophobic". The increase of the γ AB of a solid corresponds to an increase of its hydrophilicity. The PTFE was chosen because of its very low γ AB . The surface free energy components have been determined by measuring the contact angle of the solid surfaces with a polar (water) and a dispersive solution (Bromonaphtalene). The surface energy values of the as-deposited PTFE were 0.9 mJ/m 2 and 19.3 mJ/m 2 for respectively polar and dispersive components. Then, PEL layers were deposited on the collector to increase both surface energy components. The deposition of the cationic PEL leads to a decrease of the initial PTFE z-potential (−60 mV) to values close to zero, while the anionic PEL deposition does not change the negative z-potential (−60 mV). Surface properties of the PTFE and of the T2 collectors present very similar z-potential values while the T2 collector has a higher surface free energy component value, i.e. polar component increasing by a factor 5 and the dispersive components increasing by only 1.4. 12,19 The increase of the surface free energy has been directly measured by the F-D curves using different functionalized gold-coated tips (Supplementary Figure 1). The adhesion force between the PFT (polyfluorotetraethylene) modified Au-coated silicon tip and respectively the PTFE and T2 collector were measured as 642 ± 200 nN and 248 ± 100 nN (Supplementary Figure 2). This higher adhesion force between the PFT-modified Gold on the tip and the hydrophobic PTFE is attributed to the hydrophobic interacting forces resulting from their very low polar surface energy component (the tip-surface new interface is energetically more favourable than the two surfaces alone). The average adhesion force of a polyethylene oxide (PEO)-modified Au-coated tip on both the PTFE and the T2 collector (not shown) was 100 ± 76 nN indicating that the hydrophilic surface functionalization reduces the strength of interaction between the surfaces, consequently reducing their force of adhesion. The atomic force microscopy (AFM) analysis indicates that collector surfaces are rather homogenous with roughness increasing from 0.29 nm to 0.85 nm showing that the morphology of the surfaces is not dramatically affected by the formation of the polyelectrolytes layers. The z-potential was measured for different pH. A negative z-potential was obtained for the whole range of pH, especially for the PTFE non-modified and the PSS layers, and closer to neutral for the different PDDA layers. This result can be explained by the fact that the PDDA is positively charged and the PSS and PTFE negatively charged. A more comprehensive overview of the surface characterisation of the collectors is given in ref. 12 . Finally, the set of collectors presents the following features: 1. Relatively low surface roughness, root mean square (RMS) roughness <2 nm. 2. Surface chemical homogeneity, as indicated by the X-ray photoelectron spectroscopy (XPS) and time-of-flight secondary ion mass spectrometry (Tof-SIMS) analysis. 12 3. Surface homogeneity at the nanoscale level in terms of the adhesion properties as indicated by the force spectroscopy measurements. 4. A surface Zeta-potential which is negative (for the PSS terminated surfaces) or slightly negative (for the PDDA terminated surfaces). 5. Different values of LW and AB components. Nanoparticles characterisation. Supplementary Figure 3 (TEM) images obtained by spotting 4 μl AuNPs stock solution onto a C/Formvar TEM grid and dried overnight in air. Although the morphologies observed were mainly spherical, a certain degree of faceting characteristic of large AuNPs was also observed. Size distributions determined by TEM image analysis on over than 100 objects were narrow with an average value of 60.2 ± 7.9 nm. The AuNPs were functionalized with different amounts of PEO-ligands in order to produce different surface coverage on the surface i.e. different hydrophobic characters. The selected ligand chain was functionalised with a SH-group at one terminus to provide a strong affinity for gold, a central non-polar alkyl chain to provide to the structure the ability to self-assemble in dense layer that excludes water due to the hydrophobic core and a PEO sequence to enhance stability in water. ## illustrates typical examples of transmission electron microscopy The PEO surface coverage has been first characterised by UVvis adsorption measurements (Supplementary Figure 4). The results show that the increase of PEO content in the solution results in a red-shift of the surface plasmon resonance wavelength from 543 nm for uncoated to approximately 546 nm for all the functionalized PEO-Au NP samples (Table 1). The shift from the uncoated and the PEO coated surfaces might be due to slight differences in the bulk refractive index of the dispersant. Dynamic light scattering (DLS) also confirms the successful functionalization of AuNPs 60 nm with an increase of the hydrodynamic diameter from 62.7 nm for pristine AuNPs to 68.7 nm for the highest PEO concentration (Table 1). Z-potential measurements show a slight decrease in the negative surface charge from −45.7 mV for pristine AuNPs to −39.4 mV for the AuNPs coated at the highest PEO concentration. Differential centrifugal sedimentation (DCS) shows that the sediment time increases with the PEO concentration in the solution. This shift toward higher sedimentation times is the net result of (1) the increase of NP diameters and (2) the decrease of the NP apparent densities due the PEO binding. By combining the sedimentation times measured by DCS and the DLS diameter, 20 the mass of absorbed PEO molecules on the AuNPs as a function of the PEO concentration in solution and therefore the number of molecules per NP can be calculated. The values are reported in Table 1. A theoretical full coverage of the AuNPs by the PEO molecules can be calculated by dividing the available gold surface area of the particles by the footprint of a alkanethiolate molecule (21.4 2 ), i.e. the minimum space occupied by an absorbed molecule as as estimated by electron diffraction studies of monolayers of alkanethiolates on Au(111) surfaces. 21 We must underline that the obtained value of the coverage represents a rough approximation, since it corresponds to a perfect packing of the ligand on the surface and doesn't take into account physical effects such as the steric repulsion of the PEO chains, the gold surface inhomogeneity, the possibility different orientation due to the PEO chains interaction with the gold surface, occupying more space, etc. Using the TEM diameter (60.2 nm) and considering the AuNPs as perfect spheres, we calculated a maximum number of 55 × 10 4 PEO molecules per AuNP. According to Table 1 this number is reached for a volume of PEO added to the AuNPs of 3.2 µl. Any additional PEO added to the solution would potentially lead to the formation of an excess of PEO on the surface of the AuNPs, i.e. multilayers. More interestingly, the PEO functionalized AuNPs were characterised by two other techniques: protein adsorption by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and affinity to the water phase by water/octanol partition. The amount of proteins detected by the SDS-PAGE on the AuNPs was very low, since the total amount of AuNPs available was very small. However, it was found that a detectable quantity of Albumin was found on both the pristine AuNPs and the AuNPs treated with low volumes of PEO (Supplementary Figure 5, Supplementary Note 2). By plotting the SDS-PAGE Albumin band intensity as a function of the volume of added PEO, it is clear that the intensity decreases and it goes to zero for a PEO volume larger than 3 µl. This means that the PEO surface coverage obtained by adding a volume of 3 µl to the solution was sufficient to prevent any protein adsorption on the surface of the particles. A similar trend is observed in the water/octanol partition coefficient. Briefly, experiments were carried out both with pristine AuNPs of 60 nm and the differently PEO-functionalized AuNPs at a concentration of 0.5 mM of gold. As expected, pristine hydrophobic AuNPs coated with weakly bound citrate surfactant interacted strongly with octanol resulting into complete aggregation clearly visible in the interphase (Supplementary Figure 6). By increasing the degrees of PEO-functionalization, a progressive decrease in the amount of aggregation together with higher amount of NPs in the aqueous solution can be observed (increases of red colour intensity). For volume of added PEO larger than 3 µl, the AuNPs did not interact at all with Octanol as a result of their hydrophilic character. These results confirm the decrease of hydrophobicity as the PEO coverage increases and a saturation of the surface for 3 µl of thiolate added to the suspension. The proportion of NPs in water was calculated by measuring the intensity of the red colour component in Supplementary Figure 6 for each vial and normalizing to the intensity of the red colour of the original solution of NP with the same concentration in water. These two results suggest that, as calculated by DCS/DLS combined measurements, a full monolayer of PEO is reached by adding a volume of PEO larger than about 3 µl to the pristine AuNPs solution. The addition of larger volumes of PEO did not further modify the surface properties of the AuNPs. The results from these three experiments are summarized in Fig. 2. Furthermore, we characterised the hydrophobicity/hydrophilicity degree of the Au-Citrate stabilised and of the Au-PEO NPs, by contact angle measurements performed on a dried film of NPs. Contact angle measurements show that Au-citrate NPs are hydrophobic with a corresponding contact angle with water of about 95°. The AuNPs with the addition of 7.2 µl of PEO are on the other hand very hydrophilic, with a contact angle of 23°. A similar behaviour was observed on the flat Au surface, where the pristine Au surface showed a contact angle of 84°with water and the PEO-functionalized Au flat surface a contact angle of 40°. The previously described combination of characterisation techniques of the AuNPs allowed us to classify the AuNPs according to their effective PEO surface coverage. The surface coverage is linearly proportional to the amount of PEO added to the pristine AuNPs solution between 0 and 3.2 µl of volumes. The proportionality factor, k, between volumes was added and the PEO surface coverage was calculated as follows: The AuNPs samples A, B, C, D, F, G and L, corresponding to the PEO surface coverage indicated in Table 1 were used to study the binding affinity with the different collectors. Nanoparticles binding study. The AuNPs affinity has been calculated by directly measuring the real-time binding kinetics of the NPs to the different collectors. The binding curves are built by using a script created in the software ImageJ allowing to automatically detecting the NPs as round objects in the video sequence. Typical sequence images of the adsorption of NPs on the T0 collectors are shown in Fig. 3, for the Au-citrate NPs (hydrophobic) and for the Au-100% PEO coated NPs (hydrophilic). Each round object in each frame corresponds to one AuNP; for each frame, the NPs are counted and the position of each NP is recorded. It is either possible to detect the total number of NPs for each frame or to track the NP motion on the surface. Only the NPs perfectly in focus with the surface plane are detected with a focal depth of 4000 nm. Taking this into account the nanoparticles were detected in the volume above 4000 nm from the surface: these particles are moving much faster that the time resolution of the camera (>ms). Hence we are able to observe only the nanoparticles slowed down by the contact with the surface, which are moving or rolling. The number of NPs detected is then related to the affinity of the NPs to that particular surface. It should be noted that AuNPs with different PEO coverage are characterised by the same core size and a similar negative Zpotential. The choice of negatively charged collectors (terminated by the anionic polyelectrolyte) and buffer conditions was made in order to minimize any possible electrostatic interaction between the nanoparticles and the surfaces while keeping the stability of the colloids. Uncoated Au-citrate NPs dispersion stability is maintained also thanks to their very negative value of the zpotential while the increasing PEO coating coverage contributes to the NPs stability by steric repulsion. In the absence of an energy barrier, the number of NPs reaching the surface as a function of time Γ Ads ðtÞ is determined by the diffusion constant, D e via Eq. 2: Where Σ is the area of measurement and C n is the bulk concentration of NPs. The maximum velocity of adsorption v max of the NPs of radius r and bulk concentration C n on an area Σ is then Eq. 3 shows that, in the absence of an energy barrier, the NPs adsorption velocity is determined by the radius of the NPs, through their diffusion constant. 22 If an energy barrier, ΔG max , is present, the local concentration of NPs, C loc at a distance lower than the energy barrier distance is related to the bulk concentration by Eq. 4: Eq. 4 is the classical Maxwell-Boltzmann distribution which determines the distribution of objects in the presence of an energy barrier. 22 The generalized NPs adsorption velocity is then That substantially means that the NP adsorption velocity is reduced by a factor e ΔG max =kT in the presence of an energy barrier as compared to the maximum velocity. The v max can be calculated by Eq. 3, knowing the size of the NPs or it can be measured directly by observing the rate of arrival of the negatively-charged NPs to a positively-charged surface (such as T1 or preferably T3 since one layer of polyelectrolyte on the PTFE is not sufficient to guarantee full surface coverage). On the other hand, by measuring the velocity in the presence of an energy barrier one can calculate ΔG max =kT as The ratio v v max can vary between 0 and 1 and it is a direct For the 60 nm AuNPs the maximum adsorption rate is obtained when the activation energy barrier is zero, and the exponential in Eq. 5 is equal to 1. In the cases where the gravitational transport is not negligible for the timescale observed (dense and large NPs) Eq. 2 should be corrected with a factor proportional to the time and the sedimentation velocity. where with ρ NP and ρ f being the NP and fluid mass density respectively, and η the fluid viscosity. Eq. 7 takes into account that the local concentration of the NPs is increasing with time due to the sedimentation of the NPs. The adsorption curves are fitted by a modified Eq. 9: where α 1 and α 2 are respectively the linear and quadratic factors (for the polynomial function in t 1/2 ). The maximum adsorption rate which depends on the geometrical properties of the NPs is plotted in Fig. 4 in comparison with the kinetics obtained for the different NPs for the different collectors. The rate of adsorption of the NPs on the different collectors is decreasing while increasing the acid-base (polar) component of the collectors. This is more evident for the non-PEO-coated NPs. Then the rate of adsorption is found decreasing drastically with the increasing PEO% coating on the NPs. For a nominal coating of PEO larger than 25% of the NPs, the adsorption rate is close to zero. The principal information that can be extracted from Table 2: on the T4 collector, the rate of adsorption is zero for all the NPs, the rate of adsorption is larger on the collector T0 than on the collector T2 for all the NPs and the difference of the rate of adsorption between T0 and the collector T2 is decreasing by increasing the PEO coating. Calculation of the surface free energy component of the nanoparticles. According to the XDLVO theory 23 , the total interaction energy G tot can be expressed as where G el G AB and G LW are energies relative to the electrostatic, acid-base and Lifshitz-Van der Waals interactions respectively. The three potential depends on the distance between the NP and the surface. Electrostatic interaction energy is where d is the separation distance between the NP and the surface, ζ N and ζ S are the surface charge of the nanoparticle and the collector surface respectively. 1/κ is the double layer thickness, which is expressed from Eq. 12. κ ¼ e 2 εkT where ε is the permittivity of the medium, e is the charge of electron, k is the Boltzmann constant, T is the temperature, z i is the valency of the ions i, and n i is their number per unit volume. The Lifshitz-Van der Waals ΔG LW components to the free energy of interaction between a nanoparticle and surface are calculated following the extended DLVO theory: where d is a separation distance between the NP and the surface, and r is the radius of the nanoparticle. H the effective Hamaker constant for the NP-collector-water system, which can be expressed as Looking at the dependence of the G LW on the distance d, the Lifshitz-Van der Waals force then active for very short distances (d < 1 nm), only for the particles that are able to overcome the energy barrier. While the analytical expressions for the electrostatic potential and the Lifshitz-Van der Waals potentials are well known and commonly accepted, the acid-base interaction potential is mainly represented as an empirical formulation based on experimental observations 24 and direct measurements of the interaction potential between two surfaces (sphere-sphere, sphere-plane, plane-plane) in a polar medium or in an electrolytic solution. The G AB is including all forces involving the structural reorganization of the water molecules around two surfaces, depending on the degree of wettability of the surfaces involved. For a sphere-plane system where d 0 is the minimum separation distance between the NP and the surface, taken generally as 0.158 nm for many different kinds of substrates and d the separation distance in nm. G AB is defined as a short-range acting potential, having an exponential decrease with the distance. The field of interaction of the potential is mainly determined by the correlation length λ, expressed in nm. Various values for λ have been reported in literature, ranging from 0.2 nm to up to 13 nm 24 . The nature of the two interacting surfaces intervenes in the AB potential with the term ΔG AB , which can be expressed as where the term ffiffiffiffiffiffiffi γ AB i p refers to the polar component of the surface free energy for N = nanoparticle, W = water and S = surface. The interaction energy maps were calculated using the function wizard included in the software OriginPro 2015. The values for the interaction energy are given in kT units (1 kT = 4 × 10 −21 J). The total interaction energy is expressed in Eq. 10 for the interaction of a NP and a surface as a function of the distance and of the hydrophobic correlation length λ. Even if the range of possible and measured value in literature for λ is broad (between 0.6 nm and 13 nm), we decided to keep it in the range between 0.6 nm and 2 nm, in order to take into account possible influences of the radius of curvature of the NPs and of the roughness of the collectors 24 . The heat-maps for the 15 nm, 60 nm and 200 nm hydrophobic NPs on hydrophobic collector are shown in Supplementary Figure 7 and illustrate the influence of the value of λ on the potential profile as a function of the NP size (Supplementary Note 3). The blue colour map indicates that the interaction energy between the NPs and the surface is equal to or lower than 0 kT. When interaction energy between the NP and the surface is equal to or larger than 5 kT the colour map is marked as red. 5 kT is considered as the threshold value for an energy barrier that would inhibit completely the adsorption of NPs to the surface. The corresponding adsorption velocity would be v _5 kT = 0.0067. v_max. The heat maps for 60 nm bare AuNPs with T0 hydrophobic and T2 and T4 hydrophilic collectors are presented in Fig. 5. To illustrate the variation of the energy barrier, in assuming a value of 1.65 nm for λ, (dashed line in Fig. 5) the energy barrier maximum is found lower for the T0 (about 0.5 kT) than for the T2 and T4 collectors. The position of the maximum in z is also depending on the collector, being larger for T0 (about 13 nm) and smaller (or closer to the surface) for T2 and T4. For each collector, the energy barrier γ AB N must verify the following system of non-linear equations: where the first equation in bracket corresponds to the value of the maximum of the energy barrier and the second equation corresponds to the condition of the maximum (first derivative equals to zero). According to Eq. 6, the maximum of energy barrier determines the reduction of the velocity of adsorption with respect to the maximum velocity (when the energy barrier is zero). The varying parameters to verify the system in Eq. 17 are: the NPs-surface distance z i which depends on the collector. The z i is the NP-surface distance at which the energy barrier has a maximum, as a function of the collector, the hydrophobic interaction characteristic distance, λ and the polar component of the surface free energy of the NPs, γ AB N . In order to solve the non-linear system of equations in Eq. 17 we need at least two collectors, in order to have four equations with four unknowns, namely z 1 , z 2 , λ and γ AB N . The system of equations, for each couple of collectors has been numerically solved using the value boundaries listed in Supplementary Table 1. The boundaries were chosen as follows: z i is the typical range of interaction of the electrostatic forces; the γ AB N values were chosen between the value corresponding to PTFE (hydrophobic) and the one of water (the maximum hydrophilicity). The tolerance for the G tot was set at 10% of the value calculated between a hydrophilic particle and a hydrophilic collector. The numerical solution for the T0-T2 couple of collectors for the different AuNPs is listed in Table 3. The results obtained from the solution of the system of equations for the T0-T2 couple of collectors are physically meaningful. The fitted value for λ is very similar for the different NPs and it is very close to the one obtained graphically from Fig. 5. A value for each NP of γ AB N is then determined with this method. According to what was expected and also measured by the contact angle on the dried particles pellets, the γ AB N is the lowest for the Au-citrate NPs (1.57 mN/m 2 ) and increases for the PEO coverage percentage. For coverage larger than 25% the AuNPs are largely hydrophilic with a large value for the polar component of the surface free energy (30.95 mJ/m 2 ). The graphical evolution of the γ AB N with the PEO coverage is shown in Fig. 6. A second type of model NPs has been used, to verify the validity of the proposed method. 200 nm hydrophilic SiO 2 NPs were modified with different chemical groups to increase the surface hydrophobicity. Three functional groups were used to modify the NP surface, consisting of alkyl chain of different length: propyl, butyl and dodecyl groups. The SiO 2 NPs modified with dodecyl groups were not stable and started to aggregate, forming dimers and trimers as shown by the DLS measurement. The propyl and butyl functionalized SiO 2 NPs were stable for days as shown by DLS measurement. The adsorption rate of pristine SiO 2 and epoxy modified NPs on T0 (and consequently on T2 and T4) was zero. The adhesion of propyl and butyl functionalized SiO 2 NPs to T0, the hydrophobic collector, was slow but different from zero, and it was evaluated as α 1 = 2.15 t −1/2 and α 1 = 2.17 NP/t −1/2 , respectively. The maximum adsorption rate, calculated with Eq. 2, is 232 NP/t 1/2 , meaning that the potential barrier is relatively large, around 6.97 kT and 6.98 kT. The results are summarized in Table 4. This result is confirmed by the contact angle measurement with water of a silica surface functionalized with propyl and butyl groups: the contact angle is about 10°and 15°for the propyl and butyl, respectively, confirming the highly hydrophilic nature of these surfaces. ## Discussion In this work a method is presented for the quantitative characterisation of nanoparticles hydrophobicity by measuring their affinity towards specifically functionalized surfaces. The determination of the affinity of NPs towards substrate surfaces with different hydrophobicity degrees enables the direct characterisation of the NPs having unknown surface functionalization and residual hydrophobicity in a direct manner. The quantification of the surface energy of the NPs is possible by comparing the evaluation of the affinity of the NPs with different collectors and the comparison with the XDLVO theory, which takes into account the hydrophobic forces. The general method is highly sensitive to variation of the surface free energy polar components and could be adapted for the direct evaluation of the stability of the anti-fouling coatings which are usually used to prevent the agglomeration of the NPs in biological complex media and living bodies and which, through the formation of the protein corona, may lead to inflammatory responses and/or uncontrolled clearance of the NPs. The proposed method allows the quantitative characterisation of NP hydrophobicity in solution and thus is potentially highly relevant to important applications in the field of nanomaterial safety assessment in consumer and industrial products. ## Methods Surface preparation. Silicon wafers (Si(100); diameter, 50 mm; resistivity, 1−20 Ω cm) supplied by ITME (Warsaw, Poland) were used as the substrate for the whole study. Before modification the wafers were washed with ethanol and water and dried under nitrogen flow. Surface modification. The silicon substrate was modified by different deposition processes in order to tune the surface hydrophobicity. First, polytetrafluoroethylene (PTFE) coating was plasma deposited to generate a hydrophobic surface. The deposition was performed using pure octofluorocyclobutane (C 4 F 8 ) as the gas precursor at a pressure of 3.5 Pa, applying a power of 142 W for 5 min 12 . In order to tune the surface hydrophobicity, a layer-by-layer deposition of two polyelectrolytes (PELs) was then performed. The PTFE modified substrates were incubated for 2 min in poly(diallyldimethylammonium chloride) (PDDA) 2% solution in water or in poly(sodium 4-styrene sulphonate) (PSS) 2% in water for the self-assembly deposition of each PEL layer, starting from PDDA (positively charged) and alternating with PSS (negatively charged). After each step, the substrate was rinsed with milliQ water and dried under nitrogen flow. The surfaces obtained are referred to as T1, T2, T3 and T4 depending on the number of PEL layers deposited. Collector surface characterisation. In order to fully characterise the collector surfaces, several techniques were used. The thickness and refractive index of each deposited layer were measured by Ellipsometry (Vase VUV TM J.A. Woollam Co.). The contact angle and surface energy of the surfaces were determined by using a Digidrop TM goniometer with 2 liquids (water and 1-bromonaphtalene). The surface topography and roughness of the surfaces were measured using AFM (NT MDT Russia). Finally, Z-potential measurements were performed for a range of pH from 3 to 10 with a step of 0.5 in order to characterise the surface charge. The Zpotential was measured for independent samples just after adjusting the pH of the dispersions with either 1 M NaOH or 1 M HCl and at a total NaCl concentration of 1 mM to keep the conductivity at approximately 1 mS/cm. AFM was also used to directly measure the different adhesion forces present between the collector surfaces and functionalised tips to ensure that the surface properties evaluated were constant over a 1 μm range. A conventional silicon tip for contact mode (NT MDT C-Probe, spring constant k = 0.02 N/m and nominal radius of curvature of 15 nm) was employed. The instrument returns the values for the deflection of the cantilever in nA (the difference of current at the 4-quadrant optical detector of the AFM). This value was translated in force values (nN) by calculating the actual deflection of the cantilever in nm and by multiplying the value for the elastic constant of the cantilever used. The cantilever was coated with a 20 nm-thick layer of gold by magnetron sputtering (a 1 nm layer of titanium was previously deposited to ensure the adhesion between the gold and the silicon). The gold-coated tip was then immersed in 1 mM ethanol solutions of thiolated alkyl group terminated with PEO (Polyethylenoxide). The tip was then rinsed in ethanol and ultrapure water and gently dried in flowing N 2 . The gold-coated and PEO-functionalised tip was then used to perform so-called "Force-distance curves" (F-D curves). Briefly the tip was brought in contact with the surface in a certain location, and then the tip was retracted by a few micrometers by means of a piezoelectric crystal. Then a series of approaching-retracting curves were acquired in an area surrounding the contact position and the deflection of the cantilever was recorded as a function of the piezo-position. The measurements were performed in ultrapure water. A minimum of 100 curves was acquired. The adhesion force between the tip and the surface was measured for each curve as the difference between the zero-force line and the minimum of the "snap-off" force. The adhesion forces were then averaged for all the curves (see Supplementary Note 1). Nanoparticle synthesis functionalisation and characterisation. Two sets of NPs, Au and SiO 2 were used for the experiments. AuNPs were synthesised (see below) using gold (III) chloride trihydrate (>99.9%), and trisodium citrate dihydrate (>99.9). Surface modifications of SiO 2 particles was done using (3-glycidyloxypropyl)trimethoxysilane (GPTMS) (GPTMS, >98%), propylamine (>99%), butylamine (>99%) and dodecylamine (>99.5%). All reagents were purchased from Sigma-Aldrich and used as received without further purification. Ligand 2-(2-[2-(11-mercapto-undecyloxy)-ethoxy]-ethoxy)-ethoxy-ethoxy-ethoxy-ethoxy-acetic acid (PEO-ligand) was purchased from ProChimia and kept under N 2 and in the freezer at −20 °C. SiO 2 NPs of approximately 200 nm (NS-0200A) at a concentration of 7.8 × 10 12 NP/mL were purchased from MSP Corp. Centrifugal filter units (Amicon, USA) were washed three times with milliQ-water at 3500 rcf 10 min before their use. All solutions were prepared with ultrapure water (Millipore Milli Q system, resistivity, 18.2 Ω cm). The synthesised AuNP dispersions were stored in the fridge at 4 °C. Synthesis of the AuNPs. AuNPs of 60 nm in diameter were prepared by a seeding-growth four-step procedure. Initially, AuNP seeds of approximately 15 nm in diameter were synthesised with a modified Turkevich method 25 using a specialised microwave Discover apparatus (CEM Corporation, USA). Briefly, 5 mL of aqueous gold (III) chloride trihydrate (10 mM) were added to 95 mL of milliQwater in a single-necked 100 ml round bottom flask equipped with a magnetic stirrer and a glass condenser column. The flask was mounted in the microwave, heated rapidly (<1 min) to 97 °C while stirring and then held for 5 min using a maximum microwave power of 150 W. Under vigorous stirring, 2.5 mL of sodium citrate (100 mM) was injected and the reaction mixture was maintained at 97 °C for 20 min after which the reaction vessel was rapidly cooled to 60 °C with compressed air and then allowed to cool to room temperature. The nominal concentration of gold in the AuNP dispersion was 0.5 mM. The 15 nm AuNP seeds were first grown to 30 nm, and then from 30 nm to 45 nm. The last stage of the synthesis to 60 nm AuNPs was carried out by the regrowth method from the 45 nm AuNPs. The as-synthesised seeds (25 mL) of AuNPs of 45 nm were diluted in milliQ-water (60 mL) and mixed with 2.8 mL of sodium citrate (100 mM) and 1.25 mL of aqueous gold (III) chloride trihydrate (10 mM). The pH of the resulting solution was adjusted to a value of 9.0 with aqueous NaOH (200 mM, 0.42 mL) and the mixture was heated to 60 °C for 48 h in order to produce AuNPs of approximately 60 nm in diameter. The nominal concentration of gold in final AuNP dispersion was 0.24 mM. The 60 nm AuNP dispersions were purified twice by centrifugation (2500 rcf, 20 min, 4 °C) followed by redispersion in the same volume of MilliQ water, and Functionalisation of SiO 2 NPs with different hydrophobic ligands. Functionalisation of commercial SiO 2 NPs was performed as described in the literature 27 . Briefly, a sample of nano-silica NS-0200A (0.5 mL) was diluted in water (1.5 mL) and the pH increased with the addition of 1 M NaOH (1 µL). GPTMS (0.5 µL, 2.3 μmol) was immediately added and the reaction mixture was stirred at room temperature for 24 h. Modifications of epoxy-functionalised SiO 2 NPs were obtained by the addition of 100 mM solutions in water of propylamine, butylamine and dodecylamine (final concentrations of 2.24 mM) at pH = 9. The mixtures were left to react for 24 h and all the samples were then purified using centrifugal filter units of 10 kDa MWCO Amicon Ultra-15 (3500 rcf, 5 min) via two washing steps with water in order to eliminate the excess of non-conjugating molecules. DLS measurements showed no major differences in the mean hydrodynamic size for any of the samples, which was a good indicator of the colloidal stability of these solutions. Surface charges of pristine and functionalised SiO 2 NPs were determined by Z-potential measurements at a value of pH = 7.5 after adjusting with 1 M HCl and at a conductivity of 1 mS/cm after adjusting with 1 M NaCl. The final nanoparticle dispersions were diluted in a buffer (Phosphate buffer (PB) 10 mM, pH = 7.5) to a final concentration of 1.4 × 10 11 NP/mL (i.e. 55 times less than the original concentration). Nanoparticle characterisation. AuNPs were imaged using a transmission electron microscope (TEM, JEOL 2100, Japan) at an accelerating voltage of 200 kV. The samples were prepared by placing a drop (4 µL) onto ultrathin Formvar-coated 200-mesh copper grids (Tedpella Inc.), followed by drying in air at room temperature. For each sample, the size of at least 100 particles was measured to obtain the average and the size distribution. Digital images were analysed with the ImageJ software, using a custom macro to perform smoothing (3 × 3 or 5 × 5 median filter), a manual global threshold and the automatic particle analysis of ImageJ. The programme can be downloaded from http://code.google.com/p/psa-macro. A circularity filter of 0.8 was used to exclude agglomerates that could occur during drying. UV-vis adsorption spectra were recorded with an Evolution 300 UV-vis spectrophotometer (Thermo Scientific, USA) at room temperature. DLS and Zeta-potential measurements were obtained using a Zetasizer Nano ZS instrument (Malvern Instruments, UK). Hydrodynamic diameters were calculated using the internal software analysis from the DLS intensity-weighted particle size distribution. Differential centrifugal sedimentation. DCS measurements (instrument model DC24000UHR, DCS Instruments Inc, USA) were performed in an 8-24 wt% sucrose density gradient with a disc speed of 22,000 rpm. Each sample injection of 100 µl was preceded by a calibration step using certified polyvinyl chloride (PVC) particle size standards with a weight mean size of 280 nm. Contact angle. The contact angle with water was measured using a Digidrop Contact Angle metre (GBX, France). Contact angles of the differently functionalised AuNPs were measured by spotting a drop of sample solution (10 µL) on a silicon surface to form of confluent film. Then, a 0.5 μL water droplet was deposited onto this film and the contact angle was measured. Contact angles were also measured on flat Au-surfaces functionalised with PEO-ligand. Au deposition on silicon wafers (3 min, ~100 nm Au) was performed by physical vapour deposition. The surfaces were cleaned with sonication in EtOH, followed by several washing steps with EtOH and MilliQ H 2 O. They were then functionalised using 10 mM aqueous solution of PEO-ligand, and dried. A 0.5 μL droplet of either water or 1-bromonaphtalene was deposited on the surface and the contact angle measured. For measurements of contact angles of flat SiO 2 -surfaces with hydrophobic ligands, silicon wafers were treated with O 2 plasma (treatment time = 5 min) and then exposed to 1 mM aqueous solution of GPTMS at room temperature for 24 h. After washing with MilliQ H 2 O, wafers were exposed to 2.24 mM solutions in water of propylamine, butylamine and dodecylamine at pH = 9 for 24 h at room temperature before further washing with MilliQ H 2 O. After drying, a 0.5 μL water droplet was deposited on the surface and the contact angle measured. SDS-PAGE gel electrophoresis. AuNPs of 60 nm diameter, both pristine and PEO-functionalised were incubated with human serum (Sigma-Aldrich) for 24 h at 37 °C. The mixture was centrifuged (10,000 rcf, 5 min, 4 °C) and the supernatant carefully removed. The NP pellet was subsequently washed with 1× PBS (Gibco). This washing procedure was repeated three times. The final pellet was suspended in 20 µL Pierce Lane Marker reducing sample buffer (ThermoFisher) and incubated at 95 °C for 5 min. After a short spin down, the supernatant was loaded in 12% SDS polyacrylamide gel and run at 110 V, 25 mA in 1× SDS running buffer. After electrophoresis, the gel was Coomassie stained. Scanned gel images were analysed with the ImageJ software using sharp + smooth + brightness/contrast adjustment, followed by splitting of colour channels (best contrast for Red and Green). Quantification of the proteins was made by calculating pixel intensity in the central band (lower intensity = dark = more proteins) and plotted as the inverse of the pixel intensity. Nanoparticles adsorption study by dark-field microscopy. Dark field microscopy (DFM) videos were recorded to measure the NP adsorption rates on the different collectors. Image analysis was performed using the open source graphics software ImageJ (http://rsb.info.nih.gov/ij/). A special data processing protocol was developed to enable automatic frame-by-frame analysis of the collector surface coverage by the NPs. The coverage boundaries were identified as red and green stains on a black background and only those with a size between 7 and 200 pixel units and circularity value between 0.20 and 1.00 were taken into account in calculating the degree of surface coverage (using the Analyse Particles function of ImageJ). After background correction, noise reduction (Despeckle, Kalmann Stack Filter with values 0.05-0.80), brightness, contrast and colour balance adjustment, the Unsharp Mask filter (radius 4.0 and mask weight 0.7) was applied. The image was then converted to an 8-bit file, this being required to adjust the threshold and analyse the particles with the provided function. This procedure allowed real-time kinetics analysis with a microfluidic chip, calculating the coverage degree of each substrate as a function of time. The association rate could then be determined, providing a quantitative measurement of the affinity of the NPs for the different collectors. These real-time measurements were carried out in static mode by using commercial slides provided by Ibidi (Sticky-Slide IV, Germany) with a channel volume of 30 µl and cell height of 0.4 mm. The channel was filled with the NPs solution using a syringe and the subsequent analysis of the collector surface was done with DFM. Numerical solver. The Solver of Microsoft Excel © was used for finding the numerical solution of a system of non-linear equations using the GRG nonlinear solving method. The equations and the boundaries chosen are described in the results and discussion section, since they are part of the developed method.

chemsum

{"title": "Direct quantification of nanoparticle surface hydrophobicity", "journal": "Nature Communications Chemistry"}

one-pot_fabrication_of_ag_@ag2o_core–shell_nanostructures_for_biosafe_antimicrobial_and_antibiofilm_

## Abstract: Microbial contamination is one of the major dreadful problems that raises hospitalization, morbidity and mortality rates globally, which subsequently obstructs socio-economic progress. The continuous misuse and overutilization of antibiotics participate mainly in the emergence of microbial resistance. To circumvent such a multidrug-resistance phenomenon, well-defined nanocomposite structures have recently been employed. In the current study, a facile, novel and cost-effective approach was applied to synthesize Ag@Ag 2 O core-shell nanocomposites (NCs) via chemical method. Several techniques were used to determine the structural, morphological, and optical characteristics of the as-prepared NCs. XRD, Raman, FTIR, XPS and SAED analysis revealed a crystalline hybrid structure of Ag core and Ag 2 O shell. Besides, SEM and HRTEM micrographs depicted spherical nanoparticles with size range of 19-60 nm. Additionally, zeta potential and fluorescence spectra illustrated aggregated nature of Ag@Ag 2 O NCs by − 5.34 mV with fluorescence emission peak at 498 nm. Ag@Ag 2 O NCs exhibited higher antimicrobial, antibiofilm, and algicidal activity in dose-dependent behavior. Interestingly, a remarkable mycocidal potency by 50 μg of Ag@Ag 2 O NCs against Candida albican; implying promising activity against COVID-19 white fungal post-infections. Through assessing cytotoxicity, Ag@Ag 2 O NCs exhibited higher safety against Vero cells than bulk silver nitrate by more than 100-fold.Earth's biosphere is occupied by a plethora of microorganisms which encompass several categories, including bacteria, archaea, yeast, molds, algae, viruses and protozoa. Several benefits are provided by them in maintaining a balanced ecosystem, such as oxygen generation, nutrient supplementation, organic material decomposing and bioactive compounds production. Nonetheless, their pathogenicity represents a serious problem for public health and the entire ecosystem. The microbes causing infectious diseases are ubiquitous through several routes such as food manufacturing machines, water purification systems and polluted medical devices 1 . As a result, various antimicrobial agents, particularly antibiotics, were developed to combat the spread of pathogen-causing infections. However, the intense and widespread abuse of such biocides led to emergence of multi-drug resistant microbes (MDR). Recently, nanotechnology with its related products opens different avenues to face and solve the MDR threat. The metal and metal oxide nanoparticles either sole or in nanocomposite structures increased the antimicrobial activity by the virtue of expanding spectrum of enhanced features 2 .Silver nanoparticles (Ag Nps), are among the most attractive nanomaterials have been widely used in a range of biomedical applications, including diagnosis treatment, drug delivery, medical device coating, and for personal health care 3 . For years, knowledge about silver's ability to kill harmful bacteria has made its nanoparticles popular for creating various products. Silver has many advantages, including the fact that it is non-toxic to humans at very low doses and the use of silver was a common expedient for cooking procedures and for preserving water from contamination. Previous studies have presented improved bactericidal activity for lower nanoparticle sizes associated with higher surface area of the nanomaterials 4 . Silver ions are known to specifically react with the metabolic enzymes inside the bacteria; causing growth suppression. An oxide shell has also been demonstrated to boost biocidal action 5 . The preparation process of good quality nanoparticles (NPs) is vital to ensure their multidirectional effectiveness 6 . Several chemical and physical means were employed to synthesize NPs, the physical one involving laser ablation of a solid target in water 7 , condensation or evaporation and the thermal treatment of Ag NPs in an organic solvent at temperatures up to 360 °C in the gas atmosphere 8 . In the laser ablation procedure, the lack of any chemical reagents provides a unique benefit, but it is an expensive method. Chemical methods are alternatively applied, in which metal nanoparticles sizes are reduced leading to the formation of minute metal clusters. The synthesis of the nanoparticles in solution has important advantages as the ease with which the design, shape and size of the nanoparticles can be precisely controlled 9 . In the present study, Ag@Ag 2 O core-shell nanocomposites (NCs) were synthesized via a simple chemical method. The as-prepared NCs were characterized structurally, morphologically and optically. Thereafter, the antimicrobial and antibiofilm efficiency of Ag@Ag 2 O nanocomposite against planktonic and biofilm-forming pathogens were evaluated. Additionally, the biocompatibility of Ag@Ag 2 O nanocomposites was assessed. ## Materials and methods Methodology. The formation process of nano-sized silver composites (Ag@Ag 2 O) powder is a simple and safe system using alkali chemical techniques. 0.1 N of silver nitrate (AgNO 3 , 99.97%, Sigma Aldrich) aqueous solution which is used as a precursor of the silver element. It's added drop-wise to an alkali mixture solution, which contains [2.5 Wt% of Potassium Hydroxide (KOH, 99.97%, Sigma Aldrich), 10 Vol% of n-propanol (NPA) and deionized water, with stirring]. The reaction temperature is kept constant at 70-80 °C for 2 h. The solution was constantly stirred with a magnetic stir bar, until the solution turned into a grey colloidal suspension; indicating the fulfillment of the chemical reaction. Then, the precipitate powder is filtered, and dried at 60 °C overnight, as shown in Fig. 1a. Characterization methods. The structural, composition and morphological properties of the (Ag@Ag 2 O) NCs composite powder were investigated using X-ray diffraction (XRD), Raman spectroscopy, Fourier-transform infrared spectroscopy (FTIR), scanning electron microscopy (SEM), high resolution transmission electron microscopy (HRTEM), zeta potential and Fluorescence spectra. X-ray diffraction measurement was performed using Shimadzu 7000 XRD, with CuKα radiation (λ = 1.54 ) generated at 30 mA and 30 kV with a scanning rate of 4° min −1 and 2θ values ranged between 25° and 80°. Raman spectrum was obtained at an excitation wavelength of 532 nm using Raman spectroscopy (Senterra, Germany). For the determination of the chemical bonds formed during the preparation process, Fourier Transform Infrared Spectrophotometer (FTIR, Bruker Corporation, Ettlingen, Germany) is used. The powder product morphology was investigated using Scanning Electron Microscopy [SEM, JEOL (JSM 5300)]. However, high resolution transmission electron microscope TEM (HR-TEM, JEOL-2100, Japan) was employed to examine morphology, high resolution d-spacing of the different structures, electron diffraction and mapping of silver and oxygen elements. X-Ray photospectroscopy (XPS) measurement was carried out using PHI 5000 Versa Probe III Scanning XPS Microprobe with Monochromatic Al source ranged from 0-1486.6 eV. Electrostatic potential was determined by the DLS technique using Zetasizer Nano ZS (Malvern Instruments, Worcestershire, UK) and the data were analyzed by Zetasizer software 6. Finally, the fluorescence spectrum was recorded at an excitation wavelength of 250 nm on fluorescence spectrophotometer (Agilent, G9800A, USA). Evaluation of the as-prepared Ag@Ag 2 ONCs as anti-bio-film agent. The inhibitory effect of Ag@ Ag 2 ONCs and AgNO 3 (5, 10, 25, 50 and100 µg/mL) against P. aeruginosa and S. aureus biofilms were assessed using tissue culture plate method. A sterile polystyrene 96-well microplate was seeded by 100 μL of tryptone soy broth (TSB) containing 10 8 CFU/mL of each tested strain. Simultaneously, two controls were run in parallel; positive control wells (medium containing a bacterial culture) and negative control wells (sterile TSB only). After 24 h of static incubation at 37 °C, washing, fixation and staining of the remained biofilm were carried out by 95% ethanol and 0.25% crystal violet, respectively. The absorbencies of adhered cells were measured spectrophotometrically at 595 nm. All the experiments were carried out in triplicate and the results are expressed as mean ± SD 11 . The following equation was employed to calculate inhibition percentage of biofilm formation where A represents the absorbance of the positive control wells and A 0 reveals the absorbance of the treated wells containing an antimicrobial agent. Biofilm disintegrating assay. The potential of Ag@Ag 2 ONCs to degrade the already formed biofilms by P. aeruginosa and S. aureus were examined in comparison to AgNO 3 . Firstly, the bacterial lawn (10 8 CFU/mL) was inoculated into 96-well microplates and incubated statically at 37 °C for 24 h to permit biofilm formation. Secondly, the well contents were discarded aseptically. The diluted Ag@Ag 2 ONCs and AgNO 3 to concentrations (5, 10, 25, 50 and100 µg/mL) were added to each well. The incubation, processing, quantification and disintegration percentage of the biofilms were performed as previously described. All the experiments were carried out in triplicate and the results are expressed as mean ± SD. As stated by Cremonini et al. 12 the biofilm was deemed strong, medium and low at optical density (OD) ˃ 2, 1 ˂ OD ˂ 2 and 0.5 ˂ OD ˂1, respectively. Antagonistic effect of Ag@/Ag 2 O NCs on the algal growth. The inhibitory effect of Ag@Ag 2 O NCs was evaluated against Chlorella vulgaris by adding (5, 10, 25, 50 and100 µg/mL) in parallel to exact concentrations of AgNO 3 . The algae were propagated in sterilized Bold's basal media (BBM) medium; incubated at 25 °C under illumination with daily cycles of 12-h light and 12-h night for 7 days 13 . The cell count was assessed with a hemocytometer under a light microscope (Olympus BH-2, Japan). The inhibition percentage was calculated as mentioned in Eq. (1), and the results are expressed as means ± SD. Investigation of the cytotoxicity of Ag@Ag 2 ONCs comparing with silver nitrate against normal cells. Normal mammalian kidney epithelial cells (Vero) were used to detect cytotoxicity of the studied compounds. Vero cell line was cultured in DMEM medium-contained 10% fetal bovine serum (FBS), seeded as 4 × 10 3 cells per well in 96-well cell culture plate and incubated at 37ºC in 5% CO 2 incubator. After 24 h for cell attachment, serial concentrations of Ag@Ag 2 O NCsand silver nitrate were incubated with Vero cells for 72 h. Cell viability was assayed by MTT method 14 . Twenty microliters of 5 mg/mL MTT (Sigma, USA) was added to each well and the plate was incubated at 37 °C for 3 h. After removing the MTT solution, 100 µL DMSO was added and the absorbance of each well was measured at 570 nm using a microplate reader (BMG LabTech, Germany). The effective safe concentration (EC 100 ) value (at 100% cell viability) of the tested compounds was estimated by the Graphpad Instat software. ## Results and discussions Structural analysis and chemical bonds formation. X-ray diffraction (XRD). The X-ray diffraction (XRD) spectrum of nano-composite (Ag@Ag 2 O)NCsis given in Fig. 2a 16 .On the other hand, the close overlap between Ag and Ag 2 O diffraction peaks and the difficulty to distinguish between the Ag + and Ag 0 peaks at the diffraction angle of 38.1° inferred a formation of a hybrid structure 17,18 . Sajjad Ullah et al. 18 ) in the samples. The structure could possibly have an Ag 2 O shell with Ag as the core with a decreasing gradient of oxygen from the surface to the core 6 . Despite the simplicity of preparation method, the silver element needs a special medium during its preparation. The individual crystallite size (t) was calculated using Scherrer's formula 19 given by Eq. ( 2). where k is the Scherrer's constant (0.89-0.9), λ is the X-ray wavelength, β is the full width at half maximum (FWHM) and θ is the Bragg angle 19 . According to Eq. ( 2), the sample crystallite size for the plan (111) is calculated and is found to be approximately 18.6 nm. Raman analysis. The molecular structure and phase identification of the Ag@Ag 2 Ocore shell are explored using Raman spectra. Figure 2b shows the Raman spectrum for the prepared nanopowder in a range from 0 to 4500 cm −1 . Two major peaks are clearly detected; the first one at 74.1 cm −1 with an intensity of 7900 cps for Ag 0 (Ag lattice vibrational mode) and the other one at 1046.7 cm −1 with an intensity of 583 cps for Ag 2 O (Ag-O stretching/bending modes) 20 . Fourier-transform infrared spectroscopy (FTIR):. FTIR analysis reveals the functional groups of the Ag@/Ag 2 O nanocomposite synthesized using alkali chemical treatment (Fig. 2c).The broad band at3400 cm -1 indicates the O-H stretching vibrations of the hydroxyl groups 21 corresponding to H-bonded alcohols and also to intramolecular H bonds which are most probably from water molecules 22 . The peaks at 2357 cm −1 and 1655 cm −1 prove the existence of O-H carboxylic acids 23 and OH bending, respectively 24 . The band at 1387 cm −1 assigned to O-H bend of carboxylate 25 . The absorption band on 675 cm −1 is due to Ag-O stretching mode, which corresponds to Ag-O vibration in Ag 2 O.Furthermore, the appearance of follower peak at 868.68 cm −1 corresponds to the metaloxygen vibrations for the formation of (Ag@Ag 2 O) NCs 15 ; thus, synchronizing with the aforementioned XRD results; confirming the formation of (Ag@Ag 2 O) NCs. (2) t = k. /β. cos θ www.nature.com/scientificreports/ X-ray photoelectron spectroscopy (XPS). The XPS data for the chemically prepared Ag@Ag 2 O NCs is illustrated in Fig. 3. Figure 3a shows the general survey analysis of the nanopowder, which exhibits a major detected peak of the Ag3d at a binding energy of 368.34 eV with an atomic ratio of 48.5%. Also, O1s peak is detected at a binding energy of 530.81 eV with an atomic ratio of 29.75% and K2p peak is observed at a binding energy of 293.5 eV with an atomic ratio of 2.18%. Finally, C1s peak is measured at a binding energy of 285.21 eV with an atomic ratio of 19.56%. In Fig. 3b, the high resolution of the Ag3d spectrum displays two main strong bands. Such two bands can be further de-convoluted into two pairs of sub-peaks. The peaks at 367.98 eV with an atomic ratio of 44.01% and 373.96 eV with an atomic ratio of 28.75% are respectively assigned to Ag 0 (3d5/2 and 3d3/2). The other set of bands is detected at 367.38 eV with an atomic ratio of 10.48% and 373.6 eV with an atomic ratio of 10.36% are attributed to Ag + (3d5/2 and 3d3/2, respectively) in the nanocomposite. Figure 3c confirms the oxidation of the silver nanoparticles through the existence of the O1s spectrum at 529.39 eV with an atomic ratio of 47.64%, at 530.77 eV with an atomic ratio of 25.32% and at 531.4 eV with an atomic ratio of 23.19%. Finally, the results confirm that there are two different configurations of silver species, namely Ag 2 O and Ag, which is consistent with many published reports . The detected elemental carbon in the main survey analysis may have originated from the ambient atmosphere itself. The adsorption of hydrocarbons from the surrounding atmosphere, which results in the creation of a thin carbon layer on surfaces, is most likely the source of the carbon contamination 29 . result is nearly consistent with the result from XRD patterns in Fig. 2a. The silver oxide may have formed in as a solution mixture containing potassium hydroxide and n-propanol which have high oxidation potential 30 . Also, the time spent since the beginning of the reaction, i.e. when adding silver nitrate to the oxidized mixture and until the end of the reaction is not enough to produce the silver oxide in its final form. Thus, it is an incomplete reaction that results in the precipitation of the silver nanopowders. This step entails the formation of a layer of silver oxide on the surface of the silver powder nanoparticles as a result of remaining in the oxidizing solution for a longer time 31 . Thus, it is logical to form an Ag/Ag 2 O core shell compound of a spherical nature as a result of the lattice mismatch between silver metal and silver oxide 6 . However, the aggregation is more likely to occur due to too small size as shown in Fig. 4a and b. Generally, the smaller particle size is usually more beneficial for www.nature.com/scientificreports/ the antibacterial activity. Because the particle size is smaller, many more particles will be easily adsorbed on the surface of the bacterial cell membrane, and then successfully attack the cell, ultimately destroying the physiological functional groups of the cell 32 . ## Morphological analysis. Scanning electron microscope (SEM) Transmission electron microscopy (TEM). TEM has been employed to characterize the size, shape, morphology and crystallinity of the synthesized Ag@Ag 2 O NCs. Zeta potential. The surface charge of Ag@Ag 2 O core shell was determined from Zeta potential by applying voltage across a pair of electrodes at either end of a cell containing the particle dispersed. The charged particles are attracted to the oppositely charged electrode and assessing the Zeta-potential value by − 5.34 mV (Fig. 5a). The Ag@Ag 2 O NCs show slightly low surface charges which tend to form agglomerates 33 . Moreover, the low surface charges of Ag@Ag 2 O NCs reflect the urgent requirement of a capping agent to prevent such agglomeration and keep nanocomposites stable for a long time 34 . However, upon antimicrobial application and cytotoxicity evaluation, the examined NCs were freshly prepared and examined after a short time of preparation (within 48 h of preparation). Subsequently, the prepared NCs, within such time, didn't exhibit aggregation and were still stable. Additionally, several reports 35,36 synthesized AgNPs and other metal-NPs in the same range of zeta and also exhibited antimicrobial activity. Fluorescence spectra. The fluorescence emission peak of Ag@Ag 2 O NCs was detected using an excitation wavelength of 250 nm and appeared at about 498 nm in the visible range as shown in Fig. 5b. This fluorescence emission peak may be attributed to the relaxation of the electronic motion of surface plasmons 37 . The sharpening behavior in the peak may be due to the core shell structure and coverage of Ag by Ag 2 O, which prevents the nanopowder from combining with any water molecules as well as continuing the oxidation process 38 . ## The chemical mechanism. Based on the preceding experimental data, it is worth mentioning to explain the chemical mechanism of the nanocomposite (Ag@Ag 2 O) formation as demonstrated in Eq. ( 3). The reaction of silver nitrate with potassium hydroxide produces silver hydroxide via the following mechanism 24 : (3) 4)- (7). Briefly, a part of AgOH may be reacting with the n-propanol, which acts as a wetting agent that decreases the recombination rate and the generation of silver propanoate (Ag-O 2 CCH 2 CH 3 ), as shown in Eq. ( 4), which is inferred from FTIR spectra as a sharp peak at 1655 cm −1 and 3400 cm −1 as shown in Fig. 2c 39 . Meanwhile, Ag-O 2 CCH 2 CH 3 is reacted with the hydroxyl group of KOH producing silver ions (Ag + ) in a continuous oxidation process [Eq. ( 5)]. The silver ion reacts with water and n-propanol in an alkaline medium via the presence of OHgroup to produce silver element (core); as shown in [Eq. ( 6a)]. Additionally, some of the silver ions re-interact with water and n-propanol for producing silver hydroxide as in [Eq. (6b)]. Therefore, the unstable silver hydroxide product (AgOH) is reduced to silver oxide (Ag 2 O shell) as shown in [Eq. ( 7)]. ## OR Antimicrobial efficiency of Ag@/Ag 2 O NCs against planktonic pathogens. Considering the health problems associated with microbial contamination, it is vital to find out effective antimicrobial agents that are able to control their outbreak. Thus, the current study is concerned with the antimicrobial activity of Ag@ Ag 2 O NCs against some prokaryotic and eukaryotic pathogens. The sensitivity of the examined pathogens to different concentrations of Ag@Ag 2 O NCs is shown through agar diffusion assay. Figure 6a and b illustrates the comparative results of antimicrobial activities of the Ag@Ag 2 O NCs and their precursor. Also, it demonstrated a dose-dependent manner in which the antimicrobial activity of different concentrations of Ag@Ag 2 O NCs against E. coli, B. cereus and C. albicans as representative models of pathogenic Gram-negative bacteria, Grampositive bacteria and Fungi, as well as C. vulgaris control before treatment with Ag@Ag 2 O NCs and C.vulgaris after treatment with 50 μg/mL of Ag@Ag 2 O NCs are shown in Fig. 7A-E respectively. Generally, Ag NPs displayed considerable effectiveness indicated by halo zones which exceeded 1 mm, where any antimicrobial agent was evaluated as "good" atan inhibition zone greater than 1 mm 40 . For all the examined pathogens, inhibition halos were directly proportional to the concentration of AgNPs. In addition, Gram-positive strains seemed to be more resistant than Gram-negative strains. That could be attributed to the lipophilicity of Ag NPs according to different cell wall polarity and compositional variations 41 . As revealed by Pazos-Ortiz et al. 42 the thickness of the cell wall increases the resistance of bacteria to the exposed NPs. The thick peptidoglycan layer of the Gram-positive bacteria's wall, which is composed of teicoic acids and lipoteicoic acids, restricts the diffusion of NPs. Moreover, the tolerance response of each microbe depends on its metabolic properties. However, the cell wall of the Gram-negative bacteria is composed of thinner peptidoglycan layer together with lipoprotein and lipopolysaccharide, which together represent 25% of its mass. It is noteworthy to mention that the nosocomial infections and enteric fever are associated with P. aeruginosa, E. coli and S. typhi, respectively. Therefore, their inhibition is a pivotal issue. In agreement with our results 42,43 reported low reduction in S. aureus count (CFU/mL) and also halo zone in comparison to Gram-negative bacteria upon treatment by Ag@Ag 2 O NPCs. Besides, Ag@Ag 2 O NPCs biosynthesized by aqueous leaf extract of Eupatorium odoratum (EO) exhibited antagonistic performance coincident with the obtained results of current study 41 . In the www.nature.com/scientificreports/ same sense, D'Lima et al. 6 reported that Ag/Ag 2 O hybrid nanoparticles showed a considerable zone of inhibition against P. aeruginosa; declaring the enhancement of antibacterial activity upon combination with carbenicillin. In contrast, other studies reported higher susceptibility of Gram-positive bacteria for NPs treatment than Gram-negative one 11,44 . Remarkably, a considerable halo of mycostasis was noticed against C. albicans. Despite the oligodynamic nature of silver ions, which is due to their higher activity at minute concentrations, a potent antifungal efficiency of 50 μg of Ag@Ag 2 ONCs exhibited upon comparing with its precursor (Fig. 6); implying effectiveness in the treatment of COVID-19 post infections. Such fungal infections appeared recently in the second wave in India, in particular in patients who were put on mechanical ventilation in intensive-care units. The fungicidal property of Ag@Ag 2 ONCs could be assigned to the damage of the glycoprotein-glucan-chitin cross-linkage of fungi cell wall followed by sever alterations in cellular biochemistry 11,45 . In addition, it has been suggested that Ag nanoparticles interact with the proteins of the plasma membrane, which is responsible for keeping trans-membrane electrochemical potential gradient such as H + ATPase protein. Such interaction leads to alterations of normal protein conformations and malfunctioning by blocking the regulation of H + transport across the membrane, which ultimately hindering growth, restraining respiration and ending with death . In coincidence with our results, Mallmann et al. 49 highlighted similar results with inhibitory influence of Ag@ Ag 2 O NCs against several species of Candida. Otherwise, Elemike et al. 41 demonstrated the dominant biocidal effectiveness ofAg@Ag 2 O NCs in bacteria than fungi. Evaluation of the as-prepared Ag@Ag 2 ONCs against biofilm formation, biofilm disintegration and algal growth. Biofilms are multicellular sessile microbial communities embedded in a self-produced extracellular polymeric matrix (EPS) (e.g. DNA, proteins and polysaccharides) and attached toa living or inert substratum or interface. Actually, the viscoelastic nature of the EPS represents a serious concern, especially in water pipes, water purification systems and also in medical devices. Where, the biofilms have the capability to withstand different stress factors by the virtue of such feature. Hence, nanotechnology invasion has provided a significant tool to eradicate such problem at both environmental and medical levels 50 . The inhibitory effect of different concentrations of as-synthesized Ag@Ag 2 ONCs and their precursor salt on biofilm formation/ disintegration of both Gram-positive and Gram-negative bacteria was illustrated in Table 1. As noticed, P. aeruginosa biofilm was less susceptible for both treatments and under formation/ disintegration conditions, in comparison to S. aureus biofilm. As revealed by Hoseini -Alfatemi et al. 51 , P. aeruginosa and S. aureus biofilms were inhibited by 10 and 1 mg/mL of AgNPs, respectively; which makes our study characteristic. Where, 100 µg/mL suppressed (98.7% and 87.5%) and (93.1 and 74.8%) of S. aureus and P. aeruginosa biofilm synthesis and disintegration, respectively. Interestingly, Gram-negative biofilms were comparatively more resistant to antibiofilm treatments than Gram-positive as reported in several studies 42,51,52 . Generally, Ag@Ag 2 O NCs exhibited antibiofilm activity via several routes including, destruction of initial planktonic phase, damage of aggregated/sessile phase, disruption of EPS matrix, increasing of hydrophobicity of EPS and inhibition of quorum sensing system 53 . What is more, the inhibitory effect of Ag@Ag 2 ONCs against algal growth of C. vulgaris was studied. C. vulgaris is involved among other algal genera which are responsible for various environmental issues such as eutrophication and biofouling, especially in the availability of high concentrations of contaminants and in association with direct sunlight 53 . As illustrated in Table 1, Ag@Ag 2 O NCs exhibited a drastic algicidal effect on the proliferation and viability of algae with 98.4% growth inhibition. Severe damage of chloroplasts could be proposed due to yellowish to pale green color of algal growth in the presence of Ag@Ag 2 ONCs. Meanwhile, the control culture (without Ag@Ag 2 ONCs) appeared green and flourished during 7 daysof incubation as shown in Fig. 7D and E. Disintegration of algal cell organelles, thylakoid disorder and plasmolysis are common features associated with the destructive effect of Ag@Ag 2 ONCs on algal cell as stated by Duong et al. 54 . Therefore, the employment of Ag@ Ag 2 ONCs in restriction the algal blooms could result in constraining of their environmentally adverse influence. As general observations, Ag@Ag 2 ONCs exhibited greater inhibitory activity than its precursor against all examined microbial forms. That could be assigned to the small size of nanoparticles and in relation to surface area. As pointed out by 55 , the antagonistic activity of NPs derived from their penetration ability which depends on www.nature.com/scientificreports/ sizes that are less than 100 nm. In addition, the biocide activity of Ag@Ag 2 ONCs uplifted linearly with increasing in Ag@Ag 2 O NCs concentration, which implies dose-dependent manner. However, NPs type, concentration, size, aggregation state, surface charge, synthesis conditions and tested microbe consider being governing parameters influencing of the effective doses 51 . Broadly, several strategies could be ascribed for NPs to display their toxicity against different microbial forms. The first strategy begins from puncturing and perforating the first protective barrier of the cell, which is cell wall, by interacting with its anionic components such as neuraminic acid, N-acetylmuramic acid, and sialic acid. However, as long as the NPs are smaller than 80 nm, their passage to cell membrane and later inside the cell is facile; causing phospholipid peroxidation, polysaccharides depolymerization and subsequently membrane detachment and integrity destruction 10,56 . At this stage, cell permeability increases followed by intracellular components leakage and proton motive force dissipation. Once NPs occupies intracellularly, more destructive features were exerted concerning metabolism and biochemical activities 10 . AgNPs showed higher affinity for binding with thiol group of amino acids; forming extra -S-S-bonds. By such way, deformation of protein configuration occurs, leading to proteins denaturation and ribosomes inactivation 56,57 . Further, NPs bind with nucleic acids such genomic and plasmid DNA; causing blockage of DNA replication and repair processes. With continuous release of Ag + ions and their oxide from Ag@Ag 2 O NCs, set of reactions (e.g., Fenton and Haber-Weiss reactions) are continuously and intensively generating Reactive Oxygen Species (ROS) such as hydroxyl radicals (OH − ), superoxide radicals (O 2 − ) and singlet oxygen ( 1 O 2 ). Under such oxidative stress, massive damage to the cell takes place and eventually lead to cell death. Tee et al. 58 and Pazos-Ortiz et al. 42 referred to the complexity of the mechanisms by which NPs exhibit their antagonistic influence. Figure 1b represents schematic illustration on the destructive effect of Ag@Ag 2 O NCs against different microbial forms. ## Cytotoxicity assessment. After 72 h of incubation of the Ag/Ag 2 O NPs and silver nitrate precursor with normal renal epithelial Vero cells, it was found that their estimated safe doses on cell viability were 13.43 ± 1.63 µg/mL and 0.075 ± 0.001 µg/mL, respectively. This indicates that Ag/Ag 2 O NPs ismore safe than silver nitrate source. However, at 100 µg/mL of Ag/Ag 2 O NPs or silver nitrate caused death in Vero cells by 79.69% and 91.09%, respectively, as it is shown in Fig. 8a. Moreover, severe collapse in the normal spindle shape of silver nitrate-treated cells, at 25 µg/mL, confirmed its cytotoxicity in comparison to the normal morphology of Ag/Ag 2 O NPs-treated cells and untreated control healthy cells (Fig. 8b) 14 . The lower cytotoxicity of the prepared NPs, at < 13 µg/mL, may be related to their particle size (≥ 40 nm), negatively particle charge and high agglomeration potential (Fig. 4a,b) which results in increasing their size thus decreasing their cellular uptake and diminishes ROS generation 59 . In support of this issue, Liu et al. 60 found that Ag NPs with size of 55 nm generated less ROS than 15 nm Ag NPs. Moreover, silver NPs' tendency to agglomeration increases in culture medium 61 . Besides, based on the previous finding, corona formation which is mediated by adsorption of fetal bovine serum (FBS), from culture medium, on silver NPs, mainly limits their cytotoxicity via reducing their cellular uptake 59,62 . All these factors contribute to minimize the cytotoxicity effect of Ag/Ag 2 O NCs on normal cells. This higher safety of Ag/Ag 2 O NPs on human normal cells (Fig. 1c) lends credibility to their biomedical applications compared to bulk silver nitrate. ## Conclusion Due to the globally identified antibiotic resistance among clinical pathogens, novel antimicrobial materials are needed to circumvent drug resistance. In this study, we have demonstrated a facile chemical method to fabricate Ag@Ag 2 O core-shell nanocomposites and their antibacterial, antifungal and antibiofilm activities against a wide range of microbial pathogens were examined. Structural, morphological and optical properties were studied using different techniques. XRD, Raman, FTIR, XPS and SAED indicated the formation of a hybrid NC structure with a crystalline nature. SEM and HRTEM showed the evidence of Ag@Ag 2 O with a spherical core-shell structure and its particle size ranging from 19 to 60 nm. Furthermore, the antagonistic properties of Ag@Ag 2 O core-shell and its precursor AgNO 3 were compared in the range of 5-100 μg/mL. The image data declared the sensitivity order of pathogens versus examined Ag@ Ag 2 O as follows: S. typhi ˃ P. aeruginosa ˃ E. coli ˃ B. cereus ˃ S. aureus. Besides, a noticeable antifungal potency of Ag@Ag 2 O was observed at 50 μg/mL. Additionally, its antibiofilm activity and disintegration capability were increased with elevation of concentration. Generally, a dose-dependent behavior could describe the inhibition of examined pathogens by Ag@Ag 2 O. Eventually, the cytotoxicity of the NC was analyzed by Vero cells and its effective safe concentration value was estimated to be about 36.31 ± 1.53 µg/ml. The promising structural features and biocidal activity of Ag@Ag 2 O opens up employment in various technological sectors.

chemsum

{"title": "One-pot fabrication of Ag @Ag2O core\u2013shell nanostructures for biosafe antimicrobial and antibiofilm applications", "journal": "Scientific Reports - Nature"}

concentration-dependent_<i>rhombitrihexagonal_tiling</i>_patterns_at_the_liquid/solid_interface

## Abstract: We report STM investigations on a linear oligophenyleneethylene (OPE)-based self-assembling Pd(II) complex 1 that forms highly-ordered concentration dependent patterns on HOPG. At high concentration, 2D lamellar structures are observed whereas the dilution of the system below a critical concentration leads to the formation of visually attractive rhombitrihexagonal Archimedean tiling arrangements featuring three different kinds of polygons: triangles, hexagons and rhombi. The key participation of the Cl ligands attached to the Pd(II) centre in multiple C-H/Cl interactions was demonstrated by comparing the patterns of 1 with those of an analogous non-metallic system 2. From bee honeycombs to ancient Roman mosaics and Moorish wall tilings, periodic polygonal patterns are ubiquitous in arts and science both for ornamental and technological purposes. 1 Tessellations of surfaces, i.e., the tiling of planes using one or various types of regular polygons, were frst classifed by Kepler in 1619. 2 Regular tilings feature a single type of polygon (triangle, square or hexagon) repeated infnitely whereas less frequent semiregular Archimedean tilings (AT) combine at least two different polygons placed edge-to-edge around a vertex. Scanning tunnelling microscopy (STM) provides a perfect tool to observe such patterns at the molecular level 3 and furthermore constitutes a link to nanodevice technology. 4 For instance, and besides various 1D patterns and/or 2D networks based on a wide variety of self-assembled systems, 5,6 AT arrangements have been visualized by STM. These systems, however, have been limited to trihexagonal (Kagomé) tilings, 6 in which two hexagons and two triangles are alternated on each vertex. Very recently, lanthanidebased polyphenyl systems have been demonstrated to form exciting snub square tilings combining two different (trigonal and square) polygons on Ag(111) through STM. 7 In this article, we move a step further towards complex tessellations at the liquid/solid interface by realizing a special type of rhombitrihexagonal AT patterns that feature up to three different polygons (triangle, rhombus and hexagon). Our system not only represents one of the most complex patterns ever visualized by STM but it can also be transformed into lamellar structures above a critical concentration. 8 To achieve this goal, we have taken advantage of recent fndings from our group on a self-aggregating oligophenyleneethylene (OPE) 9 -based Pd(II) derivative 1 (Fig. 1). 10 Complex 1 exists in a monomeric state in nonpolar solvents below z1 10 4 M. Above this critical concentration, micrometre-sized elongated supramolecular structures are formed, in which the OPE-based ligands of 1 adopt a more rigid conformation around the Pd(II) ion to maximize p-p and Pd/Pd interactions with neighbouring units in the stack. Encouraged by these results, we questioned whether the distinct conformation of the units of 1 in solution depending on the concentration would also be reflected in a different packing mode at the liquid/solid interface, as observed for other metallosupramolecular p-stacks. 11 To that end, a drop of a concentrated solution of 1 (c ¼ 5 10 4 M) in 1-phenyloctane was placed onto HOPG and investigated by STM. The images show the formation of highly ordered lamellar structures consisting of alternated bright and dark fringes (Fig. 1 and S1 †). In contrast, the pyridine-based ligand (precursor of 1) alone does not form any organized structures on HOPG even at millimolar concentration, thus revealing the importance of an extended aromatic surface and the presence of a Cl-Pd(II)-Cl fragment to produce organized ad-layers. Within the lamellar structures, the spots with higher local density of states can be assigned to individual aromatic rings belonging to the OPE segments of 1 whereas the darker striations attached to the edges of the aromatic system correspond to parallel-aligned dodecyl chains (Fig. 1). The unit cell of this pattern is highlighted in yellow in Fig. 1 and the corresponding parameters are: a ¼ 0.76 AE 0.02 nm, b ¼ 6.81 AE 0.06 nm, and a ¼ 75.0 AE 3.0 . On the basis of these dimensions, only four out of six dodecyl chains from each molecule (two outer and two inner) are adsorbed on the substrate whereas the remaining two chains are most likely embedded in the supernatant. 12 The density of the lamellar pattern was calculated to be 0.20 molecules per m 2 (plane group p1). 13 Remarkably, the parallel orientation of the OPE units at the HOPG/1-phenyloctane interface closely resembles that observed in the associates in nonpolar solutions, although one has to note that in the former assemblies the aromatic rings lie on the HOPG surface whereas in solution these are stacked on top of each other. These observations infer that monolayer formation is largely driven by adsorbate-substrate (epitaxial) and adsorbate-solvent (solubility) interactions. 14 Similarly to the solution behaviour, we questioned whether dilution of the system below a given concentration (1 10 4 M) would lead to a distinct molecular arrangement at the liquid/ solid interface. Fig. 2 shows the STM images of 1 obtained from a 100-fold more diluted solution (c ¼ 1 10 6 M) of 1 at the HOPG/1-phenyloctane interface. Interestingly, no sign of lamellar structures was observed at this concentration. However and to our surprise, a highly-ordered periodic pattern comprising three types of polygons (hexagons, rhombi and triangles) can be visualized (Fig. 2, S3 and S4 †). 15 These results bring to light that the structural phase transition in solution and at the solid/liquid interface occurs in a similar concentration range (below 1 10 4 M). On closer scrutiny, we noticed that the geometric shapes are separated from one another by bright segments that correspond to the aromatic rings of the molecules of 1 (Fig. 2a and b). Within this arrangement, each hexagon is sharing its edges with 6 rhombi and its vertices with 6 triangles leaving no gaps and overlaps, as shown in Fig. 2b and c. This exotic surface tessellation (plane group p6) resembles one of the AT of the Euclidean plane, the 3.4.6.4. rhombitrihexagonal tiling. 2 The unique difference from the regular rhombitrihexagonal tiling is the presence of rhombi instead of squares. The density of the AT patterns corresponds to 0.16 molecules per m 2 . By careful analysis of the STM images, we found out that the edges of all polygons are nearly equivalent in length (1.7 AE 0.2 nm) (Fig. 2b). This distance matches that of the aromatic backbone of a pyridine-substituted OPE ligand obtained by theoretical calculations (Fig. S10a †). Indeed, some individual aromatic rings can be distinguished in the magnifcation shown in Fig. 2b. According to our STM investigations, the Pd(II) centres are located at the vertices of the polygons, as all edges are occupied by the aromatic segments (see model in Fig. 2c). The repeat unit is represented by an equilateral triangular motif consisting of three molecules, whose edges are successively oriented towards the Cl-Pd(II)-Cl centre of a neighbouring unit within the triangle, as shown in Fig. 2c. Six such triangular subunits further pack into a hexagonal motif, thereby delineating an inner hexagonal cavity surrounded by six triangles and six rhombi in an alternated fashion (Fig. 2b and c). The dimensions of the unit cell are a ¼ 6.5 AE 0.1 nm, b ¼ 6.5 AE 0.1 nm, and a ¼ 60 AE 3 whereas the distance between the Pd centres within each triangular motif extracted from STM measurements was found to be 2.5 AE 0.2 . According to this dimension, the edge of one molecule and the Cl-Pd(II)-Cl fragment of a neighbouring unit are distant enough to enable the interaction between the central OCH 2 group of one molecule and the Cl ligand of the other one by C-H/Cl interactions (see proposed model in Fig. 2c and S5, S6 †). As shown by us 16 and others, 17 metal-bound chlorine atoms have a strong propensity to interact with polarized C-H groups through hydrogen bonding interactions both in the crystalline state 18 and in solution. 16 In our system, only the methylene groups attached to the electronegative oxygen heteroatoms of the side chains are polarized enough to interact with such hydrogen-bonding Cl acceptors. Thus, on the basis of these considerations, STM analysis and theoretical calculations, two C-H/Cl interactions on either side of every molecule of 1 represent, along with the interaction of aromatic and aliphatic segments with the HOPG lattice, 19 the driving force for AT formation. It is worth noting that due to their lower tunnelling efficiency, a clear visualization of the dodecyl chains has not been possible. 20 We hypothesize that the alkyl chains will be concentrated in all polygonal cavities to maximize their interaction with the HOPG surface, which ultimately facilitates the AT formation. This situation is clearly possible in the hexagonal cavities, in which up to 12 dodecyl chains can be accommodated, two per monomeric unit (Fig. 2c). The cavities of the rhombi are slightly smaller and we postulate that four chains (see Fig. 2c and S5 †) can occupy these areas. Finally, the relatively high electron density observed in the triangular voids suggests that these areas are also considerably flled with alkyl chains. However, due to their smaller size compared to the rhomboidal and hexagonal cavities only partial adsorption of the chains is possible, whereas other parts protrude above the adsorbate into the phenyloctane layer. 21 Our proposed model (See Fig. 2c, S5 and S6 †) clearly shows that a maximum of 6-7 carbon atoms from each dodecyl chain ft in the triangular voids without inducing severe steric effects or distortions in the AT arrangement. As particularly apparent in Fig. 2a and b, the majority of the molecules feature a nearly perfect linear geometry, indicating that the pyridine-based ligands are arranged with a 180 angle around the Pd(II) ion. There are, however, some areas in which some slightly bent molecules can be observed. This is evident in Fig. 2d and S4 † top, in which a small distortion of the ideal 180 angle is observed in few molecules resulting in a ring-like appearance. We also observed that this bending is not periodic but rather randomly distributed over the whole HOPG surface. The phenomenon of molecular curvature of systems exhibiting an extended p-conjugated surface has been previously observed for different classes of molecules. 22 In a particularly relevant example, Beton, Anderson and co-workers have recently reported on a novel 2D supramolecular organization of cyclic porphyrin systems by STM. 23 They describe the encapsulation of one cyclic polymer in a folded state into another unfolded polymer. The folded polymer undergoes bending where the subsequent strain is adequately made up by stacking stability. Accordingly, the bending energy of the complex in the AT patterns was calculated using the following equation: where K is the bending coefficient, l is the length of the molecule and R is radius of curvature in the molecular arrangement. We applied this equation to calculate the bending energy of our 1 taking into account that the bending coefficient corresponds to 0.03 nN nm 2 (for monolayer systems), the molecular length is 3.8 nm and R can be approximated to 4 nm. The relatively small estimated energy (3.56 meV) required for the bending around the metal ion is well compensated by the high stability of the multipolygonal tiling that is attributed to C-H/Cl interactions and alkyl chain packing. In order to fnd out to what extent the existence of a Cl-Pd(II)-Cl fragment and thus, the participation of C-H/Cl forces is influencing the AT formation, we have investigated a non-metallic OPE-based analogue 2 through STM. OPE 2 (Fig. 3) 10 is equivalent in size to complex 1 (see geometry-optimized structures in Fig. S11 †). However, the Cl-Pd(II)-Cl fragment has now been replaced by an alkyne functionality. This slight modifcation is expected to prevent C-H/Cl interactions and, consequently, the formation of multipolygonal patterns. Similarly to 1, OPE 2 forms one-dimensional associates in nonpolar solvents above 1 10 4 M, although the propensity of this system to aggregate is considerably reduced compared to 1. 10 In fact, when a 5 10 4 M solution of 2 in phenyloctane was used for STM under equivalent conditions to those of 1, no lamellar patterns were observed, highlighting again that the solution and interface behaviour are comparable. Moreover, dilution of the sample up to 10 6 M did not lead to any changes in the molecular packing on HOPG, as shown in Fig. 3 and S8. † Regardless of the concentration, a highly regular grid-like pattern consisting of bright segments of 4.0 AE 0.2 nm in length is observed (Fig. 3a and b). The good agreement between this length and that extracted from molecular modelling (3.83 nm, Fig. 3c and S11 †) supports that these fragments correspond to the aromatic OPE core of 2. In contrast to 1, the absence of a relatively flexible Pd(II) centre increases the rigidity of the system to the point that bending of the molecules cannot be realized. Similarly to 1, the alkyl chains have not been visualized and are most likely adsorbed onto HOPG occupying the empty spaces between the OPE segments, as depicted in the model shown in Fig. 3c and S9. † The repeat unit comprises four molecules that delimit a cavity with a quadrilateral shape (Fig. 3b), yielding a unit cell whose parameters are a ¼ 5.2 AE 0.2 nm, b ¼ 5.2 AE 0.2 nm, and a ¼ 74.0 AE 3.0 (plane group p4). However and in contrast to 1, no rhombitrihexagonal structures are formed. This is influenced by the absence of chlorine ligands that can participate in weak C-H/Cl hydrogen bonding interactions with polarized CH 2 groups. As a result, the grid-like pattern formed by 2 should be stabilized by other weak interactions. According to the proposed model shown in Fig. 3c, the patterns are maintained by weak CH/O forces between the hydrogens of the aromatic rings connected to the central triple bond and the oxygen atoms of the dodecyloxy chains. On this basis, each molecule interacts with four neighboring molecules through a total number of eight C-H/O contacts: four of them involving four of the oxygen atoms of the peripheral chains and the remaining four involving the two central aromatic rings, two on each side, thus creating a uniform structure exhibiting a network density of 0.15 molecules per nm 2 . ## Conclusions In summary, we have observed distinct patterns through STM by exploiting the self-assembly behaviour of an OPE-based Pd(II) complex 1. Above a critical concentration, the units of 1 are arranged in a parallel fashion into lamellar patterns. In more diluted solutions, however, the involvement of the Cl-Pd(II)-Cl fragment of 1 in C-H/Cl interactions with oxygen-polarized CH 2 groups of the side chains along with surface effects of the HOPG lattice lead to one of the most complex tessellations ever visualized by STM: a special type of semiregular rhombitrihexagonal tiling. The key influence of the Cl-Pd(II)-Cl center is demonstrated by investigating a related non-metallic compound 2. Our fndings bring to light that unconventional non-covalent forces such as C-H/X interactions may become relevant enough to strongly influence pattern formation. Such surface tessellations with uniform porosity may be exploited for the encapsulation of guest molecules on surfaces, providing access to surface-active 2D or 3D assemblies, as recently shown by Tait, Flood and co-workers.

chemsum

{"title": "Concentration-dependent <i>rhombitrihexagonal tiling</i> patterns at the liquid/solid interface", "journal": "Royal Society of Chemistry (RSC)"}

microstructural_control_of_polymers_achieved_using_controlled_phase_separation_during_3d_printing_wi

## Abstract: Controlling the microstructure of materials by means of phase separation is a versatile tool for optimizing material properties. In this study, we show that ink jet 3D printing of polymer blends gives rise to controllable phase separation that can be used to tailor the release of drugs. We predicted phase separation using high throughput screening combined with a model based on the Flory-Huggins interaction parameter, and were able to show that drug release from 3D printed structures can be predicted from observations based on single drops of mixtures. This new understanding gives us hierarchical compositional control, from droplet to device, allowing release to be 'dialed up' without any manipulation of geometry. This is an important advance for implants that need to be delivered by cannula, where the shape is highly constrained and thus the usual geometrical freedoms associated with 3D printing cannot be exploited, bringing a hitherto unseen level of understanding to emergent material properties of 3D printing. Often it is not possible to exploit design freedoms due to limitations in the manufacture or the implementation of a device. A pertinent example is the long-term subdermal delivery implant. This is usually cylindrical with a size (maximum diameter 2 mm, length range 1 to 4 cm) defined by a combination of implantation method and anatomical positioning . Currently, such devices are manufactured by a process which heats a blend of polymer and active pharmaceutical ingredient (API) to around 100°C, extrudes and then cuts them to size. However, current systems are not personalizable, nor is it possible to combine multiple drugs into a single treatment. One route to achieving personalization is through 3D printing. Recent advances in 3D printing have shown it can be used for controlling drug elution, most commonly through variation in geometry, or variation in composition . Whilst Fused Deposition Modelling and Binder Jetting are popular and show promise, for the manufacture of implants they are limited by their resolution. Ink jet based 3D printing, however, offers multiple benefits including its scalability, high resolution and importantly, drop by drop deposition that can provide both control over material properties at the microscale as well as the ability to co-deposit multiple materials (drugs). In this work, we exploit the latter to develop microstructural control at the sub droplet level which then permits the precise tailoring of the drug release. This microstructuring only emerges as a function of the drop by drop deposition inherent to ink jet printing. In developing this concept, we report the creation of a library of multicomponent inks, whose diversity of physicochemical properties allows for the range of phase separation behaviour required for tailoring drug release. We show that, by understanding the 3 mechanisms that drive formation of this microstructure, we can predict microstructure that arises out of the ink jet printing process and reliably design and manufacture implants for tailored release. ## Materials Library To create the library of printable, functional materials, we first synthesized a range of low molecular weight (5 kDa) biodegradable oligomers with the following head-terminal group combinations, -OH, MA-OH and MA-A (Figure S2). The end functionalities were varied to control; (a) the drug release by influencing chain-end centered degradation, and (b) reactivity. The oligomers (PCL, PLA and PTMC) were synthesized from three core monomers (caprolactone, DL-lactic acid and trimethylene carbonate), offering a range of degradation rate/mode, crystallinity and thermal/mechanical properties. These polymers are widespread in the biomedical industry, and selected for an easier pathway to adoption compared to completely new materials. The oligomers were synthesized by ring opening polymerization using metal free organocatalysis, chosen for low toxicological impact of any subsequent medical devices. The library was created by combining nine hydrophobic oligomers in a 1:1 ratio with two relatively hydrophilic reactive solvents, n-vinyl pyrrolidone (NVP) and poly(ethylene glycol diacrylate) (PEGDA) 250 Mw, giving a total of 18 inks (Figure S3). After degradation studies, we formulated with a combination of NVP and PEGDA to further tune the degradation behaviour and enhance the structural integrity of the fast release formulations. ## Screening materials at the drop scale (10 picolitres) Microarrays of single drops of each of our 18 inks were rapidly deposited and cured on a glass slide using a high throughput (HTP) method, ready for characterization (see supplementary). In this single drop screening (SDS) images were taken of single 200 um spots (Figure 1B), each representing the deposited drops within ink-jet processes. Drops' surface chemical and microstructure/phase separation properties were evaluated using optical microscopy, time-of-flight secondary ion mass spectrometry (ToF SIMS) and automated peak force quantitative nanomechanics (QNM) atomic force microscopy (AFM). This analytical combination provided the phase separation and oligomer distribution data (Figure 1B and Figure S4) required to create a phase separation 'taxonomy' which was then related to a Flory-Huggins interaction parameter (χ) prediction of the likelihood of phase separation (see supplementary, Table S1-2) (Figure 1C). Our observations indicated that the mixtures exhibit three different types of microstructures, (a) homogeneous, completely interspersed mixture (b) dispersed droplets indicating nucleation and growth of domains and (c) spinodal decomposition (core-shell) mixture (Figure 1B). Consequently, the materials were ordered according to their χ values and compared to the observed microstructure taxonomy (Figure 1C). Broadly, the material microstructure conformed to classification by χ and approximate χ values demarking boundaries between microstructure types were identified: (a) below 0.025 presented one single phase (b) with values between 0.025 and 0.06 phase separated into dispersed droplets and (c) above 0.06 exhibited a core-shell microstructure. These material combinations were further screened using a HTP methodology designed to assess "printability" to eliminate those materials that cannot be easily printed using ink-jet printing (Table S3-5). ## Macro-scale cast samples (15 microliters) Degradation rate: To determine the influence of microstructure on the cured ink's macromaterial/degradation, cast cylinders (length 4 mm, radius 0.5 mm) of the 18 inks were prepared and subjected to degradation studies. The duration of degradation was sixteen weeks and the mass loss rates are shown in Figure 2A and Figure S5. As most of the structures printed with NVP didn't maintain their integrity following curing, we replaced the NVP only formulations with PEGDA/NVP solvent combinations. These were used in the biocompatibility and drug release studies to enable faster degradation than PEGDA alone, whilst maintaining some structural integrity. Drug Release: A cardiovascular disease hypertension active, trandolapril, was used as a model drug in drug release screening from candidate formulations (see supplementary). Here, the χ parameter was used to predict which ink component trandolapril has more affinity with and would migrate toward (Table S6). With PEGDA alone, Figure 2C, we observed that the fastest release profiles were those from PCLMA/PEGDA and PTMC/PEGDA, each of which presented a core-shell microstructure in the SDS. PTMCMA/PEGDA exhibited the next fastest release and presented a dispersed droplets microstructure. In each of these cases, χ analysis predicted that trandolapril had more affinity for and would migrate toward the PEGDA (Table S6). In contrast PTMCMAA/PEGDA, PLAMA/PEGDA and PLA/PEGDA had significantly slower release, and χ forecasted greater active affinity for the oligomers over PEGDA, suggesting that the release is significantly slowed due to hydrophobicity of these materials. Therefore, we hypothesized that when using PEGDA alone, cast drug release rates are predominantly controlled by their affinity for PEGDA and that the microstructure provides a secondary tuning parameter. In contrast, all formulations with PEGDA/NVP, with the sole exception of PTMCMAA, exhibited very similar dissolutions rates which were seemingly unaffected by the microstructure, i.e. the drug release appeared to be dominated by the PVP behaviour. The AFM and optical images of the PTMCMAA/PEGDA/NVP formulation surfaces indicate segregation behavior quite unlike those observed in other formulations, suggesting a complex set of interactions that may be leading to exceptional drug release behavior. ## Macro-scale 3D ink-jet printing of drug eluting samples The SDS and cast samples evaluation were used to guide the choice of materials for 3D inkjet printed drug releasing devices. Our screens indicated that two "levers" controlled deposition for an API, namely NVP inclusion and microstructure. Thus, we chose formulations predicted to show a range of release rates and therefore were guided by the release from cast materials (leading to the addition of NVP to PCLMA/PEGDA, and PTMCMA/PEGDA) and also by the SDS and degradation data (leading to use of PLAMA/PEGDA/NVP). We also incorporated a second active, the cholesterol lowering drug pitavastatin, into these formulations to demonstrate the predictions' effectiveness for multiple actives. Pitavastatin offers an effective contrast since it has a similar log P to trandolapril (3.97 and 3.45 respectively) but different aqueous solubility (0.426 mg/L and 2.5 mg/L respectively). The amount of drug released from 3D printed constructs was quantified on day 3, 15, 30 and 60, and 3D OrbiSIMS was used to unambiguously identify each compound (Figure S11) and ToF-SIMS micro-scale resolved maps of bulk composition (via depth analysis) were used to understand the role of phase separation on drug distribution and release (Figure 3B). Drug release and ToF-SIMS assessment showed that both PCLMA/PEGDA and PTMCMA/PEGDA constructs exhibited similar extended-release profiles and material distributions within the construct (Figure 3B). The latter were similar to those observed in the SDS microarray, indicating its reliability in predicting 3D printed material separation. ToF-SIMS (Figure 3B) also confirmed the affinity between the API and the PEGDA within a formulation, as we predicted from the χ values (Table S6). This indicated that microstructure dominated the drug release behavior in these printed samples, with the release likely via diffusion from exposed PEGDA, whilst also suggesting that the SDS, cast and printed samples have the same microstructure. Using PCLMA, PTMCMA and PLAMA oligomers with NVP/PEGDA resulted in a range of drug release profiles that broadly followed the degradation rates of the core polymers. We noted that PCLMA/PEGDA/NVP's release was not statistically different to that from PCLMA/PEGDA, whilst using PTMCMA resulted in a substantial increase and PLAMA gave full release in < 20 days. This behavior, in combination with insights from ToF-SIMS 3D mapping (Figure S12, Figure 3B) leads us to propose that, whilst cast samples are not a reliable guide to release when using NVP, the similarity in microstructure in SDS and 3D printed samples indicates the reliability of SDS as a guide to performance when manufacturing via printing. In each case (with or without NVP) the microstructure is a determining factor, but via a different mechanism. When using NVP, ToF-SIMS confirmed that the drug and PVP are homogeneously distributed, as expected when using χ values to inform PVP:drug compatibility. Thus, the release is driven by PVP dissolution rather than diffusion of the drug within the PVP and will be controlled by the extent of the exposed PVP, a feature dependent on the microstructure. This degradation is also dependent on the oligomer degradation speed, i.e. rate of exposure of further PVP surface area, leading to the rapid release seen when using PLAMA (Figure 2A). ## Conclusions We have shown that the microstructure formed in our polymer blends was process dependent and arose as a function of the drop-by-drop deposition technique. As a result, we demonstrated that it is possible to functionally tailor the composition of 3D printed constructs to successfully control the release of drugs incorporated within them. We selected suitable 3D printing inks using complementary HTP methodologies that allowed us to screen for various desired properties and down select formulations from these screens. Screening of behavior in single drops, combined with the Flory-Huggins interaction parameter provided a prediction of phase separation, and thus drug release, in 3D printed structures. In summary, we demonstrated a reliable toolkit for the development of formulations suitable for 3D printing that can be used to tailor long term drug release on demand. Carbosynth. In all cases the vials were dried in an oven at 50 °C overnight prior to use, and the HEMA and DCM were stored over molecular sieves and under an inert atmosphere. Benzyl alcohol (BA) and hydroxyethylmethacrylate (HEMA) initiated ROP of the oligomers. The oligomers were synthesized by ring opening polymerization using metal free organocatalysis, chosen for low toxicological impact of any subsequent medical devices. ROP experiments were performed adopting 'standard laboratory' conditions, i.e. ambient temperature and atmosphere. -OH ended macromers were initiated using BA, -MA and MA-A macromers were initiated using HEMA. Macromers were synthesized following the procedure by Ruiz et al . Briefly, 1000 mg of cyclic monomer (caprolactone, lactide or trimethylene carbonate) and BA or HEMA ([M]:[I] or DP0 ratios targeted to produce final molar masses of 5000 Da were weighed into a vial, which had been dried in an oven at 50 °C overnight and capped with a rubber septum. DCM (5 ml), was then added via syringe and the mixture was allowed to dissolve at room temperature (RT) for 5-10 minutes. Varying amounts of catalyst (1% mol/mol of TBD for lactide and trimethylene carbonate, 2 % mol/mol of TBD for caprolactone) were then added to trigger the ring-opening process. Reactions were observed to occur in time-frames ranging from 15-120 minutes, according to the monomer:initiator :solvent :catalyst adopted ratios. The reaction was terminated by catalyst deactivation upon adding an acidic solution and the polymer purified by means of multiple precipitation steps and dried in a vacuum oven. ## MA-A macromers end capping. The MA-OH macromers were further functionalized with an acrylate end using a Stenglich coupling esterification following the same procedure by Taresco, et al . Briefly, PCLMA, PLAMA or PTMCMA (0.2 mmol) and DMAP (0.04 mmol) were added to DCM (5 ml) at room temperature in a glass vial under magnetic stirring until complete dissolution. A second solution was prepared by dissolving 1 mmol of EDC and 1 mmol of acrylic acid in 2 ml of DCM. After dissolution, both solutions were mixed. The reaction was allowed to stir for 48 hours. The modified macromers were purified under multiple precipitation steps and dried in a vacuum oven. ## Materials library preparation Our library of inks was composed of a new set of biodegradable, UV curable materials, which were screened using high throughput methodologies to identify key characteristics suitable for use as an implant, such as printability, biodegradability, cytotoxicity and drug elution (Figure S1).The library was created by combining the nine hydrophobic macromers with two different relative hydrophilic reactive solvents in a 1:1 ratio (w/v) resulting in a total of 18 inks. The solvents functioned as diluents and cross-linkers. Polyethylene glycol diacrylate (PEGDA) 250 Mw and n-vinyl pyrrolidone (NVP) were chosen as the reactive solvents. PEGDA is commonly used as a plasticizer to reduce the glass transition temperature of polymers such as PLA, which helps reduce viscosity without the need of using high temperatures. It is also non-degradable so will not be depleted from the structure. Meanwhile, NVP was selected because once polymerized into poly(N-vinylpyrrolidone) (PVP) it has the ability to form a water-soluble composite structure with insoluble active substances and improve the release and solubility of drugs. Additionally, NVP is also known to increase the reactivity of acrylate resins and will degrade in hydrolytic conditions so be removed from any printed structure during use. All the formulations were a 50% w/v solution of the macromer (s) in the solvent (s). Whilst the reactive solvents both functioned as diluents, NVP is monofunctional so gives rise to linear polymer chains and PEGDA is difunctional so functions as a cross-linker/branching monomer so giving rise to a 3D network structure. Additionally, both were chosen as the reactive solvents because they are commonly used in pharmaceutical formulations owing to their ability to interact with hydrophilic and hydrophobic solvents, polymers and drugs, and they exhibit different degradation behavior. Formulations contained 1% Irgacure 2959 as photoinitiator. ## Microarray preparation. DMF (75% w/v) was used as a non-reactive solvent for all the formulations in this experiment in order to study their properties in high throughput without the need to optimize viscosity in advance. The microarrays were prepared on polyHEMA coated glass slides using the using XYZ3200 dispensing station (Biodot) and quilled metal pins (946MP6B, Arrayit) under argon atmosphere (< 2000 ppm oxygen) maintaining between 40 and 50% relative humidity. Each spot had an average diameter of 200 µm. The spots were UV polymerized under argon atmosphere for 10 min after printing. To remove the solvent, glass slides were dried the in vacuum oven for a week. Atomic Force Microscopy. Height, Peak Force error, DMT modulus, logDMT modulus, adhesion, deformation and dissipation images were simultaneously acquired by Peak Force QNM-AFM measurements (Bruker Fast Scan). Images of 5x5 m per spot were recorded by using a programmable stage. AFM cantilevers with a nominal stiffness nominal k= 40 N/m (RTESPA 300) were used. A Poisson's ratio of 0.3 was used in all cases. Three images were acquired per polymer spot throughout the micro array. The spring constant of each cantilever was estimated by using the thermal tune. Sample standards of polystyrene (PS) were also used to validate tip characterization. Images were analyzed using the NanoScope Analysis software. ToF SIMS: ToF-SIMS of microarray samples was carried out using a TOF.SIMS IV instrument from IONTOF GmbH (Muenster, Germany). ToF-SIMS analysis of positively charged secondary ions was carried out using a TOF.SIMS IV system from IONTOF GmbH (Münster, Germany) using 25 keV Bi3 + ion beam operated in the high current bunched mode delivering 0.3 pA with 100 µs cycle time, resulting in a mass range between 0 and 694 u. Secondary ion maps were acquired using the stage raster mode. The whole area was scanned once with one shot per pixel, ensuring static conditions. ## Flory-Huggins interactions parameter. To investigate the phase separation in pin printed droplets, we used a combination of the Flory-Huggins theoretical model and experimental characterization methods. The Flory-Huggins parameter (χ) describes the excess free energy of mixing and governs phase behavior for polymer blends and block copolymers . In order to calculate the χ value we first obtained the Hansen solubility parameter of the individual components of the formulation using the HSPiP program, were the δd, δp, δh and δTOT were obtained using the DYI tool of the software. We calculated the χ values following the procedure described by Imre et al. by using equation 1, where Vr is the molar volume of the repeating unit of the oligomer, R is the gas constant, T the absolute temperature and δ1 and δ2 are the total solubility parameter (δTOT) of the solvent and the oligomer respectively. The phase separation taxonomy was created from observations from the printed spots and the χ values. The boundaries were chosen such that we included all the samples that exhibited the dispersed droplet phenomena. This resulted in two exceptions that showed either core-shell or homogeneous microstructure, which reflected the fact that the boundaries were not hard, and could be influenced by other physical properties such as viscosity and curing rate; we estimated the likely error in the boundary by calculating the average difference between χ at the boundary and at the exceptions, resulting in an approximate error of ±0.01. Printability screening. To investigate the printability of the inks we used a HT method developed by Zuoxin et al where the viscosity and surface tension are measured using a liquid handler and the printability calculated using the Ohnesorge number (Z=1/Oh). The Ohnesorge number has been identified as the appropriate grouping of constants to characterize drop formation . Reis & Derby used numerical simulation of drop formation to propose 10 > Z > 1 for stable drop formation . To identify printability at different temperatures, eighteen based inks formed by the combination of the nine different macromers mixed with PEGDA and NVP were selected and screened using ranges from 40 ºC to 70 ºC. Degradation. This study was performed on the 18 primary inks. Cast cylindrical samples with dimensions of 4 mm length and 1 mm radius were used for this test. ## Degree of conversion. Samples were analyzed with a Perkin Elmer Frontier FTIR-ATR spectrometer (Seer Green, UK) from 4000 cm −1 to 600 cm −1 with a scan resolution of 2 μm and step size of 0.5 cm −1 . Three scans were collected for each sample. Prior to sample spectrum collection, a background was collected on the clean ATR crystal. The degree of curing was calculated by quantifying the reduction of the C=C acrylate stretches (1636 cm -1 ) and the CH 2 acrylate twist 810 cm -1 when the macromers were combined with the reactive solvent PEGDA. The degree of conversion of on the samples mixed with NVP was calculated by looking at the reduction of the C=C vinyl groups of the NVP (1639 cm -1 ). Cytotoxicity (Extract test). To test biocompatibility (Figure 2B), we performed an indirect cytotoxicity test for a period of 30 days to determine any evolving cytotoxicity of leached products, either through residual monomers or products emerging through polymer degradation. BJ6 fibroblasts were grown in Dulbecco's modified eagle medium (DMEM) supplemented with 10% (v/v) foetal calf serum, 1 % MEM non-essential amino acids solution (Sigma-Aldrich), and 1% antibiotics/antimycotics (100 units/mL penicillin, 100 mg/mL streptomycin, and 0.25 mg/ml amphotericin B; Life Technologies). Cells were cultured until they reached 80% confluency and subsequently detached from the culture surface using trypsin/EDTA (0.25%/0.02% w/v), centrifuged at 200 x g for 5 min and resuspended in culture medium. Cells were seeded in a 96 well plate at a density of 5,000 cells per well and allowed to attach for 24 hours before the cytotoxicity experiments. A new seeded well plate was used for each time point. Triplicates of each formulation cast samples were sterilized under UV light (0.05 mW/cm 2 , 265nm) for 50 minutes and transferred into a 48-well plate. Each well containing a sample was filled with 1 ml of culture medium. Samples were incubated in the medium for a total of thirty days to allow for leaching of any cytotoxic components. After day 1, day 3 and day 30 of incubation, 200 µl of the supernatant were transferred in triplicate to the cells seeded in the 96-well plates. Cells cultured in in standard medium were used as negative control. Cells were incubated for 24 hours with the supernatant with cells cultured in fresh culture medium used as a negative control. Cytotoxicity was measured using Presto BlueTM (Invitrogen) following the manufacturer's instructions. The fluorescent signal was measured with an automated microplate reader (Tecan) using an excitation wavelength of 560 nm and an emission wavelength of 590 nm. For the cytotoxicity percentage calculations, the fluorescent background control was first subtracted from all the samples. Then the percentage was calculated by multiplying the fluorescence of each sample by 100 and then dividing the total by the average fluorescence of the negative control. Drug release study. The drug release profile was screened for a period of eight weeks on 16 of the formulations, with PCLMAA/PEGDA and PLAMAA/PEGDA being eliminated as they were not within the printable range. The drug trandolapril was selected for this screening. Formulations containing 0.65% w/v of trandolapril were casted in the same way than for the degradation study. Samples were transferred to individual vials containing 3 ml of phosphate buffered saline solution and placed in an incubator at 37˚C. 500 µl of the PBS solution were collected and each timepoint and filtered (0.45µm) for the HPLC analysis. The PBS solution was refreshed at each timepoint. For the drug release studies of the 3D printed samples the formulation were prepared in the exact same way than the cast ones. HPLC. Samples were characterized with an Agilent (Santa Clara, USA) HPLC Series 1260 system, equipped with an auto sampler, degasser, UV lamp and multi-diode array detection. A wavelength of 210 nm was used to quantify trandolapril and 280 nm for pitavastatin. Method mobile phase compositions were 65% buffer and 35% acetonitrile (Fisher HPLC gradient grade). Phosphate buffer was composed of 6.8 g/L monobasic potassium phosphate (anhydrous, Sigma Aldrich) adjusted to pH 3.0 with phosphoric acid (85-90%, Fluka). An Ultimate LP-C18 column (5 μm, 25 cm × 4.6 mm diameter) was used to separate the samples at 40 °C. A flowrate of 1 mL/min using a 10 μL injection volume was implemented; runtime was 10 min. Trandolapril stock solutions were prepared by sonicating trandolapril/pitavastatin (nominally 1 mg, Carbosynth) in 10 mL methanol (Fisher HPLC grade) and diluting the volume with dissolution media in a 10 mL volumetric flask. Standards were prepared with the stock solution and dissolution media. Printing. The formulations were printed using a Dimatix Materials printer (DMP-2830 Fujifilm). The printer was enclosed in a metallic environment box and filled with nitrogen gas. The oxygen level was kept between 0.25 ± 0.05% during the printing process to minimize the inhibition effect caused by oxygen during the free radical photo-polymerization curing procedure. A 10pL disposable printhead, Dimatix Materials Cartridge (DMC-11610, Fujifilm) was used for printing. In-line UV curing was applied at the cartridge height immediate after each swath of ink droplets are deposited, by using a LED UV unit (365nm, 800mW/cm 2 , Printed Electronics Limited, Tamworth, UK) attached and move with the printhead unit. The printing temperature was set to 28°C. The sample was printed at 30 µm for the first layer and reduced to 20 µm for all the following layers. The height of the printhead was set to 700 µm with an increment of 9 µm after each layer printed. The samples were 3D printed using an inkjet printer (Dimatix DMP 2800) on the substrate polyethylene naphthalate. Samples dimensions were 5x5x 1 mm. Individual sessile droplet size when deposited varied depending on the mixture being processed. ToF-SIMS of printed samples was carried out using a 3D OrbiSIMS (hybrid SIMS) instrument from IONTOF GmbH (Muenster, Germany). Secondary ion mass spectra were acquired in negative ion polarity with delayed extraction mode using a 30 keV Bi3 + primary ion beam delivering 0.3 pA.The ToF analyser was set with 200 µs cycle time, resulting in a mass range between 0 and 3493 mass units .For the surface spectra, the primary ion beam was raster scanned over different areas with the total ion dose kept under the static limit of 10 13 ions/cm 2 . The 3D depth profiling data were acquired in dual-beam mode by raster scanning the primary ion beam over regions of up to 150 x 150 µm 2 at the centre of 300 x 300 µm 2 sputter craters formed using an argon gas cluster ion beam (GCIB). The GCIB was operated with 20 keV and 2000 atoms in the cluster delivering a pulsed 5 nA beam current. The analysis was performed in the "non-interlaced" mode with a low-energy (20 eV) electron flood gun employed to neutralise charge build up. 3 sputter frames were performed per cycle with 15 analysis scans per cycle and a pause time in between cycles of 0.5 s. Optical profilometry was used to determine the crater depth after ToF-SIMS depth profiling experiments and calibrate the depth scale. Scans were obtained using a Zeta-20 optical microscope (Zeta Instruments, CA, USA). All maps were produced using SurfaceLab and 3D visualisations were produced using the simsMVA software . Intensities were normalised by total ion counts to correct for topographic features. The final 3D representations were created by combining rendered isosurfaces ranging from 40% to 90% of the maximum normalised intensity for each ion. orbiSIMS of a cross section of a multi-layer printed sample containing all compounds of interest was carried out using a 3D orbiSIMS (hybrid SIMS) instrument A 20 keV Ar3000 + imaging GCIB of 5 µm diameter was used as primary ion beam, delivering 18 pA . We saw no significant reduction in cell viability on day 3 when compared to the control group (cells cultured in standard medium). However, there was a 10% viability reduction on day 30 when the cells were cultured with medium from the PLA based samples (Figure 2B), most likely due to the acidic degradation by products. This images shows pH changed observed in the medium after been incubated with the samples for 30 days. The change in pH was detected due to the presence of phenol red in medium. The lighter the colour the more acidic the medium. Table S6. Flory-Huggins interaction parameter of the different components of the formulations with trandolapril and pitavastatin. When the formulations contained PEGDA/NVP (1:1) as the reactive solvent, there was an initial burst release of more than 20% in all cases, directly attributable to the dissolution of PVP when immersed in an aqueous environment.

chemsum

{"title": "Microstructural control of polymers achieved using controlled phase separation during 3D printing with oligomer libraries: dictating drug release for personalized subdermal implants", "journal": "ChemRxiv"}

evolutionary_importance_of_the_intramolecular_pathways_of_hydrolysis_of_phosphate_ester_mixed_anhydr

## Abstract: Aminoacyl adenylates (aa-AMPs) constitute essential intermediates of protein biosynthesis. Their polymerization in aqueous solution has often been claimed as a potential route to abiotic peptides in spite of a highly efficient CO 2 -promoted pathway of hydrolysis. Here we investigate the efficiency and relevance of this frequently overlooked pathway from model amino acid phosphate mixed anhydrides including aa-AMPs. Its predominance was demonstrated at CO 2 concentrations matching that of physiological fluids or that of the present-day ocean, making a direct polymerization pathway unlikely. By contrast, the occurrence of the CO 2 -promoted pathway was observed to increase the efficiency of peptide bond formation owing to the high reactivity of the N-carboxyanhydride (NCA) intermediate. Even considering CO 2 concentrations in early Earth liquid environments equivalent to present levels, mixed anhydrides would have polymerized predominantly through NCAs. The issue of a potential involvement of NCAs as biochemical metabolites could even be raised. The formation of peptide-phosphate mixed anhydrides from 5(4H)-oxazolones (transiently formed through prebiotically relevant peptide activation pathways) was also observed as well as the occurrence of the reverse cyclization process in the reactions of these mixed anhydrides. These processes constitute the core of a reaction network that could potentially have evolved towards the emergence of translation.T he biosynthesis of peptides involves aminoacyl adenylates (aa-AMPs), formed through the reaction of ATP with a-amino acids (aas) (Fig. 1), that are subsequently used to aminoacylate tRNA. Their standard free energy of hydrolysis value DGu9 5 ca. 270 kJ mol 21 , determined for Tyr-AMP 1 , ranks them among the energy-richest biochemicals. Aa-AMPs possess a phosphate group transfer potential much higher than ATP 1 and might then constitute adenylating agents as well as aminoacylating agents 2,3 . The otherwise unfavourable 1 reaction of ATP with a-amino acids (K 5 3.5 3 10 27 ) is driven towards completion by selective stabilization of aa-AMPs in the active sites of aminoacyl tRNA synthetases (aaRSs). They usually remain sequestrated by the enzyme and are not released in solution before reacting with tRNA. The importance of this process can be appreciated by considering that the set of aaRS enzymes, responsible for the association of amino acids with their cognate tRNAs, actually holds the key of the genetic code. The evolutionary path through which adenylates were introduced in the process remains unidentified. In addition of being thermodynamically unfavourable, the spontaneous reaction is indeed very slow in the absence of enzyme 4,5 , so that the emergence of the biochemical amino acid activation pathway remains unexplained before a set of catalysts (very probably ribozymes) could lead to an embryo of the genetic code for prebiotically available amino acids 6 . In spite of this obstacle, the evolution of this pathway from an abiotic process of random peptide formation via the polymerization of a-amino acid mixed anhydrides with phosphate (aa-PMAs) or phosphate esters (aa-PEMAs) and adenylates (aa-AMPs) has prompted much work [7][8][9][10] . However, the abiotic formation of adenylates or their analogues from phosphate anhydrides did not receive any experimental support. As a matter of fact, the claim 11 that ATP is capable of driving the polymerization of aamino acids on clays through aa-AMP intermediates turned out to be non-reproducible 12 . Though the genetic code might have evolved late in the hypothesis of an ''RNA world'' without needing ATP activation as shown by the successful selection of ribozymes capable of aminoacylating RNAs using either amino acid esters 13 or activated RNAs 14 , an early co-evolution involving the chemistries of nucleotides and amino acids is consistent with the comparatively higher abundance of the latter as the products of abiotic processes. Therefore, selecting the co- evolutionary option, the elucidation of the potential evolutionary process through which aa-AMPs could have been introduced requires the identification of simple pathways capable of leading to these intermediates. A likely possibility is the reaction of a-amino acid N-carboxyanhydrides (NCAs) with inorganic phosphate 15 and its esters including adenylates that takes place spontaneously at moderate pH 16,17 (Fig. 2a). This possibility is supported by the role of NCAs deduced from the literature 2 and the disclosure of realistic abiotic pathways for their formation during the last decade 18,19 . Since the activation of the C-terminus in peptides has recently been identified as a plausible prebiotic pathway and involves the formation of 5(4H)-oxazolone intermediates 20 , it is reasonable that similar mixed anhydrides with phosphates involving acylated amino acids (acyl-aa-PEMAs) or peptides (peptidyl-PEMAs) could be formed by reaction of the energy-rich cyclic intermediate (Fig. 2b). The occurrence of abiotic pathways leading to aa-PEMA or peptidyl-PEMA must have preceded their involvement in chemical evolution. However, the low stability of these mixed anhydrides and the availability of highly reactive cyclic intermediates prone to polymerize more easily renders their role in early abiotic processes of peptide formation highly questionable. The kinetic stability of aa-AMPs and of other aa-PEMAs has been studied in aqueous solution leading to contradictory results in the literature . Of particular interest with regard to an evolutionary context is the description of a highly efficient CO 2 -catalyzed path of hydrolysis . No definitive mechanism has been proposed but the intermediacy of NCAs is highly probable 2,25,26 since other activated amino acids (nitrophenyl esters, thioesters) proved to undergo conversion into NCAs in hydrogen carbonate buffers 25 . This analysis casts doubts on the possibility that aa-AMPs constitute efficient monomers for the abiotic formation of peptides in aqueous solutions 2,3,26 since most early Earth aqueous environments are likely to have contained CO 2 or HCO 3 2 . The present investigations were aimed at providing data on the efficiency of the CO 2 -promoted pathway (Fig. 3a) in aqueous solution at neutral pH and in the presence CO 2 concentrations compatible with early Earth environments and at clearly identifying the NCA as an intermediate. They address both the issues of the stability of aa-AMPs and of other aa-PEMAs and that of the path of peptide formation. They demonstrate the prevalence of the CO 2 -promoted pathway in the hydrolysis of adenylates. More importantly, using model amino amide reactants, they additionally demonstrate that peptide bond formation takes place predominantly from the cyclic intermediates rather than directly from the mixed anhydrides ruling out any possibility of considering the latter as direct peptide precursors at early stages of chemical or bio- chemical evolution. Lastly, considering NCAs as likely precursors of aa-AMPs and aa-PEMAs, the hypothesis of an abiotic formation of non-coded peptides through these mixed anhydrides becomes unnecessary. The evolution of translation must then have proceeded through a pathway independent from abiotic polymerization. This work also addresses the more general goal of understanding the stability of phosphate mixed anhydrides of amino acids and peptides in aqueous media at moderate pH. As a matter of fact, though Nacylation is an obvious way to prevent CO 2 participation, another intramolecular path of breakdown through 5(4H)-oxazolones is possible in the case of acyl-aa-PEMAs (Fig. 3b). Therefore, the issues of the importance of the NCA and 5(4H)-oxazolone pathways in the reactions of the corresponding mixed anhydrides (Fig. 3) are raised as well as that of the potential role of these cyclic intermediates as potential prebiotic precursors of these mixed anhydrides (Fig. 2). The consequences of these chemical pathways as factors determining early biological evolution of amino acid activation processes and their constraints on the contemporary biochemistry of adenylates will also be discussed. ## Results Experiments were carried out from model systems derived from Omethylated tyrosine 5 (Fig. 4) likely to be representative of the reactivity of usual amino acid derivatives. The UV-absorption of the tyrosine side chain (l max 5 273 nm) was selected to monitor reactions by HPLC at a reasonably low (0.05-1 mM) concentration range in which activated intermediates have a lifetime sufficient for their behaviour to be determined. Furthermore, phenol methylation was introduced to simplify analyses by avoiding any side-reaction of this group. Reactions were carried out in non-nucleophilic MES or MOPS buffers at pH values of 6.5 or 7.5, respectively, whereas 50 mM phosphate or methyl phosphate buffers were used for studying the transient formation of mixed anhydrides. Analyses were performed to monitor the reaction progress of samples stored in the HPLC systems located in a room maintained at the temperature of 20uC. Fast reactions were monitored by withdrawing 1 mL samples from the reaction medium and the reaction was blocked by addition of a formic acid solution to bring the pH to a value below 4 (Supplementary information). NCAs as intermediates of aa-PEMA reactions promoted by CO 2 . The hydrolysis of methyl phosphate mixed anhydride 1b was studied in buffered solutions in the presence of varying contents of CO 2 / HCO 3 2 . The reaction rates were observed to strongly depend on the presence of CO 2 as shown by a c.a. 4 fold increase in rate using pH 6.5 MES buffers previously equilibrated with air as compared with a solution flushed with N 2 for 60 min (Fig. 5, panel A). The rates could be reduced by further c.a. 35% by extensive degasification through cycles of freezing at 295uC/gas removal under vacuum/ melting in a closed vessel. Under the conditions of the experiment displayed in the panel A of Fig. 5, the starting material 1b (HPLC retention time, r.t. 4.6 min, method A) disappeared slowly and several species containing the methoxyphenyl moiety (l max 273 nm) were observed, namely the free amino acid 5 (r.t. 8.4 min) representing the main product of hydrolysis but also several peaks corresponding to the dipeptide H-Tyr(Me)-Tyr(Me)-OH (r.t. 22.7 min) and the diketopiperazine cyclo-Tyr(Me)-Tyr(Me) (r.t. 23.6 min), very probably resulting of the cyclization of the mixed anhydride H-Tyr(Me)-Tyr(Me)-OPO 3 Me 2 , which has not been properly identified. The presence of these two products was confirmed by HPLC-MS analysis ([M 1 H] 5 373.2 at r.t. 1.52 min and 355.2 at r.t. 1.88 min, method C). By contrast, the addition of 2 or 10 mM NaHCO 3 to the buffer led to the fast disappearance (#1% after 3 min) of the mixed anhydride 1b as monitored by HPLC analysis (Fig. 5, panel B). An intermediate (r.t. 23.1 min, method A) formed in proportion yields as high as .06 min, respectively, method C). By contrast reduced amounts of diketopiperazine cyclo-Tyr(Me)-Tyr(Me) formed confirming that the starting material lifetime was not sufficient for it to behave as a polymerization initiator leading to a dipeptide mixed anhydride prone to cyclization 27 . Under these conditions involving the presence of HCO 3 ## 2 , the polymerization into peptides thus proceeds through the NCA rather than directly from the starting material. An NCA intermediate was also observed to form rapidly at pH 7.5 in 100 mM MOPS buffers in the presence of added HCO 3 2 (Supplementary Information, Fig. S1). This behaviour indicates that the formation of long peptides from adenylates reported in the literature 9,10 results probably from the polymerization of NCAs rather than from that of adenylates. The conversion of aminoacyl adenylates into NCA in the presence of CO 2 /HCO 3 2 was investigated starting from the Tyr(Me) derivative 1c (Supplementary Information, Fig. S2). The conversion of 1c into NCA was observed to proceed with rates similar to that observed for mixed anhydride 1b. The release of AMP (r.t. 1.5 min, method A) accompanying the formation of NCA 3 could be detected by HPLC allowing the reaction to be monitored at 50 mM concentrations of reactant 1c (r.t. 6.8 min, method A). The lifetime of the adenylate decreased with increasing concentrations of CO 2 /HCO 3 2 (t 1/2 , 80 min, ,25 min, and ,2 min at pH 6.5 in N 2 -flushed buffer, air equilibrated buffer and in the presence of 500 mM HCO 3 2 , respectively). At pH 7.5 the lifetime of adenylate 1c was reduced to less than 1 min in the presence of 500 mM HCO 3 2 , which means that this mixed anhydride is likely to be converted into NCA within a few seconds at concentrations of CO 2 /HCO 3 2 above 2 mM and at pH value close to neutrality, which are representative of the present day ocean or physiological fluids. It is worth noting that this lifetime is not sufficient for peptides to be significantly formed by a direct reaction with adenylate so that any observation of peptide products under these conditions results for the most part from the intermediacy of NCAs. At pH 4, the hydrolysis of mixed anhydride 1b was much slower (t 1/2 5 ca. 550 min) and CO 2 catalysis was not observed (Supplementary Information, Fig. S3). This result is consistent with the results obtained by Kluger from alanyl ethyl phosphate 24 . The protonation of the amino group of 1b increases the electrophilic character of its acyl group and then the rates of nucleophilic attack, but it also prevents any possibility of reaction with CO 2 according the pathway of Fig. 3a. The hydrolysis of the acetylated mixed anhydride 2b was indeed observed to be slower (t 1/2 , 950 min at pH 6.5) and was not affected by addition of 10 mM NaHCO 3 (Fig. 6) in a way consistent with this explanation and with previously reported analyses 22 . However, it is important to emphasize that the CO 2 -catalyzed pathway does not only constitute a process leading to the deactivation and the hydrolysis of mixed anhydrides since peptide formation can be improved significantly by this means. As a matter of fact, with regard to peptide formation, the prevalence of the NCA pathway was demonstrated by studying the model reaction of 1 mM mixed anhydride 1b with 5 mM glycinamide either in a nitrogen-flushed sample or in the presence of 2 mM NaHCO 3 (Fig. 7). Importantly, less than 2 min were sufficient for the starting material to be exhausted in the presence of carbonate, whereas CO 2 removal increased the reaction times to much higher values (t 1/2 , 50 min) and reduced the final yield in dipeptide (Fig. 7). This reaction remained faster than that observed for the acetylated mixed anhyd-ride 2b (t 1/2 , 260 min) unable to undergo the conversion into NCA, but that will be demonstrated below to partly undergo cyclization into 5(4H)-oxazolones. These experiments carried out using glycinamide for mimicking a growing peptide chain show that the polymerization of adenylates and other aa-PEMA is improved in the presence of CO 2 by the occurrence of the NCA pathway owing to both the higher reactivity of the latter intermediate and its ability to suppress diketopiperazine formation. The interconversion of 5(4H)-oxazolones and acyl-aa-PEMA and peptidyl-PEMA. The reaction of Ac-Tyr(Me)-OH-derived oxazolone 4 in methyl phosphate-buffered aqueous solution (pH 6.5) at 20uC was monitored by HPLC and compared with the hydrolysis of mixed anhydride 2b in MES buffers (Fig. 6). Comparable rates were observed and the intermediate of the 5(4H)-oxazolone 4 reaction was identified in situ by HPLC-ESI-HRMS (negative mode, calcd for C 13 H 17 NO 7 P 2 , 330.0743; found 330.0747) as the mixed anhydride 2b. A similar behaviour was observed from a reaction of inorganic phosphate (Supplementary Information, Fig. S5). The hydrolysis of mixed anhydride 2b was monitored by HPLC at 20uC in buffered solutions (Fig. 6). The reaction was also carried out in D 2 O to detect any hydrogen/ deuterium exchange resulting from the transient formation of 5(4H)-oxazolone 20,28 and compared to the product of a similar reaction of pure oxazolone 4 (Table 1). The values obtained demonstrate the occurrence of an intramolecular pathway already suspected from the higher rate of conversion of acylated aa-AMPs compared to simple acyl-adenylates 29 . At pH values below 5, the hydrolysis of anhydride 2b (Supplementary Information, Fig. S4) has been observed to become faster in a way similar to the observation made by Lacey's group for Ac-Phe-AMP 22 . The identification of an intramolecular pathway made in the present work strongly suggests that the acid catalysis of acyl-aa-PEMA hydrolysis is the consequence of a facilitated cyclization from a good neutral phosphate leaving group. However, the absence of H/ D exchange from the reaction of neither acyl-aa-PEMA 2b nor 5(4H)-oxazolone 4 at this pH (Table 1) prevented any determination of the actual pathway of hydrolysis of mixed anhydride 2b. Similarly, we analyzed the degree of D/H exchange during the reaction of 2b with L-Ala-NH 2 in D 2 O at pH 6.5 (Table 1). The observation of a partial deuteration of the two diastereoisomers of the dipeptide product demonstrates that even when a better nucleophile is present, the a-proton is exchanged to a significant extent before the subsequent reaction of the 5(4H)-oxazolone takes place. The fast reaction of acyl-aa-AMP 29 and other acyl-aa-PEMA results therefore, at least for a noticeable part, from a transient conversion into 5(4H)-oxazolones. Interestingly, the different degrees of deuteration of the two diastereomers indicate that the intramolecular path of Fig. 3b has a higher stereoselectivity as compared to the direct path (the reactants 2b and 4 were prepared under a racemic form 28 ). ## Discussion As regards aa-PEMA reactions, it is noteworthy that CO 2 catalysis proceeds through a pathway involving induced intramolecularity 30 . This kind of process shares one of the most important components of enzymatic activity, which corresponds to the utilization of binding energy to non-reacting portions of the substrate to bring about catalysis 31 . It was also proposed to constitute the easiest path for enzyme evolution under the name of uniform binding 32 and is moreover necessary for enzymes to exceed a physical limit 33 . Induced intramolecularity has also been used to drive highly stereoselective catalysis in organic synthesis 34,35 . The efficiency of this kind of catalysis relies on the rates of intramolecular reactions 36 . Carbon dioxide present at total concentrations of ca. 30-40 mM in pH 6.5 solutions equilibrated with air (as deduced from the Henry's coefficient of CO 2 37 and the pK a of carbonic acid) brings about a rate increase sufficient to render the catalytic pathway largely predominating, which is remarkable by considering a simple three-atom molecule compared to the efficiency of enzymes 38 . The ease of formation of 5-membered cycles from a-amino acid mixed anhydrides is also demonstrated by the conversion of acyl-aa-PEMA into 5(4H)-oxazolones. These experiments demonstrating that the NCA path is prevailing at pH values close to neutrality in solutions equilibrated with air at present atmospheric levels of CO 2 (ca. 0.04%) suggest that the pathway must be overwhelming in natural environments with higher contents. The experiments at 2 mM HCO 3 2 are representative of present day ocean total concentration of dissolved carbonate 39 showing that the lifetime of aa-PEMA is expressed in tens of seconds in these media at pH 7.5. In biological media, with total carbonate concentrations approaching or exceeding 10 mM, the lifetime of mixed anhydrides would be even shorter. The early atmosphere had a CO 2 content that remains poorly constrained 40 but values similar to the present atmospheric levels 41 , or representing up to hundred times this value 40,42 , are often considered. Under these conditions, aa-PEMAs would be rapidly converted into NCA before any direct conversion into peptides could take place, which discards the earlier proposed contribution of aa-AMPs in the formation of prebiotic peptides . Moreover, a less efficient polymerization ability of aa-PEMA and the diketopiperazine side-reaction make them improbable peptide precursors. The possibility that a very low content of CO 2 in the atmosphere could have transiently permitted mixed anhydrides to be stabilized 23 is made unlikely because it would have also required a very efficient removal of the most part of CO 2 in the whole ocean ($2 mM in HCO 3 2 ). On the contrary, the development of the activation pathway leading to translation must have occurred in an environment in which the role of NCA was unavoidable rather than in a local environment in which the mixed anhydrides were preserved from the presence of CO 2 and HCO 3 2 by any kind of geochemical processes. NCA can be considered not only as intermediates of the degradation pathway of adenylates but also as precursors of any kind of aa-PEMA mixed anhydrides including adenylates as well as precursors of peptides through a pathway suppressing diketopiperazine side-reaction. From this point of view, the catalysis by carbon dioxide may lead to a fast exchange among different energy-rich species capable of linking activated amino acids to phosphorylating species. This distribution of energy in a reaction network, that may have anticipated the role of ATP as an energy currency, ensured a global far from equilibrium situation that was essential even at early stages of chemical evolution 43 www.nature.com/scientificreports and nucleotide chemistries 44 the CO 2 -catalyzed pathway may then constitute a key-element in the systemic integration of the two subsystems 45 . The fast conversion of adenylates, and more generally mixed anhydrides aa-PEMAs, into NCAs at low concentrations of CO 2 in water questions the way through which the biochemical amino acid activation evolved. As a matter of fact, aa-AMPs, possibly produced from ATP through ribozyme activity 46 , would rapidly be converted into NCAs impeding the evolution of translation. Conversely, the catalytic activity of aaRSs might have evolved by acting on the thermodynamically favourable reverse reaction of aa-AMPs (formed spontaneously from NCAs) as a primitive pathway to produce ATP 2,3 . One could argue that the NCA pathway of Fig. 3a is still active in living cells but this speculation is not supported by any experimental data. However, the mechanism of pretransfer editing of misactivated aaRSs (through which adenylates are hydrolyzed) remains uncertain 47 . Any possible release of adenylates from the active site to solution 48 during this step would lead to the formation of the corresponding NCA within seconds. Whatever NCA is actually or not a biochemical metabolite, the present results indicate that living organisms probably had to limit the importance of the release of adenylates into solution after translation evolved since a conversion into NCA would certainly lead to random aminoacylation of pending amino groups likely to be harmful to protein functional integrity. From this point of view, the N-formylation of methionine needed to initiate ribosomal peptide synthesis in bacteria might be considered as a remnant of a period in which NCA could be released in the cytoplasm. Therefore, we conclude that the potential formation of NCAs at least influenced the development of the translation apparatus and that of the aaRS family of enzymes in order to avoid random aminoacylation and that the NCA pathway must be taken into account in evolutionary studies. Our analyses confirm the observations made by Lacey that CO 2 is a very efficient catalyst for the conversion of adenylates. However, taking into account the probable role of NCAs and the diversity of processes made available through their intermediacy leads us to the very different conclusion that the process could be favourable to the development and evolution of life rather than solely detrimental to the role of adenylates as intermediates of peptide formation. It is also worth noting that acyl-aa-PEMA that were considered by Lacey as blocked equivalents of aa-AMPs 22,23 does actually not constitute models of the reactivity of their parent compounds since they also undergo a spontaneous cyclization into 5(4H)-oxazolone. The transient formation of 5(4H)-oxazolone intermediates may be responsible for their efficiency in peptide formation 20 . The mixed anhydrides formed from free amino acids as well as peptide segments turn out to constitute unlikely precursors of peptides since their reactions are actually preceded by a very efficient cyclization into uncharged intermediates that thus constitute better electrophilic agents. This observation can be related to the evolutionary advantage of phosphate derivatives 49 that is partly related to their negative charge reducing spontaneous hydrolytic degradation with respect to their enzymepromoted reactions. From this perspective, their involvement required specific and efficient catalysts. However, the fact that NCA and 5(4H)-oxazolone also constitute precursors of mixed anhydrides through spontaneous processes provides a potential path through which these intermediates may have led for example to aminoacyl esters of RNA at predisposed locations 16,23,50 . ## Methods Reagents and solvents were purchased from Bachem, Sigma-Aldrich, or Euriso-Top and used without further purification. Starting materials and products samples were prepared according to standard procedures and characterized by 1 H, 13 C and 31 P NMR spectrometry and HRMS (Supplementary Information). NMR analyses were performed on a Bruker Avance 300 apparatus. HPLC analyses were performed on a Waters Alliance 2690 system with a photodiode array detector 996 using a Thermo Scientific BDS Hypersil C18 5 mm 2.1 3 50 mm column; mobile phase: A: H 2 O 1 0.1% TFA, B:CH 3 CN 1 0.1% TFA; flow rate: 0.2 mL/min and two different gradients; method A: 0 min (5% B), to 15 min (15% B), 25 min (60% B) and 26 min (100% B); method B: 0 min (5% B), to 10 min (20% B), 11 min (100% B). HPLC-ESI-MS analyses were carried out on a Waters Synapt G2-S system connected to a Waters Acquity UPLC H-Class apparatus equipped with a Acquity UPLC BEH C18, 1.7 mm 2.1 3 50 mm column; method C: A: H 2 O 1 0.01% formic acid, B: acetonitrile 1 0.01% formic acid; flow rate: 0.5 mL/min; linear gradient 0% to 100% B over 3 min.

chemsum

{"title": "Evolutionary Importance of the Intramolecular Pathways of Hydrolysis of Phosphate Ester Mixed Anhydrides with Amino Acids and Peptides", "journal": "Scientific Reports - Nature"}

electrochemical_proton_intercalation_in_vanadium_pentoxide_thin_films_and_its_electrochromic_behavio

## Abstract: This work examines the proton intercalation in vanadium pentoxide (V 2 O 5 ) thin films and its optical properties in the near-infrared (near-IR) region. Samples were prepared via direct current magnetron sputter deposition and cyclic voltammetry was used to characterize the insertion and extraction behavior of protons in V 2 O 5 in a trifluoroacetic acid containing electrolyte. With the same setup chronopotentiometry was done to intercalate a well-defined number of protons in the H x V 2 O 5 system in the range of x = 0 and x = 1. These films were characterized with optical reflectometry in the near-IR region (between 700 and 1700 nm wavelength) and the refractive index n and extinction coefficient k were determined using Cauchy's dispersion model. The results show a clear correlation between proton concentration and n and k. ## Introduction Nowadays there is a high demand for materials with welldefined, tunable optical properties to realize new optical modulation devices, e. g. optoelectronic switches and phase or intensity modulators as optical data processing is getting more and more important for communication and information technologies. Silicon photonic integrated circuits (PICs) are getting more and more important in this field in which the near-IR region is of great interest. Recently, self-holding optical actuators for silicon photonic waveguides have been proposed, these activators can maintain the switching state without a constant supply of energy. Different types of materials have been exploited which comprise phase change, insulator-metal phase transition, memristor-like plasmonic structure and electrochromic materials. Electrochromic materials are able to reversibly change their optical properties, i. e. refractive index and absorption coefficient by inserting charge which causes redox reactions in the material. Intercalation is commonly done using lithium cations or protons since excellent reversible coloration of the materials in the visible range can be effected with these ions. Some wellknown electrochromic oxides for the visible range are tungsten trioxide (WO 3 ), niobium pentoxide (Nb 2 O 5 ), molybdenum trioxide (MoO 3 ) and vanadium pentoxide (V 2 O 5 ). V 2 O 5 has been used in a broad field of different applications, for example as catalysts in oxidation reactions, in sensors as gas sensing material, as insertion electrodes for lithium ion batteries and as already mentioned above for smart window applications as electrochromic material. There, the electrochromic effect has only been studied with lithium intercalation and in the visible range of the electromagnetic spectrum. In previous works we examined the electrochromic properties of lithiated V 2 O 5 in the near-infrared (near-IR) region. Nevertheless, there is only few information available regarding insertion of protons. Wruck et al. and Ottaviano et al. describe that the insertion of a proton is possible in V 2 O 5 and that it is similar to Li + -intercalation but they only give details about the lithiation. In 1998, a solid-state proton battery was introduced by Pandey et al. where vanadium pentoxide was used as cathode material within a mixture of carbon and lead dioxide. Therefore, it is already known that V 2 O 5 may serve as a reversible proton intercalating material. Liu et al. presented insertion of hydrogen in the form of intercalated protons accompanied by excess electrons in the conduction band of vanadium pentoxide by treating it in a hydrogen containing atmosphere. In their work, they investigated the optical properties of V 2 O 5 in the visible range in dependence of the hydrogen content in the gas mixture in a Pd/V 2 O 5 device. Thus, they showed that vanadium pentoxide irreversibly changes in a first formation cycle of insertion and extraction of hydrogen but then remains optically passive for the subsequent cycles. By electrochemical insertion of protons, V 2 O 5 has never been tested before as an optically active and tunable material. In 2013, Malini et al. presented electrochromism of thin films of CeVO 4 , a mixed oxide sythesized from cerium dioxide and vanadium pentoxide. They inserted and extracted protons electrochemically via hydrochloric acid containing electrolyte and measured the transmittance in the ultraviolet and visible range in-situ. They showed that there is a decrease of the transmittance with rising H + content, giving a clear hint that V 2 O 5 can act as an electrochromic active material. To conclude, previous research has been focused on characterizing the electrochromism of vanadium pentoxide mainly in relation to lithiation, there is still a lack of information about the electrochromic properties of pure vanadium oxide with electrochemical proton intercalation especially in the near-IR region. Accordingly, it is of interest to investigate the electrochromic behavior as a function of concentration of intercalated protons. Therefore, we prepared vanadium pentoxide thin films via direct current magnetron sputter deposition and investigated in detail their electrochemical and electrochromic properties in the near-IR region in terms of proton insertion and extraction. ## Results and Discussion The sputtered and annealed V 2 O 5 films all exhibit an orthorhombic structure well matching literature data. An x-ray diffraction (XRD) spectrum of an as prepared V 2 O 5 film (600 nm thickness) on bare silicon is presented in Figure 1. No representative reflex of vanadium dioxide (VO 2 ) can be found so we conclude that all produced thin films consist of pure crystalline V 2 O 5 . In a next step we investigated the electrochemical performance in a proton containing electrolyte. Therefore, we have chosen trifluoroacetic acid in a solution of tetrabutylammonium perchlorate in propylene carbonate for the following reasons: In contrast to many other acid solutions, this composition does not dissolve the V 2 O 5 thin film. Nevertheless, trifluoroacetic acid is a strong acid with a pK a -value of 0.23, its solubility in various organic solvents is excellent and in our case it is serving as proton source. Tetrabutylammonium perchlorate is needed as conducting salt to guarantee good conductivity of our electrolyte system, whereupon especially the cation is too big to be intercalated in the structure of V 2 O 5 . As solvent propylene carbonate was chosen, because it is a well-known solvent for electrolytes in lithium-ion batteries, exhibiting good electrochemical stability. According to Ottaviano et al. and Tong et al., in an electrochemical experiment the intercalation and deintercalation mechanism of protons in V 2 O 5 can be described as While intercalating a proton, V 5 + is reduced to V 4 + and the proton should coordinate with the oxygen atom to form a hydroxyl species. Figure 2 shows a cyclic voltammetry (CV) measurement of a V 2 O 5 thin film (1.2 μm thickness) in the voltage range between 0.1 and 1.3 V versus silver chloride electrode (Ag j AgCl) and a scan rate of 1 mV • s 1 . A well-defined reduction peak and two oxidation peaks are observable, which indicate that at least the deintercalation of protons is taking place in a two-step mechanism, whereas a second peak in the reduction area could not be observed. The overall coulombic efficiency of the system is 89 %, indicating that the intercalation and deintercalation process is not fully reversible. This assumption is supported by the fact that the peak current densities are decreasing with repeated cycling. Regarding the relatively high number of protons For this purpose, a 300 nm thin film of V 2 O 5 was investigated with chronopotentiometry. The film was reversibly loaded with � 0.5 μA to a state corresponding to H 0.05 V 2 O 5 and back to H 0 V 2 O 5 for 40 cycles. After a total of five formation cycles the complete system reacts reversible by constantly ranging between 1.15 V (deintercalated, x = 0) and 0.55 V (intercalated, x = 0.05) vs. Ag j AgCl, as shown in Figure 3. By this means, repeated cycling of vanadium oxide films in proton conducting electrolyte has proven to be fully reversible in a well-defined range of x at least to x = 0.05. Comparable results can also be obtained for higher constant currents up to 50 μA, which enables faster switching between different values of x (in the case of a 300 nm thick 1 × 1 cm film less than one minute), making H x V 2 O 5 a promising system for switches or tuning devices. Graphs of chronopotentiometry measurements done with higher currents than 0.5 μA (5 μA and 50 μA) can be found in the supplementary data. After proving the electrochemical functionality of the H x V 2 O 5 system, the optical behavior of the thin films was investigated as a function of the proton content. Therefore, chronopotentiometry was used with a constant current of � 0.5 μA for a well-defined period of time to set an accurate value of x, before and after each experiment the cell voltage was measured until open circuit potential (OCP) was reached (OCP values see Table 1). Between every intercalation and deintercalation the cell was dismounted and the thin film was cleaned with ethanol. The sample was investigated with reflectance spectroscopy before and after the intercalation and additionally after the deintercalation. Figure 4 shows two chronopotentiometry graphs, one for the intercalation and one for deintercalation. The belonging reflectance graphs of the same sample with x = 0, x = 0.1 and deintercalated back to x = 0 are also shown in the same figure. It is visible that there is a definite deviation between intercalated and the deintercalated state. Furthermore, a very good reversibility of the intercalation and deintercalation process is evident since the reflectance spectrum after the deintercalation step matches very well to the spectrum measured before intercalation. This also proves the reversibility of the electrochemical process up to x = 0.1. Not only the reversibility, but also the stability or the maintaining of the optical properties of the intercalated state is of great importance for use as an optical switch. Therefore, a sample of H 0 V 2 O 5 was intercalated to H 0.1 V 2 O 5 , the sample was cleaned and the reflectance was measured directly afterwards. Then, the sample was allowed to rest in ambient conditions, whereas the reflectance was measured after six, nine and twenty days. In Figure 5 the reflectance curves are shown and it is observable that the reflectance changed a little bit between the freshly intercalated sample and the curve measured after six days. One reason for this could be the non-ideal cell-setup, as it was not possible to contact the whole area of the V 2 O 5 thin film with liquid electrolyte without directly contacting the current collector (see also in Figure 7). It is assumed that shortly after intercalation the layer is not yet in total equilibrium. This is also the reason why we decided not to examine the time response of the thin films with additional chronoamperometric measurements. Nevertheless, the reflectance curves after six, nine and twenty days are in good accordance to each other, no change in the reflectivity can be seen on further storage. Thus, the intercalated films are stable for the whole period of 20 days and are able to maintain their optical properties in the intercalated phase. In a next step we investigated if a systematic trend in the change of the optical properties can be seen by varying the concentration of protons. For this reason, the sample was intercalated to a certain amount of x in several steps from x = 0 to x = 1. Between the intercalations the sample was always taken out of the electrochemical cell and the reflectance was measured. We discovered that there is a clear correlation between the number of protons intercalated in the V 2 O 5 thin film and the optical properties which can be seen in Figure 6 (for small proton concentrations, x = 0.02 to x = 0.08 see the supporting information). Overall, the reflectance decreases with increasing proton concentration. In addition, the maxima and minima of the reflectance curves are shifted to lower wavelengths on the xaxis indicating a change in the refractive index n value with proton intercalation. The intercalation and deintercalation process is reversible up to a proton concentration of 0.1 < x < 0.2. With higher amounts, the reflectance graph of the proton free state (x = 0) cannot be obtained anymore and is shifted to lower reflectance values (reflectance graphs are shown in the supporting information). Figure 7 gives an impression of the irreversible change of the optical properties in the visible range of the spectrum, showing photographs of a H x V 2 O 5 thin film sample without protons and with a proton concentration of x = 0.2 and x = 0.6. There is a clear color change observable in the whole area where the sample was in direct contact with the electrolyte. It was impossible to reach the initial state again. In the following the focus is concentrated on lower amounts of x, namely in the range between x = 0 and x = 0.2 to stay in the reversible range of intercalation, which is especially interesting for optical devices. To determine the refractive index n and the absorption coefficient k of the measured samples, Cauchy's dispersion model was used according to equations (3) and ( 4). Figure 8 shows the Cauchy fit for a reflectance measurement done on a V 2 O 5 thin film with a proton concentration of x = 0.1. As can be seen the fit accuracy is excellent and reaches a goodness of fit value higher than 99 %. The received graphs for the n and k values dependent on the wavelength are presented in the inset. It is observable that the values for the refractive index n are constantly decreasing in a wavelength range between 700 and 1700 nm while the graph of the absorption coefficient k exhibits a maximum at 850 nm. At higher wavelengths, the values are also decreasing. According to this example of fitting all other measured reflectance graphs were evaluated. The obtained graphs for n and k are summarized in Figure 9, where it can be seen that there are just results for samples with a number of intercalated protons between x = 0 and x = 0.2. Beyond that no additional values for n and k could be obtained. It is noticeable that the limit of reversibility was determined between 0.1 < × < 0.2 and the reflectance curves can only be fitted up to x = 0.2, whereas the goodness of fit is dramatically decreased for x = 0.2. The shape of the obtained curves of n and k in dependence of the wavelength is comparable for all values of x and a clear trend can be observed. The n curves are decreasing with increasing wavelengths. A maximum for k is observable before the values are also decreasing with increasing wavelength for all obtained data curves of different intercalated amounts of x. In general, the refractive index n is decreasing with higher amounts of x incorporated in the structure of V 2 O 5 and the absorption coefficient k is increased. This trend can be seen more clearly by plotting the n and k values against the x values of H x V 2 O 5 for different wavelengths (cf. Figure 10). There, a nearly linear behavior of both n and k is clearly visible. For comparison with literature values of e. g. lithiated V 2 O 5 , it is helpful to determine ~n/ ~x and ~k/ ~x at a distinguished wavelength. According to ~n/ ~x is 1 and ~k/ ~x is 1.43 for Li x V 2 O 5 at a wavelength of 1550 nm. For our investigated system H x V 2 O 5 ~n/ ~x is round about 0.93 and ~k/ ~x round about 0.15 at the same wavelength. Comparing both systems, the change in n is a little bit lower for H x V 2 O 5 , but in the same order of magnitude, whereas the change in k is smaller for H x V 2 O 5 . ## Conclusions V 2 O 5 thin films were successfully prepared with dc magnetron sputter deposition and following annealing. It has been proven that V 2 O 5 is an excellent material for electrochemical proton intercalation and that the system H x V 2 O 5 can be used as an electrochromic cathodic material in the near-IR region. The reflectivity and therefore the n and k values of the material can be influenced systematically with proton insertion. The behavior of H x V 2 O 5 is comparable with Li x V 2 O 5 although the absolute change of the optical constants, especially the absorption coefficient k, is smaller with proton intercalation than with lithiation. However, one advantage over Li x V 2 O 5 is that it is easier to handle in non-inert environments, as there is no need for moisture-sensitive materials like metallic lithium. To conclude, H x V 2 O 5 is a material which is excellent for use in future optical devices.

chemsum

{"title": "Electrochemical Proton Intercalation in Vanadium Pentoxide Thin Films and its Electrochromic Behavior in the near\u2010IR Region", "journal": "Chemistry Open"}

programming_hydrogel_with_classical_conditioning_algorithm

## Abstract: Living systems are essentially out of equilibrium, where concentration gradients are kinetically controlled by reaction networks that provide spatial recognitions for biological functions. They have inspired life-like systems using supramolecular dynamic materials and systems chemistry. Upon pursuing ever more complex life-inspired systems, mimicking the ability to learn would be of great interest to be implemented in artificial materials. We demonstrate a soft hydrogel model system that is programmed to algorithmically mimic some of the basic aspects of classical Pavlovian conditioning, the simplest form of learning, driven by the coupling between chemical and physical processes. The gel can learn to respond to a new, originally neutral, stimulus upon classical conditioning with an unconditioned stimulus. Further subtle aspects of Pavlovian conditioning, such as forgetting and spontaneous recovery of memory, are also achieved by driving the system outof-equilibrium. The present concept demonstrates a new approach towards dynamic functional materials with "life-like" properties. Biological systems have inspired biomimetic materials with fascinating properties, e.g., toughness, structural colors, catalytic activity, and superhydrophobicity 1,2 . In future, materials are foreseen to mimic ever more complex functional and responsive properties of dynamic biological systems. Various model systems for dynamic dissipative out-of-equilibrium self-assemblies have been presented . To allow a major progress towards "life-like" materials, the importance of 2 systems chemistry has been emphasized for the structural and temporal programming of the functionalities by coupled chemical reactions under out-of-equilibrium conditions . One of the most relevant biological functions deals with the concept of "learning" 12 . In its full biological form, it involves formidable complexity. However, systematic approaches have shed light on its simplest forms, i.e., habituation, sensitization, and classical conditioning 13 . Adopting a still more reductionist approach, a generic question could be posed: whether even inanimate materials could be designed to show programmed responses mimicking some elementary aspects of learning, in resemblance to systems based on synaptic electronics or biochemical circuits . To address the above question, we explore whether artificial materials could be designed to show responses that algorithmically mimic the classical (Pavlovian) conditioning 18,19 . In Pavlov's seminal experiment, an unconditioned dog salivates (unconditioned response, UR) upon seeing food (unconditioned stimulus, US), while ringing a bell (neutral stimulus, NS) does not lead to salivation. However, upon conditioning by simultaneously ringing the bell and showing the food, a conditioned response (CR) to the neutral signal is associatively learned, after which salivation can be triggered also by the bell. The process involves the association of the two stimuli stored in the memory, so that the original response can be triggered by a new stimulus. Hydrogels have been shown to be relevant model systems for out-of-equilibrium and systems chemistry 5,9 . Here we introduce a hydrogel that is programmed to mimic classical conditioning, which allows melting of the gel in response to an originally neutral stimulus (light) upon associating the light with an unconditioned stimulus (heat). Further aspects of the classical conditioning, such as forgetting and spontaneous recovery of memory after extinction, were 3 mimicked by temporally programming the gel response using coupled chemical reactions, which drives the hydrogel out-of-equilibrium. ## Results and discussion Programming hydrogel with classical conditioning algorithm. The hydrogel consists of pHsensitive gold nanoparticles (AuNPs) embedded in a soft hydrogel network of agarose, and a merocyanine-based photoacid (Fig. 1a). The photoacid allows reversible photo-switching of the solution pH between 5.4 and 3.8 in the aqueous solution (0.2 mM) upon irradiation in the visible range (Supplementary Fig. 1) 20,21 . The AuNPs are modified by lipoic acid chosen to respond to pH changes caused by the photoacid 22 . The composition of the system, such as the size of AuNPs and the gel concentration, has been optimized for fast response under mild conditions (Supplementary Fig. 2-6). The Pavlovian response of the hydrogel is shown in Fig. 1 b-e. Heating (US) above the melting temperature (Tm ~ 33 °C, Supplementary Fig. 7) induces gel melting as the unconditioned response (Fig. 1b), whereas the AuNPs remain well dispersed as confirmed by TEM imaging and UV-Vis measurement. A clear plasmonic band can be seen at 520 nm before and after heating (Fig. 1b, right panel). This is due to the sufficient electrostatic stabilization of the carboxyl groups at pH above 5, since heating doesn't affect the pH of the gel significantly (Supplementary Fig. 1). As the neutral stimulus we used laser irradiation at 635 nm (140 mW cm -2 ), mixed with 455 nm LED light (25 mW cm -2 ). The photoacid absorbs strongly in the range between 380 nm and 460 nm, so that efficient photoacid activation can be achieved by the LED light. Besides, the intensity of the LED is chosen to be just sufficient for the activation of photoacid, in order to avoid contributing significantly to the photothermal heating inside the gel. Under irradiation, the original unconditioned gel does not melt due to the low absorption at 635 nm and thus insufficient heating (Fig. 1c, Supplementary Fig. 8). Note that in this case the interparticle electrostatic repulsion is in 4 fact also reduced upon the pH decrease, but the plasmonic band remains at 520 nm, suggesting dispersed particles due to the stabilizing gel matrix. The presence of the gel network hinders the diffusion of the AuNPs, and thus no significant self-assembly into chains takes place even when the photoacid is activated. The crucial step to achieve conditioning is the self-assembly of the AuNPs by simultaneous exposure to light and heat (Fig. 1d). Once the gel melts upon heating, the AuNPs regain mobility and self-assemble into linear aggregates triggered by the irradiation-induced pH change (Supplementary Fig. 9, 10). The formation of linear assemblies of AuNPs is dependent on different parameters such as ligand composition, pH, solvent, or salt 23 . In our system, the use of lipoid acid is important to achieve the linear self-assembly. It has been proposed that anisotropic electrostatic repulsion of AuNP dimers formed in the initial aggregation stage accounts for the rather linear configuration of the aggregates 24 . In the gel, the linear self-assemblies of AuNPs remain stable even after the pH recovery (light off) and re-gelation of the agarose (Fig. 1d), which we attribute mainly to interparticle hydrogen bonding and/or van der Waals attraction 25 . The self-assembled AuNPs can be only separated by increasing the pH of the solution to above 8, as we show in the later sections. The resulting spectral change of self-assembly is the appearance of a new plasmonic band around 635 nm due to the coupling of the AuNPs in the linear self-assemblies 26 . Note that non-specific aggregation would not result in a defined new band at longer wavelengths. This spectral change serves as the memory, which enables significantly enhanced photothermal heating at 635 nm due to the thermoplasmonic properties of AuNPs 27,28 , and the gel thus melts upon irradiation (Fig. 1d,e). Hence the system has evolved to a new state, where upon conditioning the AuNPs are self-assembled from individual particles into chains, and the gel melting occurs upon the newly learned stimulus, i.e. light irradiation (CS). For the generalization of the Pavlovian concept, Fig. 2 suggests the underlying logic circuit for the Pavlovian gel. In order to "learn", the material must possess a memory module that can be triggered by external stimuli, and a read-out mechanism that modifies the behavior upon switching of the memory 16 . In the gel, the memory is the spectral change due to the linear self-assembly of AuNPs that can be switched on by conditioning ("learning" AND gate). This AND gate is achieved by the incorporation of the AuNPs/photoacid pair into the gel network, so that the self-assembly of the AuNPs is only possible when light and heat are both present. This ensures that learning is exclusively based on the association of two stimuli. The OR gate is necessary to sustain the memory once it has been switched on, which is accomplished by the stable self-assembly of AuNPs in the gel. On the other hand, the "recalling" AND gate is achieved by the photothermal effect due to the coupled surface plasmon resonance of the AuNPs, through which the material is able to respond to irradiation. Importantly, the Pavlovian gel can be distinguished from the conventional shape memory materials, where the memory is the equilibrium permanent shape that can be recovered from a temporary shape in the kinetically trapped non-equilibrium state. Shape memory materials do not really "learn" to respond to a new stimulus. Timing dependence of the conditioning process. In biological systems undergoing Pavlovian conditioning, the efficiency of learning is highly dependent on the timing between applied stimuli. US and NS may take place simultaneously, the NS may precede the US, or the US may precede the NS, denoted as simultaneous, forward, or backward conditioning, respectively. Forward and simultaneous conditioning are the fastest, while backward conditioning is less effective or even inhibitory 29 . This is presumably because that the NS no longer predicts the appearance of US in the case of backward conditioning, where the NS is applied after the US 29 , so that such an association will not be beneficial to the organism. The association process in the Pavlovian gel shown in Fig. 3a-c is in line with such observations. When the irradiation (NS) precedes or coincides with the onset of heat (US), the "learning efficiency", as manifested by the increase in absorbance at 635 nm of the gel, is comparable in forward and simultaneous conditioning (Fig. 3a,b). In contrast, backward conditioning is less effective, and the increase of absorbance is roughly 70% of that resulting from the simultaneous conditioning process (Fig. 3c). This can be attributed to the viscosity increase of the solution upon removal of the heat 30 , which slows down the diffusion and thus the self-assembly of the AuNPs. Consequently, after backward conditioning the gel does not melt upon irradiation, since the temperature stays below the Tm of the gel, though the photothermal effect is stronger compared to the unconditioned gel (Supplementary Fig. 11). The ability to differentiate the temporal relationship between the unconditioned and neutral stimuli is intriguing, reminiscent to the timing dependence of classical conditioning 29 . Forgetting and spontaneous recovery of the memory using out-of-equilibrium processes. The above results allow to mimic some aspects of the classical conditioning in equilibrium state, where the memory stays unchanged after conditioning. In a step further, we expect that programming the time domain of the response and potentially driving the system out-of-equilibrium are needed to mimic more subtle aspects. The classical conditioning may involve several stages, such as acquisition, extinction, and spontaneous recovery 18,31 . In Fig. 4, we show the possibility of temporally programming the memory of the Pavlovian gel to achieve "forgetting" and "spontaneous recovery" using coupled chemical reactions. In Fig. 4a-c, the gel contains additional 20 mM of urea and 5 µg mL -1 of urease as an internal "clock" to trigger the forgetting process. The urease catalyzes the hydrolysis of urea, resulting in the production of 2 eq. of ammonia and 1 eq. of carbon dioxide. The temporal profile of the solution pH can thus be programmed depending on the ratio and concentration of the two components 32 . In the gel, the pH remains almost unaffected by the presence of urea/urease on the time scale of conditioning (~ 10 min), which thus enables acquisition of the memory. Yet the conditioned gel is not in the equilibrium state. The forgetting process takes place as the result of urea hydrolysis, which slowly increases the pH to around 8.8 in 12 hours. The gel was left in the liquid state to facilitate the disassembly of the AuNPs triggered by the pH change, resulting in the drop of absorbance (memory) as shown in Fig. 4b,c. The high pH required for the disassembly could be the result of interparticle van-der-Waals attraction / hydrogen bonding that has to be counter-balanced by strong electrostatic repulsion from the deprotonation of the carboxyl groups on the AuNP surfaces. The recovery of the absorbance at 635 nm is significant, indicating that the AuNPs are well protected by the ligands during the selfassembly process. In addition, the absorbance around 450 nm dropped as a result of the deprotonation of photoacid at high pH. The forgetting profile is reminiscent of the Ebbinghaus' forgetting curve: the memory decreases with time, yet a small portion of it is retained over an extended period 33 . As a result, the gel no longer melts upon irradiation (Supplementary Fig. 11), and can be considered as having forgotten the conditioning. The conditioned memory can also be extinguished by an external stimulus followed by a spontaneous recovery (Fig. 4d-f). The extinction is induced by a chemical cue containing a potassium phosphate buffer (K3PO4) and methyl formate, offering a new possibility to control the memory inspired by the extinction process in biological systems, which is triggered by repeated exposure to the NS without US. After addition of the chemical cue to the conditioned gel, the pH of the melted gel increases instantaneously to 11.8 due to the buffer (20 mM), which leads to disassembly of the AuNPs and thus fast extinction of the memory. Subsequently, the spontaneous hydrolysis of the methyl formate (240 mM) results in the formation of formic acid and thus a controlled decrease of the pH to below 5.0 in 20 hours 34 . As the pH decreases, self-assembly of the AuNPs again takes place spontaneously, where the gel was kept in liquid state to allow diffusion of the AuNPs. The absorbance at 635 nm therefore gradually recovers after the extinction (Fig. 4e). The decrease of the absorbance around 530 nm is due to the protonation of the photoacid during the recovery. This kinetically controlled pH change using phosphate buffer and methyl formate thus enables the extinction of the memory and the following spontaneous recovery. After extinction, the gel does not respond to irradiation, but subsequently regains the ability to melt upon irradiation once recovered (Supplementary Fig. 11). The memory processes shown in Fig. 4 are essentially out-of-equilibrium, in contrast to the equilibrium "learning" demonstrated in Figs. 1-3, where the memory remains unchanged after conditioning. The acquired memory through conditioning is only in a temporally stable state in Fig. 4a-c, which gradually decays with time due to the kinetically controlled chemical reaction. The system reaches equilibrium only after the memory is mostly lost (disassembly of the AuNPs). The same applies to the process in Fig. 4d-f, where the memory is temporarily extinguished and then recovers spontaneously. These results demonstrate the possibility to further manipulate the (primitive) memory of the gel system, inspired by real cognitive processes that operates under outof-equilibrium conditions. ## Conclusion Summarizing, we have shown that an inanimate soft hydrogel can be designed to exhibit responses resembling classical conditioning, which has been considered as one of the elementary forms of learning. Therein, the material learns to respond to an initially neutral stimulus (light) through an associative conditioning process, during which the material is exposed to both neutral (light) and unconditioned (heat) stimuli. Forgetting and recovery of memory can be achieved by introducing kinetically controlled coupled chemical reactions to the system, which can be discussed in the context of systems chemistry and out-of-equilibrium processes. The Pavloviantype sol-gel transition in hydrogels may find applications, e.g., in intelligent drug delivery or cell culture, and the concept may be extended to other material systems beyond gels following the demonstrated logic flow diagram, based on other functional groups and fields, e.g., using magnetic fields. Admittedly, the content of learning in the demonstrated Pavlovian hydrogel is prescribed to preselected stimuli when compared to the complex adaptive behavior in biological systems capable of responding to a wide variety of stimuli 35 . Yet our systems offer selectivity towards stimuli, and allow considerable flexibility for new properties and functions to be engineered (e.g., forgetting/recovery). The possibility to program Pavlovian conditioning in a hydrogel, even including forward and backward conditioning, as well as forgetting/recovery under out-ofequilibrium conditions, is conceptually intriguing. We envision that designing complex conditioning behaviors coupled with engineered physical properties of materials may provide

chemsum

{"title": "Programming hydrogel with classical conditioning algorithm", "journal": "ChemRxiv"}

sorbent_track:_quantitative_monitoring_of_adsorbed_vocs_under_in-situ_plasma_exposure

## Abstract: Sorbent-TRACK is a new device developed to monitor adsorption and surface oxidation of pollutants under direct plasma exposure. It is based on direct transmitted Fourier Transformed Infrared (FTIR) spectroscopy. A pyrex reactor under controlled gas pressure and composition is inserted on the infrared beam of a commercially available Nicolet 5700 FTIR spectrometer. A substrate holder is located on the optical path of the infrared beam. A thin pellet of a dedicated catalyst (CeO 2 in the present work) is inserted in a substrate holder and can be exposed to direct plasma treatment using a Dielectric Barrier Discharge. The time resolution of Sorbent-TRACK is limited by the time resolution of the Nicolet 5700 FTIR spectrometer and close to 30 s. The dynamic of the adsorption and plasma oxidation of acetone and isopropanol on CeO 2 are studied and intermediates are monitored. Performances and sensitivity of Sorbent-TRACK are reported Adsorption and oxidation of acetone leads to production of adsorbed isobutene and acetic acid, where oxidation of isopropanol gives mainly to adsorbed acetone, mesityl oxide and acetate. An increase of the plasma power leads to an increase of the isopropanol and acetone oxidation rate and a related increase of the production of adsorbed intermediates.The combination of a Non-Thermal Plasma with a catalyst or sorbent has been studied for many years for various applications such as exhaust gas purification (soot oxidation and NO x reduction) 1 , Volatile Organic Compounds (VOC) removal for indoor air purification 2 and more recently for CO 2 valorization 3,4 .Indoor air purification is a major health issue as well as a rising market. In 1983, World Health Organization has defined Sick building syndrome (SBS), usually reported by occupants in certain buildings or specific rooms 5,6 . Sources located inside the building, such as adhesives, carpeting, wood products and cleaning products may emit VOCs 7 . Techniques for VOCs control in exhaust air streams include such as adsorption 8 . thermal and catalytic oxidation 9 , and photocatalysis 10 . Such methods may be cost-inefficient and difficult to operate when low concentrations of VOCs need to be treated in indoor air, typically lower than 1 ppm 8 .Non-thermal plasma (NTP) is an effective way to produce oxidative species in air, at ambient temperature and at a low energy cost [11][12][13] . Energetic electrons generated in non-thermal plasmas can collide with carrier gases, forming highly reactive species such as free radicals and excited atoms, molecules and ions. When combined with a catalytic sorbent, a NTP triggers some surface oxidative reaction of adsorbed VOCs. In some cases, synergetic effects are observed in the sense that, for example, the removal rates obtained with a plasma-catalytic setup are higher than those predicted by simply adding the effects of plasma and catalyst 2,14 . The combination of NTP with heterogeneous catalysts can be divided into two categories depending on the location of the catalyst: in-situ plasma catalysis (IPC) and post-situ plasma catalysis (PPC). The latter is a two-stage process where the catalyst is located downstream of the plasma reactor, while the former is a single-stage process with the catalyst being exposed to the active plasma.Numerous studies have been dedicated to studying the performances of Plasma-Catalyst coupling. However, to the day, most studies have focused on the gas phase analysis; the destruction of an injected VOC is reported and often a carbon balance is calculated. Only few articles have reported a dynamic monitoring of species adsorbed by infrared on the catalytic surface under plasma exposure (IPC 15,16 or PPC 17,18 ).Infrared spectroscopy (IR) is for that purpose very appropriate since it is fast and sensitive. In a recent paper Barakat et al. 18 investigated the evolution of adsorbed phase species by post-situ plasma regeneration (PPC) using Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS). It has been proven that the complete oxidation of isopropanol and acetone is mainly limited by the acetone oxidation rate. DRIFT spectroscopy was coupled to conventional gas phase analysis using FTIR. A reaction scheme was proposed for explaining the formation of those adsorbed intermediates and gaseous products. Similarly, Sauce et al. 17 studied acetaldehyde catalytic ozonation in the post-situ plasma-catalysis process 19 in which, an acetaldehyde saturated Ag/TiO 2 /SiO 2 surface 20 was brought into contact with a NTP. These studies show that combining monitoring of the gas phase and of the surface greatly improves the understanding of surface oxidation mechanisms. Two recent papers have monitored adsorbed species in in-plasma configuration PPC: Rivallan et al. 16 process by operando FTIR spectroscopy when the catalyst is in direct contact with a Dielectric Barrier Discharge (DBD-IR reactor). Measurements were combined with a DBD on an Al 2 O 3 surface to enable in situ and time resolved monitoring of the oxidation of pre-adsorbed IPA. Although no quantitative concentration could be given, very interesting oxidative mechanisms could be deduced. Stere et al. 15 reported a newly developed DRIFT-MS system for the investigation of non-thermal plasma (NTP) assisted heterogeneously catalyzed reactions, when a plasma jet is used in contact with a Ag/Al 2 O 3 catalytic surface. The results provided further evidence of the role of NTP in promoting the performance of the Ag catalyst at low reaction temperature. These works emphasize that in situ monitoring developments are possible and necessary. However, so far, quantitative monitoring of the surface has not yet been published in the case of the In-plasma configuration, when the plasma is in direct contact with the sorbent/catalytic surface. A newly developed system Sorbent-TRACK dedicated to quantitative analysis of adsorbed species on various sorbents under direct plasma exposure (IPC). In the present paper, test pollutants are isopropanol (IPA) and acetone. The catalyst is CeO 2 which is considered as an effective promoter in thermal catalytic reactions due to its high oxygen storage capacities and redox properties between Ce 4+ and Ce 3+ . CeO 2 can also act as a local source/ sink of oxygen species for due to its high bulk oxygen mobility and oxygen vacancies 21 . In the present work, a two-step protocol is followed to investigate surface reactions: i) First, the adsorption of the test pollutants on CeO 2 is monitored without plasma; reactive adsorption products may also be identified. ii) then, the gas phase inlet of the test pollutants is turned off and the plasma is turned on; NTP oxidation of the test pollutants adsorbed on ceria is monitored as well as adsorbed oxidation intermediates. ## Experimental Set Up Sorbent-TRACK cell. The Sorbent-TRACK cell as shown in Fig. 1a consists of a cylinder of glass carrying a toroidal sample holder in its center, where the catalyst is placed in the form of a self-supported wafer. The heating system guarantees a maximum temperature of 573 K on the sample. The tightness can be obtained by using Kalrez O-rings between the terminal KBr windows and the extremities of the cell. The plasma reactor consists of a Pyrex glass tube of 5.8 mm inner diameter, 8 mm outside diameter. The inner electrode (H.V.) consists of a copper wire of 30 μ m and the outer electrode is a copper wire of 30 μ m placed in the center of sample holder. The electrical parameters (U a , U m ) were measured via two high voltage probe (LeCroy, PPE20KV-CC) connected to a digital oscilloscope (LeCroy WaveSurfer 64Xs-A, 600 MHz). A measurement capacitance (Cm) of 680 pF is placed in series with the DBD reactors. The equivalent electrical scheme is represented in Fig. 1b. Catalytic materials and gas flow set-up. Catalytic CeO 2 powder with a specific surface of 84.41 ± 5.32 m 2 /g was pressed into self-supported wafers (Ø = 7 mm, m ∼ 60 mg cm −2 . thickness of 0.3 mm); experiments were carried after activation of the pellet at 473 K for 2 h and then cooling down to room temperature. The VOCs in this study are 2-Propanol (34959 -CHROMASOLV ® , for HPLC, absolute, 99.9%) and acetone (270725 -CHROMASOLV ® , for HPLC, ≥ 99.9%) both prepared by Sigma-Aldrich. Certified gas cylinders are supplied by Air Liquide. The regulation of the gas flow was insured using Brooks mass flow controllers. Synthetic air was used to prepare the carrier gas flow. The Sorbent-TRACK cell is connected to a flow set-up, as shown in Fig. 2. Gases are introduced into the lines by mass flow controllers. A two way valve allows the selection of either air or N 2 as the main carrier gas. The flow is then divided, into two gas lines, one of which is the main flow and the other through the cryostat containing the liquid VOC. The flowmeter attributed to the main gas flow can go up to 2000 mL/min (± 1%, i.e. is less accurate below an imposed flow of 20 mL/min) while the one used for VOC dilution has a maximum of 10 mL/min and can be used accurately up to 0.1 mL/min. The temperature of cryostat and the flow through the cryostat were respectively 0 °C and 5 ml/min for IPA and − 0 °C and 3.9 ml/min for acetone. The main flow was adjusted to have a total flow of 500 ml/min. Spectra acquisition using FTIR spectroscopy. IR spectra were collected with a Nicolet 5700 FTIR spectrometer equipped with a MCT detector. The spectra have been treated by the Nicolet OMNIC software. Two spectra per minute are collected with Omnic software with 16 scans per spectrum and a spectral resolution of 0.5 cm −1 . For calibration curves and quantification, TQ Analyst 8 from thermo scientific was used. Spectra of species adsorbed on the catalytic samples are acquired as follow: the infrared beam emitted by the FTIR source is collimated through the entrance window of Sorbent-TRACK, propagates through the catalytic sample, and exits though the exit window to be detected on the MCT detector. Alternatively, Sorbent-TRACK may be replaced by a 10 m optical-path White cell to calibrate the concentration of VOCs in the gas phase. In these conditions (no catalytic samples), the detection limits of this analytical tool have been determined as two times the signal/noise ratio in the region of interest and are: 80 ppb for acetone, 90 ppb for IPA, 20 ppb for CO 2 , 10 ppb for CO and 15 ppb for O 3 . ## Plasma generation. The injected power is obtained by the Lissajous figures 22 , corresponding to the plotting of the transported electric charge through the discharge as a function of the applied periodical voltage. Experimentally, the charge is delivered from the voltage drop across the reactor and the average electric energy dissipated in a discharge cycle is the area of the characteristic Lissajous figure. Figure 3 shows such Lissajous figures obtained by applying a voltage of 15 kV, 9 kV and 5 kV respectively and a frequency of 50 Hz in a typical DBD reactor (Fig. 1c,d, respectively off and on) under air flow by means of a sinusoidal power supply. Figure 4 shows the injected power as a function of applied voltage and constant frequency (50 Hz) under dry air flow at 1 bar. ## Results and Discussions Quantitative measurement of adsorbed species using Sorbent track. We will estimate the quantity of adsorbed IPA and acetone using the Beer's Law. In the solid or gaseous state, the Beer's law express as: where ε is the absorption coefficient, L the pathlength, and C the volumic concentration of the sample (mol/L). Acetone and IPA absorbance are first measured in the gas phase using a 10 m White cell in order to correlate absorbance and absolute number of absorbing molecule on the optical path. Calibration curves were obtained by passing the standard gases in air at different known concentrations, through the 10 m White gas cell. The fitting function is shown in Fig. 5. The spectral region selected for acetone calibration is 1207-1197 cm −1 . A similar calibration is made for IPA in the 992-944 cm −1 spectral region. Experimentally, we find Where σ acetone and σ IPA are the LxC product expressed in mol.cm −2 . In the following, absolute values of adsorbed concentration will be estimated for IPA and acetone. Acetone adsorption and in situ plasma oxidation study. Adsorption of acetone. 200 ppm of acetone diluted in air is sent into the Sorbent-TRACK system, with a total flow rate of 500 mL/min. In Sorbent-TRACK, acetone molecules may be located on the catalyst and in the gas phase. Hence the absorbance is: σ S acetone + σ Gacetone are the optical depth of acetone in the solid and gaseous phase respectively. In order to evaluate the contribution of the gas phase molecules to the Absorbance in Sorbent-TRACK, the dynamic of absorbance is recorded. Figure 6 shows the infrared spectra of the CeO 2 surface upon acetone adsorption. Spectra are recorded in the range 3000-1200 cm −1 and the clean CeO 2 spectrum is subtracted to the spectra collected during adsorption. At t = 1 min, as shown in Fig. 6, there is a new band at 1699 cm −1 . The volume of sorbent-TRACK being 460 ml, the filling time is about 1 min. Hence the spectrum recorded at 1 min by sorbent track is mainly in gas phase. The contribution of gas phase of acetone is about 3% of acetone concentration at 96 min. In the following, the concentration of adsorbed acetone has been corrected from the gas phase contribution. At t = 96 min, CeO 2 sample is saturated by acetone. The bands at 2971, 2926, 1699, 1365, and 1236 cm −1 are assigned to adsorbed acetone on the CeO 2 particle surface 23 . These bands are respectively assigned to ν s-CH3 , ν as-CH3 , ν C=O , δ s-CH3 and ν C-C vibration modes of molecularly adsorbed acetone. The bands at 1628 and 1423 cm −1 , are assigned to v C=O and δ CH vibrations in − CH 2-C=O groups of diacetone alcohol-like species 24 . Furthermore, the bands at 1574 cm −1 along with the bands at 1554 cm −1 are characteristic of the ν C=O and ν C=C vibration modes of mesityl oxide. The different steps of acetone adsorption on CeO 2 are summarized below 24 as shown in Fig. 7. This stage correspond the sorbent track system filling and is dominant by gas phase acetone. From 1 min, the acetone surface coverage rate increases while mesityl oxide is still not yet product. As soon as the surface coverage of acetone reaches 0.29 (42 μ mol/g), the production of mesityl oxide starts. This behavior suggests that mesityl oxide formation is controlled by a threshold regarding acetone surface coverage. The same behavior is noticed by El Maazawi et al. 25 using the Langmuir adsorption technique and Barakat et al. 18 by DRIFT on TiO 2 surface. They report a threshold acetone surface coverage of 0.3 and 0.35 respectively, which is consistent with our result. In the adsorption step, the total amount of adsorbed acetone reached 145 μ mol/g. When the pollutant is removed, the catalyst is flushed by dry air to remove the reversibly adsorbed acetone which is about 40 μ mol/g. The value of acetone at the end of flushing step give the amount of acetone irreversibly adsorbed of 105 μ mol/g. Adsorbed acetone oxidation by in-situ plasma. The infrared spectra obtained upon in-situ plasma exposure of acetone saturated CeO 2 surface are shown in Fig. 8 (applied voltage 14 kV, frequency 50 Hz, injected power 133 mW). Exposing the surface to plasma induced species results in a decrease in the intensities of acetone (1699 cm −1 (ν C=O )) and of mesityl oxide (1574 cm −1 ν C=O and 1554 cm −1 ν C = C ), related bands which is mentioned above. Furthermore, the adsorption peaks at 1540, 1440, 1386 and 1305 cm −1 are related to the acetate species in the spectra 26 . Whereas those at 1540 and 1440 cm −1 are assignable to antisymmetric and symmetric ν COO− , the adsorption at 1386 and 1305 cm −1 are due to the δ CH3 vibrations and the adsorption at 1467 and 1421 cm −1 are due to the δ CH vibrations. On the other hand, the absorptions at 1762 and 1300 cm −1 can account for ν C=O and δ OH of AcOH molecules. The components at 1590, 1456 and 1355 cm −1 suggest the presence of isobutene which is also observed by Rivallin et al. 16 . They propose that acetone decomposition only occurs after its aldolization on γ -Al 2 O 3 into mesityl oxide which fragments into acetaldehyde and isobutene. However, in our case, the formation of acetate surface species is rather favorable. Indeed, the acid-base pair sites (M n+ − O 2− ) is available on CeO 2 , but not on Al 2 O 3 . Zaki et al. 27 indicated that strong base sites are necessary for the enolate formation ( scheme 1), whereas strong lewis acide sites are essential to stabilize the reaction intermediates. Coordinated acetone molecules may also be activated for a nucleophilic attack on α -carbons, leading to the splitting of a methyl group in form of CH 4,g and formation of acetate surface species 26 . or can be activated for a bi-molecular reaction 28 . Acetic acid is believed to be formed by the following reaction: The temporal evolution of the peaks characteristics of acetone and mesityl oxide and its decomposition products isobutene and acetate acid (Fig. 9) are followed during plasma treatment under pure air at different injected powers. The acetone peak intensities are normalized with respect to the highest intensity of the corresponding experiment (i.e. the intensity of acetone before turning on the plasma) and mesityl oxide, isobutene and acetic acid are reported in arbitrary units. The peaks followed for the species evolution are 1699 cm −1 for acetone, 1554 cm −1 for mesityl oxide, 1726 cm −1 for acetic acid and 1590 cm −1 for isobutene. Data presented in Fig. 9 lead to the following observations: (i) The acetone and mesityl oxide consumption rate increases with increasing injected power (Fig. 9a,b). (ii) The rate of formation of all oxidation products (acetic acid and isopropanol) also increases with increasing injected power (Fig. 9c,d). (iii) Acetic acid initially accumulates on the surface but is further slightly oxidized as its surface coverage decreases with plasma exposure (Fig. 9c). (iv) The trend of acetic acid and isobutene on the CeO 2 surface are not similar (Fig. 9c,d). Points (i) and (ii) imply that the evolutions of all the adsorbed species are dependent on injected power. In addition, points (iii) and (iv) could imply that the formation reaction pathway is not the same where isobutene is mainly formed by reaction 2 and acetic acid results from reaction 1, 2 and 3. ## Isopropanol adsorption and in situ plasma oxidation. Adsorption of isopropanol. Similarly to what is discussed in the previous section, during adsorption process, IPA molecules are in gas phase and adsorbed phase. The related absorbance is therefore: (the gas phase contribution is about 5% according to the method described in section of acetone). Figure 10 shows that IR spectra recorded by sorbent track during IPA adsorption on CeO 2 . The introduction of IPA in the system leads to the appearance of several new absorption bands located at 3342 (broad), 2980, 2930, 2880, 1469, 1420, 1388, 1328, 1279, 1237, 1159, 1125 and 981 cm −1 . Generally, the bands appearing at 2970, 2930, 2880 cm −1 correspond to C-H (ν CH3 ) symmetric and asymmetric stretch modes of the different adsorbed IPA species. These bands are accompanied by symmetric C-H bends (δ as CH3 ) at 1469 cm −1 , anti-symmetric C-H bends (δ as CH3 ) at 1328 cm −1 . The skeletal C-C vibration appears at 1237 cm −1 . Similarly to TiO 2 18 , the bands at 1159 and 1125 cm −1 show two distinct δ C-O vibrations, respectively representative of a dissociative and non dissociative IPA adsorption on CeO 2 . These two adsorption modes are, associated with surface isopropoxide ions, whereas the 981 cm −1 band, strongest, originates from intact IPA molecules hydrogen-bonded to the surface. The dissociative adsorption is also supported by the production of OH groups in the high wavenumber range. Indeed, a large band with a maximum at 3342 cm −1 corresponding to the OH stretch (ν OH ) of interacting hydroxyl groups increases with IPA breakthrough on the surface while the bands at 1388 cm −1 also indicates strongly non dissociated IPA. In addition to what was observed on CeO 2 , new peaks appear at 1699, 1590 and 1556 cm −1 . These peaks are attributed to acetone and its condensation products which have been discussed above. Figure 11 presents the quantitative evolution of IPA and acetone during the adsorption of 100 ppm of IPA followed by flushing under air. The surface is saturated by IPA (166 μ mol/g) from 20 min while the quantity of adsorbed acetone still increases. As the site for IPA is saturated, the new generated acetone should be adsorbed by a different site than IPA. Arsac et al. evidenced 29 that an acetone molecule formed by the photocatalytic oxidation of IPA adsorbed on a "S 1 " site cannot remain adsorbed on the TiO 2 surface: it must either desorb rapidly as gaseous acetone or diffuse on the surface to be adsorbed on "S 2 " sites, specific to acetone, or on the "S 1 " sites liberated by the removal of IPA surface species during oxidation. The formation of mesityl oxide is not observed because the adsorbed acetone concentration is still low, about 7 μ mol/g. It was shown in Fig. 7 that as soon as the surface coverage of acetone reaches 0.29 (45 μ mol/g), the production of mesityl oxide starts. The flushing step results in the evacuation of IPA, which amounts to 24 μ mol/g. NTP oxidation of isopropanol. The infrared spectra obtained upon in-situ plasma exposure of IPA saturated CeO 2 surface are shown in Fig. 12 (applied voltage 9 kV, frequency 50 Hz, injected power 32 mW). Exposing the surface to the plasma results in a increase in the intensities at 1696, 1367 and 1235 cm −1 corresponding to the acetone formation on CeO 2 , concomitantly with disappearance of the previous IR features of the IPA (C-H vibrations (ν C-H ) in the high wavenumber range and at δ C-O 1159 and 1125 cm −1 ). On the other hand, other broad components appear at 1670-1550 should be contributed to the acetone condensation at CeO 2 surface, as discussed in section of acetone. The evolution of IPA and its oxidation products, acetone, mesityl oxide and acetate apparent surface coverage during the in-situ plasma exposure is plotted as a function of treatment time in Fig. 13. The initial and rapid consumption of IPA must result from its direct reaction with the oxidative peroxide species induced by plasma to yield acetone and its condensation product, mesityl oxide. Recall from Figs 10 and 11 that the adsorption of IPA on CeO 2 resulted in the formation of acetone species with the metal oxide. The same phenomenon is believed to take place under plasma exposure, whereby IPA reacts with via adsorbed atomic oxygen to give acetone. In parallel, strongly adsorbed acetone, mesityl oxide and acetates gradually accumulate on the surface at the first 20 min. The surface coverage of the three adsorbed oxidation intermediates decreases with treatment time, implying that the species are oxidized. Adsorbed acetone concentration increases during the 70 min experiments. It comes of course from the IPA oxidation and from the creation of free adsorption sites for acetone during the process. The acetone accumulation is related to the mesityl oxide and acetate decomposition rate and should be the limiting step for IPA decomposition. ## Conclusion Operando IR spectroscopy was found to be an ideal technique for studying plasma catalytic coupling. Coupling this technique with in-situ plasma allowed real-time monitoring of both the pollutant adsorbed on the surface of the catalyst and the by-products produced. The sorbent-TRACK system made in our laboratory provides access to the activity, selectivity and mechanism of the process and could provide a quantitative analysis. To demonstrate the reliability of this technique, two VOCs were studied in this work: acetone and isopropanol. The dynamic of adsorption of IPA or acetone on ceria followed by their oxidation under plasma exposure has been successfully monitored. Direct adsorption capability was estimated using Trans-FTIR analysis. It was seen that on CeO 2 , Acetone initially adsorbs molecularly on Lewis acid sites (Ce 4+ ) of the metal oxide surfaces. When its surface coverage reaches ~30%, two adsorbed acetone molecules react through an aldol condensation to yield mesityl oxide. IPA adsorbs molecularly via hydrogen bonds and by heterolytic dissociation, with a proton going to a surface lattice oxygen and an alkoxide to a surface ion, where the cation acts as a Lewis acid site and the surface oxygen ion acts as a Lewis base. Both two VOC could be decomposed by in-situ plasma, acetone and the condensation species are decomposed into isobutene, acid acetic and gas phase CO 2 . The complete oxidation of the intermediate species (isobutene and acid acetic) occurs via redox reactions with the oxide surface. The oxidation of IPA on CeO 2 results in adsorbed acetone, mesityl oxides and acetates along with gas phase CO 2 .

chemsum

{"title": "Sorbent track: Quantitative monitoring of adsorbed VOCs under in-situ plasma exposure", "journal": "Scientific Reports - Nature"}

visualizing_changes_in_mitochondrial_mg²⁺during_apoptosis_with_organelle-targeted_triazol

## Abstract: Magnesium is one of the most abundant metals in cells and is essential for a wide range of cellular processes.Magnesium imbalance has been linked to a variety of diseases, but the scarcity of sensors suitable for detection of Mg 2+ with subcellular resolution has hampered the study of compartmentalization and mobilization of this ion in the context of physiological and pathological processes. We report herein a family of fluorescent probes for targeted detection of free Mg 2+ in specific intracellular organelles, and its application in the study of programmed cell death. The new sensors feature a triazole unit that plays both structural and electronic roles by serving as an attachment group for targeting moieties, and modulating a possible internal charge transfer process for ratiometric ion sensing. A probe decorated with an alkylphosphonium group was employed for the detection of mitochondrial Mg 2+ in live HeLa cells, providing the first direct observation of an increase in free Mg 2+ levels in this organelle in the early stages of Staurosporine-induced apoptosis. ## Introduction Magnesium is essential for numerous cellular processes, playing a role in activation of enzymes, structural stabilization of nucleic acids and proteins, modulation of ion channels, and as a second messenger. 1,2 In mammalian cells, Mg 2+ is the most abundant divalent cation, with a total concentration typically maintained in the mid-millimolar range in most cell types. 3,4 Abnormal levels of serum or cellular magnesium have been linked to various conditions including cardiovascular disease, diabetes, neurodegeneration, and cancer. Despite the importance of Mg 2+ homeostasis in human health, details of the mechanisms that regulate the concentration of this ion at the cellular and subcellular level have remained partially obscure, primarily due to the paucity of efficient tools for the measurement of Mg 2+ with the required spatial and temporal resolutions. 10 In particular, the ability to study intracellular ion distribution and mobilization between subcellular domains has been hampered by the scarcity of probes capable of reporting organelle-specifc levels of Mg 2+ . In this regard, Oka and coworkers developed a rosamine-based Mg 2+ turn-on indicator that spontaneously localizes to mitochondria. 11 More recently, the same group reported a related turn-on biarsenical dye that can be anchored to tetracysteine-tagged proteins expressed in specifc compartments, thus enabling the visualization of Mg 2+ dynamics upon mitochondrial membrane depolarization. 12 Genetically encoded protein-based FRET fluorescent sensors reported by Merkx and coworkers have been targeted to other intracellular compartments. 13 A general platform suitable for organelle-targeted ratiometric detection of Mg 2+ with small-molecule indicators, however, is still lacking. The activation of apoptotic pathways bears close connection with cellular homeostasis of divalent cations, with Ca 2+ playing a major role in regulation of the intrinsic (mitochondrial) pathway. The role of Mg 2+ , on the other hand, has not been clearly established. Changes in cytosolic Mg 2+ concentration have been observed in glycodeoxycolate-induced apoptosis of hepatocytes, 17 during proanthocyanidin/doxorubicin-induced apoptosis in K562/DOX cells, 18 and in Fas ligand-induced apoptosis of B lymphocytes. 19 In the latter example, an increase in cytosolic free Mg 2+ was found to be independent of the extracellular concentration of the metal, which led to the hypothesis that mitochondria could be acting as an intracellular source. Until now, however, the dynamics of mitochondrial Mg 2+ during apoptosis have not been observed directly in whole cells. In this report, we introduce a new family of fluorescent sensors for targeted ratiometric detection of Mg 2+ in organelles of interest (Fig. 1), and present the frst direct observation of the changes in free Mg 2+ levels in mitochondria during early stages of Staurosporine-induced apoptosis in HeLa cells. ## Results and discussion Sensor design and synthesis 1,2,3-Triazoles assembled by copper catalyzed alkyne-azide cycloaddition (CuAAC) 20 have been used extensively as structural linkages in fluorophore biocojugation, but only recently have their electronic features been exploited to influence the properties of fluorescent labels and sensors. 21 We envisioned a sensor design incorporating a 1,2,3-triazole moiety as part of the fluorophore, replacing the oxazole group in furaptra 22 and related 'fura' dyes. 23 The triazole is thus intended to serve a dual purpose, namely, a structural role as an attachment group between fluorophore and an organelle-targeting moiety, and a possible electronic role as a modulator of an internal charge transfer (ICT) process for fluorescence-based ion sensing. We synthesized an alkynyl-functionalized benzothiazole, 5, to be employed as a precursor for rapid assembly of targeted ratiometric sensors via CuAAC (Scheme 1). This compound was obtained from 2-aminobenzothiazole 2, which was prepared by modifcation of a protocol reported by Metten and coworkers. 24 The amino function was converted by diazotization and treatment with potassium iodide, followed by Sonogashira coupling with trimethylsilylacetylene and subsequent deprotection. The late stage click reaction with the resulting alkyne may be used to tune the chemical and biological properties of the fnal sensors with minimum synthetic effort, based on sensible choice of azide. With the goal of evaluating the performance of our sensor design, model sensors 7a and 7b were prepared by reaction of alkyne 5 with benzyl-and phenylazide, respectively, followed by ester hydrolysis. Cycloaddition was performed on the esterprotected sensors in order to minimize residual copper binding to the metal-recognition unit, which may interfere with metal sensing in subsequent studies. ## Spectroscopic properties of uorescent sensors Photophysical characterization of sensors 7a,b was conducted in aqueous buffer mimicking physiological ionic strength (Table 1). The new triazole-based probes show large Stokes shifts in aqueous solution, and respond to Mg 2+ with a signifcant blue shift in the fluorescence excitation and emission maxima (Table 1 and Fig. 2). These observations are consistent with a destabilizing effect of the cation on an excited state characterized by a large dipole moment. A similar response is observed with the related furaptra (Mag-fura-2) dye, 22 suggesting a common ICT mechanism with the nitrogen of the metalrecognition unit acting as a donor. 25 This notion is currently being investigated computationally. Both benzyl and phenyl derivatives 7a and 7b exhibit similar absorption and fluorescence emission wavelengths in their metal-free and -bound forms (Table 1), but the phenyl derivative exhibits a lower quantum yield than the benzyl derivative, well below 10%. We postulated that rotation around the triazole-phenyl bond may provide an efficient non-radiative decay pathway for 7b, via distortion of the excited state 26 and/or access to a non-emissive twisted intramolecular charge transfer state. 27 Attempts to test this hypothesis through measurements in solvents of increasing viscosity proved inconclusive. However, incorporation of sterically demanding isopropyl substituents on the ortho position of the phenyl ring (see derivative 7c, Scheme 1), which increase the barrier of rotation and disrupt a possible coplanar arrangement of phenyl and triazole rings, resulted in the recovery of the fluorescence quantum yields to values comparable to those of benzyl derivative 7a. Compounds 7a and 7c are useful for ratiometric detection of Mg 2+ (Fig. 2 and S1-S3, ESI †), with apparent dissociation constants in the low millimolar range at 25 C (K d,Mg 2+ ¼ 8.8 AE 0.4 and 9.5 AE 0.4 mM for 7a and 7c, respectively). On the other hand, the difference in brightness for the metal-free and -bound forms of phenyl derivative 7b makes it more suitable for a turnon application ($13-fold turn-on, K d,Mg 2+ ¼ 7.8 AE 0.2 mM). It is important to note that these indicators detect free Mg 2+ , and do not respond to bound forms of the ion such as MgATP. In this regard, the fluorescence response of a solution of compound 7c treated with increasing amounts of Mg 2+ in the presence of 18.4 mM ATP (Fig. S5 †) can be modelled by considering a single binding event for the complexation of Mg 2+ by the sensor. The fluorescence ratio expressed as a function of [Mg 2+ ] free , calculated from the amount of total magnesium and dissociation of MgATP (K d ¼ 50 mM (ref. 30)), matches the isotherms obtained in the absence of the ATP (Fig. S5B †). The optical properties of derivatives 7a-c were also tested in the presence of high concentrations of other biologically relevant divalent metal ions, including Ca 2+ , Mn 2+ , Fe 2+ , Co 2+ , Ni 2+ , Cu 2+ , and Zn 2+ (Fig. S6-S8 †). The metal selectivity of the new probes is comparable to that of other related o-aminophenol-N,N,O-triacetic acid (APTRA)-based metal ion indicators. 23,31 In addition to Mg 2+ , the compounds respond to midmicromolar concentrations of Ca 2+ (Table 1), thus could be employed as low-affinity Ca 2+ indicators for the study of systems with particularly high concentrations of this ion. For compounds 7a and 7c, the changes in spectral properties upon Ca 2+ coordination are similar to those observed in the presence of Mg 2+ (Fig. S9 and S11 †). For 7b, on the other hand, binding of Ca 2+ leads to a blue shift in excitation with no signifcant increase in the emission efficiency, i.e. no turn-on response is obtained (Fig. S10 †). The compounds also respond to the micromolar concentrations of Zn 2+ tested. 32 With few exceptions, however, the typical sub-nanomolar intracellular concentrations of this ion should not interfere with Mg 2+ detection. 33 Finally, the sensors are insensitive to variations in pH in the 5.5 to 8.0 range (Fig. S13 †). ## Targeted, organelle-specic sensing of free Mg 2+ With insight gained from the model compounds characterized in vitro, we focused on the design of a mitochondria-targeted sensor. Mitochondria are regarded as intracellular reservoirs of Mg 2+ , and are invoked often as central players in the regulation of Mg 2+ homeostasis due to their ability to take up and extrude this metal ion in a respiration-dependent manner. 11,34,35 Thepotential caused by the proton gradient across the mitochondrial membrane can be exploited to direct the accumulation of small-molecules to this organelle. With this feature in mind, a derivative functionalized with a lipophilic cationic alkylphosphonium group 36 (Mag-mito, Scheme 2) was prepared. This targeted sensor shows similar photophysical properties and metal response as those displayed by the analogue compound 7c, devoid of the targeting moiety (Table 1 and Fig. S4 and S12 †). a Measurements performed in 50 mM PIPES, 100 mM KCl, pH 7.0 at 25 C. Molar absorptivity coefficients, fluorescence quantum yields and dissociation constants are averages of three determinations; numbers in parenthesis represent the uncertainty on the last signifcant fgure. N.D. ¼ not determined. b Quinine sulfate in 0.5 M H 2 SO 4 (F 347 ¼ 0.546) 28,29 was employed as a fluorescence standard. Mag-mito was tested for the excitation ratiometric imaging of mitochondrial Mg 2+ in live HeLa cells by widefeld fluorescence microscopy, using flter sets available for Mag-fura-2 and the Ca 2+ -sensitive analog Fura-2 (Fig. 3). To facilitate cell loading of the compound, the metal-binding carboxylate groups were masked as acetoxymethyl (AM) esters, which are readily cleaved by intracellular esterases after probe uptake. 37 Cells were incubated with 1 mM of the sensor for 30 min at room temperature, rinsed, and then allowed to incubate for another 30 min for full de-esterifcation of the internalized probe. Successful targeting of the desired organelle was evidenced by a Pearson correlation coefficient of 0.83 in the co-localization analysis with MitoTracker green FM (Molecular Probes, Fig. 3F). 38 This analysis was conducted over the three-dimensional volume of the cell, reconstructed from a z-stacked series of images (Fig. S14 †). To the best of our knowledge, this is the frst example of targeted ratiometric detection of mitochondrial Mg 2+ with a fluorescent probe. 39 For comparison, the non-targeted analog 7c, devoid of the alkylphosphonium group, was tested under the same conditions. This sensor showed relatively unselective staining of various compartments (Fig. 3H-J), with a correlation coefficient of 0.55 for the co-localization analysis with the reference mitochondrial stain. The ability of the indicators to respond to changes in intracellular Mg 2+ concentrations was confrmed by collecting two sets of images of cells stained with non-targeted compound 7c, before and after treatment with non-fluorescent ionophore 4-bromo-A-23187 (Molecular Probes) and 20 mM of MgCl 2 for 60 min. An increase in the average fluorescence ratio per cell ($20%, Fig. 4 and S15 †) was observed in response to the increase in intracellular free Mg 2+ concentration mediated by the ionophore. Furthermore, the fluorescence excitation spectrum of 7c-loaded HeLa cells treated with ionophore and 50 mM EDTA for 30 min was acquired on a plate reader, showing a red-shift consistent with decreasing concentrations of intracellular Mg 2+ (Fig. S16 †). ## Mitochondrial changes in free Mg 2+ during apoptosis With a probe capable of detecting free Mg 2+ in mitochondria, we investigated the changes in ion levels in these organelles during apoptosis induced by Staurosporine (STS) in HeLa cells. Live cells pre-loaded with Mag-mito were treated with 1 mM of the alkaloid on the fluorescence microscope stage, and monitored over the course of 120 min (Fig. 5). MitoTracker green was employed to confrm the localization of the Mg 2+ probe and a caspase indicator was used to verify apoptosis, whereas ethidium homodimer-1 was used to rule out possible cell lysis from necrosis. Changes in the fluorescence ratio of the sensor revealed a roughly threefold increase in concentration of free Mg 2+ , which plateaued at 2.6 mM within 10 min and decreased slowly after $25 min as the process continued (Fig. 5B). Signal of the sensor and MitoTracker started to appear diffuse after approximately 40 min of observation, likely due to dye leakage upon depolarization of the mitochondrial membrane that makes the estimation of ion concentration less reliable at later points. Morphological changes associated with apoptosis such as mitochondrial fragmentation and cell blebbing were also observed. The caspase indicator became activated after $90 min, revealing the downstream events of the apoptosis cascade (Fig. S17 †). For comparison, no signifcant changes were observed in cells treated with vehicle over the same period of time, showing a basal mitochondrial level of 0.8 mM free Mg 2+ that remained constant throughout the experiment. Given the weak Ca 2+ binding ability of APTRA-based sensors, we sought to rule out possible Ca 2+ -induced signal in our experiment by comparing the fluorescence response of Magmito with that obtained with a genetically encoded Ca 2+ -specifc indicator. We conducted a similar experiment with HeLa cells transiently expressing cameleon 4mtD3cpv, which has been optimized for the detection of Ca 2+ in mitochondria. 40 The protein-based FRET indicator revealed Ca 2+ elevations in mitochondria clusters starting after 30-40 min of treatment with the drug (Fig. 5C). The clear differences in the onset and duration of the Ca 2+ signal in comparison with the response obtained by Mag-mito are consistent with the detection of Mg 2+ , and not Ca 2+ , by the small molecule probe. Another control experiment was conducted by adding tris-(2-pyridylmethyl) amine (TPA), a rapid picomolar Zn 2+ chelator, 41 15 min after induction of apoptosis. The fluorescence ratio did not show a decrease within the typical response time of the chelator, ruling out the interference of Zn 2+ in our measurement (Fig. S18 †). To the best of our knowledge, these results represent the frst direct observation of changes in mitochondrial free Mg 2+ during programmed cell death. The source of this pool of free Mg 2+ is unknown at this time, but it could be attributed to its release from bound forms abundant in the mitochondrion (e.g. MgATP), or to an extra-mitochondrial origin. Signifcantly, studies conducted with isolated mitochondria by Martinou and coworkers have shown that Mg 2+ may potentiate the release of cytochrome c from these organelles, 42 thus hinting to the possible relevance of an early increase in free Mg 2+ in the apoptotic cascade. ## Conclusions The ability to study metal compartmentalization and mobilization in cells in the context of physiological and pathological processes depends on the availability of fluorescence indicators that enable rapid detection of the ions with subcellular resolution. We have designed a new family of triazole-based fluorescent probes for targeted ratiometric detection of Mg 2+ in intracellular organelles by fluorescence microscopy. The sensors are rapidly assembled by copper catalyzed alkyne-azide cycloaddition between an alkynyl benzothiazole, functionalized with an APTRA Mg 2+ recognition unit, and an azide-functionalized organelle-targeting group of choice. The resulting triazole moiety plays both structural and electronic roles in the new sensors, by serving as an attachment group to organelle-targeting moieties and participating in a possible ICT process useful for ion sensing. With appropriate changes to the metalbinding functionality, the sensor design presented herein may be adapted for the targeted detection of other cations of biological relevance. We developed a sensor functionalized with a lipophilic cationic alkylphosphonium group, i.e. Mag-mito, which displays selective localization in mitochondria thus enabling the targeted ratiometric imaging of free Mg 2+ within these organelles in live cells. A time-course fluorescence imaging study conducted on HeLa cells treated with Staurosporine provided the frst direct observation of an increase in free Mg 2+ levels in mitochondria during early stages of apoptosis. The onset of this change appears to precede Ca 2+ entry into the organelle. Future studies will be aimed at identifying the origin and destination of this mitochondrial pool of free Mg 2+ and its influence in the downstream events in the apoptotic cascade.

chemsum

{"title": "Visualizing changes in mitochondrial Mg²⁺during apoptosis with organelle-targeted triazole-based ratiometric fluorescent sensors", "journal": "Royal Society of Chemistry (RSC)"}

analysis_of_reactive_oxygen_and_nitrogen_species_generated_in_three_liquid_media_by_low_temperature_

## Abstract: In order to identify aqueous species formed in Plasma activated media (PAM), quantitative investigations of reactive oxygen and nitrogen species (ROS, RNS) were performed and compared to Milli-Q water and culture media without and with Fetal Calf Serum. Electron paramagnetic resonance, fluorometric and colorimetric analysis were used to identify and quantify free radicals generated by helium plasma jet in these liquids. Results clearly show the formation of ROS such as hydroxyl radical, superoxide anion radical and singlet oxygen in order of the micromolar range of concentrations. Nitric oxide, hydrogen peroxide and nitrite-nitrate anions (in range of several hundred micromolars) are the major species observed in PAM. The composition of the medium has a major impact on the pH of the solution during plasma treatment, on the stability of the different RONS that are produced and on their reactivity with biomolecules. To emphasize the interactions of plasma with a complex medium, amino acid degradation by means of mass spectrometry was also investigated using methionine, tyrosine, tryptophan and arginine. All of these components such as long lifetime RONS and oxidized biological compounds may contribute to the cytotoxic effect of PAM. This study provides mechanistic insights into the mechanisms involved in cell death after treatment with PAM.Low temperature plasmas generated at atmospheric pressure have been studied in the past few years for their applications in the field of medicine and biomedicine. Short and long lived reactive oxygen (ROS) and nitrogen species (RNS) can be generated by non-thermal plasmas in either gaseous or aqueous forms when primary plasma species (ions, electrons, radicals, and dissociated molecules) interact with a liquid phase 1-3 . It is noteworthy that plasmas can induce either cell proliferation for low doses or cell death by apoptosis for high doses of exposure 4,5 . Low temperature plasmas have therefore been studied intensively for wound healing 6 , sterilization 7 , blood coagulation 4 , dental treatment 8, 9 and also for the inactivation of various cancer cells from breast 10 , head and neck 11 , ovarian 12 , lung 13 , prostate 14 or colorectal tissues [15][16][17] . It has been shown that plasma species can inactivate cancer cells either directly (interactions of gaseous species with cells) or indirectly when using a previously-prepared plasma-activated liquid media (PAM). In addition, such plasma treatments can selectively inactivate cancer cells without really affecting normal cells 16,[18][19][20] . Interestingly, PAM have numerous advantages: i) they allow selective treatment of internal organ cancer tissues which are difficult to reach by the gaseous species and requiring endoscopes or catheters; ii) they present minimal toxicity for normal tissues; and iii) they remain stable several days after their preparation if they are stored at the right temperature 13,16,21 . In PAMs, reactive oxygen and nitrogen species (RONS) have been shown to induce cancer cell apoptosis 11,12 , although the cell death pathways at a molecular level have not yet been clearly elucidated, some studies have suggested mitochondrial dysfunction 13,22 . Identification and quantification of the aqueous RONS generated in PAMs could therefore shed light on the mechanisms of action of PAM with regard to tumor eradication and wound healing. We have already described the genotoxic and cytotoxic effects of PAMs on colon adenocarcinoma multicellular tumor spheroids and its selective action on cancer cells elsewhere 16,20 .Furthermore, most of plasma devices described in the literature which are used to generate PAMs are based on dielectric barrier discharge setups (DBD) using RF, AC or pulsed power supplies with different carrier gases such as helium 10,15,16 , helium with oxygen 11,14 or argon 12,13 . The nature and quantity of the plasma species generated depend on the type of plasma device used and the carrier gas composition. Hence, there is a real demand for the identification and quantification of the aqueous RONS (superoxide anion radical, hydroxyl radical, singlet oxygen, nitric oxide, hydrogen peroxide, nitrite/nitrate, etc.) generated when the gaseous plasma species impact the liquid media. Several techniques are currently used to identify and quantify the gaseous plasma products, including optical emission spectroscopy (OES) or laser induced fluorescence . However, aqueous plasma by-products with a short lifetime are difficult to quantify using OES 15,26 , and other methods including chemical dosimetry or fluorescent probes 27,28 electron paramagnetic resonance (EPR) spectroscopy appear more appropriate. The later technique requires specific spin traps to allow detection of some aqueous plasma by-products 31 . The overall goal of this study is the analysis of several plasma-induced free radicals in three liquid media, i.e. Milli-Q water and a cell culture medium, DMEM, with and without fetal calf serum (FCS), which are exposed to a low temperature plasma jet generated by a DBD setup using helium carrier gas at atmospheric pressure 26 . These different liquid media are activated by the plasma jet using different exposure times. The RONS investigated are hydrogen peroxide (H 2 O 2 ), hydroxyl radical ( • OH), singlet oxygen ( 1 O 2 ), superoxide radical (O 2 •− ), nitric oxide (NO • ) and nitrite/nitrate anions (NO 2 /NO 3 ). H 2 O 2 was measured using a fluorometric kit and NO 2 /NO 3 with a colorimetric kit, while the three other free radicals were quantified using EPR spectroscopy and spin traps due to their short life spans. ## Results and Discussion Hydroxyl Radical ( • OH) produced in liquid media by He plasma jet. In order to characterize the formation of reactive oxygen species in different media such as Milli-Q water, a cell culture medium, DMEM, without or with fetal calf serum (FCS) upon exposure to cold plasma, spin trapping experiments were carried out using 5,5-dimethyl-1-pyrroline N-oxide (DMPO) as spin trap. Figure 1 shows the electron paramagnetic resonance (EPR) spectrum of DMEM containing DMPO exposed to the plasma jet for 150 s. Whatever the medium, the EPR spectrum recorded corresponds to the superposition of two signals. The first and the most intense signal yielded the following hyperfine coupling constants a N = a H = 15 G, g = 2.0056 and peak intensity ratio of 1:2:2:1 (Fig. 1, #) which is attributed to DMPO-OH adduct 3,30,31, . The second signal (Fig. 1, *) is a triplet having a N = a H = 15 G, g = 2.0056 and peak intensity ratio of 1:1:1. This signal is also observed in untreated media (Milli-Q water or DMEM +/− FCS) (data not shown). Under similar experimental conditions, this second signal was also been observed by Tresp et al. 34 and was assigned to DMPO-CH 3 adduct. DMPO-CH 3 spin adduct can only originate from the spin trap itself since no carbon-based molecules are present in Milli-Q water. This assumption seems to be confirmed by the absence of correlation between the signal intensity and the exposure time of the media to the plasma jet. • OH radical can be formed by the reaction of an oxygen atom with an H 2 O molecule at the liquid surface 38,39 or from the solvation of gaseous • OH produced in the gas phase of plasma giving rise to the production of aqueous • OH radicals. In a previous work, the presence of oxygen atom and OH radical species was validated by observing their specific radiation using emission spectroscopy 15 This indicates that both pathways may be involved in our experimental conditions. This assumption is supported by the data published by Gorbanev et al. 40 showing that with dry feed He plasma, OH radical may be produced both in the gas phase and in the sample. The concentration of OH radical was assessed by exposing the different media with DMPO spin trap to cold plasma. As shown in Fig. 2, the concentration of DMPO-OH adduct increases linearly with the plasma exposure time (R² water = 0.99, R² DMEM = 0.98, R² DMEM + FCS = 0.92) indicating an increase of OH radical production. Kanazawa et al. 27 have already shown similar linear relationship between the amount of OH radical in water and the time of exposure to He plasma jet. For this purpose, they used chemical dosimetry based on the reaction of terephthalic acid with OH radical to generate a fluorescent molecule 27 . The concentration of DMPO-OH is evaluated around 4.12 µM, 3 µM and 0.4 µM after 150 s exposure to He plasma in water and DMEM +/− FCS, respectively. The amount of DMPO-OH detected in water is higher than in both biological media. Indeed, OH radicals generated in water are surrounded only by water molecules and other types of radicals. In cell culture media, OH radicals can oxidize organic components such as amino acids, vitamins and proteins 41 . These oxidation reactions will reduce the quantities of OH radical trapped by DMPO. Chemical reactions between amino acids and OH radicals were predicted by reactive classical MD simulations 42 and observed by high-resolution mass spectrometry 43 . We have estimated the half-life of DMPO-OH in the three media considered Milli-Q water and DMEM +/− FCS. To estimate the half-life of DMPO-OH, the signal decay of DMPO-OH was studied from 120 s to 2000 s after plasma exposure. The data were fitted with a polynomial profile of the 3rd order. The results summarized in Table 1 show that the half-life of DMPO-OH adduct depends on the composition of the medium. In Milli-Q water, the adduct half-life was evaluated at 820 s which is very close to the 870 s found in the literature 31 . In DMEM +/− FCS the half-life was about 1043 s and 442 s, respectively. The stability of the adduct DMPO-OH is affected by the change in pH and/or the presence of nitric acid 32,44 . The results suggest that DMEM makes it possible to maintain the pH constant at around 7.4 after exposure to plasma leading to an increase in the stability of the DMPO-OH adduct. In the presence of FCS, its stability is around 2 or 2.3 times lower than in water and DMEM respectively. This might be explained by the fact that the adduct can bind to proteins such as albumin which have documented anti-oxidant properties and/or by the presence of higher quantities of nitric acid than in water and DMEM as we will show in the section nitrite/nitrate anions of manuscript. Nitrite may react with OH radical to form peroxynitrite. ## Formation of Superoxide Anion Radical (O 2 •− ) by the plasma jet in media. DMPO used as a spin trap in EPR experiments can react with many ROS under different rate coefficients and trapping times. The EPR signal obtained after plasma exposure (Fig. 1) of the different media containing DMPO is clearly defined as the signal of the DMPO-OH spin adduct. However the mechanism is more complex than it seems. DMPO shows a significant preference for the hydroxyl radical (k OH > 10 9 M −1 s − 1 31, 34 ) but can also scavenge superoxide anion to form the spin adduct DMPO-OOH which has a much lower reaction rate (k O2− < 10 2 M −1 s − 1 31, 34 ). The difficulty lies in the fact that adduct DMPO-OOH tends to decompose rapidly in DMPO-OH, making it difficult to detect. To highlight the presence of superoxide anion in the media after plasma treatment, superoxide dismutase (SOD) was added before plasma exposure. Found in almost all aerobic organisms this enzyme is a metalloprotein that catalyzes the dismutation of superoxide anion radical into molecular oxygen and hydrogen peroxide. The addition of SOD (150 units of SOD per well) to the media leads to the dismutation of the superoxide anion and thus to inhibition of DMPO-OOH spin adduct formation 31 when exposed to plasma. In water, the addition of SOD leads to a decrease of about 10.2 ± 3.34% in the concentration of DMPO-OH adduct (Fig. 3). This indicates that a small part of the DMPO-OH signal comes from the decomposition of DMPO-OOH into DMPO-OH. It is interesting to note that the amount superoxide anion produced increases linearly with the exposure time. This evolution was also observed by Arjunan and Chyne 45 using Tempo-9AC (fluorescent probe) and DBD plasma generated in humid ambient air. Superoxide anion is detected in water but in a much lower concentration than OH radical. H 2 O 2 is generated inside the plasma and is delivered into the medium 40 . In our previous work 16,20 , we showed that hydrogen peroxide produced in PAM remains stable when PAM is stored at 4 °C for at least 7 days. H 2 O 2 may be considered to be a central genotoxic agent produced during the exposure of media to a He plasma jet. This activity may be attributed to its ability to diffuse through cell membrane and to generate OH radical through a Fenton reaction catalyzed by iron containing proteins. Hydroxyl radical is known for its high reactivity with cellular components, its potent ability to induce DNA damage such as guanine oxidations, DNA cleavage and to cause cell death. ## Quantification of the The formation of H 2 O 2 by He plasma jet was characterized and quantified using a fluorometric Hydrogen Peroxidase Assay kit (procedure described in Material and Methods). As shown in Fig. 4, the H 2 O 2 concentration in Milli-Q water and cell culture media increases linearly with the exposure time to the plasma jet (R² water = 0.93, R² DMEM = 0.97, R² DMEM + FCS = 0.99). The linear increase in hydrogen peroxide concentration versus exposure time was also reported by Adachi et al. using an Argon plasma jet and a DMEM cell culture medium 13 . In the three different media, the hydrogen peroxide concentration can reach 1.60 mM ± 0.12 mM for 150 s of plasma exposure time. This is also very close to the concentration determined in our previous studies 16,20 . The H 2 O 2 quantities found are similar to those obtained by Tresp et al.using argon plasma to treat saline solution, buffered saline solution and cell culture medium 33 . In contrast to OH radical and superoxide anion radical, the quantities of hydrogen peroxide produced do not depend on the type of the medium treated. This result suggests that H 2 O 2 will not be consumed via a Fenton reaction in DMEM +/− FCS. Singlet Molecular Oxygen ( 1 O 2 ) induced in the liquid media exposed to the plasma jet. It has been shown that singlet oxygen may be generated in water or cell culture media after the treatment of these liquid media by He plasma jet and Air DBD plasma 3,31,45 . Wu et al. suggested that singlet oxygen was produced in plasma and diffused into the solvent 31 . In order to characterize the formation of singlet oxygen in the different media after plasma exposure, EPR experiments were performed using 2,2,6,6-tetramethylpiperidine (TEMP) as the spin trap. As shown in Fig. 5a, the EPR signal of Milli-Q water exposed for 150 s to plasma in the presence of TEMP is characterized by 3 peaks with an intensity ratio of 1:1:1 triplet with hyperfine coupling constants a N = a H = 16 G, g = 2.0059. This signal, detected in the three media, is attributed to the TEMPO spin adducts indicating the formation of singlet oxygen or ozone upon exposure to the He plasma jet. In water, the amount of TEMPO increases linearly with the exposure time (R² water = 0.98, R² DMEM = 0.95, R² DMEM+FCS = 0.88) (Fig. 5b). The addition of NaN 3 (50 mM), a selective scavenger of singlet oxyimen in water before plasma exposure induced a decrease of around 80% and 51% of the formation of TEMPO after 30 s and 150 s of exposure. This data clearly shows that the formation of TEMPO results from the oxidation of TEMP by singlet oxygen. Moreover, in our plasma jet configuration, with pure He, ozone cannot be detected along the plasma jet by using the UV absorption method at 253.7 nm. This indicates that ozone concentration is very low or negligible. This observation is also in good agreement with the investigations of Kawasaki et al. 49 . Both data show that involvement of ozone to the oxidation of TEMPO is negligible in our experimental conditions. The concentration of TEMPO produced in water was evaluated to be about 13.85 µM after 150 s exposure to the plasma jet. The intensity of the TEMPO spin adduct signal in both DMEM and DMEM + FCS is negligible when compared to the case of Milli-Q water. This may be explained by the fact the singlet oxygen is a very high reactive oxygen species known to oxidize cysteine, methionine and tryptophan free amino acids and proteins 50 . Singlet oxygen is consumed in both DMEM and DMEM with FCS. It is worth noting that singlet oxygen contributes to the formation of oxide radicals in PAM leading to a depletion of cellular nutrients. The formation of these oxidized by-products in DMEM +/− FCS may also contribute to the cytotoxic effects of PAM. Hydrogen radical ( • H) induced in liquid media by plasma treatment. Hydrogen radical was detected by EPR spectroscopy using α-phenyl-N-tert-butylnitrone (PBN) spin trap. The EPR signal obtained during plasma treatment of DMEM is shown in Fig. 6a. This signal is formed by 9 peaks with an intensity ratio of 1:2:1:1:2:1:1:2:1 and with hyperfine coupling constants a N = 16.2 G, a H = 10.3 G, g = 2.0776. It may be attributed to the PBN-H spin adduct. The PBN spin adduct of the • H atom was also observed with cold atmospheric dry-helium 40 and argon 32 plasma. Moreover, a small magnitude signal associated with PBN-H is also observed in Milli-Q water and cell culture media not exposed to the low temperature plasma jet (Fig. 6b at time = 0). Signal intensity was totally independent of exposure time in Milli-Q water whereas it increased linearly in DMEM +/− FCS (R² DMEM = 0.96, R² DMEM + FCS = 0.97). These results suggest that the He plasma jet produces a small quantity of hydrogen radical. This is in total agreement with the spectroscopy studies showing that the He plasma jet generate very small amounts of gaseous H 15 . ## Exposure Time (s) TEMPO Concentration (µM) In the case of DMEM +/− FCS, the formation of hydrogen radical may be partly attributed to the reaction of ROS such as OH radicals with amino acids 42,51 . As shown in Fig. 6b, the formation of • H is higher in DMEM than in DMEM + FCS. This difference could be explained by the presence of albumin in DMEM + FCS, a protein known for its antioxidant properties and its ability to trap free radicals 52 . This can lead to a decrease in the amount of ROS able to react with the amino acids in DMEM + FCS. Nitric Oxide ( • NO) formed in liquid media by plasma jet exposure. Nitric oxide was detected by indirect EPR spectroscopy using the Carboxy-PTIO spin trap that can react with nitric oxide to produce Carboxy-PTI and • NO 2. 53, 54 . Carboxy-PTIO and carboxy-PTI are stable molecular radicals that are detectable by EPR spectroscopy and have their own spectral signature. As shown in Fig. 7a the experimental signal of Carboxy-PTIO which is composed by 5 peaks with intensity ratio of 1:2:3:2:1 and hyperfine coupling constants a N = a H = 8.1 G, g = 2.0068 (this signal is obtained without any plasma treatment). Figure 7b shows experimental EPR signal obtained after exposure of 166 µM C-PTIO in Milli-Q water to the He plasma for 150 s. This signal consists of two superimposed radical spectra. The first one was identified as the remaining C-PTIO which does not react with nitric oxide (the simulated spectrum extracted from Fig. 7b is given in Fig. 7c). The second was the EPR signal of C-PTI (the simulated spectrum extracted from Fig. 7b is given in Fig. 7d) and presented 7 peaks with an intensity ratio of 1:1:2:1:2:1:1 and hyperfine coupling constants a N = 9.8 G, a H = 4.4 G, g = 2.0068. The presence of C-PTI in water after plasma exposure confirms the generation of nitric oxide. As shown in Fig. 8, the formation of CPTI in Milli-Q water increases in a linear fashion with (R² = 0.98) exposure to the He plasma jet. The amount of • NO cannot be estimated from the amount of C-PTI formed because C-PTI may be reduced by both hydroxyl and hydrogen radicals to the parent C-PTIO 55 . Nitric oxide in liquids arises after solvation of gaseous nitric oxide produced in the plasma plume observed using emission spectroscopy in our earlier study 15 . Production of nitric oxide in liquids has already been observed in Milli-Q water after its exposure to helium 56 or argon plasma 35 and also in buffered saline solution treated with air plasma 57,58 . In aqueous solution, nitric oxide is a highly reactive species that may react with oxygen to produce nitrite (NO 2 − ) (Reaction 1) 59 . Formation of NO 2 − in media is known to induce a decrease in the pH of the solution. As shown in Fig. 9, we observed a drastic decrease in the pH of water from 6.5 to 4.5 upon plasma exposure. This result suggests the formation of reactive nitrogen compounds such as nitrous acid, nitric acid and peroxynitrous acid . Nitrous acid, which is in acidic equilibrium with nitrite, may decompose in acidic medium into nitric oxide and nitrogen dioxide (Reactions 2 and 3) 58 . It is also known that nitrite is not stable under acidic conditions. In water, nitric oxide may be regenerated via reaction 3. In contrast, in cell culture media the amount of NO radical detected is very low. Due to the presence of a buffer, for example phosphate buffer at pH 7.4 in these media, the change of pH will be very slight upon plasma exposure. In phosphate buffer, the pH varies by less than one pH unit 62 At physiological pH conditions, nitrite will be more stable and subsequently reactions 2 and 3 will not occur or at least be limited in DMEM +/− FCS. This may explain the low concentration of NO in both cell culture media compared to water. Another possible explanation for the low concentration of NO in biological liquids could be the high reactivity of this radical with bio-macromolecules 63 . NO radicals and related species are able to modify proteins through chemical reactions without involving enzymes. Nitric Oxide groups can bind to a transition metal found in protein or thiol residues of amino acids such as cysteine 64 . Tyrosine is one of the main target of RONS 65 . Nitric oxide can also provide nitrogen sources for the formation of the nitro group (NO 2 ). The presence of proteins and amino acids in culture medium can explain the low nitric oxide concentration detected in DMEM +/− FCS. ## Nitrite (NO 2 − ) and Nitrate (NO 3 − ) anions generated in liquid media by plasma jet. The formation of nitrite and anions during exposure of the different media to a plasma jet was investigated and quantified. The concentration of nitrite in Milli-Q water and culture media (Fig. 10a) increases linearly with exposure time to plasma (R² water = 0.92, R² DMEM = 0.97, R² DMEM+FCS = 0.97). The nitrite concentration in DMEM with fetal calf serum is 2.87 ± 0.47 and 18.77 ± 2.26 times higher than in DMEM and in Milli-Q water respectively. It is more difficult to quantify the nitrate anion concentration. Figure 10b shows the results in Milli-Q water and serum-free medium only. No anion values can be estimated for plasma jet exposure time of less than 90 s. Above 90 s of treatment, the anion concentration seems to increase linearly with the exposure time. As observed for nitrite, the nitrate anionconcentration is higher in DMEM than in Milli-Q water after 150 s of plasma treatment. Significant uncertainties in the nitrite concentration and the inability to quantify anion in DMEM + 10% FCS could be explained by the interactions between the species produced in this medium by plasma jet and Griess reagent or by the nitrite concentration which is lower than the limit of detection (i.e. 2.5 μM). Moreover, the negligible values obtained during anion quantification may be due to the non-linearity of the calibration curves (see materials and methods) that increased significantly due to the dilution factor. Nitrite and anions concentrations in PAM seem to be affected by the composition of the different media. Both nitrogen species are formed in plasma treated media through the dissolution of nitrogen oxides formed in plasma jet. Their formation and their stability in water and DMEM +/− FCS will depend on different parameters such as the pH of the solution, the presence of amino acids, metallo-proteins and the production of ROS by the He plasma jet. As mentioned above, we have observed in water a drastic decrease in pH, about 2 units, after 150 s exposure to He plasma. In these acidic conditions, nitrous acid (which is one of the major source of nitrite NO 2 ## − ) is not stable. It will decompose rapidly into nitrogen dioxide which may subsequently react with hydroxyl radicals produced by the He plasma jet. This chemical reaction leads to the formation of peroxynitrous acid which is not stable in acidic pH and converts into stable nitrate NO 3 − . Lukes et al. 59 have shown that, under acidic conditions, nitrite will react with hydrogen peroxide to generate peroxynitrite and subsequently nitrate anion. Girard et al. 62 have clearly identified the formation of peroxynitrite anion in physiological pH under cold atmospheric plasma exposure. This cascade of chemical reactions may explain both the higher level of nitric oxide and the lower quantity of NO 2 − in water than in DMEM +/− FCS. These reactions will be less efficient in buffered media with a pH of around 7. Hence, the nitrite concentration in DMEM +/− FCS will be higher than in water. Moreover nitrate/ nitrite anions can be the targets of short lifetime ROS such as hydroxyl radical 66 leading to the formation of peroxynitrite. In complex media such as DMEM +/− FCS, this reaction will compete with the oxidation of biomolecules by ROS. However, the production of peroxynitrite in biological media via this pathway will be minor. Also, the higher concentration of nitrite observed in DMEM + FCS than in DMEM may be explained by the presence of copper proteins such as cytochrome c in FCS that can contribute to the oxidation of nitric oxide into nitrite 67,68 . The major source of anion in water and DMEM is the formation of nitrite anion and peroxynitrite. However, anions and nitrite are poorly reactive species with regards to bio-macromolecules. It is mostly ONOO−, NO 2 • , N 2 O 3 that will induce protein and DNA damage such as nitration and nitrosylation of amino acid residuals, nucleic acids and they will be partly responsible for the cytotoxic effect of plasma jet. In DMEM +/− FCS, these RNS will react with the biomolecules causing a decrease in the conversion of peroxynitrite in nitrate anion in these media. Taking into account all the factors influencing the reactivity of RNOS in solutions, the levels of nitrate anion observed should be water> DMEM > DMEM + FCS as was the case here. It has been shown that nitrate and nitrite anions can be recycled in NO in cells 69 . These inorganic anions produced in PAM are therefore potential sources of NO radicals. Depending on their concentrations, reactive nitrogen species (RNS) are known to have both deleterious and beneficial effects on cell dynamics 70 . Interestingly, the RNS produced by cold atmospheric plasma jets and the quantity produced seem to induce the death of cancer cells in particular offering interesting selectivity between healthy and cancerous cells 20,70 . Plasma treatment of aqueous solutions of amino acids. Oxidation of aqueous solutions of tyrosine, tryptophan, methionine and arginine after He plasma treatment were studied. The oxidized products were analyzed by HPLC coupled with mass spectrometry (HPLC-QTrap 4500-MS). In addition to the peaks corresponding to the reactants, the mass chromatograms of solutions exposed to plasma showed peaks corresponding to hydroxylation and nitration of tyrosine and tryptophane, sulfoxidation of methionine and hydroxylation of arginine (Figs S1-4 Supplementary data). All of these chemical modifications of amino acids by cold plasma were previously reported by Takai et al. 43 . Figure 11 shows that, in descending order, the reactivity of the 4 amino acids is methionine > tryptophan > arginine > tyrosine. Methionine is totally degraded after 30 s exposure to cold plasma. Tyrosine is the only amino acid which is nitrated. All of these data are in accordance with the extensive literature on the reactivity of hydroxyl radical and RNS with amino acids and proteins. These data partially explain the difference in reactivity observed between biological media and water during cold plasma treatment and their different biological activity with regards to cancer cells. ## Conclusion We have characterized and quantified the formation of radical species such as hydroxyl radical, superoxide anion, singlet oxygen, nitric oxide and long lifetime RONS such as H 2 O 2 and nitrite/nitrate anions in three culture media (Milli-Q water, DMEM and DMEM with FCS) exposed to a DBD plasma jet using helium at atmospheric pressure as a carrier gas. We have shown that the composition of the medium has a major impact on the pH of the solution during plasma treatment, on the stability of the different RONS that are produced and on their reactivity with biomolecules. Reactions of RONS with DMEM +/− FCS generate oxidized products which may be toxic for cells. Our data indicate that beside the production of long lifetime RONS, oxidized biological compounds form and accumulate in PAM leading to a decrease in essential nutrients for cell growth. All of these components such as long lifetime RONS and oxidized biological compounds may contribute to the cytotoxic effect of PAM previously observed on HCT116 spheroids 16,20 . Moreover, this suggests that the cytotoxicity of OH radical produced by He cold plasma jet can be mainly due to the production of cytotoxic chemicals in DMEM +/− FCS rather than a direct effect on cellular constituents. Moreover, long lifetime species (H 2 O 2 and nitrites/nitrates) can penetrate into cells and can be potential precursors of intracellular reactive oxygen species. Indeed, these species can lead in turn to the formation of • OH radical via a Fenton reaction involving H 2 O 2 and NO • synthesis via recycling of nitrites/nitrates anions by the cells 56 . • OH radical and NO • are highly cytotoxic and genotoxic for cells. Therefore, under the present plasma exposure conditions, the obtained PAM can necessarily generate many potential cytotoxic and genotoxic by-products which have interesting applications, particularly for cancerous cell inactivation. ## Materials and Methods Plasma jet device. Figure 12 shows a diagram of the plasma jet device based on a dielectric barrier discharge configuration already detailed elsewhere 15,71 . In short, two aluminum tape electrodes are wrapped around a quartz tube and connected to a High-voltage mono-polar square pulses generator. A power supply with the following characteristics is applied: 10 kV voltage, 9.69 kHz frequency and 1 µs pulse duration. Helium gas flows through the quartz tube at a flow rate of 3 L min −1 . ## Preparation of Plasma Activated Medium (PAM). Plasma activated medium (PAM), was produced by exposing 100 µL of water or cell culture medium DMEM with or without fetal to the He plasma jet. These different media were exposed for up to 150 s in 96 well plate, leading to a 20 µl decrease in volume after plasma exposure (data not shown). Plasma exposures were performed under the same experimental conditions (applied voltage, frequency, pulse duration and gas flow) and at the same distance of 2 cm between plasma jet tube output and the upper-surface of the liquid medium. EPR spin-trapping spectroscopy. Electron paramagnetic resonance spectroscopy (EPR) is a technique based on the magnetic resonance between an unpaired electron and an external magnetic field. Due to the very short lifetime of radicals, EPR measurement is often used with spin-trap reagents. The reaction of a radical with the spin-trap reagent leads to the formation of a longer-lived spin adduct. EPR spectra were recorded with a Bruker ESP 500E spectrometer at room temperature. The following instrumental settings were employed for the measurements: central field: 3516 G; sweep width: 100 G; microwave frequency: 9.87 GHz; modulation frequency: 100 kHz; microwave power: 5.15 mW; scanning time: 84 s; number of scans: 6. Figure 13 gives EPR calibration performed using an aqueous solution of a stable radical, TEMPO, in concentrations ranging from 0-50 µM 40 . The spin trapping reagents, 5,5-dimethyl-1-pyrroline N-oxide (DMPO), 2,2,6,6-tetramethylpiperidine (TEMP), α-phenyl-N-tert-butylnitrone (PBN), 2-(4-Carboxyphenyl)−4,5-dihydro-4,4,5,5-tetramethyl-1H-imidazol-1-yloxy-3-oxide potassium salt (C-PTIO), D-Mannitol (OH* scavenger), and superoxide dismutase (superoxide anion scavenger) Sodium azide (NaN 3 ) were purchased from Sigma-Aldrich. Reactive oxygen species such as hydroxyl radical ( • OH) superoxide anion (0 2 •− ), singlet oxygen ( 1 O 2 ), nitric oxide radical (NO • ) and others including hydrogen radical (H • ) are presumed to be produced in the different media upon their exposure to plasma. To characterize their formation, the spin trapping reagents DMPO •− into H 2 O 2 and O 2 and directly related to the intensity of the DMPO-OH EPR signal. The DMPO-OH intensity was estimated directly from the half-height of the second peak of the quartet EPR spectra obtained for different media. All measurements were performed in triplicate for each sample and the changes in the EPR signal were monitored for 2000 seconds after plasma exposure. Easyspin (MATLAB library) 72 and WinSim2002 software were used for EPR spectra simulations [available online https://www.niehs.nih.gov/research/resources/software/tox-pharm/tools]. Hydrogen peroxide (H 2 O 2 ) assay. H 2 O 2 concentrations in PAM were quantified using a fluorometric Hydrogen Peroxidase Assay kit (Sigma-Aldrich Co., Ltd). This kit uses horseradish peroxidase and a red fluorescent peroxidase substrate (λex: 540 nm, λem: 590 nm) and allows quantification of H 2 O 2 in a range of concentrations between 0 and 10 µM. PAM were diluted 1/50 before each measurement in order to achieve an adequate concentration of H 2 O 2 . Calibration curves (see Fig. 14) were plotted for the different liquid media used for the preparation of PAM from an initial 3% hydrogen peroxide solution to avoid any influence of medium absorption on the fluorescence measurements. Samples in black 96-well plates were analyzed using a CLARIOstar fluorescence plate reader (BMG LABTECH) at room temperature. ## Detection of Nitrite/Nitrate anions (NO 2 − /NO 3 − ). NO 2 /NO 3 concentrations in PAM were assayed with a colorimetric Nitrite/Nitrate Assay kit (Sigma-Aldrich Co., Ltd) using Griess reagent and nitrite reductase. The kit was initially planned for measurements in a range of concentrations between 0 and 100 µM of NO 2 − and NO 3 − . PAM were diluted 1/20 before each measurement in order to have an adequate concentration. Calibration curves (see Fig. 15) were plotted for each medium from initial solutions of NaNO 2 and NaNO 3 to take into account the influence of the medium absorption on the measurement. Sample absorbance at 540 nm was analyzed using a CLARIOstar plate reader (BMG LABTECH) at room temperature. ## Mass spectrometry analysis. The LC/MS system was equipped with an HPLC chromatograph (HPLC Agilent 1100 series) and a triple quadrupole mass spectrometer (QTRAP Applied Biosystems). HPLC analyses were performed using a Waters X Bridge C18 (3.5 µm) column (2.1 × 150 mm), and a gradient elution starting with 5% acetonitrile and 95% ammonium acetate at a flow rate of 0.6 mL min −1 rising at 15 min a plateau corresponding to 50% acetonitrile and 50% ammonium acetate for 5 minutes. The mass spectrometer was equipped with an electrospray ion (ESI) source (turbo ion spray (TIS) and was operated in positive mode. Nitrogen served as auxiliary, collision gas, and nebulizer gas. The detection was scan mode with a step size of 0.1 atomic mass unit (amu) and a scan range of 50-500 amu. Mass chromatograms, i.e., representations of mass spectrometry data as chromatograms (the x-axis represents time and the y-axis represents signal intensity), were registered using different scan ranges. The prepared concentration of each amino acids (tryptophan, tyrosine, methionine, arginine) was 0.2 mM in MilliQwater. Each solution was treated by He plasma at various time of exposure (0-240 s).

chemsum

{"title": "Analysis of reactive oxygen and nitrogen species generated in three liquid media by low temperature helium plasma jet", "journal": "Scientific Reports - Nature"}

a_general_protocol_for_the_accurate_predictions_of_molecular_13_c/_1_h_nmr_chemical_shifts_via_machi

## Abstract: Accurate prediction of NMR chemical shifts with affordable computational cost is of great importance for rigorous structural assignments of experimental studies. However, the most popular computational schemes for NMR calculation-based on density functional theory (DFT) and gauge-including atomic orbital (GIAO) methods-still suffer from ambiguities in structural assignments. Using state-of-the-art machine learning (ML) techniques, we have developed a DFT+ML model that is capable of predicting 13 C/ 1 H NMR chemical shifts of organic molecules with high accuracy. The input for this generalizable DFT+ML model contains two critical parts: one is a vector providing insights into chemical environments, which can be evaluated without knowing the exact geometry of the molecule; the other one is the DFT-calculated isotropic shielding constant. The DFT+ML model was trained with a dataset containing 476 13 C and 270 1 H experimental chemical shifts. For the DFT methods used here, the root-mean-squarederivations (RMSDs) for the errors between predicted and experimental 13 C/ 1 H chemical shifts are as small as 2.10/0.18 ppm, which is much lower than the typical DFT (5.54/0.25 ppm), or DFT+linear regression (4.77/0.23 ppm) approaches. It also has smaller RMSDs and maximum absolute errors than two previously reported NMR-predicting ML models. We test the robustness of the model on two classes of organic molecules (TIC10 and hyacinthacines), where we unambiguously assigned the correct isomers to the experimental ones. This DFT+ML model is a promising way of predicting NMR chemical shifts and can be easily adapted to calculated shifts for any chemical compound. NMR spectroscopy is a powerful tool to interrogate chemical structure and stereochemistry because the magnitude of the chemical shift for the target atom reflects its local chemical environment 1 . Accurate predictions of NMR chemical shifts with respect to experimental values are highly valuable for structural elucidations, especially when a straightforward assignment is lacking, and which sometimes may result in misleading conclusions 2 . Currently, calculations of isotropic shielding constants based on density functional theory (DFT) 3 and gauge-including atomic orbital (GIAO) 4 have become routine procedures for chemists. However, the performance of these calculations is often uncertain and strongly depends on the computational model used, such as exchange-correlation functionals, basis sets, solvents, or conformational effects 5 being unable to unambiguously assign structures. Higher level theories such as coupled cluster (CC) may be more accurate, but currently they are too expensive to be applied on medium-or largesize molecules, i.e., 10 1 -10 2 atoms 6 . In recent years, linear regression (LR) 3 has been used for the correction of chemical shift predictions, complementing the aforementioned DFT and GIAO methods . The LR between calculated isotropic shielding constants () and experimental chemical shifts () is expressed in a simple form:  = slope×+intercept, where "slope" and "intercept" are fitted with respect to a considerable amount of data. Although the accuracy of such DFT+LR model is still limited (vide infra), it implies that a statistical or a more advanced data-driven approach may be used to improve the NMR chemical shift predictions. To abstract enough information from the large number of existing experimental NMR chemical shifts, an elegantly designed model augmented with an efficient data science tool could be very valuable. Machine learning (ML) is emerging as a powerful data-driven approach in the fields of chemistry and physics 10 , and has gained wide application in building highly accurate force fields 11 , accelerating global optimization 12 , assisting catalysts and material development 13 and predicting molecular electronic structures 14 , to mention a few. Moreover, ML has also been applied to improve the accuracy of NMR predictions of molecules , proteins , and solids 22,23 . However, these ML methods have been applied to only a small set of systems, and contain only a few input features (for example, one single set) or are trained only with calculated (instead of experimental) data, which greatly limits their applicability and accuracy. Some prediction models cannot be systematically improved because of a large but less diverse dataset, making them less robust outside the dataset used to fit them. In this study, we focus on building a model with the intent of obtaining a more accurate, robust, sustainable, and applicable toolset for 13 C/ 1 H NMR chemical shifts that can be applied to any molecular system. Given a molecule, the model will read the molecular information (input features) and predict its NMR chemical shifts (output labels). The model relies on a fully connected, multi-layer DNN. Other ML objects such as convolutional NN (CNN) were not used because DNN is chemically relevant and provides quick and robust answers, being much more accurate for a chemical problem with high quantitative demand 24 . Success of an ML model is to a large extent determined by the design of its input features. It is logical to include chemical environments of the target atom into the input feature. An algorithm to efficiently encode them into a vector is therefore needed. In previous studies of NMR predictions, chemical environments are represented by some mathematical functions of atomic coordinates 15,22,23 . They are biased toward the geometrical information. Accurate NMR predictions based solely on the geometrical information cannot be achieved, because more NMR-related chemical information like electronic structure is only implicitly represented. An experienced chemist can already estimate NMR chemical shifts of a molecule with insight into the chemical environment manifested in atom types, hybridization states, the electronegativities of connected atoms, the ring strain and aromaticity, and so on. These chemical properties can be explored for atoms in an organic molecule even before knowing its exact geometry without expensive calculations. For example, the Gasteiger charge 25 , which can be evaluated using only the bonding information. Inspired by this analysis (see Fig. 1), the chemical information of an atom A in a molecule can be described by collecting numerical values into a vector 𝐯 𝐴 , termed as chemical environment descriptor in this paper. The following eight properties were found to be able to efficiently encode chemical environments: atomic number (identifying atoms), Gasteiger charge 25 (reflecting electronegativity equilibration), the total valence (revealing bonding state), the size of the minimal ring the atom stays in (revealing geometrical information), the Crippen logP contribution 26 (measuring lipophilicity contribution), the Crippen molar refractivity contribution 26 (measuring steric effect), the topological polar surface area, and the Labute approximate surface area 27 (both measuring solubility contribution). These eight descriptors were found to be necessary and sufficient in expressing the chemical properties of the various atoms within their respective chemical environments. It is also critical to explicitly consider the bonding atoms of A because they directly affected its NMR chemical shifts. Therefore, the input feature xA of atom A can be constructed by its vector v as well as the vectors associated with the atoms it is bound to. In the current implementation, up to four atoms can bond with atom A. As a result, the chemical environment descriptor part of xA is a 5 (4+1 atoms)×8 (descriptors)=40-dimensional vector. When A has less than four bonding atoms, the remaining components of xA are packed with zeros. Although xA already contains a large amount of chemical information, exploratory studies showed that it was still insufficient to give an accurate prediction (vide infra). This is because some different chemical environments cannot be distinguished. An example is that for the molecule in Fig. 1, xA is identical for its different diastereoisomers. Therefore, additional dimensions of chemical information are needed. That explains why the model must include DFT calculations, the results of which depend explicitly on the actual geometry, the electronic state, and even the applied solvents in calculations, etc. The calculated isotropic shielding constants are directly related to NMR chemical shifts; therefore, it is taken as an additional component of xA. Finally, xA becomes a 41-dimensional vector after the addition of the isotopic screening constant. With vector xA as input features and predicted NMR chemical shifts as output labels for molecules in the dataset, a DNN can be trained with respect to a dataset containing 476 13 S1) were used to calculate isotropic shielding constants; and for each DFT method, a DNN was trained. For comparison, a DFT+LR model was also constructed using the same dataset. For all molecules in the dataset, their calculated isotropic shielding constants and predicted and experimental NMR chemical shifts can be found in Tables S2 and S3. The performance turned out to be only weakly dependent on the DFT functionals used for calculating isotropic shielding constants (Table S1). Hereafter, all discussions are based on models with Method 2 unless otherwise mentioned, since it has the lowest computational cost in practice. Our DFT+ML model showed high effectiveness in significantly reducing the errors compared to either the pure DFT or DFT+LR approaches (see Figs. A more encouraging observation comes from the comparison of error distributions between the different methods. As shown in Figs. 2, S3, and S4, the distributions of errors of our DFT+ML model are Gaussian-like. In contrast, both DFT and DFT+LR methods have a lot of outliers with large errors (for 13 C: error > 5 ppm; 1 H: error > 0.5 ppm). Combined with the observation that the performance is essentially independent of the functional used, the ML-based approach seems to successfully eliminate a large part of systematic errors (for example, DFT's tendency to overestimate NMR chemical shifts 28 ), leaving only random errors which should have a Gaussian distribution. Procedures based on simple numerical corrections such as DFT+LR also tend to have only limited predictive success as they cannot take into account the dependence on the chemical environment. Overall, the use of a more sophisticated method like a DNN based model is essential in capturing the dependence on the chemical environments and are more successful. One natural product, limonene and some of its isomers (Fig. S5), were used to further validate the DFT+ML model. The results are summarized in Fig. S6 and Tables S4 and S5 and show excellent agreement between predicted and experimental values for 74 13 C and 47 1 H NMR chemical shifts with small RMSD values of 2.02 and 0.20 ppm, respectively. To further evaluate the robustness of our DNN input feature, we conducted several independent types of training of different DFT+ML models. The results are summarized in Fig. 3 and Tables S6 and S7. For 13 C/ 1 H NMR chemical shift predictions, without using any chemical environment descriptors or DFT calculations, the DNN could only reduce the errors to 4.59/0.22 and 11.56/0.45 ppm, respectively, showing little improvement over DFT+LR (4.77/0.23 ppm). Therefore, both input features are essential for accurate NMR chemical shift predictions. In the DFT+ML model, DFT calculations give an overall estimation of the chemical shifts, further supporting our thesis that the inclusion of the chemical environment is critical to overcome the accuracy barriers beyond the corrections by LR. Compared with two recent models, the DFT+ML model outperforms the other models in terms of RMSD and maximum error (see Table 1). The models in Refs. 15 and 16 used a sorted Coulomb matrix and a set of rules as input features, respectively. As mentioned above, the lack of explicit electronic structure information limits their accuracy. Note that both models used more than 5000 NMR chemical shifts for training. Nowadays, constructing an accurate ML model with small datasets has become the subject of intense research. For example, using active learning, an accurate neural network force field could be constructed with only 10% of the original dataset 29 . In the DFT+ML model, the introduction of a DFT-calculated isotropic shielding constant as an additional input component not only increases the performance of the model, but also efficiently reduces the size of the required dataset. We used two prototypical classes of organic compounds with rich stereochemistry to demonstrate the ability of this model to accurately predict 13 C NMR chemical shifts: (1) TIC10 and (2) hyacinthacines. Their extended conjugated ring structures are also part of many systems relevant to separations, such as functionalized graphene 30 or lignins 31 , where also NMR is often used in determining particular chemical components. ## TIC10: The bioactive TIC10 (1a in Fig. 4) was reported in 1973 to be able to induce the expression of a target for cancer treatment 32 . In the initial report 33 , 1a was originally misassigned as 1b, while the actual 1b and another isomer 1c are not biologically active. The elucidation of their structures was merely conducted by mass spectrometry, and their full characterization had not been realized 7 . The structural assignment from experiments is challenging, and misassignment can lead to significant losses for industrial drug design. Here, our DFT+ML model predicted the 13 C NMR chemical shifts of structures 1a-1c (see Table S8). The errors plotted in Fig. 4 show excellent agreement with experimental data for structure 1a (RMSD: 0.83 ppm; much lower than that of 2.80 ppm in a previous study based on DFT+LR model 7 ). Large deviations were evident for structures 1b and 1c, either from the specific positions or from the overall errors. Therefore, the DFT+ML model can unambiguously assign 1a to the target molecule. Hyacinthacines: This is a natural product from Muscari Armeniacum that can inhibit glycosidase 34 . It contains several chiral centers (Fig. 5), and the absolute configuration of A1 isomer (Fig. 5) was unknown when this class of compounds was first synthesized. Again, NMR was used as a tool for structural assignment. Because there are not many molecules with carbon-nitrogen bonding types in the current dataset, the accuracy of our model for such molecules is limited. For example, the prediction accuracy for isomers of Nevirapine (a nitrogen-containing compound) using the DFT+ML model is close to that of the DFT+LR model (see Table S9 and Fig. S7). However, one way to improve the prediction accuracy lies in the selective extension of the original dataset. As mentioned above, the role of chemical environment descriptors in the framework of DNN should be underscored. Therefore, we augmented the training dataset with isomer 3 and isomer 5 of hyacinhacine and retrained the DNN. Using this updated DFT+ML model, the predicted values (shown in Table S10) match well with experimental data, and the errors for 13 C/ 1 H were within a reasonable range (less than 2.81/0.15 ppm; see Table S10-5), comparable to M. M. Zanardi et al.'s prediction 35 based on the DP4 approach. It is worth noting that the experimental chemical shifts for these isomers were measured in CD3OD and D2O (shown in Table S10), while most of the experimental chemical shifts in our dataset were measured in DMSO. Even though the transferability between different solvents has been investigated in our previous studies of 15 N NMR predictions 8 , which were based on the DFT+LR approach, the use of a different solvent could be a possible source for prediction error. To sum up, a DFT-based computational scheme augmented with a deep neural network is proposed as a promising computational model for the prediction of NMR chemical shifts of organic molecules. The number of layers in the model was optimized by manually adjusting the layers and comparing the error to a large set of experimental data. The model demonstrated high accuracy for 13 C/ 1 H chemical shift predictions compared to other empirical approaches and outperformed two previous ML-based models. The effectiveness of this novel approach was also tested with various types of structural assignments, indicating its robustness and reliability for chemical studies. Its prediction accuracy can be further improved via selective extensions of the original dataset. In future work, we will seek to improve the accuracy of the predictions of other molecular properties, which are correlated with the chemical environment of the target atoms. ## Computational Methods Dataset. A dataset of 476 13 C and 270 1 H experimental chemical shifts was compiled (see Supporting Information). It contains data collected by us and other researchers 3,36 . This dataset covers various types of bonding environments for target atoms, and it can be sustainably augmented for further applications by experimental researchers. Input feature calculations. The chemical formulas of organic molecules were represented by the SMILES codes 37 . The chemical environment descriptors (vector v) were calculated using RDKIT 38 . The isotropic shielding constants (σ) were performed using four different levels of theory as shown in Table S1. These methods were recommended in previous studies because of their reliability 3,7 . Two steps were involved: (1) geometry optimization in the gas phase, and ( NMR calculation with an SMD-implicit solvent model 39,40 . These calculations were carried out with Gaussian09 41 . Training the DFT+ML model. After exploration, a 7-hidden layer DNN (41×32×64×128×256×512×128×16×1) was built for this model. The input and output layer are xA and the predicted NMR chemical shift A, respectively. The hidden and output layer nodes used sigmoid and rectified linear unit (ReLU) activation functions, respectively. The weight and bias parameters were initialized with the LeCun uniform random approach 42 and zeros, respectively. The dataset was randomly divided into a training one and a test one with a ratio of about 0.9:0.1. The DNN was trained using the ADAM update method 43 with default parameters with TensorFlow 44 . ## ASSOCIATED CONTENT The Supporting Information is available free of charge via the Internet at http://pubs.acs.org. Experimental and predicted 13 C/ 1 H NMR chemical shifts used in this study (PDF) Datasets for training the neural network (PDF)

chemsum

{"title": "A General Protocol for the Accurate Predictions of Molecular 13 C/ 1 H NMR Chemical Shifts via Machine Learning", "journal": "ChemRxiv"}

theoretical_and_computational_validation_of_the_kuhn_barrier_friction_mechanism_in_unfolded_proteins

## Abstract: A long time ago, Kuhn predicted that long polymers should approach a limit where their global motion is controlled by solvent friction alone, with ruggedness of their energy landscapes having no consequences for their dynamics. In contrast, internal friction effects are important for polymers of modest length. Internal friction in proteins, in particular, affects how fast they fold or find their binding targets and, as such, has attracted much recent attention. Here we explore the molecular origins of internal friction in unfolded proteins using atomistic simulations, coarse-grained models and analytic theory. We show that the characteristic internal friction timescale is directly proportional to the timescale of hindered dihedral rotations within the polypeptide chain, with a proportionality coefficient b that is independent of the chain length. Such chain length independence of b provides experimentally testable evidence that internal friction arises from concerted, crankshaft-like dihedral rearrangements. In accord with phenomenological models of internal friction, we find the global reconfiguration timescale of a polypeptide to be the sum of solvent friction and internal friction timescales. At the same time, the time evolution of inter-monomer distances within polypeptides deviates both from the predictions of those models and from a simple, one-dimensional diffusion model.Protein folding may be justifiably viewed as a finite-size first-order phase transition, with folding kinetics following a classic nucleation mechanism 1, 2 . This view, however, masks the enormous complexity of folding dynamics at the molecular scale, where an unfolded protein chain samples a very large number of conformations before reaching the native state. Yet this number of conformations actually pales compared to the astronomically large number of conformations that the unfolded chain can access in principle. Levinthal considered this paradox of exhaustive search taking longer than the age of the universe, proposing that some combination of non-equilibrium dynamics and locally favorable interactions lead to biologically accessible folding times 3 . Subsequently, Anfinsen and Go pursued a purely thermodynamics view, emphasizing the congruency between stabilizing local and tertiary interactions, resulting in cooperative transition from the unfolded phase to the folded state 4,5 . A modern statistical mechanical theory of protein folding has built upon these insights, suggesting that, in general, energy landscapes of globular proteins are globally correlated, where not only the native state is unusually low in energy compared to random conformations of the protein chain, but also conformations that partially resemble the native structure are also rather low in energy 6, 7 . This funnel-like organization of globular proteins' conformational substates is extremely unlikely for random protein sequences, being a result of sequence evolution over billions of years 6 . The energy landscape theory of protein folding raises a new issue of local energetic ruggedness, where deep energetic traps might kinetically arrest the folding process 8 . Hence, a quantitative criterion of the ratio of the funnel depth to the magnitude of local energetic ruggedness determines whether a particular protein sequence will be foldable at laboratory or biological timescales, with an important related energy scale controlling the coil-to-globule collapse temperature 7,9 . The funnel theory is mesoscopic, leaving room for various alternative polymer theories of microscopic chain dynamics. Among interesting questions are whether the unfolded state is in a swollen (coiled) or molten globular state, and in the former case, whether folding is preceded by globular collapse or takes place concomitantly with it. Hence, both analytical and numerical models rooted in polymer theory can provide important insights into the nature of dynamics in the unfolded ensemble and into the specific collective processes driving the folding reaction. These approaches are particularly fruitful in addressing kinetic questions, where many prior studies of protein folding assumed either one-dimensional diffusion along some reaction coordinate 8 (which is postulated, equated to an experimental signal, or computed based on an optimality criterion 10 ) or relied on discrete kinetic networks , which are often deduced from all-atom simulations. One-dimensional models are attractive because of their direct connection to experimental data, but, because they lump all the complexity of protein dynamics into a few empirical parameters (such as the effective diffusion coefficient), they often lack molecular insight. Moreover, the assumption of one-dimensional diffusion is not always justified . Discrete state models, on the contrary, often involve too much molecular and kinetic information to offer adequate insight. Polymer theory models offer an attractive middle ground between all atom descriptions and low-dimensional models. Such models highlight certain universal properties of chain molecules that are independent of their precise chemical identity. Of course, therein lies their weakness: they ignore sequence effects or specific intrachain interactions; yet despite this drawback they are often remarkably successful in accounting for many structural and dynamical features of proteins. For example, despite sequence diversity, the radius of gyration of most chemically denatured proteins was found 19 to scale with the chain length N as N 0.6 , in accord with Flory's scaling law for random coil 20 (although this scaling may not be applicable to proteins under physiological conditions 21,22 ). Likewise, we showed that the experimental observations of the dynamics of loop formation within both unfolded proteins and single-stranded DNA obey simple, universal relationships derived from polymer theory 23 . Polymer theory makes simple, experimentally testable predictions regarding protein dynamics in the unfolded state. In particular, the global relaxation time τ r (or reconfiguration time, in the language adopted by the protein folding community) of a sufficiently long unstructured polymer chain in solution is predicted to be comparable to the time it takes the chain to diffuse over a length equal to its own size 24 : Here we can take the mean-square end-to-end distance, R 2 , as a measure of the size, D tr is the translational diffusion coefficient, and c is a numerical factor that depends on the specific model and on the precise definition of the reconfiguration time. If, for example, τ r is identified with the timescale associated with the slowest relaxation mode in the Rouse model, then one finds 25 c = π −2 /3. If, as in some studies, τ r is defined as the end-to-end vector relaxation time 26 , or as the end-to-end distance relaxation time, the resulting values of c differ from the above one by only a numerical factor of order one (see ref. 17 and the Results section). We will refer to the limit where equation (1) is satisfied as the Rouse-Zimm (RZ) regime, as it is the regime where the standard Rouse or Zimm theories of chain dynamics apply 25 . Equation (1) provides a reasonable order-of-magnitude estimate of the reconfiguration time of chemically denatured proteins, but it manifestly underestimates the reconfiguration times of some of the intrinsically disordered proteins or proteins that are unfolded near native conditions 27 , presumably because of intra-chain interactions or some other new effects. The remarkable feature of equation ( 1) is its insensitivity to any microscopic details of protein dynamics -a result known as the Kuhn theorem 24 . Indeed, assuming that D tr depends only on the solvent temperature, viscosity, and the protein radius of gyration (as would be implied by the Stokes-Einstein formula), these parameters also completely determine the reconfiguration time, a characteristic of a protein's internal dynamics. The Kuhn theorem is true under the assumption of sufficiently long chains, which, of course, does not necessarily apply to finite-length polypeptides. It is then instructive to consider the opposite limit of a very short peptide, say a di-peptide. In this case, significantly different configurations of the molecule result from its ϕ− and ψ− dihedral angles occupying distinct regions of the Ramachandran plot. The characteristic time over which such a molecule significantly changes its conformation (as quantified, for example, by the fluctuation timescale of its end-to-end distance) is then controlled by the typical time τ dih it takes the molecule to jump into a different Ramachandran plot region. Such jumps usually involve activated barrier crossing. In contrast, internal barriers do not affect the time of equation (1) (except through the relatively weak dependence of the average chain dimensions on the precise shape of the dihedral energy landscape). Following de Gennes 24 , we will refer to the short chain limit, where the chain dynamics timescale is dominated by τ dih , as the barrier friction limit. For notational brevity, we will reserve the term "barrier friction" to refer to this specific mechanism that involves overcoming dihedral barriers, keeping, however, in mind that many other intra-chain interactions may also give rise to microscopic kinetic barriers. Barrier friction is related to (but not necessarily identical with) the notion of internal friction in proteins. The most common experimental definition of internal friction is as a viscosity-independent component of the friction; that is, if the dependence of τ r on the solvent viscosity η is of the form τ r = τ i + aη, then τ i is the internal friction timescale 28,29 . Since dihedral rotations within a solvated polypeptide chain are, in general, mediated by the hydrodynamic drag on its various parts, there is no a priori reason to expect that τ i would be entirely controlled by dihedral rotations. Nevertheless, recent computational evidence suggests that (i) dihedral relaxation times are weakly dependent on solvent viscosity 30,31 and (ii) the height of the dihedral barriers controls the internal friction timescale 30 . These observations may at least partially explain why the two different definitions of friction, internal friction (operationally defined as solvent viscosity independent component of reconfiguration time) and barrier friction (stemming from hindered dihedral rotations) may coincide 27 . Experimental studies that probed not only the slowest relaxation time but also the entire spectrum of relaxation times 27,32 further suggest that the dynamics of unfolded proteins can be interpreted in terms of the Rouse or Zimm models with internal friction 18,24,33,34 (RIF and ZIF), whose physical foundation is the Kuhn picture of barrier friction 35 . RIF and ZIF provide a semi-phenomenological description of internal friction effects and build on the classic Rouse and Zimm models of polymer dynamics. Both Rouse and Zimm models are coarse-grained descriptions of polymers, which are represented as connected beads subjected to Brownian motion in solution. In addition to the solvent-induced forces, RIF and ZIF introduce an internal friction force, which resists deformation of inter-bead bonds. The physical mechanism of this force is that described by Kuhn: stretching a chain segment requires rearrangement of one or several dihedral angles within this segment, which is accomplished via activated barrier crossing. Bazua and Williams predicted a RIF/ZIF-type internal friction force using a rotational-isomeric-state model 35 , but even simpler arguments 24 predict that the internal friction force on a chain segment must be inversely proportional to its length, l. Specifically, imagine that a stretching force F is applied to the ends of a segment, causing dihedral rotations that preferentially lead to a more extended segment state. The average velocity u at which the ends will be moving apart must be proportional to the number of dihedral transitions per unit time, which is proportional to the segment length l. In the linear response regime, it is also proportional to the force, so we have u ∝ lF or F ∝ u/l, resulting in a friction coefficient inversely proportional to l. Now, the relaxation time of the segment can be estimated as this friction coefficient divided by the segment's effective spring constant. Since this spring constant is, likewise, inversely proportional to l 24 (assuming a chain with Gaussian statistics), this results in a single timescale that is independent of chain length: this is the internal friction timescale τ i ; if, as suggested by simulations, this timescale is itself viscosity independent, it should coincide with the zero-viscosity intercept of τ r (η). Despite these developments there is currently little consensus about what exactly internal friction is and how it is related to microscopic chain dynamics. Specifically, the following questions remain open: (1) The microscopic mechanisms of internal friction in unfolded proteins (and even in simpler polymeric systems 36 ) remain elusive. In particular, while the idea that dihedral rotations lead to internal friction effects is not new, no quantitative connection between the microscopic parameters describing the dynamics of the dihedrals and the experimental measures of the internal friction (such as the internal friction time τ i within the RIF/ZIF picture) is known. It is tempting to equate τ i with the dihedral hopping time τ dih , but little experimental or theoretical evidence exists in support of this idea -just because the two timescales are related does not mean they are the same! In fact, in a series of papers by Allegra and collaborators , an entirely different model was proposed to explain polymer relaxation in terms of dihedral dynamics; this model is at odds both with the relaxation spectrum predicted by RIF or ZIF and with the assertion that internal friction can be described in terms of a single, chain length independent timescale. Likewise, other models of polymer dynamics 24,36,40,41 postulate alternative internal friction mechanisms and do not necessarily lead to a single internal friction timescale. (2) A related question is concerned with the inherent limitations of the RIF/ZIF models. Despite their success in fitting experimental data, these models have a fundamental flaw: they fail to describe the rotational dynamics of a polymer chain and predict that, in the limit of high internal friction, the chain rotation timescale would coincide with τ i . This is obviously not true: even if the chain configuration is frozen with all dihedrals angles fixed, it can still rotate, with solvent friction determining the rotational timescale. It was thus argued 42 that RIF and ZIF should be viewed as one-dimensional models that are only applicable to the internal chain dynamics. But since the concept of a dihedral angle is meaningless in a one-dimensional space, the connection between dihedral dynamics and the ZIF or RIF parameters becomes even more vague. (3) Most models of internal friction force postulate, without justification, additivity of internal and solvent friction. Yet an alternative common view of internal friction based on diffusion on rough landscapes 43 postulates multiplicativity of the two effects! Without the additivity assumption, no theory exists that simultaneously includes both internal and solvent friction and interpolates between the solvent friction dominated and internal friction dominated limits; How, then, does the reconfiguration time τ r depend on chain parameters in the (arguably most relevant experimentally) intermediate case between these two limits? (4) While the Rouse or Zimm models provide a reasonable view of polymer dynamics in the absence of internal friction effects, a mechanistic description of the chain dynamics in the internal-friction-dominated regime is lacking. In particular, although it is reasonable to view chain reconfiguration in this limit as resulting from dihedral rotations, whether such rotations must occur in a concerted fashion or can be independent of one another is an unsettled issue. Independent dihedral rotations require large motions of the entire polymer chain and, therefore, they must entail high solvent friction, but concerted rotations, while reducing the friction, involve higher activation energies 44 . Simulations indicated that dihedral rotations are concerted, involve simultaneous dihedral hops, and lead to localized, crankshaft-like movements of the chain 30 ; however, whether such correlations are essential to explaining internal friction remains an open issue -if all the dihedrals changed independently, would that also lead to internal friction? Are there any experimentally measurable consequences of correlated dihedral motions? In this paper, our aim is to address these questions, first for simpler, peptide-like model homopolymers, and then to extend the analysis to atomistic models. Our simulations show additivity of the Kuhn-type barrier friction and the Rouse/Zimm friction effects; moreover, when the barrier friction component of the protein reconfiguration time is equated with the RIF/ZIF internal friction time τ i , it is found to coincide, within a numerical, chain length independent factor, with the dihedral hopping time τ dih . This establishes a relationship between a phenomenological internal friction timescale postulated by RIF or ZIF and the microscopic chain dynamics. We further show that this relationship is only possible if dihedral angles change in a correlated fashion. Given the earlier findings that the dihedral relaxation timescale only shows weak viscosity dependence 30,31 , our results further reconcile the experimental definition of internal friction as the zero-viscosity intercept of the reconfiguration time with the concept of Kuhn-type barrier friction. Finally, although the ZIF and RIF models correctly reproduce many qualitative features of reconfiguration dynamics, our results suggest that their utility for the analysis of experimental data has limitations. In particular, in contrast to the prediction of diffusive end-to-end distance dynamics in the high internal friction limit, all of the peptides studied here show subdiffusive dynamics. ## Methods Three types of simulations were used in this work: atomistic simulations of short polypeptides, Langevin dynamics simulations of coarse-grained, C α -only peptide models, and kinetic Monte Carlo simulations of a rotational isomeric state models (RISM). Atomistic simulations. Molecular dynamics simulations were performed using the GROMACS software package, version 4.5.5 45 , using the Amber03 parameter set 46 and an extended simple point charge (SPC/E) explicit water model. Starting from the NMR structure of the 66-residue Thermotoga maritima CSP (pdb access code 1G6P), the initial peptide models were built by cutting the protein into six equal 11-residue peptide fragments. Since studying shorter peptide fragments amplifies sequence effects (which may otherwise be averaged out in longer polypeptides), we have also performed simulations of an 11-residue peptide fragment with the Gly-Ser repeat, which is often viewed as a model polypeptide with properties close to those of a random coil 42,47 . Finally, to assess the role of dihedral rotations on the reconfiguration timescale, we have studied a Gly-Ser repeat of the same length but with a reduced dihedral-barrier. After initial minimization and equilibration, production runs of 2 μs were performed, as in an earlier study 30 , at T = 300 K and P = 0.138 atm using the modified Berendsen thermostat 48 and the Parrinello-Rahman barostat 49 . Further details are described in the Supplementary Information (SI). Coarse-grained model. Our random-coil, C α -only homopeptide model is similar to those described earlier 26, and represents each amino acid residue as a single bead; the model generally employs a 3-letter alphabet for the amino acid sequence, consisting of hydrophobic, neutral, and polar beads; however, to describe the unfolded polypeptide, sequences of various lengths consisting entirely of neutral beads were used. The potential energy of the chain included a harmonic spring potential describing the chain connectivity, a harmonic bending potential imposing the constraints inherent to peptide geometry, and a repulsive r −12 potential that accounts for the excluded volume interactions. As in the original studies where this type of model was introduced 52 , the dependence of the energy on the dihedral angles ϕ is described by a potential of the form where the dihedral barrier height ε was varied to study how the dynamics of hindered rotations affects the peptide's global relaxation timescale. The chain dynamics was governed by a Langevin equation. See the SI for further details. ## RISM simulations. In the (alpha-carbon only, coarse-grained) rotational isomeric state model (RISM), the configuration of the polypeptide chain is entirely specified by its dihedral angles, {ϕ 1 , ϕ 2, …, ϕ N−2 }, where N is the number of monomers. Same geometry (i.e. same bending angles) was assumed as in Langevin dynamics of the coarse-grained model described above, but the dihedrals were the only degrees of freedom in the RISM. Each dihedral was assumed to undergo jumps between three equivalent states 1, 2, 3, with the same value of the jumping rate coefficients, The stochastic time evolution of each dihedral was computed using the standard kinetic Monte Carlo scheme. Further details are given in the SI. ## Results The transition from the barrier friction limit to the Rouse/Zimm regime. In Kuhn's original argument, the effect of barrier friction on the global relaxation dynamics will become increasingly small as the chain becomes longer, because the barrier friction decreases with the increasing chain length (see Introduction) while the hydrodynamic friction increases. As a result, the global reconfiguration time τ r approaches the limit where it obeys the Rouse or Zimm model and, accordingly, grows as a power law with the chain length N (cf. equation ( 1)), while the internal friction characteristic time τ i remains independent of N. The transition between these limits can also be observed by studying the solvent viscosity dependence of τ r ; however simulations of proteins 31 indicate that the dihedral relaxation times show a weak, but non-negligible solvent viscosity dependence thus complicating deconvolution of the two effects. In a computational (as opposed to real) experiment, there is an alternative (and often easier) approach: keep the chain length fixed but vary the height of the hindered rotation barrier to control the dihedral rotation time. This latter approach was used in ref. 30 to prove that reduced dihedral angle barriers lead to enhanced conformational mobility of an unfolded protein under poor solvent conditions. To systematically study this transition in the absence of complications resulting from sequence effects and to avoid the high computational cost of simulating long polypeptide chains at high solvent viscosities, we first resorted to coarse-grained simulations. Specifically, we used Langevin dynamics simulations of a C α -only coarse-grained model of a random-coil homopeptide, as described earlier 26, . Explicit treatment of water molecules is, however, required in order to reproduce the correct viscosity dependence of the dihedral dynamics 31 ; thus solvent viscosity effects cannot be adequately captured in this treatment. Nevertheless, the relationship between the overall reconfiguration time and τ dih is easily investigated in this approach by varying chain length and the magnitude of the dihedral barrier. The dihedral energy landscape of our model peptide is controlled by a force-field term V dih = ε(1 − cos3ϕ)/2, which has a 3-fold symmetry, and whose minima are separated by barriers equal to ε. For chains with N = 10, 20, 33, and 66 monomers, we computed the relaxation times of the end-to-end vector R, end-to-end distance R = |R|, and each dihedral angle as a function of ε (Fig. 1). The relaxation time of each of these quantities was defined as where C(t) is the autocorrelation function given by for X = R or R and for a dihedral ϕ. Note that, unless C(t) is an exponentially decaying function, there is no unique definition of the associated characteristic time τ -the above heuristic definition is, however, commonly used. We further note that the dihedral relaxation time depends on the location of the dihedral within the chain, with the dihedrals belonging to the chain extremities moving faster than those in the middle (see SI, Fig. S2); however, the difference was always less than a factor of two. Figure 1 reports the average dihedral relaxation times, with the dihedrals belonging to the chain extremities (2 outer dihedrals at each chain end) excluded from the average. Consistent with the picture of activated barrier crossing, the dihedral relaxation time τ dih increases exponentially with ε (Fig. 1, green); it is further found to be independent of chain length N. The end-to-end distance relaxation time τ EE (Fig. 1, red) is greater than τ dih (except for the shortest peptide considered) and nearly independent of the dihedral barrier at low values of ε. In contrast, at high values of ε, τ EE becomes shorter than τ dih ; moreover, we observe direct proportionality between the two quantities, τ EE ∝ τ dih (i.e., the green and the red lines become parallel in the logarithmic plot of Fig. 1). This indicates that the barrier friction limit is attained, where the chain remains essentially frozen until a dihedral changes, and where, as a result, the timescale τ dih controls the global relaxation dynamics as measured by τ EE . Conversely, the low ε limit is associated with the Rouse model behavior (note that, since our Langevin dynamics simulations did not include hydrodynamic interactions, we are comparing the results with the Rouse rather than the Zimm model). In support of this conclusion, the end-to-end vector relaxation time τ V measured in this limit (Fig. 1, black) is close to the slowest relaxation time estimated using the Rouse model (given by equation (1), where the translational diffusion coefficient is given by D tr = k B T/(Nξ 0 ) and ξ 0 is the monomer friction coefficient), as expected to be the case for the Rouse model 25,42 -see SI, Table S1. Moreover, at low values of ε, the vector relaxation time τ V is longer than τ EE by a factor of ~3 (see SI, Table S1), a result consistent with previous theoretical predictions for the Rouse chain 17 . However, as the dihedral barrier increases, τ V shows only a weak increase and eventually becomes shorter than τ EE . This result is easy to understand: even if all the dihedral angles are frozen, the chain can still rotate, with its rotational relaxation controlled by the solvent friction (while the end-to-end distance does not significantly fluctuate). We note that the Rouse model with internal friction fails to properly account for the rotational dynamics of the chain and predicts the end-to-end vector relaxation to be determined by internal friction in this limit 42 . Since most experimental measurements of internal chain dynamics probe absolute distances between different chain segments and are insensitive to orientational dynamics, we now focus on the end-to-end distance relaxation time τ EE . We have discussed above that this time is proportional to the Rouse time τ R (defined here as the longest relaxation time of a Rouse chain, equation ( 1)) at low values of the barrier to hindered rotations or for long chains; it is proportional to the dihedral relaxation time τ dih in the opposite limit. Hence, it is natural to try the simple linear combination of the form to interpolate between the two limits. Indeed, equation ( 2) is found to globally fit the data well for all chain lengths and dihedral barriers, as shown in Fig. 2. Here the optimal values of the parameters, a = 0.26 and b = 0.15, describe the data for all peptide lengths and all dihedral barriers. Note that the value of a is comparable to the value expected using analytic estimates for a Rouse chain. Indeed, an analytic approximation due to Portman 17 predicts that, in the long-time limit, the end-to-end distance autocorrelation function for a Gaussian chain reaches its asymptotic limit at twice the rate at which the end-to-end vector autocorrelation function decays, implying τ EE ≈ 0.5 τ V . Combined with the relationship 42 τ V ≈ 0.8 τ R , this yields τ EE ≈ 0.4 τ R . Given nonexponentiality of the correlation functions, however, the precise definition of the relaxation time affects the expected value of the proportionality coefficient between the two timescales. Implications for experimental studies of internal friction. While here we have been able to observe the transition from the Rouse regime to a barrier-controlled regime by varying chain length and the magnitude of the dihedral barrier, those are not common experimental variables. Rather, two most common ways to observe deviations from solvent-dominated dynamics are to study how the protein reconfiguration time depends on the solvent viscosity η 27 and how the reconfiguration time of a shorter segment of the entire protein depends on its length 26 . Empirically, experimental viscosity dependence of τ EE (η) is often close to linear 27 , and so the zero viscosity intercept of this dependence provides an operational definition of internal friction. Moreover, this linear dependence suggests additivity of the contributions from internal friction (equal to the zero-viscosity intercept) and the solvent controlled friction (which is directly proportional to viscosity). This is consistent with the additivity of the barrier friction and the Rouse-Zimm-type friction effects predicted by Equation (2). It is, however, not possible to directly compare the experiments with coarse simulations for two reasons. First, since Langevin dynamics simulations in the overdamped regime were used to obtain Eq. 2, both τ R and τ dih are proportional to the solvent friction within the coarse model employed here. Second, if the dihedral relaxation time τ dih is viscosity dependent, the zero viscosity intercept of τ EE (η) predicted by Eq. 2 cannot be simply equated with the term bτ dih . Indirect comparison between predictions of Eq. 2 and experimental studies is still possible if we consider additional information coming from atomistic studies. Specifically, atomistic simulations from the Best group 31 and our own work 30 show only a weak solvent viscosity dependence of τ dih . Assuming the validity of equation ( 2) as applied to real proteins, then, the viscosity dependence of the observed reconfiguration time should mostly result from its first term. In contrast, the internal friction time τ i , identified with the second term of equation ( 2), should not show significant viscosity dependence, which is consistent with experimental observations 27 . Identification of the dihedral relaxation time with the internal friction time τ i is further supported by a more detailed analysis of intra-chain dynamics. For example, while at low values of the dihedral barrier the reconfiguration time between the mid-segment of the chain and its end is shorter than that between the chain ends, in accord with the predictions of the Rouse and Zimm models 26 , the two times converge as the dihedral barrier increases (See SI, Fig. S3). This observation agrees with the prediction of RIF that, at high internal friction, relaxation modes of different wavelengths have approximately the same relaxation time τ i ## 27 . Likewise, it agrees with the argument (see the Introduction) that a chain whose dynamics is dominated by barrier friction exhibits a single, segment-length-independent timescale. ## Concerted or uncorrelated dihedral rotations?. Having described the transition between the solventand barrier-friction dominated regimes, we now focus on the barrier friction limit and try to further elucidate the microscopic origins of equation (2). In particular, we would like to know whether the rotations of different dihedrals tend to occur independently of one another or are correlated, and whether such correlations (or their lack) can be deduced from the observed end-to-end dynamics. Independent dihedral rotations (particularly those occurring near the middle of the polypeptide chain) would involve large swinging motion of two parts of the chain, thereby entailing both steric clashes and high solvent friction. But while this argument is often used to justify correlated dihedral hops that result in more localized, crankshaft-type motions of the polymer, such concerted changes of the dihedrals must require higher activation barriers 44 -the outcome of this tradeoff between lower friction but higher activation barrier is unclear in advance and may depend on chain length and the magnitude of the dihedral barrier. To understand the connection between local dihedral changes and the global dynamics, it is helpful to first consider the one-dimensional toy model introduced by Hall and Helfand 54 . In this model, the polymer is a one-dimensional chain of N bonds, with each bond having a length jumping between two possible values, l − and l + . These jumps mimic the dihedral rotations in 1D; the result of each jump is a change of the total end-to-end distance by ±Δl ≡ ± |l + − l − |. Let us further assume that the jumps of each bond can be described by a first-order kinetic scheme, − +  l k l k . The total chain length undergoes a one-dimensional random walk with a step Δl and with an average frequency v = kN, since N bonds are each jumping independently. At short enough times, the mean square displacement of such a random walker is given by ν . Equating this with 2Dt defines an effective end-to-end diffusion coefficient D = NΔl 2 k/2 = (1/4)NΔl 2 /τ bond , where τ bond = (2k) −1 is the relaxation time of a single bond. Importantly, this diffusion coefficient is proportional to chain length N. After a sufficiently long time, the chain length will adopt a Gaussian distribution with a mean N(l + + l − )/2 and a variance ρ 2 = NΔl 2 /4. The relaxation time of the end-to-end distance can then be estimated as a time it takes the random walk to travel the distance ρ: This result is a one-dimensional analog of the barrier friction limit, with the bond relaxation time being analogous to τ dih . Since this argument can be applied not only to the entire chain but also to any of its segments, this simple calculation provides yet another explanation of why single, segment independent relaxation timescale emerges from microscopic conformational dynamics in this limit. Extension of these arguments to 3D is somewhat tricky. Unlike the 1D case, where the step size Δl along the end-to-end distance direction is fixed, the change of this distance as a result of a change in a particular dihedral depends both on the dihedral and on the current chain configuration. To make progress, let us assume that one can characterize the random walk along R by an average value of the step size Δl instead. The frequency of individual dihedral hopping v is equal to the inverse of the mean dwell time in one of the dihedral states. If we represent the kinetics of an individual dihedral by a 3-state system, with a rate of jumping between adjacent states equal to k, then this mean dwell time is (2k) −1 and so v = 2k (the factor of two comes from the fact that there are two possible states that a dihedral can jump to). The effective diffusion coefficient is then given by D = NΔl 2 × (2k/2). The dihedral relaxation time τ dih for our 3-state model can be estimated as the inverse of the lowest eigenvalue of the corresponding 3 × 3 rate matrix and is equal to τ dih = (3k) −1 . This gives D ≃ (1/3)NΔl 2 /τ dih . The relaxation time of the entire chain can now be estimated as the time to diffuse the distance comparable to the root-mean-square end-to-end distance, equal to ρ = ~Ll n l ( ) 1 2   , where L is the polypeptide contour length, l k is the length of its Kuhn segment, and n k = L/l k is the number of the Kuhn segments in the chain. This gives where σ is the peptide bond length. There are two possibilities now. In the case of crankshaft-type moves where a pair of dihedrals change simultaneously such that only a local chain segment is affected while there is no global rearrangement of the entire chain, Δl is a geometry dependent number that would be typically much smaller than ρ and comparable to the Kuhn length. For example, assuming Δl = l k we find τ r = (3/2)τ dih (σ/l k ). The ratio l k /σ is the number of peptide bonds within one Kuhn segment, whose value is ~3 for the peptides studied (see the SI, Table S1). Thus the fact that τ r is shorter than τ dih and so the factor b in equation ( 2) is less than one is naturally explained within this picture. The second possibility arises where all dihedrals change independently. When a single dihedral changes its value (and assuming no other relaxation mechanisms present), the ensuing large-scale pivoting motion results in a distance change Δl that is comparable to the dimensions of the chain itself. In this case, of course, the motion of end-to-end distance cannot be viewed as diffusion. A better model would be one where each dihedral rotation causes the chain to completely lose the memory of its end-to-end distance (over the time of a pivoting motion). The effective reconfiguration time, therefore, is comparable with the inverse frequency of dihedral transitions, which is equal to The heuristic prediction that the barrier friction time is inversely proportional to the chain length in this case is verified explicitly in the SI (see Fig. S6) using a rotational isomeric state model of a polypeptide with independently changing dihedrals. Since no such chain length dependence is observed in our simulations, we conclude that concerted dihedral motions dominate the barrier friction time, in accord with ref. 30. Similar analysis based on atomistic simulations leads to the same conclusion -see the following discussion. We have also observed temporal correlations between sequence distant dihedrals directly in the following way. Consider a pair of dihedrals labeled, say, i and j. If the flipping of each dihedral is an independent Poisson process with a characteristic time , then the distribution of the time lag t between the flip of i and subsequent flip of j is exponential 55 , . In contrast, correlation between the two events will lead to deviations from this exponential distribution. Indeed, while the dynamics of each individual dihedral is well described by a Poisson process, the lag time distributions for the pairs of dihedrals that are close to one another in sequence deviates from exponential, showing positive correlation between their jump times (see the SI, Fig. S5). The correlation disappears at large sequence separation. This behavior is similar to the earlier findings of a finite correlation length in atomistic simulations of the cold-shock protein 30 . ## Dihedral dynamics vs. global dynamics in the cold shock protein and its short fragments. Having developed insights about the connection between global peptide dynamics and microscopic timescales associated with dihedral rearrangements, we next examine whether any of these carry over to atomistic models of proteins and polypeptides. To this end, we have used the already published data on the dynamics of the cold shock protein (CSP), simulated, at a fully atomistic level in explicit solvent, using two different force fields. One simulation 30 uses a conventional force field, as described in the Methods section. This simulation will be referred to as CSP1. Recently, Piana et al. 56 proposed to modify water dispersion interactions in order to achieve better agreement with experimental estimates of protein dimensions-we use one of their trajectories from ref. 56 (specifically, the one displaying the best agreement with the experimental single-molecule FRET data 27 ) -we will refer to this as CSP2. In addition, we have performed explicit-solvent atomistic simulations of 8 short polypeptides, each 11 residues long. The first 6 were fragments of the cold-shock protein previously studied experimentally 27 and via molecular dynamics simulations 30 . Our rationale for choosing these fragments of a well studied protein is to examine how changing the length of a polypeptide (from N = 66, which is the full length of CSP, to N = 11) affects the dynamics. In particular, the arguments presented in the Introduction and leading to the Kuhn barrier friction picture as well as to the ZIF/RIF models predict that the internal friction timescale τ i would remain unchanged as a result of this length change; these predictions are, however, based on the idealized picture of highly localized conformational changes within homopolymer chains and are not rigorously justified. At the opposite extreme, a model where all the dihedrals change independently and where, consequently, a single dihedral jump may lead to a global rearrangement of the entire chain predicts that τ i would become longer for such shorter peptides (see equation ( 5)), provided that the dihedral relaxation times stay the same. Of course, polypeptides of only a few Kuhn segments in length may not be adequately described by simple polymer models. Moreover, sequence-dependent effects should be significant if not dominant for such short peptides. In order to assess the role of such effects, we also performed all-atom simulations of an 11-residue peptide with the Gly-Ser repeat, a system that is often deemed to be a model random coil polypeptide. Finally, to study how the height of the dihedral barrier affects the dynamics, we studied the same Gly-Ser repeat sequence with its dihedral barrier reduced by a factor of 2. The simulation results are summarized in Table 1. Each of the polypeptides showed a spectrum of dihedral relaxation times, with dihedral autocorrelation functions decaying on timescales from ~0.1 ns to hundreds or even thousands of nanoseconds (See the SI, Fig. S41). Some of the dihedrals within the CSP fragments either stayed unchanged during the course of a 2-microsecond-long simulation or exhibited very few rotations, preventing us from getting a converged autocorrelation function and estimating the resultant relaxation time. As a result, estimation of individual dihedral relaxation times for all the angles is not possible even with a trajectory as long as 60 microseconds, let alone the 2-microsecond-long simulations of the shorter peptides. To improve the statistics in estimating autocorrelation functions of the dihedral angles, we have computed the mean autocorrelation function by averaging it over all dihedrals, as in ref. 30. From this function, a single, average dihedral relaxation time was estimated. These results are reported in Table 1. The average dihedral relaxation times for CSP fragments were typically shorter (but of the same order of magnitude) than the full CSP1 simulated with the same force field. The longer timescales of rotational dynamics in the longer peptide presumably originate from stronger steric interactions 30 . The mean dihedral relaxation time for the modified Gly-Ser repeat with a lower dihedral barrier was a factor of ~4 faster, supporting the view that activated crossing of the dihedral barrier controls the timescale of dihedral dynamics. Somewhat unexpectedly, the dihedral relaxation time for CSP2 was found to be nearly an order of magnitude longer than for CSP1, despite the less compact CSP2 conformational ensemble 56 . This difference is presumably due to the difference in the force fields used in the two simulations. The simulated peptides are in the barrier friction regime. The end-to-end vector relaxation times τ V for all of the short peptides are much (3-10 times) shorter than their end-to-end distance relaxation times τ EE . At the time, all values of τ V are further within a factor of 3 from the time τ RZ estimated from equation (1), where the translational diffusion coefficients D tr for each peptide were inferred directly from the simulations (Table 1). Recall from the above discussion that we expect τ V to be a factor of ~3 longer than τ EE in the limit where Rouse/ Zimm-type friction dominates. These observations show that all of the peptides are far away from this limit and that internal dynamics of dihedrals controls the relaxation of their end-to-end distance. The end-to-end distance relaxation is comparable to (and correlated with) the dihedral relaxation time. Furthermore, both τ dih and τ EE are proportionally shorter for the CSP fragments as compared with CSP1 (Fig. 3). Let us recall that the Kuhn barrier friction model and the related RIF/ZIF models (in the limit of high internal friction) predict independence of the global relaxation timescale on chain length; this prediction, however, is contingent on the independence of the dihedral barrier crossing time on chain length, which is not the case here. Since both τ dih and τ EE increase when chain length increases from N = 11 to 66, this, in fact, is consistent with Kuhn's barrier friction picture. Moreover, this behavior implicates concerted dihedral motions as controlling the end-to-end distance dynamics, since τ EE is expected to become shorter with increasing N in the case where dihedrals change independently (cf. equation ( 5)). Consistent with our coarse-grained simulations, which predict proportionality between τ dih and τ EE in the barrier friction limit (equation ( 2)), peptides simulated atomistically also show direct proportionality between these two timescales (Fig. 3). The proportionality factor b is, however, different, being close to 1 (Fig. 3), as opposed to b = 0.15 found in coarse-grained simulations. This difference between atomistic and coarse-grained models may be purely geometrical in nature: recall that the coarse model employed in our work (see the Methods section) consists of alpha-carbons only and thus has only one dihedral angle per amino acid residue. In contrast, microscopic description of a real polypeptide involves two dihedrals (ϕ and ψ) per residue. Notwithstanding this difference, the proportionality factor between τ dih and τ EE is found to be independent of the length of the chain in both cases. This key observation supports the Kuhn-type barrier friction view, with a single, chain length independent reconfiguration timescale (see Introduction). End-to-end dynamics of polypeptides is subdiffusive, reflecting memory effects and deviations from RIF/ZIF predictions. A common description adopted by most experimental studies that probe relative motion of polypeptide chain segments is that of one-dimensional diffusion in an effective potential, which is determined by the entropic elasticity of the chain 8, . Although it is known that the monomer motion of a polymer chain is not simple diffusion 15,25,62 , an effective diffusion coefficient that depends on the time and/or length scale of the process of interest still can sometimes be introduced to describe this process 15,58, . The monomer motion of a Rouse chain, in particular, is subdiffusive at intermediate timescales that are longer than the monomer relaxation timescale but shorter than the Rouse time, with the mean square monomer displacement scaling as ∆ ∝ . R t t ( ) 2 0 5 . In the high internal friction limit, in contrast, the Rouse model with internal friction predicts simple diffusive dynamics 42 . The possibility of dihedral rotations associated with large-amplitude monomer displacements introduces a different kind of non-diffusive dynamics. Consider, for example, the (already discussed) model where the end-to-end distance R loses the memory of its previous value every time a dihedral rotation occurs. The statistics of dihedral jumps is Poisson, with the average number of jumps per unit time equal to ν. Given no memory of the previous configuration after a jump, the new value of R is simply a random number drawn from the equilibrium probability distribution p(R). Let R 1 be the end-to-end distance at time t = 0 and R 2 at time t. The probability that R 2 = R 1 is equal to the probability that a jump did not happen during the time interval t, which is e −vt for a Poisson process. The probability that R 2 is different from R 1 is thus (1 − e −vt ). Averaging over these two possibilities (the jump did not or did happen) and taking advantage of the statistical independence of R 1 and R 2 in the case where the jump did happen, one obtains at short times. Therefore, the dynamics of R appears to be simple diffusion. Examination of higher-order moments, however, reveals that it is not! Indeed, the value of any moment ∆R t ( ) n 2 is proportional to the probability (1 − e −vt ) of at least one dihedral transition during the time t and, therefore, exhibits exactly the same time dependence. In contrast, for the simple diffusion we have ∆ , so that the ratio with α ≃ 0.4-0.7 for both the atomistic and coarse-grained simulations (Fig. 4; only atomistic data are shown). This, again, supports the picture where independent dihedral rotations leading to large changes in the end-to-end distance are improbable. A more surprising finding is that the values of α are significantly less than the value (of 1) expected for simple diffusion both in the Rouse limit (where this is expected) and in the barrier friction limit. This indicates a subdiffusive, and, therefore, non-Markov process with prominent memory effects and contradicts the RIF or ZIF, which predict subdiffusion in the Rouse/Zimm limit but diffusive dynamics in the barrier friction limit 42 . To gain further insight into peptide dynamics in the barrier friction regime, we computed the mean square displacements of the end monomers of the Gly-Ser construct as , where R i (t) is now a vector describing the position of the first (i = 1) or last (i = 11) alpha-carbon of the peptide (Fig. 5). At short time t it is found to grow slightly slower than linearly (α ≈ 0.9), while approaching a strictly diffusive behavior and converging with the linear dependence for the mean square displacement of the peptide's centroid. In other words, the movement of the end monomers is nearly diffusive even at short times, in contrast to the predictions of the Rouse or Zimm models but consistent with those of RIF and ZIF in the high internal friction limit. Since the relative distance measured between two statistically uncorrelated diffusing particles also undergoes simple diffusion, the much stronger deviation from simple diffusion observed for the end-to-end distance indicates that the displacements of the peptide ends are strongly correlated at short times. The source of such correlations is easily understood (albeit not explaining the subdiffusive behavior per se): when the time t is much shorter than the dihedral relaxation times, the displacements of the peptide atoms are mostly due to its overall rotation. Indeed, at short times, the mean square change in the end-to-end distance, ∆R 2 , is much smaller than the corresponding changes in the positions of its end monomers. Moreover, the mean square change in the end-to-end vector, ∆R 2 , is much greater than ∆R 2 , again indicating that, at short times, the end-to-end vector mostly rotates without changing its magnitude. It is important to note that RIF and ZIF do not describe this rotational dynamics correctly 42 in the internal friction dominated limit: they make the unphysical prediction that the rotational relaxation time is independent of solvent hydrodynamics and equal to τ i . ## Discussion The subject of internal friction or viscosity has been extensively debated and never satisfactorily settled in the polymer physics of 1970-90's 24, 35-37, 40, 41 , even for relatively simple polymeric systems such as hydrocarbons. Different physical mechanisms were invoked to explain internal friction effects and a number of distinct polymer models were proposed. In light of these complications one may wonder if a simple polymer-theoretical description is at all possible, especially in the case of unfolded proteins, whose energetics, involving hydrogen bonding, hydrophobic interactions, and other mechanisms 65 is more complicated. Our study, while revealing limitations of simple models, also points toward near universal features exhibited by protein dynamics. One surprising finding is the subdiffusive, rather than diffusive monomer motion. While such breakdown of the simple diffusion model has been predicted, e.g., on the basis of the Rouse model 15,62,64 , internal friction effects treated using the Rouse model with internal friction (RIF) were expected to restore diffusive dynamics 42 in the high internal friction limit, a prediction that is at odds with the present study. A possible explanation of this discrepancy is the fact that RIF, being essentially a one-dimensional model, does not adequately describe bond rotation. Indeed, it predicts internal friction effects, when they are dominant in the global reconfiguration dynamics, to dominate the rotational dynamics of the chain as well -this prediction is clearly absurd as even when all the internal degrees of freedom of the chain are frozen (i.e., infinite internal friction) it can still undergo rotational relaxation entirely controlled by the solvent hydrodynamic friction. Since the end-to-end distance of the chain changes in response to three-dimensional motion of the internal chain segments, it is conceivable that RIF would be inaccurate in describing the dynamics of the chain's end-to-end distance. At the same time, RIF (and the related ZIF) captures other aspects of protein dynamics. For example, consistent with these models, high internal friction leads to near-degeneracy of the relaxation timescales, where internal chain segments reconfigure on the same time scale as the entire chain 27,30 (see Fig. S3). ## Figure 5. Mean square displacement of the end monomers of the Gly-Ser repeat construct grows nearly linearly at short times (α ≈ 0.9), approaching the strictly linear dependence and converging with the mean-square displacement of the peptide's centroid (defined as the average position of the peptide's alpha carbons) at long times. Despite this almost purely diffusive motion of the peptide's ends, the relative distance R between them undergoes subdiffusive motion. The much lower values of ∆R 2 , as compared to the monomer mean square displacements, indicates that, at short times, the latter are dominated by rotations of the polymer chain. Moreover, in the case of protein dynamics within a coarse-grained model we see that the global reconfiguration obeys a simple RIF-like relationship, equation (2), provided that the internal friction time τ i is identified, to within a chain-length-independent proportionality factor, with the dihedral relaxation time τ dih . Equation (2) provides a simple interpolation between the regime where the solvent friction dominates, achieved when the chain is sufficiently long, and the barrier friction regime, where the reconfiguration dynamics is controlled by dihedral rotations. Equation ( 2) is also consistent with atomistic simulations of peptides and proteins, although in the latter case we have only been able to explore the barrier friction limit. Our study further supports the view that dihedral rotations in a polypeptide are concerted 30 . The most direct evidence of this comes from the correlation between the times at which the dihedrals that are close in sequence undergo rotations -see Fig. S5. Note that similar correlation was found in atomistic studies 30 . Further, indirect (but experimentally testable) evidence comes from the weak chain length dependence of the reconfiguration time (in the barrier friction limit), which contradicts chain-length dependent reconfiguration times predicted by a model where the dihedrals change independently. Finally, the subdiffusive character of the end-to-end distance dynamics observed in all of our simulations also contradicts the independent dihedral jump picture. Our study shows that measurements of internal friction (such as the ones reported in ref. 27) provide information about the timescales of microscopic motion within the chain, specifically, its dihedral relaxation. But a key question remains to be answered: what properties of the polypeptide chain determine those timescales? Why do different proteins of comparable length and at the same conditions display different values of the internal friction time τ i and, hence, different timescales of dihedral dynamics? Furthermore, what determines the length scale over which dihedral flips are correlated? A parallel can be made here with the problem of first-principles prediction of the dry friction coefficients, which is still an open issue. Experiments, atomistic simulations, and theoretical efforts will be required to further elucidate the microscopic basis of internal friction.

chemsum

{"title": "Theoretical and computational validation of the Kuhn barrier friction mechanism in unfolded proteins", "journal": "Scientific Reports - Nature"}

the_key_role_of_the_latent_n–h_group_in_milstein's_catalyst_for_ester_hydrogenation

## Abstract: We previously demonstrated that Milstein's seminal diethylamino-substituted PNN-pincer-ruthenium catalyst for ester hydrogenation is activated by dehydroalkylation of the pincer ligand, releasing ethane and eventually forming an NHEt-substituted derivative that we proposed is the active catalyst. In this paper, we present a computational and experimental mechanistic study supporting this hypothesis. Our DFT analysis shows that the minimum-energy pathways for hydrogen activation, ester hydrogenolysis, and aldehyde hydrogenation rely on the key involvement of the nascent N-H group. We have isolated and crystallographically characterized two catalytic intermediates, a ruthenium dihydride and a ruthenium hydridoalkoxide, the latter of which is the catalyst resting state. A detailed kinetic study shows that catalytic ester hydrogenation is first-order in ruthenium and hydrogen, shows saturation behavior in ester, and is inhibited by the product alcohol. A global fit of the kinetic data to a simplified model incorporating the hydridoalkoxide and dihydride intermediates and three kinetically relevant transition states showed excellent agreement with the results from DFT.Scheme 1 Reversible activation of hydrogen mediated by RuPNN dearom . ## Introduction Catalytic transformations relying on metal-ligand-cooperative hydrogenation or dehydrogenation of polar substrates have seen a dramatic expansion in development over the past decade and a half, following the disclosure by Milstein and co-workers of the PNN-pincer-ruthenium complex RuPNN dearom (Scheme 1), shown to be active under relatively mild conditions for the hydrogenation of esters to alcohols, 1 as well as the reverse reaction, the acceptorless dehydrogenative coupling (ADC) of primary alcohols to esters. 2 This complex has since been applied to a wide range of mechanistically related transformations. 3 In the original reports, the dearomatized complex RuPNN dearom was shown to react reversibly with hydrogen at room temperature to give the rearomatized dihydride complex RuPNN H2 . Based on this observation, a mechanism was proposed that involved this heterolytic cleavage of hydrogen as a key step in catalytic ester hydrogenation. In the years since the initial reports on RuPNN dearom , many researchers have studied the effect of catalyst structure on activity in ester hydrogenation, and several highly optimized catalysts have been discovered that give more than 10 000 turnovers at full substrate conversion. 4 Common to almost all of these elite catalysts is an N-H functional group on the ligand. In some cases, the N-H group was demonstrated to be essential for high catalytic activity through the synthesis of control ligands where N-H was replaced with N-Me, N-Bn, or O. 4a,c,e In many cases, minimum-energy pathways involving deprotonation of the N-H group have been identifed through density functional theory, 4c,f,h,i,5 although recent computational work has suggested that in some cases, the N-H group may function in catalysis as a hydrogen-bond donor without being deprotonated on the catalytic cycle. 5b,6 Although many of the most highly active catalysts for ester hydrogenation feature an N-H functional group, several structural motifs lacking an N-H group have also been reported. In particular, several ruthenium catalyst variants featuring dialkylamino side groups like RuPNN dearom have shown activity. 4d,7 During mechanistic studies of our previously reported 7c,d,8 CNN-pincer-ruthenium catalysts for ester hydrogenation, we made a surprising observation: precatalysts featuring NEt 2 or N i Pr 2 side groups underwent an unexpected dehydroalkylation reaction, releasing an equivalent of ethane or propane early on in catalytic reactions. 9 The observation of catalytic induction periods concomitant with the release of alkane established that dehydroalkylation was a necessary step in formation of the active catalyst. Milstein's catalyst RuPNN dearom , which features an NEt 2 side group, also showed an induction period for ester hydrogenation, and released ethane concomitantly with the onset of catalytic activity. By heating our CNN-and Milstein's PNN-pincer precatalysts in the presence of tricyclohexylphosphine, we were able to trap the products of dehydroalkylation as fve-coordinate ruthenium(0) complexes, where the dialkylamino side group was transformed to an imine functionality (PNN variant shown in Scheme 2). RuPNN imine , the ruthenium(0) derivative of RuPNN dearom , is dramatically more active as a precatalyst for ester hydrogenation than its precursor, and is by some measures the most efficient catalyst currently known for ester hydrogenation, giving in excess of 10 000 catalytic turnovers at room temperature with no added base. Several catalysts have been reported to operate at 4d,10 or near 4b,c,5a,11 room temperature, but require signifcant quantities of strongly basic additives such as NaO t Bu and KO t Bu. Conversely, several catalysts operate without the need for added base, but require temperatures of 80 C or higher. 1,5g,12 To further probe the catalyst speciation under operating conditions, we monitored the reaction of RuPNN imine with hydrogen at room temperature. Under 10 bar H 2 , RuPNN imine converts quantitatively to the dihydride complex RuPNN HEt , involving a net hydrogenation of the imine functional group and a net oxidative addition of H 2 to the ruthenium center (Scheme 2). As RuPNN HEt contains a ruthenium hydride and N-H group, we proposed 9 that it may be an active catalytic intermediate in reactions initiated by RuPNN imine and RuPNN dearom , operating by a metal-ligand-cooperative mechanism analogous to that proposed for other elite N-H-containing catalysts. The discovery of the dehydroalkylative activation of RuPNN dearom has potentially broad mechanistic implications. Because we have demonstrated that RuPNN dearom is not a kinetically competent intermediate and must undergo dehydroalkylation prior to being catalytically active for ester hydrogenation, it is unlikely that the originally proposed mechanism 1 is correct. Three reports in the literature apply density functional theory to predict the mechanism of ester hydrogenation or the reverse ADC, catalyzed by RuPNN dearom . 13 Since these studies rely on the catalytic intermediacy of RuPNN dearom or RuPNN H2 , they too are unlikely to be correct. More broadly, RuPNN dearom has been applied as a catalyst for a wide range of related hydrogenation and dehydrogenation reactions, including amine-alcohol coupling, 3a-d couplings of amines 3e or alcohols 3f with esters, organic carbonate hydrogenation, 3g carbon dioxide hydrogenation, 3h and amide a-alkylation with alcohols. 3i All of these transformations are conducted at or above 100 C, where the dehydroalkylation reaction occurs rapidly (t1 2 ¼ 6 min at 100 C). As such, it is possible that RuPNN dearom is inactive prior to dehydroalkylation in these systems as well, which may call into question the DFT studies of these transformations. 13a,14 Kinetic studies have the potential to validate or falsify the fndings from computation, as they show conclusively what species are consumed or released on the pathway from the turnover-frequency-determining intermediate (TDI) to the turnover-frequency-determining-transition state (TDTS). 15 For example, recent kinetic investigations of metal-catalyzed hydroformylation 16 and ketone hydrogenation 17 have provided deep insight into the underlying reaction mechanisms. Although catalytic ester hydrogenation and its microscopic reverse, ADC, have been studied intensively through computational methods, detailed experimental mechanistic investigations, especially kinetic studies, are scarce. In studies of an iridium-bipyridine catalyst system, Brewster, Sanford, and Goldberg determined the dependence of turnover number at low conversion on the concentrations of catalyst, hydrogen, and ester. 12h However, as reactions were not monitored over time, this study did not probe for potential activation of catalyst observed as an induction period, and did not probe for potential product inhibition. Filonenko, Pidko and coworkers reported time-course studies of ester hydrogenation catalyzed by CNCpincer-ruthenium complexes, but did not determine the detailed dependences of the rate on concentrations. 4e O and Morris also reported time-course studies for ester hydrogenation catalyzed by ruthenium complexes of NHC-amine ligands, but also did not determine a rate law. 5a In this paper, we describe a computational and experimental mechanistic study of ester hydrogenation catalyzed by RuPN-N imine . We report the crystallographic characterization of the key dihydride intermediate RuPNN HEt , and the synthesis and characterization of the ruthenium-hydrido-alkoxide RuPNN-HOEt , which represents the catalytic resting state and TDI. Detailed kinetic studies show an induction period at room temperature during which RuPNN imine is converted to the active form, and after which the reaction shows frst-order dependence on [catalyst] and [hydrogen], frst-order saturation behavior in [ester], and a transition from inverse second-order to inverse-frst-order inhibition by the product alcohol. All of this kinetic behavior, as well as the overall rate of reaction, is consistent with the minimum-energy pathway calculated using density functional theory. ## Computational mechanistic analysis Background Previous computational studies of the mechanism of ester hydrogenation catalyzed by transition-metal complexes with a pendant N-H functional group have converged on a bicyclic pathway, 4c,f,h,i,5 which can be separated into three linear sequences (Scheme 3). In the hydrogen-activation sequence, the H-H bond is cleaved heterolytically, placing a hydride on the metal center and a proton on the basic nitrogen center. In ester hydrogenolysis, these hydrogen atoms are transferred to the ester substrate, and cleavage of the C-O bond facilitated by the catalyst produces an aldehyde intermediate and one product alcohol molecule. In the aldehyde-hydrogenation sequence, the intermediate aldehyde is reduced to alcohol by the hydrogenated form of the catalyst. In one important variation on this scheme, it is possible that the nitrogen remains protonated throughout catalysis if an exogenous alkoxide base participates in hydrogen cleavage directly. 5b,6 We chose ethyl acetate as an appropriate model ester for computational study, for several reasons: (1) its use is wellprecedented in both experimental and computational work; (2) it is small enough to minimize issues resulting from a large number of potential conformations; (3) it is large enough to appropriately model the steric interactions of common ester substrates with the catalyst. In particular, we expected the energetics of ethyl acetate hydrogenation to closely mimic those of hexyl hexanoate, which we employed in the kinetic studies described below. ## Consideration of plausible resting states We began our study by comparing the relative free energies of plausible catalyst resting states, in each case considering the effect of hydrogen-bonded product ethanol molecules (Scheme 4). Dihydride species g, g1, and g2 were identifed, and we found a small (<2.0 kcal mol 1 ) effect of binding to ethanol on the standard-state free energy. The calculated structure of g very closely matches the crystal structure of RuPNN HEt , described below. We found that unsaturated intermediates c, c1, and c2, related to the dihydride compounds by formal loss of H 2 , were approximately 4-5 kcal mol 1 higher in energy than their dihydride counterparts. We also considered hydridoalkoxide species a1 and a2. Although the "free" species a1 is similar in energy to its unsaturated counterpart c1, the hydridoalkoxide a2 with an additional hydrogen-bonded ethanol molecule is much more stable than c2, and is identifed as the catalyst resting state. As described below, the calculated structure of a2 closely matches the crystal structure of RuPNN HOEt , and the experimental kinetics are consistent with a2 as the catalyst resting state. ## Pathway for hydrogen activation We began by probing a range of possible pathways for the activation of H 2 , informed by the rich history of prior work on related reactions. Our search for the MEP for H 2 activation covered metal-ligand-cooperative heterolytic cleavage involving the N-H functional group, as proposed by Noyori and coworkers for their seminal carbonyl hydrogenation catalysts, 18 both with and without explicit ethanol molecules to serve as proton shuttles. Noyori-type mechanisms for H 2 activation have been identifed for a range of catalysts for ester hydrogenation or ADC. 4c,f,h,i,5a,c-j,6d We also carefully searched for pathways involving cooperative activation of H 2 through the ruthenium center and an ethoxide anion, which Dub et al. showed can Scheme 3 Two linked catalytic cycles for ester hydrogenation. Scheme 4 Species considered as plausible resting states. Energies given represent standard-state free energies in kcal mol 1 at 298.15 K relative to a2. bypass the deprotonation of the N-H group on the pincer ligand. 5b,6 Lastly, we exhaustively examined pathways for H 2 activation involving deprotonated CH 2 linkers on the pincer ligand, as originally proposed by Milstein and coworkers, 1 and identifed by DFT in many studies. 13a,14a-e,g,19 Fig. 1 shows the pathway we identifed with the lowest overall barrier, which we describe as a "proton brigade" because of the involvement of two ethanol molecules in the stepwise cleavage of the H-H bond. Beginning with the resting state a2, whose experimental characterization is described in the next section, a double proton transfer through a2-b2-TS (ref. 20) gives the Ndeprotonated species b2 with a neutral ethanol molecule coordinated to Ru. Then, this ethanol dissociates to give the unsaturated species c2, which transfers a proton back to nitrogen to give d2 before binding H 2 in the s-complex f2, in which the nitrogen is protonated and an ethoxide anion is stabilized by two hydrogen bonds. Then, H 2 is cleaved through the proton-shuttle transition state f2-g2-TS, resulting in the dihydride species g2 (with two associated ethanol molecules) or g1 (with one associated ethanol). Although this pathway does involve temporary deprotonation of nitrogen between a2 and d2, the reformation of the N-H bond is not concerted with H 2 cleavage. We also located a pathway connecting a2 to d2 where the N-H bond remains intact (Fig. S1 †), and a pathway connecting c2 to f2 where hydrogen coordination occurs before proton transfer to nitrogen (Fig. S2 †), both with slightly higher barriers. The MEP identifed for hydrogen activation requires passing through f2-g2-TS at 15.0 kcal mol 1 . Notably, this protonbrigade pathway relies on the inclusion of two explicit ethanol molecules, both for the low overall barrier and for the stepwise proton-shuttle mechanisms. For comparison, we calculated analogous pathways involving only one ethanol molecule as proton shuttle and involving no ethanol molecules. These pathways, both concerted, are described in detail in the ESI (Fig. S3 †), and were found to proceed through higher overall barriers of 18.6 and 25.3 kcal mol 1 , respectively. We also searched exhaustively for pathways involving the activation of H 2 mediated by deprotonated CH 2 linkers of the pincer ligand, both with and without explicit ethanol molecules as proton shuttles. These pathways, described in detail in the ESI, † would implicate dearomatized species similar to RuPNN dearom as key intermediates in catalysis. All mechanisms of this nature were found to proceed through higher barriers for H 2 cleavage, with a lowest identifed barrier of 25.6 kcal mol 1 for mechanisms involving the NCH 2 linker and 23.7 kcal mol 1 for the PCH 2 linker. In summary, our work shows that the presence of the N-H functional group is essential for activation of hydrogen with a low barrier. The N-H group is temporarily deprotonated in our MEP, but a pathway where the N-H group remains protonated and instead serves to stabilize intermediates and transition states through hydrogen bonding is energetically similar, and cannot be excluded by DFT. Pathways involving deprotonation of CH 2 linkers have signifcantly higher barriers and can be excluded. ## Pathway for ester hydrogenolysis The ester hydrogenolysis portion of the catalytic cycle involves the hydrogenation of the carbonyl functional group and cleavage of the C-O bond, ultimately releasing one product alcohol molecule and an aldehyde intermediate. Prior work on many systems has identifed the transfer of hydride from the metal to the carbonyl carbon as a key initial step, which generates a hemiacetaloxide intermediate. Two principal pathways for cleavage of the C-OEt bond have emerged, which have been shown to have similar barriers for related catalysts. These pathways differ by the coordination of either the aldehyde oxygen 4c,f,i,5d,f,h,6d or the alkoxide oxygen 4h,5a,h,i,6d,21 to Ru during C-O cleavage. In our system, we fnd these mechanisms to have nearly identical barriers, as described below. The pathway shown in Fig. 4 is an example of the former mechanism. Beginning from the dihydride intermediate g1, the ester replaces the hydrogen-bonded alcohol molecule to give the reactant complex h, which transfers hydride from Ru to C to give the C-H s-complex i, where the hemiacetaloxide oxygen is stabilized by hydrogen-bonding to the N-H. Then, proton transfer from N to O gives j, 20 which rearranges to place the hydroxyl oxygen on Ru in k, followed by reprotonation of nitrogen and subsequent hydrogen bond formation to give the Ru-bound hemiacetaloxide complex m. Then, cleavage of the C-O bond, concerted with proton transfer from N back to O, gives n, a loosely-bound complex of the product ethanol and intermediate acetaldehyde. Replacement of the aldehyde with another ethanol molecule gives c2, which connects back to the hydrogen activation pathway in Fig. 1 and completes the frst hydrogenation cycle. The ester hydrogenolysis pathway shown above proceeds through a highest barrier of 17.4 kcal mol 1 , which is the free energy of the intermediate species n. Notably, intermediates k and n and transition states k-l-TS and m-n-TS all have essentially identical free energies of 17.0-17.4 kcal mol 1 . Thus, flux through this sequence is limited by both the cleavage of the O-H bond in k (essentially barrierless in the forward direction) and the cleavage of the C-O bond in m (essentially barrierless in the reverse direction). 20 We also identifed a different pathway for C-O cleavage, with a nearly identical overall barrier of 18.1 kcal mol 1 , which directly places the newly formed ethoxide rather than the aldehyde on ruthenium (Fig. S8 †). Similar to the transformation identifed by Hasanayn and termed a hydride-alkoxide metathesis, 21 this pathway would predict identical kinetic behavior as the one shown below in Fig. 2. In the process of establishing the twin minimum-energy pathways described in Fig. 2 and S8, † we characterized diastereomeric sequences where the ester initially coordinates to Ru through the opposite face, and pathways involving an explicit ethanol molecule. As described in detail in the ESI, † we found slightly higher overall barriers for the diastereomeric pathways (Fig. S10 and S12 †) and similar overall barriers for pathways involving an explicit ethanol molecule (Fig. S11 and S14 †). We also calculated a ruthenium-free pathway for the conversion of the hemiacetal to ethanol and acetaldehyde (Fig. S15 †), and fnd a much higher barrier of 36.4 kcal mol 1 , in line with previous work. 6d,14h In summary, we fnd that ester hydrogenolysis proceeds in our system by well-precedented mechanisms for ruthenium-pincer catalysts possessing an N-H group, and that the decomposition of the hemiacetal is mediated by the ruthenium-pincer catalyst. ## Pathway for aldehyde hydrogenation The fnal portion of the catalytic cycle involves hydrogenation of the aldehyde intermediate to give the second equivalent of alcohol product. This sequence, as mediated by a ruthenium 1. Note that the standard-state free energy of 13.7 kcal mol 1 reported here for c2 corresponds to release of acetaldehyde and binding of ethanol from n, whereas the free-energy of 8.2 kcal mol 1 reported for c2 in Fig. 1 is calculated against the ethyl acetate and dihydrogen reactants. hydride complex with a pendent N-H functional group on the ligand, has been studied extensively through DFT in the context of ester hydrogenation, but also has a longer history dating back to the original Noyori catalysts, which were highly efficient for aldehyde and ketone hydrogenation. 18 For ester hydrogenation catalysts, the aldehyde hydrogenation step is generally found to have a lower barrier than the ester hydrogenolysis step, which along with the thermodynamic instability of the aldehyde with respect to reactants, is consistent with the lack of buildup of aldehyde in catalytic reactions. For our catalytic system, we identifed the pathway shown in Fig. 3, beginning with coordination of the aldehyde to form r. This is followed by stepwise transfer of hydride and proton to the substrate from the ruthenium and nitrogen centers, respectively, giving intermediates s and t. Dissociation of the C-H s-complex gives c1, which connects back to the hydrogen activation pathway. In some recent studies, 6,13c,22 proton transfer from the ligand to the alkoxide oxygen was calculated to have a higher barrier than proton transfer from an exogenous alcohol molecule, which enables the construction of a pathway for hydrogenation where the ligand N-H group (or CH 2 linker) is never deprotonated. Pathways like this may have been missed in earlier work because of the optimization of structures without a solvent model, which can favor concerted proton/hydride transfer pathways and disfavor ion-pair intermediates such as s. In our work, conducting geometry optimizations using a toluene continuum solvent model allowed the identifcation of the intermediate s. As proton transfer from N to O through s-t-TS is barrierless 20 and strongly exergonic in our system, we did not search extensively for additional pathways for conversion of the aldehyde to ethanol. ## Summary and predicted kinetics In summary, we have identifed MEPs for hydrogen activation, ester hydrogenolysis, and aldehyde hydrogenation in ester hydrogenation catalyzed by RuPNN HEt , which forms in situ from RuPNN imine as we have shown experimentally. 9 Fig. 4 shows a simplifed energy diagram depicting key intermediates and transition states relevant in predicting the kinetics of hydrogenation through the energetic span model. 15 The hydridoalkoxide complex a2 is predicted to be the turnover-frequencydetermining intermediate (TDI). The highest-energy transition states in the hydrogen activation and ester hydrogenolysis sequences are f2-g2-TS and m-n-TS, respectively. Although intermediate n is calculated to be higher than m-n-TS by 0.3 kcal mol 1 , 20 we have used m-n-TS in our kinetic analysis for consistent application of transition-state theory to calculate rate constants. The 2.1 kcal mol 1 free-energy difference between f2-g2-TS and m-n-TS is likely within the error of the DFT method, especially considering the changes in molecularity involved: between a2 and f2-g2-TS, a hydrogen molecule is consumed, and between f2-g2-TS and m-n-TS, two ethanol molecules are released and ethyl acetate is consumed. The energetic span model allows a prediction of the rate of catalysis based on the free-energy difference between the TDI and the TDTS, which is an effective barrier for catalytic turnover. 15 Taking our DFT results at face value, the TDTS is m-n-TS when the reactants, EtOAc and H 2 , as well as the product EtOH, are at their standard states of 1 M. In this scenario, the energetic span is 17.1 kcal mol 1 , which is qualitatively consistent with a catalytic system that turns over rapidly at room temperature. The predicted rate law, based on the species consumed and released between the TDI and TDTS, is represented by eqn (1). In our experimental kinetic analysis described below, we have taken the above simplifed model as a starting point, and additionally we fnd saturation behavior at high [ester], consistent with a switch to f2-g2-TS as TDTS under these conditions. ## Effect of explicit ethanol on kinetics Motivated by past work showing the key involvement of protic solvent in heterolytic hydrogen cleavage 23 and by the complicated dependence of our catalytic rate on alcohol concentration (as described below), we examined the effect of including explicit ethanol molecules in the hydrogen activation and ester hydrogenolysis pathways described above. The complete pathways are included in the ESI. † Fig. 5 shows a summary of the effect of explicit ethanol molecules on the free energies of key intermediates and transition states that determine the kinetics. Taking the computed free energies at face value, several predictions can be made about the kinetics. First, the hydridoalkoxide intermediate a2 interacts strongly with an ethanol molecule from solution, so the "free" complex a1 does not represent a signifcant fraction of the resting catalyst speciation, even at very low ethanol concentration. Second, the minimum-energy pathway for hydrogen activation goes through f2-g2-TS and includes two ethanol molecules as a "proton brigade", although a pathway through e1-g1-TS, with only one ethanol molecule as a proton shuttle, is only 3.6 kcal mol 1 higher. Third, as the energies of g, g1, and g2the dihydride intermediates with 0, 1, and 2 ethanol molecules includedare above the energy of a2 and below the energies of the transition states, their specifc energies and relative ordering are not kinetically relevant. Last, ester hydrogenolysis proceeds through a very similar free-energy barrier whether an explicit ethanol molecule is included (m1-n1-TS, 17.4 kcal mol 1 ) or not (m-n-TS, 17.1 kcal mol 1 ). This model formed the basis for our kinetic analysis, described below. ## Synthesis of proposed intermediates RuPNN HEt . The computational studies described above predict that the catalyst resting state will be a hydrido-alkoxide species such as a2, stabilized by hydrogen-bonding to a product alcohol molecule. Dihydride species such as g, g1, and g2 are predicted to be key intermediates, but are less stable by several kcal mol 1 and are expected to have low steady-state concentrations once even small amounts of alcohol product build up in ester hydrogenation reactions. We previously demonstrated (Scheme 2) that the precatalyst RuPNN imine converts quantitatively to the dihydride RuPNN HEt under hydrogen pressure, 9 although the reversibility of this reaction on removal of hydrogen prevented easy isolation of RuPNN HEt . Recently, Gusev reported a clever method to isolate the dihydride product RuPNN H2 formed by reaction of Milstein's original precatalyst RuPNN dearom with hydrogen: a solution of the dearomatized species was placed under hydrogen in an unstirred pressure vessel, in a solvent mixture that dissolved RuPNN dearom completely but allowed the product dihydride to crystallize. 13c Gratifyingly, we found that the same procedure allowed us to successfully isolate RuPNN HEt in crystalline form (Scheme 5). RuPNN HEt is isolated as yellow crystals, which are moderately stable at room temperature, but can be stored under argon at 37 C for extended periods without decomposition. Although RuPNN HEt is stable even under air as a solid, it decomposes rapidly when dissolved in degassed benzene-d 6 at room temperature. As we previously characterized RuPNN HEt fully in solution under H 2 pressure, 9 we did not attempt to repeat the spectroscopic characterization in the absence of H 2 . The crystals of RuPNN HEt formed by the method described above were suitable for characterization by X-ray crystallography. Fig. 6 shows the molecular structure in the solid state. As was concluded based on our previous spectroscopic characterization, 9 RuPNN HEt features a nearly octahedral ruthenium(II) center bound to two hydride ligands, carbon monoxide, and a PNN-pincer ligand with an aromatic pyridine fragment flanked by CH 2 P( t Bu) 2 on one side and CH 2 NHEt on the other. The structure is closely analogous to that of RuPNN H2 as recently reported by Gusev, 13c except for the change from NEt 2 to NHEt. The mechanism of double hydrogenation from RuPN-N imine to give RuPNN HEt is not obvious, and is the subject of an ongoing experimental and computational study. RuPNN HOEt . Although dihydride species such as RuPNN H2 have been proposed as the resting states in catalytic hydrogenation reactions, 5h,13a,24 our calculations indicate that alkoxide a2 is more stable than the dihydride g1 by 2.9 kcal mol 1 . In recent work, Gusev demonstrated experimentally that RuPNN H2 converts rapidly to alkoxide species on addition of alcohols, and showed computationally that the ethoxide species was more stable than the dihydride by 0.4 kcal mol 1 . 13c We observed analogous reactivity for RuPNN HEt : although it decomposes in benzene-d 6 with no other additives, RuPNN HEt rapidly converts to the hydrido-alkoxide species RuPNN HOEt when dissolved in benzene-d 6 containing ethanol, with visible evolution of hydrogen gas (Scheme 6). NMR spectra taken immediately after reaction show one clean species. RuPNN HOEt was fully characterized by NMR spectroscopy at 25 C in benzene-d 6 . The hydride signal appears as a doublet at 15.8 ppm. At room temperature, broad signals are observed for the methylene and hydroxyl hydrogens of free ethanol. Signals for the bound ethoxide, N-H, and the PCH 2 hydrogen syn to the ethoxide are not observed, as they are in rapid exchange with hydrogens from free ethanol. To characterize RuPNN HOEt in the absence of this chemical exchange, 1 H NMR spectra were recorded from 90 C to 20 C in toluene-d 8 (see the ESI † for spectral images). At 50 C, the above chemical exchanges are slow on the NMR time scale, and distinct resonances are observed for free ethanol, bound ethoxide, the N-H, and all four CH 2 linker hydrogens. Single crystals of RuPNN HOEt suitable for X-ray crystallography were obtained by slow evaporation of a pentane solution containing a small amount of ethanol. Although crystals could be reproducibly obtained in this manner, the instability of RuPNN HOEt in the absence of an excess of ethanol coupled with its high solubility in solvents with a wide range of polarities have thus far prevented its bulk isolation as a solid. Fig. 7 shows the structure of RuPNN HOEt . In the solid state, the ethoxide ligand is syn to the N-H group, which is pseudo-axial and is 2.21 A from the ethoxide oxygen, indicating a weak intramolecular hydrogen bond. A molecule of ethanol is present in the asymmetric unit, and the O-H hydrogen interacts with the ethoxide oxygen through hydrogen-bonding with a distance of 1.73 A. The solid-state geometric parameters for RuPNN HOEt are remarkably similar to the computationally optimized structure a2, which was the lowest-energy hydrido-alkoxide conformation we were able to locate that included one explicit ethanol molecule. The rapid conversion of RuPNN HEt to RuPNN HOEt at room temperature is consistent with our DFT study above. This transformation is the reverse of the hydrogen activation shown in Fig. 1, which is predicted to proceed in the reverse direction with a free-energy barrier of 12.1 kcal mol 1 , proceeding from g1 through f2-g2-TS. The complete formation of RuPNN HOEt from RuPNN HEt suggested that, as predicted by computation, the alkoxide species might be more stable under catalytic conditions, and would hence represent the resting state and TDI. To confrm this, we placed a solution of RuPNN HOEt formed in situ from RuPNN HEt and ethanol under 10 bar H 2 in a high-pressure NMR tube. No conversion back to the dihydride species was observed, consistent with the prediction from computation that RuPNN HOEt is the dominant resting state throughout the catalytic reaction. ## Kinetics As we noted in the introduction, DFT studies of catalytic ester hydrogenation are widespread but kinetic studies are rare. Because the computed mechanism and energies make clear predictions about the kinetics, the latter provide an important check on the former. Based on our computed mechanism, the following predictions can be made. First, because hydrogen activation occurs between the TDI and TDTS, the reaction should be frst-order in hydrogen. If a dihydride intermediate such as g1 were more stable than the intermediate preceding hydrogen activation (a2 in our work), the reaction would follow zero-order kinetics in hydrogen. Second, because an ester molecule is consumed between the TDI and TDTS, the reaction should be frst-order in ester. If the barrier for hydrogen activation were much higher than the barrier for ester hydrogenolysis, the reaction would follow zero-order kinetics in ester. If these two barriers are similar in energy, saturation behavior is possible. Third, because alcohol product is released between the TDI and TDTS, the reaction should be inhibited by the buildup of alcohol. The precise dependence of the rate on [alcohol] is not unambiguously predicted by computation, because of the multiple pathways available as shown in Fig. 5 above. Lastly, and very importantly, the overall rate of reaction should be approximately consistent with the overall barrier predicted by DFT, which is 17.1 kcal mol 1 in our system. Although the hydrogenation of ethyl acetate to ethanol was ideal for our computational study and for the isolation of the hydrido-alkoxide intermediate RuPNN HOEt , we chose to conduct detailed kinetic studies using hexyl hexanoate as substrate, because both the ester reactant and alcohol product have very low volatilities at room temperature, which minimizes the possibility of evaporation of reactant or product at any point during the setup, reaction, or analysis. As both RuPNN HEt and RuPNN HOEt were unstable in solution, we conducted kinetic studies using RuPNN imine as precatalyst. We previously determined that isopropyl alcohol was an ideal solvent for obtaining high catalytic rates and turnover numbers in practical ester hydrogenation catalyzed by RuPNN imine , 9 but we decided to conduct kinetic studies in toluene for two reasons: (1) the RuPNN imine precatalyst is only sparingly soluble in isopropyl alcohol, posing difficulties with the preparation of stock solutions and occasionally causing clogging in the stainless-steel tubing used for removing aliquots from the reaction mixture; and ( 2) as described below, the hexanol product was found to inhibit turnover, and the analysis of this observed product inhibition was most straightforward if no other alcohols were present in solution. In kinetic experiments, we monitored the conversion of hexyl hexanoate to 1-hexanol at 25 C by gas chromatography, with tetradecane as an internal standard. We began with the standard conditions shown in Scheme 7, then varied the initial concentrations of RuPNN imine , hexyl hexanoate, and hexanol, as well as the hydrogen pressure in independent experiments. The plots labeled "without preactivation" in Fig. 8 shows a typical kinetic trace under our standard conditions. Over approximately the frst 45 minutes of the reaction, the rate increases, after which apparent pseudo-frst-order consumption of ester is observed, as the plot of ln[ester] vs. time is linear after this point. During the 45 minute induction period, aliquots are dark purple, indicating the presence of the strongly absorbing precatalyst RuPNN imine , and become pale yellow as the precatalyst is converted to the resting state, which we propose is a hydridohexyloxide species analogous to RuPNN HOEt . To determine if the observed induction period can be explained by the activation of RuPNN imine with hydrogen, we conducted a preactivation experiment where we frst pressurized a solution of RuPNN imine with hydrogen (20 bar) for 90 minutes, which results in formation of the dihydride complex RuPNN HEt . 9 Then, the hexyl hexanoate substrate was transferred into the pressure reactor and its conversion to 1-hexanol was monitored at 25 C. As the plots labeled "with preactivation" in Fig. 8 demonstrate, the reaction follows apparent frst-order kinetics without an induction period, giving a nearly identical k obs value to what is observed without pre-activation of the catalyst. Importantly, this experiment rules out the possibility that sigmoidal kinetics come from acceleration of the reaction by the product alcohol, as proposed by O and Morris for their catalyst. 5a We also attempted to use RuPNN HEt directly as a precatalyst, but partial decomposition prior to the introduction of hydrogen pressure hindered our attempts to obtain reliable kinetic data. Because it was much more operationally convenient to assemble kinetic experiments in parallel without a catalyst preactivation step, we elected to conduct further kinetic trials using RuPNN imine as the precatalyst, using only the data after the 45 minute induction period to develop the kinetic model for the activated catalyst. To determine the partial order in catalyst concentration under the standard conditions, we repeated the experiment with a range of initial concentrations of RuPNN imine (Fig. 9). The same initial induction period followed by pseudo-frst-order behavior was observed, and k obs was taken as the slope of the linear portion of the plot of ln[ester] vs. time. A plot of k obs vs. [RuPNN imine ] 0 is linear with an intercept of zero, indicating that the reaction is frst-order in [ruthenium] after the induction period is complete. To determine the partial order in hydrogen, we repeated the standard experiment varying the hydrogen pressure (Fig. 10). In all experiments, a constant hydrogen pressure was maintained as aliquots were removed. Again, k obs was determined based on the linear portion of the plot of ln[ester] vs. time, and again a plot of k obs vs. hydrogen pressure gave a line with an intercept of zero, indicating that the reaction is frst-order in hydrogen under these conditions. We then repeated the experiment with initial ester (hexyl hexanoate) concentrations varied over a wide range from 0.05 M to 0.75 M. At high [ester] 0 , we observed a change from pseudo-frst-order to pseudo-zero-order behavior, consistent with saturation kinetics. When [ester] 0 is 0.25 M or less, apparent frstorder kinetic behavior is observed in each individual experiment, but k obs increases dramatically at lower [ester] 0 , consistent with inhibition by the product 1-hexanol. Although it may be counterintuitive that pseudo-frst-order kinetic behavior in [ester] is observed when the buildup of alcohol product inhibits the reaction, in our system the increasing product inhibition is roughly cancelled out by saturation in [ester], resulting in apparent frst-order behavior in each experiment. To directly probe for inhibition by the product hexanol, we repeated the experiment with a range of initial 1-hexanol concentrations, and we found that the rate decreases as [hexanol] 0 is increased, consistent with product inhibition. To deconvolute the effects of saturation in [ester] and inhibition by the product alcohol, we developed a numerical model of the reaction progress using the program Copasi. 25 Numerical modeling is seeing increased use in the analysis of kinetics in catalytic systems. 16b,26 When used in combination with DFT, kinetic modeling offers the potential to validate mechanisms and refne the energies predicted by DFT. 16b,26e In developing our model, we included the data after the 45 minute induction period from a total of 18 kinetic experiments, where the initial ester, alcohol, and ruthenium concentrations were varied, as well as the hydrogen pressure. Our model takes the standardstate free energies of the kinetically relevant intermediates and transition states as inputs (see Fig. 5 above), and computes the time course for ester hydrogenation, given the initial concentrations of ruthenium catalyst (a2), hexyl hexanoate, hexanol, and hydrogen. Hydrogen concentration, calculated from its known solubility in toluene at 25 C and the appropriate pressure, 27 was held fxed in the model. Because the activity coefficients of alcohols are known to vary signifcantly over the range of 0 to 1.0 M in non-polar solvents, 28 we used the activity of 1-hexanol rather than its molarity, as estimated following a model developed by Li and coworkers for 1-hexanol in benzene. 29 In attempting to reproduce the kinetic data with our model, we set the relative free energy of a2 to zero and compared a range of scenarios adjusting the remaining energies, in an attempt to achieve the best overall ft while including the smallest number of adjustable parameters. We found no better ft by allowing the free energies of a1, e1-g1-TS, g2, g1, or g to be adjusted. On the other hand, allowing adjustment of f2-g2-TS, m-n-TS, and m1-n1-TS was essential to obtaining a good global ft. Further, entirely excluding the pathway through e1-g1-TS had no detrimental effect on the ft. Our optimized model is depicted in Fig. 11, and the global ft to the kinetic data is shown in Fig. 12. The free energies of the dihydride species g, g1, and g2 were taken from the DFT results and held constant. The free energies of the transition states f2-g2-TS, m-n-TS, and m1-n1-TS were allowed to vary; ftted values are shown in Fig. 11. Overall, the kinetic data are very well-reproduced by this simplifed model, with minimal adjustment of the free energies obtained from DFT. Interestingly, the free energy of the hydrogen activation transition state f2-g2-TS was adjusted upward by 2.4 kcal mol 1 , while the energies of the ester hydrogenolysis transition states were adjusted downward by 1.2 and 1.4 kcal mol 1 , indicating that the standard-state activation barrier for hydrogen cleavage is the higher of the two, in contrast to the prediction from DFT. The model correctly reproduces the frst-order dependence on hydrogen pressure, the frst-order dependence on [Ru], and the saturation kinetics in [ester]. The inclusion of two similar-barrier pathways for ester hydrogenolysis allows a transition from second-order inhibition by alcohol at low [alcohol] to frst-order inhibition at higher [alcohol], consistent with the data. Importantly, the relatively small adjustment of the transition-state energies, less than 3 kcal mol 1 in each case, indicates that the barriers from DFT calculations are consistent with overall rate of the catalytic reaction. The above model satisfactorily reproduces the effect of [hexanol] on the rate of reaction essentially by assuming strong hydrogen bonding to the resting state a2 and weaker interaction with the transition states for ester hydrogenation. However, we do not claim that this model fully accounts for the behavior of the alcohol in the system, which likely includes medium polarity effects and more complicated interactions with the reacting species. ## Effect of added isopropyl alcohol Although we have conducted the above kinetic experiments in toluene, we determined previously 9 that isopropyl alcohol was an ideal solvent for the reaction, giving rates approximately 2-3 times faster than toluene and THF. To probe the accelerating effect of isopropyl alcohol further, we repeated our standard kinetic experiment with varying amounts of isopropyl alcohol added, up to 0.75 M. For comparison, pure isopropyl alcohol is 13.1 M. As shown in Fig. 13, we observe moderate but clear acceleration of the reaction with added isopropyl alcohol, consistent with our prior fndings and in contrast with the inhibiting effect of added 1-hexanol. Although we have not tried to probe this effect further, it likely originates from a different balance of stabilization of the resting state and transition-states by the two alcohols, potentially through specifc interactions and/or medium polarity effects. For example, the activity coef-fcient of 1-hexanol is expected to decrease with increasing [isopropyl alcohol], which should reduce 1-hexanol inhibition and accelerate the catalytic reaction. We note that the detailed rate dependence determined above for hexyl hexanoate, especially the effect of the alcohol, does not necessarily extend to the hydrogenation of all other esters. ## Disproportionation of aldehydes to esters As Gusev has reported recently, 30 catalysts for ester hydrogenation that produce an aldehyde intermediate can also be active for the catalytic disproportionation of aldehydes to esters, which effectively operates by running the ester hydrogenolysis pathway in reverse and the aldehyde hydrogenation pathway in the forward direction. Rearranging the pathways in Fig. 2 and 3 gives the MEP shown in Fig. S16 in the ESI, † with an overall barrier of 12.9 kcal mol 1 . Therefore, our calculations predict that aldehyde disproportionation should be rapid at room temperature. To test this prediction, we dissolved 1-hexanal in benzene-d 6 with 0.2 mol% RuPNN HEt at room temperature, and monitored by 1 H NMR. After 10 minutes, the aldehyde was completely consumed and hexyl hexanoate was the major product (Scheme 8). Further studies of this disproportionation reaction are in progress. ## Discussion We previously demonstrated that the ubiquitous RuPNN dearom is not kinetically competent as a catalyst for ester hydrogenation, instead converting from an inactive precatalytic form with an NEt 2 side group to an active form RuPNN HEt , which features an NHEt group that is essential for catalytic activity. 9 In this work, we have presented a plausible minimum-energy pathway, identifed through computation and validated experimentally through kinetic characterization and isolation of two key intermediates, RuPNN HEt and RuPNN HOEt . Our computations demonstrate that the N-H functional group plays a key role in the exceptional room-temperature activity of this catalyst. The N-H group is deprotonated and re-protonated in our MEPs for hydrogen activation, ester hydrogenolysis, and aldehyde hydrogenation, although in the frst two cases we identifed nearly isoenergetic pathways where the N-H group acts only as a hydrogen-bond donor without being deprotonated. A thorough search for alternative pathways where a CH 2 linker is involved in hydrogen activation identifed a minimum barrier of 23.7 kcal mol 1 , compared to 15.0 in our MEP. Because of the widespread application of RuPNN dearom in catalytic transformations and the corresponding widespread study of its reactivity by DFT prior to our disclosure of its facile dehydroalkylative activation, this system provides a unique case study in how the application of DFT in the absence of complementary experimental data can lead to the proposal of incorrect reaction mechanisms. Three studies we are aware of report a complete pathway for ester hydrogenation catalyzed by RuPNN dearom . In 2017, Zhang and coworkers reported a mechanism for the hydrogenation of ethyl benzoate catalyzed by RuPNN dearom . 13b In their work, RuPNN dearom was identifed as the resting state, and the highest barrier occurred in ester hydrogenolysis, giving an energetic span of 27.2 kcal mol 1 . In 2011, Wang and coworkers reported a study comparing the activity of RuPNN dearom for the acceptorless dehydrogenative coupling (ADC) of alcohols to give esters against the coupling of amines and alcohols to give amides, rationalizing the preference for the latter pathway over the former. 13a Reversing the ADC process predicts an overall energetic span of 38.5 kcal mol 1 for ester hydrogenation, from a dihydride resting state after hydrogen activation to a proton-transfer TDTS along the ester hydrogenolysis pathway. In 2020, Gusev reported a revised mechanism for ester hydrogenation and the reverse ADC, aided by the experimental identifcation of a hydridoalkoxide species as the proposed resting state. 13c In that study, the energetic span from the hydridoalkoxide TDI to the TDTS, a Hasanayn-like 21 hydride-alkoxide metathesis transition state, was 31.8 kcal mol 1 . As all three of the above studies rely on the on-cycle intermediacy of either RuPNN dearom , RuPNN H2 , or both, the proposed mechanisms cannot be correct, as we have shown that RuPNN dearom , which converts rapidly to RuPNN H2 under hydrogen pressure, 1 is inactive in ester hydrogenation prior to undergoing dehydroalkylation. 9 With our present demonstration that the experimental free-energy barrier to catalytic turnover is only 17.4 kcal mol 1 , the computed pathways above can also be excluded because the barriers they predict are implausibly high. Although it is not always explicitly stated, a common flter for the plausibility of reaction mechanisms calculated by DFT or other quantum-chemistry methods is a qualitative agreement of the overall reaction barrier with the observed rate of reaction. When detailed kinetic information is not available, Scheme 8 Disproportionation of reaction barriers must be estimated knowing only the catalyst loading, reaction time, and temperature. In the case of ester hydrogenation catalyzed by RuPNN dearom , Milstein's initial disclosure reported a turnover number of 100 in 4 h at 115 C. 1 If one assumes that catalyst induction is rapid and the turnover frequency is constant over the reaction time course, an overall barrier (energetic span) of 26.7 kcal mol 1 can be estimated. However, the barrier for turnover can be substantially overestimated if the catalyst undergoes a slow activation followed by very rapid turnover, as we showed is the case for this system. 9 This overestimation makes the above mechanisms, especially those proposed by Gusev and Zhang, appear plausible even though they predict barriers that are much higher than the actual barrier for catalytic turnover. It is worth revisiting a broader implication of the fndings we report here. As we described in the introduction, the majority of elite catalysts for ester hydrogenation and the reverse ADC of alcohols feature an N-H group with a key role in promoting catalysis. In this work and in a prior study, 9 we demonstrated that Milstein's catalyst RuPNN dearom and NEt 2 -substituted CNNpincer analogs developed in our group 7d are initially inactive, and must convert to an NHEt form to be catalytically active. Recently, Khaskin, Gusev, and coworkers 22 have shown that the same is true for a related bipyridyl PNN-pincer catalyst originally reported by Milstein and coworkers. 31 In this case, a pyridine ring is hydrogenated to a piperidine, again providing a latent N-H functional group with a key role in catalysis. Before these reports, initial proposals 1,2,31 and many computational studies 13a,14 pointed to the reversible deprotonation of a CH 2 linker as a key step in catalysis, but these proposals should potentially be reevaluated in light of the new fndings. Importantly, our work does not completely rule out the potential involvement of CH 2 linkers in other processes. It is experimentally known that the addition of H 2 to RuPNN dearom occurs reversibly at room temperature (Scheme 1). 1 Our calculations (Fig. S7 †) indicate that this process has a barrier of 23.7 kcal mol 1 under the conditions of ester hydrogenation, which is too high to account for the fast room-temperature turnover the activated catalyst exhibits in this process, but could be accessible at a higher temperature in a different process. We are continuing to probe these possibilities in computational and experimental studies of related catalytic transformations. ## Conclusion We previously demonstrated that Milstein's pincer-ruthenium catalyst for ester hydrogenation and related reactions is activated by dehydroalkylation to give the active form, which contains an NH functional group that is essential for catalysis. 9 In this work, we have presented a detailed computational and experimental study of the mechanism of ester hydrogenation catalyzed by this activated form, and conclude that participation of the N-H functional group is key in hydrogen activation, ester hydrogenolysis, and aldehyde hydrogenation. The catalyst speciation, the overall rate of reaction, and the dependence of the rate on the concentrations of reactants and products determined by experiment are in agreement with the mechanism predicted by DFT.

chemsum

{"title": "The key role of the latent N\u2013H group in Milstein's catalyst for ester hydrogenation", "journal": "Royal Society of Chemistry (RSC)"}

morita–baylis–hillman_reaction_of_acrylamide_with_isatin_derivatives

## Abstract: The Morita-Baylis-Hillman reaction of acrylamide, as an activated alkene, has seen little development due to its low reactivity. We have developed the reaction using isatin derivatives with acrylamide, DABCO as a promoter and phenol as an additive in acetonitrile. The corresponding aza version with acrylate and acrylonitrile has also been developed resulting in high product yields. ## Introduction The Morita-Baylis-Hillman (MBH) reaction is an important carbon-carbon bond-forming reaction . It involves the coupling of an activated alkene with an electrophile (usually aldehydes or imines) in the presence of a catalyst (Figure 1). The reaction is organocatalytic, atomically economical and operationally simple in nature. Most importantly, it results in the synthesis of densely functionalized molecules, also called MBH adducts. These are versatile synthons as they constitute several functionalities within close proximity, which aids in further synthetic transformations. Thus, as expected, the reaction has emerged as a powerful synthetic tool. It has seen exponential growth in several directions involving not only the application of the MBH adducts, but in the development of reaction as well . Although a very useful reaction, it does have some limitations such as a slow reaction rate, and is effected by electronic parameters and steric effects. Although the reaction has been wellexplored with aldehydes, the reaction with ketones is somewhat problematic. For a successful reaction to occur, the ketones require activation either by the presence of an α-activating group , the use of Lewis acid or application of high pressure . Similarly, in the case of activated alkenes, the acryl system shows differences in reactivity upon slight structural modifications. In such a system, the enone and acrylonitrile are more reactive, while with the acrylate reaction is relatively slow. Furthermore, there is a decrease of reactivity with acrylamide due to the reduced Michael acceptor tendency of alkene, which retards the attack of the catalyst on alkene, thus hindering the initiation of a reaction (Figure 1). Thus acrylamide has least contributed to the success of this reaction in the last four decades. In an effort to address the slow reaction rate of acrylamide, Hu et al. used dioxane/water in a 1:1 ratio, while Aggarwal et al. used quinuclidine in methanol to carry out the MBH reaction of acrylamide. Connon et al. utilized phenol and/or a H 2 O/t-BuOH 7:3 system for rate acceleration and Guo et al. used aryl activation . Other reports made use of reactive aldehyde , post-MBH modifications , an organometallic approach and other strategies . With comparatively few reports with respect to the significant literature on other activated alkenes in the MBH field, acrylamide thus requires further development and expansion of its scope. This is especially relevant given the fact that they have been extensively used in drug design , polymer chemistry and are popular synthetic templates . For further comparison to other acryl systems, acrylamide also offers extra valencies at nitrogen, which can be used for appending other functionalities/groups for intramolecular transformations. Other reports have used this feature for the development of an intramolecular MBH reaction: Corey et al. (total synthesis) , Pigge et al. (ruthenium complexes as an electrophile) , and Basavaiah et al. . Isatin has been the favored template not only for the spectrum of biological activities it provides, but also with respect to the development of methodologies . In the field of MBH, it has been used both for reaction development and application of its MBH-derived adduct including spiro frameworks . It is therefore anticipated that the development of the MBH reaction using acrylamide and isatin would not only expand the scope of acrylamide, but would also contribute to the expansion of the synthetic potential of isatin. a All reactions were carried out at rt with 0.5 mmol of 2a using 2 equiv of catalyst in 0.5 mL of the designated solvent. b Isolated yields. c 2 equiv of PhOH was used. d 5 equiv of PhOH was used. e Reaction was performed at 55-60 °C. ## Results and Discussion Initially N-phenylacrylamide (1a) was selected as a substrate for the development of the MBH reaction. This approach, together with the activation of acrylamide (by delocalization of lone pair electrons of nitrogen), was implemented in an attempt to directly vary the electronic properties of the acryl system and to expand the substrate scope. The reaction between 1a and N-methylisatin (2a) was carried out in the presence of DABCO using acetonitrile as the solvent (Table 1, entry 1). Although the reaction was slow and produced low yield (31%), the formation of the product 3aa with starting material remaining was nevertheless positive. In order to find the best conditions, several reactions were carried out. Increasing the reaction time to 5 days resulted in 56% yield (Table 1, entry 2). An increase in the loading of acrylamide (2 equiv, in order to generate more enolate) was helpful and resulted in 71% yield (Table 1, entry 3). The use of phenol [36, as a mild acid (Table 1, entry 4), further increased the yield to 92%. A further increase in the reaction time (5 days) and addition of more phenol (5 equiv, Table 1, entry 5 and entry 6) did not affect the yield. Scheme 1: Substrate scope of the MBH reaction for various isatin and acrylamide derivatives. All reactions were carried out at rt with a 0.5 mmol isatin derivative in acetonitrile (0.5 mL). Yields presented are isolated yields. a Acetonitrile (0.75 mL) was used for better dissolution. With the goal to reduce synthesis time, other catalysts along with different solvents were tested, but none led to a better result. The use of TPP as a catalyst required longer time and even after 10 days, TLC showed considerable amounts of starting material (Table 1, entry 9). The use of DMF as a solvent did not result in a pure product (Table 1, entry 15). Although heating to 55-60 °C did reduce the reaction time, this was accompanied by the generation of impurities along with a reduction in yield (Table 1, entry 16 and entry 17). Finally, using two equivalents of acrylamide, DABCO and phenol each (using acetonitrile as a solvent) at rt was identified as the best condition (Table 1, entry 4). Given these optimized conditions, the substrate scope for the developed protocol (Scheme 1) could be evaluated. It was found that the reaction was compatible with various substrates including different N-protected groups (methyl (3aa), benzyl (3ba) and propargyl (3ca)) on isatin and different substitutions on the periphery of the aryl group of isatin as well as acryl-amide. Electron-withdrawing groups resulted in a higher reaction rate (3ad, 3da, 3ea, one day each and 3ed 0.5 day). The reaction was found to work well with electron-rich (4-methyl, 2-methoxy), electron-deficient (3-chloro) and neutral aryl groups on acrylamide. Similarly, 5-chloro and 5-bromo-substitution on isatin gave similar yields in a similar time. The compound 3ea resulted in a single crystal , which further confirmed the structure (Figure 2). After establishing the synthetic potential of the protocol, the aza version of the corresponding reaction using Boc imine 4a and acrylamide as an activated alkene system was investigated. Accordingly, 4a was reacted with N-phenylacrylamide (1a) in the presence of DABCO and MeCN as a solvent. However, the reaction mixture resulted in complicated TLC results. A change of substrate (N-benzyl-protected isatin), catalyst (DMAP) or other solvents (THF, DCM, dioxane), gave no different result. However, remarkably, when the activated alkene was changed from acrylamide to methyl acrylate, formation of required aza-Scheme 2: Substrate scope of the aza-MBH reaction for various isatin derivatives. All reactions were carried out at rt with 0.5 mmol of isatin derivative in acetonitrile (0.5 mL) and yields are isolated yields. MBH product 5aa in high purity and in 91% yield in just 3 hours of reaction (Table 2, entry 1) was achieved. Encouraged by these results, the focus was shifted to the development of this aza-Morita-Baylis-Hillman reaction using isatin-derived ketimines . This reaction could also lead to the construction of tertiary benzylic amines and would help in the development of yet another fundamental reaction with commonly used Michael acceptors and inexpensive catalysts. As mentioned earlier, the application of the MBH adduct has greatly contributed to the success of the MBH reaction, as it necessitated quick access to these adducts for the rapid development of other methodologies. Optimization of the conditions and parameters revealed DABCO as a superior catalyst (Table 2). The reduction in the loading of methyl acrylate and catalyst, or dilution of the solutions did not have any major effect on time or yield. Thus, DABCO (0.25 equiv) along with a All reactions were carried out at rt with 0.5 mmol 4a. b Isolated yields. c Acetonitrile (0.25 mL) was used. d Acetonitrile (0.5 mL) was used. methyl acrylate (3 equiv) in acetonitrile as a solvent was identified as the best condition (Table 2, entry 9). Next, the reaction on different substrates was further explored. The protocol was found to work consistently, delivering the product with a short reaction time and in high yields (Scheme 2). The reaction scope was expandable to other activated alkene (acrylonitrile) and to other isatin derivatives with substituents on nitrogen (methyl, benzyl) and on the aryl ring (H, 5-chloro). ## Conclusion We have developed the Morita-Baylis-Hillman reaction of acrylamide with isatin derivatives. The reaction is facile and high yielding. However, the aza version of the reaction with N-phenylacrylamide as a substrate was not successful and led to a complicated reaction mixture. In contrast, the corresponding reaction with acrylate and acrylonitrile was very facile, clean and high yielding. We are currently investigating the development of the aza version with acrylamide and isatin-derived imine.

chemsum

{"title": "Morita\u2013Baylis\u2013Hillman reaction of acrylamide with isatin derivatives", "journal": "Beilstein"}

synthesis_and_characterization_of_the_fluorescentself-assembled_structures_formed_by_benzothiazolone

## Abstract: We report the synthesis and characterization of self-assembled structures formed by4-Choro-2(3H)-benzothiazolone (VK) to panchromatic fibers and its application as cell imaging tool. The aggregation properties ofthe synthesized compounds have been studied extensively under different solvent and concentrationand theirmorphologies examined at supramolecular level was observed by microscopic techniques like optical microscopy, fluorescence microscopy, and atomic force microscopy(AFM). Interestingly, the selfassembled structures formed by VKreveal panchromatic emission properties andshow blue, green and red fluorescence under different excitation filters. The intensity of the fluorescence observed was blue>green>red and the dye interestingly do not show any fluorescence quenching, on the other hand reveal photoactive properties under green channel. The mechanisms of formation of the self-assemblies were studied through different techniques like concentration dependent NMR and,UV visible spectroscopy and fluorescencemicroscopic studies.Finally, the utility of VK for cell imaging applications is demonstrated and it can be noted that VK can be efficientlyup taken by mammalian cells and the stained cells reveal panchromatic emission under blue, green and red channel. ## Introduction: Molecular self-assemblyisconsidered asvery fast-growing field of research considering its crucial significance in the field of material science, 1-4 chemistry 3,5,6 and biological science 3,7,8 for diverse applications. The study of molecular assembly of any compound is important since their organization at supramolecular level plays a decisiverole for ascertaining their potential applications as in drug delivery and sensing . 9-10 The main forces responsible for the formation of self-assembled structures arethe noncovalent interactions like vander Waal's, hydrogen bonding and pi-pi stackingwhich results in formation of well-defined morphologies throughbottom-up approach. Moreover, the self-assembling properties of the compound also influences its photophysical characteristics. which include its conductivity, optoelectronic properties 17,18 and its dispersion capability, 19,20 as well as the lattice packing of the molecules. 21,22 The selfassembling properties of the material is also affected by the various parameters such as concentration, 23,24 temperature, pH, 25,26 solvent, 27,28 and lattice arrangement. The importance ofthe self-assembly/aggregation on the photophysical properties of the material can be broadly understood by two terminologiesaggregation induced emission 34 and aggregation caused quenching 35 . We have earlier demonstrated synthesis and characterization of biomolecular self-assemblies and its applications as sensing and drug delivery tools. Recently, our group reported a new pyridothiazole based AIEE probe for sensitive detection of amyloids. 42 Herein, we report the self-assembly of 4-Choro-2(3H)-benzothiazolone (VK)to fibers exhibiting panchromatic emission. The self-assembly studies are done under varying solvent and concentration and the mechanism of its formation assessed by NMR and UV visible spectroscopy, Fluorescence spectroscopy.Notably,VK is an intermediate which iswidely used for the preparation of benazolin-ethyl, a well-knownherbicide used in agriculture.It has immense applications in pharmaceutical and agrochemical industries, agrochemical some of the noted products based on VK scaffold being Chlobenthiazone, Benazolin-ethyl,±)-Mevashuntin. 32 to name a few. Hence, the study pertaining to aggregation and photophysical characteristic of this benzothiazolone conjugate is particularly significant for assessing its implications in drug delivery and sensing. ## Results and discussions VK was synthesised in five steps through already reported synthetic methodologies. 32 . In the first step condensation reaction is carried out between 2-chloraniline and benzoyl isothiocyanate. The step is followed by alkaline hydrolysis which yields 2-chlorophenyl thiourea. Further, oxidative cyclization reaction is carried out using Br 2 to yield 4- Scheme 1 Chemical structure and synthesis of4-Choro-2(3H)-benzothiazolone. The synthesized VK was characterized by 1 H and 13 C NMR spectroscopy and LCMS mass analysis. The purity of VK was ascertained by HPLC. After achieving sufficient characterization and purity, the self-assembling properties of VK were analysed in water at various concentrations. The self-assembly studies of VK were done by dissolving VK in DMSO to make a stock solution of 50mM. The solution was subsequently diluted to 1mM by dissolving it in deionized water. The self-assembling behaviour of VK (1mM) was studied by atomic force microscopy (AFM). AFM analysis reveals VK assemble to fibers which also show branching (Figure 1). The fibers were having diameters in microns range and their length ranged to several micrometers. The fibers were soft in nature as can be assessed by the AFM tip force when recorded in tapping mode. The fibres of VK could also be observed by simple optical microscopy since their dimensions were large. A concentration dependent analysis of fiber morpholgu was done using optical and fluorescence microscopy at 1mM, 3mM, 5mM and 7mM concentrations As the concentration is increased fibers appear to become more thick and form tape like structure at high concentrations. Interestingly, under green filter these fibers also reveal intense fluorescence without the use of any dye in fluorescence microscopy. Hence, it may be inferred that VK is fluorescent in nature and hence assemble to fluorescent fibers. Interstinngly, when the fibers were observed in red filter a slight fluorescence with relatively less intensity than green channel could be observed which suggested VK might exhibit panchromatic fluorescence. Further, to understand the mechanism through which VK assemble to panchromatic fibers we resorted to concentration dependent NMR studies. These studies were done to understand the role of pi-pi stacking The study revealed that as we increased the concentration of VK, there was an upfield peak shift in the aromatic region between 7 to 9 ppm which could be observed by the concentration dependent NMR spectra (Figure 4b-4d). As the concentration increases. the intensity of peak is also increased and there is a simultaneous aromatic peak shift up field, indicating increase shielding of the protons due to pi-pi stacking which causes increase in the electron density aroung aromatic nucei. As the concentration of the VK solution is increased there is enhanced pi-pi stacking and molecules are stacked more and more close causing upfield shift. 3324, 34, 35 in different solvents like dimethyl sulphoxide, methanol and water. 24,33 It was surmised that as the polarity of solvents will change their will be a shift in spectra which can reveal the crucial role of hydrogen bonding in the self-assembly process. The UV visible study of VK revealed as the polarity of solvent increases their was peak broadening. The UV graph of VK, was studied in water, methanol and DMSO. Water has maximum polarity due to its high dialectic constant followed by methanol, DMSO, As we record the UV spectra of VK in three different solvent we observed that theUV graph display two peak in between interval of 10 nm, UV study in DMSO reveal two peaks at 282 and 289 nm, with a broad hump which might pointed to the presence of aggregates formed due to pi-pi stacking., When the UV spectra of VK was recorded in methanol, ,the peak at 282 nm observed in the case of DMSO shifts to 284 nm and peak at 289 shifts towards 292 nm. Further when UV spectra of VK were recorded in water there was maximum peak shift to 285 nm and 295 nm respectively, indicating as the solvent polarity is increased there is a red shift and also also more peak broadening indicating increased aggregation due to pi-pi stacking and hydrogen bonding. Further, concentration dependent microscopy studies presente in Figure 3 Hence, the mechanism of self-assembly of VK can be explained by pi-pi stacking and hydrogen bonding and can be represented graphically as in Figure 8. structures indicate their potential application as dye for bio-imaging. Hence, cellular uptake of VK was studies and the cells were imaged to identify its applications. The fluorescence of the VK compound was concentration-dependent (Figure 9). VK showed significantly high fluorescence on excitation with 488 nm laser and almost negligible fluorescence on excitation with 633 nm laser. However, VK's fluorescence intensity on excitation with 405 nm laser was significantly increased by increasing the concentration from 500 µM to 1 mM. This may be explained, since for bio imaging application confocal laser is fived at Ex of 405 nm, while VK real Ex is 360 nm as discussed in Figure 3. Hence a relatively less blue fluorescence is visible since during bio-imaging the optimal excitation wavelength is changed so that there are no harm to cells due to high energy radiations, Under green filter optimal fluorescence is observed due to Ex of 488 nm which is used in fluresence microscopy analysis of fibers too. The fluorescence in red region was also low since here again the excitation was 633 nm which is different from 540-560 nm used in the florescence microscopy. Hence from the cellular imaging data it may be inferred that VK can be potentiall used as bio-imaging dye in future. ## Synthesis and characterization of VK VK-Stage-4 crude(7.0 g, 37 mM, 1.0 eq.) was dissolved in 36 % concentrated HCl and heat at temperature 110 to 115 0 C for 10 to 12 hours on oil bath with water condenser, The reaction monitor shows that the formation of small particle after hydrolysis and product formation, during the progress there was the loss of HCl, so need to be added more 36 % HCl, monitor the TLC shows in 50% ethyl acetate: Hexane showed that the starting material consumed, stop the heating and allow to cool the reaction mixture at room temperature again cooled the reaction mixture at 0 to 5 0 C temperature and filtered the reaction mixture in the cooling then after dry the wet cake in rotavapor under reduced pressure which gives 6. ## Aggregation study of VK The aggregation study of VK has been study in DMSO: water mixture which deciphered that VK is itself fluorescence active in solution form in DMSO, When the fraction of water has been increased from 10 to 100 %, at beginning the fluorescence has been increased the till 50 % water fraction while after 50 % fraction fluorescence gradually decreased and at 90 % water fraction fluorescence complete quenching observed which clearly reveal that the aggregation caused quenching phenomenon. Which also supported that the aggregation caused quenching take place due to the formation excimers in the solution, the main cause of formation of excimers due to pi-pi stacking 36 which hindered the radiative emission of VK when it comes from S1 to S0 state as per Jablonski diagram. ## In vitro bioimaging assessment Human breast cancer (MDA-MB-231) cells were grown in DMEM cell culture media supplemented with 10% FBS and 1% antibiotic in a 5% CO2 incubator. First, the cells were seeded onto a glass coverslip and fixed with 4% paraformaldehyde. Later the fixed cells were treated with 500 µM and 1 mM VK for 15 min at room temperature. The coverslips were then mounted on a glass slide, and images were captured in Leica confocal microscope. ## Concentration NMR NMR spectra have been recorded at 1, 3, 5, 7, 9 mg/mL in DMSO-d6, for VK. The study was carried out in the Avance neo 400 MHz NMR Instrument. which has been performed to the check effects of concentration on pi-pi stacking of VK. 24, UV Visible Fluorescence spectroscopy

chemsum

{"title": "Synthesis and characterization of the fluorescentself-assembled structures formed by Benzothiazolone conjugates and applications in cellular imaging", "journal": "ChemRxiv"}

synthesis_of_montbretin_a_analogues_yields_potent_competitive_inhibitors_of_human_pancreatic_α-amyla

## Abstract: Simplified analogues of the potent human amylase inhibitor montbretin A were synthesised and shown to bind tightly, K I ¼ 60 and 70 nM, with improved specificity over medically relevant glycosidases, making them promising candidates for controlling blood glucose. Crystallographic analysis confirmed similar binding modes and identified new active site interactions. The healthcare burden of diabetes mellitus has steadily risen over the past century, with an increase from 108 million adult cases worldwide in 1980 to 422 million cases in 2014. 1 Strikingly, type II diabetes accounts for 90-95% of all cases. 2 The inhibition of carbohydrate catabolism has been explored as a means of mediating post-prandial blood glucose levels for the treatment of type II diabetes. 3 Targeting of carbohydrate catabolism provides two benefts: beyond directly modulating blood glucose levels it can also promote weight loss in prediabetic individuals to prevent the onset of diabetes. Miglitol, voglibose and acarbose are three inhibitors of carbohydrate catabolism that are currently in medical use. These compounds act as inhibitors of the mammalian gut a-glucosidases, thereby mediating post-prandial blood glucose levels. 4 While effective in preventing hyperglycemia, the resulting displacement of di-and tri-saccharides into the lower gut produces unwanted side effects arising from their strong osmotic effects within the large intestine and their rapid processing by anaerobic bacteria, resulting in diarrhea, nausea, and abdominal discomfort. 5 Human pancreatic a-amylase (HPA) catalyzes the endohydrolysis of ingested starches into maltose and maltotriose 6 and its activity has been positively correlated with post-prandial blood glucose levels. Consequently the targeted inhibition of this enzyme has been highlighted as a means of managing postprandial blood glucose levels while avoiding side effects associated with general a-glucosidase inhibition. 7 Most designs of glycosidase inhibitors have been based upon azasugar scaffolds in which a nitrogen atom replaces O5 or C1 of the sugar ring in the 1 binding site. 8 In this manuscript we explore a completely new pharmacophore for a-amylase inhibition involving stacked phenolic rings that interact closely with the conserved active site carboxylic acids. The parent molecule, montbretin A (MbA), is a complex flavonol glycoside produced by Crocosmia croc-osmiiflora that acts as a potent and selective inhibitor of HPA (K I ¼ 8 nM). 9 This inhibitor has been tested in ZDF (Zucker diabetic fatty) rats and produced a signifcant decrease in blood glucose plasma levels, highlighting this compound as a promising lead for the therapeutic reduction of post-prandial blood glucose levels. 10 However its large-scale extraction and purifcation from the corms of Crocosmia crocosmiiflora is a challenging multistep process and requires a substantial amount of biomass. Consequently the design and synthesis of novel inhibitors based upon MbA's structure has been proposed as a means of generating chemically accessible structures of similar potency and specifcity. 11 Montbretin A contains a myricetin flavonol core glycosylated at the 3-and 4 0 -positions (Fig. 1a). myricetin A-ring. 11 Meanwhile, minimal interactions were observed with the myricetin B-ring or any of MbA's carbohydrate appendages. Step-wise chemoenzymatic degradation of MbA had yielded a 'minimum inhibitory structure' termed mini-MbA (Fig. 1b). This inhibitor contained only the myricetin flavonol and caffeic acid linked through a D-glucopyranosyl-(b-1 / 2)-L-rhamnopyranose disaccharide, and had a K I ¼ 400 nM (ref. 12) towards HPA (lone myricetin and ethyl caffeate have K I ¼ 100 mM and 1.3 mM, respectively 9 ). The retained potency of the simplifed mini-MbA structure suggested that a suitable synthetic analogue could achieve equivalent levels of potency, thus mini-MbA was used as inspiration for our synthesis. It was not clear however, whether such a compound would retain the required specifcity for HPA over the gut alpha-glucosidases given the active site similarities. Quercetin was selected to replace myricetin as the flavonol core in our analogue synthesis since it, and its glycosylated derivatives, are substantially less expensive than other flavonols. Quercetin differs from myricetin only in hydroxylation of the B-ring and, as noted earlier, there are very few interactions between HPA and this ring. 13 In mini-MbA, the caffeic acid and myricetin groups are connected by a disaccharide, which forms a bridge of 7 atoms. This region formed no direct interactions with the enzyme, though it likely provides some conformational constraints. As a result, a variety of different chemical structures could be explored for the linker region. Requirements for linker length and rigidity could be assessed, as well as the possibility of incorporating new functionalities to create favorable interactions with the enzyme. Quercetin-3-O-rutinoside was used as a convenient and inexpensive starting material for the synthesis since the pre-existing sugar at C3 could be used to differentiate the hydroxyls in a protecting group strategy. Quercetin-3-O-rutinoside was thus reacted with benzyl bromide in the presence of K 2 CO 3 , followed by cleavage of the 3-O-glycoside by treatment with HCl/ethanol to yield benzyl quercetin 2, freeing the 3-OH for subsequent modi-fcation. 14 Analogues with simple linkers were frst synthesized to test the basic concept of linking the two phenolics (Fig. 2). Optimal linker length and polarity was then explored through a series containing three, fve, seven and eight atoms. Analogues M01 to M03 contained simple alkyl linkers of three, fve and seven atoms long while M04 contained a similarly flexible eightatom triethylene glycol linker. Focus then changed to design of linkers that incorporate appropriate conformational constraints while seeking out favourable interactions of the side chains with active site residues. To that end analogues M05 to M11 were generated containing an amino acid linked to quercetin's 3-OH through a propyl chain (Scheme 1). Tri-O-benzyl quercetin 2 was functionalized with 1-bromo-3chloropropane, 1-bromo-5-chloropentane, 1-bromo-7azidoheptane, or 1-bromo-8-azido-triethylene glycol depending on the analogue synthesized. Treatment with tetrabutylammonium azide produced a terminal azide, which was reduced with trimethylphosphine. In the cases of analogues M01 to M04, this could be coupled directly to caffeic acid through use of activated pentafluorophenyl caffeic ester 1. For analogues M05 to M11, the propylamine derivative 4 propyl was coupled to one of six chosen Fmoc-L-amino acids (with appropriate side chain protecting groups) (Scheme 2). 15 The Fmoc group could be deprotected with 20% piperidine in dichloromethane followed by a second coupling with the pentafluorophenyl caffeic ester 1. Deprotection of the benzyl ethers and amino acid protecting groups was achieved through treatment with BBr 3 . While this synthesis allowed the construction of a small library of analogues, the fnal deprotection step in particular was prohibitively low yielding, posing a signifcant barrier to future applications. The continued study of any promising analogues required an improved synthesis, as will be explored below. The eleven analogues were tested as inhibitors of HPA as shown in Table 1. Analogues M01 and M02 both had a K I ¼ 4 mM, demonstrating that joining the flavonol and caffeic acid groups with a simple linker could increase potency by 25 times compared to the lone flavonol (K I $ 100 mM). 9,13 However, these analogues were substantially less potent than mini-MbA, despite having the potential to form all the same interactions within the amylase active site. M03 was no more potent than the lone flavonol, suggesting that its three-atom linker was too short to allow for proper orientation of the phenolic moieties in the active site. M04, which contained an eight-atom triethylene glycol-based linker (TEG), also failed to offer a signifcant increase in potency over the lone flavonol. Despite having a linker of similar length and flexibility to those of M01 and M02, the incorporation of a triethylene glycol-based chain in place of a simple alkyl chain led to a 21-fold drop in potency, potentially due to the differences in polarity. While amino acid based analogues M05, M07, and M08 offered only meagre increases in potency compared to the lone flavonol, M06 fared better with a K I ¼ 1 mM, perhaps due to the limited conformational range of its proline pre-organising the ligand for binding. Introduction of aromatic residues into the linkers also afforded an increase in binding affinity. The phenylalaninebased analogue M09 bound 22-fold tighter than the unfunctionalized glycine-based analogue M05, suggesting that its aromatic side chain likely creates favorable p stacking interactions in the active site, contributing 1.8 kcal mol 1 to binding affinity. Installation of hydroxyl groups onto the phenyl ring, as seen with the tyrosine-based analogue M10 and the DOPA-based analogue M11, produced a further 40 to 47-fold increase in potency relative to M09, equating to a further $2.5 kcal mol 1 contribution to the binding affinity. With inhibition constants of 70 and 60 nM, M10 and M11 bind approximately six times more tightly than the mini-MbA that inspired their synthesis. Thus not only have we replicated the affinity of the natural product with a simple, synthetically accessible analogue, but also we have substantially improved upon its binding affinity. To gain insights into the binding modes of the best of these analogues, and particularly to understand the greatly improved affinity of M10 and M11, X-ray crystallographic analysis of analogues M06 and M10 in complex with HPA was pursued. The resulting crystal structures (PDBs 6OCN and 6OBX at 1.15 and 1.30 resolution) revealed the analogues in non-covalent complex within the enzyme active site. Importantly, an overlay of the M06/HPA and M10/HPA structures with that of mini-MbA/HPA showed essentially complete overlap of the flavonol and caffeic acyl moieties in each case, simplifying future design (Fig. 3a). Proline-containing analogue M06 created all of the polar contacts within the HPA active site that were previously observed with mini-MbA (Fig. 3b). Hydrogen bonding was observed between the caffeic acid catechol and R195, D197 and E233 of the enzyme. The 7-OH of M06's flavonol formed hydrogen bonds with D197 and H201. A hydrogen bond was also observed between the C4 carbonyl of the flavonol and the side chain of T163. Meanwhile, the 3 0 -OH of the flavonol formed a hydrogen bond with the side chain of H201, as was also observed in the mini-MbA/HPA complex. This indicates that the more cost-effective quercetin was sufficient to form the necessary interactions within the active site. The absence of this 4 0 -O-disaccharide on mini-MbA opens this position up for hydrogen bonding with K200 in the active site, an interaction that is also seen with M06. In addition to these polar contacts, a p-CH stacking interaction appears to form between the proline side chain of M06 and W59 of the enzyme. Despite its high degree of alignment and reproducibility of polar contacts, M06 is still 2.5 times less potent than mini-MbA, highlighting the importance of MbA's glycosidic appendages in stabilization of a pre-stacked conformation. Analogue M10 also formed all of the same hydrogen bonds within the amylase active site previously observed for mini-MbA and M06 (Fig. 3c). In addition, and presumably largely responsible for its affinity enhancement, the tyrosine of M10's linker formed hydrogen bonds with the main chain carbonyl and side chain hydroxyl of T163. These are accompanied by a number of CH-p-interactions between the tyrosine side chain and residues W59, Q63, and L165 of the enzyme. To probe its specifcity towards HPA, M10 was tested as an inhibitor of ten other glycosidases, including human intestinal maltase-glucoamylase and sucrase-isomaltase, and two intestinal bacterial a-amylases (Table 2). The analogue showed no inhibitory activity towards these a-glucosidases at a concentration over 100-fold higher than its K I value for HPA. M10 did show some inhibitory activity against the nonrelevant S. cerevisiae a-glucosidase and Agrobacterium bglucosidase. However, the binding affinity of M10 towards yeast a-glucosidase is actually lower than that of lone quercetin (IC 50 ¼ 7 mM). 16 Comparison of these results with those of previous specifcity analyses of mini-MbA and MbA indicated that M10 was in fact more selective for HPA over the intestinal glucosidases, since mini-MbA and MbA inhibited sucraseisomaltase and B. fbrisolvens a-amylase, respectively. 9,11 This bodes well for the use of M10 as a selective amylase inhibitor. It also implies that this pharmacophore may have restricted application to other glycosidases. Inhibitor M10 demonstrates highly promising HPA inhibition, however the synthetic route that led to its discovery has a number of shortcomings that would hamper its development as a potential pharmaceutical. As such, an alternative approach was devised (Scheme 3). Selective azide reduction of 3 propyl could be carried out in the presence of 6 to give amide 7, which could then be globally deprotected to give amine 8. Treatment of crude amine 8 with 1 then delivered M10 in 49% yield over four steps, with no purifcation necessary between steps. This improved procedure delivers M10 in eight steps from rutin (24% overall yield), with further optimization certainly possible. ## Conclusions The synthesis and kinetic analysis of MbA analogues M01 to M11 provides new insight into the structural requirements for HPA inhibition. The fact that most of the analogues bound more weakly than mini-MbA highlights the importance of MbA's constrained carbohydrate linker. Despite forming no direct interactions in the active site it appears to pre-stack the flavonol and caffeic acid moieties for optimal interaction. Initial exploration of the effect of incorporation of rigidity into the analogue linkers was achieved with the proline linker of M06. This increased potency some 60-fold compared to the nonfunctionalized glycine linker, with more possibly achievable by freezing out other motions. Affinity enhancement through the acquisition of stabilising interactions was explored through the incorporation of phenolic side chains to recruit some of the hydrophobic and hydrophilic interactions ordinarily developed by the starch substrate in the amylase active site. 17 Indeed the phenolic linkers of M10 and M11 take advantage of both such features, rendering the inhibitors six times as potent as mini-MbA, and delivering a higher ligand efficiency than either MbA or mini-MbA (Table S1 †). Further, and perhaps surprisingly, the specifcity of inhibition is retained despite the reduction in complexity. Thus M10 and M11 represent amylase inhibitors that are both much more synthetically accessible and indeed more potent than the mini-MbA that inspired their development. With further optimization of their synthesis underway, pharmacokinetic and efficacy studies with these inhibitors are planned to determine their suitability for therapeutic usage. ## Experimental section Please refer to ESI. †

chemsum

{"title": "Synthesis of montbretin A analogues yields potent competitive inhibitors of human pancreatic \u03b1-amylase", "journal": "Royal Society of Chemistry (RSC)"}

bioaccessibility_of_organic_compounds_associated_with_tire_particles_using_a_fish_in_vitro_digestive

## Abstract: Tire and road wear particles (TRWP) account for an important part of the polymer particles released into the environment. There are scientific knowledge gaps as to the potential bioaccessibility of chemicals associated with TRWP to aquatic organisms. This study investigated the solubilization and bioaccessibility of seven of the most widely used tire-associated organic chemicals and four of their degradation products from cryogenically milled tire tread (CMTT) into fish digestive fluids using an in vitro digestion model based on Oncorhynchus mykiss. Our results showed that 0.06% to 44.1% of the selected compounds were rapidly solubilized into simulated gastric and intestinal fluids within a typical gut transit time for fish (3 h in gastric and 24 h in intestinal fluids). The environmentally realistic scenario of coingestion of CMTT and fish prey was explored using ground Gammarus pulex. Coingestion caused compound-specific changes in solubilization, either increasing or decreasing the compounds' bioaccessibility in simulated gut fluids compared to CMTT alone. Our results emphasize that tire-associated compounds become accessible in a digestive milieu and should be studied further with respect to their bioaccumulation and toxicological effects upon passage of intestinal epithelial cells. ## Introduction: Tire and road wear particles (TRWP) are produced during abrasion of tires on road pavement. Low amounts of small-sized TRWP (<10 µm) enter the atmosphere during use but between 95 -99% of total emitted TRWP are expected to be deposited on the road side 1 and be transferred into the nearby soil, from which a fraction will eventually enter the water streams 2 . A modelling study estimated that 49% of TRWP emitted on the road would reach the freshwater system in the French Seine basin 3 . Field measurements suggest that levels of TRWP decrease between its emission source and the aquatic environment with concentrations of 0.1 -100 g kg -1 on the road side, 0.5 -1.2 g kg -1 in river sediment and 0.5 -5 mg L -1 in river water 4 . TRWP are heterogeneous particles composed of rubber polymer, minerals, bitumen and various chemicals originating from the road environment or from the rubber itself 5,6 . They are susceptible to environmental weathering, leading to changes in physical properties and chemical composition of the particles 4,7 . For instance, metals, such as Pb, Mn, Co, Cr, Ba, and Ni, were measured as traces in the tire rubber but also in higher concentration in TRWP, revealing the contribution of the road constituents to the overall metal burden of TRWP 5,6,8 . Several organic chemicals are added to tire rubber to facilitate polymerization during manufacturing or to increase the performance and longevity of the tires during use. Among many other compounds, 2-mercaptobenzothiazole (MBT) and 1,3-diphenylguanidine (DPG) are intensively used as vulcanization agents; they can represent up to 0.5 % of the tire rubber 9 . Phenylenediamine compounds, such as N-isopropyl-N′-phenyl-p-phenylenediamine (IPPD) and N-(1,3-dimethylbutyl)-N′phenyl-1,4-phenylenediamine (6PPD), are also commonly used as antioxidants and antiozonants (up to 4% of the tire tread) in the final product to prevent cracking and degradation of the rubber during wear 10 . Highly aromatic oils used in rubber manufacturing commonly include polycyclic aromatic hydrocarbons (PAHs), some of which are classified as carcinogenic. The use of PAHs by the tire industry has been regulated by the EU Directive 2005/69/EC since January 2010. Accordingly, tire tread may no longer contain more than 10 µg∑8PAHs g -111 . However, lower levels of regulated PAHs as well as unregulated PAHs could still be present in tire tread. Finally, Zinc oxide is commonly used as a sulphur vulcanization catalyst during the curing process of rubber and represents up to 2.5 mass% of the final tire composition 12,8 . The potential toxic impact of these tire-associated chemicals for aquatic biota has been mainly assessed by laboratory experiments with organisms exposed to aqueous leachates of unaltered tire particles . A few studies explored the toxicity of aged tire particle leachates 19,20 . These studies were conducted with heterogeneous experimental conditions (temperature of leaching, pH of leaching solution and salinity), which could impact the solubilization and further bioaccessibility of the tireassociated compounds. Overall, the studies investigating the toxicity of tire particle leachates led to contrasting results. These could originate from the variability of the exposure conditions but may also suggests species-specific sensitivity 14,17,18, . A recent study incriminated a 6PPD oxidation product, namely 2-((4-methylpentan-2-yl)amino)-5-(phenylamino)cyclohexa-2,5-diene-1,4-dione (6PPD-Q), as responsible for acute toxicity to Coho salmon (Oncorhynchus kisutch), threatening the local population of this fish species in urban creeks of Seattle (United States of America) 25,26 . 6PPD-Q was also found to be highly toxic for two other salmonid species (Brook trout (Salvelinus fontinalis) and Rainbow trout (Oncorhynchus mykiss) ) but not for five other fish species and two crustacean species , suggesting a species-specific mode of action for this chemical. A recent study found microplastics, including tire particles, in the stomach content of several wild fish species, showing that fish can also ingest TRWP 31 . Studies investigating direct effects of TRWP remain scarce 21,32,33 which highlights the need to investigate the effects of the particles themselves in addition to the effects of leachates 34 . Indeed, several studies showed that the solubilization of polymer-bound chemicals was enhanced in fish gut fluids compared to water and could promote bioaccessibility of the chemicals for uptake into the circulatory system . More specifically, we demonstrated in a previous article that bioaccessibility of Zn from tire particles was enhanced by the organic components of the fish gut fluids and assessed the effects of coingestion of food organic matter on Zn bioaccessibility in fish gut 8 . Nonetheless, the bioaccessibility of organic compounds associated with tire particles, which could contribute to the toxic effects observed in several studies, remains poorly investigated. Therefore, this study used a fish (Rainbow trout) in vitro digestion model and cryogenically milled tire tread (CMTT) as a surrogate material for environmental TRWP in order to (i) determine the solubilization kinetics of several commonly used antioxidants, vulcanization aids and transformation products from unaltered and artificially aged CMTT into simulated gastrointestinal fluids of fish and (ii) assess the overall bioaccessibility (defined as the soluble fraction of the chemical available for uptake) of these organic compounds in fish gut with and without coingestion of food organic matter. ## Materials The generation of CMTT was previously described by Masset et al. (2021) 8 . Briefly, the upper layer of the tire tread from Pirelli® (Sottozero 3), Michelin® (Primacy 3) and Bridgestone® (Saetta Touring 2) tires (ratio 1:1:2, respectively) were cut into small pieces of 1 cm 3 using industrial scissors and a water jet machine and cryogenically milled using a model A Hammer Mill (Pulva®). The particles were collected and stored in amber glass vials in darkness at room temperature. Artificially aged CMTT were generated by thermooxidation, following the protocol of Klöckner et al. (2021) 41 . More details regarding the physicochemical characteristics of CMTT (size distribution, electron microscopy images) are presented in figures S1 and S2. The composition of the fish simulated gastric fluid (SFGASTRIC) and simulated intestinal fluid (SFINTESTINAL) used in this study was the same as in Masset et al. 2021 8 (table S1). Briefly, both SFGASTRIC and SFINTESTINAL consisted of a luminal buffer adapted from Leibovitz's L-15 cell culture medium to mimic the composition of the lumen of fish intestine. The digestive fluids were designed to be used in combination with a cell line isolated from Rainbow trout intestine, the RTgutGC, which is cultured using L-15 medium 42 . Purified pepsin (Sigma-Aldrich®) was added to the luminal buffer at a concentration of 12.5 U mg -1 of protein and pH was adjusted to 2 with 32 % HCl to obtain SFGASTRIC. A concentration of 4 mg mL -1 of porcine bile extract (Sigma-Aldrich®) and 2 mg mL -1 of pancreatin (Sigma-Aldrich®) was added to the luminal buffer to obtain SFINTESTINAL with a pH of 7.4. Control experiments were performed in mineral water (MW) (Evian®) (composition in table S2) for comparison with digestive fluids. ## Determination of particle and digestive fluids characteristics A qualitative analysis of particle morphology was performed on CMTT and aged CMTT before and after in vitro digestion using scanning electron microscopy (SEM) (GeminiSEM 300, Zeiss®). This was done in order to assess the general morphology of the particles and to investigate any morphological changes at the surface of the particles that might result from the aging process or from the in vitro digestion. Surface tension of the SFINTESTINAL was measured with a goniometer (EasyDrop, Kruss®) and the presence, size and stability of micelles in the digestive fluids was assessed by dynamic light scattering and measurements of the zeta potential of the solutions using a Zetasizer ZS®. The concentration of dissolved organic carbon (DOC) was measured in the digestive fluids using an organic carbon analyser (vario TOC cube, Elementar®). ## CMTT organic chemical composition determination In order to determine the total concentration of selected antioxidants and vulcanization agents in CMTT, preliminary tests showed that ultrasound-assisted extraction resulted in poor recovery as some compounds were strongly bound to the rubber matrix and required harsher extraction conditions. Therefore, CMTT spiked with deuterated internal standards (benzothiazole-d4, aniline-d5, diphenylurea-d10 and 6PPD-Q-d5 and a mix of 16 deuterated PAHs) were Soxhlet-extracted with 150 mL of methanol for 16 h, followed by 150 mL of dichloromethane for another 16 h. Both fractions were combined and evaporated to 2 mL using a rotavapor (Büchi®) and passed through a 0.45 µm Glass Fiber Filter (GFF). An aliquot of 1 mL was prepared without further clean-up for direct analysis with Ultra Performance Liquid Chromatography coupled with a tandem mass spectrometer (UPLC-MSMS). Another aliquot was passed through a chromatographic column filled with 3 g of silica-gel previously activated at 180°C for 8 h, eluted with 50 mL of hexane and concentrated with a rotavapor to a volume of 2 mL. Finally, the extracts were concentrated near dryness under a gentle stream of nitrogen and solvent exchanged to 500 µL of isooctane for analyses of PAHs with Gas Chromatography coupled with a tandem mass spectrometer (GC-MSMS) (more details in section "Chemical analyses"). ## In vitro digestion experiment Solubilization kinetics. Solubilization kinetics of organic chemicals were investigated with both CMTT and aged CMTT in SFGASTRIC and SFINTESTINAL separately to investigate the effects of pH and of the composition of the fluids on the kinetics. The in vitro digestion was performed at 10 g of CMTT L -1 of digestive fluid by introducing 150 mg of CMTT in amber glass vessels containing 15 mL of SFGASTRIC or SFINTESTINAL. The digestion was carried out at 20°C under gentle agitation for 3 h (SFGASTRIC) and 24 h (SFINTESTINAL). At regular time, a 15 mL sample was collected and centrifuged at 950 g-force for 5 min to remove large CMTT particles and bile aggregates and the supernatant was filtered through 0.45 µm GFF filters. All experiments were conducted in triplicates and control experiments consisting of leaching of 10 g of CMTT L -1 for 24 h were performed in mineral water (Evian®) for comparison with digestive fluids. Experimental blanks with digestives fluids and mineral water without CMTT were also prepared and analysed. Coingestion experiments. The environmentally realistic scenario of coingestion with food was explored with unaltered CMTT as unaltered and aged CMTT exhibited close composition and behaviour in digestive fluids (see Results and Discussion). In this experiment, a sequential in vitro digestion was chosen. It consisted of a 3 h incubation in SFGASTRIC to mimic the transit time in the fish stomach followed with a 24 h digestion in SFGASTRIC + SFINTESTINAL estimated as an average transit time in the fish small intestine 43 . Four g of Gammarus pulex (G. pulex), used as a surrogate for fish prey, were ground with a mortar and pestle and 0.4 g of CMTT (food/CMTT ratio = 10) were placed in digestion vessels. Twenty mL of SFGASTRIC was added in the vessels and the digestion was performed at 20°C under gentle agitation. After 3 h, 20 mL of SFINTESTINAL was added to the vessel and the pH was adjusted to 7.4 with stepwise addition of NaOH. The digestion was stopped after 27 h in total (3 h in SFGASTRIC and 24 h in SFGASTRIC + SFINTESTINAL) and all samples were centrifuged at 3000 rpm for 5 min and passed through 0.45 µm GFF filters. Control experiments were performed with ground G. pulex alone and with CMTT only for comparison. All experiments were performed in triplicates. ## Chemical analyses For each digestate sample, a sub-sample (1 mL) was spiked with deuterated internal standards (benzothiazole-d4, aniline-d5, diphenylurea-d10 and 6PPD-quinone-d5) and analysed without further clean-up with UPLC-MSMS. A second sub-sample (12 mL) was collected for PAHs analysis and was spiked with deuterated internal standards (mix of 16 deuterated PAHs) and liquid/liquid extracted twice with 10 mL of dichloromethane. Then, the extracts were concentrated with a rotavapor to 2 mL and followed a similar purification and preparation protocol as described for the CMTT extracts (section "CMTT organic chemical composition determination"). The following tire-associated compounds in the CMTT extracts as well as in the in vitro CMTT digestates were analysed with an UPLC-MSMS (Xevo TQ MS, Waters®): Benzothiazole (BT), 2hydroxybenzothiazole (HBT), 2-mercaptobenzothiazole (MBT), aniline (ANI), 1,3-diphenylguanidine (DPG), N-(1,3-dimethylbutyl)-N′-phenyl-1,4-phenylenediamine (6PPD) and 2-((4-methylpentan-2yl)amino)-5-(phenylamino)cyclohexa-2,5-diene-1,4-dione (6PPD-Q). The EPA's 16 priority pollutant PAHs were analysed with a GC-MSMS (TSQ Quantum XLS Ultra, Thermo Scientific®). Six calibration standards were analysed for each batch of samples (1 ng mL -1 to 500 ng mL -1 , linearity R² >0.99). Details regarding the chemicals used, UPLC-MSMS, GC-MSMS methods and QA/QC for chemical analyses of CMTT particles and simulated gastrointestinal extracts are provided in text S1. Details regarding the synthesis and quality control of the 6PPD-Q produced in-house are provided in text S2. ## Statistical analyses All statistical tests were performed using R ver. 3.5.0. Tentative fitting of four models (logarithmic kinetic, diffusion-controlled kinetic, 0 th and 1 st order kinetics) were performed for each compound. Differences in chemical concentrations following ingestion with or without coingestion of food were tested using t-tests or Kruskal-Wallis test for non-normally distributed data. ## CMTT organic chemical composition Eleven compounds were quantified in unaltered CMTT and aged CMTT extracts (table 1). In unaltered CMTT, 6PPD represented 31.0 mg g -1 (3.1% of the CMTT mass). One of its oxidation by-products, 6PPD-Q was detected in much lower amount (14 µg g -1 , 0.0014%) which can be explained by the fact that CMTT was not exposed to oxidative conditions prior to analyses. However, in aged CMTT, the concentration of 6PPD was reduced (13.1 mg g -1 , 1.3%) and that of 6PPD-Q increased (30 µg g -1 , 0.0030%). Similarly, MBT and BT concentrations decreased by 71 and 61% respectively, whereas HBT concentration increased by 71%. These results show that the artificial aging of CMTT led to chemical modification of the particles via oxidation processes, as has previously been described 41,44,45 . They also reveal the high degradability of 6PPD resulting in the production of several transformation products, among which 6PPD-Q. Contrastingly, the PAH content in unaltered CMTT and aged CMTT was similar, likely due to the low vapor pressure of these congeners preventing volatilization from the surface of the CMTT and to their weak oxidation under controlled atmospheric conditions. The PAH profile was dominated by 4 PAHs, namely phenanthrene (PHE), fluoranthene (FLT), pyrene (PYR) and benzo(g,h,i)perylene (BPY) (table 1). These 4 PAHs represented more than 80% of the total measured PAH content of the particles (the full 16 PAHs profile analysed is provided in figure S3). A qualitative observation of the CMTT by electronic microscopy did not reveal any visible alteration of the surface of the particles or formation of cracks from the aging process (figure S1). However, it has been shown that exposure to oxidative conditions of microplastics could lead to modification of the particles surface microstructure 46 , possibly not detectable with electron microscopy. Therefore, we investigated whether the desorption of tire-associated chemicals in simulated digestive fluids was impacted by aging. ## Solubilization kinetics In vitro digestions of unaltered CMTT and aged CMTT in SFGASTRIC and SFINTESTINAL were performed in both types of fluids separately to reveal the underlying mechanisms responsible for the solubilization of the tire-associated compounds. The solubilization kinetics from unaltered CMTT for all compounds are presented in figure S4. To facilitate comparison between compounds present in different concentrations in CMTT (table 1), the results were expressed as the bioaccessible fraction (%), which was calculated as follows: ## 𝑏𝑖𝑜𝑎𝑐𝑐𝑒𝑠𝑠𝑖𝑏𝑙𝑒 𝑓𝑟𝑎𝑐𝑡𝑖𝑜𝑛 (%) = 𝑚 𝑓𝑓 𝑚 𝐶𝑀𝑇𝑇 * 100 With 𝑚 𝑓𝑓 = the mass of the compounds solubilized in the digestive fluid at the end of the digestion (µg) and 𝑚 𝐶𝑀𝑇𝑇 = the nominal total mass of chemical based on measuring extracts of CMTT in the digestion vessel (µg). All compounds except PAHs were rapidly solubilized in SFGASTRIC within the 3 h digestion time (figure 1, figure S3, table S3). The compounds' solubilization kinetics were best fitted by a logarithmic or a diffusion-controlled model 47,48 , suggesting that at least two mechanisms were involved in the solubilization from the rubber matrix (figure 1). With one exception, the data revealed a fast solubilization within the first hour of digestion before a pseudo-equilibrium was reached. The exception was DPG, for which a constant solubilization was observed during the 3 h digestion time (figure 1b). PAHs were not detected in SFGASTRIC, indicating very poor solubilization potential for these hydrophobic molecules. Concentration of ANI at the end of the digestion in SFgastric was very high (2.5 mg mL -1 ), i.e. at 449% of the total ANI content in CMTT introduced in the digestion vessel, suggesting that ANI was formed during the digestion. Indeed, as nitrobenzene is widely used within the rubber industry and is a recognized precursor for ANI production under acidic condition 52,53 , it is possible that the excess of ANI measured in SFgastric was formed via reduction of nitrobenzene at pH = 2 (figure S5). An alternative explanation could be that ANI was formed due to degradation of DPG 9 . In SFintestinal, all compounds were rapidly solubilized during the in vitro digestion time of 24 h. All solubilization kinetics were best fitted by a logarithmic model except for DPG for which the solubilization was best fitted by a diffusion-controlled model (figure 1b). PAHs were poorly solubilized in SFintestinal (solubilization rate k = 1.1 x 10 -2 -5.4 x 10 -2 ) (table S3). However, tentative fitting with the 4 models tested in this work was not satisfactory as a peak of concentration in the SFintestinal was reached after 3 h and concentrations decreased afterward until the end of the digestion (24 h) (figure S3). One possible explanation could be the high affinity of the PAHs for organic particulates, hence, the compounds would adsorb on bile particulates which were removed by filtration before analysis. The solubilization kinetics of chemicals from aged CMTT are presented in table S3. The aging treatment did not lead to significant changes in the solubilization rates of most compounds in the three solutions (water, SFgastric and SFintestinal). The solubilization rate only increased slightly for HBT in SFgastric and SFintestinal, whereas it decreased for DPG in water and SFgastric and for 6PPD in SFgastric (table S3). These results suggest that only a minor alteration of the polymer matrix occurs during the aging process and confirms the lack of visible physical changes observed by electron microscopy. Indeed, the artificial aging treatment by thermooxidation did not result in significant modification of the solubilization potential of the tire-associated compounds. In contrast, a strong oxidative treatment using potassium persulfate as a surrogate for aging treatment of tire particles led to morphological modifications of tire particles and solubilization of antibiotics adsorbed on these tire particles was reduced 54 . Chemical aging with potassium persulfate is very harsh and likely poorly representative of environmental aging whereas thermooxidation is only one of the weathering processes that could occur in the environment. More research on the impact of environmentally representative aging on solubilization of tire-associated chemicals is needed. Overall, for the more polar compounds, such as ANI, benzothiazoles and DPG, the bioaccessible fraction was higher than for the more hydrophobic PAHs (figure 2). This difference could be explained by the lower hydrophobicity of the former chemicals. Furthermore, the solubilization potential of the polar compounds was similar in water, SFGASTRIC and SFINTESTINAL, meaning that solubilization was hardly affected by the presence of enzymes and by the low pH of the SFgastric or by the presence of bile constituents in SFintestinal compared to mineral water. Contrastingly, the solubilization of PAHs was strongly affected by the nature of the digestion fluids. PAH concentrations were below LQ in the SFgastric and in water but were quantifiable in SFintestinal. Nonetheless, only a small fraction of the total PAH content of CMTT (0.06 -0.25%) was bioaccessible compared to the more polar compounds (> 1%). The critical micelle concentration in SFINTESTINAL calculated from contact angle measurements was approximately equal to 2000 mg bile L -1 (figure S6). The SFINTESTINAL used in this study contained 5000 mg bile L -1 , thus well above the critical micelle concentration, indicating that SFINTESTINAL was a micellar solution. The presence of stable micelles was confirmed by dynamic light scattering, which revealed a high concentration of micelles with an average size of 154.8 nm and a zeta potential of -10 mV. Micelles-mediated solubilization has been demonstrated for bisphenol A 26 and PCBs 41 and is also likely responsible for the solubilization of hydrophobic organic compounds such as PAHs. These results are consistent with previous studies that showed greater solubilization of hydrophobic contaminants, such as PAHs and PCBs, from sediment in gut fluids compared to water, both in vitro and in vivo 56,57 but also from various types of microplastics 58,59 . Micelles-mediated processes along with the hydrophobic nature of the gut fluids were also pointed out in these prior studies as drivers of the solubilization. In contrast, solubilization of only one (Butyl benzyl phthalate) of 12 estrogenic compounds was enhanced under simulated fish gut conditions compared to water 60,61 , indicating that the mechanisms facilitating solubilization of organic chemicals in digestive fluids might be compoundspecific. The in vitro digestion of CMTT along with surrogate fish prey (G. pulex) affected the solubilization of the tire-associated compounds. When comparing the digestion of CMTT alone to the coingestion scenario, only the solubilization of BT was not impacted by the addition of food (figure 3). The solubilization of ANI, HBT, MBT, 6PPD-Q and DPG was reduced by a factor of 1.8 to 5.6 in the coingestion scenario. In contrast, the solubilization of 6PPD, FLT, PYR and BPY was enhanced by a factor of 2.3 to 2.7 with the addition of food. The DOC of both fluids reached 12700 mg L -1 in the coingestion scenario but only 1250 mg L -1 without coingestion of food organic matter. Reduced solubilization of the most polar compounds in the presence of food could be due to the increased hydrophobic properties of the gut fluids due to the solubilization of organic matter from the food particles, preventing solubilization of the hydrophilic compounds. Oppositely, the enhanced solubilization of the PAHs and 6PPD were related to the more hydrophobic nature of the fluids in the coingestion scenario and to the presence of higher DOC concentration. The effect of DOC or dissolved organic matter (DOM) on the solubilization of PAHs from microplastic particles has been studied and the dissolution of Phenanthrene was enhanced by 3.7 fold when DOM increased from 0 to 1000 mg L - ## 162 . From a broader standpoint, the presence of increasing levels of DOC in aqueous solution favours the transfer of hydrophobic organic compounds from polymers 63,64 and nonaqueous phase liquids 65 to water. Finally, lower solubility of polar drugs and enhanced solubility of nonpolar drugs was observed in the fed state compared to the fasted state in simulated human intestinal fluids 66 . This corroborates our findings and highlights that solubilization of hydrophobic compounds in simulated gut fluids is not only controlled by the intra-particle diffusion but can also be impacted by external mass transfer as a function of the fluid's characteristics. As a consequence, a strong exponential relationship between the bioaccessible fraction and the octanol-water partition coefficient (Kow) of the compounds was found for both scenarios (with or without coingestion of G. pulex) (figure 4). The decreased solubilization of the more polar compounds and the increased solubilization of the apolar compounds in the coingestion scenario is emphasized by the lower slope of the regression line compared to CMTT digestion only (slope = -0.34 with CI95%: [-0.39; -0.30] and slope = -0.52 with CI95%: [-0.59; -0.45], respectively). This highlights that both the compound properties (log Kow) and the fluid's composition drive the solubilization of the tire-associated compounds in our model fish digestive fluids. It should be noted that 6PPD-Q was highly solubilized with regards to its hydrophobic properties (log Kow = 3.96 67 ) (figure 3 and 4). As 6PPD-Q is likely mainly formed at the surface of the CMTT particles by oxidation of 6PPD, its solubilization into the gut fluids was probably facilitated compared to other compounds that are distributed homogenously in the CMTT particles and for which bioaccessibility was lower. ## Limitations and environmental implications In our in vitro digestion experiments, only a small to moderate percentage of tire-associated compounds (between 0.4 % and 11.3 %) was found to be solubilized and bioaccessible in the simulated gastrointestinal fluids within a representative gut transit time for fish. These values are low compared to studies investigating the bioaccessibility of PAHs from microplastics where this parameter was assessed following loading of exogenous chemicals on the test material by an adsorption step 68 . Desorption of compounds adsorbed on the surface and micropores of a polymeric matrix is likely to be faster compared to tire-related compounds. Indeed, tire-related compounds are part of the blend of the polymer matrix, are homogeneously distributed within the particles and are more strongly bound to the matrix. Nonetheless, the in vitro digestion of 10 g of CMTT L -1 of digestive fluids without coingestion of G. pulex resulted in marked concentrations of tire-associated compounds, such that they can be compared with LC50 available in the literature. Concentrations of 6PPD, DPG, 6PPD-Q and MBT approached or were above LC50 values determined for salmonid species (table 2). It should be noted that LC50 were determined in vivo and account for all exposure routes whereas our study focused only on the exposure via the digestive track. The LC50 for 6PPD-Q was determined for rainbow trout and that recent studies showed contrasting toxicity of this compound for other fish species and crustaceans (LC50 from 0.095 -309 µg L -1 25-28,30 ) suggesting species-specific mode of actions for this compound 29 . 1.96 (27) One limitation of our study is the use of a high concentration of CMTT in the in vitro digestion experiments (10 g L -1 ) and that a scenario with a low food/CMTT ratio of 10 was tested. Although ingestion of TRWP by aquatic organisms, including fish, has been demonstrated 31 , the level of exposure of fish to TRWP remains poorly documented and the concentration of CMTT and the food/CMTT ratio used in our study are likely overestimated compared to an environmentally realistic scenario. Moreover, we used a static in vitro digestion setup mimicking a finite bath digestion scenario. This type of experiment does not account for the passive diffusion of the compounds across the small intestine which will cause a disequilibrium between the CMTT and the digestive fluid and create a concentration gradient for further solubilization of contaminants from CMTT . Another limitation of this study is that it relies on the use of CMTT as a surrogate for environmental TRWP. It has been demonstrated that the chemical content of TRWP is not identical to that of pure tire tread due to encrustation of minerals and organic constituents originating from the road pavement 5,8 . The different surface areas of CMTT and TRWP could impact the solubilization kinetics of the associated compounds as well as their overall bioaccessibility. Furthermore, TRWP will undergo various types of weathering (thermooxidation, photodegradation, mechanical shear stress, biodegradation) once released in the environment that may affect its chemical and physical properties 7 . The effects of aging on tire particles was addressed in this study via exposure of CMTT to thermooxidative conditions, which led to a small fraction of 6PPD being converted into 6PPD-Q (0.09%) (table 1). It is likely that other yet unknown transformation products of tire-associated chemicals were formed. Nevertheless, the bioaccessibility of tire-associated compounds did not vary significantly in artificially aged CMTT and unaltered CMTT. As in our case, the artificial aging of CMTT only consisted of thermooxidative conditions and other weathering mechanisms (photooxidation, biodegradation) could come into play. Further studies should take these other aging processes into account and investigate the bioaccessibility of organic compounds from such aged tire particles and TRWP. Overall, our study shows that the ingestion of CMTT by fish, as a surrogate for environmental TRWP, could lead to exposure of a cocktail of identified and probably other still unknown chemicals. The coingestion of food organic matters impacted the bioaccessible fraction of the chemicals in the fish digestive fluids. Thus, the bioaccessibility and further uptake of tire-associated compounds by the epithelial cells and related toxicity to fish based on refined environmental concentrations should be investigated. Finally, the bioaccumulation potential of the more hydrophobic tire-associated chemicals (6PPD-Q, DPG) needs to be determined. ## Supporting information: Details regarding the chemical analyses of the tire-associated compounds, the physical and chemical characteristics of Cryogenically Milled Tire Tread (CMTT) and of the simulated gastrointestinal fluids, surface tension of Simulated Intestinal Fluid (SFINTESTINAL), solubilization kinetics of tire-associated compounds from unaltered and artificially aged CMTT in water, SFGASTRIC and SFINTESTINAL, the profile of the 16 measured PAHs in unaltered CMTT and aged CMTT can be found in the supplementary information.

chemsum

{"title": "Bioaccessibility of organic compounds associated with tire particles using a fish in vitro digestive model: solubilization kinetics and effects of food co-ingestion", "journal": "ChemRxiv"}

nickel-catalyzed_trifluoromethylthiolation_of_csp²–o_bonds

## Abstract: While nickel catalysts have previously been shown to activate even the least reactive Csp 2 -O bonds, i.e. aryl ethers, in the context of C-C bond formation, little is known about the reactivity limits and molecular requirements for the introduction of valuable functional groups under homogeneous nickel catalysis. We identified that due to the high reactivity of Ni-catalysts, they are also prone to react with existing or installed functional groups, which ultimately causes catalyst deactivation. The scope of the Ni-catalyzed coupling protocol will therefore be dictated by the reactivity of the functional groups towards the catalyst. Herein, we showed that the application of computational tools allowed the identification of matching functional groups in terms of suitable leaving groups and tolerated functional groups. This allowed for the development of the first efficient protocol to trifluoromethylthiolate Csp 2 -O bonds, giving the mild and operationally simple C-SCF 3 coupling of a range of aryl, vinyl triflates and nonaflates.The novel methodology was also applied to biologically active and pharmaceutical relevant targets, showcasing its robustness and wide applicability. ## Introduction Owing to nickel's non-precious nature and its higher reactivity in the frst elementary step of cross coupling cycles, i.e. the oxidative addition, the feld of homogeneous Ni-catalysis has long been considered promising, yet also challenging. 1 This is because difficulties have frequently been encountered in taming nickel's reactive nature to achieve desired selectivities and scope. 2 In spite of that, in recent years there has been impressive progress in the activation of the least reactive bonds, such as aromatic ethers or aryl fluorides. 3 However, these milestones typically featured the conversion of C-OMe (or C-F 4 ) to inert C-C or C-H bonds. 5,6 By contrast, less is known about the reactivity limits and molecular requirements for the installation of potentially reactive functional groups. We therefore envisioned that a computationally assisted development 7 of an unprecedented Nicatalyzed protocol for C-heteroatom bond formation presents an ideal challenge to (i) identify the general reactivity requirements for efficient Ni-catalysis and (ii) demonstrate the viability of applying computational tools to assess substrate scope. As a suitable test case, we focused on the nickel-catalyzed trifluoromethylthiolation of Csp 2 -O bonds. 8 The SCF 3 group makes molecules more lipophilic, increasing their membrane permeability and bioavailability. 9 These properties are of considerable interest in a pharmaceutical and agrochemical context. Consequently, numerous efforts have been undertaken to synthesize aryltrifluoromethyl sulfdes. 10,11 In particular the direct catalytic introduction of SCF 3 is an attractive approach. While aryl halides 12 or boronic acids 13 have successfully been converted to C-SCF 3 via metal catalyzed crosscoupling strategies or oxidative protocols, 14 to date, there is no report of a direct and catalytic trifluoromethylthiolation of Csp 2 -O bonds. ## Results and discussion Given the widespread abundance of phenols, the tri-fluoromethylthiolation of phenol derivatives would be highly attractive for synthetic diversity. In this context, the scope could in principle range from more activated derivatives (e.g. aryl tri-flates) to the least reactive derivatives, i.e. aryl ethers which are present in biomass feedstocks (such as lignin 15 ). 6 However, while Ni-catalysis has recently been successfully utilized to activate aromatic ethers, 3 we hypothesized that there might be a fundamental reactivity conflict in introducing SCF 3 : the created SCF 3 -product would be expected to be inherently more reactive towards oxidative addition 16 which may impede further transformation. To test this, we subjected Ni(cod) 2 /dppf to PhSCF 3 1 (see Fig. 1). We recently showed that this system triggers the mild trifluoromethylthiolation of aryl chlorides, proceeding via Ni (0) / Ni (II) catalysis with [(dppf)Ni(cod)] formed as the active catalyst. 12e In accordance with our hypothesis, the reaction of the [Ni (0) ] catalyst with PhSCF 3 is indeed seen, even under mild reaction conditions (45 C), as judged by 31 P-NMR spectroscopic analysis. A complete disappearance of the characteristic 31 P-NMR singlet signal of [(dppf)Ni (0) (cod)] (33.8 ppm) 12e occurred, and the formation of a new species was seen that appears as two triplets at 30.8 ppm (with J ¼ 23.0 Hz) and at 22.1 ppm (with J ¼ 37.6 Hz) by 31 P-NMR spectroscopic analysis (see Fig. 1). While our efforts to structurally characterize the latter by X-ray crystallography have so far been unsuccessful, the formed species clearly constitutes a catalyst deactivation product. The subjection of this species as a catalyst (or also stoichiometrically) in the trifluoromethylthiolation of aryl chlorides did not yield ArSCF 3 . This indicates that oxidative addition by a [Ni (0) ] catalyst to the product is facile and eventually leads to catalytically inactive species. To achieve productive catalysis and high overall conversion, it is therefore of utmost importance to avoid this deactivation process. Our computational assessment 17 of the oxidative addition of [(dppf)Ni(cod)] to Ph-SCF 3 1 suggests an activation free energy barrier of DG ‡ ¼ 19.2 kcal mol 1 , and it uses the M06L method with a CPCM solvation model to account for toluene and the mixed 6-311++G(d,p) and LANL2DZ (for Ni, Fe) basis set. 17,18 This value now sets the bar for the possible reaction scope. The 'to-be-transformed' bond must show a barrier lower than 19.2 kcal mol 1 to avoid catalyst loss via an unproductive reaction with the product (ArSCF 3 ). ## Identication of suitable leaving groupscomputational assessment & experimental tests We subsequently undertook computational studies to identify matching leaving groups 'OR' (Fig. 2) that would show the desired greater reactivity than the Csp 2 -SCF 3 bond. For the cleavage of the C-O bonds, mechanistic support for Ni (0) /Ni (II)5i,6 and also Ni (I) -catalysis 19 has previously been reported. However, on the basis of our previous mechanistic study 12e and the observation that (dppf)Ni (I) Cl is ineffective as a catalyst in C-SCF 3 bond formation, 12e,20 as a frst approximation, we calculated the activation barrier of oxidative addition using [(dppf) Ni (0) (cod)] to a range of phenol derivatives (Ph-OR), with R ¼ alkyl (ether), R 0 C]O (pivalate), SO 2 R 00 (sulfonic esters). Fig. 2 presents the results. This computational assessment suggests that in the context of C-O to C-SCF 3 conversion, the inherently high reactivity of C-SCF 3 only allows for triflate precursors as suitable starting materials. Alternative C-O leaving groups that have previously been employed in the Ni-catalyzed construction of inert C-C bonds, such as aryl ethers (OMe), mesylates (OMs), tosylates (OTs) or pivalates (OPiv) 3,6 are predicted to be incompatible with Ni (0) -catalyzed trifluoromethylthiolation, as they would generally be less reactive than Ar-SCF 3 , hence favoring catalyst deactivation via reaction with the product. 21 To experimentally test this computationally predicted trend, we subjected Ni(cod) 2 /dppf along with the easily accessible SCF 3 -source (Me 4 N)SCF 3 to Ar-OR derivatives (in toluene at 45 C), ranging from the predicted low (aryl ether) to high (aryl triflate) reactivity (Fig. 2). In accordance with expectations, at best, a low conversion was seen for phenyl mesylates (5%), tosylates (1%) or pivalates (0%). In stark contrast, phenyl triflate showed excellent conversion to PhSCF 3 (83%). We additionally followed the conversion ArOTf / ArSCF 3 with ReactIR®. This analysis showed that the transformation was rapid, being essentially complete in 1.5 h with only little increase in conversion over the subsequent hours (see ESI, Fig. S1 †). We also analyzed the reactions of those substrates that showed little conversion (#5%), i.e. ArOMs and ArOTs, by 31 P-NMR spectroscopic analyses. We observed that essentially all of the [Ni (0) ] catalyst had transformed to the catalytically inactive species described in Fig. 1 within 3 h reaction time. This clearly highlights that while [Ni (0) ] is in fact capable of reacting with Ph-OMs or -OTs, the catalyst is rapidly consumed as soon as some of the more reactive PhSCF 3 molecules are generated. This corroborates with the strict requirement of suitably matching functionality and tailored reactivity progression from a "more" to "less reactive" functionality. ## Computational assessment of functional group tolerance We subsequently set out to test for the generality of the iden-tifed Ni-catalyzed trifluoromethylthiolation of activated C-O bonds and computationally assess the functional group (FG) tolerance (see Fig. 3). As we determined a barrier of DG ‡ ¼ 14.4 kcal mol 1 for the oxidative addition of [(dppf)Ni (0) (cod)] to Ph-OTf, all additional functional groups (FG) in the substrates will only be compatible if the reactivity of the C-FG bond is lower than that of Ph-OTf. The computational results depicted in Fig. 3 suggest a tolerance of the protocol to ketone functional groups, C-C or benzylic C-O bonds. In all cases, the requirement of DG ‡ C-FG > 14.4 kcal mol 1 is fulflled. Even aromatic C-CN bonds that were previously shown to be reactive under Ni-catalysis conditions 22 are predicted to be compatible. ## SCF 3 -coupling of aryl triates On the basis of this computationally guided substrate scope, we subjected a range of aryl triflates to standard catalysis conditions. Table 1 presents the results. A number of aryl-and heteroaryl triflates were coupled in good to excellent yields. The transformation was compatible with ketone (6, 7 and 8, Table 1), ether (9) and cyano (5) functional groups. Two heterocyclic examples (10, 11) were also trifluoromethylthiolated in good yields (see Table 1). We next searched for bioactive molecules of greater complexity that would fulfl our reactivity requirements and show compatibility with the computationally predicted scope. Estrone (an estrogenic hormone), 6-hydroxy flavanone (a plant secondary metabolite used inter alia as an antioxidant) and dtocopherol (vitamin E) show an excellent functional group match, containing predominantly ketone and benzylic C-O bonds that are predicted to be less reactive than C-OTf and C-SCF 3 . Trifluoromethylthiolation was successfully accomplished in 62-96% yield, highlighting the potential of this method for pharmaceutical applications (see Scheme 1). ## SCF 3 -coupling of vinyl triates Vinyl SCF 3 -compounds are also of signifcance, fnding applications as herbicides for example. 23 However, the current methodological repertoire to access these compounds relies predominantly on indirect strategies 24 or requiring stoichiometric amounts of metal. 13b, 25 The direct construction of C vinyl - SCF 3 in a catalytic manner would be a highly attractive approach. It has been accomplished via the Cu-catalyzed tri-fluoromethylthiolation of vinyl boronic acids with electrophilic SCF 3 -sources. 13c-e In a nucleophilic context, the catalytic installation of C vinyl -SCF 3 is limited to vinyl iodides and requires harsh reaction conditions (110 C). 26 A mild Ni-catalyzed conversion of readily accessible C vinyl -OR derivatives to C vinyl -SCF 3 would thus substantially widen the synthetic repertoire. Our calculation of the barrier for the oxidative addition of [Ni (0) ] to C vinyl -SCF 3 indicated DG ‡ ¼ 18.8 kcal mol 1 . This barrier constitutes the upper limit for the reactivity of a potential leaving group (OR). C vinyl -OPiv and C vinyl -OMs show higher or similarly high barriers for oxidative addition (DG ‡ ¼ 22.1 and 17.7 kcal mol 1 ) and are hence ruled out. C vinyl -OTf on the other hand is predicted to be highly reactive (DG ‡ ¼ 5.2 kcal mol 1 ) and should hence be a compatible match. After applying standard catalysis conditions, 27 we successfully transformed a number of vinyl triflates to the corresponding trifluoromethylthiolated counterparts (see Table 2). The protocol proved to be compatible with a heterocyclic moiety (20, Table 2), a benzyl protecting group (17), and was successful for fully aliphatic (15) as well as conjugated (18, 19) vinyl triflate derivatives. Compound 19 (Table 2) was afforded in a slightly lower yield (44%). However, upon closer inspection, it became clear that this was related to the inherent instability of the vinyl triflate starting material. ## Assessment of aryl and vinyl nonaates We therefore shifted our attention to potentially more stable analogues and considered nonaflates. 28 Both, aryl and vinyl nonaflates are computationally predicted to be compatible with Ni-catalyzed trifluoromethylthiolation, showing similarly low or even lower barriers for oxidative addition by [Ni (0) ] than the corresponding triflates (DG ‡ ¼ 4.8 for addition to C vinyl -ONf and DG ‡ ¼ 10.6 kcal mol 1 for addition to Ph-ONf). In accordance with these computational predictions, excellent conversions to aryl-and C vinyl -SCF 3 were observed (see Table 3). Particularly notable is the synthesis of 19 0 (Table 3) which was now highyielding (as opposed to its preparation in Table 2), reflecting the greater robustness of vinyl nonaflates over vinyl triflates. 29 ## Conclusions The inherently high reactivities of Ni-catalysts may be fundamentally at conflict with introducing a wide range of functional groups, as shown here for the introduction of the pharmaceutically and agrochemically valuable SCF 3 group. We identifed that the reaction of the Ni-catalyst with the desired product, ArSCF 3 , triggers undesirable catalyst deactivation reactions that ultimately inhibit catalysis. The overall substrate scope is therefore dictated by the reactivity of the desired functionality towards the catalyst (here: C-SCF 3 ). The application of computational tools allowed for the identifcation of matching Table 2 Ni(0)-catalyzed trifluoromethylthiolation of vinyl-OTf a a Ni(cod) 2 (5.5 mg, 0.02 mmol), dppf (11.1 mg, 0.02 mmol), vinyl triflate (0.2 mmol), (Me 4 N)SCF 3 (52 mg, 0.3 mmol), PhCN (20.6 mg, 0.2 mmol), 27 toluene (1 mL), under inert atmosphere, isolated yield. b Yield determined by 19 F-NMR analysis using PhCF 3 as the internal standard. functional groups in terms of suitable leaving groups and tolerated functional groups. As a result, the frst Ni-catalyzed C-SCF 3 coupling of aryl and vinyl C-O bonds has been developed. Given the highly reactive nature of C-SCF 3 , only those C-OR derivatives of even greater reactivity, i.e. triflates and nonaflates, allow for efficient C-SCF 3 coupling. The protocol is mild, general and operationally simple. Given that computational methods, software and hardware have evolved to a level, at which calculations can nowadays frequently be done faster than experiments, 30 we anticipate that the herein applied approach will fnd applications in the development of, but not limited to, homogeneous Ni-catalysis.

chemsum

{"title": "Nickel-catalyzed trifluoromethylthiolation of Csp²\u2013O bonds", "journal": "Royal Society of Chemistry (RSC)"}

environment-controlled_post-synthetic_modifications_of_iron_formate_frameworks

## Abstract: New hybrid iron-formate perovskites have been obtained in high-pressure reactions. Apart from the pressure range, also the liquid environment of the sample regulates the course of transformations. Formate α-DmaFe 2+ Fe 3+ For6 (Dma = (CH3)2NH2 + , For = HCOO -), when compressed in oil and in isopropanol at 1.40 GPa, transforms to a new phase γ, different than that phase β obtained at low-temperature. In glycerol phase α can be compressed to 1.40 GPa, but then reacts to DmaFe 2+ For3, with all Fe(III) cations reduced, surrounded by amorphous iron formate devoid of Dma cations. Another mixed-valence framework Dma3[Fe 2+ 3Fe 3+ For6]2•CO2, can be produced from phase α incubated in methanol and ethanol at 1.15 GPa. These pressure-induced environment-sensitive modifications have been rationalised by the volume effects in transforming structures, their different chemical composition, voids, ligands and cation oxidation states switching between Fe(II), Fe(III), their high-and low-spin states as well as solubility, viscosity, molecular size and chemical character of pressure transmitting media. The topochemical redox reactions controlled by pressure and the liquid environment offer new highly efficient, safe and environmentfriendly reactions leading to new advanced materials and their post-synthesise modifications.The hybrid inorganic-organic frameworks depending on the cations, ligands, linkers their connectivity and voids, display various attractive properties. [1][2][3][4][5][6][7][8][9][10][11][12][13] In these multifunctional materials, the manifold functionalities lead to new cross-effects of higher order, both physical and chemical in nature. Particularly interesting are post-synthetic modifications (PSM) of metal-organic frameworks (MOF's). 14,15,[24][25][26][27][28][29][30][16][17][18][19][20][21][22][23] The PSM can be used for fabricating in-situ, under specific conditions, new materials with required properties. The PSM solid-state transitions and reactions are often initiated by physical stimuli. Owing to the self-contained reaction space, requiring no additional substrates, the PSM can be invaluable for green technologies and for obtaining sophisticated advanced materials on requested sites. This topochemical reactions can be stimulated by light, temperature and pressure. The typical high-pressure effects, such as tighter molecular packing, increased density, compressed coordination bonds, shorter distances to the ligands and modified crystal fieldall can destabilise material structure. Several rules were formulated for describing the pressure effects in various compounds, including elements and minerals by Prewitt and Downs. 40 Their rule 4 states that the coordination number increases with pressure, and rule 5 that the non-metal atoms (usually oxygen in minerals) are stronger compressed than cations. 40,41 Most recently we showed that these rules also describe the effects of pressure on the coordination polymers (CPs) and porous MOFs: The increased coordination number in the high-pressure phases of CPs originates from the stronger compression of anions compared to central cations; and this rule can be extended to the temperature dependence of time-averaged volumes of central and ligand atoms. 42 Hence the inverse relation between the pressure and temperature effects was observed in numerous reactions and transformations of CPs. 42,43, Presently, we further explore the pressure-induced PSMs. By choosing the mixed-valence perovskite-like iron-formate framework with dimethylamine cations (Dma), we have introduced into the system additional instabilities of low-spin and high-spin states, the oxidation changes between Fe 2+ and Fe 3+ cations and possible relocation, disproportionation and reactions involving the formate and amine ions. The strongly increased energy of interactions can result in electronic transitions, spin-state changes and even in redox reactions. Spin crossover (SC) phenomena in transition-metal compounds under pressure were reviewed by Drickamer et al. 57,58 Since then the SC transformations have been observed in many complexes of divalent and trivalent iron, cobalt, nickel, chromium and manganese. 42 However to our knowledge, no pressure-promoted SC transformations coupled to the redox reactions in MOFs have been reported. Here we describe several high-pressure transformations involving electronic transformations for the [(CH3)2NH2 + ][Fe 3+ Fe 2+ (HCOO -)6], hereafter denoted as α-DmaFe 2+ Fe 3+ For6 (Figure 1). Moreover, we have observed that the PSMs of iron formates strongly depend on their liquid environments. ## Discussion The structural features of the starting material are essential for understanding its PSMs. The ambient-pressure hybrid perovskite α-DmaFe 2+ Fe 3+ For6, of trigonal space group P3 ̅ 1c (Table 1), is stable in different hydrostatic media nearly up to 1.40 GPa. As previously reported, 59,60 all iron cations in α-DmaFe 2+ Fe 3+ For6 are octahedrally coordinated by formate anions in anti-anti configuration. Every second cavity of the framework is occupied by the Dma cation and every other is empty. At ambient pressure, above 155 K, 59,60 the Dma cation is disordered in this way that the two methyls reside on the 3-fold axis and around the axis, the nitrogen atom is distributed between three equivalent sites. Below 155 K, the onset of ordering of Dma cations breaks the symmetry and transforms the crystal to phase β, of space group R3 ̅ c. 61,62 In the framework, each Fe(III) cation is surrounded by six octahedrally arranged Fe(II), each of which is likewise surrounded by six Fe(III) cations. Bonds Fe(II)-O and Fe(III)-O can be distinguished by their length, 2.12 and 2.01 , respectively, as the Fe(II) cation is richer in one electron and hence larger than Fe(III). We have established for a series of iron formates the bondlength difference of about 0.1 between Fe(II)-O and Fe(III)-O is hardly affected by the pressure up to 1.2 GPa. Whereas the continuous compression reduces the Fe-O bond distance by less than 1% GPa -1 , the electronic transition between Fe(II) and Fe(III) as well as the SC transition result in over 10% Fe-O distance changes. Depending on the pressure range and solvent, α-DmaFe 3+ Fe 2+ For6 transforms its framework and electronic states of the iron cations in three different ways (Figure 1). The different ionic radii in ferric and ferrous cations make the Fe-O distances highly responsive to the pressure changes. The extent of Fe-O bond changes accumulated of the monotonic compression, electronic transitions and strains in the crystal lattice is illustrated for selected iron formates in Figure 2. ## Monotonic compression of α-DmaFe 2+ Fe 3+ For6 The high-pressure behaviour of α-DmaFe 2+ Fe 3+ For6 by itself is quite unique. Up to 1.0 GPa phase α is hardly compressed along plane (001) and then, up to 1.40 GPa, it displays the extremely rare effect of negative area compressibility (NAC) 56, (Figure 3). On approaching the phase transition at 1.40 GPa both the NAC along plane (001) and the strong linear compression along 68,69 The insets graphically represent the strain tensors of the crystals at the low-pressure limit of their stability regions. 70 The magnetic properties of α-DmaFe 3+ Fe 2+ For6 are consistent with high-spin configurations of the iron cations. 60 The configuration of electrons in the partly filled 3d shell affects not only the potential energy and size of the cation, but also its reactivity. According to Hund's rule, the electronic configuration of maximum-multiplicity (highest-spin) is favoured. Thus, the ferrous cation Fe(II) adopts the outer shell 3d 6 4s o , and it changes to 3d 5 4s o for the cation reduced to ferric Fe(III). Both these cations can further differentiate into high-spin or lowspin states. The high-spin ferric state, with five orbitals each containing one electron, is spherical ( 6 A1). The high-spin ferrous state is in an asymmetric ( 5 T2) ground state. On the other hand, the low-spin ferrous state is spherically symmetric ( 1 A1), while the low-spin ferric state ( 2 T2) is not. In compressed high-spin α-DmaFe 2+ Fe 3+ For6, the ligand field increases and the interelectronic repulsion (Racah parameter) decreases. 71 Thus, on one hand, the high-pressure increases the energy of all orbitals; and on the other, the interelectronic repulsion can be reduced by the spin-pairing energy. The lower energy of interelectronic interactions of the antiparallel spins, consistent with Pauling's principle, can be also associated with the reduction of the volume of electronic orbitals, with the reduction of ionic and atomic radii. These effects combine the high/low-spin transitions with the pressure stimuli. Other external stimuli, like temperature and pH of the solvent, can further affect the ligand field. Such combined effects can cumulate for triggering phase transitions and topochemical reactions. Generally, the coordinating electronegative atoms or anions are larger and stronger compressed than cations. Consequently, high-pressure usually favours the equilibrium of smaller anions and neutral structural units, even when achieved at the cost of a small increase of the cations 4). Moreover, the pressure can trigger a composition change reducing the overall volume of the crystals and its environment, treated as a whole system. The volume reduction can be achieved either by phase transitions as well as topochemical reactions. For example, it was shown that high-pressure, through the chemical reactions, promotes the reduced oxidation states in the series of oxides Fe2O3, Fe3O4, Fe4O5, Fe5O7, Fe7O9 and FeO. 57,58, For the iron formates under this study, the oxidation states of the Fe cations are the derivative of the crystal structure, ligands conformation, voids and of the compounds composition and it possible changes. Several types of pressure-induces transformations of iron formates reported in the literature and this study illustrate the interplay of various structural components and their compressed environment. ## Compression in Daphne 7474 oil and Isopropanol At 1.40 GPa, α-DmaFe 2+ Fe 3+ For6 reversibly transforms to phase γ, of monoclinic space group P21/c (Figure 2, Table 1). In phase γ, ## Squeezing Dma off α-DmaFe 2+ Fe 3+ For6 in glycerol In glycerol α-DmaFe 2+ Fe 3+ For6 is compressed like in Daphne oil and isopropanol up to 1.4 GPa (Figure 3), but then several spectacular macroscopic effects are clearly visible (Figure 7). The black single crystal of phase α gradually changes its shape and becomes transparent between the crystal surface and the dark central part. This colour change of the outer part and its multi-grain texture is clearly different from α-to-γ phase transition. On releasing pressure the transparent region is reduced till the sample is fully black again. This process could be repeated several times, after which the initial sharp edges of the sample become first rounded and next shattered. The X-ray diffraction measurements confirmed that in glycerol phase α is compressed like in Daphne oil and isopropanol, but then a sudden transformation takes place to a new phase (Figure S1). Only one single crystal diffraction pattern was present above 1.40 GPa, despite that two different (transparent-outer and dark-internal) phases were clearly visible. By solving the crystal structure of the new diffraction pattern, we found that this is DmaFe 2+ For3 (Table 1), of trigonal space group R3 ̅ c. DmaFe 2+ For3 was synthesised before at ambient pressure and it forms transparent crystals, 80 which is consistent with the Fe(II) oxidation, and it could be an indication that the transparent-outer part of the sample corresponds to the new compound (Figure 7). The compression of DmaFe 2+ For3 was measured by Collings et. al. and it is consistent with our measurements in glycerol (Figure 2b). However, in our experiment, the DmaFe 2+ For3 crystal disappears below 1.40 GPa and α-DmaFe 2+ Fe 3+ For6 appears again. This it is apparent that the pressure-induced topochemical reaction in the single crystal of α-DmaFe 2+ Fe 3+ For6 compressed in glycerol leads to two products at least, one of which has been identified by X-ray diffraction as DmaFe 2+ For3 and other(s) is(are) amorphous or in the form of fine powder, as no other diffraction pattern was observed. This reaction can be described by equitation: 𝛼-DmaFe 2+ Fe 3+ For 6 81 The Fe 2+ For2 crystals are transparent and their formation is consistent with the outer part of the sample above 1.40 GPa. The pressure-induced reduction of iron from Fe(III) to Fe(II) was described by Drickamer et. al. 71 With increasing pressure the energy of 3d orbitals (placed against the ligand orbitals) lowers, at the same time increasing iron affinity for electrons. The formate anions possess empty π orbitals, they can bond to the (filled) t2g(π) orbitals of the metal atom by the back donation of metal electrons. The reduction is accomplished at the point, when the electron from a nonbonding ligand orbital is transferred to the metal 3d antibonding orbital. This conversion can lead either to a free radical formed at a ligand site, or to an electron hole circuling between the adjacent ligands. 71 The reversible transformations observed by us visually (Figure 7) and by X-ray diffraction in the diamond-anvil cell (DAC) on increasing and reducing the pressure through 1.40 GPa contrast with the reported stability of DmaFe 2+ For3 and Fe 2+ For2 compounds. However, both these transparent phases are separated be the transition zone of collapsed strongly strained (black) layer of phase X. Apparently, this dark region on the release of pressure has the potential to expand into the framework with cages ready to and receive back the Dma cations, which reconstructs the DmaFe 2+ Fe 3+ For6 phase α. ## Irreversible CO2 templated reaction in MeOH, EtOH We detected no signs of anomalous behaviour for α-DmaFe 2+ Fe 3+ For6 compressed in methanol and ethanol to about 1.10 GPa. However, at 1.15 GPa, the crystal slowly dissolved and small cubic prisms precipitated (Figure 8). We recovered them to ambient pressure and established by X-ray diffraction that they are formed of Dma3Fe 2+ 3Fe 3+ For12•CO2. Their cubic symmetry of space group Im-3 and the structure is analogous to a group of metal M(III) formates of general formula where M=Mn, Fe, Al, Ga and In. 82,83 They adopt the ReO3-type structure, built of FeO6 octahedra sharing formate anions at the vertices. The neutral CO2 molecules and Dma cations trapped in the cages equilibrate the 3:1 ratio of Fe 2+ and Fe 3+ cations (Figure 9). It appears that the main reason for the different effect of methanol and ethanol is that they better dissolve α-DmaFe 2+ Fe 3+ For6 than isopropanol, Daphne oil and glycerol. Consequently, phase α dissolves and some formates decompose to CO2 and H2. This reaction involves an electron transfer from the formate non-bonding level to the metal dπ orbitals, according to: The crystal structure of Dma3Fe 2+ 3Fe 3+ For12•CO2 is similar to that of DmaFe 2+ For3 in this respect that all cages are filled. However, in Dma3Fe 2+ 3Fe 3+ For12•CO2 every 4 th cage is filled with the neutral CO2 molecule. It is known that many formate salts of the general formula M For (H2O)x are prone to decarboxylation. For example, hydrated nickel formate decarboxylases at about 473 K, yielding the fine powder of pure metallic nickel. 84 Moreover, the catalytic effect of Fe(BF4)2•6H2O for the dehydrogenation of formic acid is highly efficient. 85 These liquids are different in several respects, and it appears that the most significant properties in this respect are molecular volume, viscosity and the hydrostatic limit. The liquids also differ in the types of intermolecular interactions, with the cages and that all these properties affect the penetration of molecules into the framework. 87 The highly viscous Daphne 7474 and isopropanol consist of large molecules, unlikely to penetrate the α-DmaFe 2+ Fe 3+ For6 structure. It appears that glycerol is similar, but it is much more hydrophilic in interactions. Finally, small molecules od methanol and ethanol can penetrate the pores and in this way affect the compression. ## Conclusions When exposed to different external stimuli, the hybrid iron-formate perovskite α-DmaFe 2+ Fe 3+ All these effects can be applied for planning the PSMs. Most importantly, this study on α-DmaFe 2+ Fe 3+ For6 and its PSM products has revealed a variety of transformations and reactions that broaden the general understanding about the chemistry in extreme conditions, and in particular the subtle effects most relevant to soft and highly sensitive hybrid metal-organic perovskites. ## Synthesis of α-DmaFe 2+ Fe 3+ For6 Black single crystal of α-DmaFe 2+ Fe 3+ For6 was synthesised according to the procedure reported by Zhao et. al. 59 1 gram of the FeCl3•6H2O crystals were dissolved in mixture of DMF and 88% formic acid and heated at 140°C for several days in autoclave. 89 After that, black crystals of DmaFe 2+ Fe 3+ For6 were mechanically separated, washed and dried. The Dma cations caged in the formate framework come from the hydrolysis of DMF used as the solvent for reaction. ## Synthesis of Dma3Fe 2+ 3Fe 3+ For12•CO2 Dma3[Fe 2+ 3Fe 3+ For6]2•CO2 was synthesised from α-DmaFe 2+ Fe 3+ For6 compressed to 1.15 GPa in methanol or ethanol. During 5 days, the phase α crystal gradually dissolved and the reaction yielded small green cubic crystals (Figure S3). Single crystal X-ray measurements were performed on a SuperNova diffractometer with a micro-focus source (CuKα = 1.54178 ) revealed that the new product is Dma3Fe ## High-pressure Structural Measurements A black trigonal single crystal of the α-DmaFe 2+ Fe 3+ For6 has been mounted in a Merrill-Bassett diamond-anvil cell (DAC) chamber, 90 then filled with isopropanol, Daphne 7474 oil, glycerol and methanol, ethanol and isothermally compressed. The pressure inside the DAC was calibrated by the ruby-fluorescence method. 91,92 The gaskets of 0.3 mm stainless steel foil with spark-eroded holes 0.5 mm in diameter were used. The DAC was centered by the gasket-shadow method. 93 The crystal compressed was measured by single crystal X-ray diffraction on an Xcalibur Eos-CCD and Kuma Eos-CCD 4-circle diffractometers (MoKα= 0.71073 ). The optimum diffractometer settings for measuring the reflection intensities were applied. 94 Crystallographic data were collected and preliminarily reduced with the CrysAlisPro Version 1.171.33. 95 The structure was solved by direct methods in program Shelxs and refined with Shelxl using the Olex2 suite. The final crystal data are summarised in Tables 1 and 2 (cf.

chemsum

{"title": "Environment-controlled post-synthetic modifications of iron formate frameworks", "journal": "ChemRxiv"}

water_adsorption_on_tio2_surfaces_probed_by_soft_x-ray_spectroscopies:_bulk_materials_vs._isolated_n

## Abstract: We describe an experimental method to probe the adsorption of water at the surface of isolated, substrate-free TiO 2 nanoparticles (NPs) based on soft X-ray spectroscopy in the gas phase using synchrotron radiation. To understand the interfacial properties between water and TiO 2 surface, a water shell was adsorbed at the surface of TiO 2 NPs. We used two different ways to control the hydration level of the NPs: in the first scheme, initially solvated NPs were dried and in the second one, dry NPs generated thanks to a commercial aerosol generator were exposed to water vapor. XPS was used to identify the signature of the water layer shell on the surface of the free TiO 2 NPs and made it possible to follow the evolution of their hydration state. The results obtained allow the establishment of a qualitative determination of isolated NPs' surface states, as well as to unravel water adsorption mechanisms. This method appears to be a unique approach to investigate the interface between an isolated nano-object and a solvent over-layer, paving the way towards new investigation methods in heterogeneous catalysis on nanomaterials. Titanium dioxide (TiO 2 ) is undoubtedly one of the most studied materials owing to its technological relevance to various fields, such as photonics, electronic devices, self-cleaning materials and photocatalysis [1][2][3][4] . Considerable research effort has been devoted to the understanding of the link between the surface properties of TiO 2 and water adsorption mechanisms on its surface [5][6][7][8] . These mechanisms are known to be essential in photocatalytic processes, affecting for instance charge recombination rates 9,10 . The water -TiO 2 interface is thus of crucial importance, and the proper control of the surface properties of TiO 2 appears to be even more crucial, as soon as we reach the nanometer scale where the surface-to-bulk ratio is considerably larger than in the infinite solid. Several studies, based on techniques such as high resolution scanning tunneling microscopy (HRSTM) 7,11,12 or X-ray Photoelectron Spectroscopy (XPS) 13,14 , have been reported in the literature and aimed at understanding and controlling TiO 2 nanoparticles (NPs)' surface properties. However, despite the wide interest devoted to this subject, the photocatalytic activity of TiO 2 NPs and its correlation with water adsorption is still a matter of debate both theoretically and experimentally because of the complex interplay between the surface structure at the atomic scale and the nature of the adsorption mechanisms 5,7, . The degree of complexity is also enhanced by the fact that the NPs are usually deposited on a substrate, resulting in sample modifications during the deposition process itself, interactions between the substrate and nanosystem under study, and sample charging effects. Identifying the key factors influencing the adsorption mechanisms and mastering the degree of hydration of TiO 2 NPs are important challenges for photocatalysis. Here we address the water adsorption problem on the surface of isolated TiO 2 NPs using a novel experimental technique, which has recently proved its efficiency in the characterization of isolated nano-objects . Our approach consists in using synchrotron radiation (SR) based soft X-ray electron spectroscopy to analyse the properties of a collimated beam of differently hydrated NPs generated and focused to the interaction region with the SR by an Aerodynamic Lens System (ADLS). This experimental approach offers the opportunity of avoiding any interaction between the sample and a substrate, thus giving access to the sole, intrinsic information about the NP surface. Several questions have been addressed to evaluate the feasibility of controlling the hydration state of freestanding TiO 2 NPs and to achieve insight into the factors which can influence the water adsorption mechanisms on isolated TiO 2 NPs in the gas phase. ## Results A variety of studies was preliminary conducted in view of the structural characterization of the commercial TiO 2 nanopowder. The TEM images obtained for the TiO 2 nanoparticles reveal an important size dispersion of the nanometer grains (Fig. 1), ranging from 20 nm to 120 nm. The sample from Sigma Aldrich, made of a mixture of the two prevalent crystalline phases of TiO 2 , namely rutile and anatase (Fig. 1a) shows clearly the presence of two distinguishable morphotypes on the grain structurea faceted and a cluster-like structure -which can be attributed to the two different crystalline phases present in the sample. This assumption is supported by the absence of such an inhomogeneity on the micrograph performed on a commercial sample of pure anatase TiO 2 nanopowder (MK Impex Corp.) (Fig. 1b). A systematic TEM study of different samples additionally shows that the amount of anatase and rutile phases present in the mixture is not equivalent. The X-ray diffraction (XRD) patterns of all samples (not shown here) evidence a predominant anatase phase in the mixture with large anatase (101) and (200) peaks, along with a weak rutile (110) peak. Using Spurr and Myers formula 21 , the fraction of anatase in the commercial nanopowder was thus evaluated to be 0.8. Some DFT calculations and experimental studies have shown that the water sorption mechanisms are dependent on the crystalline structure 22 , as well as on the orientation of nanocrystals 5,23,24 . However, as our gas-phase experiment results to an averaging of the contributions from all crystal orientations, the difference between anatase and rutile becomes meaningless. The mixture sample of 100 nm was thus chosen to be close to a "realistic" sample commonly used in applications. Even if the anatase-rutile NPs mixture of 25 nm (P25) is the most used in commercial photocatalysis systems, it is poorly focused with our ADLS. Consequently, the choice of the average size of nanopowder sample (100 nm) was mainly guided by our aerodynamic focusing efficiency requirements. Soft x-ray photoelectron spectroscopy was initially performed on dry TiO 2 NPs sprayed out by the nanopowder aerosoliser and then hydrated thanks to the setup described in the Methods section. In order to ensure the dryness of the particles prior to aerosolisation, an annealing process has been performed, similarly to the protocols used in bulk surface science. For annealing, the nanopowder has been kept in a vacuum oven at 150 °C under N 2 atmosphere during 24 h. The temperature has been chosen deliberately low in order to avoid any crystalline phase transition, according to previous observations 25,26 . The effect of annealing on the O 1s XPS spectra can be seen in Supplementary Figure S1 online. The annealed nanopowder was then transferred in the aerosoliser chamber before being sprayed through the ADLS. In these conditions, the exposure of the nanopowder to the ambient moisture is strictly limited; however a brief exposure during the transfer to the aerosoliser chamber remains possible. Figure 2 displays O 1s XPS core-level spectra obtained for annealed nanopowder before hydration (a) and during hydration (b) by water evaporation as described in the Methods section. The incident photon energy used to record the O 1s spectra was 630 eV, leading to an inelastic mean free path of about 0.6 nm in TiO 2 , as determined by using Seah's equation for inorganic compounds 27 . The experimental resolution (originating from the convolution of the monochromator bandwidth and of the electron spectrometer resolution) was about 960 meV, resulting in a total FWHM in the range 1.2-1.5 eV for each component (see Supplementary Table S2 online). It is worth stressing that NPs samples present a non-cleaned, non-oriented surfaces which are all measured at the same time, and each of the spectral lines represents an ensemble of atomic arrangements in the NPs with slightly different chemical environments (leading to slightly different binding energies). Also, on top of the instrumental and lifetime broadenings, linewidths are affected by phonon and final state vibrational broadenings (FSVB). In case of absorbed species the FSVB can lead to broader photolines than their gas phase counterparts due to the formation of new vibrational modes 28,29 . For example several OH-vibrational modes having frequencies up to 0.5 eV have been reported on hydroxylated TiO 2 surfaces 30 . A systematic energy calibration has been performed, by recording the Ar 2p photoemission lines in the gas phase. A Voigt profile was used to fit the data, and the background was assumed to have a Shirley-type shape 31,32 . In order to achieve a reasonable fit of the broad structure originating from O 1s photoemission, four symmetric peaks with their experimental linewidths as fixed parameter were used. For information, the error bars representing the standard deviation have been reported for the individual fitting components. The relative energy positions of the components have also been kept constant (within the error bars), except for the higher binding energy (BE) peak whose position is shifted as discussed below. Hence, after Shirley background subtraction and deconvolution, the spectrum gives rise to four components as shown in Fig. 2. The positions of each component used to fit the spectra can be found in Supplementary Table S3 online. The main peak at 533.8 ± 0.2 eV BE is interpreted as originating from bulk oxygen atoms in the TiO 2 NPs lattice, and the component at 535.6 ± 0.2 eV BE is linked to the adsorption of water on the TiO 2 surface, mainly as hydroxylated chemisorbed species, which is in agreement with several previous bulk studies 14,25,31,33,34 and confirmed by the fact that before hydration, this component is substantially reduced (Fig. 2a) and increases as a function of the water temperature during hydration (Fig. 2b). Let us point out that even without hydration, a small residual component attributed to the position of OH species still remains in the spectrum (a). This can be linked to the spectroscopic signature of the two-fold coordinated O-bridging, which has been already evidenced by Bullock et al. and other groups 32,35 . However, a small hydration due to ambient moisture is not totally excluded to explain the presence of this peak. Two different OH groups have been distinguished in the literature of hydrated TiO 2 surfaces 32,35,36 : OH-groups bond to 5-coordinated Ti 4+ cations forming Ti-OH basic groups, and acidic OH-groups linked to the bridging oxygens, here called O br H. Sham and Lazarus reported already in 1979 a study of hydrated rutile (001) surface, where they observed these chemisorbed acidic and basic groups as well as physisorbed components in the O 1s XPS 36 . Evaluating from the spectra they present, the BE shifts from the bulk O 1s component are approximately + 1.6 eV and + 2.6 eV for O br H and Ti-OH components, respectively. Perron et al. reported a binding energy shift of + 1.3 eV for O-bridging component, and + 2.5 eV for Ti-OH. Unfortunately, we cannot resolve these peaks, and only one broad component To fit the experimental data, two more components have to be added: an intermediate peak at 537.4 ± 0.2 eV BE which results from the adsorption of molecular water in the upper layer, as previously shown by Sham and Lazarus 36 , H. Perron et al. 32 and other groups 37 , and a higher BE component whose position relative to the bulk component seems to fluctuate. This peak can be related to oxygen from organic contaminants at the surface of TiO 2 NPs resulting either from the annealing process 38 or from air exposure. The energy location of the carbon contamination signature is known to depend on the nature of the adsorbate and to vary for different organic contaminants as a function of the chemical partner of the carbon (e.g. oxygen or hydrogen) 39 . This chemical shift can thus be assigned to the different neighboring of the organic adsorbent species before (a) and during (b) hydration. It is important to stress that the energy scale of all spectra is referenced to the vacuum level, due to the gas phase configuration. Consequently, a systematic shift varying between 3.5 eV and 4 eV of the whole spectrum is to be considered to compare with the data from the literature dealing with deposited NPs, which corresponds to the TiO 2 work function. Evaluations of the peak areas relative to the bulk component for different hydration levels are shown in Fig. 3. The relative humidity (RH) measured for each step of water-heating temperature is also reported. The curves reveal that the peak weight at 535.6 eV related to the OH species adsorbed is strongly dependent on the RH, whereas the H 2 O peak seems to stay constant whatever the hydration level is. It has been shown that molecular physisorbed water is easily desorbed under ultrahigh vacuum conditions 32,37 , and hence cannot be directly linked to the hydration level. The higher BE peak is also independent of the state of hydration which supports the assignment to oxygen atoms from organic contaminants, whose signature is also confirmed by the slightly grey color observed on our nanopowder samples after annealing. To validate this interpretation of the spectra, additional XPS measurements have been performed for chemical specification of our sample with a monochromatized Al Kα source on TiO 2 NPs deposited on a substrate, as described in the Methods section. Figure 4 displays the O 1s XPS spectra obtained with the latter setup. Charge effects are compensated thanks to a flood gun correction and a systematic calibration by recording C1s position. Fitting has been performed with the commercial Thermo scientific Avantage software, using a Voigt line shape. The bulk oxygen peak arises at 530.2 eV BE in perfect agreement with the position observed by Hugenschmidt et al. 40 . The BE shift of the overall spectrum with regards to Fig. 2 -corresponding to gas phase configuration -enables to extract a work function equal to 3.6 eV. As was the case for the isolated nanoparticles, a second component arises at 1.6 eV higher BE upon hydration (blue spectrum) which is fully consistent with the peak position attributed to OH chemisorbed species in Fig. 2, observed at + 1.8 eV from the bulk O peak. This peak was verified to be independent from any C-contamination, by recording C 1s XPS spectra before and after hydration. It has to be stressed that due to a lower surface sensitivity in the latter conditions, the NPs were saturated with liquid water in order to get a signal from surface-adsorbed water, resulting in a higher coverage in this case. It might also be noted that no other component is observed at higher energy, contrary to the spectra shown in Fig. 2. This confirms the assignment of the peak at 537.4 eV BE in Fig. 2 to physisorbed molecular water in the upper layer, which tends to be more easily desorbed under the present higher vacuum conditions. It explains why this peak has been evidenced only at low temperature 41 or by tilting the sample 32 . Ti 2p XPS spectra have also been recorded on freestanding TiO 2 NPs, before and after hydration of previously annealed TiO 2 NPs. The Ti 2p core-level spectra (Fig. 5) do not reveal any obvious signature of water adsorption. However, some studies report the presence of a shoulder at the lower binding energy side of the Ti 2p doublet 31,33,42,43 attributed to reduced species (Ti 3+ and Ti 2+ ) in non-stoichiometric defective TiO 2 films. This asymmetry is observed to be quenched after subsequent water exposure 33,43 , resulting in a "healing" of surface defects as shown by Wang et al. 33 . The absence of such an asymmetry in our situation and the high similarity between the two spectra in Fig. 5 reveals that these Ti 3d states are absent or below the detection threshold in our NPs and suggests that they do not take part in the adsorption path. Thus, we can conclude that the Ti-OH are not most likely dominating species on the However, a clear evidence of water adsorption can be seen in the valence spectra. Indeed, Fig. 6a depicts the valence spectra obtained on solvated TiO 2 NPs sprayed out by atomization (blue spectrum) and on an "as-received" nanopowder sprayed out with the commercial nanoaerosoliser (green spectrum). The peak at 15.9 eV BE corresponds to Ar 3p valence states, and has been deliberately kept for calibration purposes. The spectrum corresponding to solvated TiO 2 NPs gives rise to three molecular states of water in the valence band region, labelled as 1b 1 , 3a 1 and 1b 2 -as previously described by M.A. Henderson 44 -which are absent or strongly attenuated in the spectrum of dry NPs. Water peak assignment was based on Kimura's et al. 45 valence states study. The presence of molecular water states in the valence band region (blue spectrum) is assigned to the atomization procedure of NPs in water suspension. However, it is difficult to distinguish the signal of adsorbed surface water from gas phase water also present in the interaction region, because the binding energy shift between them is very small in the valence band region. To better illustrate water adsorption in the valence band region, the spectra have been normalized relative to the top of the valence band (Fig. 6b) and energy calibration was achieved using the Ar 3p photoline. A difference between dry and solvated NPs cases is evidenced: the band bending of the valence edge accompanying the hydrated state and the shift towards higher BE is in full agreement with the observation made by Kurtz et al. 46 on bulk hydrated TiO 2 surfaces. The same behaviour was highlighted with DFT calculations 47 and was attributed to a solvation of TiO 2 surface states due to water adsorption. ## Discussion Considerable effort has been dedicated to unravel the reactivity of water on TiO 2 surfaces. However, the literature suffers from insufficient information regarding the nanoparticles' case. Even on the well-known bulk situation, several factors have been shown to influence the water sorption scheme. Some studies have argued for a coverage-dependency on the adsorption mechanism, with a two-step process in which a low coverage dissociative adsorption is followed by a molecular adsorption at higher coverage 16,43 . Another point of view defends influence of the sample temperature on the adsorption scheme 46 . However, a large consensus seems to be achieved for a molecular adsorption on defect-free surfaces, whereas dissociation occurs only on defective surfaces 24 especially at O-vacancies (O vac ) 8,11,25 . Walle et al. 41 have recently shown that dissociation can also take place on vacancy-free surfaces, resulting in a mixed dissociative and molecular water adsorption qualified as a "pseudodissociated state" at monolayer coverage, where the temperature of the substrate plays a crucial role in the experimental observation of such a mixed state. SR-based XPS of a rutile (110) single crystal was recorded to follow the hydration signature of a bulk solid in ultra-high vacuum conditions for comparison with the isolated NPs experiment using the same photon energy (Fig. 7a). In good agreement with the experiments of Walle et al. 41 these results show that (1) no trace of molecular water can be visible at room temperature (RT) whereas it is clearly evidenced at −193 °C through the structure around 534 eV, (2) a small shoulder appears at 532.2 eV at both temperatures, and seems to be related to OH species chemisorbed at O vac . Contrary to Walle's experiment, the present study is made on a defective surface with O vac -as observed with the STM characterization performed on the rutile (110) monocrystal (not shown here) -supporting the view that in this case, OH species are preferentially adsorbed at O vac sites. For comparison, the data obtained with our isolated nanoparticles (mixture of rutile/anatase) are shown in Fig. 7b on the annealed nanopowder progressively hydrated. As previously discussed, the spectrum corresponding to dry-annealed NPs, as well as the spectra associated to a controlled-hydration, display four components with a clear signature of OH and H 2 O -as supported by the fit -the whole hydration procedure being made at RT. Moreover, still in contradiction with the bulk situation, the hydration is accompanied by a strong increase of the OH component -with a weak effect on the H 2 O peak -giving rise to the second peak at 535.6 eV, which corresponds to the small shoulder observed on the (110) surface (Fig. 7a). In a classical paper by Sham and Lazarus 36 , chemisorbed and physisorbed water on a freshly introduced ambient sample of TiO 2 (001) surface were observed. When the sample was allowed to stay in UHV vacuum conditions for a week, the high energy side of the O 1s spectrum was substantially decreased, indicating that these features corresponded to physisorbed water. A similar effect can be observed in our spectrum of the ( 110) surface (Fig. 7a) evidencing the presence of physisorbed water: when the temperature is decreased, the residual moisture of the experimental chamber condenses on the TiO 2 surface and the physisorbed molecular water peak becomes evident. Additionally, in their study Sham and Lazarus mechanically scraped the surface before exposing it to water, thus creating a lot of surface defects. To record the O 1s XPS they used Mg Kα radiation, resulting to higher kinetic energies and thus longer escape depth of the photoelectrons, and in order to be more sensitive to the surface, the sample was tilted which clearly enhanced the intensity of the hydroxylated peaks. The effect is very similar to what can be seen in the evolving hydration process of NPs (Fig. 7b). However, compared to our results obtained on a rutile (110) monocrystal and to the (001) surface studied by Sham and Lazarus, the NPs data show stronger increase of the OH-species. The first explanation for this difference in intensity of the OH component can be attributed to the coverage, which is higher in the case of isolated TiO 2 NPs than with a well-controlled surface, the hydration process being hardly quantifiable in our gas phase environment. This coverage dependency is moreover confirmed by the presence of a molecular H 2 O peak in Fig. 7b, which can be observed only at low temperature on the experimental data of the (110) surface (Fig. 7a) or on Walle's study 41 . Decreasing the temperature might result in adsorption of residual water on the surface of TiO 2 and prevent from complete desorption of adsorbed species as shown in other studies 24,37,41 . This supports the idea that at higher coverage the adsorption can also occur molecularly on the OH sublayer acting as "anchor sites" for H 2 O molecules, as reported by Yamamoto et al. 43 for the TiO 2 (110) surface. Indeed, the presence of the molecular H 2 O component was also experimentally evidenced by H. Perron and co-workers 32 on a (110) rutile surface at RT where the hydration process was performed during 24 h, resulting in a presumed higher hydration amount than in Fig. 7a, and more comparable with our isolated NPs hydration conditions. Let us point out that the relative coverage is moreover maximized at the nanometer scale, where the surface-to-bulk ratio is larger, resulting in an exaltation of the water-peaks, which become almost comparable to the bulk component as the water layer thickness increases. A clear indication for the occurrence of a dissociative adsorption mechanism is hence visible through our NP spectra via the appearance of a RH-dependent OH component, and this is consistent with the fact that our commercial nanopowder sample might contain O vac at the NPs' surface, especially after the low-temperature annealing process carried out to dehydrate the nanopowder. The presence of induced-surface defects is confirmed by the appearance of a shoulder in the valence gap region for the annealed nanopowder samples (Fig. 8), which was absent from the non-annealed sample. These valence gap states are usually attributed either to O vac or Ti interstitials 37 . However Yim et al. 48 have shown that these states are mainly resulting from O vac rather than Ti interstitials, which is consistent with the fact that no Ti 3+ component was observed in our Ti 2p core-level spectrum. A DFT study has also attributed this peak at + 1 eV above the VB of TiO 2 to a "poorly solvated" configuration of OH species 49 . Based on this XPS study, Fig. 9 summarizes the main adsorption schemes which can occur at the surface of a TiO 2 NPs and evidenced through XPS spectra measurements. The dissociative mechanism (1) at O-vacancies (grey O atoms) have been shown to be dominating, resulting in a highly hydroxylated surface. A proton transfer on a neighbouring O-site can subsequently occur (black O atoms), resulting in a OH weight which is twice the density of defects, as already shown 11,46 . Related to this "chemisorption path" a molecular H 2 O adsorption which could contribute to the peak arising under hydration (Fig. 7b) is not totally excluded (2). Increasing the coverage, the H 2 O physisorption takes place in the upper layers (3), which is enhanced by the presence of underlying OH sites. Another dissociative mechanism (4) via 5-coordinated Ti atoms adds also the amount of O br H species when the proton is transferred from Ti to the neighboring O-bridging atom (schematically represented by black arrows). This pathway is concluded to be a smaller contribution compared to the mechanism (1), based on the fact that the Ti 2p XPS spectra remained unchanged during the hydration process. Also the fitted OH-related peak in the hydrated nanoparticle XPS shows a BE shift closer to the O br H value from literature, but one has to be careful when comparing the importance of these methods based on O 1s XPS since both mechanisms (1) and ( 4) contribute to the O br H species. For the first time, we have experimentally demonstrated the presence of water at the surface of freestanding TiO 2 NPs. To achieve this, we used aerodynamic focusing and SR-based X-ray photoelectron spectroscopy. More importantly, we have shown that it is possible to control the amount of water adsorbed at the NPs' surface and to unravel the adsorption mechanism. A comparison with the bulk case showed that the water signature is exalted in the O 1s XPS spectra through the OH component, attributed to a higher coverage and the higher surface-to-bulk ratio, enhancing the weight of the water components. This result is relevant and of high importance to the photocatalysis research, taking into account that the dissociation of water is favored at the surface of TiO 2 NPs. The high surface sensitivity obtained in our experimental conditions have proved to be crucial to disentangle the role of the surface state into the adsorption mechanism. Finally, consistent with previous studies on bulk TiO 2 surfaces, we have observed a clear evidence for a mixed dissociative and molecular adsorption mechanisms, explained by the high coverage obtained in our experimental hydration conditions for nanomaterials. ## Methods SR based soft X-ray electron spectroscopy measurements. The experiments were carried out at the French national SR laboratory SOLEIL (Saint-Aubin, France) at the ultrahigh resolution soft X-ray PLEIADES beamline (9-1000 eV), where soft X-rays with any kind of polarization can be generated using an Apple II -type permanent magnet HU80 (80 mm period) undulator starting from 60 eV. PLEIADES is dedicated to the spectroscopic studies of isolated species ranging from atoms 50 , ions 51 and molecules to proteins 56,57 , clusters 58 and nanoparticles 20,59 . The photoelectron spectra were recorded with a commercial VG-Scienta R4000 spectrometer based on a hemispherical electron analyzer whose detection axis is perpendicular to the propagation direction of the SR. The pass energy and entrance slit are selected according to the experimental resolution targeted for each measurement and the polarization vector of the linearly polarized SR was chosen to be parallel to the electron detection axis. The basic idea of the experiment is to create a collimated beam of isolated TiO 2 NPs in interaction with SR under high vacuum conditions. Briefly, a flow of nanoparticles is sprayed out in the aerosol phase using a carrier gas (usually Ar or N 2 ) and the resulting solid aerosol is focused thanks to an ADLS and injected through a 2 mm skimmer into the high vacuum chamber of the photoelectron spectrometer setup where the pressure is kept around 2 × 10 −6 mbar during the experiment. The principle of the ADLS was widely described by Zhang et al. 60 or McMurry et al. and its relevance to the study of isolated nanospecies has already been demonstrated at the PLEIADES beamline 20,59,64 . Complementary measurements of single crystal TiO 2 have been performed at the TEMPO beamline at the French national SR facility SOLEIL (A. Naitabdi et al.). TEMPO is dedicated to the spectroscopic studies of condensed matter in the soft X-ray region (50-1500 eV), supplying a good complementarity to the gas-phase measurements performed at PLEIADES. The comparison is discussed in section "Discussion". Hydration of the nanoparticles beam. To study water/TiO 2 surface interaction and to control the hydration level of the TiO 2 NPs, two different experimental methods have been used. The first method was based on the drying of solvated nanoparticles. Commercial TiO 2 NPs (Sigma Aldrich, mixture rutile and anatase, diameter < 100 nm) are initially kept in a distilled water suspension at a concentration of 5 g/L, and droplets are generated by atomization of the liquid suspension using a commercial atomizer setup (Atomizer model 3076, TSI Inc.). The droplets are then dried through a silica gel diffusion dryer and a tube furnace (Vecstar) before being transmitted into the ADLS system. The second method was based on the hydration of a dry particles beam generated with a commercial aerosoliser (Naneum Aerosolizer PA100S). The nanopowder aerosol is generated by concentrating high velocity vibrating jets of Ar under high pressure (6 bars) at the powder surface inside a vortex shaker aerosolisation chamber. Nitrogen gas was flown through a heated bubbler filled with deionized-water to hydrate the flow of NPs before the injection in the ADLS. The moisture level was controlled by varying water temperature and the flow rate of N 2 . Water temperature was measured thanks to a type-K thermocouple and the humidity level was measured in-situ with a commercial humidity sensor (Sensirion, kit EK-H5). ## Classical XPS measurements. A better characterization of our nanopowder has been achieved by additional XPS measurements carried out at the French laboratory "Institut Lavoisier de Versailles", on deposited TiO 2 NPs, using the same sample which served for the SR studies. The spectra were collected using an XPS apparatus (Thermo scientific) with a monochromatized Al Kα source, and the pressure in the analysis chamber was kept around 2 × 10 −8 mbar.

chemsum

{"title": "Water adsorption on TiO2 surfaces probed by soft X-ray spectroscopies: bulk materials vs. isolated nanoparticles", "journal": "Scientific Reports - Nature"}

heterologous_expression_and_characterization_of_functional_mushroom_tyrosinase_(abppo4)

## Abstract: Tyrosinases are an ubiquitous group of copper containing metalloenzymes that hydroxylate and oxidize phenolic molecules. In an application context the term 'tyrosinase' usually refers to 'mushroom tyrosinase' consisting of a mixture of isoenzymes and containing a number of enzymatic side-activities. We describe a protocol for the efficient heterologous production of tyrosinase 4 from Agaricus bisporus in Escherichia coli. Applying this procedure a pure preparation of a single isoform of latent tyrosinase can be achieved at a yield of 140 mg per liter of autoinducing culture medium. This recombinant protein possesses the same fold as the enzyme purified from the natural source as evidenced by single crystal X-ray diffraction. The latent enzyme can be activated by limited proteolysis with proteinase K which cleaves the polypeptide chain after K382, only one The latent enzyme can amino acid before the main in-vivo activation site. Latent tyrosinase can be used as obtained and enzymatic activity may be induced in the reaction mixture by the addition of an ionic detergent (e.g. 2 mM SDS). The proteolytically activated mushroom tyrosinase shows >50% of its maximal activity in the range of pH 5 to 10 and accepts a wide range of substrates including mono-and diphenols, flavonols and chalcones.Tyrosinases form an ubiquitous family of metalloenzymes and are found in all domains of life 1 . They are of central importance for the pigmentation in vertebrates as the reactions catalyzed by tyrosinase provide the starting material for melanin biosynthesis 2 . Tyrosinases catalyze the ortho-hydroxylation of monophenols to o-diphenols (monophenolase or cresolase activity, EC 1.14.18.1) as well as the subsequent two-electron oxidation to the respective o-quinones (diphenolase or catechol oxidase activity, EC 1.10.3.1), which is coupled with the reduction of molecular oxygen to water 3,4 . The active site of tyrosinase is composed of two copper ions which are coordinated by three histidine side chains each 5 forming a type III copper center 6 . Activation of molecular oxygen is affected by binding of dioxygen to the type III copper center in a characteristic 'side-on' bridging mode (µ-η 2 :η 2 ) 7-9 . Hydroxylation and oxidation of one monophenol to the corresponding o-quinone requires one molecule of O 2 , while two o-diphenols can be oxidized to o-quinones per molecule of O 2 consumed (see Fig. 1).The products of the reactions catalyzed by tyrosinase -o-quinones -are highly reactive and consequently commonly unstable in the biological environment they are generated in 10 . Therefore, they do participate in a number of non-enzymatic, spontaneous reactions. The best studied among those reactions are the formation of high-molecular-weight adducts, especially melanin 10-14 , Michael-type nucleophilic 1,4-additions 15,16 and the direct coupling of two quinones ('phenol coupling') 15,17,18 . As tyrosinases can oxidize both small phenolic molecules and phenolic moieties of larger molecules (e.g. proteins) they have found a plethora of biotechnological applications in e.g. organic synthesis, determination of phenolic analytes, bioremediation as well as in medicine, food processing and engineering of (bio)materials 19,20 .Most of these application utilize tyrosinase isolated from fruiting bodies of the common white mushroom Agaricus bisporus ('mushroom tyrosinase') 20 , supposedly mainly due to its ready commercial availability 21 . However, it should be noted that the purification protocols used to prepare the available commercial preparations do usually not yield homogenous tyrosinase and these are therefore likely to contain one or more unspecified 'extras' like laccase, β-glucosidase, β-xylosidase, cellulase, chitinase and xylanase activities 22,23 . The purity of these preparations is further compromised by the fact that Agaricus bisporus possesses genes coding for six different tyrosinases (AbPPO1 -AbPPO6) 24,25 . Of those six genes, at least two (AbPPO3 and AbPPO4) [25][26][27] are expressed in significant amounts in the fruiting bodies which serve as the source material for the commercial preparations of 'mushroom tyrosinase' . Considering these limitations of the widely used tyrosinase preparations in terms of both purity and batch-to-batch variability, an alternative enzyme source that manages to provide a single isoform of tyrosinase in a pure form and at a constant quality would promote both basic (protein-)biochemical research and biotechnological applications of tyrosinase. Herein, we describe such a protocol, yielding pure AbPPO4 in its latent form as well as providing access to the mature, active form of the enzyme. ## Results Sequence of AbPPO4. The initial PCR on the A. bisporus cDNA with primers enframing the ORF for AbPPO4 (AbPPO4_fwd and AbPPO4_rev) yielded two distinct bands at 1.8 kbp (expected size of the gene) and 1.4 kbp (Figure S3). By cloning and sequencing both molecules were identified as the expected gene for AbPPO4. While the larger one (1836 bp) contained the complete gene encoding amino acids M1 to F611, the smaller band (1401 bp) was missing 435 bp corresponding to 145 amino acids. The loss of these bases occurred in such a manner that the reading frame was conserved and amino acids A436 to A580 were deleted from the translation product. This deletion starts in the middle of exon number 6 and spans the last intron as well as more than twothirds of the last exon in the gene for AbPPO4. On the protein level the missing amino acids are all located in the C-terminal domain and contain the CXXC-motif as well as approximately half of the putative trans-membrane helix 27 . Expression of this construct was attempted but for all the conditions tested the heterologous protein was found exclusively in the insoluble fraction and no tyrosinase activity could be detected in the cell lysate. With respect to the published sequence for the AbPPO4-gene 28 the cloned full-length gene contains 23 mutations among which four are non-silent (see Table 1). Of those two are located in the main domain and two at the end of the C-terminal domain. The two variations in the main domain are already known to be compatible with enzymatic activity as they were also encountered in the enzyme purified from the natural source 27 . Those and the two remaining changed amino acids as well as all the changes to the amino acid sequence in the second sequence (with the exception of the deletion of 145 amino acids) are predicted to be non-detrimental to the enzyme's function by SIFT 29 . Expression of the full-length construct (up to F611) confirms this prediction as the heterologous enzyme was found to be fully active. Expression and purification of AbPPO4. The latent form of the tyrosinase (up to T565) 27 was expressed as a fusion protein with an N-terminal tag, namely glutathione-S-transferase from Schistosoma japonicum 30 . Initial expression attempts employing induction by addition of IPTG yielded a big amount of heterologous protein which was however almost exclusively in an insoluble form. Cultivation at lower temperatures and the use of autoinduction medium 31 did produce a very small fraction of fusion protein in soluble form so that enzymatic activity could be detected in the cell lysate after 3 days of reaction. The addition of 500 mM NaCl to the cultivation medium did decrease the specific growth rate of the production strain by 30% causing a marked increase in total cultivation time but did also increase the fraction of soluble protein by several orders of magnitude. This expression protocol as described in Methods yields around 200 mg of fusion protein per liter of medium after capture and purification by affinity chromatography employing the affinity of GST to glutathione immobilized on cross-linked agarose. Removal of the fusion partner by the specific protease HRV 3 C was usually quite efficient with yields between 80% and 100% corresponding to approximately 110 to 140 mg of latent tyrosinase per liter of expression culture. The preparations were still active after 1 year of storage at 4 °C in the used 10 mM HEPES pH ## Sequence Mutations relative to GQ354802.1 # 28 AbPPO4 full length C21T, T168C, T306C, T362C (V121A), T483C, A504C, G536A (S179N), A540C, T717C, T735C, G1089A, C1104T, C1131G, G1218A, C1359T, T1449C, C1458T, A1521G, C1650T, T1686C, T1704C, G1717A (V573I), G1783A (A595T) C21T, G97A (V33I), G133T (A45S), T168C, G301A (V101I), G324C, T362C (V121A), T483C, A504C, G536A (S179N), A540C, G563A (R188K), C618T, C620G (A207G), T171C, T735C, G1089A, C1131G, T1172A & C1173A (L391Q), G1783A (A595T) Table 1. Sequences of the cloned genes. # The sequences are given as mutations relative to GQ354802 (mRNA for the reference sequence for AbPPO4, Uniprot: C7FF05) with non-silent mutations shown in bold and followed by the respective altered amino acid in parentheses. Sequence numbers start with 1 at the A of the start codon and M of the peptide chain for nucleobases and amino acids, respectively. 7.5 buffer. For long-term storage freezing in liquid nitrogen 32 is recommended as the preparations tend to darken after a few months of storage. This darkening, however, does not cause the loss of enzymatic activity. While the full-length construct did behave almost identical as the latent version of AbPPO4 in heterologous expression, the construct encoding only the main domain of the tyrosinase (up to S383) 27 did not exhibit any tyrosinase activity. Activation of latent AbPPO4. Proteolytic activation of the latent tyrosinase was tested with trypsin and proteinase K and both proteases induced tyrosinase activity in the latent enzyme. The activation by proteinase K was more efficient and did yield an essentially pure preparation of active tyrosinase while even after prolonged incubation with trypsin a second species of approximately 52 kDa remained in the reaction mixture (Figure S4). The digestion with an unspecific protease resulted in a low yield of the activation of 25-50% corresponding to 19-34 mg per liter of culture media. The activated protein appeared as a single peak in the size exclusion chromatogram and had a purity of >95% as estimated from SDS-PAGE (Fig. 2). Enzymatic activity could be induced in the latent tyrosinase by treatment with ionic detergents. The cationic detergent cetylpyridinium chloride (CPC) was more effective in activating latent AbPPO4 than the anionic sodium dodecyl sulfate (SDS). Maximal activation was achieved by applying 0.1 mM of CPC and reached a level which was 25% higher than that at the optimal SDS-concentration of 0.6 mM (see Fig. 3). The activity level for activation by CPC remained fairly constant at concentrations higher than 0.1 mM but above 2 mM the latent tyrosinase was prone to precipitation. For SDS the enzymatic activity decreased slightly above 0.6 mM but remained constant at circa 90% of the maximal level for up to 5 mM SDS. Above this concentration the enzymatic activity did slowly decrease but no protein precipitation was observed even at SDS levels above 100 mM. ## Analysis of the purified tyrosinase by intact protein mass spectrometry. ESI-MS of the acidified (causes the loss of copper) 27 but otherwise intact protein (see Fig. 4) yielded a mass of 65096.7 ± 0.30 Da for the latent tyrosinase which matches the calculated mass of 65096.46 Da for the complete sequence (see Fig. 5) from the N-terminal glycine at position −8 to the C-terminal threonine 565 with one thioether bridge (−2.016 Da) and one closed disulfide bridge (−2.016 Da). The presence of a closed disulfide bridge was also reported for the enzyme isolated from the natural source after electrospray ionisation in positive mode 27 but the disulfide bridge was found in the open form in the crystal structure 33 . The only free cysteine residues in the protein are very close to each other in the C-terminal domain of the enzyme (C462 and C465). Since positive electrospray ionisation of free cysteine has been shown to generate mainly protonated cystine (which was attributed to oxidation processes occurring during positive mode electrospray) 34 a similar mechanism may also be observed for the pseudo molecular ions of AbPPO4. The mass spectra of the activated tyrosinase indicated the presence of two protein variants with determined masses of 44181.5 ± 0.51 Da and 44449.6 ± 0.48 Da, respectively. These masses indicate proteolytic cleavage after lysine 382 (calculated: 44449.35 Da) and additionally for most of the tyrosinase molecules the removal of the first three N-terminal amino acids (Gly-Pro-Leu) from the vector-derived region (calculated: 44182.03 Da). Crystal structure of latent AbPPO4. Latent tyrosinase crystallized in the monoclinic space group C 1 2 1 with 4 chains in the asymmetric unit giving rise to a unit cell of a = 287.30 , b = 52.09 , c = 152.66 , α = 90.00°, β = 98.03° and γ = 90.00°. The obtained crystals diffracted to a resolution of 3.25 (for further statistics see Table S2). In contrast to the crystals obtained with the enzyme isolated from the natural source which contained two different chains in the asymmetric unit (one latent and one active protein) 33 and did only form in the presence of sodium hexatungstotellurate(VI) (Na 6 [TeW 6 O 24 ] • 22 H 2 O, TEW) 35,36 , the recombinant enzyme formed crystals containing exclusively the latent tyrosinase. Inspection of the electron density gave no indication for any deviation from the sequence shown in Fig. 5. The recombinant enzyme assumes the same fold as the tyrosinase isolated from the natural source and their active centers as well as the surrounding amino acids are virtually identical (see Fig. 6). 1) are underlined and marked in bold. The amino acid side chains which form hydrogen bonds with the TEW polyoxyanion in the crystal of the enzyme purified from the natural source (HKKE starting at H116) are found in equivalent positions in the crystal of the recombinant enzyme but the lysine side chains are considerably more flexible as indicated by the lack of electron density beyond the C β -atoms. In the crystal containing TEW the interactions mediated by TEW are crucial for the stacking of the monomer chains while in the crystal of recombinant latent AbPPO4 this motif does not play a significant role for the formation of the lattice with the closest interchain contact being longer than 6 . The missing contribution from these strong interactions may be one decisive factor limiting the attainable crystallographic resolution 35,37 which is 0.49 worse than in the crystal structure obtained using TEW. Characterization of the activated AbPPO4. The optimal pH for the enzymatic conversion of L-tyrosine was found to be 6.8 (Figure S5). Enzymatic activity is found starting from pH 4 and extends beyond pH 10.5. In the pH-range from 5 to 10 the enzyme retains more than 50% of its activity at the optimal pH. The activated AbPPO4 catalyzes both reactions observed for tyrosinase (Fig. 1). The catechol oxidase activity proceeds typically with a rate two orders of magnitude faster than the hydroxylation and oxidation of monophenols (Table 2 and Figure S7). Of the tested substrates the enzyme exhibits the highest affinity and the lowest reaction rate for L-tyrosine. The reaction with catechol shows a decrease in rate for catechol concentrations higher than 10 mM which is probably due to the effective suicide-inactivation of tyrosinase by this substrate 38,39 . Evidence for such a loss of activity during catalysis has also been presented for the type III copper enzyme aurone synthase from Coreopsis grandiflora Coreopsis grandiflora acting on sulfuretin (which does contain a catechol moiety) 40 . Besides the substrates chosen for kinetic measurements AbPPO4 was also tested with and does accept tyrosol, chlorogenic acid, p-coumaric acid, 4-tert-butlycatechol, octopamine, 4-methylcatechol, resorcinol, hydroquinone, protocatechuic acid, 3,4-dihydroxyphenylacetic acid, pyrogallol and 3-methoxyphenol as well as the flavonols fisetin, quercetin, its glycoside rutin, the flavanone naringenin and the chalcons isoliquiritigenin and butein (see Figure S6 and Figure S8 for structural formulas of substrates accepted by AbPPO4). Activated AbPPO4 discriminates between enantiomers of tyrosine showing pronounced differences in the rate of the tyrosinase reaction. For tyrosine 1 mM of the L-enantiomer is converted at a rate of 1.22 ± 0.0073 U mg −1 , which is 2.58 ± 0.020 times faster than the rate on D-tyrosine. A slight increase in enantioselectivity is seen for the methyl ester of tyrosine for which the respective value is 14.3 ± 0.11 U mg −1 for L-tyrosine methyl ester representing a ratio of 3.70 ± 0.050 relative to the rate for the D-enantiomer. ## Discussion For bacterial tyrosinases productivities in the gram per liter range have been demonstrated with the application of an optimized fed-batch strategy 41 but for eukaryotic tyrosinases a yield of 4 to 6 mg per liter of culture is already considered large-scale 42 and the same value was also reported for a related plant enzyme 43 . Expression strategies targeting fungal tyrosinases did usually rely on fungal hosts for the expression of soluble and active tyrosinase . Substantial expression yields per liter of culture were reported for the secreted TYR2 from Trichoderma reesei overexpressed homologously (1 g l −1 in batch fermentation) 45 or in Komagataella pastoris (24 mg l −1 ) 46 and for a tyrosinase from Pycnoporus sanguineus produced heterologously in Aspergillus niger (20 mg l −1 ) 44 . As fungi do possess the necessary molecular tools to activate latent tyrosinases all the isolated enzymes were in their active form 47 . Bacterial expression of fungal tyrosinase does provide access to latent tyrosinases but was hampered by insufficient solubility of the expressed proteins which were also prone to enzymatic inactivity 28 . Enzymatically functional protyrosinase from Pholiota microspora was expressed in Escherichia coli 48 but no value for the yield was reported. Recently, protyrosinase from Polyporus arcularius was produced in the same host with a yield of 54 mg latent tyrosinase per liter of culture medium 49 . Here, the latent form of AbPPO4 is produced by E. coli at a yield of 110 to 140 mg per liter of culture. The conversion of latent tyrosinase into the enzymatically active form is coupled to the proteolytic removal of the C-terminal domain which shields the active site of the tyrosinase 47,50 . The causal agents for this key-step in the maturation of fungal tyrosinases are still elusive and for only four fungal tyrosinases the exact location of this crucial event is known. Neurospora crassa TYR is activated in vivo by cleavage after F408 51 , active TYR2 of Trichoderma reesei extends up to G400 45 , TYR1 from Pholiota nameko is cleaved after F387 52 and the C-terminal residue of active AbPPO4 is S383 27 . All those cleavage sites are found 31 or 30 (T. reesei TYR2) amino acids after the tyrosine motif (Y-X-Y/F or Y/F-X-Y) which is found close to the end of the central domain in all tyrosinases 53 . Preceding the cleavage site by 4 amino acids, the YG-motif, which is conserved in fungal tyrosinases 20 , is present. The activated AbPPO4 presented herein was cleaved by proteinase K after K382, which is only one amino acid away from the in vivo activation site S383 27 . This activated tyrosinase should therefore be an excellent model for the native enzyme. Digestion with proteinase K has also been used as a purification method for tyrosinase from mice 54 and apple 55 . This stability against proteinase K digestion of tyrosinases from three different kingdoms of life suggests resistance against proteolysis by serine proteases as a general feature of tyrosinases. Besides proteolytic activation enzymatic activity may also be induced in latent tyrosinases by exposing them to acidic conditions 56 or detergents like SDS 57 . Employing such a system, the enzymatic activity may be kept dormant in a preparation for a prolonged period of time until it is needed at which point it may be induced by the simple addition of a detergent. Latent AbPPO4 was activated by both the cationic detergent CPC and the anionic detergent SDS (see Fig. 3). Activation by CPC did yield a circa 25% higher activity than SDS-activation did and required only one sixth of the detergent concentration. For routine analysis during purification we employed SDS (at a concentration of 2 mM) as CPC did cause precipitation of preparations still containing significant concentrations of foreign proteins and, at concentrations above 2 mM, also AbPPO4 itself. Activated AbPPO4 retains >50% of its activity at the optimal pH 6.8 in the range of pH 5 to pH 10 providing a wide range of possible reaction conditions. For most of the characterized tyrosinases this range is considerably more narrow, e.g. for the heterologously expressed tyrosinase from Polyporus arcularius it is found between pH 5-6 49 and the homologously overexpressed T. reesei TYR2 was found to be almost fully active in the range of pH 6-9.5 45 . The kinetic characterization of activated AbPPO4 shows low specificity and high reaction rates for the tested substrates, especially the diphenols (see Table 2). Kinetic parameters of recombinant fungal tyrosinases on L-tyrosine were reported for AbPPO2 produced in Saccharomyces cerevisiae (K m = 0.302 µM, k cat = 11.39 s −1 ) 58 and for MelB from Aspergillus oryzae produced in E. coli (K m = 43 µM, k cat = 49 s −1 ) 59 , making these enzymes both more specific towards L-tyrosine as well as faster on this substrate than activated AbPPO4. For TYR1 from P. nameko the kinetic parameters on tyrosine could not be determined due to insufficient solubility of the substrate 52 . For L-DOPA a slightly higher number of values (enzyme name: K m in µM | k cat in s −1 ) are reported for the recombinant tyrosinases from A. bisporus (AbPPO2: 1.22 | 141) 58 , P. nameko (TYR1: 1930 | 478) 52 , P. arcularius (Photo-regulated tyrosinase: 1040 | 223) 49 and T. reesei (TYR2: 3000 | 22) 45 . In comparison to those enzymes activated AbPPO4 is less specific for L-DOPA and much faster on this substrate. ## Conclusions In conclusion, a protocol for the production of mushroom tyrosinase was established which is able to produce both latent and active tyrosinase in a pure form. Using this protocol it is possible to provide quantities of mushroom tyrosinase sufficient for providing even larger research projects with a defined tyrosinase preparation that does not suffer from the isoenzyme mixture and the side-activities frequently present in commercial preparations of tyrosinase isolated from mushrooms. ## Methods If not indicated otherwise the chemicals used were purchased from Sigma-Aldrich (Vienna, Austria) or Carl Roth (Karlsruhe, Germany) and were at least of analytical grade. The methyl esters of L-and D-tyrosine were ## Substrate λ max /nm ε max /M −1 cm −1 K m /mM k cat /s ## RNA-extraction and cDNA-synthesis. RNA extraction and cDNA synthesis have been performed according to standard procedures 60 and are described in detail in the SI. Cloning of the AbPPO4 gene. The gene encoding AbPPO4 was cloned out of frame into the expression vector pGEX-6P-1 (GE Healthcare Europe; Freiburg, Germany) and was brought in frame by PCR-based mutagenesis (for details see SI). ## Mutagenesis of AbPPO4. As the enzyme isolated from the natural source did not contain the last 46 amino acids that are encoded by its gene 27 , the cloned gene was adjusted accordingly. Towards that end the base triplets encoding the respective amino acids were deleted from the gene using the Q5 ® Site-Directed Mutagenesis Kit (NEB). Removal of the sequence corresponding to amino acids A566 to F611 was accomplished with the two primers lAbPPO4_fwd and lAbPPO4_rev, while E384 to F611 was removed using lAbPPO4_fwd and aAbPPO4_ rev resulting in the bases encoding just the main domain of the tyrosinase 27,33 . Expression of AbPPO4. AbPPO4 was expressed in E. coli BL21(DE3) in auto-inducing medium, namely ZYM-5052 without the trace element solution 61 but with an additional 500 mM of NaCl. Cultures were grown at 20 °C in shaking flasks in media containing 100 l −1 Na-Ampicillin for 20 h. Then, 0.5 mM copper sulfate was added and expression was continued for 20 more hours (for details see SI). Isolation and purification of the recombinant tyrosinase. The pelleted cells were washed with 10% of the original culture volume of 9 g l −1 NaCl in ddH 2 O, repelleted by centrifugation (8 min @ 3000 × g and 4 °C) and resuspended in lysis buffer (25 mM HEPES, 150 mM NaCl, 5 mM Na 2 MgEDTA set to pH 7.3 with NaOH with the following three components being added immediately before use: 2 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride (PMSF) and 100 mg l −1 hen egg white lysozyme) at 100 g of wet cells per l. Cells were disrupted by 2 passages through a french press 62 at a cell pressure of 850 bar. The samples were cooled on ice in between the passages through the french press and fresh PMSF was added to a final concentration of 2 mM in two portions at the end of each passage. The lysate was cleared by centrifugation (15 min @ 30800 × g and 4 °C), filtered (0.45 µm pore size PES membrane) and applied to a 5 ml GSTrap FF column at a flow rate of 0.5 ml min −1 . The column was kept at 4 °C and 50 mM Tris-HCl, 150 mM NaCl pH 8.0 (@ 4 °C) was used as the mobile phase for elution of unbound material while the same solution with 20 mM reduced glutathione served as the elution buffer. The eluted fractions containing the fusion protein were concentrated by ultrafiltration (Vivaspin ® 20, 30 kDa molecular weight cut-off) which was also applied for buffer exchange. GST was cleaved from the fusion protein by the action of picornain 3C (human rhinovirus serotype 14 protease 3C, HRV 3C) which was applied as a fusion-protein with GST (production protocol in the SI). 1 µg of protease was applied per 150 µg of GST-AbPPO4 fusion protein and the proteolysis reaction was allowed to proceed for at least 18 h at 4 °C in the running buffer of the affinity chromatography supplemented with 1 mM DTT in order to preserve the activity of the cysteine protease HRV 3C. After enzymatic cleavage, the tyrosinase was separated from the fusion partner as well as the protease by a second passage through the affinity column using identical conditions as for the first chromatographic step. The column flow-through containing the protein of interest was concentrated by ultrafiltration (Vivaspin ® 20,30 kDa molecular weight cut-off) and its buffer was exchanged to 10 mM HEPES pH 7.5 (@ 4 °C). This preparation was diluted to a concentration of 20 g l −1 and stored at 4 °C until use. ## Enzymatic activity assay. Tyrosinase activity was routinely assayed on 1 mM L-tyrosine in 50 mM sodium citrate buffer pH 6.8 at 25 °C. For activation of the latent enzyme 2 mM SDS were included in the assay mixture 27 . The monitored species was dopachrome at 475 nm (ε 475 = 3600 M −1 cm −1 ) 63 , volumetric enzymatic activities were calculated from the linear part of the absorption-time curves (after the lag-phase but before the subsequent reactions towards melanin contribute significantly). One unit of enzymatic activity (U) was defined as the amount of enzyme that catalyzes the conversion of 1 µmol of substrate per minute of reaction. Intact protein mass spectrometry. For ESI-MS, which was done at the Mass Spectrometry Centre at the University of Vienna, the buffer of the protein solutions was exchanged to 5 mM ammonium hydrogen carbonate pH 7.8 by repeated ultrafiltration (Vivaspin ® 500, 30 kDa molecular weight cut-off). For introduction into the nanoESI-QTOF mass spectrometer (maXis 4 G UHR-TOF from Bruker, Billerica, MA, USA; providing a mass accuracy better than 5 ppm) by a syringe pump (KDS 100 from KD Scientific, Holliston, MA, USA) @ 3 µl min −1 the protein solutions were diluted to approximately 1 µM in an aqueous solution containing 2% (v/v) acetonitrile and 1‰ (v/v) formic acid. ## Enzyme kinetics. A spectrophotometric assay detecting the appearance of the product in the reaction solution was applied for the determination of the kinetic parameters of AbPPO4. Since the o-quinones generated by the enzymatic action of tyrosinase do not give rise to a stable and soluble product they were trapped by the potent nucleophile 3-methyl-2-benzothiazolinone hydrazone (MBTH). MBTH couples to o-quinones via its amino group generating reasonably stable adducts that remain soluble and are easily detected photometrically 64 . Absorption curves and spectra were recorded on a Shimadzu UV-1800 spectrophotometer applying 1 cm cuvettes which were kept at 25 °C by a Julabo F25 MH thermostat in a circulating water-bath. Kinetic measurements were done in a total volume of 1 ml containing 50 mM sodium citrate buffer pH 6.8, 5 mM MBTH, 2% (v/v) N,N-dimethylformamide and different concentrations of the substrates to be tested as well as AbPPO4 (0.23 to 46 nM). Molar absorption coefficients were determined from rapid oxidation of small concentrations of substrate under standard assay conditions applying tyrosinase concentrations in the µM range. K m and k cat , the two parameters of the Michaelis-Menten model 65 , were calculated from the steady-state rate of product formation for the different substrate concentrations tested. Measurements were performed in triplicate and the reciprocals of the variances of the observed slopes were used as weights for the nonlinear regression applying the Levenberg-Marquardt algorithm 66 as implemented in the program Dataplot (version 11/2010). Initial estimates for the two free parameters were generated by applying the Hanes-Woolf linearization of the Michaelis-Menten equation 67 . Proteolytic activation of AbPPO4. Latent AbPPO4 was converted into its active form by treatment with proteinase K. The protease was used at a ratio of 1:10 (equivalent to 45 µg of proteinase K per mg of latent AbPPO4) in a reaction buffer containing 50 mM Tris-HCl pH 8 @ 25 °C, 100 mM sodium ascorbate and 20 g l −1 of latent AbPPO4 for a total reaction time of 90 min. The reaction was stopped by addition of 2 mM PMSF after which the solution was concentrated to less than 70 µl by ultrafiltration (Vivaspin ® 500, 30 kDa molecular weight cut-off) and applied onto a size exclusion column (Superdex 200 Increase from GE Healthcare) equilibrated with 50 mM sodium citrate pH 6.8 and run with the same buffer at 4 °C and with a flow rate of 0.5 ml min −1 . The eluted fractions possessing tyrosinase activity were pooled and concentrated by ultrafiltration (Vivaspin ® 500, 30 kDa molecular weight cut-off). Protein crystallization, X-ray diffraction and model building. Conditions for the growth of AbPPO4 crystals were refined based on an initial hit obtained with sodium cacodylate and PEG 4000 in a hanging drop vapour diffusion setup. Single crystals suitable for crystallography grew over the course of a few days in hanging drops initially made up of 1 µl of a 10 g l −1 solution of latent AbPPO4 in 10 mM HEPES pH 7.5 mixed with 1 µl of reservoir solution containing 50 mM sodium cacodylate pH 5.8 and 13% (w/w) PEG 4000 which was equilibrated via vapour diffusion with 1 ml of reservoir solution at 293 K. Crystals were harvested using Kapton ® loops (Hampton Research, Aliso Viejo, CA, USA), soaked with cryo-protectant (50 mM sodium cacodylate pH 5.8 and 40% (w/w) PEG 4000) and plunged into liquid nitrogen where they remained until the diffraction experiment. Data were collected at beamline ID-23 at the European Synchrotron Radiation Facility (ESRF, Grenoble, France) at 100 K applying a wavelength of 0.873 (14.2 keV) and a PILATUS2 3 M detector. Data reduction (for details see SI) was carried out using XDS 68 . Final model quality was evaluated by the MolProbity server 69 and the model has been deposited in the PDB under entry number 5M6B.

chemsum

{"title": "Heterologous expression and characterization of functional mushroom tyrosinase (AbPPO4)", "journal": "Scientific Reports - Nature"}

near_uv-visible_electronic_absorption_originating_from_charged_amino_acids_in_a_monomeric_protein

## Abstract: Electronic absorption spectra of proteins are primarily characterized over the ultraviolet region (185-320 nm) of the electromagnetic spectrum. While recent studies on peptide aggregates have revealed absorption beyond 350 nm, monomeric proteins lacking aromatic amino acids, disulphide bonds, and active site prosthetic groups are expected to remain optically silent beyond 250 nm. Here, in a joint theoretical and experimental investigation, we report the distinctive UV-Vis absorption spectrum between 250 nm [3 ¼ 7338 M À1 cm À1 ] and 800 nm [3 ¼ 501 M À1 cm À1 ] in a synthetic 67 residue protein (a 3 C), in monomeric form, devoid of aromatic amino acids. Systematic control studies with high concentration non-aromatic amino acid solutions revealed significant absorption beyond 250 nm for charged amino acids which constitute over 50% of the sequence composition in a 3 C. Classical atomistic molecular dynamics (MD) simulations of a 3 C reveal dynamic interactions between multiple charged sidechains of Lys and Glu residues present in a 3 C. Time-dependent density functional theory calculations on charged amino acid residues sampled from the MD trajectories of a 3 C reveal that the distinctive absorption features of a 3 C may arise from two different types of charge transfer (CT) transitions involving spatially proximal Lys/Glu amino acids. Specifically, we show that the charged amino (NH 3 + )/carboxylate (COO À ) groups of Lys/Glu sidechains act as electronic charge acceptors/donors for photoinduced electron transfer either from/to the polypeptide backbone or to each other. Further, the sensitivity of the CT spectra to close/far/intermediate range of encounters between sidechains of Lys/Glu owing to the three dimensional protein fold can create the long tail in the a 3 C absorption profile between 300 and 800 nm. Finally, we experimentally demonstrate the sensitivity of a 3 C absorption spectrum to temperature and pH-induced changes in protein structure. Taken together, our investigation significantly expands the pool of spectroscopically active biomolecular chromophores and adds an optical 250-800 nm spectral window, which we term ProCharTS (Protein Charge Transfer Spectra), for label free probes of biomolecular structure and dynamics. ## Introduction The characteristic electronic absorption profles of proteins/ amino acids in aqueous media show broad features in the UV region (185-320 nm) of the electromagnetic spectrum. The most distinctive absorption features of the spectra, typically seen around 255-280 nm, are attributed to chromophores present in the sidechains of aromatic amino acids like Trp around 280 nm (3 $ 5600 M 1 cm 1 ), Tyr around 275 nm (3 $ 1420 M 1 cm 1 ) and Phe around 257 nm (3 $ 197 M 1 cm 1 ). Weak contribution from disulphide bonds between 250 nm (3 $ 360 M 1 cm 1 ) to 320 nm (3 $ 6 M 1 cm 1 ) has also been reported. The peptide bond in proteins has a strong absorption around 190 nm (3 $ 7000 M 1 cm 1 ) and a weak absorption between 210 and 220 nm (3 $ 100 M 1 cm 1 ). Further, chromophores present in prosthetic groups and metal-ligand centres localized at enzyme active sites absorb in the visible (beyond 400 nm). 2 Accordingly, proteins devoid of aromatic amino acids, disulphide bonds, and active-site chromophores are expected to remain optically silent at wavelengths beyond 250 nm. 8,9 Several years ago, our group had reported signifcant absorption in the 250-350 nm region for poly-L-Lys solutions and in Lys-rich proteins such as human serum albumin (HSA) after accounting for the absorption from dominant chromophores like Trp. 10 Interestingly, the UV absorption band beyond 320 nm is absent in many proteins, particularly those that have low Lys content. 10 Further, novel absorption spectra (l max $ 270 nm) were also seen in concentrated (0.5-1.0 M) pure aqueous solutions of L-Lys monohydrochloride (Lys$HCl), 11 and corroborated by other researchers later. 12 The results above are intriguing because Lys has an aliphatic sidechain ending with a primary amine that cannot possibly absorb in the near-UV region in its monomeric form. Recently, several investigators have also reported unusual UV-Vis absorption/visible fluorescence beyond 350 nm from protein powders, high concentration protein solutions, and peptide aggregates lacking aromatic amino acids. The suggested mechanisms for the unusual absorption and fluorescence include backbone H-bonding and proton transfer. Our previous studies suggest that proteins rich in charged amino acids may also absorb between 250 and 350 nm even if they lack aromatic amino acids. 10 Establishing a quantitative link between nonaromatic protein amino acid sequence content and the UV-Vis absorption features above 320 nm would open up a new spectral window to probe prominent proteins of biomedical relevance. In particular, charged amino acids Lys and Arg are integral constituents of DNA and RNA binding proteins such as histones, spliceosomal proteins and transcription factors. His is an important constituent of glycoproteins. 22 Glu rich proteins play important roles in estrogen receptor binding. 23,24 Finally, intrinsically disordered proteins which play a key role in several regulatory events inside the human cell are rich in charged amino acids. 25 These considerations motivate a systematic investigation of protein absorption spectra beyond 320 nm and For monomer models we extracted a dimer containing Lys/Glu/Gly, mutated the other amino acid to Gly and capped the amino and carbonyl ends with hydrogens. The same procedure was followed for each monomer in Distally Separated (DS) in sequence pairs. Nearest Neighbour (NN) in sequence residue pair dimers were directly capped with hydrogens. its dependence on charged amino acid content. Given the strong prominent spectral features of Trp, Tyr and Phe, it is desirable to investigate the UV-Vis absorption spectra within a model protein devoid of these aromatic amino acids. In this study, we carried out systematic experimental and theoretical investigations on the UV-Vis absorption spectrum (250-800 nm) of a small (67 residue), monomeric, synthetic protein devoid of aromatic amino acids (a 3 C). The structure of a 3 C has been determined previously by NMR experiments 26 to be a three helix bundle (Fig. 1a). The protein is rich in charged amino acids (54% of the sequence) which comprise of 17 Lys, 17 Glu, and 2 Arg residues (Fig. 1a). Despite lacking conventional aromatic chromophores, a 3 C exhibited moderate absorption features in the 250-320 nm region (3 ¼ 7338 AE 191 M 1 cm 1 at 250 nm) and a distinctive long tail spanning the entire visible spectrum up to 800 nm (3 ¼ 964 AE 129 and 501 AE 66 M 1 cm 1 at 450 and 800 nm, respectively). We carried out control absorption spectra measurements in high concentration aqueous solutions of all non-aromatic amino acids including those present in a 3 C and confrmed that charged amino acids possess unique characteristic absorption features extending beyond 320 nm. Time-dependent density functional theory (TDDFT) electronic structure calculations on amino acids of the a 3 C protein sampled from classical molecular dynamics (MD) trajectories revealed that Lys and Glu amino acids may produce broad absorption spectral profles. Analysis of the computed spectra showed that Lys and Glu amino acids possess charge transfer (CT) transitions involving the amino (NH 3 + )/carboxylate (COO ) groups of their sidechains and the polypeptide backbone. Our MD simulations further highlighted the spatially proximal (4-6 ) interactions between charged amino/carboxylate groups of Lys/Glu sidechains facilitated by the three dimensional (3D) protein fold. We show that such spatial interactions between charged residues can modulate the spectral transitions above 300 nm and create a long tail in the a 3 C absorption spectra. Finally, we attempt to connect the experimentally observed changes in the a 3 C absorption spectra triggered by changes in solvent pH and temperature with alterations in the separation between charged amino (Lys) and carboxylate groups (Glu) of a 3 C in the 3D space. We term the new 250-800 nm absorption spectra from CT transitions in charged amino acids as ProCharTS (Protein Charge Transfer Spectra). ProCharTS provides new label free spectral markers to track the structure and dynamics of proteins rich in charged amino acids which can complement traditional techniques based on aromatic chromophores. In the manuscript, we provide evidence for ProCharTS spectral signatures in proteins containing aromatic amino acids. Thus, the method is generally applicable to study natural proteins or protein domains which are rich in charged amino acids irrespective of their aromatic amino acid content. ## Search and selection of a 3 C for experimental and computational studies We scanned all available PDB codes from RCSB Protein Data Bank (http://www.rcsb.org) and corresponding FASTA sequence to examine their charged amino acid content. Based on our exhaustive search, 2-mercaptoethanol-a 3 C protein (PDB ID: 2LXY), was selected for further studies based on high content and close proximity of Lys residues. The a 3 C protein (Fig. 1a) contains 17 Lys residues. Out of these, 14 Lys pairs are within 10 distance. More details on this search and selection procedure is presented in the ESI (Section S1.1). † ## Experimental methods All the amino acids and the control protein samples of the highest purity available were purchased from Sigma Aldrich Chemicals Pvt. Limited, Bengaluru, India. Control proteins were Human Serum Albumin (HSA; cat #A1887) and Hen Eggwhite Lysozyme (HEWL; cat #L6876). Other chemicals and reagents of high purity analytical grade were procured from Merck India Limited. Expression and purifcation of a 3 C. The recombinant a 3 C was over expressed in E. coli and purifed as per the methodology described elsewhere. 26 The purifcation of protein was monitored (see Section S1.2 of ESI †) at each step by SDS-PAGE (Fig. S1a of ESI †). Purity was further confrmed by reverse phase HPLC (Fig. S1b of ESI †) and the mass ascertained by Electrospray Mass Spectrometry to be 7462.883 Da (Fig. S1c of ESI †). The lyophilized sample of the protein was used for further experiments. Solid phase peptide synthesis of peptides containing Lys. Peptides with varying distance between the Lys residues (NH 2 -Gly-Lys-Lys-Gly-CONH 2 , NH 2 -Gly-Lys-Ala-Lys-Gly-CONH 2 , NH 2 -Gly-Lys-Ala-Ala-Lys-Gly-CONH 2 ) were synthesized by standard Fmoc/tertiary-butyl orthogonal protection strategy using solid phase peptide synthesis. The syntheses were performed manually on a Stuart blood tube rotator. Peptides were synthesized such that each peptide had two Lys residues with variable separation in sequence. The steps involved in the synthesis and characterization are described in detail in the Sections S1.3 and S1.4 of ESI. † Unless stated specifcally, all reactions were carried out at room temperature. UV-Visible absorption spectra. The absorption spectra for all non-aromatic amino acids, Lys containing peptides, and poly-L-Lys were recorded at room temperature (25 C) on a double beam Lambda-25 UV-Vis Spectrophotometer (Perkin Elmer, USA) using a UV quartz cell of 10 mm path length. Flatness of baseline was ascertained before and after all measurements by running the blank solution in sample and reference cuvette chambers. Spectra were acquired with multiple scans (3-5) between 250 and 800 nm (fxed 1 nm bandwidth) and averaged subsequently. For recording temperature dependent spectra at 25 C and 85 C, Varian Cary-100 double beam spectrophotometer equipped with a Peltier-based sample temperature controller was used. The sample was thermally equilibrated at high temperature for at least 30 minutes prior to recording absorption spectra. The Lys containing peptides and Poly-L-Lys$HCl samples were dissolved in deionized water. The amino acids, viz. Ala, Arg, Glu$Na, Asp$K, Gly, Lys, Lys$HCl, Pro, and Ser were dissolved in deionized water, while Asn, Cys, His, Ile, Leu, Thr and Val were dissolved in 0.1 N HCl as they were insoluble in pure water. For all the scans, 1 M concentrations were employed unless otherwise stated. Control studies of pH dependence of Lys solutions are provided in the ESI (Fig. S4). † The a 3 C protein was dissolved in deionized water and the absorption spectra (200-800 nm) were recorded for different concentrations (5-105 mM) of the protein. Pure deionized water was kept as blank control for the measurements. Protein concentrations were calculated using the Lowry method and confrmed by measuring the difference in far UV absorbance (A 215 A 225 ). 27,28 We carried out pH dependent studies on a 3 C (85 mM) dissolved in pure deionized water by gradual addition of either 0.1 N NaOH or 0.1 N HCl to the protein solution. Absorption spectra from 250 to 800 nm were recorded as stated earlier. Circular dichroism measurements. CD measurements were carried out on a 3 C at 25 C and 85 C on a spectropolarimeter (Make: JASCO, Model: J-1500, JASCO Inc., Maryland, USA). The scans were recorded from 300 to 190 nm with data pitch of 0.1 nm, bandwidth of 2 nm; thinning scale was kept at 9 and the dynode voltage never exceeded 0.6 kV. Three scans were recorded for each sample and deionized water served as blank in all the cases. Quartz Cuvette (Make: JASCO) with 1 mm path length with transmission range up to 190 nm was used for recording all the measurements. ## Computational methods Molecular dynamics (MD) simulations of a 3 C. We carried out MD simulations on fully solvated atomistic models of the a 3 C protein using the NAMD program 29 (version 2.9) and the CHARMM27 force feld. 30 The initial structure used in the simulations was an NMR derived structure (PDB code: 2LXY) captured with mercaptophenol ligated at the C32 site. The ligand was removed during processing to carry out simulations of mercaptophenol free a 3 C. The Protein Data Bank (PDB) structures had 31 frames and we chose frame 15 (this frame had the maximum number of Lys residues within 10 of each other) as the reference structure for simulations. First hydrogens were added to the structure using the psfgen utility in the VMD program 31 and the protein was solvated (TIP3P water model) inside a rectangular water box of dimensions $67 56 60 3 . The volume for a single protein molecule mimics a protein solution of $7.4 mM concentration. This is worked out as follows: a water box of volume 67 56 60 3 ¼ 225 120 10 30 m 3 solution contains 1 protein molecule. This implies that 1000 litre (1 m 3 ) of water contains 1/(225 120 10 30 ) molecules. Thus, 1 litre of water contains 1/(225 120 10 30 1000 6.022 10 23 ) mol ¼ 7.38 10 3 mol of protein. The system was neutralized by adding 2 Cl ions. Following standard equilibration protocols (see Section S1.5 of ESI †) a 110 ns MD NPT production run was carried out generating snapshots at interval of 2 ps. The a 3 C protein structure was found to be stable in the 3-helix bundle form along the trajectory. Data from the last 100 ns of the MD production run was used for analysis shown in the manuscript. Electronic structure calculations. We computed the absorption spectra of amino acid fragments (see below) extracted from 100 a 3 C snapshot structures sampled from the last 100 ns of the MD production run (see procedure below and Fig. 1b). For 100 conformations of each amino acid fragment listed below, absorption spectra were calculated using TDDFT with a CAM-B3LYP 32 functional and the 6-31++G(d) basis set on all atoms in the Gaussian 09 program. 33 In all TDDFT calculations, the frst 100 to 200 lowest energy electronic transitions were calculated. Since some systems exhibited transitions deep in the visible range of wavelengths, more than 100 transitions were required to cover the absorption spectral range. Difference electron density plots were calculated using the Multiwfn 3.3.8 software 34 and visualised (isovalue set to 0.0004) in GaussView 5.0. 35 Calculations of monomer, dimer, and peptide spectra. The amino acid fragments included monomers (Gly, Lys, and Glu), dimers (Lys-Lys, Glu-Glu, Lys-Glu, Lys-Ala, Lys-Ile, Lys-Leu, Lys-Val, Lys-Cys), pairs of dimers (Lys:Gly-Lys:Gly, Glu:Gly-Glu:Gly, Lys:Gly-Glu:Gly), and tetramers (Gly-Gly-Gly-Gly). The dimers and dimer pairs represent models for interactions among nearest neighbour (NN) and distally separated (DS) amino acid pairs in sequence (data in Fig. 6 and S13 of ESI †). For each amino acid fragment extracted from the trajectory, dangling bonds were capped using the psfgen module in VMD with modifed C terminus (CHO group) and N terminus (NH 2 group). NN interacting pair dimer fragments (data in Fig. S13 of ESI †) and Gly tetramers (data in Fig. S5 and S6 of ESI †) were capped directly. However, for monomers (data in Fig. 4) and DS pairs (data in Fig. 6), we capped the charged amino acid backbones with Gly to better represent the polypeptide backbone in the protein environment. The procedure involved extraction of Lys/Glu and its adjacent residue and then mutating the adjacent residue in the fragment to Gly. For DS pairs the capping procedure described above was applied to both charged amino acids of the pair (Fig. 1). NN and DS pairs of charged amino acid residues were chosen based on the distance of amino nitrogen (N A ) and carboxylate carbon (C C ) atoms of their sidechain. Three cases of (1) strong, (2) intermediate, and (3) weak interactions were considered based on distance ranges extracted from the RDF plots for Lys N A and Glu C C atoms (data in Fig. 5b). Spectra calculations modelling the effects of the environment. For selected amino acid fragments described above we recomputed the TDDFT spectra after including explicit water molecules and other charged chemical species in the vicinity of the fragment representative of the polar environment of the protein surface. The following procedure was used to construct these models: for Lys, Glu monomers and DS Lys-Lys, Glu-Glu and Lys-Glu dimer pairs, 10 MD snapshots (from within the 100 used for vacuum calculations) were chosen. We included explicit waters and/or Glu carboxylate groups within 3-6 from either the Lys N A or Glu C C atoms (monomers) or the geometric centre of interacting N A and C C atoms (dimers). For the case of Lys-Lys, and Lys-Glu dimer pairs we also examined the effect of water position (the electronic coupling effect of the water to the dimer pair) on the spectra by manually placing the water molecule at different distances from the charged amino/ carboxylate groups. In order to examine the effect of changes in protonation states of interacting dimers, we selected single representative MD snapshots of DS Lys-Lys and Lys-Glu dimer pairs and recomputed the TDDFT spectra after either deleting H atoms from the NH 3 + groups or adding H to the COO groups. We covered all 3 possible deprotonation sites for the amino group N A atoms and the 2 possible protonation sites for the carboxylate group O atoms. Characterization of transitions. Two measures were used to characterize the transitions as charge transfer (CT) transitions or non-CT transitions. The frst measure is the average holeelectron separation distance, Dr: 34,36 Dr ¼ where F is the molecular orbital (MO) and the index i and j go over all occupied and vacant MOs respectively. Here K i j ¼ P i j + Q i j where P i j and Q i j are excitation (i / j) and de-excitation (j ) i) confguration coefficients. The second measure is the distance between the centroid of the hole and electron distribution (D CT ), defned as: 34,37 with where, the index a represents Cartesian components (X, Y, Z) and r electron/hole is the electron/hole density distribution. The two measures were calculated with Multiwfn version 3.3.8 34 and then the following conditions were used to categorize the transitions: (i) CT transition when Dr > 2 and D CT > 1 , (ii) non-CT transition, when Dr ¼ <2 or if Dr > 2 and D CT < 1 . Depending on the overlap integral between the hole and electron distribution, non-CT transitions can be further classifed into: (1) local excitations: large overlap integral values, and (2) Rydberg transitions: small overlap integral values. In our analysis (data in Fig. 4 and S15 of ESI †), we simply classify the transitions as CT and non-CT transitions. Further, transitions for which the Dr was within 5% of the threshold (2 ) value were classifed as borderline transitions. Critical discussion of modelling methods and assumptions. In our computational protocol we sampled chromophore (amino acid monomers and interacting dimers) conformations from fully solvated and atomistic MD simulation trajectories of the a 3 C protein. Thus chromophore confgurations used in the TDDFT calculations are completely compatible with the solution phase measurements. The objective of our computational studies in this manuscript is to examine the nature of transitions in charged amino acids and the modulation of their spectral range with conformational fluctuations and side-chain associations. To this end, we carried out electronic structure calculations on more than 2500 conformations (each panel of spectral profles in Fig. 4, 6, and associated ESI † shows data from 100 or more structures) of relevant amino acid chromophores sampled from the MD simulations. The two central conclusions from our computational studies relate to physical properties of the chromophores: (1) charged amino acids possess intrinsic CT transitions due to the directional electric feld created by the excess charge on their sidechains, and (2) interactions between the charged sidechains alter the nature of charge donor-acceptor states to modulate the absorption spectral range of such CT transitions. These conclusions are further supported by our control experimental data comparing the spectra of charged vs. non-charged amino acids (data in Fig. 3). We critically examined capping strategies for amino acids in our electronic structure calculations using control calculations on extended peptide backbones and alternative capping models using methyl groups 38 (Fig. S5-S7 of ESI †). Based on these calculations, we concluded that extending the polypeptide backbone by adding an extra Gly unit to the C-terminal end of amino acid fragments provided robust converged spectra above 250 nm. Our control calculations show that further extending the backbone by adding Gly units or using methyl capping groups (Fig. S5 and S6 of ESI †) only alters backbone transitions around 200 nm. However, backbone transitions and signature CT transitions involving charged amino acid sidechains/ backbone near and above 250 nm are not altered by either extending the backbone or changing the capping groups. We do fnd a small number of fctitious non-CT transitions above 300 nm localized on our capping groups (Fig. 4 and S7 of ESI †). However, we were able to cleanly identify and separate these non-CT transitions above 300 nm from our characteristic CT transitions of interest (see Fig. 4 and S15 †). We have chosen the TDDFT method to compute the UV-Vis spectral profle for the amino acid chromophores. The method scales well with system size and has been shown to provide reasonable results for simulating UV-Vis spectra of organic chromophores and amino acids. 17, We employed a range corrected exchange-correlation (XC) functional (CAM-B3LYP) which provides a reasonable description of charge transfer excitations and UV-Vis spectra in dipeptides and tripeptides. 32,42 Environmental/solvation effects are typically included in spectra calculations through continuum dielectric or quantum mechanics (QM)/molecular mechanics (MM) models. 40, These models are best suited for the description of isolated (spectrally distinct) chromophores (e.g. dyes or aromatic amino acids) embedded in a solvent and/or protein medium. Even in such situations, it is advisable to use continuum solvation models coupled with an explicit QM description of solvent molecules interacting with the solute. 40,46 Depending on the nature of the transitions, hundreds of solvent molecules may be required to converge spectral trends. 44,46 A reasonable criterion for choosing the size of the QM region is to include all residues/ solvent molecules whose charge distribution changes signifcantly during the chemical reaction of interest. 47 In other words, the MM region should only include those molecules whose charge distribution is fxed during the reaction. Our system shows highly anisotropic solvation with concentrated charged moieties dynamically interacting on the surface of a protein along with bound waters. Here, during photoexcitation, all charged amino acids (and maybe even bound waters) will show signifcant changes in their charge distribution (since the reaction involves CT transitions). Thus, the choice of the QM/ MM boundary is non-trivial in our case. To examine environmental effects, a rational way forward is to systematically increase the size of the QM region accompanied by sampling confgurations with different charge states. Thus, in our TDDFT spectra calculations, we focus on developing explicit QM models to describe environmental effects. Specifcally, we considered the effect of the polar protein surface on the spectral features by examining different charged sidechain states and by including explicit water molecules and charged sidechains in the vicinity of the chromophores. ## Results and discussion UV-Vis spectra of a 3 C reveals signifcant absorption spanning 250-800 nm We investigated the UV-Vis spectra between 250 and 800 nm for different solution concentrations of a 3 C ranging from 5 to 105 mM. The molar extinction coefficient (Fig. 2a) reveals moderate absorption (3 ¼ 7338 AE 191 M 1 cm 1 at 250 nm) features in the 250-300 nm region which decay gradually with a distinctive long tail that extends into the visible region (3 ¼ 964 AE 129 and 501 AE 66 M 1 cm 1 at 450 and 800 nm, respectively). The observed spectral features (the tail region) are clearly not due to scattering, as demonstrated by the poor overlap of the observed spectra with a simulated Rayleigh scattering profle which follows a (1/l 4 ) dependence (Fig. 2b). Further, the absorbance at different wavelengths varies linearly with concentration (Fig. 2ainset), arguing against any contribution arising from protein intermolecular interactions to the spectra. Indeed, the monomer proteins are likely to be farther than 20 nm from each other, on average, at 105 mM concentration. While absorption above 320 nm was seen previously in proteins rich in charged amino acids such as HSA (3 ¼ 1546 M 1 cm 1 at 325 nm), 10 the spectra below 320 nm was masked by strong contributions from Trp and Tyr residues. In this regard, a 3 C clearly stands out as it is completely devoid of aromatic amino acids and rich in charged amino acids. Thus, the spectral features of a 3 C, even in the 250-300 nm range, are novel as they do not arise from aromatic chromophores. Further, the report of absorption beyond 350 nm for a monomeric protein lacking aromatic amino acids or active site chromophores is unprecedented. The ProCharTS spectra, demonstrated above for a 3 C, is generally applicable for the study of natural proteins rich in charged amino acids irrespective of their aromatic amino acid content. In Table 1 we present new data for two proteins HSA and HEWL for which absorption spectra were previously reported up to 350 nm. 10 Table 1 shows that both a 3 C and HSA (both rich in charged amino acids) display absorption features which match well above 300 nm and extend up to 800 nm. In contrast, HEWL which has a lower percentage of charged residues in its sequence relative to HSA or a 3 C, lacks signifcant absorption features beyond 320 nm (Table 1). While HSA contains Tyr, Trp, and Phe residues, the absorption from these aromatic amino acids is expected to sharply drop beyond 320 nm. In contrast, the ProCharTS spectrum extends up to 800 nm providing signatures well resolved from that for Tyr and Trp in HSA. Thus it is evident that even in presence of aromatic amino acids, contribution from ProCharTS persists and remains conspicuous between 320 and 800 nm. Further, the presence of ProCharTS has implications even for the spectral ranges overlapping with that from aromatic amino acids. The signifcant absorption from charged amino acids around 280 nm (a 3 C shows 3 ¼ 4531 AE 133 M 1 cm 1 at 280 nm) should be taken into account when interpreting aromatic amino acid absorbance at 280 nm to quantify protein solutions (e.g. Near UV absorbance at 280 nm to estimate protein concentrations 27 ). Further, the ProCharTS profle broadly overlaps with the emission profle of fluorescent chromophores (such as Trp) or dyes. Thus, in addition to monitoring the absorption profle changes directly, the decay kinetics of fluorescent probes may also be used as a spectral marker to follow the dynamics and interactions of charged amino acids within protein folds. To examine the role of the protein fold in producing the observed spectral features of a 3 C (Fig. 2a), we studied the absorption spectra of a mixture of a 3 C amino acids at proportions (see Table S2 of ESI † for amino acid concentrations) present in a 105 mM protein solution. These samples do not show any signifcant absorption in the 250-800 nm region in stark contrast to a 3 C polypeptide chain linking together the same amino acids (Fig. 2b). This implies that the protein fold may play a crucial role in the origin of the observed novel UV-Vis spectral features. In the next section we carry out further experimental studies to highlight the role of charged amino acids and the protein fold in producing the absorption spectrum of a 3 C in Fig. 2a. ## Charged amino acid and peptide solutions show signicant absorption above 350 nm To examine the sequence specifcity of the UV-Vis absorption from a 3 C, we studied the absorption spectra of high concentration solutions of all non-aromatic amino acids including those present in the a 3 C sequence (Fig. 3). We fnd signifcant absorption between 250 and 400 nm for charged amino acid solutions of Lys, Glu monosodium salt (Glu$Na), Arg, Asp potassium salt (Asp$K) and His (Fig. 3a). In contrast, uncharged amino acid solutions of Ala, Asn, Ile, Leu, Met, Pro, Ser, Thr, and Val show negligible absorption in this range (Fig. 3b). Note that Lys$HCl has a molar absorptivity $6 times smaller than that for pure Lys solutions lacking the hydrochloride ion. The decrease in absorption due to the presence of ions supports participation of the charged amino acid sidechains in the photoinduced electronic transitions. For instance, the hydrochloride ion may screen the sidechain charge to reduce the net absorption of the sample. A similar reasoning suggests that pure Glu (insoluble in aqueous medium) should have a higher molar absorptivity than that measured for its monosodium salt solution. Since Lys$HCl and Glu$Na have very similar absorption intensities (0.23 and 0.20 respectively, at 270 nm), the molar absorptivity of pure Glu may match that of pure Lys. Previously, proton-transfer as well as the hydrogen bonding had been suggested by Pinotsi et al. 17 as possible mechanisms for fluorescence of amyloid aggregates containing b-sheet architectures and lacking aromatic amino acids. However, we found no dependence of the absorption features (3) of high concentration Lys solutions on the pH of the medium over a broad pH range of 2-12.5 covering the Lys amino group pK a (Fig. S4 ## of ESI †). The insensitivity of the absorption features on a Measured in deionized water. b Too low to be measured accurately. pH argues against the participation of protons in the initial photoexcitation process of high concentration charged amino acid solutions. While Pinotsi et al. do not provide data on the effect of pH on the excitation spectra, 17 our observations do not rule out the possibility of participation of hydrogen bonding or proton transfer reactions in the subsequent excited state relaxation and fluorescence processes. The results in Fig. 3 clearly suggest a possible role for sidechains of charged amino acids (Lys, Arg, Asp, Glu, His) behind the UV-Vis absorption between 250 and 400 nm. However, there are notable differences in the spectral features in Fig. 3a versus the absorption profles of the a 3 C protein (Fig. 2a and b). The tail of the charged amino acid spectra (beyond 320 nm) extends up to $500 nm. Short peptides containing Lys placed at different separations in the peptide sequence and poly-L-Lys$HCl solutions also show similar absorption features (Fig. 3c). In contrast, the long tail of the a 3 C absorption spectrum extends up to 800 nm. Thus, a possible role of the protein fold in the origin of tail spectra between 400 and 800 nm (Fig. 2a) is anticipated. Further, a comparison of molar extinctions coefficients of pure Lys amino acid solutions (3 ¼ 1.42 M 1 cm 1 at 270 nm) with that for a 3 C (3 ¼ 5808 M 1 cm 1 at 270 nm) reveals that $4000 fold enhancement in absorptivity at 270 nm is achieved by the folded a 3 C protein structure. We note, however, that a 3 C also contains other charged amino acids (Glu and Arg) besides Lys. To summarize, the results in this section demonstrate the ability of charged residues either in amino acid form or within extended peptide chains to absorb in the near UV. The protein fold further enhances the spectral range for the charged amino acids and the subsequent sections examine the possible mechanisms of enhancement. Computed UV-Vis absorption spectra for Lys and Glu monomers show charge transfer transitions Since the a 3 C protein is rich in both Glu and Lys, we carried out TDDFT electronic structure calculations on 100 structures of each charged amino acid sampled from MD simulations of a 3 C (see Methods for a discussion on our modelling assumptions) to simulate their absorption spectra between 200 and 800 nm (Fig. 4 top row). Here we discuss the spectra of monomers with their backbone amide units capped with Gly to represent the extended backbone present in the protein environment (see Methods for our capping strategy). The application of electronic structure calculations to MD sampled structures has proven to be effective for calculating UV-Vis spectra and electronic couplings for CT in organic molecules. 39,40, We visualize the lowest energy transitions of Lys, Glu, and Gly through difference density plots which show the location of hole (pink) and electron (blue) density on each amino acid fragment (Fig. 4 bottom row). The computed Gly spectra (Fig. 4a) extends up to $250 nm with the lowest energy transition at 248 nm delocalized over the entire backbone unit. A decomposition of the transition into constituent single orbital transitions (Fig. S8 of ESI †) reveals that it predominantly involves the frontier orbitals (HOMO/ LUMO/LUMO+5) delocalized over the backbone. Note that the weak transitions above 300 nm are spurious, arising due to the truncated form of the peptide backbone used in our calculations (see discussion below about the capping group effects). The simulated spectrum of Lys monomer (Fig. 4b) shows transitions in the same spectral range as the Gly control (Fig. 4a) extending to slightly higher wavelengths (up to 270 nm). In contrast, the Glu monomer spectra is distinct, displaying prominent transitions up to 450 nm (Fig. 4c). Difference density plots for the lowest energy Lys transitions around 270 nm show the electron density decreases (pink) on the peptide backbone and increases (blue) on the charged amino group and sidechain of Lys (Fig. 4b). This represents a CT transition involving the Lys backbone and its charged amino group. The positive charge on the Lys sidechain amino group, makes it a favorable location for the frontier unflled orbitals of Lys (e.g. the LUMO in Fig. S8 of ESI †), thereby populating the amino group with charge acceptor states. The average HOMO-LUMO gap for the Lys structures is reduced by $2 eV with respect to that for Gly (Fig. 4 bottom row). Thus, photoinduced CT transitions should be characteristic and unique to amino acids with charged sidechains and their derivatives. Since CT transitions are highly sensitive to the nature of the charge donor/acceptor states and the chemical structure of the sidechain separating them, each charged amino acid (Lys, Glu, Arg, Asp, and His) is expected to show distinct absorption features. Indeed, the Glu spectra (Fig. 4c) shows transitions over a much greater spectral range relative to Lys, extending into the visible wavelength range (up to 450 nm). Further, since the Glu sidechain is negatively charged, the flled orbitals (e.g. the HOMO in Fig. S8 of ESI †) are placed on the carboxylate group. Thus, for Glu, the direction of photoinduced CT is opposite that for Lys, from the sidechain carboxylate group to the polypeptide backbone (Fig. 4c bottom row). Differences between the charged sidechain groups (carboxylate COO for Glu vs. amino NH 3 + for Lys), different extents of hyperconjugation involving the charged groups, the presence of lone pair electrons for Glu, and the shorter sidechain for Glu (2 CH 2 links vs. 4 in Lys), all contribute towards the difference in spectral features for Lys and Glu monomers. The average HOMO-LUMO gap for the Glu structures is reduced by $4 eV and $2 eV with respect to that for Gly and Lys respectively. We characterized all transitions in the simulated spectra for each amino acid through two measures of spatial separation of charges (see Methods): (1) charge separation indices (Dr) 36 given by eqn (1), and (2) distance between hole and electron centroids (D CT ) 34,37 given by eqn (2). The middle rows in Fig. 4 show the percentage of CT vs. non-CT transitions within 5 nm wavelength windows over the whole absorption spectral range for all three amino acids. The data show that transitions above 200 nm for Glu and Lys monomers are rich in CT transitions. It is of course possible to get signifcant photoinduced charge separation on the extended polypeptide backbone, so that CT transitions are not exclusive to charged amino acids. For instance, the control Gly spectra (Fig. 4a) also seem to produce CT transitions above 200 nm. However, analyses of these transitions reveal that these states are actually localized backbone transitions showing spurious charge separation due to small orbital contributions of the capping groups (Fig. S7 of ESI †). However, this contamination appears to be restricted to transitions close to 200 nm and the lowest energy transitions around 250 nm are not signifcantly affected (see comparison of transitions around 250 nm for Gly dimer and tetramer in Fig. S5 of ESI †). Control calculations on Gly tetramers and Gly dipeptides truncated with methyl capping show (Fig. S6 of ESI †) that the spurious CT character of the transitions above 200 nm are diminished along with a blue shift in the spectra. In our calculations on tetramers and dipeptides, we fnd CT transitions around the 150-180 nm wavelength range (Fig. S6 of ESI †) consistent with previous computational reports of dipeptide and tripeptide spectra. 32,38,51 Finally, a small number of spurious non-CT transitions which are completely localized on capping groups (Fig. 4 and S7 of ESI †) appear consistently around 320 nm in the spectra of all three amino acids (Gly, Lys and Glu) which can be clearly identifed and distinguished from CT transitions. To assess the impact of the capping groups on the signature CT transitions of charged amino acids, we carried out control calculations on Glu peptides with different backbone extensions/capping (Fig. S6 of ESI †). These control calculations show that the effect of the capping group is negligible for transitions beyond 250 nm. For the Glu spectra, the most pronounced changes with change in backbone extension occur around 200 nm for transitions localized on the backbone (consistent with results for Gly peptides in Fig. S6 of ESI †). In contrast, the prominent CT absorption of Glu at lower energies (above 250 nm) is not altered in terms of both peak intensities and the spectral range (Fig. S6 of ESI †). We thus conclude that the effect of the capping groups is mostly restricted to backbone transitions around 200 nm with negligible effect on the signature backbone/ sidechain CT transitions of the charged amino acids above 250 nm. To summarize, charged amino acids, Lys and Glu, in monomeric form produce characteristic CT transitions. The intensities and spectral range of transitions for monomeric Lys and Gly with extended backbone are very similar and these amino acids are not distinguishable on the basis of their absorption spectra. The electronic properties of the monomeric Lys/Glu chromophores is neither able to explain the full spectral range of the transitions seen in high concentration Lys solutions (250-400 nm) nor that seen for the a 3 C protein (extending up to 800 nm). In the following sections, we explore higher order sidechain interactions between the charged amino acids within a 3 C which shed light on the role of the protein fold in dramatically extending the spectral range of Lys/Glu CT transitions. ## MD simulations of a 3 C reveal signicant interactions between Lys and Glu sidechains The NMR structures for the a 3 C protein (Fig. 1a) show several Lys and Glu residue pairs placed in close proximity. We thus investigated the interactions of Lys/Glu sidechains within the a 3 C protein fold using classical atomistic MD simulations of the solvated protein (see Methods). As discussed previously, even at the maximum concentration of a 3 C employed in our experiments (105 mM), we expect the protein to remain in monomer form. Accordingly, our simulations comprised of a single a 3 C molecule immersed in water box of volume $22 500 3 with periodic boundary conditions. We generated radial distribution function (RDF) plots capturing the range of pairwise atomic separations in our MD simulation trajectory (see Fig. 5a): (1) Lys amino nitrogen (N A -N A ) atom pairs, (2) Glu carboxylate carbon and Lys amino nitrogen (C C -N A ) atom pairs, and (3) Glu carboxylate carbon (C C -C C ) atom pairs. Further, we created 2-D contact maps displaying the average separations for these atom pairs representing Lys-Lys, Glu-Lys, and Glu-Glu sidechain interactions over the MD trajectory. The Lys N A -N A RDF plot (Fig. 5b) shows peaks around 4.5 and 7 which is surprising as two positively charged sidechains should repel each other. This observation is reinforced in the Lys-Lys contact map (Fig. 5c) which reveals multiple sets of amino group interactions, wherein the average N A -N A separation is lower than 7 over the MD trajectory. A visualization of the dynamics of Lys residue pairs during the MD trajectory reveals that the interactions of Lys amino groups are mediated Water molecules mediate Lys-Lys sidechain interactions through hydrogen bonding and by screening the Lys amino group charges as hypothesized previously. 10 We note that the mediation of charged sidechain interactions by polar and charged chemical species (water, Glu sidechains) will also include electronic effects which are not captured by MD simulations 52 but can signifcantly modulate spectral features. We discuss such effects in the subsequent sections. The C C -N A RDF shows a peak around 3.5 (Fig. 5b), corresponding to strong salt bridge interactions between the Lys amino group and the Glu carboxylate group. The time scales associated with Lys-Lys or Glu-Lys sidechain interactions vary from picoseconds to a few nanoseconds in our MD trajectory (Fig. S11 of ESI †). The Glu C C -C C RDF plot shows peaks at $6 and $9 , indicating weaker interactions between Glu sidechains relative to that between Lys sidechains. Note that sidechain interactions for a 3 C include both nearest neighbor residue pairs (NN pairs) or distally separated residue pairs (DS) in the protein sequence (Fig. 1b). Both DS and NN interactions tend to show similar separations between the amino/ carboxylate groups (Fig. S12 in ESI †), but the electronic coupling strengths for such interactions differ so as to produce signifcant spectral differences (vide infra). Interactions between Lys and Glu sidechains can extend the spectral range of CT transitions from charged amino acids to wavelengths above 300 nm We generated TDDFT based 200-800 nm spectra (see Methods) for NN and DS Lys-Lys, Glu-Glu and Lys-Glu residue pairs sampled from the a 3 C MD trajectory. In these calculations we retained an extended dimer backbone for DS fragments (details in Methods) wherein the backbone of each of the two residues of a DS pair was extended to include the backbone of the adjacent peptide units. We examined data (Fig. 6) for DS pairs (corresponding data for NN pairs are shown in Fig. S13 of ESI †) for three different separations of amino/carboxylate groups chosen on the basis of RDF data (Fig. 6). Specifcally, for each Lys-Lys, Glu-Glu, and Glu-Lys residue pair, we processed 100 conformations each with N A -N A , C C -C C , and C C -N A atom pair separations corresponding to red (strong interactions), green (intermediate interactions) and blue (weak interactions) shaded ranges in the corresponding RDF plots (Fig. 6 panels (a4-c4)). Both Lys-Lys and Glu-Glu sidechain interactions create new low energy transitions (panels (a1-a3) and (b1-b3) in Fig. 6) which extend the absorption range seen for Lys and Glu monomers by 100-300 nm towards the visible region. The spectral range for DS Lys-Lys pairs (Fig. 6, panels (a1-a3)) extends up to 550 nm (strong interactions), 500 nm (intermediate interactions), and 350 nm (weak interactions). The NN data (Fig. S13 of ESI †) show similar but weaker extensions of the spectral range for interacting Lys-Lys pairs; corresponding spectral ranges for NN interactions (Fig. S13 of ESI †) are curtailed to 450 nm (strong interactions), 440 nm (intermediate interactions), and 310 nm (weak interactions), respectively. Similar trends are observed for DS vs. NN Glu-Glu pair spectra (Fig. 6 panels (b1-b3) and S13 of ESI †). Characterization of transitions in the spectra on the basis of charge separation measures (eqn (1) and ( 2)) for Lys-Lys and Glu-Glu interacting pairs (both DS and NN) reveals mostly CT transitions beyond 250 nm (Fig. S15 of ESI †). For both DS and NN pairs, the lowest energy transitions involve CT between the extended backbone and the amino/carboxylate group of Lys/Glu (difference density plots in Fig. 6 panels (a1-a3) and (b1-b3), S13 of ESI †). A decomposition of the lowest energy transition for the most strongly interacting DS Lys-Lys pairs shows that (Fig. S14 of ESI †) these transitions are predominantly frontier orbital transitions with the LUMO delocalized between the interacting amino groups. For DS and NN Glu-Glu pairs, the direction of CT is opposite to that seen for Lys-Lys pairs with the charge donor states (pink) located on sidechain carboxylate groups. For both NN and DS Glu-Glu pairs, interactions between sidechain carboxylate groups are weaker than that for Lys amino groups. A decomposition of the lowest energy transitions (Fig. S14 of ESI †) for the most strongly interacting Glu-Glu pairs reveals that while these transitions are predominantly frontier orbital transitions, the HOMO is localized to one of the carboxylates of the interacting Glu-Glu pairs. The interactions between Lys-Glu residue pairs lead to starkly different spectral profles from that produced by Lys-Lys and Glu-Glu interactions. Weakly interacting Lys-Glu residues produce prominent transitions up to 800 nm, whereas strong interactions between the oppositely charged amino and carboxylate groups limit the transitions to 350 nm in the computed spectrum (Fig. 6 panels (c1-c3)). For weakly interacting Lys-Glu interactions, the lowest energy transitions involve CT transitions from the negatively charged (charge donor) Glu carboxylate group to the positively charged (charge acceptor) Lys amino group (Fig. 6 panel (c3) and S13 of ESI †). In contrast, strong interactions between Lys amino and Glu carboxylate groups create a neutral moiety through the formation of a salt bridge which curtails the spectral range of CT transitions (Fig. 6 panels (c1) and (c2)). Thus, in contrast to the Lys-Lys/Glu-Glu case, extension of CT spectral range to lower energies is inversely proportional to the strength of DS Lys-Glu interactions. We fnd that the NN Lys-Glu pairs are unable to form salt bridges due to geometry constraints and therefore show only weak interactions with spectral features that extend beyond 800 nm in the computed spectrum (Fig. S13 of ESI †). To summarize, the results in this section show that the association of charged sidechain amino/carboxylate groups in Lys/Glu residue pairs can greatly extend the spectral range of CT transitions observed for Lys/Glu monomers beyond 300 nm. Mechanism of CT transitions in a 3 C and modulation of absorption features by the protein and solvent environment Photoinduced CT can be described in the framework of a three component Donor(D) -Bridge(B) -Acceptor(A) molecular complex 50,53,54 wherein the D and A components are electron donating and electron accepting groups respectively. In contrast, the B component electronically couples the D and A components. The absorption of light by such a molecular complex can lead to a CT transition when electrons are transferred from D to A. The photoinduced CT may either proceed through the creation of a locally excited state on the donor or directly transfer charge from the donor to the acceptor. The CT transitions in charged amino acids subscribe to the latter model (Fig. 7a). While solvation coordinates couple to both donor and acceptor states, the vertical CT transition energies depend only on the ground state solvation confguration (Fig. 7a). The electronic coupling between D and A which determines the relative energies of the ground (j G ) and excited (j E ) state depends critically on the chemical structure of B. Thus, a quantitative description of intensities and peak positions of CT transitions should include a rigorous description of both solvation and the D-A electronic coupling through B states. In our study, we propose two different types of CT transitions (Fig. 7) to explain the a 3 C absorption profle: (1) peptide backbone to sidechain CT (PBS-CT), and (2) sidechain to sidechain CT (SS-CT). In PBS-CT, the peptide backbone and the charged amino/carboxylate groups of Lys/Glu residues act as the D/A components, and the sidechain alkyl groups act as the B component of the D-B-A complex (Fig. 7b). In SS-CT, the negatively charged Glu carboxylate groups act as D and the positively charged Lys amino groups act as A (Fig. 7c). Here, however, the B component is variable and depends on protein and solvent dynamics. Depending on the D-A distance, the B component could include variable number of water molecules and/or other sidechain groups. For both (PBS-CT and SS-CT) transitions the Lys/Glu charge plays a crucial role in dictating the direction of charge transfer. In the previous sections, we showed that Lys/Glu CT transition energies may be shifted to lower energies (above 300 nm) due to pairwise interactions with NN or DS Lys/Glu residues. The polar protein environment can further influence the CT energies of such charged amino acid dimers by introducing higher order interactions involving other charged sidechains and/or bound water. Further, the pK a of some of the interacting sidechains may be altered. Below we frst discuss how the dimer interactions lead to the spectral shifts observed for charged amino acids and then discuss the higher order effect of the environment on the spectra. The photoinduced CT in monomer Lys and Glu residues is a PBS-CT process (Fig. 7b). Association of charged sidechains can modulate PBS-CT (DS Lys-Lys example in Fig. 7b) by altering the relative stabilities of the j G and j E as a function of distance between these groups. The association of groups with like charges (Lys-Lys amino groups or Glu-Glu carboxylate groups) should destabilize j G due to unfavorable electrostatics. In contrast, such associations should stabilize j E due to a higher probability of placing/removing electrons from the Lys amino/Glu carboxylate groups during PBS-CT. The net result is a lowering of the energy gap for photoinduced PBS-CT transitions, commensurate with decreasing distance between charged sidechains. The average HOMO-LUMO gap for Lys-Lys dimers (Fig. S16 of ESI †) is lowered by $1 eV as the distance between their amino groups is reduced from around 6-7 to 3-4 . Likewise, the average HOMO-LUMO gap for Glu-Glu dimers (Fig. S16 of ESI †) is reduced by $0.7 eV as their carboxylate group separation reduces from around 8-10 to 4-5 . In contrast, when groups with unlike charges interact, the mechanism of CT changes to SS-CT (Fig. 7c). In this case strong interactions (Lys-Glu salt bridges) should stabilize j G (favorable electrostatics) and destabilize j E (neutralization of charges). Thus, for Lys-Glu interactions the SS-CT transition energy is lowered commensurate with increasing distance between the amino acid sidechains. The average HOMO-LUMO gap for Lys-Glu dimers (Fig. S16 of ESI †) increases by more than 2.5 eV as the Lys-Glu amino-carboxylate distance reduces from around 5-6 to the salt-bridge forming distance (3-4 ). We next examined spectral features of charged amino acid dimer pairs in the presence of explicit water and other chemical species in the vicinity. DS Lys-Lys dimer spectra are shifted to higher energies by around 100-150 nm upon including neighboring waters and/or Glu carboxylate groups in the calculations (Fig. 8 panels (a-c)). Similar blue shifts are also seen for DS Glu-Glu and DS Lys-Glu pairs with explicit water (Fig. S17 of ESI †). While inclusion of explicit waters induces only blue shifts in the DS Lys-Lys spectra, inclusion of carboxylate groups additionally leads to more intense transitions above 300 nm. These high intensity transitions arise from photoinduced SS-CT (Fig. 8 panels (b) and (c)) between carboxylate and amino groups not present in the vacuum Lys-Lys dimer spectra. For calculations with explicit waters, the DS Lys-Lys/DS Glu-Glu spectral shifts converge upon including 5 closest waters (Fig. 8d and S17 of ESI †), retaining a red shift of $100/150 nm for the lowest energy transitions relative to that for Lys/Glu monomer. The difference density plots (Fig. 8 and S17 †) show that nature of the lowest energy transitions are also unaltered (PBS-CT for Lys-Lys or Glu-Glu dimers and SS-CT for Lys-Glu dimers) upon inclusion of explicit water in the calculations. The spectral shifts for dimer pairs are highly sensitive to the position of waters with respect to the Lys amino groups. In Fig. 8e, we show that for the case of a single explicit water bridging the Lys-Lys pair the extent of spectral blue shifts introduced by the water can be reduced dramatically as the water is placed closer to the Lys amino groups. Similar trends are seen for Glu-Glu and Lys-Glu dimer spectra computed with explicit water (Fig. S17 of ESI †). These results clearly demonstrate that waters can both enhance and reduce the electronic coupling between charged sidechains (bridge effect vs. the polarization effect). We note, that a previous study comparing solvation of amino and carboxylate groups in ab initio and classical MD simulations showed that classical MD overestimates the number of water molecules interacting with charged groups and underestimates the electronic coupling between the charged sidechain moieties. 52 Thus, QM/MM calculations of solvated dimers replacing waters with point charges will overestimate solvent polarization effects while ignoring bridge electronic coupling contributions. Further, our QM calculations of solvated dimers with waters sampled from classical simulations also likely overestimate solvent polarization effects while underestimating bridge electronic coupling contributions. In our classical MD simulations of a 3 C carried out at pH 7, we assumed standard pK a values for the Lys amino and Glu carboxylate groups. Thus, all Lys and Glu residues are charged. However, given the high concentration of charged species and their dynamic encounters, it is possible that sidechains may exchange protons to change their charged states. Thus, in addition to the dimers pairs with both sidechains charged (doubly charged pairs) it may also be possible to fnd Lys-Lys, Glu-Glu, and Lys-Glu dimer pairs, wherein one of the monomer sidechains is uncharged (singly charged pairs). Following the analysis of dimer spectra presented earlier in this subsection, if one of the amino acids in a Lys-Lys or Glu-Glu dimer is neutralized, we anticipate stabilization of j G (less electrostatic repulsion) and destabilization of j E (lower charge on D and A) leading to a spectral blue shift towards that of the charged monomer amino acid. Indeed, we fnd (Fig. 8f) that the DS Lys-Lys spectral range, extending up to 550 nm for doubly charged pairs, is blue shifted, extending up to $300 nm when one of the amino groups is deprotonated in our calculations. The lowest energy transitions are sensitive to the position of the proton shared by singly charged Lys-Lys pairs and appear much more intense (relative to doubly charged dimer spectra) between 250 and 300 nm. The difference density plots for the lowest energy transition for singly charged Lys-Lys pairs reveal that both PBS-CT and SS-CT are operational. SS-CT occurs from the uncharged to the charged Lys amino group due to the short distance between the sidechains. For Lys-Glu salt bridge pairs, deprotonation of either Lys amino groups or protonation of Glu carboxylate groups also shifts the spectra to resemble that for the charged monomer (Fig. S18 of ESI †). Thus, for short Lys-Glu separations (salt bridge), the spectra should blue shift when the carboxylate group is protonated and red shift when the amino group is deprotonated. For all singly charged dimer pairs (Lys-Lys, Glu-Glu, and Lys-Glu) with well separated sidechains, we anticipate PBS-CT the charged monomer of the pair to be more competitive than SSCT between monomers producing spectra which resembles that for the charged monomer in the pair (see data for Lys-Glu in Fig. S18 of ESI †). To summarize, the a 3 C spectrum (Fig. 2) can be rationalized in terms of the light absorption by a range of D-B-A chromophores involving charged amino acids and two distinct types of photoinduced CT transitions (PBS-CT and SS-CT). The chromophores show diversity in terms of the electronic character of the D, B, and A groups. We have further computed absorption profles (Fig. S19 of ESI †) for NN Lys-AAA dimer pairs (AAA ¼ Ala, Val, Ile, Cys and Leu), which together with the Lys-Lys, Lys-Glu, and Glu-Glu pairs represent all Lys containing NN dimer species present in a 3 C. Other than the charged amino acid dimers (Lys-Lys, Lys-Glu, and Glu-Glu) studied in this section, no other dimer species shows signifcant absorption beyond 300 nm. Thus, our calculations highlight the role of the association between charged amino acid side-chains in producing the long tail absorption of a 3 C above 300 nm (Fig. 2a). Our computational results predict that the spectral range of the a 3 C ProCharTS profle should be sensitive to the interactions between Lys/Glu sidechains. Our analysis suggests two clear reaction coordinates that modulate the spectral range of Pro-CharTS: (1) distance between sidechains of charged amino acids, and (2) the sign of charge between interacting sidechains. Thus, we anticipate that the spectral range of ProCharTS will extend to lower energies (longer wavelengths above 300 nm) as the order of interactions between sidechains with like charges (Lys n ; Glu n ; n ¼ order) increases. In contrast, the spectra will be curtailed to higher energies (shorter wavelengths below 300 nm) when sidechains with unlike charges interact strongly (Lys-Glu salt bridges). Based on these observations we reasoned that perturbations of the protein tertiary fold which alter the Lys/Glu sidechain interactions should modify the UV-Vis absorption spectral profle of a 3 C. To verify this, we employed two approaches. In the frst approach, CD and absorption spectra for a 3 C was recorded over a temperature range of 25-85 C. The CD spectra (Fig. 9a) reveal that the protein retains a signifcant fraction of its a-helical structure even at temperatures as high as 85 C. In contrast, the UV-Vis absorption profle of a 3 C shows sensitivity to temperature (Fig. 9b) increasing by 1.2-2 fold between 300 and 500 nm. Fig. 9d shows that the temperature induced changes in spectral profle are non-uniform between 250 and 500 nm, distinct from the uniform and linear (Fig. 2a) changes induced by varying protein concentrations. The Bjerrum length (l B ¼ q 2 3k B T ; 3 ¼ dielectric of the medium, T ¼ temperature, k B ¼ Boltzmann's constant) for sidechain interactions in Lys-Lys and Glu-Glu dimers should decrease with increasing temperature as thermal energy compensates for electrostatic repulsion. In contrast, strong Lys-Glu interactions (salt bridges) should be destabilized as entropic contributions increase with temperature. Both factors, increase in Lys-Lys/Glu-Glu dimer associations and increase in Lys-Glu separations, rationalize the 90-120% increase in intensity for the a 3 C Pro-CharTS band between 300 and 500 nm as the temperature increases from 25-85 C. Contributions from Lys-Lys/Glu-Glu at higher energies (200-300 nm) should also go up. However, contributions of monomers and Lys-Glu salt bridge species will decrease (increase in Lys-Lys/Glu-Glu dimer formation and increase in Lys-Glu separation) in this wavelength range as temperature increases. These compensating factors can rationalize the modest 20% increase for the spectrum around 270 nm. In the next approach, our objective was to ascertain the role of charge in NH 3 + and COO groups on the protein CT spectra. For this purpose, we altered the pH of the medium to extreme limits (pH 1 and 13), so that the protein contained only one charged species. Under these conditions, both Lys-Glu salt bridges and long range SS-CT between Lys-Glu pairs (which dominate the spectra at longer wavelengths) should not exist. Further, in absence of electrostatic attraction between oppositely charged pairs of NH 3 + and COO groups, the protein structure is expected to be destabilized and likely unfolded, such that DS dimer interactions (Lys-Lys at pH 1.5 and Glu-Glu at pH 13) are reduced while NN dimer interactions (there are multiple adjacent pairs and triplets of Glu and Lys in the sequence), may still persist. NN dimer spectra (Lys-Lys and Glu-Glu only) are blue shifted with respect to that for their DS dimer counterparts (Fig. 6 and S13 of ESI †). Further, calculations on singly charged dimer pairs show spectra which are signifcantly blue-shifted with respect to that for their doubly charged counterparts (Fig. S18 of ESI †). Taken together, all these factors suggest that the ProCharTS absorption should be reduced at lower energies under extreme pH conditions. Indeed, Fig. 9c shows that absorption in the range 310-800 nm has nearly diminished (a dramatic >70% dip), both at pH 1.5 and 13, in comparison to the spectrum at pH 5.5. In summary, the experimental pH variations clearly validate the critical role played by charged Lys-Lys, Glu-Glu, and Lys-Glu interactions contributed by the protein fold to the a 3 C ProCharTS absorption in the near UV-Visible range. The temperature variation, on the other hand, emphasizes the sensitivity of the a 3 C ProCharTS absorption intensity to perturbation of tertiary Lys-Lys, Glu-Glu and Lys-Glu sidechain contacts. In order to extract detailed structural information on proteins from ProCharTS, a careful computational mapping of spectral intensities and peaks to geometric parameters (distances and angles) of specifc chromophores is required. Such mappings should account for spectral shifts due to the environment (see previous subsection) and must be benchmarked against experimental constraints. Our study represents a frst step in this direction and opens the door for both computational and experimental investigations for mapping the ProCharTS spectral profle to biomolecular structure and dynamics. ## Conclusions Using a 3 C as a model, we have unambiguously demonstrated that monomeric proteins lacking aromatic amino acids can display signifcant UV-Vis absorption with notable features between 250 and 300 nm and a long tail that can extend up to 800 nm (Fig. 2). We have presented several lines of evidence (both experimental and theoretical) to show that charged amino acids (Lys and Glu) can produce the observed spectral features. Through experimental control studies on high concentration solutions of non-aromatic amino acids and Lys containing peptides (Fig. 3), we showed that charged amino acids possess distinctive spectral features beyond 250 nm. Our computational analysis on Lys and Glu amino acids extracted from MD generated structures of a 3 C revealed CT transitions between 250 and 450 nm in the computed TDDFT spectra (Fig. 4). The CT transitions involve the amino (NH 3 + )/carboxylate (COO ) groups of Lys/Glu sidechains and the peptide backbone. Classical MD simulations revealed dimer and higher order interactions between Lys amino and Glu carboxylate groups imposed by the protein fold (Fig. 5). The interactions between charged amino acid sidechains were found to strongly modulate the computed CT absorption spectral profle (Fig. 6) and can account for the broad 250-800 nm absorption of a 3 C (Fig. 2). We described two specifc mechanisms of photoinduced CT (PBS-CT and SS-CT) involving Lys and Glu amino acids which are operational in a 3 C (Fig. 7) and their modulation by the polar solvent/protein environment (Fig. 8). Finally, we experimentally demonstrated the sensitivity of the a 3 C absorption spectrum to temperature and pH induced structural changes of the protein fold (Fig. 9). Our results connect UV-Vis absorption in proteins to the charged amino acid content of protein sequences for the frst time and rationalize hitherto unexplained experimental observations of absorption beyond 300 nm in Lys-rich proteins. The novel assignment of CT transitions to the 250-800 nm region in the absorption profle of proteins opens up a new spectral window (ProCharTS) to develop intrinsic spectral markers to monitor structure and dynamics of proteins rich in charged amino acids, such as nucleic acid binding proteins or intrinsically disordered proteins, irrespective of their aromatic amino acid content.

chemsum

{"title": "Near UV-Visible electronic absorption originating from charged amino acids in a monomeric protein", "journal": "Royal Society of Chemistry (RSC)"}

structure-astringency_relationship_of_anacardic_acids_from_cashew_apple_(anacardium_occidentale_l.)

## Abstract: Cashew apple presents a characteristic astringency. However, the compounds responsible for this characteristic were not described yet. A cashew apple extract was added to a BSA solution and the compounds before and after precipitation were analyzed by UPLC-QTOF/MS E . The extract astringency was measured on a 5-point scale (0: non astringent and 4: extremely astringent). Among the phenolics detected anacardic acids were identified and evaluated for their astringent effect. In the sensorial tests the cashew apple extract was considered very astringent (average of 2.5). A mixture of anacardic acids, had an average of 1.76 (astringent). The three isolated anacardic acids were evaluated. The in silico experiments were performed to analyze mainly the steric factor associated to the binding. The sensory results were confirmed by in silico analysis, indicating that a higher unsaturation degree of the aliphatic chain leads to an astringency increase. ## Introduction The cashew apple is a peduncle which supports the cashew nut. It can be consumed in natura, but also has good characteristics for processing due to its juicy pulp, high sugar and vitamin C content; and flavor. Despite its nutritional and functional potential, the cashew apple still presents low consumption when compared to other fruits, mainly due to its astringency. 1 Astringency is defined as a set of wrinkling sensations of the oral epithelium after exposure to substances such as aluminum or tannins. This sensation can be perceived by consumers as a "puckered" taste and throat irritation. Despite the importance of astringency for some products, the mechanisms of this attribute are not well known, so it is necessary to deepen the methodologies of astringency study. 2 The most accepted mechanism to explain how astringency occurs was proposed by Siebert, Carrasco, & Lynn, 1996, 3 in which the protein has a fixed number of sites to which the tannins can bind, while each polyphenol also has its fixed number of bonds. When the total number of polyphenol and protein bonds are the same, the largest complex and maximum precipitation will be produced. 2 New sensory and analytical techniques have been developed and used together in an effective procedure for the screening of non-volatile compounds important for the taste of food. This approach, combining instrumental analysis and human response led to the discovery of several previously unknown compounds such as the bitter and astringent compounds of different products. To solve the problem of astringency, it is necessary to identify the compounds present in the cashew apple that are responsible for this characteristic and, thus, to develop methodologies, extraction systems or even genetic modifications in cashew clones, aiming to decrease or eliminate the astringent compounds. The objective of this work was to identify the cashew apple components responsible for its astringency using sensory, instrumental and computational analysis. ## Reagents The reagents used were methanol HPLC (purity ≥ 99.9%, LiChrosolv ® , Germany); acetonitrile HPLC (purity ≥ 99,9%, Tedia, Fairfield, OH, USA); glacial acetic acid P.A. (purity ≥ 99,7%); genistein and methanol P.A. (purity ≥99,8%), purchased from Sigma-Aldrich (Saint Louis, MO, USA); and purified water from an Mili-Q system (Millipore, São Paulo, Brazil). ## Cashew Apples Cashew apples from clone CCP 09 were used for the compounds extraction. The fruits were harvested on the Embrapa experimental field located at Pacajus-CE, Brazil (4°11'26,62'' S; 38°29'50,78'' W), harvested in 2017 (September to November). After being sanitized the peduncles were freeze-dried and grinded. The samples were packed under vacuum and stored at -20º C until further use. ## Extraction of cashew apple phenolics Freeze-dried cashew apples (50 g) were extracted with methanol-water 60:40 (v/v) in an ultrasonic bath (Ultrasonic Cleaner 1400, Thornton/UNIQUE, São Paulo, Brazil), at 40 kHz, 100W, temperature of 25º C for 30 min. The mass:volume ratio used was 1:10 (m/v), being the extraction performed with ten replicates. Subsequently, the samples were centrifuged at 2,944 g for 15 min and the supernatants combined. The extract was dried under reduced pressure at 40° C, followed by freeze-drying to assure methanol removal. ## Protein Precipitation Protein precipitation was performed on the methanolic extract of the cashew apple with bovine serum albumin (BSA), according to the methodology described by Hagerman & Butler, 1978, 7 with adjustments. To 1.0 mL aqueous solution of the cashew apple extract was added 2.0 mL of BSA solution (1.0 mg.mL -1 ) in a 15 mL centrifuge tube. After vortexing for 1 min, it was allowed to stand 24 h at 8 o C for precipitation. After precipitation, the complex was centrifuged (2,944 g for 15 min) obtaining the supernatant (non-complexed phenolics) and the precipitate. The precipitate was gently washed with water, centrifuged (2,944 g for 10 min) and the precipitate was extracted with methanol on an ultrasound bath (5 min) and centrifuged. The extraction process was repeated four times and the combined methanolic extract was dried under reduced pressure at 40° C and freeze-dried (cashew-protein precipitate extract). To obtain sufficient phenolics for sensory analysis, the above protein precipitation process was carried out on a larger scale, respecting the proportions of methanolic extract and protein, only that BSA was substituted by an aqueous solution of commercial gelatin. ## UPLC-QTOF-MS E profile The analysis was performed using an Acquity UPLC (Waters, Milford, MA, USA) system, coupled with a Quadrupole/TOF (Waters) system. 8 A Waters Acquity UPLC BEH column (150 × 2.1 mm, 1.7 μm) was used, with the column temperature set at 40 °C. The binary gradient elution system consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B). The UPLC elution conditions were optimized as follows: linear gradient from 2% to 95% B (0-15 min), 100% B (15-17 min), 2% B (17.01), 2% (17.02-19.01 min), a flow of 0.4 mL.min -1 , and a sample injection volume of 5 μL. The chemical profiles of the samples were determined by coupling the Waters ACQUITY UPLC system to a QTOF mass spectrometer (Waters) with the electrospray ionization interface (ESI) in negative ionization mode. The ESI − mode was acquired in the range of 110-1180 Da, with a fixed source temperature of 120 °C and a desolvation temperature of 350 °C. A desolvation gas flow of 500 L.h -1 was used for the ESImode. The capillary voltage was 2.6 kV. Leucine enkephalin was used as a lock mass. The MS mode used Xevo G2-XS QToF. The spectrometer operated with MS E centroid programming using a tension ramp from 20 to 40 V. The instrument was controlled by the MassLynx 4.1 software program (Waters Corporation, USA). The samples were spiked with genistein (1ppm) internal standard. ## Fractionation of anacardic acids Anacardic acids were obtained by preparative HPLC fractionation of cashew nut shell liquid (CNSL) as described by Oiram Filho, Zocolo, Canuto, Silva Junior, and de Brito, 2019. 9 The compounds present in the anacardic acid mixture were isolated on a reverse phase chromatographic column Waters SunFirePrep C18 OBD (100 x 19 mm x 5 μm). The mobile phase was used in an isocratic mode using methanol, water and acetic acid in the proportion (90:10:1), run time of 40 min, and a flow of 3 mL.min -1 , at 25 ºC. The injection volume was of 1 mL at a concentration of 100 mg.mL -1 . The chromatograms were monitored at a wavelength of 280 nm. The yield obtained for the triene, diene and monoene anacardic acids were 22.1, 13.3 and 17.5 %, respectively. The purity of each anacardic acid isolated was monitored by HPLC 10 and the values were 98.93 %, 72.67 % and 79.68 % for triene, diene and monoene, respectively. The purity values of the compounds were satisfactory since, in the literature, studies reported isolation of phenolic compounds from purities between 75 and 99%. 11 ## Sensory Analysis The sensorial test was performed by previously selected and trained panelists, using test protocols approved by a Research Ethics Committee under Opinion n° 147.279. Before the tests were run, the panelists were asked to sign a Free and Informed Consent Form (TCLE). The analyzed samples were: methanolic cashew extract (ME), cashew protein precipitate extract (PPE), anacardic acid mixture (AnMix) and anacardic acids (An1, An2, An3). The samples were solubilized in bottled water with ºBrix and pH adjusted for the mean values of in natura cashews (7.1 and 4.15, respectively). The concentrations were defined according to the phenolic concentration values found in the literature for the cashew apple, in the range of 1 to 2 mg.mL -1 . 12 All samples were analyzed with repetition and the minimum interval between sessions was 10 min. The concentrations for sensory analysis were 2 and 5 mg.mL -1 for the methanolic cashew extract (ME2 and ME5 respectively); 1 and 2 mg.mL -1 for the cashew protein precipitate extract (PPE1 and PPE2 respectively); 1 mg.mL -1 for anacardic acid mixture (AnMix), anacardic acid 15:1 (An1), anacardic acid 15:2 (An2); and anacardic acid 15:3 (An3). Five mL of samples were served in 50 mL cups in a monadic sequence. The panelists were asked to put all the container contents in the mouth and let it stand for 10 s, roll the solution through the mouth, exposing it to all taste buds and buccal mucosa (at least 3 rotations) and then spit the solution into a container. After 15 s, the panelists marked the perceived astringency intensity on a 5-point scale (0 = not astringent, 1 = little astringent, 2= astringent, 3 = very astringent; and 4 = extremely astringent). The minimum and maximum extremes of the scale were previously determined in training. ## Statistical Analyses The results obtained in the sensorial astringency tests were submitted to analysis of variance (ANOVA) with the following sources of variation: sample (SAMP), assessor (ASSE) and the interaction SAMP X ASSE, being the assessor considered as a block. Significant differences between means were determined by the Ryan-Einot-Gabriel-Welschand Quiot test (REGWQ) with confidence interval of 95% (α = 0.05). The analyses were performed using the statistical program XLSTAT v. 18.1 (Addinsoft). ## Computational method The structures of the three anacardic acids (ene-derivatives of the salicylic acid) were built and optimized using Avogadro (version 1.0.3) using MMFF94 force field 13 and all of them based on the benzoic acid residue found in the active site of 6DHB. 14 Topology of the ligands for MD simulation were generated via the CGenFFserver. 15 As the penalty scores from CGenFF were lower than 50, no further optimization was taking in account. All MD simulations were carried out using the GROMACS software package (version 5.0.2). 16 An initial model of each protein-ligand complex in dodecahedral box filled by TIP3P water was constructed using the editconf and solvate tools of GROMACS. Z-Length of simulation box was determined by water thickness, minimum water height on top and bottom of the system was set to 10 . The net charge on the system was neutralized by adding Na + ions. The charmm36 force field 15, was used for all systems and simulations. The system was gradually relaxed according to position and angle restraint conditions to reach equilibrium (300 K, 1 atm). Then, 10 ns NPT (constant number of atoms, pressure, and temperature) simulation without any position restraint with 2 fs time step was performed. In NPT simulation, temperature and pressure were regulated using the V-rescale thermostat algorithm 21 and the Berendsen barostat algorithm, 22 respectively. The time constant for the temperature and pressure coupling was kept at 0.1 and 2.0 ps, respectively. The pressure was coupled with isotropic scheme with isothermal compressibility of 4.5×10 -5 bar -1 . The short-range nonbonded interactions were computed for the atom pairs within the cutoff of 1.2 nm, while the long-range electrostatic interactions were calculated using particle-mesh-Ewald summation method with fourth-order cubic interpolation and 0.16 nm grid spacing. The same method was reproduced for all simulations. All PDB files of the ligands are transcripted in the Supporting Information section. The files of the protein and protein-ligand complexes are provided in the same section as PDB files. ## Results and Discussion Table 1 shows the compounds that were tentatively identified in cashew extracts samples based on their fragment ions as well as a comparison with data from the literature. Among the identified compounds, pentagaloyl hexoside, a precursor for the formation of more complex ellagitannins and gallotannins was present in all samples analyzed. Three ancardic acids with C15 alkyl chain length and different degrees of unsaturation (tri, di and mono-unsaturated) were identified in the extracts, and one anacardic acid with C17 alkyl chain was present in the methanolic extract of the cashew apple and in the cashew protein precipitate extract. Anacardic acids are phenolic compounds derived from salicylic acid, and due to their aliphatic chain, have lipid characteristics. 23 They are present in higher concentration in the cashew nut shell liquid. 10 In cashew apple the concentration of these compounds varies from 0.20 to 0.51 %. 24 As shown in Table 1, the phenolics that precipitate alongside the protein had the same profile as the cashew methanolic extract. However, the anacardic acid 1 7:1 was not detected on the protein non-complexed fraction, probably due to its low concentration. The combination of sensory analysis with analytical techniques (bioguided isolation) has been of great importance for the recognition of several compounds that influence the sensorial characteristics of food products. In order to identify the chemical markers for astringency in the cashew apple, the astringency test was performed with the samples mentioned in the previous sections. Due to the presence of anacardic acids in the protein precipitate extract, a sensory analysis of a mixture of major anacardic acids (approximately 50 % for An3, 20 % for An2 and 30 % for An1), as well as the isolated compounds, was performed. A significant difference was observed among the samples regarding the intensity of astringency perceived by the sensory panelists. The means were compared by means of the REGWQ test; whose results are shown in Fig. 1. Cashew apple methanolic extracts at concentrations of 2 and 5 mg.mL -1 presented low astringency, scoring below 1.0, between 'not astringent' and 'little astringent' in the 5-point scale. The cashew protein precipitate extract in the concentration of 1 mg.mL -1 scored 1.12 ('little astringent'), however, when its concentration was doubled, the astringency perception increased, reaching 2.50 points, considered between "astringent" and "very astringent" by the panelists. Those results make clear that the protein precipitation was selective to concentrate the astringent compounds in the extract, since the sensory astringency perceived in cashew protein precipitate extract is statistically greater than in methanolic cashew extract. The anacardic acid mixture (concentration of 1 mg.mL -1 ) showed an intermediate astringency (1.75) between the two protein precipitate extracts, but not differing statistically from cashew protein precipitate extracts (1 and 2 mg.mL -1 ). Although the sensory panel was trained with reference samples for astringency intensity, the individuals reported a difficult to classify the samples as very or extremely astringent, probably because they were accustomed to the high astringency of cashew apples. This fact may lead the use of a shorter interval on the scale, with a maximum of 4 points, thus reducing the sensitivity of the test in detecting significant differences among the samples. For this same reason, there was also no difference between the anacardic acid mixture and the isolated anacardic acids. The triene anacardic acid (C15:3) reached 2.01 points ("astringent"), diene anacardic acid (C15:2) scored 1.63, and monoene anacardic acid (C15:1) averaged 0.86 ("little astringent"). However, monoene and triene anacardic acids differed statistically from each other. The panelists described a sensation of stinging followed by throat irritation after tasting the anacardic acids. The throat irritation and roughness in the mouth, named astringency subqualities, are common in the perception of astringency in the human palate and are even perceived after ingestion of cashew juice. In lack of a specific protein or receptor responsible for the astringency sensation and in the attempt to justify the stronger interaction of the anacardic acid triene we have used the T-cell immunoglobulin and mucin-domain containing-3 crystallized with a benzoic acid residue (6DHB). As this protein have great affinity for derivatives of the salicylic acid, 14 we have conducted the in silico experiments to analyze mainly the steric and electrostatic factors associated to the energy of binding (Table 2). The decomposition of the short-range energies, from Coulombic and Lennard-Jones models, shows the major stability of the triene in comparison to the mono and the diene. The total interaction energy, Coulombic plus Lennard-Jones (and propagating the error according to the standard formula for addition of two quantities) for the triene (C15:3) gives a total value of -155.3892 ± 8.5 kJ mol -1 , lower than -117.3290±25.2 kJ mol -1 and -146.5156±13.1, for mono and diene, respectively, confirming the indicatives shown by the decomposed parts. In a complementary way, we have analyzed the values of RMSD (data not shown), calculated from the protein backbone to the structure of the ligand. The data show how much the ligand binding pose has changed over the course of the simulation, adding more information regarding the more instable nature of the complex involving the monoene, which indeed make sense, considering the higher translational freedom degree of the hydrocarbon tail, compared to the more rigid triene. Fig. 2 shows information for the 10 ns simulation concerning the three anacardic acids. First of all, we can point the instability of the protein-ligand complex involving the monoene (Fig. 2a), once it departs from the active site after about 5 ns of simulation. The di and triene, although stays attached, we, could see a lesser interaction between the side chain of the triene, more rigid, compared to the diene, following the energetic parameters presented in Table 2. For an even more detailed look at how the ligands are interacting with 6DHB, we have computed the distance between the carboxyl and hydroxyl groups of anacardic acids and the amino group of three very important aminoacids of the active site of 6DHB (Fig. 3). Exactly as reported by Golebiowski, Fiorucci, Adrian-Scotto, Fernandez-Carmona, & Antonczak, 2011, 25 studying the astringency of tannins, the main nature of the binding process of the astringency lies on the formation of hydrogen bonds, and for that, the backbone of the protein is more important than its side-chains. Another characteristic, also corroborated by Cala et al., 2012 26 is that the ligand (in their case tannins) has preferentially been found in the hydrophilic site of some proteins segments responsible for the astringency response. We have considered here that a hydrogen bond is formed when the donor and the acceptor are at most 3.5 apart (≤ 0.35 nm). As shown in Table 3, the stronger hydrogen bond is formed from the triene to the aspartate residue (residue 98 from 6DHB), all values indicate the more favorable formation of hydrogen bonds between the triene and 6DHB. As highlighted by Fig. 3, we can clear see that the hydroxyl group has more stereo advantage regarding the hydrogen bond to 6DHB than the carboxyl group. Docking studies of anacardic acid and different proteins have been performed. For the matrix metalloproteinases, MMP-2/gelatinase A and MMP-9/gelatinase B, placed the head group in the aliphatic pocket, with the carboxylate group functioning as a zinc-binding group and forming a hydrogen bond to the active site of MMP-2; and the hydroxyl group of anacardic acid also forms a hydrogen bond to backbone oxygen of Ala192. 27 The anacardic acid carboxylate group also functions as a zinc-binding group in MMP-9 and forms a hydrogen bond to the Glu402 side chain, while the hydroxyl group of anacardic acid forms a hydrogen bond to backbone oxygen of Ala189. With parasitic sirtuins was observed the pose of anacardic acid in the TcSIR2rp1 pocket forming hydrogen bonds between the carboxylic group of the ligand and the side chain of Arg50. 28 Regarding the estrogen receptor α (ERα)-expressing breast cancer cell lines it was proposed that the alkyl chain of anacardic acid may be an important factor, in combination with the salicylic moiety, for high affinity for the ERα DBD and no affinity for the ERα LBD. 29 Anacardic acid interaction with the steroid receptor coactivator (Src)/focal adhesion kinase (FAK) was evaluated and mechanistically, it was proposed that it could dock into the hydrophobic pocket of Src and FAK protein. 30 The anacardic acid interaction with SIRT isoforms, which are class III histone deacetylases (HDACs) also revealed that it made hydrogen bonds, through its carboxyl group and hydroxyl group. It also verified that the rigidification of the tail could promote stable hydrophobic interactions with the pockets, decreasing the flexibility, and therefore the entropy of the systems. 31 In conclusion, the astringent effect of anacardic acids was observed and described for the first time. The protein precipitation method revealed a profile similar to the extract, especially phenolics in the protein precipitate as revealed by UPLC-ES-QTOF-MS E . The isolation of anacardic acids allowed the sensory evaluation and the ranking of astringency of these compounds was established based on the unsaturation pattern. Apparently, a higher unsaturation degree of the anacardic acid aliphatic chain leads to an increase on astringency. The triene was more astringent when compared to the monoene. The sensory data was corroborated by in silico analysis of the interaction energy of anacardic acids and a mucin, which demonstrated a stronger interaction of triene as compared to monoene anacardic acids. Footnote: ME1= Methanolic cashew extract (2 mg.mL -1 ); ME2= Methanolic cashew extract (5 mg.mL -1 ); PPE1= Cashew protein precipitate extract (1 mg.mL -1 ); PPE2= Cashew protein precipitate extract (2 mg.mL -1 ); AnMix= Anacardic acid mixture(1 mg.mL -1 ); An3= Anacardic acid 15:3 (1 mg.mL -1 ); An2= Anacardic acid 15:2 (1 mg.mL -1 ); An1= Anacardic acid 15:1 (1 mg.mL -1 ). Astringency scale ranging from 0 to 4, being: 0 = not astringent and 4 = extremely astringent.

chemsum

{"title": "Structure-astringency relationship of anacardic acids from cashew apple (Anacardium occidentale L.)", "journal": "ChemRxiv"}

role_of_the_dispersion_force_in_modeling_the_interfacial_properties_of_molecule-metal_interfaces:_ad

## Abstract: We present density functional theory calculations of the geometry, adsorption energy and electronic structure of thiophene adsorbed on Cu(111), Cu(110) and Cu(100) surfaces. Our calculations employ dispersion corrections and self-consistent van der Waals density functionals (vdW-DFs). In terms of speed and accuracy, we find that the dispersion-energy-corrected Revised Perdue-Burke-Enzerhof (RPBE) functional is the ''best balanced'' method for predicting structural and energetic properties, while vdW-DF is also highly accurate if a proper exchange functional is used. Discrepancies between theory and experiment in molecular geometry can be solved by considering x-ray generated core-holes. However, the discrepancy concerning the adsorption site for thiophene/Cu(100) remains unresolved and requires both further experiments and deeper theoretical analysis. For all the interfaces, the PBE functional reveals a covalent bonding picture which the inclusion of dispersive contributions does not change to a vdW one. Our results provide a comprehensive understanding of the role of dispersive forces in modelling molecule-metal interfaces. ## M olecular nanostructures on solid surfaces, especially on metals, have received great attention in recent years . Accurate modelling of the properties of organic molecule-metal interfaces is of great importance to build, modulate, and utilize these molecular nanostructures . Because some of the constituent molecules have a small or vanishing electrical dipole moment, the dispersion force is a major part for the van der Waals (vdW) force between these molecules and other nanostructures. Dispersion forces are not included in conventional density functional theory (DFT), which poses a challenge to the theoretical description of the interfacial properties of these molecules on solid surfaces. However, the inclusion of dispersion forces has been realized in recent developments of DFT and has been applied to a number of organic molecule-metal interfaces . Still, the role of the dispersion correction in determining the geometry, adsorption energy and bonding picture of an interface remains poorly understood. Dispersion forces have been modelled in Perylene-3,4,9,10-tetracarboxylicacid dianhydride (PTCDA)/Ag(111) 21 , thiophene/Cu(110) 22 , pyrazine/Cu(110) 23 , azobenzene/Ag(111) 24 , and C 60 /Au(111) 25 using semi-empirical dispersion corrections (DFT-D) 24,25 , an ab-initio C 6 parameter with the consideration of many-body screening effects 26 , and van der Waals density functionals (vdW-DF) 17,18 , both by a post-Generalized Gradient Approximation (GGA) method 27 and later by a self-consistent method 24 with a strongly reduced computational cost 28 . To analyze geometries, experimental results from X-ray standing waves (XSW) for PTCDA/Ag(111) 29 and azobenzene/Ag(111) 30 are available for comparison with theory. Neither vdW-DF nor DFT-D reproduces accurately the XSW experiment on PTCDA/Ag(111) 21 , but an ionic final state (IFS) approximation that considers electron-core-hole interactions does. For azobenzene/Ag(111), out of two DFT-D and four vdW-DF methods, only the Tkatchenko-Scheffler (TS) scheme 13 yields a reasonable Ag-N distance 30 . It is still an open issue whether the XSW measured results are for the ground state . If the electronic relaxation after core-hole creation is sufficiently fast, the measured geometry should be identical to the groundstate structure, or if the molecule-substrate bonding is sufficiently strong then the geometry should be very close to that of the ground state. Otherwise, an excited-state method such as IFS should be considered to reproduce the XSW measurements. An accurate description of adsorption energies has recently been achieved by vdW-DF methods 21,22,24,35 . The choice of the exchange functional together with the vdW correlation functional was found to be of vital importance for an accurate calculation of interaction energies and bond lengths 35 . These approaches seek the best-match exchange functional to use together with the vdW correlation functional and have generated new exchange functionals including C09 36 , optPBE 19 , optB88 19 and optB86b 20 . The non-local correlation functional can also be optimized, as in vdW-DF2 18 . These exchange functionals have usually been tested by a ''standard test'' of the S22 set for molecules 19 , and by lattice constants, bulk moduli and atomization energies for solids 20 . Although these functionals have also been tested recently for interfaces, including graphene and Ni(111) 35 , their performance for organic molecule-metal interfaces remains unknown. In addition, the role of the exchange functional in DFT-D, a much cheaper method, has not yet been appreciated. The dispersion correction has mostly been employed together with the PBE functional 37 , which overestimates adsorption energies quite considerably. In fact different theoretical results remain in conflict over the appropriate bonding picture. Covalent bonding was identified using standard DFT for both pyrazine/Cu(110) 38 and C60/Au(111) 39 , and the bonding picture of C60/Au(111) revealed by vdW-DF 27 is also covalent. However, for pyrazine/Cu(110) the bonding was found 23 to be due solely to dispersion forces, contradicting the DFT result. Resolving such a complex situation calls for a systematic investigation. Because a large amount of experimental data is available for these systems , in this paper we report DFT calculations for thiophene on Cu(111), Cu(110) and Cu(100) surfaces. We consider the dispersion force at the DFT-D and self-consistent vdW-DF levels, and include the electron-core-hole interaction by the IFS approximation. The geometry of the adsorbed thiophene molecule was relaxed using standard PBE, different DFT-D and vdW-DF methods and standard PBE with the IFS approximation. Using the calculated interface structures, the corresponding adsorption energies were computed using PBE and all DFT-D and vdW-DF methods as were the electronic structures. The local density of states (LDOS) and differential charge density (DCD) were found to support unambiguously a covalent bonding picture for thiophene/Cu, in contrast to a previous suggestion of vdW bonding for thiophene/Cu(110) 22 . The choice of methods (functionals) used in the prediction of structural and energetic properties for these interfaces is summarized in the discussion section. Our results improve the overall understanding of the role of dispersion forces and electron-core-hole interactions in organic molecule-metal interfaces. ## Results Structural properties of thiophene/Cu(111). Our calculated results for the structural relaxations of thiophene adsorbed on three Cu surfaces are summarized in Table 1 and Figure 1. Six configurations of adsorption site and molecular orientation were considered. The top site, where the S atom is located on the top of a Cu atom, shown in Figure 1(a), was found to be the most energetically favored, which is consistent with experiment 40,41 . In experiments, the tilt angle a between the molecular plane of thiophene and the Cu surface varies from 12u to 45u when the coverage increases from 0.03 ML to 0.14 ML 40 . Because the Cu(111) surface was modeled using a 3 3 3 supercell, equivalent to a coverage of 0.11 ML, the results should be comparable with the tilting angle a 5 26 6 5u measured at 0.1 ML 40 . The PBE calculation gives 21.6u, within the experimental error bar. Although the PBE Cu-S bond length of 2.68 A ˚is slightly larger than the experimental values of 2.62 6 0.03 A ˚40 and 2.50 6 0.02 A ˚41 , this approach is known to be prone to overestimate bond lengths. An attractive potential, such as the dispersion correction with semi-empirical pairwise C 6 -R 6 coefficients introduced by Grimme (G06) 11 , overcomes the overestimated Cu-S distance. It slightly shortens the Cu-S bond length to 2.51 A ˚, but drastically reduces the tilting angle to 7.6u [Figure 1(c)]. Two categories of interactions can be inferred from the PBE results, namely Cu-S covalent bonding and the relatively weak Cu-p interaction. The Cu-p interaction cannot adequately resist the attraction from the dispersion correction, leading to a tilting angle significantly smaller than the experiment. The C 6 -R 6 coefficients depend on the atoms involved but not on the functional employed in the calculation. To balance the overenhanced Cu-p interaction, we employed a more repulsive functional, namely RPBE (RPBE-G06). One flat and one tilted configuration were found to be energetically degenerate (within 3 meV), which implies a competition between the Cu-S and Cu-p interactions. Similar behaviour was also found in the optPBE-vdW and optB88-vdW results. As shown in Figure 1(d) and (e) and Table 1, the flat configuration has a negative tilting angle and a rather large Cu-S bond length, a RPBE-G06-flat 5 21.5u and d RPBE-G06-flat 5 3.17 A ˚; in the tilted configuration the bond length, d RPBE-G06-tilted 5 2.62 A ˚, is very close to experiment but the angle a RPBE-G06-flat 5 13.2u is roughly 10u too small. A similar degeneracy of tilted and flat configurations was previously reported in a DFT investigation of thiophene on Cu(100) 50 . The C 6 coefficients in the TS dispersion correction are based on the ground-state electron density 13 , which offers a better transferability than G06. However, the replacement of G06 by TS does not cause a qualitative change in the relaxed atomic structures, increasing the Cu-S bond length by only 5-10%. Although these PBE and DFT-D results are reasonable, they are not satisfactory for predictive purposes. Thus we adopted a more sophisticated method to model the thiophene/Cu(111) interface, namely self-consistent vdW-DF including four combinations of exchange and non-local correlation functionals. Three of the exchange functionals were used together with the same correlation functional, namely optPBE-vdW, optB88-vdW and optB86b-vdW, while vdW-DF2 is an exception with a non-local correlation functional which was optimized especially for organic molecules. The geometry revealed by the first three functionals is similar to that of RPBE-G06: in the tilted configuration, the Cu-S bond length varies from the shortest value of 2.53 A ˚for optB86b-vdW to the longest value of 2.88 A ˚for optPBE-vdW. The trend of the bond lengths is consistent with the nature of these exchange functionals, in that the most attractive is optB86b and the least attractive optPBE; this result may be read from the exchange enhancement factor F x as a function of the dimensionless density gradient s in the range between 1.0 and 2.0 19,20 . The vdW-DF2 functional performs very well for the S22 set 18 owing to its optimization for molecules, but it fails to reproduce the thiophene/Cu(111) structure, giving rise to a rather large adsorption height, which can also be deduced from F x (s), and to a negative tilt angle. The angles predicted by the other three methods are also significantly lower than the experimental result (Table 1). The fact that none of these sets of theoretical bond lengths and tilting angles achieves good agreement with experiment implies that the XSW measurements may reflect the structure of the interface in an effectively charged state. Thus we considered the effects of electron-core-hole interactions by introducing three different effective charges of 0.5 e, 1.0 e and 1.5 e. This effective charge causes an abrupt change of the adsorption configuration: with an effective charge of 0.5 e, the molecule moves away from the surface (d 5 3.05 A ˚) and becomes more flat (a 5 14.8u). A charge of 1.0 e pushes the molecule back to the tilted configuration, which suggests that the screening charges are transferred into a Cu-S bonding state, strengthening the Cu-S covalent interaction. Such an enhanced Cu-S interaction results in an sp 2 to sp 3 transition of the sulphur hybridization, leading to a more tilted molecule. The bond length d 5 2.56 A ˚and the tilting angle a 5 22.0u, shown in Table 1, are both within the error bars of the experiments . Structural properties of thiophene/Cu(110). The top site is, again, the most favorable adsorption site for thiophene/Cu(110). The Cu-S bond length of 2.39 A ˚(PBE value, see Figure 1(h)) indicates a covalent bond, which differs by only 0.02 A ˚from the previously reported theoretical value of 2.41 A ˚22 . The relaxed atomic structure shows that an adjacent C, marked by a red square in Figure 1(g), also bonds covalently to the Cu atom underneath, a result confirmed by the differential charge densities, as elucidated later. None of the methods more advanced than PBE (listed in Table 1), except vdW-DF2, find significant changes to the Cu-S bond length, which ranges from 2.32 A ˚to 2.45 A ˚. This is distinctively different from the case of thiophene/Cu(111), where the tilt angle and bond length appear to be very sensitive to the method. The shorter bond lengths suggest much stronger covalent Cu-S and Cu-C bonds. Thus the same energy correction due to dispersive contributions is minor compared with the bonding energy, resulting in much smaller changes of bond length than on the (111) surface. The IFS estimate of the core-hole interaction, which obviously changes the adsorption geometry of thiophene/Cu(111), shortens the Cu-S bond length only slightly, from 2.40 A ˚(0.5 e) to 2.30 A ˚(1.5 e). Only the vdW-DF2 method predicts a Cu-S bond length at least 0.3 A ˚larger than the others, similar to the (111) cases. The tilt angle varies from 3.0u (PBE-G06) to 20.7u (PBE-CH-1e) between methods. This wide range of angles can be ascribed to a transition in electronic hybridization from sp 2 to sp 3 on the carbon adjacent to the S atom, as shown in Figure 1(k) and similar to the case of CuPcF 16 /Ag(111) 31 . No structural information is available from experiment for this interface. The arrangement of atoms on Cu(110) shares with Cu(100) the feature that their ½1 10 directions are identical, and so measured values for thiophene/Cu(100) may provide a meaningful comparison with the calculated results for Cu(110). The experimental bond length in thiophene/Cu(100) is 2.42 6 0.02 A ˚44 , which falls in the range from 2.32 A ˚to 2.45 A ˚of our theoretical values (other than vdW-DF2), implying good agreement on the Cu-S bond length. Structural properties of thiophene/Cu(100). Early theoretical studies of this interface predicted the Top site to be the most favorable for adsorption 50 . Experiments, however, suggest that the Bridge site is preferred. In our calculations the Top site (Figure 1(l) left) is at least 30 meV more stable than the Bridge site (Figure 1(l) right) for all methods. With the exception of vdW-DF2, the bond length of the Top site configuration is insensitive to the different corrections considered (Table 1), similar to thiophene/Cu(110). The IFS method returns similar structures for all effective charges. All methods used reveal that the Top configuration is more stable than the Bridge by several tens of meV (Table 2). At the Bridge site, there is an obvious dependence of the Cu-S bond length on the different functionals, which cause it to vary from 2.25 A ˚to 3.34 A ˚. As in thiophene/Cu(111), PBE underestimates the interaction between the molecule and the surface, with a Cu-S bond length of 3.19 A ˚, while PBE-G06 overestimates this interaction and RPBE-G06 suppresses it again. The TS correction shows a slightly larger bond length than that of G06. For the vdW-DF series, vdW-DF2 gives the longest bond length (3.34 A ˚), and optB86b the shortest (2.41 A ˚), repeating the trend observed for the (111) surface. All of these methods show small or even negative tilting angles. In the IFS results, as the effective charge increases, the Cu-S bond length decreases from 3.38 A ˚(0.5 e) to 2.85 A ˚(1.0 e) and 2.63 A ˚(1.5 e), while the tilt angle increases. These results indicate that the dominant contribution to the molecule-surface interaction for the Top-site configuration is due to Cu-S and Cu-C covalent bonds, whereas for the Bridge-site configuration the Cu-p interaction dominates. It is somewhat surprising that all vdW-DF functionals favour the Top site. Recent scanning tunneling microscopy (STM) observations suggested that the Cu(100) surface is much more reactive than Cu(111) for phenyl-based molecules. One may speculate that during the XSW measurements 44 , x-ray irradiation and thermal excitation may result in a dissociative attachment of thiophene on Cu(100), leaving a C 4 carbon chain/ring adsorbed on the surface and a single S atom located at a Bridge site. Such dissociative attachment has been observed for a similar molecule (TB-TTF, comprised of phenyl-and vinyl-groups and S atoms) on Cu(100) 54 . Further experimental and theoretical efforts are required, especially a direct low-temperature STM observation of thiophene adsorbed on Cu(100). Adsorption energy. We calculated the adsorption energies of thiophene adsorbed on all three surfaces using the different functionals (other than PBE-CH-1e) and compared these with the available experimental data, as summarized in Table 2. The experimental adsorption energy for thiophene on Cu(100) is 20.63 eV 43 . The adsorption energy for the other two surfaces is unavailable, but it can be estimated from other experiments. The adsorption energy of benzene on Cu(111) 51,52 is 20% lower than that on Cu(110) 53 , as deduced from the temperature programmed desorption (TPD) temperatures of 225 K (111) and 280 K (110). The adsorption energy of a molecule on Cu(110) is usually very close to that of Cu(100), within a difference of 10%, as noted above. Thus one may infer that the adsorption energy of thiophene on Cu(111) is approximately 0.5 eV. Table 2 shows that the PBE functional, as expected, underestimates the adsorption energies for all surfaces and sites, especially for the (111) surface. Adding either the G06 or the TS dispersion correction to PBE, however, considerably overestimates this energy, but leads to relatively accurate energies when applied to RPBE. This result can be ascribed to a better error cancellation between the overly attractive dispersion correction and a more repulsive exchange-correlation functional. The adsorption energies of the (111), ( 110) and (100) surfaces revealed by RPBE-G06 are 20.46 eV, 20.57 eV and 20.59(20.52) eV (Table 2). By considering the correction to the zero-point energy, which is approximately 50 meV, the RPBE-TS values agree even better with the experiments. VdW-DF functionals also give very reasonable results. Although vdW-DF2 overestimates the Cu-S distance very strongly, the predicted adsorption energies are only some tens of meV smaller than the corresponding experimental values. The other three functionals slightly overestimate the adsorption energy, by values from 0.1 eV (optPBE-vdW) to 0.3 eV (optB86b-vdW). The sequence of energy values predicted by the four vdW-DF functionals is consistent with the Cu-S bond lengths. The DFT-D methods, such as RPBE-G06, were found to be very robust for calculating adsorption energies. Table 3 shows adsorption energies calculated by RPBE-G06 for structures relaxed by all the methods considered in this work. The energy varies within 0.09 eV, 0.10 eV and 0.23 eV respectively for the (111), ( 110) and (100) surfaces, and thus is quite insensitive to the structural differences. The columns ''RPBE'' in Table 3 list the DFT portion of the adsorption energy and columns ''G06'' the contribution from the G06 dispersion correction. The DFT portion is always positive because RPBE is such a repulsive functional, while the attractive dispersion correction opposes this repulsion to yield a very reasonable adsorption energy. It is remarkable that the repulsion and attraction are enhanced or weakened simultaneously for most structures: the energy changes from RPBE and G06 are very similar over a wide range of adsorption distances, resulting in a rather ''stable'' adsorption energy. In the case of thiophene, the molecule is predicted to bond to substrates more strongly than it physisorbs. There are some extreme cases where the molecule-substrate interaction is by physisorption, e.g. 6,13-pentacenequinone(P2O)/Ag(111) 55 , in which the molecule-substrate distance optimized by PBE is 4.0 A ˚, but the RPBE-G06 value is 0.9 A ˚shorter. Such a drastically shortened distance increases the adsorption energy only from 1.15 eV to 1.45 eV, indicating that the RPBE-G06 method gives reasonable adsorption energies for a wide range of molecule-substrate distances even for non-bonding interfaces. Electronic density of states. Experimentally measured electronic structures of metal-molecule interfaces are usually reproducible by standard DFT calculations, e.g. PTCDA/Ag(111) 6,29 . As discussed above, including dispersive contributions causes noticeable changes to the Cu-S bond length (d) and the molecular tilt angle (a), which can be expected to cause an appreciable variation of electronic structures. We have investigated this effect for both the Cu(111) and Cu(110) surfaces, because the largest difference of 0.8 A åmong the predicted Cu-S bond lengths was found on the (111) surface and the smallest, ,0.1 A ˚, on the (110) surface (Figure 2). Figure 2(a) shows that structural differences scarcely modify the appearance of the DOS for the occupied states of the (111) interface. The largest shift of those states in the energy range within 2.0 eV below E Fermi is only 0.2 eV. However, the situation for the unoccupied states is more complicated. The LUMO-substrate hybridized state (h-LUMO) moves from approximately 2.0 eV (PBE/optPBE-vdW) to 1.5 eV (PBE-G06), although the Cu-S bond length is shortened by only 0.2 A ˚. The shorter Cu-S bond lengths correspond to lower-energy states. In addition, a band broadening is observable in the shaded area of Figure 2(a), which is enhanced as the bond length decreases, suggesting a strengthening thiophene-Cu(111) interaction. Although the structural difference between the two energetically degenerate (flat and tilted) configurations revealed by methods including RPBE-G06 and optPBE-vdW is quite significant (Table 1), Figure 2(b) indicates that both configurations have nearly identical DOSs. They differ only in that the band broadening vanishes and the energy of state h-LUMO moves slightly higher in the flat configura- --20.63 The unit of energy is eV. Results are listed for the top site on ( 111) and ( 110) surfaces and for both top and bridge sites on the (100) surface. The DOS shows no appreciable charge transfer from the Cu surfaces to thiophene, a situation different from PTCDA and other polycyclic aromatic hydrocarbons, such as pentacene, on Cu, Ag and Au surfaces 31,38,50 . In these systems, the lowest unoccupied molecular orbital (LUMO) interacts with substrate states and a charge transfer occurs from the surface to the LUMO, causing a Fermi-Level pinning (the LUMO-surface bonding state is located just below the Fermi Level 6 ). In thiophene/Cu, the h-LUMO state is located well above the Fermi Level, at 1.9 eV and 1.2 eV respectively for the PBE results on Cu(111) and Cu(110). However, the S atom bonds to the Cu surface through covalent Cu-S bonding and the hybridized states originate from the molecular LUMO and Cu surface states. The position of the d-band centre is a well-established property which determines the reactivity of transition-metal surfaces 56 . The closer is the d-band centre to the Fermi Level, the more reactive the metal surface is expected to be. Table 4 summarizes the computed values of E d-band -E Fermi for the (111), ( 110) and (100) surfaces with three representative functionals, i.e. PBE, RPBE-G06 and optPBE-vdW. The positions predicted by these functionals are nearly identical for each surface, which indicates again that the valence band is insensitive to the choice of exchange functional in thiophene/Cu interfaces. The adsorption energies on different surfaces (considering the Top site for (100)) are, as expected, completely consistent with the position of the d-band centre. The only exception is the PBE result, where the (110) surface has a lower d-band centre and larger adsorption energy compared with the (100) surface, a result explicable by the generally poor performance of PBE in predicting adsorption energies. Charge density. when vdW effects are included, the DOSs of all the interfaces largely preserve their original features, which implies no changes to the covalent or vdW nature of thiophene-Cu bonding. Figure 3 shows the differential charge densities (DCD), defined as r DCD 5 r Thiophene/Cu -r Thiophene -r Cu , for PBE, PBE-G06 and optPBE-vdW (flat) calculations on the (111), ( 110) and (100) interfaces. Charge reductions (cold colors) were found near both Cu and S (C) atoms. These reduced charges accumulate in a volume between the S atom and the Cu atom beneath it (hot colors), indicating a typical Cu-S covalent bonding picture for both interfaces, regardless of the method used in the calculation. Even if the Cu-S bond length is as large as 3.28 A ˚, characteristic covalent bonding features remain observable in the flat configuration of the optPBE-vdW-relaxed (111) interface, shown in Figure 3(c). In --20.63 The unit of energy is in eV. Columns RPBE, G06 and RPBE-G06 refer to the DFT portion, dispersion correction and total value of the adsorption energy, respectively. addition to the Cu-S bonding, similar Cu-C covalent bonding is also illustrated for thiophene/Cu(110) in Figure 3(d)-(f). Although a vdW bonding picture for thiophene/Cu(110) was suggested by a vdW-DF calculation on the basis of adsorption energies 22 , ''covalent-like'' bonding can be found in the electronic structure of C 60 /Au(111) deduced by the same method 27 . The situation in thiophene/Cu(100) is similar to the other two interfaces, even though the discrepancy between theory and experiment over the adsorption site remains. In our work, optPBE-vdW explicitly shows covalent bonding, which implies that the other two vdW-DF methods, optB88-vdW and optB86b-vdW, should also suggest covalent bonding, because the Cu-S bond lengths they predict are shorter than in optPBE-vdW. We close the presentation of our results by discussing the effects of core-hole interactions on the charge distribution. The IFS implementation of these interactions succeeded in reproducing the adsorption structure of thiophene-Cu(111) and Figure 4 shows the screening charge induced by the core-hole. This is concentrated primarily around the C atoms, consistent with the fact that electron-core-holes are created on these atoms. The accumulated charges transfer to the LUMO and thus strengthen the Cu-S bonding, resulting in a shorter Cu-S bond length and a larger tilt angle. No significant differences are visible among the four configurations in Figure 4, and core-hole contributions do not resolve the puzzle of the adsorption site for thiophene/Cu(100). ## Discussion The thiophene-Cu interaction predicted by PBE is rather weak: the binding energy is underestimated significantly and the Cu-S bond length slightly overestimated due to the missing vdW correlation and dispersion correction. The inclusion of the G06 dispersion correction increases the binding energy too much, and the more repulsive RPBE functional is required to balance this correction in both the adsorption energy and the molecule-substrate separation. Dispersion corrections shorten this separation significantly in thiophene/Cu(111) and thiophene/Cu(100) at the Bridge site, where the molecule-substrate interaction is relatively weak, whereas for adsorption via the relatively strong Cu-S and Cu-C covalent bonds, as at the Top site in Cu(110) and Cu(100), the separation changes only slightly. The coefficients C 6 and R 0 in the TS correction were derived from GGA charge densities, which offer a better transferability, and thus it is recommended to use these if applicable. Our results, however, show that RPBE-TS overestimates slightly both the molecule-substrate separation and the adsorption energy. It has recently been shown that the inclusion of a repulsive many-body screening effect within TS results in a smaller adsorption energy 26 . We found that the choice of a proper exchange functional is crucial to the performance of vdW-DF. Our results show that the predicted adsorption energy varies inversely with the predicted molecule-substrate distance. The sequence of calculated adsorption energies is optPBE-vdW , optB88-vdW , optB86b-vdW for these three functionals sharing the same correlation functional, consistent with their behavior for small molecules and simple solids 24,35 . The vdW-DF2 method gives the weakest bonding strength. Among the four vdW-DF schemes, optPBE-vdW optimizes the adsorption energy and optB88-vdW the molecule-substrate distance compared with experiment; by contrast the best performance is achieved by optB88 for interaction energies of the S22 set 19 and by optB86b for lattice constants 20 . The poor performance of vdW-DF2 is not surprising because it was optimized to simulate molecules 18 . Our results suggest that it is still necessary to find an even ''better-match'' exchange functional -it should be as repulsive as optPBE, with the distance range where Pauli repulsion operates being similar to optB88. If thiophene bonds tightly to the Cu surface, for example through covalent bonds at the Top site of Cu(110) and Cu(100), the moleculesubstrate bonding is so strong that the bond length should not be affected significantly when the molecule is partially charged under xray irradiation. It is thus safe to compare the measured XSW results with the ground-state DFT results in these interfaces. Otherwise, for weaker interactions like the Cu-p interaction in thiophene/Cu(111) and thiophene/Cu(100) at the Bridge site, a small amount of transferred charge could drastically alter the molecular configuration. The unit of energy is eV. This is the reason why the configuration on the (111) surface is more sensitive to the consideration of IFS than that on the (100) surface at the Top site. The agreement between theory and experiment achieved by considering IFS for thiophene/Cu(111) extends our understanding of the relevance of core-hole interactions. It does not indicate that the DFT-D and/or vdW-DF methods employed here are inaccurate, but simply that the energy range of XSW measurements exceeds the reach of these ground-state methods. IFS is an appropriate means of extending these and also constitutes a probable solution for similar discrepancies observed on certain noble metal (111) surfaces 29, . These statements are, we believe, extendable to similar organic molecule-metal interfaces. However, IFS does not resolve the discrepancy regarding the adsorption site of thiophene/ Cu(100), which calls for further experiments including neutron scattering and scanning tunneling microscopy, and also for a higher-level theoretical analysis. Turning to the electronic structure of thiophene/Cu interfaces, the standard PBE functional reveals a covalent bonding picture in all cases on the basis of the differential charge density. This statement holds even if the predicted bond lengths and adsorption energies are quite different, demonstrating that these quantities may not be the only clues determining the bonding picture. The inclusion of dispersive contributions does not change the covalent picture to a vdW one, but the energies of certain unoccupied states and the band broadening depend strongly on the method used. Further measurements of electronic structures are required to refine the ''best'' theoretical basis for reproducing the DOS. All of our results shed considerable light on the physical and technical ingredients for accurate modelling of organic molecule-metal interfaces and are invaluable in refining the exchange functionals for DFT-D and vdW-DF methods. ## Methods DFT calculations. Calculations were performed using the general gradient approximation (GGA) for the exchange-correlation potential 37 , the projector augmented wave method 45,46 , and a plane-wave basis set as implemented in the Vienna ab-initio simulation package 47,48 . The energy cut-off for the plane-wave basis was set to 400 eV for all configurations examined. Five layers of Cu atoms, separated by a 20 A ˚vacuum region, were employed to model the Cu surfaces. Supercells of sizes p(3 3 3), p(2 3 3) and p(3 3 3) were adopted to investigate the adsorption of thiophene on Cu(100), Cu(110) and Cu(111) surfaces, respectively. We anchored the S atom of thiophene on the Cu substrates with respect to the underlying sites, i.e., the Top, Bridge and Hollow sites, and rotated the molecule by the allowed symmetry operations of the different Cu surfaces. As a result, seven, six and six initial configurations were considered respectively for Cu(100), Cu(110) and Cu(111). Molecules were placed on one side of the slab with a dipole correction applied. A 6 3 6 3 1 k-mesh was adopted to sample all two-dimensional (2D) surface Brillouin zones for both geometry optimization and total energy calculation, and the results verified with one calculation using an 8 3 8 3 1 mesh. In geometry optimizations, all atoms except those for the bottom two Cu layers were fully relaxed until the residual force per atom was less than 0.02 eV/A ˚. In the light of recent developments in vdW-DF techniques, which produce a significant reduction in their computational cost to the GGA level, both DFT-D and vdW-DF methods were employed in the structural relaxation and adsorption energy calculations. Dispersion corrections from the methods of Grimme (G06) 11 and Tkatchenko-Scheffler (TS) 13 were combined with two different GGA functionals, PBE 37 and RPBE 49 , denoted respectively as ''PBE-G06'', ''PBE-TS'', ''RPBE-G06'' and ''RPBE-TS''. Four combinations of exchange and non-local correlation functionals were considered at the vdW-DF level, namely optPBE-vdW 19 , optB88-vdW 19 , optB86b-vdW 20 and vdW-DF2 18 . Ionic final state approximation. The experimentally available interfacial geometry of organic molecule-metal interfaces, with which theoretical calculations can be compared, was measured using x-rays 29,30, 44 . A core hole is created when x-rays excite a core electron from an atom of thiophene. Nearby electrons are prone to transfer to the lowest unoccupied state of the excited molecule, screening the core hole. The transferred electrons begin to relax to lower states and eventually fill the excited core level, eliminating the core hole. It has been suggested from studies of PTCDA/Ag(111) and CuPcF 16 /Ag(111) 29, that the electron transfer process is much faster than the relaxation process of the transferred electrons. Thus if excitation events happen continuously and their separation is shorter than the relaxation time of the transferred electrons, some of these electrons are not fully relaxed. This results in a dynamic electron accumulation around the molecule, effectively making it become negatively charged. The geometry of a molecule in this effectively charged state, previously denoted IFS 31 , can be very different from its neutral ground state. An accurate description of such a dynamic process for these interfaces may require timedependent DFT, which is a highly computationally demanding technique. However, it has been demonstrated that the IFS approximation, a much cheaper method, gives a good reproduction of experimental results for PTCDA/Ag(111) and CuPcF 16 / Ag(111) when used with a proper effective charge 29, . We thus considered the structural variations of thiophene adsorbed on Cu surfaces with the IFS effect included. The term ''PBE-CH-ne'' is used to indicate an IFS state with a net effective transfer of n electrons to the C atoms of thiophene, in which n may be fractional.

chemsum

{"title": "Role of the dispersion force in modeling the interfacial properties of molecule-metal interfaces: adsorption of thiophene on copper surfaces", "journal": "Scientific Reports - Nature"}

visible_light_sensitizer-catalyzed_highly_selective_photo_oxidation_from_thioethers_into_sulfoxides_

## Abstract: We report herein a visible light sensitizer-catalyzed aerobic oxidation of thioethers, affording sulfoxides in good to excellent yields. The loading of the catalyst was as low as 0.1 mol%. The selectivity was excellent. Mechanism studies showed both singlet oxygen and superoxide radical anion were likely involved in this transformation.Sulfoxides are important fragments in organic synthesis 1-4 and biologically active molecules 5 , including commercialized medicines 6,7 and antiseptics 8 . Oxidation of thioethers into sulfoxides was the most straightforward pathway for the synthesis of sulfoxides [9][10][11] . Several methods in this field were developed during the past decades, including hydrogen peroxide oxidation [12][13][14] , metal complexes-catalyzed oxidation [15][16][17][18][19][20] , organocatalytic oxidation 21 , photo oxidation [22][23][24][25][26][27] , etc. However, stoichiometric external organic or inorganic oxidants were generally required in those reactions. Thus, a large amount of environmentally unfavorable wastes were generated during the production of sulfoxides. Another issue of those methods were the low selectivity between sulfoxides and over-oxidized by-product sulfones in many cases 14 . Although some catalytic system showed high selectivity, but the catalyst was too expensive to practical applications 15,17 . With the consideration of "Green Chemistry", an environmentally friendly, energy-saving, atom-economical, and highly selective oxidation from thioethers to sulfoxides is required.Visible light has attracted wide attentions with its clean and abundant advantages. Outstanding works by MacMillan et al. showed the utilizations of visible light in organic reactions 28 . Two typical pathways normally proceeded in visible light catalysis: electron transfer and energy transfer. Ru or Ir complexes [29][30][31][32][33] and some heteroatom-containing metal-free organic dyes [34][35][36] , which trend to grab or donate an electron in its excited state, are often used as the electron transfer catalyst. On the other hand, some rigid and conjugated organic compounds [37][38][39] , which can absorb visible light photon but are not capable of grabbing or donating electrons, could be used as energy transfer catalyst. Selective oxidations of thioethers into sulfoxides catalyzed by visible light photo catalysts have been widely studied [24][25][26][27] . Those reactions could also be classified to electron transfer process and energy transfer process as mentioned above (Fig. 1). In electron transfer process, superoxide ion (O 2 − ) was the key oxidative intermediate 27 . Recently, Chao and Zhao reported a visible light-induced photo oxidation of thioethers using a dinuclear Ru-Cu complex as the catalyst 23 . While in energy transfer process, oxygen was directly excited to its singlet state ( 1 O 2 ) which served as the predominant oxidative species 25 . Notably, Vitamin B 2 Derivative could achieve this reaction via both electron transfer process and energy transfer process 22 . Although visible light-induced selective oxidations of thioethers into sulfoxides under aerobic conditions have been reported, these reactions normally required expensive photo-catalysts with relatively high loading. Reactions with higher efficiency and lower cost were still required. Based on our continuous interest in photo oxidation reactions [40][41][42] , we decided to investigate whether thioxanthone derivatives would be an effective energy transfer catalyst in thioethers oxidation. Herein, we wish to report our recent results on visible light sensitizer-catalyzed aerobically selective oxidation of thioethers into sulfoxides. ## Results and Discussion Optimization and scope investigation. In the beginning, methyl phenyl thioether (2a) was chosen as the model substrate. The initial attempt was conducted under oxygen atmosphere at rt using 5 mol% of thioxanthone (1a) as the catalyst, toluene as the solvent and purple LED as the light source. A 7% NMR yield of methyl phenyl 1). The low recovery was mainly caused by the loss during the process of rotary evaporation, since the boiling point of 2a was low. Encouraged by the initial result, a screening of solvents was carried out. Only trace amount of 3a was afforded when THF or CH 3 NO 2 was used as the solvent (entries 2 and 3, Table 1). When cyclohexane, CH 2 Cl 2 , ethyl acetate (EA) or acetone was tested, the yield of 3a was slightly increased (entries 4-7, Table 1). A dramatic improvement of the yield was observed using CH 3 CN as the solvent (entry 8, Table 1). But on the other hand, over oxidized product, methyl phenyl sulfone (4a), was also generated in a 4% NMR yield. The reaction in CH 3 OH gave an excellent yield of 3a with increased selectivity of 3a/4a (entry 9, Table 1). Thus, CH 3 OH was chosen as the best solvent. Next, modifications of thioxanthone derivatives were conducted aiming at promoting efficiency and selectivity. Thioxanthone derivatives were synthesized by the coupling of iodine compound with thiosalicylic acid followed by Friedel-Crafts reaction 43 . Reaction employing 2-chloro-thioxanthone (1b) showed better selectivity but lower yield (entry 10, Table 1). When 4-phenyl-thioxanthone (1c) was used as the catalyst, the reaction gave a 99% NMR yield of 3a with the ratio of 3a:4a being more than 99:1 (entry 11, Table 1). Then methoxy group was attached at 2-position of thioxanthone, but showed lower efficiency than 1c (entry 12, Table 1). 1,4-Dihydroxy-thioxanthone (1e) was proved to be unfeasible in this transformation probably due to its low solubility (entry 13, Table 1). Thus, 1c was chosen as the best catalyst. To our delight, decreasing the amount of 1c till 0.1 mol% still gave excellent yield and selectivity (entries 14 and 15, Table 1). Further reducing the amount of 1c to 0.01 mol% led to a sharply decreased yield (entry 16, Table 1). Finally, a series of control experiments were carried out indicating both catalyst and light was necessary for this reaction (entries 17 and 18, Table 1). Furthermore, considering the thermal effect caused by the purple LED light, the reaction was carried out at 50 °C without light. The result validated that no reaction took place at all even heated (entry 19, Table 1). Thus, Condition A (0.1 mol% of 1c, CH 3 OH, purple LED, air atmosphere, and rt) was chosen as the optimized condition for further studies. With the optimized reaction condition in hand, the scope of this oxidation was examined carefully. Some typical results are summarized in Fig. 2. Firstly, the electron effect of the aryl group in methyl aryl thioether was studied (3a-k). Excellent to good yields were obtained for methyl o-, m-or p-methoxyphenyl thioether (3b-d). Methyl 4-methylphenyl sulfoxide (3e) was formed in good yield from the corresponding reactant. In cases of substrates with halogen atom, excellent yields were obtained (3f-h). Substrates with strong electron withdrawing groups, like formyl (3i), methoxy carbonyl (3j), and nitrile (3k), were also tolerant in this reaction. When naphthyl ring was used instead of phenyl ring, a 94% isolated yield of methyl 2-naphthyl sulfoxide (3l) was generated. Secondly, we focused on the influence of alkyl group. Ethyl (3m) or cyclopropyl (3n) were applied instead of methyl. The corresponding yields were nice. Thirdly, diaryl thioether was also tolerant in this reaction, giving the corresponding sulfoxide (3o) in excellent yield and selectivity. Finally, aliphatic thioethers were examined. Di-n-butyl thioether led to an excellent yield of 3p, while tetrahydro-2H-thiopyran and tetrahydrothiophene resulted in slightly lower yields of 3q and 3r, respectively. Scale-up reaction was also conducted using 2o (Fig. 3) under Condition A. 95% of 3o was afforded. This result showed the potential in organic synthesis. ## Mechanism studies. To gain insight into the reaction mechanism, singlet oxygen quencher 1,4-diazabicyclo[2.2.2]octane (4) 44 and superoxide radical anion quencher N-tert-Butyl-1-phenylmethanimine oxide (5) 45,46 were added into the reaction system (Fig. 4), respectively. Severe inhibitions were observed in both cases. These results clearly indicated that both singlet oxygen and superoxide radical anion were likely involved in this transformation. Based on the experiment results above and literature precedents 24,26,27 , a possible mechanism was proposed as shown in Fig. 5. 1c was excited upon the visible light irradiation and then sensitized oxygen to its singlet state 24 which is more oxidative than normal triplet oxygen. Singlet oxygen could grab one electron from the lone pair electron of thioether 2, forming thioether radical cation 6 and superoxide radical anion 27 . Then 6 could react with superoxide radical anion to give intermediate 7 27 . 7 and another molecule of 2 further furnished 3 as the final product 27 . ## Conclusions In conclusion, we developed a visible light sensitizer-catalyzed highly selective oxidation from thioethers into sulfoxides under aerobic condition. This reaction employed visible light as limitless energy source and 4-phenyl-9H-thioxanthen-9-one (1c) as metal-free catalyst with the loading as low as 0.1 mol%. This reaction showed high efficiency and selectivity with broad functional group tolerance. Gram-scale reaction could also be achieved under optimized conditions in nice yield and excellent selectivity. Mechanism studies indicated that both singlet oxygen and superoxide radical anion were likely involved in this transformation via energy transfer between visible light sensitizer and oxygen. Further applications of this reaction are in progress in our group. ## Synthesis of methyl phenyl sulfoxide (3a). A solution of 1c (10 mg, 0.03 mmol) in CH 3 OH (100 mL) was prepared prior to use. 2a (124 mg, 1.0 mmol), 1c (3 mL, 0.1 mg/mL, 0.001 mmol), and CH 3 OH (2 mL) were added to a schlenk bottle which was equipped with a magnetic stirrer. The mixture was irradiated by a purple LED at rt under air atmosphere. The photoreaction was completed after 5 hours as monitored by TLC (eluent: petroleum ether/ethyl acetate = 10/1). The solvent was removed and the residue was purified by flash column chromatography on silica gel (eluent: petroleum ether→petroleum ether/ethyl acetate = 20/1→10/1→1/1) to afford 3a 18 as a solid (130 mg, 93%); 1 H NMR (400 MHz, CDCl 3 ) δ 7.68-7.63 (m, 2 H), 7.57-7.47 (m, 3 H), 2.72 (s, 3 H).

chemsum

{"title": "Visible light sensitizer-catalyzed highly selective photo oxidation from thioethers into sulfoxides under aerobic condition", "journal": "Scientific Reports - Nature"}

fabrication_of_multilayered_composite_nanofibers_using_continuous_chaotic_printing_and_electrospinni

## Abstract: Multi-material and multilayered micro-and nanostructures are prominently featured in nature and engineering, and are recognized by their remarkable properties. Unfortunately, the fabrication of micro-and nanostructured materials through conventional processes is challenging and costly. Herein, we introduce a high-throughput, continuous, and versatile strategy for the fabrication of polymer fibers with complex multilayered nanostructures.Chaotic electrospinning (ChE) is based on the coupling of continuous chaotic printing (CPP) and electrospinning. CCP produces fibers with internal multi-material microstructure. When a 2 CCP printhead is used as an electrospinning nozzle, the diameter of the fibers is further scaled down by three orders of magnitude while preserving their internal structure. ChE enables the utilization of various polymer inks for the creation of nanofibers with a customizable number of internal nano-layers. Our results showcase the versatility and tunability of ChE to fabricate multilayered structures at the nanoscale at high throughput. We apply ChE to the synthesis of unique carbon textile electrodes composed of nanofibers with striations carved into their surface at regular intervals. These striated carbon electrodes with a high surface area exhibited a 10fold increase in specific capacitance compared to regular carbon nanofibers; ChE holds great promise for the cost-effective fabrication of electrodes with supercapacitance. ## Introduction Layered micro-and nanostructures are prominently featured in nature. While varying in color, shape, size, and composition, they share a high degree of interface in one, two or three dimensions. These multi-scale and multi-material structures may provide remarkable properties and functionalities (i.e., support, structure, orientation, mass transfer, and interfacial connection) that are crucial for many applications. 5,6] These architectures have inspired countless research contributions dedicated to mimicking their unique structure and composition to harness their highly efficient and exceptional electrical, chemical, and structural properties. In the field of MEMS and nanotechnology, a growing interest exists in fabricating these multifunctional constructs for their application in optical imaging, catalysis, drug delivery, sensing, micro-motors, and water purification. Among the various nanofabrication technologies, electrohydrodynamic (EHD) jetting techniques (and electrospinning in particular) stands out due to their versatility and scalability. By altering the modality of the techniques, the jetting process can rapidly produce nanoparticulate coatings or nanofibrous mats in a simple and scalable fabrication system. With appropriate post treatment processes, the material composition of the formed nanostructures can be adjusted, allowing the synthesis of materials ranging from various polymers to metals, ceramics, and carbons. In addition to its versatility to create nanostructures of varying morphology and material composition, EHD jetting has the capability to fabricate multilayered or multiphase nanostructures through the exploitation of physical and chemical behaviors of its precursor solutions. However, current multilayer EHD jetting processes are mainly limited to techniques exploiting phase separation, such as water-oil emulsions, or chemical separation (such as that seen in block co-polymers). Herein, we describe a novel and versatile EHD jetting-based methodology for fabricating multilayered nanostructures in a simple, rapid, and continuous process. Chaotic Electrospinning (CheE) is an innovative fabrication technique that combines electrospinning with continuous chaotic printing (CCP). CCP is a recently developed technology that uses chaotic static mixers as printing-nozzles to produce micron-sized filaments with internal multilayered structures by extrusion. This technology is based on the deterministic chaotic advection of flows that allows the exponential creation of a micro-layered structure inside a static mixer in a controlled and predictable manner. Electrospinning, meanwhile, enables the transitioning from the micro-into the nanoscale domain. This transition from the micro-(CCP) to the nano-domain (electrospinning) challenges the preservation of the multilayered structure achieved by CCP within the extruded filament due to the drastic changes in velocity when the jet is formed. In this regard, we studied the role of rheological properties of the polymer combinations used in the preservation of the internal multilayered structure and found suitable conditions for fabricating electrospun fibers with multilayered nanostructure. We demonstrated the use of ChE to create carbon nanofibers (CNFs) with multilayered parallel carved patterns. These CNFs were fabricated by utilizing two polymers, one amenable of being carbonized through pyrolysis and a sacrificial one that is precisely removed during pyrolysis. The resulting carbon is a textile composed of striated carbon nanofibers (SCNFs), greatly enhancing its surface area. The augmentation in surface area provided by the unique internal structures of these CNFs, boosted their performance as electrodes for energy storage applications, such as supercapacitance. In this regard, the results presented here suggest that ChE is an effective step towards circumventing current limitations for synthesizing multiscale nanostructures and provides the necessary versatility, flexibility, and high yielding capability that is needed in many engineering applications in a cost-effective manner. ## Results and Discussion ChE combines CCP technology and far field electrospinning into a versatile and scalable nanofabrication technique for synthesizing nanofibers with internal multilayered structures. Initially, alternated micro-layers of two flowing materials are formed inside a chaotic static mixer and are then extruded through the spinneret, where the electrospinning jetting stretches the chaotic structure into the nano domain (Figure 1). Although EHD co-jetting processes for the synthesis of multi-phase structures have been widely explored, these techniques often employ chemical-or phase-based separation methods such as oil-water emulsions or block-copolymers. Alternatively, ChE achieves defined layered architectures by splitting and folding fluid streams in the laminar regime (chaotic advection). The polymer solutions remain segregated because the laminar regime is preserved both during the CCP and the electrospinning process. Thus, our fluid-based fabrication method opens a new window of unprecedented possibilities not achievable today by current chemical-or phase-separation processes. ## ChE setup We illustrate the concept and setup of our ChE system in Figure 1A. The actual setup is shown in Figure 1B. The process comprehends two main stages: CCP and electrospinning. The CCP occurs within a static mixer where chaotic advection of two different flowing polymers creates intercalated layered structures (Figure 1Ai). The static mixer used here, a Kenics static mixer (KSM), contains a sequential set of helicoidal mixing elements (Figure 1Aii) placed at 90º one with respect to other, that split and reorient the flowing materials, creating layers in an exponential fashion along the tube (Figure 1Aiii). This KSM was custom 3D-printed as a singlepiece unit containing a cap with 2 inlets, a cylindrical pipe that ends in a conical dispensing tip, and 3 KSM-elements inside the pipe. This particular setup produces filaments with 8 internal and intercalated layers of the two fed flowing materials. For the second stage, electrospinning, copper tape was wrapped around the tip of the conical dispenser. The entire CCP-spinneret system is held 10 cm above a rotating drum spinning at 5 revolutions per second (rps) that acts as the nanofiber collector. In operation, the two polymer solutions are dispensed through silicone tubing into the KSM downstream at a constant rate (0.2 mL/h) controlled by a syringe pump. A high-voltage power source drives the electric field between the drum and spinneret. The electric field stretches the polymer meniscus at the tip of the spinneret, forming the Taylor cone. The electrospinning jet emits from the Taylor cone (Figure 1Aiv) jetting the polymer solution into a continuous nanofiber as it travels to the collector and forming a nanofibrous textile (Figure 1C) with inner-microlayered structure (Figure 1D). We conducted an extensive set of preliminary ChE experiments, using different combinations of polymers (Table S1), and identified several key requirements that must be satisfied to assure the successful fabrication of multilayered nanostructured fibers. In this regard, an optimal polymer combination must enable the laminar coextrusion in the CCP system (i.e., polymers must exhibit interfacial and rheological compatibility), be electrospinnable (i.e., at least one polymer must be sufficiently conductive), and offer the proper balance between elastic and viscous forces for the system to create coherent nanofibers with internal microlayered structure. ## Nanolayered fibers fabrication and PiFM characterization Next, we demonstrate the fabrication of polymer fibers that exhibit internal layered structures at the nanoscale using two combinations of aqueous-polymer-solutions, (a) 7% poly ethylene oxide and 4% sodium alginate (PEO/SA), and (b) 12% polyvinyl alcohol and 4 % sodium alginate (PVA/SA). For the PVA/SA blend, sodium dodecyl sulfate (SDS) was added to PVA solution to increase the conductivity of the PVA/SA polymer and facilitate electrospinning. From all the polymer combinations that we analyzed (Table S1), these exhibited the best working properties to stably produce flawless (i.e., bead-less and continuous) electrospun nanofibers by ChE. The electric field was determined by the lowest working voltage that would enable a stable and continuous EHD jetting. For the PEO/SA combination, the working voltage was 10 kV, while 12 kV was used for the PVA/SA combination. Atomic Force Microscopy (AFM) was used to characterize the topography of our electrospun textiles, while Photoinduced Force Microscopy (PiFM) was employed as the primary characterization tool to visualize the alternating layers of different composition within them. PiFM is a noninvasive, highly sensitive, and versatile technique that allows nanoscale chemical imaging with 10 nm spatial resolution. A mechanical cantilever's tip interacts with the sample's surface through local light-matter interaction. The wavelength of light excitation is tuned to one that corresponds to the vibrational resonances of the material under study. The technique has been proven to be useful for the analysis of heterogeneous samples such as block copolymer samples, showing high signal-to-noise ratio. For this work, PiFM was used to identify internal layers formed within each nanofiber by differentiating between the PEO or PVA and SA layers through differences in their spectral signatures. The micrographs obtained from the AFM and PiFM analysis of the nanofibrous textiles are shown in Figure 2. Figure 2Ai and Bi show the topographical map of each nanofiber as determined by the AFM. Figure 2Aii and Bii show the juxtaposition of the AFM topography and the composition map measured by the PiFM. SA layers appear in gold ( 500-750 V), and PEO or PVA ( 0V) in blue. We were able to clearly differentiate the multiple layers of the PEO/SA and PVA/SA fibers. These results demonstrate the feasibility of successfully transitioning the CPP technique into the nano-domain. Figure 2Aiii and 2Biii show the intensity profiles along the lines indicated in AFM and PiFM micrographs. This analysis evidences that ChE was able to produce flawless fibers and preserve the multilayered internal structure developed during the CCP stage. However, not all combinations of polymer blends tested were successful. Combinations such as 7% PEO/3% SA were incapable of forming nanofibers (Figure 2 Ci), while others, such as 7% PEO/5% SA successfully form nanofibers, but no inner structure was observed (Figure 2Cii). Next, we discuss how the properties polymer solutions used in ChE greatly dictate the success of this manufacturing method. PiFM analysis of i) 7% PEO/3% SA and, ii) 7% PEO/5% SA nanofibers. D) Rheology characterization of the polymer solution tested in the ChE system. ## Influence of polymer properties on generating structured nanofibers Several conditions need to be satisfied in order to successfully produce multilayeredstructured nanofibers using ChE. Both stages of the process, CCP and electrospinning, demand the use of a polymer blend with particular properties to be successfully processed into nanofibers with internal layered architecture. For CCP, the combination of polymer solutions should be compatible in terms of interfacial tension and polarity. Additionally, both flowing materials should behave as Newtonian liquids within the window of shear rates that they experience when extruded along the mixer. This compatibility and Newtonian behavior will allow the polymer solutions to flow stably side-to-side (without mixing), creating defined layers during CCP. For the electrospinning stage, the blend must exhibit adequate ionic conductivity and viscoelastic profile to establish and sustain the electrospinning jet and produce coherent nanofibers steadily. Moreover, ChE involves additional challenges. The polymer blend is subject to a substantial change of velocity (therefore, shear-stress values) when transitioning from the micro (CCP) to the nano (electrospinning) domain. The electrospinning solution undergoes an extreme confinement effect at the Taylor cone, where the polymer meniscus at the tip of the spinneret transitions into the electrospinning jet. This transition enables a 3-order of magnitude change in the surface-to-volume ratio of the polymer solution over less than a few microns. In order to preserve the internal layered structure even when the blend is rapidly pulled from the Taylor cone, the polymers should keep flowing as Newtonian liquids within the laminar regime. Considering these requirements, we used water-polymer solutions to comply with the interfacial and polar compatibility that CCP demands. For electrospinning, we used polymer pairs previously reported as electrospinnable elsewhere In this particular case, PVA and PEO were used as carriers to confer electrospinnability to the blend, since SA itself is not electrospinnable. [6, We observed that, among all the tested polymer blends, only blends containing 4% SA were successful in producing coherent electrospun nanofibers with internal layered structure as confirmed by the PiFM analysis. Aiming to understand the mechanism behind this finding, we individually characterized the complex viscosity of all our polymer blends using a rotational, cone-plate rheometer at a wide range of shear rates values (6.28 -126 1/s) (Figure 2D). Among all polymer solutions, 4% SA was the only one showing a steady Newtonian behavior (invariable viscosity) in the range of shear-stress values tested (Figure 2D). These results suggest that 4% SA acts as a "stabilizer" of the flow and allows the blend to keep flowing in the laminar regime throughout the ChE process. Even when PVA or PEO behave as non-Newtonian materials at different shear rates, our results suggest that the presence of 4% SA provides sufficient stability to the polymer blend to preserve the laminar regime. Rheological characterization also suggests that the SA concentration has an important effect in both, electrospinability and preservation of the internal microlayered structure. When using 3% SA solutions, the polymer blend did not exhibit enough strength (viscosity) to produce coherent nanofibers, instead, the material was sprayed from the KSM to the rotating drum creating a mat formed of beads (Figure 2Ci). When using 5% SA solutions, nanofibers were successfully obtained, but no internal layered structure was evident in our PiFM analysis (Figure 2Cii). The viscosity profile of 5% SA solution showed a non-Newtonian behavior in the range of shear stress values tested (Figure 2D). The increased concentration of SA (as compared to 4% SA) may have added elastic forces that counteract when pulled from the Taylor cone, generating micro-eddies and destroying the structure formed within the KSM. The 5% SA solutions, in combination with the PEO or PVA solutions, may have rendered a viscoelastic blend incapable of sustaining the flow laminarity in the CCPelectrospinning transition, therefore non-structured (well-mixed) nanofibers were obtained. Closer examination of Figure 2Aii and 2Bii shows that the PVA/SA polymer combination exhibited more defined layers. The higher viscosity of 12% PVA as compared to that of 7% PEO solutions (Figure 2C) may have contributed to the formation of more coherent nanofibers and reduction in mixing with 4% SA, impacting favorably to the preservation of the internal structure. The rheological characterization for all the polymer solutions tested is presented in ## Figure S1. Based on these observations, a basic framework can be made for ChE. For a twopolymer-based ChE system, the preservation of the chaotic internal structures requires a polymer blend which is: 1) compatible in terms of interfacial tension and polarity, 2) electrospinnable into coherent nanofibers, and 3) Newtonian within a wide range of shear rates of the whole ChE process. According to the aforementioned criteria, ChE has the potential to be one of the most versatile nanofabrication methods for producing such multilayered structures. ## Supercapacitor fabrication and characterization To demonstrate the applicability of the ChE technique, we exploited the carbonizable properties of PVA to produce CNFs via pyrolysis, which allows the formation of internal layered cavities by the decomposition of the SA layers of the nanofiber. These as-fabricated SCNFs, with their uniquely etched cavities, feature higher surface area than typical CNFs, thus making them ideal for energy storage applications. SCNFs are electrochemically characterized and compared to CNFs to explore their use for such an application, as a supercapacitor electrode. The primary function of supercapacitors as energy storage devices, is to provide highpower density for rapid charging and discharging in many electrical/electronic-based devices. This function is achieved through the way supercapacitors store their charges in what is known as the electric double layer. The electric double layer is a nanometers-thick sheet of ionic charge that forms along the interface between the electrode and electrolyte in a supercapacitor cell. Due to the short distance of travel, the electric double layer can provide rapid dissipation and accumulation of charges. Therefore, the most direct way to increase the performance of supercapacitors is to increase this interfacial surface area, while simultaneously reducing superfluous volumetric density. In this regard, the extra surface area created through the pyrolysis of the chaotic PVA/SA nanofibers, should provide a sharp improvement in supercapacitance over conventionally electrospun CNFs. The SCNFs were produced from the ChE 12% PVA/ 4% SA polymer nanofibers with a KSM printhead with 3 mixing elements (Figure 3A). The PVA/SA and pure PVA nanofiber mats were incubated in iodine rich environment at 80 º C for 24 h to promote the stabilization of PVA's carbon backbone into polyene (Figure 3B). This step has been demonstrated in prior studies to be necessary for providing enough carbon-carbon bonds, and to prevent the disintegration of PVA during pyrolysis. Finally, the fibers are carbonized in a tube furnace under continue ultra-pure, nitrogen-rich flow at 1000 ºC (Figure 3C). Electrical contacts are made with the carbon electrodes by attaching a copper wire to its surface via a conductive silver ink, and the entire contact area is then encased in a silicone rubber to prevent corrosion (Figure 3D). A SEM comparison of the morphology between SCNFs and CNFs (Figure 3E and F), shows clear cavities in the center of the fiber structures in the SCNFs, which are not present in the CNFs samples. Half-cell supercapacitor characterization was conducted using a 3-electrode electrochemical setup. Figure 3G shows the results of the electrochemical characterization of the SCNFs and CNF electrodes. The specific capacitance can be extracted from the cyclic voltammetry (CV) (Figure 3G) with the following the Equation 1: The specific capacitance (𝐶 𝑝 ) is equal the integrated area under the cyclic voltammogram's (CVs) curve (A) divided by the product of the mass of the carbon electrodes (m), the scan rate (v) and the potential window of the CV sweep (∆𝑉). From the analysis of the CVs shown in Figure 3G, the SCNF electrodes exhibit more than double the amount of measured capacitance than the CNF. Interestingly, the small peaks visible in the CVs of both SCNF and CNF electrodes reveal the presence of faradic reactions from pseudocapacitance. Pseudocapacitance can be present in pyrolytic carbons due to retained functional groups from the carbon precursor, such as the alcohol groups in PVA. The additional faradaic reactions enhance the measured capacitance beyond that normally seen from purely electrostatic interactions, as in the case of the double layer capacitance. In order to decouple the enhancement in capacitance between psuedocapacitance and that of the electric double layer, an Electrochemical Impedance Spectroscopy (EIS) was performed (Figure 3H). By fitting the EIS plots to an equivalent circuit (Figure S2) model, the value of the double layer capacitance can be extrapolated. Once the pseudocapacitance is factored out, a greater difference in the capacitance is observed between SCNFs and CNFs, where SCNFs exhibited over 10-fold greater electrostatic capacitance than CNFs, reflecting the enhancement in surface area created by the combination of ChE and pyrolysis. In comparison with other electrospun carbon nanofiber studies, SCNFs exhibit similar specific capacitance despite no additional activation or functionalization steps after pyrolysis. These results demonstrate the great potential of this nanofabrication technology, with capability for greater enhancements in capacitance with additional metal oxides, activation and functionalization steps, and KSM elements during the ChE process. ## Conclusion In this work, we describe a new, versatile, and scalable nanofabrication process that combines CCP technology and electrospinning to produce multilayered nanofibers. These nanostructures are created by feeding two polymer streams into a customized KSM, where chaotic advection creates intercalated layered structures at an exponential rate. A large electric field is then applied between the extruding tip of the KSM and the collector, a rotating drum. The electric field induces the formation of nanofibers from the chaotic extruder outlet, which are then deposited continuously on the collector and forming a textile-like mat. SEM and PiFM techniques were used to demonstrate ChE proficiency in manufacturing multilayered nanofibers in a facile and cost-effective manner. In this study, we used PVA, PEO, and SA, which are all polymers soluble in water with interfacial compatibility that allow extrusion and structure generation by CCP. SA solutions are not themselves electrospinnable, but the addition of PVA (enriched with SDS) or PEO as carriers facilitated the electrojetting process. Among the blends of polymers tested, 7% PEO/4% SA and 12% PVA/4% SA were the ones that rendered the aimed layered architecture inside the nanofibers. We attribute this mainly to the steady Newtonian behavior of 4% SA observed in a wide range of shear stress values evaluated by rheological analysis. Consistently, this was the only polymer solution tested that exhibited Newtonian behavior throughout the whole range of shear stress evaluated. SA solutions at 4% acted as stabilizers of the flow, preserving the laminar regime of the blend through the whole ChE process, even when transitioning from the micro-domain (CCP outlet) to the nano-domain (electrospinning) at the Taylor cone. Furthermore, our results suggest that viscosity should be appropriately tuned in order to facilitate coherent nanofibers through the electrospinning process and preserve the laminar flow, therefore the internal microarchitecture. Further analysis is necessary to better understand the set of phenomena underlying ChE and their interplay. To demonstrate a potential application of the ChE, the PVA/SA chaotic nanofibers were pyrolyzed to create carbon fibers with striated cavities etched onto their surface. These striated carbon nanofibers (SCNFs) exhibited large surface area, which made them potential candidates for energy storage devices, such as supercapacitors. SCNF electrodes were then electrochemically characterized for their specific capacitance and compared with carbon nanofibers made via traditional electrospinning with only 8% PVA. Results show a 10-fold enhancement in the specific capacitance of the carbon nanofiber's electrodes obtained through ChE over other carbon nanofibers electrodes made via conventional electrospinning. This enhancement originates primarily from the increased surface area achieved through the formation of the cavities in each nanofiber, a feature that can be further enhanced increasing the number of KSM elements used. The improvement of the superconductive character obtained through this new synthesis technique, as well as the relatively mild conditions required for its operation, suggest the high potential of ChE as a cost-effective and scalable fabrication method for supercapacitors. ## Experimental Section/Methods Solution preparation: The solutions used in the ChE experiments were prepared separately by dissolving the polymers (wt. %) in deionized water. Polyvinyl alcohol (PVA) (MW 89,000 g/mol) was purchased from Sigma Aldrich (Toluca, Mexico) and dissolved in deionized water (8, 10, and 12 wt. %), both with and without the addition of (5 wt.%) sodium dodecyl sulfate (SDS) (from Sigma Aldrich, Toluca, Mexico). Sodium Alginate (SA), purchased from Sigma Aldrich, Toluca, Mexico, was mixed (at 2, 3, 4, and 5 wt.%) in deionized water. Polyethylene Oxide (PEO) (MW 300,000 g/mol) was mixed (at 7 wt.%) in deionized water. Che setup and parameters: ChE was performed by housing the prepared solutions, either PVA and SA or PEO and SA, in (5 mL) syringe barrels and fastened to the Chemyx's Fusion 200, 2channel syringe pump, which controls the flow of the solutions into the KSM element (Figure 1A). The KSM element was custom designed fabricated with a stereolithographic 3-D printer and fastened 10 cm above a custom-made rotating drum collector. The two solutions are fed to the KSM element via silicone tubing and copper tape is attached to the tip of the KSM element to create an electrode. The electric field, responsible for the formation of the fibers, is created by a Spellman CZE1000R voltage supply connected between the copper spinneret and the collector. The appropriate applied voltage was determined by finding the minimum voltage at which the Taylor cone is formed, which has a small variation due to difference in concentration of the polymer solutions (10-12 kV for PEO/SA, 11-13 kV for PVA/SA with SDS, and 18-19kV for PVA/SA without SDS). The rotating drum collector is grounded and set to spin at 5 rps to form an even rectangular textile mat. Scanning Electron Microscopy (SEM) characterization: SEM micrographs were collected with the EVO MA25 (Zeiss, Germany). For the polymer nanofibers, it was necessary to first sputter a thin coating (5 nm) of gold to eliminate distortions from charging effect. The same coating was not needed for the carbon nanofibers. Photo-induced Force Microscopy (PiFM) Characterization: PiFM is a relatively new nanomaterial characterization tool that enables nanoscale chemical imaging with down to 10 nm spatial resolution. Here, a mechanical cantilever with a tip that interacts with sample surface is exploited to probe local light-matter interaction. Using noncontact mode, the cantilever is excited at its 2nd mechanical eigenmode and stabilized nanometers from sample surface. Then, the tip-sample junction is illuminated with focused light generated by quantum cascade (QCL) with modulation frequency fm= f2 -f1. Because tip-sample intermolecular interaction is nonlinear with distance, fm mixes with f2, and beating signals at the sum (higher side band) and the difference (lower side band) are generated. In our set up, the lower side band is tuned such that it overlaps with the 1st mechanical eigenmode, where the PiFM signal is measured. The wavelength of light excitation is tuned to one of the vibrational resonances that corresponds to polymer nanofiber mats. Upon absorbing light, sample experience an increase in temperature leading to thermal stresses. Since the laser is modulated at fm, thermal expansion is also modulated at fm leading to a measurable signal at f1. In this manner, PiFM is used to probe nano-striations on micron-sized fiber. For the PEO/SA nanofibers, an excitation band for PEO at wavelength of 1100 cm -1 was used for the greatest contrast between the PEO and SA layers. In contrast, for the PVA/SA nanofibers the excitation band of PVA at 1449 cm -1 demonstrated the greatest contrast between the different material layers. ## Rheological Characterization: The rotational and oscillatory rheological parameters were measured for each polymer solution of varying concentration separately with the MCR-101 Rheometer (Anton Paar, Austria) and cone measuring plate (CP50-1-SN28724, cone radius 50.5 mm, cone angle 1 o , sample volume 1.5 mL) at 25.0 ± 0.1 o C ( Figure 3A). The viscosity was calculated with η = τ/γ, where τ is the shear stress and γ is the shear rate. During the rotational test, the shear rate was ranged from 6 to 130 s -1 . The oscillatory shear measurements were performed with an amplitude within the linear region (amplitude 10%) of the tested frequency range. Carbon Nanofiber Synthesis: Synthesis of the carbon and striated carbon nanofibers begins with the formation pure PVA nanofibers (8 wt.% PVA and 5 wt.% SDS in deionized water) via traditional electrospinning (single syringe needle, 10kV applied voltage and 10 cm working distance) and PVA/SA nanofibers (12 wt.% PVA and 5 wt.% SDS in deionized water and 4 wt.% SA in deionized water) via ChE (13 kV applied voltage at 10 cm working distance), respectively. The resultant polymer nanofiber mats were then incubated for 24 h in a 20 mL flask with 200 mg of iodine at 100 o C to induce the crosslinking of the PVA chains into polyene (Figure 3B), which strengthens the carbon chains so that the fibers can survive the pyrolysis process. The crosslinked PVA and PVA/SA mats are then placed in a OTF 1200X tube furnace under a 40 ml/min continuous nitrogen flow, which is then heated to 1000 o C at 2 o C/min and held at 1000 o C for 1 hour before letting it cool to room temperature through natural convection. Electrochemical Characterization: To electrochemically characterize carbon nanofiber, the carbon nanofibers were first cut into small 1 cm x 0.25 cm rectangular pieces and then wired to a copper wire with assistance of silver paint. Finally, polydimethylsiloxane (PDMS) is used to cover the painted area to prevent contamination of the electrochemical signal with corrosion as shown in Figure 3D. A traditional 3 electrode setup with a platinum mesh counter electrode, a Ag/AgCl reference electrode, and the wired carbon nanofiber working electrode, all held in a 50 ml Pyrex beaker. The electrochemical measurements are made in a 2 M MgCl2 solution and with the Versastat PalmSens 4 potentiostat. Capacitance measurements were calculated from the equation 1 from the data acquired from the cyclic voltammograms in Figure 3G. The Capacitance was also measured from Electrochemical Impedance Spectroscopy (EIS) by through numerical fitting to an electric circuit model as described by Bera et al.

chemsum

{"title": "Fabrication of Multilayered Composite Nanofibers Using Continuous Chaotic Printing and Electrospinning: Chaotic Electrospinning", "journal": "ChemRxiv"}

can_gold_nanoparticles_aid_electrophoretic_detection_of_sulphur_in_biomolecules?_development_of_gold

## Abstract: The prevalence, distinctive reactivity, and biological significance of sulphur-based groups in proteins and nucleic acids means that analysis of sulphur is of prime importance in biochemistry, biotechnology, and medicine. We report steps in the development of a method to detect these moieties using gold nanoparticles as adjuncts in polyacrylamide gel electrophoresis (Gold-PAGE). The chemistry of sulphur is key in biological systems. In proteins, the correct formation of disulphide bonds is frequently required for folding, and hence function, and the distinctive nucleophilicity of thiols is routinely exploited within chemical biology to enable labelling of proteins via cysteine. 1 In nucleic acids, natural modifications of RNAs include sulphurised nucleobases, 2 and the phosphorothioate linkage is an essential tool for increasing the biostability of nucleic acid therapeutics 3 a field boosted very recently by the first Phase III success of an RNAi therapy. Sulphur groups in DNA are also widely used in structural nanotechnology. 4,5 There is therefore an important need for tools to analyse the presence and chemical state of sulphur in proteins and nucleic acids. Mass spectrometry, usually the first portof-call, can often be challenging under native conditions, while other methods for assessing sulphur chemistry are either destructive (e.g. ICP-MS) or indirect (e.g. circular dichroism). Polyacrylamide gel electrophoresis (PAGE), on the other hand, can be run under native conditions. The inclusion of a mercury layer within PAGE has been shown to identify sulphurised RNAs by diminished mobility mediated by soft-soft Hg-S interactions. 6 In principle, this could give structural information, since the metal-sulphur affinity depends upon accessibility and oxidation state. However, the technique utilises highly toxic organomercury building blocks, and which are embedded within a crosslinked slab of gel, greatly amplifying waste disposal costs. Gold nanoparticles (AuNPs) are far less toxic 7 and are well known for their affinity for sulphur; the replacement of weakly bound ligands by thiols, disulphides, 8 and thioethers 9 is widely employed in their functionalisation. The integration of AuNPs into PAGE (Fig. 1a) could therefore be used to provide information on the number and chemistry of sulphur atoms in proteins and nucleic acids (Fig. 1b). This would mirror the use of boronic acids in PAGE to detect glycation. 10 AuNPs have been incorporated into hydrogels previously, but not for electrophoresis. 11 To achieve this integration, we initially looked into creation of thiolated acrylamide monomers to ensure that any overall charge on AuNPs (dependent upon surface capping) did not result in their own migration under a voltage. These monomers proved difficult to isolate due to thiol-ene self-reactivity. However, during preliminary screening of conditions it became apparent that concern about nanoparticle migration was unwarranted (Fig 1c): citrate-coated AuNPs were retained in within their original gel layer under normal electrophoresis conditions. The addition of β- mercaptoethanol (a common additive used to reduce protein disulphides) was required to transport the AuNPs through the gel. This finding also verified the displacement of weakly coordinating ligands by thiols within the gel matrix. Having established both AuNP entrapment within the polymer gel and ligand exchange, we examined the scope of nanoparticle capping systems which could be used (Fig. 2). Citrate-capped AuNPs have an overall negative charge, and are routinely used as precursors to functionalised AuNPs in aqueous conditions. They were synthesised using the Frens method, 12,13 and their size verified by dynamic light scattering (DLS, r = 9.9 ± 2.6 nm) and a citrate-specific UV-visible spectroscopy method 14 (r = 10 nm). Cetyltrimethylammonium bromide (CTAB) capped nanoparticles are also water soluble, but have a surfactant-type coating, and ligand exchange is known. 15 They were synthesised by a modified version of the Brust-Schriffen synthesis 16 and sized by DLS (r = 12.5 ± 4.2 nm). Dimethylaminopyridine (DMAP)-capped AuNPs are stabilised by weak coordinative bonds and have been used as precursors to thiolated AuNPs. 17 They were produced by Brust-Schriffen reduction followed by phase-transfer ligand exchange, and sized by DLS (r = 10.2 ± 2.9 nm) and TEM (r = 4.4 ± 2.1 nm); the TEM measurement is considered to be more accurate. All AuNP solutions displayed the typical red-purple colouration due to the plasmon resonance band and were stable in solution at 4 °C on the order of months. Gold-PAGE gels were cast for electrophoresis consisting of a standard polyacrylamide layer topped with a AuNP-containing layer (Fig. 2b-c). A series of DNA samples (Fig. 2a) was then used to assess the ability of the gels to selectively retain sulphurous functionalities. These 54mer strands (taken from an aptamer 18 ) had identical nucleobase sequences, but differing degrees and types of sulphur: none (0S), a single phosphorothioate linkage before the terminal nucleoside (1S-term), a single phosphorothioate in the centre of the strand (1S-cent), two and three of the same linkages (2S and 3S respectively), and a terminal thiol-modifier (SH). The citrate AuNPs were found to remain well dispersed throughout the polymerisation process in pH 8 trisborate-EDTA (TBE) buffer. However, there was no visible change in the retention of the strands with respect to their degree or type of sulphur chemistry (Fig. 2b). This remained the case despite variation of AuNP size (d = 8 -40 nm) and concentration (1 -60 mg Au mL -1 ). Noting that both DNA and citrate AuNPs are polyanionic and therefore could experience too much electrostatic repulsion to interact, we switched to a pH 8 tris-acetate- magnesium (TAMg) buffer in which the dications could act to screen the charge (akin to use of Mg 2+ in native PAGE to observe DNA hybridisation). However, this resulted only in aggregation of the AuNPs (presumably via neutralisation of the negatively charged citrate shell) to form large clumps within the gel (ESI Fig. S7), which did not affect DNA migration. CTAB-AuNPs were also found to aggregate during the polymerisation process, even in TBE buffer -this may be due to the weakly capped AuNPs having non-innocent interactions with the free radicals. 19 However, in this case the oligonucleotide migration was altered with respect to a non-Au gel -they now appeared as diffuse smears -but there was no discrimination according to sulphur content (Fig. 2c). This confirms that the oligonucleotides can interact attractively with the AuNPs entrapped within the gel matrix, although at this stage the importance of sulphur is not implicated. We then assessed the DMAP AuNPs in Gold-PAGE. The colloids remained well dispersed through to the end of the gelation period, although seemingly batch-sensitive changes in rate of gelation occurred. This was addressed by increasing the amount of the APS initiator used (see ESI). It was found that at a concentration of 10 mg mL -1 of Au within the upper gel layer, a progressive retention of the DNA strands according to their sulphur content could be observed, albeit in very diffuse bands (Fig. 2d). Importantly, the thiol-modified strand was retained significantly more than the phosphorothioates, indicating sensitivity to both number and type of sulphur functionality. With DMAP AuNPs, a direct interaction of the gold surface with sulphur functionality can be confirmed. However, clarity of the result varied upon repetition, and we therefore endeavoured to examine the gel itself in more detail. Firstly, the rheological properties of the gel were examined (Fig. 3) to assess whether the presence of DMAP AuNPs resulted in alteration of the gel matrix which might affect the way in which the DNA strands pass through. An amplitude sweep was performed on TBE gels at acrylamide concentrations between 5 and 20 %, and with and without 2.5 mg Au mL -1 AuNPs. These studies revealed negligible variation in the storage modulus (G') between +Au and -Au at the same acrylamide concentration, however the loss modulus (G'') was significantly lower in the presence of gold. Frequency sweeps were then performed within the linear viscoelastic region as determined by the amplitude sweeps. Again, the difference made by the presence of AuNPs was only seen in the loss modulus, which was both lower and more consistent across the range of frequencies in the presence of AuNPs. These results show that the gel retains its elasticity but loses just a little of its viscosity when AuNPs are present. For the purposes of Gold-PAGE, we can therefore conclude that the smearing is due to the chemistry of the AuNPs rather than a bulk effect on the physical properties of the gel. To further examine the structure of the gel matrix, we used optical coherence tomography (OCT). OCT can be used to non-invasively create micrometre resolution 3D models of a transparent sample based upon variations in its refractive index along the depth, without physical contact. 20 Gels consisting of an AuNP and Au-free layer were cast within a glass cuvette and volumes were produced using spectral domain OCT. 21 The interface between the layers could be clearly resolved (Fig. 4a)the non-gold gel provided no contrast, whereas the gold layer had a strong, speckled appearance. This is indicative of variations in gold concentration across the gel, resulting in modulation of refractive index. To understand this observation, we undertook transmission electron microscopy of the gels (Fig. 4b), which revealed large aggregates of AuNPs within the gel matrix, and in some places, partial agglomeration. This inhomogeneity may well be responsible for the smeared nature of the bands in Gold-PAGE in its current form: the amount of gold which each DNA strand 'sees' as it passes through the gel varies greatly, resulting in the elongated bands seen in Fig. 2d. ## Conclusions We have demonstrated the entrapment of AuNPs within polyacrylamide gel matrices, and shown that they are capable of binding sulphurous moieties within that matrix. DMAP-AuNPs were able to distinguish different levels of sulphur modification in oligonucleotides in electrophoresis, albeit with poor band quality. We anticipate that the quality of AuNP dispersal within the gel matrix is critical to reproducible and accurate detection of sulphur by Gold-PAGE. To make this method more fully applicable will therefore require embarkation on fine tuning of buffer, polymerisation conditions, and AuNP ligand structure. ## Conflicts of Interest There are no conflicts to declare.

chemsum

{"title": "Can Gold Nanoparticles Aid Electrophoretic Detection of Sulphur in Biomolecules? Development of Gold-PAGE", "journal": "ChemRxiv"}

extremely_stable_anthraquinone_negolytes_synthesized_from_common_precursors

## Abstract: Two extremely stable anthraquinone negolytes were synthesized from inexpensive precursors that potentially decrease the mass production cost. The carbon-linked anthraquinones eliminate S N 2 or S N Ar side reactions. Pairing with a Fe(CN) 6 3À/4À posolyte, they exhibited an open-circuit voltage of 1.0 V. By operating at pH 14, a record low capacity fade rate of <1% per year was demonstrated. ## INTRODUCTION The cost of solar and wind electricity has dropped so precipitously that the main barrier to widespread implementation is their intrinsic intermittency. A safe, low-cost, large-scale electrical energy storage system could enable grid-scale adoption of renewables. Among the numerous proposed technologies, redox flow batteries (RFBs) have been recognized as a potentially viable strategy to address the intermittency of renewable energy. Compared with conventional stationary rechargeable batteries (e.g., lithium-ion batteries and lead-acid batteries), RFBs use redox-active materials dissolved in liquid supporting electrolytes that are stored in external tanks and separated from the power generation stack. This separation allows for the decoupling of energy capacity from output power capacity, thereby providing the possibility of low-cost long-duration discharge. Aqueous RFBs, featuring non-flammable electrolytes, are particularly suitable for storing massive amounts of electricity. Aqueous vanadium RFBs are the most widely studied and adopted systems, but are hindered by the high cost of vanadium. In contrast, redoxactive organic molecules comprising earth abundant elements such as carbon (C), hydrogen (H), oxygen (O), and nitrogen (N) have the potential to be inexpensive alternatives to vanadium. Additionally, the structural diversity and tunability of organics enable chemists to design organics with essential properties such as high aqueous solubility, high chemical stability, fast kinetics, and appropriate redox potential. 6, 14,15 Recently, water-soluble anthraquinones 4,4 0 -([9,10-anthraquinone-2,6-diyl]dioxy)dibutyrate ( The Bigger Picture The cost of solar and wind electricity is decreasing so rapidly that a grid-scale energy storage technology will become essential. Aqueous organic redox flow batteries are a potentially safe, inexpensive substitute for lithiumion batteries and vanadium flow batteries for large-scale energy storage. Here, we report a new synthetic strategy for two extremely stable anthraquinone negolyte (negative electrolyte) molecules starting from inexpensive precursors that potentially decrease the cost when scaled up. Additionally, we demonstrate that an anthraquinone negolyte is more stable running at pH 14 than at pH 12, and is expected to be more stable in alkaline solution than in acidic or neutral conditions. Paired with a Fe(CN) 6 3/4 positive electrolyte, the anthraquinone cell exhibited a record low capacity fade rate of <1% per year. The new synthetic strategy for these highly stable anthraquinone negolytes might facilitate the commercialization of anthraquinone-based flow batteries. bis[propane-3,1-diyl])bis(phosphonic acid) (2,6-DPPEAQ) in mildly alkaline solutions have demonstrated extremely low temporal fade rates in flow batteries paired with K 4 Fe(CN) 6 . These quinones are chemically synthesized from 2,6-dihydroxyanthraquinone (2,6-DHAQ) by industry-compatible methods. However, 2,6-DHAQ and 2,7-DHAQ are always co-produced and are costly to separate. Furthermore, our previous research showed that the molecular lifetimes of anthraquinone-based electrolytes can differ by two orders of magnitude depending on the positions of their functional groups (e.g., 1,8-and 2,6-anthraquinones). 16,19 Therefore, it is important to quantify the stabilities of organic molecules with a mixture of isomers. Additionally, 2,6-DHAQ and 2,7-DHAQ are synthesized from 9,10-anthraquinone-2,6-disulfonic acid and 9,10-anthraquinone-2,7-disulfonic acid, respectively, in strong alkaline solution for 35 h at a high temperature (180 C) with a moderate yield, which is energy-intensive and costly. 20 Thus, designing low-cost and chemically stable anthraquinones is of vital importance for the commercialization of aqueous organic RFBs. Because of the inherent chemical stability of the parent structure, the addition of side chains to an anthraquinone core is usually accomplished by a stepwise procedure via anthraquinone derivatives (e.g., hydroxylated anthraquinone or chlorinated anthraquinone). 16,17,19,21 Here, we report a new synthetic route for water-soluble anthraquinones starting from a potentially inexpensive anthracene derivative, 9,10-dihydroanthracene, which can be readily produced from anthracene with a yield of almost 100%. 22,23 Anthracene, a component of coal tar, is one of the major resources for large-scale anthraquinone production. 24 The first step is a Friedel-Crafts alkylation or acylation to render dihydroanthracene and anthracene water-soluble; the last step is an oxidation to produce the corresponding redox-active anthraquinones, i.e., 3,3 0 -(9,10-anthraquinone-diyl)bis(3-methylbutanoic acid) (DPivOHAQ) and 4,4 0 -(9,10-anthraquinone-diyl)dibutanoic acid (DBAQ). Both molecules exhibit high water solubility and chemical stability at pH 12. The DBAQ negolyte (negative electrolyte) has a water solubility of 1.0 M, corresponding to a volumetric capacity of 53.6 Ah L; when paired with potassium ferrocyanide, a full cell exhibited a capacity fade rate of 0.0084% per day or 3.1% per year. DPivOHAQ has a solubility of 0.74 M; when paired with potassium ferrocyanide, a full cell exhibited a capacity fade rate of 0.014% per day or 5.1% per year. Additionally, we demonstrated that the DPivOHAQ negolyte is even more stable in strong base, exhibiting a capacity fade rate of 0.0018% per day or 0.66% per year at pH 14. Furthermore, we demonstrated that the capacity fade is due to formation of anthrone, which can convert back to anthraquinone through air exposure and can also be suppressed at high pH. Thus, these findings suggest that, through a combination of increased pH and periodic air exposure, both DPivOHAQ and DBAQ offer the possibility of decadal lifetimes in aqueous RFBs. In the following section we report methods and results, first for DPivOHAQ and then for DBAQ. ## RESULTS AND DISCUSSION Figure 1 illustrates the synthetic routes, chemical structure, Pourbaix diagram, and cyclic voltammogram (CV) of DPivOHAQ. Synthesis was achieved by first functionalizing 9,10-dihydroanthracene with the water-soluble group -C(CH 3 ) 2 CH 2 COOH, followed by an oxidation step with CrO 3 . The oxidation step is well known in industry for anthraquinone synthesis, and it can also be accomplished by other methods, such as nitric acid and air, as confirmed by the 1 H NMR of DPivOHAQ in Figure S3, that could further decrease the costs when scaled up. 25,26 However, CrO 3 oxidation is a more controllable method on a research scale to get a purer product, therefore, in this work, we use CrO 3 as the oxidant unless specifically mentioned otherwise. Compared with 2,6-DBEAQ and 2,6-DPPEAQ synthesis, the reported synthetic approach could be potentially more cost-effective. For example, assuming anthracene is the starting material, both DBEAQ and DPPEAQ require five synthetic steps in total as illustrated in Scheme S1 with an overall yield around 52% and 59%, respectively; whereas, DPivOHAQ synthesis requires only three steps with an overall yield of 81%. Moreover, the cost of side chains for DPivOHAQ is slightly lower than that of DBEAQ and substantially lower than that of DPPEAQ at sub-kg scales, as shown in Table S1. Therefore, we expect that DPivOHAQ is most likely less expensive to produce than DBEAQ and DPPEAQ at industrial scales. In addition to potentially lower synthetic costs, DPivOHAQ is functionalized with carbon-linked functional groups, which are chemically more robust than the oxygenlinked side chains in DBEAQ and DPPEAQ, minimizing the opportunity for S N 2 and S N Ar side reactions to occur. 16,17 The thermochemical stability of both oxidized and reduced forms of DPivOHAQ were evaluated at high temperature (65 C) and in strongly alkaline conditions (pH 14) for eight days. No apparent decomposition was detected from the 1 H NMR as shown in Figure S8, indicating that DPivOHAQ is quite chemically stable. A temperature of 65 C is well above anticipated operating temperatures of RFBs; thus there should be much less decomposition in real applications. Synthesis of DPivOHAQ results in a mixture of 2,6-and 2,7-isomers that does not require further separation prior to use in a battery. The Pourbaix diagram of DPivO-HAQ, shown in Figure 1B, suggests that the molecule undergoes a two-proton/ two-electron process below pH 9, a one-proton/two-electron process between pH 9 and 11, and a zero-proton process with a pH-independent potential of approximately 0.48 V versus standard hydrogen electrode (SHE) at pH > 11. Pairing a DPi-vOHAQ negolyte with potassium ferrocyanide at pH 12 should yield an equilibrium cell potential of approximately 0.98 V (Figure 1C). Electrochemical kinetics of DPivO-HAQ reduction were determined with the rotating disk electrode (RDE) method as shown in Figure S9. The charge transfer coefficient is 0.49, the diffusion coefficient is 2.4 3 10 6 cm 2 s -1 , and the kinetic rate constant is 2.5 3 10 3 cm s -1 ; the latter is much higher than is typical of inorganic redox-active materials at an uncatalyzed carbon electrode. 27 It should be noted that such a high kinetic rate constant could be a lower bound to the actual value, because a RDE system exhibits mixed kinetic and transport limits in the observed mass-transfer limited current density region. The solubilities of the oxidized forms of anthraquinones in alkaline solutions are typically lower than those of the reduced forms. The solubility of DPivOHAQ at pH 12 was determined by UV-vis spectrophotometry (Figure S10) to be 0.74 M, corresponding to a volumetric capacity of 39.7 Ah L -1 . Polarization experiments of a 0.5 M DPivOHAQ-ferrocyanide full cell at pH 12 were performed at various states of charge. The electrolytes comprised 5 mL of 0.5 M DPi-vOHAQ (negolyte) at pH 12 (10 mM KOH) and 80 mL of 0.3 M potassium ferrocyanide and 0.1 M potassium ferricyanide (posolyte) at pH 12. The cell was constructed from graphite flow plates and AvCarb carbon cloth electrodes, separated by a Fumasep E-620 (K) membrane because of its low permeability to ferricyanide and high ionic conductivity. 16,17 A peak galvanic power density of 0.34 W cm 2 was achieved at $100% state of charge (SOC) (Figure 2A). Careful attention to engineering design and construction should raise the power density even farther. 28 The open-circuit voltage (OCV) increases from 0.95 to 1.08 V as the SOC increases from 10% to $100%, and the OCV at 50% SOC of 0.99 V (Figure 2B) is consistent with the voltage expected from CV. The alternating current area-specific resistance (ASR) of the cell was determined via high-frequency potentiostatic electrochemical impedance spectroscopy (EIS), and the value was below 0.6 U cm 2 across all SOCs (Figure 2B). This is a relatively low alternating current ASR value for RFBs with alkaline electrolytes. 16,17 The polarization ASR was determined using the linear region within the voltage range 0.9-1.1 V (Figures 2A and 2B). The ASR of the membrane (0.54 U cm 2 at 50% SOC, determined by high-frequency EIS in the full cell) accounted for around 67% of the ASR of the entire cell (0.81 U cm 2 at 50% SOC, DC polarization). The capacity utilization for the negolyte (capacity-limiting side) is approximately 95% at 50 mA cm 2 with a high round-trip energy efficiency of 91.5% (Figures 2C and 2D). At a reasonable practical operation target of 80% round-trip energy efficiency, the low value of the ASR permits galvanostatic operation at around 140 mA cm 2 with an electrolytic power density of 0.16 W cm 2 , a galvanic power density of 0.13 W cm 2 , and 92% capacity utilization. The same 0.5 M DPivOHAQ-ferrocyanide full cell was used for long-term stability evaluation (Figure 3). The cell was cycled at a constant current density of G0.1 A cm 2 , and each galvanostatic half cycle was followed by a potential hold at the voltage limit (1.3 V for charge, 0.6 V for discharge) until the current density fell below 2 mA cm 2 to mitigate the effect of temporal variations in accessible capacity during full cell cycling caused by drifts in cell resistance. 29 The charge-discharge profiles near the voltage limits (Figures 3B and 3C) are quite steep and are followed by small subsequent horizontal segments during the potential holds. The horizontal segments end at 95.5% of theoretical capacity of negolyte, but the steepness followed by a small subsequent horizontal segment of the charge-discharge profiles suggests ll Chem 6, 1432-1442, June 11, 2020 1435 Article that the electrolyte had around 4.5% inactive material and that the active material is undergoing deep cycling to essentially the full SOC limits. The cell was cycled for 690 cycles at 100 mA cm 2 , which required 15.6 days to complete. The capacity retention over the 690 cycles was 99.78% with an average Coulombic efficiency greater than 99.9%, reflecting a capacity fade rate of 0.00031% per cycle or 0.014% per day (Figure 3A), i.e., 5.1% per year. This temporal fade rate is among the lowest exhibited by full cells in which organic molecules composed the capacity-limiting side: 2,6-DBEAQ fades at $0.04% per day; 2,6-DPPEAQ fades at $0.014% per day. 18 After 15.6 days of cycling at pH 12, DPivOHAQ negolyte, in the discharged state, was exposed to air for 2 h and the pH was adjusted to 14 by dissolving KOH pellets into the negolyte and posolyte without changing cell materials or setup. As a result, 81% of the lost capacity was recovered, as shown in Figure 3A. Over the additional 16 days at pH 14, the cell exhibited a capacity fade rate of 0.0018% per day, which is 6 times lower than that at pH 12. The charge-discharge voltage profiles (Figures 3B and 3C) are almost invariant, indicating no apparent change in ohmic resistance and good chemical compatibility with cell membrane and other cell components. 30 Based on the decomposition study of 2,6-DHAQ, 31 we attribute capacity fade to anthrone formation (Scheme 1). Therefore, increasing the hydroxide concentration should suppress the formation of anthrone. This expectation is consistent with the lower capacity fade rate observed at pH 14 than at pH 12. In general, the disproportionation reaction will generate OH (or consume H + ) at pH above the first pK a of the anthrahydroquinone. Therefore, anthrone formation will be disfavored under alkaline conditions relative to acid conditions and will be progressively disfavored as the pH increases (Figure S11). We interpret the sudden increase in capacity at cycle 691 as the consequence of anthrone being converted back to anthraquinone by both the pH effect and the effect of exposure to atmospheric O 2 . To confirm the major side reaction is the disproportionation of reduced anthraquinone, a fully reduced ($100% SOC) sample of DPivOHAQ at pH 12 was prepared and stored in an fluorinated ethylene propylene (FEP) vial in a glove box for 238 days, allowing the disproportionation to reach equilibrium. Indeed, some appreciable side peaks appeared in the 1 H NMR spectrum of the reduced DPivOHAQ, whereas upon re-oxidation in air, the 1 H NMR spectrum contained no observable decomposition peaks (Figure S12), indicating that the decomposition compounds can either be re-oxidized back to ## Article DPivOHAQ or are converted to other products with no observable signals above the detection limit of the NMR instrument. Therefore, high-performance liquid chromatography-mass spectrometry was performed to analyze both the reduced and the reoxidized samples (Figure S13). The anthrone species was detected in the reduced sample but not in the re-oxidized sample, in agreement with the 1 H NMR result shown in Figure S12. In the re-oxidized sample after 238 days, approximately 1.24% of the signal corresponded to an anthrone dimer, suggesting a fade rate of 1.90% per year for reduced DPivOHAQ at pH 12 after aeration. Goulet et al. showed that, by avoiding high SOC, the anthrone formation rate in 2,6-DHAQ decreased substantially. 31 They also demonstrated recovery of most of the lost capacity by air exposure. Because DPivOHAQ also decomposes via anthrone formation, we hypothesize that similar approaches will extend its lifetime significantly. We suggest that a flow cell with decades-long calendar life might be achievable with DPivOHAQ at pH 12. Another anthraquinone, DBAQ, with higher solubility, was synthesized (Figure 4A) using a similar strategy. The first step is a Friedel-Crafts acylation, followed by Wolff-Kishner reduction of the carbonyl groups to methylene. The last step is to oxidize the corresponding anthracene (or 9,10-dihydroanthracene) to the final anthraquinone form. Compared with the three steps for DPivOHAQ synthesis, four steps are required for DBAQ synthesis when starting from anthracene. Because of the low cost of succinic anhydride and given that the Wolff-Kishner reduction is well-developed in industry, the cost of DBAQ could also be low. Electrochemical kinetics studies of DBAQ were conducted by RDE techniques as shown in Figure S14A. The diffusion coefficient of the oxidized form of DBAQ was determined by Levich analysis (Figure S14B) to be 2.5 3 10 6 cm 2 s -1 . According to the Koutecky ´-Levich equation and Tafel plot, the charge transfer coefficient is 0.50, and the kinetic rate constant is 2.9 3 10 3 cm s 1 , which is still slightly higher than that of DPivOHAQ. The solubility of DBAQ was determined to be 1.0 M at pH 12, corresponding to a volumetric capacity of 53.6 Ah L -1 for the negolyte. The reduction potential is 0.47 V versus SHE at pH 12. When paired with potassium ferrocyanide, a full cell of approximately 0.97 V can be achieved. To evaluate the stability of a DBAQ negolyte, a full cell was assembled with 5 mL of 0.5 M DBAQ negolyte at pH 12 as the capacity-limiting side and a posolyte comprising 80 mL of 0.3 M K 4 Fe(CN) 6 with 0.1 M K 3 Fe(CN) 6 at pH 12 as the non-capacity-limiting side. The flow cell was constructed from graphite flow plates and carbon paper electrodes, separated by a Fumasep E-620 (K) membrane. The cell exhibited 91.7% of its theoretical capacity for the negolyte. It was cycled for 650 cycles at 100 mA cm 2 , which required 15.5 days to complete. The average capacity fade rate was 0.0084% per day, corresponding to 3.1% per year. Assuming the capacity fade is primarily due to anthrone formation then, with careful control of the pH and SOC of the DBAQ electrolyte and periodic exposure to air, DBAQ might exhibit an even lower loss rate in real-world applications. ## Article Conclusions In this report, we have demonstrated a new route to synthesize water-soluble anthraquinones with solubilizing groups attached by carbon-carbon bonds, starting from potentially inexpensive 9,10-dihydroanthracene. These anthraquinones exhibit high aqueous solubilities and low capacity fade rates of 0.0084% per day and 0.014% per day at pH 12, respectively. We demonstrated in a full cell containing a DPivOHAQ negolyte that anthrone formation is the major side reaction responsible for capacity fade and that air exposure can recover most of the lost capacity. Furthermore, by increasing the pH of the negolyte, we demonstrated the suppression of the DPivOHAQ capacity fade rate to an extremely low value of less than 1% per year. We expect that the stability of DPivOHAQ and DBAQ can be even further improved with careful control of the battery operating conditions. We suggest that strategies combining SOC limit control, precision air exposure, and pH tuning can be extended to other inexpensive anthraquinone molecules to achieve extremely low capacity fade rates, paving the way for commercializing anthraquinone-based RFBs to enable grid-scale energy storage of renewable electricity. ## EXPERIMENTAL PROCEDURES Experiment Materials 9,10-dihydroanthracene (97%), 3,3-dimethylacrylic acid (97%), and anhydrous dichloromethane were purchased from Sigma Aldrich. Anhydrous aluminum chloride (95%) was purchased from Alfa Aesar. All chemicals were used as received. 13.32 g (99.93 mmol) of AlCl 3 was suspended in $200 mL of anhydrous CH 2 Cl 2 . A solution of 6.67 g (66.62 mmol) of 3,3-dimethylacrylic acid in $20 mL of anhydrous CH 2 Cl 2 was added by syringe and the mixture stirred at room temperature ($20 C) for 0.5 h under nitrogen. Subsequently, a solution of 5.00 g (27.74 mmol) of 9,10-dihydroanthracene in $15 mL of anhydrous CH 2 Cl 2 was added to the above mixture and stirred for 48 h at room temperature ($20 C). After that, the solvent was quenched with 200 mL of 1 M aqueous HCl and stirred overnight. The organic layer was then removed and the remaining solution was filtered to afford the pale-yellow product. The yield given was 95%. 3,3 0 -(9,10-Anthraquinone-Diyl)Bis(3-Methylbutanoic Acid) (DPivOHAQ) DPivOHAC (6.00 g, 15.85 mM) was dissolved in glacial acetic acid (70 mL). Then, a CrO 3 solution (3.33 g, 33.3 mM) was added to the DPivOHAC and acetic acid mixture. The reaction mixture was heated at 90 C for 1 h. After cooling down to room temperature ($20 C), water was added to precipitate the solid. The compound was purified by dissolution in base followed by addition of acid to afford the precipitate. The yield given was 85%. The 1 H NMR spectra of 3,3 0 -(anthracene-diyl)bis(3-methylbutanoic acid) and 3,3 0 -(9,10-anthraquinone-diyl)bis(3-methylbutanoic acid) are shown in Figures S1 and S2. DBAQ-related synthesis uses a similar method, thus is provided in the supporting information. ## Full Cell Measurements Flow battery experiments were conducted with cell hardware from Fuel Cell Tech. (Albuquerque, NM), assembled into a zero-gap flow cell configuration, similar to a previous report. 17,18 Pyrosealed POCO graphite flow plates with serpentine flow patterns were used for both electrodes. Each electrode comprised a 5 cm 2 geometric surface area covered by one sheet of AvCarb carbon cloth or three sheets of SGL 39AA electrode. For DPivOHAQ-ferrocyanide full cell tests, a Fumasep E-620 (K) membrane was used to serve as the ion-selective membrane between the AvCarb electrodes. For DBAQ-ferrocyanide full cell tests, a Fumasep E-620 (K) membrane was used to serve as the ion-selective membrane between the SGL 39AA electrodes. The outer portion of the space between the electrodes was gasketed by Viton sheets with the area over the electrodes cut out. Torque applied during cell assembly was 60 lb-in (6.78 N m) on each of eight bolts. The electrolytes were fed into the cell through fluorinated ethylene propylene (FEP) tubing at a rate of 60 mL/min, controlled by Cole-Parmer six Masterflex L/S peristaltic pumps. All cells were run inside a nitrogen-filled glove bag. Cell polarization measurements, impedance spectroscopy, and charge-discharge cycling were performed using a Biologic BCS-815 battery cycler. Galvanostatic cycling was performed at G0.1 A cm 2 at room temperature ($20 C) with voltage limits of 0.6 and 1.3 V. To obtain the polarization curves, the cell was first charged to the desired SOC and then polarized via linear sweep voltammetry at a rate of 100 mV s 1 . This method was found to yield polarization curves very close to point-by-point galvanostatic holds, yet to impose minimal perturbation to the SOC of the small-electrolyte-volume cell. EIS was performed at SOCs between 10% and 100% at open-circuit potential with a 10-mV perturbation and with frequency ranging from 1 to 300,000 Hz. ## SUPPLEMENTAL INFORMATION Supplemental Information can be found online at https://doi.org/10.1016/j.chempr. 2020.03.021.

chemsum

{"title": "Extremely Stable Anthraquinone Negolytes Synthesized from Common Precursors", "journal": "Chem Cell"}

strong_electronic_coupling_of_graphene_nanoribbons_onto_basal_plane_of_glassy_carbon_electrode

## Abstract: The grafting of molecular motifs to the conductive carbon represents a promising approach towards new hybrid materials for electrocatalytic applications. Here, we investigate the electrochemical behavior of graphene nanoribbons deposited onto glassy carbon electrode using 𝜋 − 𝜋 stacking interactions. Using the bipyrimidine moiety on the nanoribbon edges as a reporter of the proton-coupled electron transfer chemistry, we illustrate that the simple electrochemical treatment of as-deposited nanoribbon generates a hybrid material that is in strong electronic communication with the conductive support. This work shows novel strategy for modifying basal plane of carbon materials and provide potential platform for incorporation of catalytic metal sites via coordination through N-functionalities of GNR. Graphitic nanocarbon materials have recently been recognized as promising electrocatalysts for a range of useful transformations, such as the reduction of dioxygen to water, dinitrogen to ammonia, and carbon dioxide to methanol. The success of these electrode materials is attributed to the high electronic conductivity of graphitic carbon, the high surface area of nanocarbon electrodes made using template-directed pyrolysis, and the presence of reactive catalytic functional groups obtained by heteroatom doping of otherwise chemically inert carbon framework. Further progress in the field requires the synthesis of nanocarbon electrodes with complex multinuclear catalytic motifs that go beyond "single-atom catalysts". Unfortunately, introduction of such well-defined catalytic functionalities is limited by the extreme temperatures required to graphitize carbon precursors, at which most functional groups undergo decomposition. A more favorable way to introduce catalytic moieties to graphitic electrodes involves post-pyrolysis modifications of carbon edge and basal sites with well-defined molecular motifs. For this purpose, several creative chemical methods have been developed to immobilize molecular catalysts to the carbonbased surfaces using covalent bonds, 𝜋 − 𝜋 stacking and electrostatic interactions. 20,21 Carbon edge sites are often modified covalently using 'click' chemistry 4,7 or aryl radical intermediates formed from diazonium salts. 5,6 Carbon basal sites are modified using 𝜋 − 𝜋 stacking interactions between polyaromatic groups, most commonly pyrene, and the carbon electrode. Here, the catalytic units are connected to pyrene through an alkyl chain. 19 These modification techniques ensure the immobilization of homogeneous catalysts, thus eliminating the need for their diffusion to the electrode surface. However, the electrochemical behavior of such immobilized catalysts usually does not differ from that observed in homogeneous analogs because the electronic coupling between the catalytic moieties and the conductive carbon electrode remains weak. More recently, chemical methods have been explored to graft molecular catalysts using strong electronic coupling to create new hybrid structures with altered reactivity and catalytic behavior. Specifically, Compton and coworkers showed that the carbon edge planes can be functionalized using ortho-quinone groups generated by the anodic treatment of the electrode surface. 22 The Surendranath group used this coupling method to attach molecular catalysts to the carbon edge sites and found that the strong electronic coupling exists between the catalytic unit and the carbon band structures. Such strong coupling resulted in interesting changes of the electrochemical behavior of the grafted hybrid relative to the homogeneous analog. For example, grafted hybrids were shown not to undergo simple outer-sphere electron transfer processes, which in turn had important implications in catalysis by enabling the proton-coupled chemistry to take place: a hydrogen-evolving Rh-based catalyst, when grafted to the carbon electrode, operates over a full pH region (0-14), whereas the homogeneous analog shows catalytic activity only in the acidic region. Similar mechanistic studies involving basal plane modifications and the strong electronic coupling regime are scarcer. Several recent reports involving immobilized metal-coordinated N4-macrocycles indicate that the 𝜋-stacking interactions may generate strongly coupled hybrids. For example, the coupling of molecular cobalt phthalocyanine, a carbon dioxide reduction catalyst, to a carbon nanotube support has been shown to affect product selectivity: two-electron reduction to CO is favored when the homogenous catalyst is used, while a six-electron reduction to methanol takes place with the heterogeneous catalyst/carbon hybrid. 26 Similar studies with immobilized metal-coordinated N4-macrocycles have shown better efficiency toward CO2 reduction to CO, as well as improved nitrogen reduction reaction efficiency. 33 However, most of those basal plane modification studies have been focused on catalytic performance rather than mechanistic investigation of their fundamental electrochemistry. Scheme 1. Structures of model compounds BPM an GNR and the products of their cathodic reduction via two-electron, twoproton (BPMH2 and GNRH2) and two-electron four-proton (BPMH4 2+ and GNRH4 2+ ) transfer chemistry. Here, we investigate the degree of electronic coupling in molecule/carbon electrode hybrids functionalized using 𝜋-stacking interactions. Specifically, we explore how the electrochemical behavior of 4,4'-bipyrimidine (BPM) changes when immobilized to the glassy carbon surface using graphene nanoribbon linkers (GNR, Scheme 1). GNRs exhibit large planar aromatic structures capable of 𝜋 − 𝜋 stacking with the basal sites of the carbon electrode, while the redox activity of bipyrimidine functionalities reports on the degree of electronic coupling between the nitrogen sites and the electrode surface. Our detailed studies of proton-coupled electron transfer (PCET) chemistry of nitrogen moieties reveal that the electrochemically treated GNRs form monolayers that are strongly coupled with the carbon electrode. This work shows a novel strategy for modifying the basal plane of carbon materials and provides a potential platform for incorporation of catalytic metal sites via coordination through N-functionalities of GNR. GNR was synthesized according to the method published by Sinitskii and coworkers (SI, Scheme S1). 35 The procedure involves the synthesis of a soluble aryl-substituted benzene monomer 8 via Diels-Alder cycloaddition/decarbonylation sequence developed by Müllen and co-workers. 36 Dibrominated monomer 8 was then subjected to Yamamoto coupling to produce soluble polymer 9. The MALDI mass spectrum of polymer 9 reveals a series of oligomer peaks, separated by 522 au (corresponding to the mass of one monomer unit), with sizes up to 16 monomer units corresponding to ~20 nm length (Figure S1, SI). The sample of polymer 9 may contain longer chains that are not detectable by mass spectrometry, and this hypothesis is consistent with previous reports of micrometer-long nanoribbons obtained using similar Yamamoto coupling procedures. 35,37 The oxidative cyclodehydrogenation of polymer 9 yielded insoluble GNR, which was characterized using solid state NMR, Raman spectroscopy, transmission electron microscopy, and X-ray photoelectron spectroscopy (XPS) (Figures S2-S5, SI). Electrochemical behavior of BPM was studied using cyclic voltammetry (CV) in aqueous medium (black traces in Figure 1 and Figure S6). Chemically reversible or quasi-reversible reduction features were observed in the cathodic region (-0.5 to 0 V vs. NHE), and the half-wave potential was found to shift to more negative values with increasing pH, indicative of PCET. The equilibrium potentials were used to construct the Pourbaix diagram in Figure 2A, which shows two distinct pH regions (0-4 and 4-14) in which the potential varies linearly with pH. The experimental data were fit to the Nernst equation for PCET, 38 which lead to the assignment of the two pH regions as follows: the pH=4-14 process was assigned to a two-electron two-proton transfer reduction to form BPMH2, while the acidic pH=0-4 region was assigned to the two-electron four-proton coupled reduction to form BPMH4 2+ (Scheme 1). This conclusion was further supported by an excellent match between the standard reduction potential 𝐸 𝑃𝐶𝐸𝑇 0 obtained from the Pourbaix diagram and the same parameter calculated using standard reduction potentials for single electron transfer processes and the relevant pKa values (more information is available in Tables S1 and S2 of Section S4B, Supporting Information, SI). The electrochemical behavior of as-deposited GNR is significantly different from BPM (red traces in Figure 1 and Figures S7 and S8). For example, the CV of GNR collected at pH=1 exhibits an intense cathodic peak at Ec= -0.752 V and a weaker anodic peak at Ea=1.074 V. This large separation between the cathodic and anodic peaks indicates a large kinetic barrier for the observed electrochemical transformation. Interestingly the approximate half-wave potential for GNR is similar (E1/2 = 202 mV) to that observed in BPM (E1/2 = 164 mV). Furthermore, the halfwave potential shifts to more negative potentials with increasing pH, suggesting that the observed transformation is the proton-coupled reduction of bipyrimidine moieties in GNR. The slow rate for the observed PCET is likely associated with the low electron and proton conductivity in GNR-aggregates in the as-deposited sample. Similar sluggish kinetics were observed in studies of phthalocyanine aggregates 29,39 and these results illustrate the need for the development of deposition methods that ensure monomolecular coverage of molecular units and good electronic communication with the carbon support. Interestingly, the cathodic (Ic) and anodic (Ia) peak currents in as-deposited GNR are not the same. For example, Ic/Ia in the CV of GNR at pH=1 is 7, indicating that some of the GNR material is detached from the electrode surface upon reduction. This prompted us to investigate the electrochemical behavior of GNR upon subsequent scans (blue traces in Figure 1 and Fig- ures S7-S9). The CVs of GNR after the first CV scan are significantly different from those for as-deposited GNR. Namely, the second-scan CVs show the appearance of new reversible and pH-dependent features in the 0 to -1 V potential range. These reversible features in the Figure 1. CVs of BPM (black), as-deposited GNR (red), and second-scan GNR (blue) in 1 M aqueous solutions at pH 1 (left), pH 7 (middle), and pH 13 (right). GC working electrode; 100 mV/s scan rate. Background scans are shown in pale red in all three panels. second scan appear only if the first scan ends at potentials more positive than the anodic process (Figure S10), suggesting that their formation requires reduction and re-oxidation of as-deposited GNR, and are assigned to the monolayer GNR deposited to the carbon support by electrochemical cycling. This assignment is based on the fact that the anodic-cathodic peak potential separation ∆E is drastically reduced in the second-scan features, suggesting that the kinetics of electron and proton transfer are significantly improved. Furthermore, the integration of CV peaks is consistent with the expected current for the 0.5 -3 monolayers GNR coverage (Section S4). Based on these experimental findings, we hypothesize that the electrochemical reduction of as-deposited GNRs causes their partial desorption and solubilization in the solution above the electrode surface. The subsequent re-oxidation of solubilized GNRs results in their redeposition to the electrode surface, causing improved electronic coupling to the carbon electrode support (as illustrated in Figure 4). This simple electrochemical treatment provides a promising method towards deposition of molecular species to the carbon electrode and can be useful to the field of molecular enhancement of heterogeneous electrocatalysis. 40 We now seek to answer the following question: does deposited GNR behave like the molecular species in solution or does it undergo the field-induced electrochemistry observed previously for molecular species that are in strong electronic communication with the electrode? To address this, we compared Pourbaix diagrams for as-deposited GNRs (first CV scan) and monolayer GNRs (second CV scan). The Pourbaix diagram of as-deposited GNRs (Figure 2B) show that their half-wave potentials are slightly more positive than the corresponding potentials for BPM, and this shift is assigned to the extended conjugation of GNR. Aside from this difference, the Pourbaix diagrams of BPM and as-deposited GNR exhibit similar pH-dependence, indicative of GNRH2 formation in the pH=4-14 range and GNRH4 2+ formation in the 0-4 region (Scheme 1). The twoelectron, four-proton reduction behavior observed in the pH=0-4 region is indicative of weak electronic coupling between GNR and the carbon support. When strong electronic coupling is achieved, each proton transfer event generates a surface charge that is immediately compensated by the electron transfer from the electrode, leading to an overall neutral product and the 59 mV/pH slope in the Pourbaix diagram. 25 Again, the additional support of the PCET assignment in weakly-coupled GNR was provided by obtaining a good match between the experimental 𝐸 𝑃𝐶𝐸𝑇 0 values and those calculated using DFT (Tables S3 and S4 in Section S4B). The Pourbaix diagram of the second CV-scan GNR (Figure 2C) is different from that obtained for as-deposited GNRs. In specific, the slope of the Pourbaix diagram remains the same throughout the entire pH=0-14 range. The change in PCET chemistry that was observed at ~pH 4 in the case of BPM and as-deposited GNR was absent. We hypothesize that this behavior indicates that the second CV-scan GNR is strongly coupled with the carbon support. As mentioned, earlier, the stronglycoupled system is expected to only undergo processes that are overall charge neutral  each protonation of the surface nitrogen sites triggers the immediate charge compensation by electrons from the carbon electrode. Thus, the absence of the twoelectron, four-proton coupled chemistry in the acidic region is indicative of the strong coupling regime. The exact number of protons (m) and electrons (n) transferred during PCET is difficult to evaluate from Figure 2C because the ~59 mV/pH slope informs only of the m/n=1 ratio. To obtain more insight into the details of the process, the Pourbaix diagram was treated using the model developed by Surendranath for field driven PCET. 24,25 Based on this model, the 𝐸 𝑃𝐶𝐸𝑇 0 can be estimated from the potential of zero free charge for the electrode/electrolyte interface and the zero-field pKa values of the surface protonation sites. We applied this model to two possible PCET scenarios, one where two protons and electrons are transferred and another where four protons and electrons are transferred (Section S4B, SI). Better match with the experiment was observed for two proton/electron PCET, indicating the likelihood that only two nitrogen centers are protonated. XPS investigations on modified carbon electrodes were conducted to provide additional insight into the electrochemical behavior of GNRs (experimental details are presented in Section S5). Electrodes were prepared by depositing GNR in different thickness regimes, to investigate both the weakly-coupled GNR aggregates and strongly-coupled GNR monolayers. The sample thickness had a profound effect on the observed N1s peaks (Figure 3A). Specifically, the thick sample showed a major peak at 398.3 eV (red), which we attribute to pyridinic N atoms. 35,41 The same peak was observed for GNR powder, indicating that it originates from GNR aggregates that are not coupled to the carbon electrode. The thick sample had additional higher binding energy (BE) features at 402.1 (blue) and 400.4 eV (green). The 400.4 eV feature was observed on the bare electrode, and it arises due to native nitrogen present in commercial carbon electrodes. 42 The 402.1 eV peak is assigned to the signal from monolayer GNR. This assignment is consistent with the appearance of the N 1s peak at higher BE, a trend that is expected for GNR in intimate electronic interaction with the electron-withdrawing electrode environment. 43,44 Meanwhile, the thin sample only shows features from the strongly-coupled monolayer GNR and the nitrogen present on the bare electrode (Figure 3A), further confirming our assignment of 398.3 and 402.1 eV peaks to those arising from aggregate and monolayer GNRs, respectively. The changes in the XPS N 1s peaks of thick GNR samples were monitored after electrochemical treatment (Figure 3B). The samples, initially at OCP, were first reduced at a constant potential of -1.7 V, then re-oxidized at a constant potential of +0.9 V. The N 1s XPS data were collected at three potentials to observe changes to the sample. Interestingly, cathodic treatment, which gives rise to PCET, does not lead to any significant shift in BE from the weakly-or strongly-coupled signals. The lack of the shift in the red signal after the -1.7 V treatment is not surprising, as it is assigned to the residual GNR aggregates on the electrode which were not electrochemically reduced. We hypothesize that the reduced GNR detaches from the electrode and thus not contribute to the XPS signal. The lack of the shift in the blue signal is more surprising, as it originates from strongly-coupled GNRs which have participated in PCET. The observed insensitivity of the N 1s BE indicates that the effect of nitrogen protonation on BE is fully compensated by the electron density redistribution within the carbon electrode, resulting in a near zero BE shift (Figure 4). This finding is similar to in situ studies on strongly-coupled Rh-functionalized GC electrodes where, unlike their molecular analogs, no change in X-ray absorption near edge structure, i.e., oxidation state, during ioncoupled ET is observed. 23,45 Similarly, subsequent anodic treatment does not induce changes to the binding energies of weakly-or strongly-coupled GNR. Importantly, integrated areas under XPS curves suggest the electrochemical treatment is enriching the ratio of strongly-to weakly-coupled GNRs as it changes from 1:12, 1:5, and 1:4 across the OCP, cathodic, and anodic panels, respectively. These results provide additional evidence that the electrochemical treatment of GNRs leads to the improved electronic contact between GNR and the carbon support. Based on combined electrochemical and XPS investigations, we hypothesize that strong coupling between GNR and the carbon support can be achieved via electrochemical cycling of drop-casted GNRs, as schematized for acidic electrolyte solution in Figure 4. Upon reduction (grey line, point 2, Figure 4), GNR is reduced to GNRH4 2+ , which detaches from the electrode and partially dissolves in the electrolyte solution. Subsequent oxidation to GNRH4 2+ (point 3) redeposits some of the GNRs onto the electrode, now in improved electronic communication with GC, showing PCET behavior of a strongly-coupled system (points 4 and 5). In summary, we provide a mechanistic investigation of PCET in GNR/carbon electrode hybrids. Our work demonstrated that strong electronic coupling can be achieved between the molecular unit and the carbon support using non-covalent 𝜋 − 𝜋 stacking interactions via simple electrochemical treatment that involves cathodic/anodic cycling of as-deposited GNR. The presence of nitrogen-containing functionalities on GNR opens up the possibility to coordinate transition metals and develop a new type of heterogeneous electrocatalysts with molecularlevel control of the catalytic units. ## ASSOCIATED CONTENT Supporting Information Synthetic procedure, characterization of GNR, CVs, Pourbaix diagrams experimental and calculated data, XPS experimental and fitting details, reduction potential and pKa calculations, coordinates of optimized structures (PDF). The Supporting Information is available free of charge on the ACS Publications website. ## AUTHOR INFORMATION Corresponding Author *Ksenija D. Glusac: Department of Chemistry, University of Illinois at Chicago, Chicago, IL, glusac@uic.edu

chemsum

{"title": "Strong Electronic Coupling of Graphene Nanoribbons onto Basal Plane of Glassy Carbon Electrode", "journal": "ChemRxiv"}

a_simple_design_for_microwave_assisted_digestion_vessel_with_low_reagent_consumption_suitable_for_fo

## Abstract: The objective of this work is to prepare a cost-effective, low reagent consumption and high performance polytetrafluoroethylene (PTFE) vessel that is capable to work in domestic microwave for digesting food and environmental samples. The designed vessel has a relatively thicker wall compared to that of commercial vessels. In this design, eight vessels are placed in an acrylonitrile butadiene styrene (ABS) holder to keep them safe and stable. This vessel needs only 2.0 mL of HNO 3 and 1.0 mL H 2 O 2 to digest 100 mg of biological sample. The performance of this design is then evaluated with an ICP-MS instrument in the analysis of the several NIST standard reference material of milk 1849a, rice flour 1568b, spinach leave 1570a and Peach Leaves 1547 in a domestic microwave oven with inverter technology. Outstanding agreement to (SRM) values are observed by using the suggested power to time microwave program, which simulates the reflux action occurring in this closed vessel. Taking into account the high cost of commercial microwave vessels and the volume of chemicals needed for various experiments (8-10 mL), this simple vessel is cost effective and suitable for digesting food and environmental samples.Most trace analytical techniques need a homogeneous liquid sample to operate. Therefore, solid samples should be completely dissolved and digested prior to analysis. On the other hand, the digestion step is the Achilles' heel of trace element analysis. If the digestion technique is too gentle, it will not break the sample matrix, therefore, the target analytes will not be released into the solution 1 . However, a method that is too harsh will not only destroy the sample matrix, but might also cause the loss of the analyte of interest. This is particularly the case with metals such as antimony, arsenic or tin 1 .One of the techniques that is often used for digestion is ashing, which requires burning of the sample matrix until it becomes ash. The ash is usually soluble in an acid solution. This technique may also cause loss of analytes at high temperatures and yield poor results. Another difficulty is that some matrices do not easily turn into ash and may not even completely burn. Therefore, some of the target elements may not be taken into the solution 1-3 .Fusion is another technique used for difficult matrices. This technique is very labor intensive and could be very expensive; furthermore, the high salt load and contamination problems due to the fluxing agent are the main disadvantages of this technique 1-3 .Microwave digestion technique is an efficient, fast, and reproducible sample preparation method. In this method, the reaction timescale is dramatically reduced. Moreover, the potential for contamination decreases compared to open digestion techniques [3][4][5][6][7][8][9][10][11][12] . The first application of microwave in sample preparation was reported in 1975. In this study, a domestic microwave oven was employed to digest biological samples in an Erlenmeyer flask 13 . Soon after, scientists applied closed vessels to digest samples in higher temperatures and pressures [14][15][16][17][18][19][20][21][22][23][24] . All the closed vessels are armed with a safety system, which is either a safety membrane or a safety valve that opens instantly to control any increase in pressure by releasing some of the gases 1 . Therefore, volatile elements may be lost when gases are released from the closed vessels and this could be a reason for the lower recoveries in digestion 1 . The amount of chemical reagent typically used for digestion of biological samples with commercial vessel/ microwave in other studies is about 8-10 mL 25,26 ; nevertheless, in this study, a maximum of 3.0 mL of chemical reagents are needed (2.0 mL of nitric acid and 1.0 mL of hydrogen peroxide). Furthermore, the aim of this research was to design a simple PTFE vessel, which is safe to be used in a domestic microwave, as commercial ones might not be available for all due to its high cost. Finally, the performance of this design is evaluated with an ICP-MS instrument by analyzing the milk powder NIST 1849a, rice flour NIST 1568b, Peach Leaves NIST 1547 and Spinach Leaves NIST 1570a. ## Results To evaluate the accuracy and applicability of the designed vessel in the proposed power to time program, the SRMs of milk powder, rice flour, peach and spinach leaves were digested and analyzed. The results found for these SRMs are compared with their corresponding certified values and reported in Tables (1-4). ## Discussion The results observed in Tables 1 and 2 shows excellent recoveries for most of the elements that exist in the milk powder and rice flour SRM samples. Nevertheless, the results shown in Tables 3 and 4 for the analysis of peach and spinach leaves SRMs, respectively revealed that the recoveries for Al, Cu, Zn, V, Mo, and Ni are not as good as the other elements. The lower recovery in this case might be contributed to the silicon content of peach and spinach leaves 6 ; these elements are not completely released in microwave digestion as jelly silicon contents can keep part of the analytes while digestion is in progress 6,27 . This reduction in the recovery of elements, which was mentioned above, is directly related to the amount of silicon content of the leaves. The mean silicon content of peach NIST 1547 and spinach NIST 1570a leaves has been reported to be 979 and 1137 mg kg −1 , respectively 6 . It is clear that lower recoveries for Al, Cu, Zn, V, Mo, and Ni observed in case of the peach and spinach leaves are caused by their higher content of silicon. As explained, in order to prevent over pressure condition in the designed vessel, a domestic microwave oven was used, facilitated with inverter technology. In conventional microwave ovens, the power of microwave is always the maximum nominal power of the oven. Hence, for generating lower powers of radiation during the time of operation, the device was turned on and off successively so that the approximate required power of radiation can be transferred into the object. Nevertheless, in the microwave oven with inverter technology the exact required power of radiation can be applied continuously on the target. The amount of chemical reagent typically used for digestion of biological samples is about 8-10 mL 25,26 ; however, in this study, a maximum of 3.0 mL of chemical reagents (2.0 mL of nitric acid and 1.0 mL of hydrogen peroxide) were used to digest 0.10 g of milk powder, rice flour, spinach and peach leaves SRMs. The recommended procedure for the digestion of SRM 1849 -Infant/Adult Nutritional Formula is digestion of 1.00 g of this sample with 10.0 mL of HNO 3 for the 75 mL MARSXpress vessel 28 . The results reported by the CEM corporation for this SRM are given in Table 5. As it is clear the Agreement to (SRM) values (%) reported by recommended procedure of CEM corporation are relatively poorer than the results obtained in this study. This can be attributed to the amount of (SRM) sample, which is 1.0 g, and the amount of recommended HNO 3 , which is 10.0 mL. This amount of HNO 3 and (SRM) sample can cause overpressure in the digestion vessel and activation of safety valve. Therefore, because of releasing over-pressurized gases, some of the target samples will be lost and this is the reason for poorer results reported by CEM corporation. ## Conclusions A simple microwave digestion vessel design was introduced for digesting some food and environmental samples; the accuracy and applicability of this design was evaluated by the analysis of milk powder, rice flour, peach leaves, and spinach leaves SRMs. In this study a power to time microwave program, which simulates the reflux action occurring in this closed vessel were applied. Acceptable agreement to certified values have been observed for the analysis of milk powder and rice flour SRMs for most elements. However, it was observed that agreement to (SRM) values for some elements in peach and spinach leaves SRMs were not as good as milk powder and rice flour SRMs. It is concluded that this reduction in the recoveries as observed in some of the elements could be attributed to the silicon contents of these leaves. Moreover, the amount of chemicals used for the digestion in this vessel was reduced to 3.0 mL instead of 5 to 10 mL recommended by commercial vessels. By taking into account the price of commercial vessels and ultra-pure chemicals, these vessels can be considered as more cost effective and more environmentally friendly. ## Methods The ICP-MS 7500ce (Agilent, USA) with an octopole reaction system (ORS) was used in this research. The Octopole Reaction System (ORS) is a Collision-reaction cell CRC containing an octopole ion guide in a stainless steel vessel which pressurized with a gas. Collision-reaction cell CRC is a technology to reduce or eliminate the effect of interferences coming from polyatomic species. This can be done by passing the ion beam ( just before the quadrupole mass filter) through a cell that can be pressurized with a collision gas of either a reactive gas (e.g., H 2 , NH 3 , O 2 ) or an inert collision gas (e.g. He). Argon gas that was used throughout the experiment is of spectral purity of (99.999%). Each day prior to the beginning of the experiment, the instrument was tuned with 1.0 μ g L −1 (Agilent, USA) tuning solution containing Li, Co, Y, Ce and Tl in 2.0% (v/v) HNO 3 and 0.50% (v/v) HCl to cover the entire mass range to assure proper sensitivity. The settings of the instrument are reported in Table 6. A 1000 W (Panasonic, Japan) microwave oven NN-ST651M, 32 L with inverter and ∅ 340 mm turntable was used throughout the experiments. An (Elga Purelab Uhq II UK) system was used to produce ultra-pure water with resistivity more than 18 MΩ.cm. A laboratory made polytetrafluoroethylene (PTFE) vessel was designed with a relatively thicker wall compared to commercial vessels while a silicone based polymer O-ring was used as a safety valve. The suggested vessel has been patented in Iran with patent no. 71522-1390/06/26, 2011 (Rima Instrument, Iran). In this design, eight vessels were placed onto an acrylonitrile butadiene styrene (ABS) holder to keep them safe and stable. The top view of the PTFE vessel is shown in Fig. 1(a); Fig. 1(b) represents the PTFE vessels with their ABS casing. The geometrical details of the vessel and their casing are shown in Fig. 1(c,d) respectively. The reagents used for the analysis and digestion of samples are 60% ultrapure nitric acid and 31% ultrapure hydrogen peroxide (Merck, Germany). The standard reference materials were milk powder (Infant/Adult Nutritional formula obtained from NIST U.S department) 1849a, NIST rice flour 1568b, NIST peach leave SRM 1547 and NIST spinach leave SRM 1570a. For the calibration plot, a multi element standard Agilent with concentration of 10 mg L −1 for Ag, Al, As, Ba, Be, Cd, Co, Cr, Cu, Mn, Mo, Ni, Pb, Sb, Se, Tl, V, Zn, Th and U, 1000 mg L −1 for Ca, Fe, K, Mg, Na and Sr were used. Concentrations of 50, 100, 300, and 1000 μ g L −1 were used for five elements of Ca, Fe, K, Mg, Na and Sr, and concentrations of 0.5, 1, 3 and 10 μ g L −1 for the rest of the elements. In Table 7 the limit of detection LOD, limit of quantitation LOQ, correlation coefficient, R 2 , and selected modes of gas are given. The LOD and LOQ values were obtained from S m 3 bl and S m 10 bl respectively, in these equations S bl is the standard deviation of blank signal and m is the slope of calibration curve. 0.10 g of each SRM sample was weighed and transferred into a PTFE vessel, then the vessels were put under the hood and 2.0 mL of ultra-pure 60% nitric acid plus 1.0 mL of ultra-pure 31% hydrogen peroxide were added into each vessel and the vessels were closed. A power to time microwave program was used for the digestion of standard reference materials. The details of power to time program used in this experiment are as shown in Fig. 2. This program was obtained by several trial and error varying the time to power program with constant amount of (SRM) samples (0.1 g) and digestion solution volume (3.0 mL) to gain maximum agreement with (SRM) values. In this program, the power of the microwave oven was increased stepwise, first to 167 W for 2 minutes and then it was increased to the maximum 333 W for 2 more minutes; afterwards the power was decreased to zero for one minute. This trend was repeated five times for milk powder and rice flour (SRMs) and seven times for peach and spinach leaves (SRMs). After digestion, the vessels were put under the hood to cool down. The contents were filtered through a 0.45 μ m PTFE, if necessary (only peach and spinach leaves SRMs), and then transferred into a 50 ml polypropylene volumetric flask and diluted with ultrapure water to the marked level. The diluted samples were stored in polyethylene vials until the time of analysis by ICP-MS. For cleaning, 3.0 mL of analytical grade nitric acid was poured into each vessel. After closing the vessels, a similar program with that used for digestion was applied to wash the vessels for 15 minutes. This vessel armed with silicone O-ring to release any over pressure that may occur because of applying higher power of microwave radiation or longer time of applying radiation. However, in this power to time microwave program, it was tried to apply power and time of radiation to digest different standard reference materials in a way to prevent the overpressure condition. This was archived by several experimental cycles. It has been reported that, when water is heated in a closed vessel microwave, the internal pressure is lower than that in closed vessels heated by conventional methods 4,26 . This is due to the vessel materials and the heating mechanism. Since the closed vessel microwave and their outer casing are microwave transparent, they remain relatively cool during the heating process. Therefore, the vessel's wall becomes cooler and causes water molecules to remove from vapor phase 4,26 . Moreover, it was reported that simultaneous application of air flow outside the digestion vessel can improve the efficiency of digestion in the microwave oven 4 . Consequently, this enhancement in condensation rate causes a decrease in internal pressures at higher temperatures 4,26 . This is similar to a reflux action, which explains the better digestion. With the same conclusion, the cooling and heating period, shown in Fig. 2, simulates the reflux action and causes very good agreement to (SRM) values. These vessels armed with silicone O-ring to release any over pressure that may occur because of applying higher power of microwave radiation or longer time of applying radiation. The PTFE vessels used in this research were equipped with a silicon-based polymer with high tensile and temperature resistance properties. In the case of enhancement of pressure at high temperature, this silicone rubber deforms to release the excess pressure. Figure 3(a) shows the silicone rubber O-ring before deformation and Fig. 3(b,c) show the silicone rubber after deformation, caused by high pressure and temperature. It is necessary to mention here that although this vessel is armed with the safety system, the amount of sample (0.1 g) and the amount of reagents used for the digestion (3.0 mL) with the recommended time power program will not cause the activation of safety system of this vessel, therefore, better recoveries can be observed in this investigation.

chemsum

{"title": "A simple design for microwave assisted digestion vessel with low reagent consumption suitable for food and environmental samples", "journal": "Scientific Reports - Nature"}

iridium_complexes_catalysed_the_selective_dehydrogenation_of_glucose_to_gluconic_acid_in_water

## Abstract: We describe an unprecedented catalytic dehydrogenation of glucose by homogeneous catalysts. Iridium (III) complexes containing the fragment [Cp*Ir(NHC)] 2+ (NHC = N-heterocyclic carbene ligand) are shown to be very active and highly selective catalysts for the dehydrogenation of glucose to gluconic acid and molecular hydrogen. Glucose is converted to gluconic acid at a catalyst loading of 2 mol%, at reflux in water, without additives and with a selectivity of over 95%. Experimental evidence obtained by 1 H NMR spectroscopy and mass spectrometry (ESI/MS) reveals the formation of iridium coordinated to glucose and gluconic acid species. A plausible mechanism is proposed, based on the experimental evidence and supported by DFT calculations. † Electronic supplementary information (ESI) available: Experimental procedures, the HPLC-MS/MS method, crystallographic data and DFT calculations. CCDC 1850450 (2), 1850451 (3) and 1850452 (4). For ESI and crystallographic data in CIF or other electronic format see ## Introduction The production of chemicals from fossil resources is not a sustainable process because it generates considerable amounts of CO 2 . As an alternative, the conversion of biomass is an attractive and sustainable synthetic protocol for the green manufacture of profitable organic compounds. Among others, glucose is an abundant and renewable feedstock for the production of initial platform chemicals. Traditional systems for glucose transformation use harsh conditions such as high temperature or strong acids. The main challenge in the use of glucose is to control selectivity due to the presence of many functional groups. For instance, in the oxidation of glucose the formation of different acids and keto acids is normally observed. In the last few years, highly active heterogeneous catalysts have been developed using O 2 as the oxidant. The Au/TiO 2 based materials are among the most selective for the conversion of glucose to gluconic acid. However, industrial conversion of glucose is carried out by enzymatic biocatalysis. Glucose is selectively converted into gluconic acid by the enzyme glucose oxidase obtained from the fungus Aspergillus niger. 15 Gluconic acid is an important building block in the production of other chemicals and it is used in the pharmaceutical and food industries. 16 The production of gluconic acid by enzymes has limitations imposed by the use of these labile natural catalysts, such as the pH, oxygen pressure and catalyst recovery. In this manuscript, we describe an alternative process for the transformation of glucose into gluconic acid using iridium molecular complexes (Scheme 1). Traditional glucose oxidation involves the presence of air or oxygen with the concomitant formation of water. In contrast, in the catalytic dehydrogenation of glucose, water is the origin of both OH − groups of the carboxylic acid and the proton that releases molecular hydrogen. There are precedents in the literature describing the conversion of alcohols and aldehydes into carboxylic acids where water plays a key mechanistic role. We have experimental evidence in favour of the formation of iridium coordinated to glucose and gluconic acid species. Based on these observations as well as DFT calculations, a feasible reaction mechanism is proposed. ## Results and discussion The catalytic dehydrogenation of glucose to gluconic acid was evaluated by using iridium(III) complexes containing the fragment [Cp*Ir(NHC)] (NHC stands for N-heterocyclic carbene). Iridium catalyst precursors were prepared by standard methods. The [Cp*Ir(NHC)(SO 4 )] complex 2 and the [Cp*Ir (NHC)(H 2 O) 2 ](OTf ) 2 complex 3 were obtained from 1 using silver salts as halide abstractors and water as solvent (Scheme 2). A characterization study of complexes 2 and 3 by single crystal X-ray diffraction confirmed the chelating coordination mode of the sulfate ligand and the coordination of water molecules. In a typical catalytic experiment, the iridium complex was added to a solution of glucose in deionized water and the mixture was heated at reflux for an appropriate time. The progress of the reaction and product distribution was monitored by ESI/MS, NMR and HPLC. Given the potentially wide product distribution that can result from glucose oxidation, preliminary monitoring of the reaction outcome was per-formed by negative ESI-MS. ‡ The disappearance of the peak due to deprotonated glucose [glucose-H] − (m/z 179) concomitant with the sole formation of a new species at m/z 195 anticipated good selectivity toward gluconic acid. Details of the method optimization and chromatographic conditions for quantitation purposes are given in the ESI. † 29,30 The HPLC-MS/MS method consists of taking small aliquots of the crude reaction mixture, diluting them with water/methanol (95 : 5) and directly injecting them into the HPLC-MS system. This methodology allows to (i) monitor glucose and gluconic acid without unnecessary work-up, minimizing the number of errors from sample pre-treatment steps, and (ii) obtain detailed temporal reaction profiles and rapid reaction optimization. Examples of the reaction profiles are shown in Fig. 1. The results showed that catalyst precursors 1 and 2 were active in the conversion of glucose to gluconic acid without additives (Table 1). This transformation is a green process carried out under mild reaction conditions using water as solvent. Even more interesting is the formation of only gluconic acid among all the potential acids and ketoacids directly obtained from glucose (entries 3 and 4). We observed that glucose dehydrogenation is sensitive to pH variation. The addition of different amounts of base led to a significant decrease in gluconic acid yield accompanied by glucose degradation to unidentified products (entries 2, 5 and 6). The mass balance of the reactions could not be rationalized under basic conditions and glucose dehydrogenation was not further investigated in basic media. In this context, it is worth noting that under basic conditions in the absence of any catalyst (entry 2), glucose degradation without the formation of gluconic acid was also observed at identical reactions times. This suggests that both the intrinsic glucose degradation pathways caused by the presence of base and gluconic acid formation are simultaneously coexisting at basic pH values. The situation is completely reversed under acidic conditions. The use of 0.25 eq. of H 2 SO 4 or HCl vs. glucose significantly accelerated glucose dehydrogenation without penalizing selectivity to gluconic acid (Fig. 1b and c). For example, under these conditions, 40% yield was obtained after 30 min whereas for the reaction at neutral pH, 40% yield was achieved after 6 h. Acidification enhanced glucose conversion and gluconic acid yield that increased from 44 to 88%. Further increasing the H 2 SO 4 content did not substantially influence either the reaction yield (entries 7-10) or the kinetics of gluconic acid formation. The use of different amounts of HCl displayed identical reaction outcomes to those observed for H 2 SO 4 . Control experiments under these reaction conditions (entry 1) revealed that the iridium complexes are essential for glucose dehydrogenation. We also explored the potential application of iridium catalysts in the selective conversion of ( poly)carbohydrates into gluconic acid. This represents a hot topic today with the main goal of affording valorised organic building-block chemicals, especially using water as solvent. 31 Starch was selected as a target ( poly)carbohydrate (Fig. 2). Starch is a polysaccharide made of glucose units bonded by β-glycosidic bonds. We used starch composed of 20-25% of amylose and 75-80% of amylopectin. In a typical reaction, starch was treated under the general conditions described for the dehydrogenation of glucose. Starch was dissolved in acidic water using 1 eq. of H 2 SO 4 per glucose unit and heated at 100 °C in the presence of 2 mol% of catalyst 1. The reaction proceeded slowly due to the hydrolysis of glycosidic bonds and after 50 h, the conversion was 80% (considering all starch as glucose units). At that time, the yield of gluconic acid was 50% and the yield of glucose was 30%. The initial results reveal the potential application of iridium complexes as selective catalysts for the transformation of biomass into added-value chemical products. The success of the process lies in the high stability of catalyst 1 in acidic medium. ## Mechanism In order to understand the general mechanism of the dehydrogenation of glucose to gluconic acid, we first addressed the a Conversions and yields determined using HPLC-MS/MS. b Selectivity to gluconic acid was calculated as (moles of gluconic acid produced/ moles of glucose reacted) × 100. chemical speciation of 1 in aqueous solution and under catalytic conditions. Compound 1 is moderately soluble in water and different species coexist. The 1 H NMR spectrum in D 2 O features broad signals for the NHC and the Cp* moieties, consistent with an equilibrium by chloride-water ligand exchange (Fig. S5 †). This equilibrium has previously been established for other piano-stool complexes. 32 The ESI/MS results also indicated that chloride cleavage and subsequent water coordination at the vacant sites took place upon dissolving Moreover, DFT calculations revealed that deprotonation of water molecules coordinated to organometallic complexes, such as in 3 2+ , is a feasible process. 33 Consequently, aqua and hydroxo species can exist in equilibrium in aqueous solutions, and it is reasonable to consider that this equilibrium can be shifted to the aqua 3 2+ species under acidic conditions. This explains the faster glucose conversion in acidic medium than at neutral pH and suggests that the 3 2+ dication is the active catalyst. For these reasons, we consider the aqua complex as the experimental input for the DFT studies. Next, we addressed the progress of the catalytic reaction by ESI/MS. The versatility of this technique for detecting shortlived or transient intermediates in mechanistic studies is well documented. In the present work, aliquots at different time intervals were taken, diluted with water and directly injected into the ESI mass spectrometer. Glucose coordination to the iridium complex was evidenced at the initial stages of the reaction by the m/z values, the isotopic pattern and fragmentation upon CID of species present in the mixture. Species formulated as [Cp*Ir(NHC)(glucose)] 2+ (m/z 302) and [Cp*Ir (NHC)(glucose-H)] + (m/z 603) were detected based on their m/z value, isotopic pattern (Fig. 3) and CID spectra (Fig. 4). It is worth noting that glucose binding at the Ir centre in the [Cp*Ir (NHC)(glucose)] 2+ species was remarkable since glucose backbone fragmentation was preferred over neutral glucose release under CID conditions, suggesting strong glucose coordination. As the reaction proceeds, a new Ir-containing species could be detected in situ by ESI/MS monitoring, formally corresponding to gluconic acid ligation to the Ir complex. A peak at m/z 619 formulated as [Cp*Ir(NHC)(gluconic acid-H)] + was observed according to its m/z value, isotopic pattern and CID mass spectrum (Fig. S4 †). A plausible mechanism for the dehydrogenation of glucose is proposed, based on the above experimental evidence and complemented by density functional theory (DFT) calculations (Scheme 3). The initial hypothesis came from mechanistic investigations into the synthesis of carboxylic acids from alcohols in water, 17,18, and from the conversion of aldehydes to carboxylic acids in the "Aldehyde-water shift" reaction. 20 For the theoretical studies, a simplified model for the substrate, namely MeCH(OH)CHO ≡ 2-hydroxypropanal, was considered, minimizing the computational cost that would arise from the consideration of multiple glucose conformations. The cycle starts with the formation of the aqua complex [Cp*Ir(NHC)(H 2 O) 2 ] 2+ (3) on the basis of the aqueous speciation described above. In order to prove that complex 3 is a competent catalytic active species, we performed the dehydrogenation of glucose using isolated 3 as a CF 3 SO 3 − salt. The results showed that complex 3 without additives is as efficient as complex 1 under acidic conditions for the conversion of glucose to gluconic acid in water. The H 2 O ligands in complex 3 are exchanged with glucose leading to species A. Species A was evidenced by ESI/MS and according to DFT calculations displays a chelate coordination mode via the carbonyl group and the hydroxo group. We hypothesize that this ordered cyclic arrangement not only confers remarkable stability to species A (as evidenced by CID experiments), but also is responsible for the propensity of the substrate to be converted selectively into gluconic acid. Experimental attempts to isolate crystals of glucose coordinated to 3 were unsuccessful. However, in a parallel experiment we observed the chelate coordination of a model substrate containing an ortho-substituted aldehyde and a hydroxyl group. The reaction of complex 1 with salicylaldehyde under basic conditions led to complex 4. Deprotonated salicylaldehyde was coordinated in a chelate mode as observed by single crystal X-ray diffraction (ESI, section S4 †). This result lends indirect support to the proposed glucose chelate coordination mode. Then, a nucleophilic attack of water on the carbonyl of the coordinated glucose produces a gem-diol species B through a low barrier transition state 1 (12.9 kcal mol −1 ). Similar species have been detected and/or postulated in the conversion of alcohols to carboxylic acids in aqueous medium. If the nucleophilic attack on a coordinated carbonyl is produced by an amine or alcohol, the formed species is an hemiaminal or hemiacetal, respectively. These species are proposed as key intermediates in many transformations and are difficult to detect and/or isolate. Gusev et al. detected the formation of such species by ESI/MS using ruthenium and osmium catalysts. 43,44 The gem-diol species rearranges to produce C that releases gluconic acid and a solvated proton (H 3 O + ) and forms an iridium hydride (D). Then, the iridium hydride (D) is protonated, forming a dihydrogen species that releases hydrogen and regenerates the catalyst. Analysis by gas chromatography confirmed the gas evolution. As the reaction proceeds, a gluconic acid-coordinated species is also observed in the ESI/MS (Fig. S4 †). This species is an out-of-cycle intermediate and results from the accumulation of gluconic acid in the reaction medium. Related acetate-ligated species have been identified and isolated in the pioneering examples of dehydrogenative oxidation of alcohols to carboxylic acid using Ru-based catalysts. 17 The highest energy step, in this case also the rate determining step, corresponds to the hydride transfer from the gem diol to the iridium through TS2 (26.3 kcal mol −1 ), forming 2-hydroxypropanoic acid and H 3 O + (Fig. 5). A similar high-energy step has been proposed in the "aldehyde-water shift reaction" using a bipyridine iridium complex. 20 The energy profile diagram consists of a low barrier TS1 followed by a high barrier TS2 that support our experimental observations by ESI/MS. Although other pathways could be considered, the experimental results and the calculations support an energetically feasible route for the dehydrogenation of glucose as described in Scheme 3. ## Conclusions Iridium complexes bearing NHC ligands are efficient catalysts for the production of gluconic acid from glucose in water and under mild reaction conditions. Glucose dehydrogenation in water represents a convenient methodology for the production of added-value chemicals from biomass. Using iridium complexes, the process is highly selective and we have not observed the formation of other acids or ketoacids. Experimental evidence obtained by 1 H NMR spectrometry and mass spectroscopy (ESI/MS) shows the formation of catalytic intermediates that contain glucose coordinated to iridium. The results support an unprecedented catalytic dehydrogenation of glucose by homogeneous iridium complexes with the concomitant formation of hydrogen gas. A plausible mechanism is proposed based on the observation of these species and sustained by DFT calculations. ## General details Complex 1 was prepared according to previously reported procedures. 45 Complex 3 was previously synthesised using wet CH 2 Cl 2 as solvent and in this manuscript, we report the synthesis of it using water as solvent. 46 D-Glucose (99%), D-gluconic acid (50% aq. sol), salicylaldehyde (98%), K 2 CO 3 (99%), AgSO 3 CF 3 (99%) and Ag 2 SO 4 (99%) were purchased from commercial suppliers and used as received. HPLC-grade methanol (MeOH) and formic acid (HCOOH, content >98%) were purchased from Scharlab (Barcelona, Spain). HPLC-grade water was obtained from distilled water passed through a Milli-Q water purification system (Millipore, Bedford, MA, USA). Nuclear magnetic resonance (NMR) spectra were recorded on Bruker spectrometers operating at 300 or 400 MHz ( 1 H NMR), 75 or 100 MHz ( 13 C{ 1 H} NMR) and 377 MHz ( 19 F NMR), respectively, and referenced to SiMe 4 or CCl 3 F (δ in ppm and J in Hz). NMR spectra were recorded at room temperature with the appropriate deuterated solvent. ## Methods General procedure for catalytic experiments. Catalytic assays were performed under aerobic conditions in 50 mL 2-neck round bottom flasks. In a typical reaction, iridium catalyst (2 mol%) was charged into a glucose solution (20 mL, 4 mg mL −1 ) with additives (H 2 SO 4 , HCl, and NaOH) if needed. The reaction flask was introduced into a pre-heated oil bath at 110 °C. At selected times 250 µL aliquots were collected directly from the flask with a syringe, diluted and kept at 4 °C until HPLC-MS/MS analysis. At the end of the reaction, pH was adjusted to 12 by adding NaOH. The solvent was evaporated and the crude was analysed by 1 H-NMR spectroscopy using deuterium oxide as solvent. Electrospray ionisation mass spectrometry (ESI/MS) and collision induced dissociation (CID) experiments. ESI-MS studies were performed using a QTOF Premier instrument equipped with an orthogonal Z-spray-electrospray interface (Waters, Manchester, UK) operated in the V-mode at a resolution of ca. 10 000 (FWHM). The drying and cone gas was nitrogen set to flow rates of 300 and 30 L h −1 , respectively. A capillary voltage of 3.5 kV was used in the positive ESI(+) scan mode. The cone voltage was adjusted to a low value (typically U c = 5-15 V) to control the extent of fragmentation in the source region. Chemical identification of the Ir-containing species was facilitated by the characteristic isotopic pattern at natural abundance of Ir and it was carried out by comparison of the isotope experimental and theoretical patterns using the MassLynx 4.1 software. Typically, aqueous solutions of compound 1 were stirred under catalytic conditions and aliquots were extracted at the required time intervals, diluted with water to a final concentration of 5 × 10 −4 M (based on the initial Ir concentration) and directly introduced into the mass spectrometer. For CID experiments, the cations of interest were mass-selected using the first quadrupole (Q1) and interacted with argon in the T-wave collision cell at variable collision energies (E laboratory = 3-15 eV). The ionic products of fragmentation were analysed with the time-of-flight analyser. The isolation width was 1 Da and the most abundant isotopomer was mass-selected in the first quadrupole analyser. Procedure for hydrogen identification. A 100 mL Schlenk flask was charged with glucose (80 mg, 0.44 mmol), H 2 SO 4 (5.9 µL, 0.11 mmol), catalyst 1 (4.3 mg, 8.8 × 10 −3 mmol) and 2 mL of water. The Schlenk flask was sealed and introduced into a pre-heated oil bath at 110 °C. After 2 h, a 25 mL sample of the generated gas was collected and the hydrogen content was qualitative analysed by gas chromatography (GS-MOL 15 meters column ID 0.55 mm TCD from J&W Scientific; N 2 as a carrier gas). ## Synthetic procedures General procedure for the synthesis of compound 2. Ag 2 SO 4 (82 mg, 0.26 mmol) was added to a solution of 1 (110 mg, 0.22 mmol) in distilled H 2 O (50 mL). The reaction mixture was introduced into a pre-heated oil bath at 70 °C and stirred overnight (16 h) in the dark. The reaction was filtered over Celite and the solvent was evaporated to dryness. The residue was dissolved in CH 2 Cl 2 (30 mL), dried over MgSO 4 and filtered over Celite. The resulting yellow solution was evaporated to dryness. Complex 2 was precipitated using hexane (20 mL), filtered and dried (yield 96.5 mg, 83%). General procedure for the synthesis of compound 3. AgCF 3 SO 3 (52 mg, 0.20 mmol) was added to a solution of 1 (50 mg, 0.10 mmol) in distilled H 2 O (10 mL). The reaction mixture was introduced into a pre-heated oil bath at 40 °C and stirred overnight (16 h) in the dark. The reaction was filtered over Celite and the solvent was evaporated to dryness to afford complex 3 as a yellow-brown solid (yield 56.4 mg, 73%). General procedure for the synthesis of compound 4. Salicylaldehyde (16 µL, 0.15 mmol) was added to a solution of 1 (50 mg, 0.10 mmol) in acetone (10 mL) in the presence of K 2 CO 3 (21 mg, 0.15 mmol). The resulting mixture was stirred overnight at room temperature. The solvent was removed and the crude material was purified by column chromatography on silica gel using CH 2 Cl 2 -acetone (1 : 1 v/v) as eluent to afford 4 as a yellow solid (yield. 37 mg, 68%). ## Conflicts of interest There are no conflicts to declare.

chemsum

{"title": "Iridium complexes catalysed the selective dehydrogenation of glucose to gluconic acid in water", "journal": "Royal Society of Chemistry (RSC)"}

efficient_chemical_fixation_and_defixation_cycle_of_carbon_dioxide_under_ambient_conditions

## Abstract: Chemical fixation of CO 2 as a C1 feedstock for producing value-added products is an important postcombustion technology reducing the CO 2 emission. As it is an irreversible process, not considered for the CO 2 capture and release. Overall, these chemical transformations also do not help to mitigate global warming, as the energy consumed in different forms is much higher than the amount of CO 2 fixed by chemical reactions. Here we describe the development of re-generable chemical fixation of co 2 by spiroaziridine oxindole, where CO 2 is captured (chemical fixation) under catalyst-free condition at room temperature both in aqueous and non-aqueous medium even directly from the slow stream of flue gas producing regioselectively spirooxazolidinyl oxindoles, a potential drug. The CO 2 -adduct is reversed back to the spiroaziridine releasing CO 2 under mild conditions. Further both the fixationdefixation of CO 2 can be repeated under near ambient conditions for several cycles in a single loop using a recyclable reagent. Means of viable development, typically relying on more sensible resource management, is a conceit challenge in front of modern human society. Sustainability level in recent economic growth requires a massive improvement as it is far from an adequate level. According to the data released by Intergovernmental Panel on Climate Change (IPPC 2018), global surface temperature has mounted by approximately 1.5 °C from 1880 to 2018, which is a phenomenon caused by anthropogenic activities, predominantly greenhouse gases like CO 2 emissions from fossil carbon to accomplish the escalating energy demand. Under this circumstances, melting of thousand years old glaciers, desertification of fertile land, rise in ocean water level and acidification of ocean water had caused enormous detriment to diverse ecological environment 1 . Scientific and technical advancements to curve atmospheric CO 2 concentration via limiting industrial emission and use CO 2 as an alternative fuel source in the renewable energy sector, had been a recurrent course of study for past few years . The reduction of CO 2 can be considered as a typical cohesive technology to rise artificial efficiency in producing various valuable hydrocarbons like formic acid, methanol, methane, and C2-C4 olefins . Several fascinating integrated protocols have freshly been reported for hydrothermal and photochemical CO 2 reduction, e.g., metal/metal oxide redox reaction based solar two-step water-splitting thermochemical cycle for CO 2 reduction via hydrogen generation . Alongside, chemical fixation of CO 2 has gained substantial importance in synthetic chemistry because CO 2 could be used as a benign, abundant, inexpensive, and renewable C1 reserve to yield a variety of value-added chemicals e.g. esters, amides, aldehydes, carboxylic acids, alcohols, organic carbonates and 2-oxazolidinones, etc . In particular, synthesis of therapeutically cherished and synthetically convenient five-membered cyclic urethanes such as oxazolidinones via cycloaddition of CO 2 with aziridines has become one of the most promising approaches in this area, because this process possess 100% atom efficiency, which exactly matches one of the most substantial criteria of green chemistry . Despite being an admirable strategy to chemically capture and recycle CO 2 , most of these protocols suffer from high energy demand and utilize costly catalysts/ionic liquids to achieve ambient or near ambient condition for CO 2 fixation, even from highly enriched CO 2 source . However, emissions from thermal power plants contain numerous gaseous components like SO 2 , NO 2 along with CO 2 41 . In these context, post-combustion CO 2 capture, release, and storage (CCS) had been the most abundantly used protocols for CO 2 purification from industrial exhausts. Various strategies are being industrialized for capture, release and storage (CCS) of CO 2 from gas streams, where gas-solid adsorption by metal-organic frameworks, gas-liquid chemiabsorption by amines and carbonation by quick/slacked lime are notable . However, chemical fixation of CO 2 from contaminated sources under mild conditions to produce industrially vibrant chemicals and products faces great defies because of two main reasons: (1) the high ionization potential (IP), and (2) the negative adiabatic electron affinity (EA) of carbon dioxide. Therefore, most of the reports use harsh reaction conditions to overcome the high thermodynamic stability and chemical inertness of carbon dioxide. Hence, the development of a costeffective and robust protocol for CO 2 capture, storage, and release in ambient conditions along with utility is highly desirable. Further, the chemical fixation is an irreversible process producing stable covalent compounds and thus, till now it could not be utilized for CO 2 capture and release. It might be a potential CCS protocol as it would produce valuable chemicals, provided the chemical fixation and the defixation (release) done under ambient conditions, the latter is an unmet challenge. Herein, we report the first regenerable chemical fixation, where CO 2 fixation by spiroaziridine oxindole under atmospheric pressure at rt (30 °C) without any catalyst producing stable spirooxazolidinone, a potential drug candidate , further reversed back (defixation) to the spiroaziridine releasing CO 2 under mild conditions. This fixation and defixation cycle can be repeated in a single loop for several times using a recyclable reagent. ## Results and discussion Uniqueness of spiroaziridine-and spirooxazolidinone oxindoles. CO 2 is an overall non-polar molecule, but the presence of net partial charges [O −δ -C +2δ -O δ ] makes its susceptibility to nucleophilic as well as electrophilic attack at carbon and oxygen, respectively. As a consequence, substrate such as epoxide and aziridine with both reactivity centres are suitable for the fixation of CO 2 20,36-40 . However, all these require high pressure, -temperature and/or catalyst/additive. Designing substrate with tuned reactivity may lower the pressure and temperature for the chemical fixation of CO 2 and may further facilitate the CO 2 release. We envision that NHfree spiroaziridine oxindole 1 could be a suitable substrate with desired reactivity as the presence of oxindole unit may enhance the nucleophilicity of aziridine-nitrogen via an electron-donating resonance effect of nitrogen of oxindole unit and/or its anchimeric assistance (Fig. 1), simultaneously these may increase the electrophilicity of the C3 center of oxindole via resonance structure 1A and/or the formation of intermediate 1B under neutral or mild basic condition . It is further envisioned that the presence of oxindole unit in spirooxazolidinone similarly will enable the release of CO 2 under acidic conditions as shown in Fig. 1) 53,54 . More importantly, the spirooxazolidinoyl oxindole is a potential drug candidate , so this CO 2 fixation could be excellent and cheap method for its production. Optimization of auto-chemical fixation of CO 2 by NH-free spiroaziridine 1a under ambient conditions. According to the presumption, we started our studies initially on synthesis of NH-free spiroaziridne oxindole 1a and its reactivity towards CO 2 under different conditions. We have developed a new and efficient method for the synthesis of NH-free spiro aziridine 1a from easily available amino alcohol 3a on successive treatment with chlorosulfonic acid (ClSO 3 H) in dioxane and aqueous KOH. The exclusive formation of NH-free spiroaziridine 1a was detected by MS and NMR analysis. With great delight, when a slow stream of CO 2 was passed through an aqueous dioxane solution of in situ synthesized spiroaziridine 1a at rt, within 30 min it produced exclusively CO 2 adduct, spiro oxazolidione 2a in excellent yield (Table 1, entry 1) without any catalyst. This might be the first report of catalyst-free spontaneous chemical fixation of CO 2 under ambient condition and also in aqueous medium. Instead of aqueous, solid KOH was also found to be suitable for the in situ synthesis of spiroaziridine 1a and subsequent fixation of CO 2 , but it took a bit more time than the aqueous-dioxane (entry 2). The dioxane was the optimized solvent for both in situ spiroaziridine formation and the chemical fixation of CO 2 . NaOH instead of KOH is also equally effective for the synthesis of spiroaziridine and subsequent CO 2 fixation (entries 3 and 4). Further, when in situ generated spiroaziridine was taken in ethyl acetate and treated with slow stream of CO 2 in absence of any base, it also gave the CO 2 -adduct within 1.5 h in 69% isolated yield (entry 5). Thus it can be concluded that the chemical fixation of CO 2 by spiroaziridine does not require base as a catalyst/promoter. Ultimately with our great delight, the auto-chemical fixation of CO 2 was successful with 12.5% CO 2 in N 2 as well as a stimulated coal flue gas (12.5% CO 2 , 7.5% O 2 and 80% N 2 ) without any appreciable loss in the yield of the adduct (entries 6-8). These took longer reaction time, may be due to low concentration and retention of CO 2 in solution. Defixation of CO 2 at near ambient conditions. We next sought to explore the possibility to regenerate the spiroaziridine via decarboxylation, which is an unmet challenge in CO 2 -chemical fixation. As per the presumption, the decarboxylation (CO 2 release) was initiated with the reaction of spiroxazolidinone in the presence of different Brǿnsted acids and the subsequent treatment of base to regenerate the spiroaziridine and its regeneration was quantified with the further chemical fixation of CO 2 leading to spirooxazolidione again. Both the CO 2 defixation and the fixation were optimized in dioxane and briefly summarized in Table 2. The regeneration of spiroaziridine 1a was detected when a dioxane solution of spirooxazolidinone was heated with triflic acid at 100 °C. The extend of formation of spiroaziridine was confirmed by its chemical fixation of CO 2 and it gave only 24% yield of the resynthesized spirooxazolidinone 2a (Table 2, entry 1). With our great delight, near quantitative formation of spiroaziridine 1a was achieved, when the compound 2a was heated only at 70 °C with HI followed by treatment with aqueous NaOH (Table 2, entry 4). This was revealed with the re-synthesis of spirooxazolidinone 2a with 94% of isolated yield. HBr was also found to act on at 70 °C, but it took longer time with incomplete conversion (entry 6). Further, to avoid the cumbersome procedure for the preparation of dioxane-HX, we Table 1. Optimization of in situ synthesis of spiroaziridine 1a and fixation of CO 2 . Chlorosulfonic acid (1.0 equiv.) was added slowly into the dioxane solution of 3a (100 mg, 0.521 mmol) and stirred at 70 °C. Reaction mixture was basified and a slow stream of carbon dioxide was passed through the solution until complete consumption of 1a. a GC-yield is determined using naphthalene as internal standard; the value in parenthesis referred to the isolated yield. b Spiroaziridine 1a was extracted with ethyl acetate and treated with slow stream of CO ## Mechanism of CO 2 defixation. In TfOH mediated decarboxylation of 2a (defixation of CO 2 ), the formation of spiroaziridinium ion 1a′ was detected by MS analysis prior to the treatment with base. However, in case of HI or NaI-H 3 PO 4 , exclusive formation of intermediate compound 3-(aminomethyl)-3-iodooxindole 4a′ and no 1a′ was observed by MS analysis prior to the reaction with base. The intermediate iodo-amine 4a′ was isolated and identified as N-tosyl compound 5a by MS and NMR analysis (Fig. 2). The intermediate compound 4a′ on treatment with base regenerated the spiroaziridine 1a. Its in situ formation was confirmed by MS and NMR analysis and further isolated as N-tosyl spiroaziridine 6a. A dioxane solution of spiro-oxazolidone 2a (100 mg, 0.46 mol) was heated under specified acidic conditions followed by treatment of base and then slow stream of CO 2 . a Conversion of 2a was determined by GC-MS analysis. b GC-yield is determined using naphthalene as internal standard; the value in parenthesis referred to the isolated yield. Excitingly the overall yield of spirooxazolidinone after five cycles was found to be excellent (overall GC yield 95% and isolated yield 90%). Again, if we deeply look into the chemical reactions involved during the release and capture of CO 2 of the process, NaI supposed to regenerate after the treatment of 4a′/1a′ with NaOH. So, in principle, NaI may be reused for the subsequent cycles. For the purpose, the first regeneration cycle with release of CO 2 was carried out as usual with the combination of NaI-H 3 PO 4 and NaOH and subsequent chemical fixation of CO 2 produced the spirooxazolidione. The subsequent cycles for the regeneration of spiroaziridine (CO 2 -release/defixation) and CO 2 -fixation were performed without further addition of NaI, only varying with the equivalent of H 3 PO 4 and NaOH (Fig. 4). Thrillingly these were smoothly continued for five cycles. It showed almost quantitative yield In some of the developed technologies, the sorbent (liquid or solid) loaded with the captured CO 2 is transported to a different vessel, where it releases the CO 2 (regeneration) either after being heated or after a pressure decrease or after any other change of conditions around the sorbent. The sorbent resulting after the regeneration step is sent back to capture more CO 2 in a cycle. This makes additional cost of the process. It will be desirable to conduct both CO 2 capture and the release in a single vessel, this is possible when both are near similar conditions. In our case, 70 °C was found to be optimum temperature for the CO 2 defixation. So, we further studied the temperature effect on CO 2 fixation. Interestingly, it showed a near horizontal line for the fixation at 5 °C, 30 °C, 50 °C, 60 °C and 70 °C, respectively, with > 95% yield in each case (Fig. 5). Inspired by the above findings of temperature effect on CO 2 fixation, we performed both CO 2 defixation and fixation at 70 °C and continued for five cycles. With our great delight, it showed almost quantitative yield of spirooxazolidinone 2a in each cycle and an excellent overall yield after five cycles. This chemical fixationdefixation (five) cycles at 70 °C are repeated for three times with a standard deviation of 0.47-1.70 (Fig. 6). The spirooxazolidonyl oxindoles are important bioactive compounds 21,22 . Thus further efforts are made to generalize the developed method for the synthesis of various spirooxazolidines by catalyst-free CO 2 fixation of in situ generated spiroaziridines (Fig. 7). Irrespective of N-protection-and substitution of arene moiety of the oxindole unit, all underwent smooth auto-chemical fixation of CO 2 providing the excellent isolated yields of the adducts 2, albeit N-benzyl and N-allyl substrates took longer time in comparison with others for the CO 2 fixation. Further, alike 1a, the spiroaziridines derived from 3b, 3e, 3f and 3j also efficiently produced the corresponding CO 2 -adducts 2b, 2e, 2f and 2j with the flue gas in similar yields as with pure CO 2 . The regioselectivity of the fixation and the structure of the compound 2 was confirmed from the single crystal X-ray analysis of the compounds 2g (Fig. 7; CCDC 1898609). All the CO 2 -adducts 2 are solid compounds with melting point > 100 °C and bench stable for a couple of months under ambient conditions. Thus the developed regenerable chemical fixation protocol can be utilized for CO 2 capture, storage and release, if and when it needed. ## Conclusion In summary, the first regenerable chemical fixation by spiroaziridine oxindole proved to be an excellent protocol for spontaneous and reversible CO 2 fixation and defixation. We have demonstrated that the CCS [CO 2 fixation and defixation cycle (regeneration)] can work well in one-pot (single vessel) for several cycles with excellent recovery using recyclable reagent under near ambient conditions. More importantly, the process regioselectively produced bioactive spirooxazolidinoyl oxindole in quantitative yields under extremely mild conditions (no extra reagent/catalyst, 1 atm., and rt). The CO 2 -adducts are stable compounds with high melting points, these can be stored for months under ambient conditions and can be reversed back to the sorbent as and when it requires. So, these findings in the ongoing research can open up a new avenue of the chemical fixation for the development ## Methods Auto-chemical fixation of CO 2 by in situ generated spiroaziridine 1a. Amino alcohol 3a (500 mg, 2.60 mmol) was dissolved in dry dioxane (8 ml) and cooled to 0 ºC. Chlorosulfonic acid (174 µl, 2.6 mmol) was added drop wise and the reaction mixture was stirred for 2 h at room temperature (rt). 14 ml of 1 M aqueous NaOH solution was added dropwise to quench the acid at 0 ºC and stirred at 70 °C for 16 h. The complete conversion to spiroaziridine was detected by MS analysis. Next, a slow stream of CO 2 was passed through the solution at rt for 30 min. After complete consumption of 1a (monitored with TLC and also by MS analysis) the dioxane was removed under reduced pressure and the residue was extracted with EtOAc (3 × 10 ml), washed with brine solution and dried over anhydrous Na 2 SO 4 . Combined organic layer was concentrated and purified by silica gel flash chromatography using EtOAc/hexanes (1:1) to afford the desired CO 2 -adduct 2a (528 mg, 93%). Note In case of stimulated flue gas (12.5% CO 2 , 80% N 2 and 7.5% O 2 ) or 12.5% CO 2 in N 2 , the stream of gas was passed through the solution for 18 h. Defixation of CO 2 from spirooxazolidinone 2a and re-fixation of CO 2 . To a solution of spirooxazolidinone 2a (150 mg, 0.69 mmol) in dry dioxane (6 ml), sodium iodide (414 mg, 2.76 mmol) and o-phosphoric acid (144 μl, 2.76 mmol) were added. The mixture was stirred at 70 °C and the consumption of 2a was monitored by TLC and GC-MS. After 5 h, aqueous NaOH solution (0.7 M, 10 ml) was added and stirred for 30 min. The exclusive regeneration of spiroaziridine 1a was confirmed by MS analysis. No spirooxazolidinone 2a and iodoamine 4a were detected in MS analysis at this stage. The crude solution containing spiroaziridineoxindole 1a was further used for the chemical fixation of CO 2 . So, the slow stream of CO 2 was passed through the solution for 30 min. The GC-MS analysis of the crude mixture with naphthalene as an internal standard showed quantitative formation of spirooxazolidinone 2a (98%). Usual work and flash column chromatographic purification as discussed in general procedure gave the compound 2a (143 mg, 95%). co 2 -defixation and fixation cycles through in situ regeneration of spiroaziridine 1a and the isolation of spirooxazolidinone 2a. To a stirred solution of spirooxazolidinone 2a (150 mg, 0.69 mmol) in dry dioxane (6 ml), sodium iodide (414 mg, 2.76 mmol) and o-phosphoric acid (144 μl, 2.76 mmol) were successively added at 70 °C and the reaction (consumption of 2a) was monitored by TLC. After complete consumption of 2a (5 h), it was brought to 0 °C and solid NaOH powder (390 mg, 9.75 mmol) was added to the reaction mixture. After attaining rt, it was stirred for additional 1 h. The stream of 100% CO 2 was passed through to the suspended mixture for 1 h at rt. The solid mass was filtered off and washed with dioxane (2 × 5 ml). The combined organic solvent was evaporated to dryness under reduced pressure. The crude compound was dissolved in dioxane (6 ml) and 150 μl of the solution was taken out for the GC-MS analysis with naphthalene (5 mg) as an internal standard. The analysis showed 97% yield of the spirooxazolidinone 2a. So the calculated amount of resynthesized 2a was found to be 148.5 mg and 150 μl of the solution contained 3.7 mg of 2a. The resynthesized compound 2a (148.5 − 3.7 = 144.8 mg) was used for the second cycle for the regeneration of spiroaziridine and the fixation of CO 2 using the same procedure as mentioned above i.e. the use of NaI-H 3 PO 4 , solid NaOH and the stream of CO 2 . The GC-yield of the second cycle was observed to be 96%. Similarly, another three cycles were carried and the GC-yields were found to be 99%, 97% and 95%, respectively. One-pot CO 2 -defixation and fixation cycles without isolation of re-synthesized spirooxazolidinone. The one-pot CO 2 -defixation and fixation cycles were carried out following the similar procedure as above without separating out the solid by-products and isolation of re-synthesized spirooxazolidinone in the intermediate cycles. To a stirred solution of spirooxazolidinone 2a (150 mg, 0.69 mmol) in dry dioxane (6 ml), sodium iodide (414 mg, 2.76 mmol) and o-phosphoric acid (144 μl, 2.76 mmol) were successively added at 70 °C and the reaction (consumption of 2a) was monitored by TLC. After complete consumption of 2a (5 h), it was brought to rt and solid NaOH powder (390 mg, 9.75 mmol) was added to the reaction mixture at 0 °C. After attaining to rt, it was stirred for additional 1 h. The stream of CO 2 was passed through to the suspended mixture for 1 h at rt. The complete consumption of in situ regenerated spiroaziridine and the formation of spirooxazolidinone 2a were monitored by TLC and MS analysis. Without separating out the solid mass and the isolation of spirooxazolidinone, another consecutive four cycles were repeated by adding the same amount of sodium iodide and o-phosphoric acid followed by solid NaOH and the stream of CO 2 for each cycle in the same pot. The consumption of the intermediate substrate and regeneration of the product were monitored during each cycles by TLC and MS analysis. At the end of 5 th cycles, the solid mass was filtered off and washed with dioxane (3 × 10 ml). The combined organic solvent was evaporated to dryness under reduced pressure. The crude compound was dissolved in dioxane (6 ml) and the GC-MS analysis with naphthalene as an internal standard showed 95% overall yield of the spirooxazolidinone 2a for the five cycles. The silica gel flash column chromatographic purification of the crude with hexanes-EtOAc (1:1) gave the spirooxazolidinone 2a (134.9 mg, 90% overall yield) as a white solid. Recycling of spiroaziridine and NaI for the fixation-and defixation of CO 2 . To a stirred solution of spirooxazolidinone 2a (150 mg, 0.69 mmol) in dry dioxane (6 ml), sodium iodide (414 mg, 2.76 mmol) and o-phosphoric acid (144 μl, 2.76 mmol) were successively added at rt and the reaction (consumption of 2a) was monitored by TLC. After complete consumption of 2a (5 h), solid NaOH powder (342 mg, 8.6 mmol) was added to the reaction mixture at 0 °C. After attaining to rt, it was stirred for additional 1 h. The stream of CO 2 www.nature.com/scientificreports/ was passed through to the suspended mixture for 1 h at rt. The complete consumption of in situ regenerated spiroaziridine and the formation of spirooxazolidinone 2a (97% GC yield) were monitored by TLC and MS analysis. For the next cycle, the reaction mixture was acidified with o-phosphoric acid (292 μl, 5.6 mmol) and stirred at 70 °C without further addition of sodium iodide. After complete consumption of the spirooxazolidinone (monitored with TLC), solid NaOH powder was added (694 mg, 17.36 mmol) and the stream of CO 2 was passed through for 1 h to reproduce the spirooxazolidinone 2a (98%, GC yield). This process was repeated for five consecutive cycles. GC-MS analysis showed almost quantitative yield of spirooxazolidinone in each stage and finally the spirooxazolidinone 2a (135.0 mg, 90%) was isolated after fifth cycle by flash chromatography using hexanes-EtOAc (1:1). Note (a) GC yield is determined by using naphthalene as internal standard; (b) the release of CO 2 from spiroxazolidinone and its subsequent regeneration using CO 2 fixation is considered as one complete cycle. (c) At constant temperature (70 °C) five consecutive cycles of CO 2 fixation and defixation was accomplished using above method (Supplementary material; General procedure 2). GC yield in resynthesis of spirooxazolidione 2a was monitored at each stage (Fig. 3). Received: 16 April 2020; Accepted: 17 July 2020

chemsum

{"title": "Efficient chemical fixation and defixation cycle of carbon dioxide under ambient conditions", "journal": "Scientific Reports - Nature"}

neutral_metal-chelating_compounds_with_high_64_cu_affinity_for_pet_imaging_applications_in_alzheimer

## Abstract: Positron emission tomography (PET), which uses positron-emitting radionuclides to visualize and measure processes in the human body, is a useful noninvasive diagnostic tool for Alzheimer's disease (AD). The development of longer-lived radiolabeled compounds is essential for further expanding the use of PET imaging in healthcare, and diagnostic agents employing longer-lived radionuclides such as 64 Cu (t1/2 = 12.7 h, β + = 17%, β -= 39%, EC = 43%, Emax = 0.656 MeV) are capable of accomplishing this. One limitation of 64 Cu PET agents is that they could release free radioactive Cu ions from the metal complexes, which decreases the signal to noise ratio and accuracy of imaging. Herein, a series of 1,4,7-triazacyclononane (TACN) and 2,11-diaza[3.3]-(2,6)pyridinophane (N4)-based metal-chelating compounds with pyridine arms were designed and synthesized by incorporating Aβ-interacting fragments into metal-binding ligands, which allows for excellent Cu chelation without diminishing their Aβ-binding affinity. The crystal structures of the corresponding Cu complexes confirmed the pyridine N atoms are involved in binding to Cu. Radiolabeling and autoradiopraphy studies show that the compounds efficiently chelate 64 Cu, and the resulting complexes exhibit specific binding to the amyloid plaques in the AD mouse brain sections vs. WT controls. ## Introduction Alzheimer's disease (AD) is the most common neurodegenerative disease. For example, in the United States more than 5 million Americans are living with AD, and this number is expected to reach 16 million by 2050. 1 Positron emission tomography (PET) is a functional imaging technique which can be used for the diagnosis of AD. 2,3 To date, 11 C-and 18 F-radiolabeled imaging agents have been tested for PET studies in AD patients, such as Pittsburgh compound B, Florbetapir, and Florbetaben. 2, However, the use of these agents is limited due to their short physical half-lives (t1/2 = 20.4 min and 109.8 min for 11 C and 18 F, respectively) and complicated syntheses. Therefore, the development of radioimaging agents containing longer-lived radionuclides is important, as it would lead to a longer-time diagnostic imaging agent with a better contrast. 64 Cu (t1/2 = 12.7 h) has become a useful radionuclide in the development of radiopharmaceuticals for imaging purposes. The half-life of 64 Cu is excellent because it is long enough to allow for imaging at late time points, but not so long that it takes weeks to completely decay. In addition, such a half-life will also allow for the imaging agents to be shipped and used in remote areas. Moreover, the radiolabeling with 64 Cu is always the last step in the synthesis of the 64 Cu PET imaging agents, thus simplifying their development. 20 One key limitation of 64 Cu PET imaging agents is that they could release 64 Cu ions in the human body, especially if some ligands have moderate Cu affinity. Also, some enzymes can reduce the chelated Cu 2+ into Cu + , which leads to further releasing of 64 Cu ions. Consequently, decreasing the free 64 Cu level requires ligands to have significantly high metal-binding affinity, limited ligand exchange kinetics, and also relatively low Cu II/I reduction potentials. In our previous report, a series of 64 Cu-PET imaging multifunctional compounds (MFCs) were obtained by linking macrocyclic chelators and Aβ-interacting fragments together. To further increase the metal-binding affinity, larger multidentate ligands were taken into consideration, such as cross-bridged 1,4,7,10tetraazacyclododecane (Cyclen) and 1,4,8,11tetraazacyclotetradecane (Cyclam). However, these ligands are too large and they tend to reduce the Aβ-binding affinity of the corresponding multifunctional compounds. Herein, we introduce a series of metal-chelating compounds (MCCs) designed by employing the strategy of incorporating the Aβ-binding fragment into the metal-chelating moiety (Fig. 1), and by utilizing simple synthetic steps to generate MCCs with high affinities for both 64 Cu-and Aβ species. ## Design and Synthesis of Metal-Chelating Compounds Metal-chelating compounds (MCCs) were designed through the incorporation strategy by merging chemical structures of Aβ binding moiety with a metal-chelating ligand. The synthetic route of MCCs starts with oxidative cyclization of 2aminothiophenol and 6-methylnicotinaldehyde, followed by Nbromosuccinimide (NBS) bromination for further conjugation with the multidentate ligands (Scheme 1). The MCCs were then chelated with 64 CuCl2 in 0.1 M NH4OAc (pH 5.5) and the 64 Cucomplexes were used without further purification (radiochemical yield > 95%). ## Acidity Constants of the MCCs Since all the compounds contain several basic functional groups, their acidity constants (pKa) were determined by UV-vis spectrophotometric titrations. For HYR-7, titrations from pH 1.0 to 11.0 reveal several changes in the spectra (Fig. 2). The best fit to the data was obtained with four pKa values: 1.06(9), 2.68(7), 5.45(3) and 8.82 (2). Based on previously reported acidity constants for amine-pyridine systems, we assigned the lowest pKa value to the protonation of the pyridine group, and the other three higher values correspond to macrocyclic amino groups, as is usually observed for TACN derivatives. For bis-HYR-7 (Fig. S7), the multivalent version of HYR-7, its lowest pKa can be assigned to the protonation of the pyridine group and two higher ones are from the TACN ligand. For HYR-8, UVvis titrations from pH 1.0 to 11.0 reveal changes in the spectra (Fig. S6) that are also best fit with three pKa values: 2.17(6), 4.66(5) and 9.46 (2). The lowest pKa value can be assigned to the protonation of the pyridine group on the Aβ binding fragment. In addition, the higher two pKa values are assigned to the amine groups, similar to HYR-7. For bis-HYR-8 (Fig. S8), the multivalent version of HYR-8, its lowest pKa can be assigned to the protonation of the pyridine group and two higher ones are from the N4 ligand. Please do not adjust margins Please do not adjust margins Spectrophotometric titrations were performed to determine the stability constants and solution speciation of Cu 2+ with MCCs. The pKa values of the ligands were included in the calculations, and the calculated values show that HYR-7 exhibits larger binding constants (logK's) with Cu 2+ than HYR-8, likely because the TACN ligand is more conformationally flexible than the more rigid N4 ligand and can adopt a favourable geometry for tighter Cu 2+ binding. Based on the obtained binding constants, solution speciation diagrams were calculated for Cu 2+ with HYR-7 and HYR-8, showing that the 1:1 Cu:MCC complex is the predominant species formed. In addition, Fig. 3 shows that the concentration of free Cu 2+ with HYR-7 is negligible even at very low pH. While adding a second pyridyl-benzothiazole arm to HYR-7 does not seem to increase the Cu complex stability constant for bis-HYR-7, bis-HYR-8 does show a higher logK for its Cu complex, suggesting that for the N4 ligand the adopted conformation allows for both pyridine N's to potentially interact with the Cu centre. For a better comparison of the Cu 2+ binding affinities for these compounds, pCu (−log [Cu] free) values were calculated at two pH values (Table 3). Interestingly, the pCu values of the MCCs are higher than the standard strong chelators such as diethylenetriaminepentacetic acid (DTPA). 38 During the titration experiments, the spectral changes observed were immediate, suggesting a fast Cu chelation. This is important for efficient 64 Cu radiolabeling, which requires fast complexation. Further stability tests were performed by adding Cu complexes to a solution of 2.5 M HCl (Fig. S13). Even under pH~0 condition, there is no sign of any decomposition in their UV-vis spectra. Overall, these results strongly suggest that all four compounds should be suitable to be used as 64 Cu chelating agents. ## X-ray Structures of Cu Complexes The Cu-HYR-7 complex was synthesized, and single crystals were obtained by the slow evaporation of a dichloromethane/ether solution. In Cu-HYR-7, the Cu center exhibits a N4Cl square pyramidal coordination environment, with three N atoms from the tacn macrocycle, one pyridine N atom, and one Cl atom (Fig. 4a). The Cu-HYR-8 complex was also obtained by the slow evaporation of an acetonitrile/ether solution under N2. In the crystal, the Cu center shows a N5 square pyramidal coordination structure, with four N atoms from the N4 macrocycle, one pyridine N atom (Fig. 4b). Most importantly, these crystal structures offer solid evidence that the N atom on the pyridine group does interact with the Cu center and thus the structure incorporation strategy is practical for the design of stronger chelating ligands. Please do not adjust margins Please do not adjust margins ## EPR Spectra of Copper Complexes To further characterize the Cu-MCC complexes, their X-band EPR spectra were recorded in frozen glasses at 77 K. The EPR spectrum of the Cu-HYR-7 mononuclear complex in a 1:3 (v/v) MeCN/PrCN frozen solution reveals a pseudoaxial EPR pattern with three different g values: gx = 2.059, gy = 2.050, and gz = 2.220 (Fig. 5). Similarly, the EPR spectrum of the Cu-HYR-8 mononuclear complex in 1:3 (v/v) MeCN/PrCN reveals a pseudoaxial EPR pattern with two different g values: gx = gy = 2.090, and gz = 2.270 (Fig. S14). For Cu-bis-HYR-7, its EPR spectrum exhibits a pseudoaxial EPR pattern with three different g values: gx = 2.059, gy = 2.058, and gz = 2.236 (Fig. S15), suggesting that the complex remains mononuclear in solution, while the EPR spectrum of Cu-bis-HYR-8 cannot be simulated well, likely due to presence of two different conformations for this Cu complex (Fig. S16). ## 5xFAD Mouse Brain Section Staining of Metal Complexes After showing that all metal-binding compounds have appreciable Cu 2+ affinity, fluorescence imaging studies of 5xFAD mouse brain sections were performed to evaluate the Aβ binding ability of the MCCs. Brain sections from 11-month old 5xFAD mice were treated with HYR-7, HYR-8, bis-HYR-7, and bis-HYR-8, respectively (Fig. S17). Interestingly, results reveal significant fluorescent staining of the amyloid plaques, as confirmed by co-staining with the CF594-conjugated HJ3.4 antibody (CF594-HJ3.4) that binds to a wide range of Aβ species. 2-4, 11, 39, 40 Then, fluorescence staining using the corresponding Cu complexes was also performed, since the actual PET imaging agents would be the radiolabeled 64 Cu complexes. All Cu(II) complexes stained well the Aβ plaques, as confirmed by CF594-HJ3.4 antibody immunostaining, suggesting that the Cu(II) complexes could be used for the detection of Aβ species (Fig. 6). ## Autoradiography Studies of 64 Cu-complexes Ex vivo autoradiography studies using brain sections of transgenic 5xFAD and age-matched WT mice were also performed to determine the specific binding to the amyloid plaques for the 64 Cu-HYR-7, 64 Cu-HYR-8, 64 Cu-bis-HYR-7 and 64 Cu-bis-HYR-8 complexes. There is a great contrast between the intensity of WT (Fig. 7a, first row) and 5xFAD (Fig. 7a, third row) for all radiolabeled complexes, especially for 64 Cu-HYR-7 with a quantified value of 6.3 (Fig. 7b). Moreover, the specific binding of the 64 Cu-labeled complexes to amyloid plaques was confirmed by blocking with the nonradioactive blocking agent B1 (Fig. S18), which led to a markedly decreased autoradiography intensity (Fig. 6a, second row). In addition, one crucial factor for PET imaging agents of AD is that they need to efficiently cross the blood-brain barrier (BBB). Log D values between 0.9 and 2.5 have been reported to be optimal for promising BBB permeability. 41 The 64 Cu-HYR-7 and 64 Cu-HYR-8 complexes show relatively low log D values because of their dicationic nature. However, with the introduction of the second hydrophobic fragment, the 64 Cu-bis-HYR-7 and 64 Cu-bis-HYR-8 complexes exhibit higher log D values, suggesting that they have ## Please do not adjust margins Please do not adjust margins the potential to cross the BBB (Fig. 7c). As a result, the latter two complexes were selected to be used in the in vivo biodistribution studies. ## Biodistribution Studies After the collection of promising in vitro results, in vivo biodistribution experiments were performed to investigate the pharmacokinetics of 64 Cu-radiolabeled complexes using normal CD-1 mice. The retention and accumulation of the 64 Curadiolabeled complexes in selected organs were evaluated at 2, 60, and 240 min post-injection. Interestingly, 64 Cu-bis-HYR-7 showed higher brain uptake at 2 min with an appreciable accumulation of the radioactivity of ~0.4 %ID/g even after 4 h (Fig. 8c, 8e). However, 64 Cu-bis-HYR-8 showed lower brain uptake at 2 min of ~0.2 %ID/g and increased accumulation in brain even after 4 h of ~0.4 %ID/g (Fig. 8d, 8f), indicating that the latter complex takes a longer time to reach the brain. Also, this difference could be due to the higher stability constant of the Cu-bis-HYR-7 complex than that of Cu-bis-HYR-8, suggesting that the N4 ligand in HYR-8 is too bulky for tighter Cu binding. Taken together, these studies strongly suggest that 64 Cu-bis-HYR-7 system with specific Aβ binding ability is promising for Aβ detection in vivo, but further structure modification is necessary to increase its BBB permeability for further application. ## Conclusions In conclusion, a series 1,4,7-triazacyclononane (TACN) and 2,11diaza[3.3]-(2,6)pyridinophane (N4)-based metal-binding compounds with pyridine arms were designed and synthesized by incorporating Aβ interacting fragments into metal-binding chelating ligands. The incorporation strategy increases the metal-binding affinity of the Aβ-interacting fragment, without the loss of Aβ specificity. Although the Log D values and BBB permeability of the investigated Cu complexes are less than optimal, this strategy could lead to improved Cu-chelating and Aβ-binding compounds for 64 Cu PET imaging in AD. One solution could be the introduction of carboxylate arms onto the N atoms of the TACN ligand. The resulting hexadentate ligand shaould form 6-coordinate, neutral Cu 2+ complexes, which are expected to be more hydrophobic and thus exhibit increased BBB permeability. 42 Overall, the employed approach based on the incorporation strategy can be applied to other 64 Cu-based diagnostic PET imaging applications in neurodegenerative diseases. This journal is © The Royal Society of Chemistry 20xx Please do not adjust margins Please do not adjust margins TOC Graphic

chemsum

{"title": "Neutral Metal-Chelating Compounds with High 64 Cu Affinity for PET Imaging Applications in Alzheimer's Disease", "journal": "ChemRxiv"}

discovery_of_phosphotyrosine-binding_oligopeptides_with_supramolecular_target_selectivity

## Abstract: We demonstrate phage-display screening on self-assembled ligands that enables the identification of oligopeptides that selectively bind dynamic supramolecular targets over their unassembled counterparts.The concept is demonstrated through panning of a phage-display oligopeptide library against supramolecular tyrosine-phosphate ligands using 9-fluorenylmethoxycarbonyl-phenylalanine-tyrosinephosphate (Fmoc-FpY) micellar aggregates as targets. The 14 selected peptides showed no sequence consensus but were enriched in cationic and proline residues. The lead peptide, KVYFSIPWRVPM-NH 2 (P7) was found to bind to the Fmoc-FpY ligand exclusively in its self-assembled state with K D ¼ 74 AE 3 mM. Circular dichroism, NMR and molecular dynamics simulations revealed that the peptide interacts with Fmoc-FpY through the KVYF terminus and this binding event disrupts the assembled structure. In absence of the target micellar aggregate, P7 was further found to dynamically alternate between multiple conformations, with a preferred hairpin-like conformation that was shown to contribute to supramolecular ligand binding. Three identified phages presented appreciable binding, and two showed to catalyze the hydrolysis of a model para-nitro phenol phosphate substrate, with P7 demonstrating conformation-dependent activity with a modest k cat /K M ¼ 4 AE 0.3 Â 10 À4 M À1 s À1 . ## Introduction Oligopeptides are involved in many functions of biological relevance, including self-assembly, molecular recognition and catalysis, holding much promise as designed components for molecular biotechnology and biomimetic materials research. This notion has inspired efforts to design functional peptide modalities as functional (gene-encodable) tags for proteinbased materials. Approaches for the discovery of short functional peptide modules include rational (computational) design, combinatorial screening methods, 10,11 dynamic peptide libraries, 12 phage display and hybrid computationalexperimental methods 17,18 that may be supported by machine learning algorithms. Phage display has successfully led to the identifcation of binding sequences towards protein surfaces, to a range of organic and inorganic nanostructures 13,15,16, and small molecules bound to a surface, sometimes via flexible linkers. The identifcation of short peptides that select for self-assembled (amyloid) structures 33,34 or precursors for catalytic self-assembly 15 has clearly shown that short peptides can selectively bind stable supramolecular structures over unassembled counterparts. In this work, we have developed methodology that brings together phage display with the feld of designed supramolecular materials. We focus on an example from a popular class of supramolecular materials where aromatic groups are used to aid the self-assembly of biomolecules, here exemplifed by 9-fluorenylmethoxycarbonyl-phenylalanine-tyrosine-phosphate (Fmoc-FpY). Although there are affinity reagents (e.g., antibodies/SH2 domains) that recognize phosphotyrosine moieties with high affinity and selectivity, the methodology proposed herein targets self-assembled structures that present the ligand of interest, the phosphotyrosine moiety, on a dynamic supramolecular surface. Although there is no known biological equivalent of this phosphorylated self-assembled material, the high-density presentation of functional groups is common in several supramolecular materials approaches. For our system, we show a frst example where we target tyrosinephosphate moieties exclusively in the self-assembled state. The ability to identify peptide sequences that can bind to specifc moieties presented by designed dynamic supramolecular structures can be useful in several contexts. For example, these peptide sequences can be incorporated into proteins that then bind to designed supramolecular structures; or identifed peptide modalities may inform multi-component self-assembly strategies of relevance to dynamic structures that may incorporate signifcant disorder. For these, reversibility and dynamics can be important, so modest K D values can be bene-fcial over the tight binding structures typically sought. We also hypothesized that selection for supramolecular ligands may give rise to peptides that bind to ligands only in the supramolecular context; such an approach would be useful to target chemical functional groups to both biological and synthetic supramolecular structures. To demonstrate our concept, we chose Fmoc-FpY as a simple target that was previously shown to form micellar aggregates 39 through aggregation of the hydrophobic Fmoc-moieties and consequent presentation of phosphotyrosine on the surface. At 20 mM Fmoc-FpY, forms aggregated micelles with a size distribution between 50-200 nm in diameter (Fig. 1 ## and S3 †). A commercially available dodecapeptide M13 phage display library was used to screen against self-assembled Fmoc-FpY (Fig. 1a) and subsequently tested for binding this target and for the ability to hydrolyze the phosphate-ester bond (Fig. 1b and c). The selected phages were tested for binding and chemical reactivity, and the binding mode and efficiency of the top three lead sequences alongside appropriate controls were analyzed by spectroscopy techniques and molecular dynamics (MD) simulations. A number of strategic mutants of the lead P7 peptide sequence were also investigated to gain understanding of the mode of interaction. ## Results and discussion Phage display peptide libraries were panned against Fmoc-FpY above the critical micelle concentration (CMC) of 20 mM. 39 During panning, the self-assembled Fmoc-FpY target was free in solution, thus maximizing interactions between phages and targets, which contrasts previous biopanning strategies where self-assembling targets were typically pre-immobilized onto surfaces (e.g. polystyrene plates, glass substrates). 33,34 A total of 33 colonies were obtained from the 3 rd round output (Table S1 †), where an enrichment of binders was observed. The 33 phage colonies were subsequently sequenced and analyzed leading to 14 new peptide sequences (Table 1). While there was no sequence consensus in the hits obtained, sequences PC11 and PC13 were repeated four and two times, respectively. The chemical diversity obtained provides general features of binders. The phage panning led to the selection of 14 dodecapeptide sequences (Table 1, Fig. S1a †). The diversity of the amino acids incorporated at each of the 12 positions of these 14 sequences was frst analyzed by the ratio between the frequency of each different amino acid at the n randomized position of the 12-mer sequence over the total 20 amino acids. This value was further normalized by the value 0.7 (ratio 14/20), corresponding to maximum of 14 different amino acids possible at n position (Fig. S1b †). Fig. S1b † shows that all positions of the 14 lead peptide sequences present an amino acid probability higher than 50%, except for position 5, which indicates a preferred type of amino acid at this position. The distribution of the amino acids observed at the 12 randomized positions for all the 14 peptide sequences was then analyzed and plotted in Fig. S1c. † Proline amino acid is statistically relevant specially between the positions at the middle of the sequence at positions 5 (43%), 7 (30%) and 9 (35%). Although there is an overabundance of proline in the phagedisplayed peptide population of the Ph.D-12™ libraries, this statistic overrepresentation is typically observed at positions 3 and 12, 40 which contrasts with the results obtained here. As discussed later, the incorporation of a proline at the middlesequence is important for peptide conformation, 5,41,42 giving rise to b-turns, which was found to be important for the induced binding towards supramolecular phosphorylated-based aggregates. All sequences (except for PC10) had at least one positively charged residue (R/K) or dyad RK/KR. Cationic amino acids are preferentially located at both termini as shown in Fig. S1c, † with a higher abundance at C-terminus, suggesting electrostatic interactions with the phosphate moiety. A distribution of negatively charged residues such as aspartic acid is also observed, near the N-terminus, which can counterbalance the presence of the positive charges in the peptide sequence. According to literature, the charged residues (R/K/D) are underrepresented in these type of phage-displayed peptides 40 and these residues are typically associated with phosphatebinding. 43 Another observation is the preferential presence of aromatic residues such as tyrosine and phenylalanine, at positions 3 and 4. This is interesting from the perspective of supramolecular aggregation interference, considering that the inner part of aggregates is enriched by aromatic fluorenyl, phenyl and phenol functional groups. Other amino acids with a frequency higher than 50% are found in a distributed manner throughout the peptide sequences, without position dependence or enrichment. These include histidine, hydrophobic residues such as leucine, alanine, valine, or polar residues such as threonine, glycine and serine. Overall, the selected peptides had several common compositional features with proline, aromaticity and charge playing a role in the selection. The analysis of phage clones binding to supramolecular Fmoc-FpY particles involved incubation of each amplifed phage clone with Fmoc-FpY at panning conditions, which was followed by buffer washing to avoid non-specifc interactions and subsequent transfer of the phage/Fmoc-FpY containing solutions to a 96-microplate for ELISA assay, to quantify the supramolecular target-bound phages (Fig. S2a †). PC3, PC7 and PC29 were found to be the top binders towards Fmoc-FpY (Fig. 1c), and interestingly, PC3 and PC7 present the highest combination of proline and aromatic residues with two and three respectively, followed by the PC29 with both two proline and aromatic residues. In contrast, the weakest binder, PC9, does not present any aromatic residue and only one proline residue. A common feature between all the peptide sequences is the presence of positively charged residues regardless of the extent of binding. In fact, PC9 is the lowest binder and the one containing the highest number of R/K residues which suggests non-specifc binding towards Fmoc-FpY. Interactions of phages towards Fmoc-FpY aggregates could also be imaged by TEM (Fig. 1b shows binding of the tip of amplifed PC7 to the supramolecular target. Several dozen TEM images were obtained of samples of the PC7 and phage control M13 with Fmoc-FpY). For PC7, we observed several examples of phage termini interacting with Fmoc-FpY particles (additional examples are shown in Fig. S3 †), while for the M13 phage none showed evidence of interaction. Although it was not selected for in our screening setup, as hydrolysis of Fmoc-FpY could not be detected, the phosphataselike catalytic activity of the identifed phage clones was also tested using para-nitrophenol phosphate (pNPP) with PC7 and PC8 observed to be signifcantly above background (Fig. 1c and S2b †). Peptides P3, P7, and P29 were synthesized by solid-phase peptide synthesis and used in their free form in the following studies aiming to quantify binding and understand the interaction with the target. The P9 peptide sequence corresponding to the weakest binding phage was also studied as a negative control. The mode of binding of the free peptides P3, P7, P29 and P9 towards assembled Fmoc-FpY was carried out in the same manner by using 1D 1 H Nuclear Magnetic Resonance (NMR). A solution of 20 mM Fmoc-FpY was titrated with increasing concentrations of the respective peptide in the range 50-1000 mM and the chemical shift changes of the Fmoc-FpY proton signals were monitored (as exemplifed for P7 and Fmoc-FpY in Fig. S4-S6 †). The signals of the peptide could not be resolved because of the difference in concentrations between Fmoc-FpY and peptides, as exemplifed for P7 in Fig. S4. † Analysis of the Fmoc-FpY chemical shifts perturbation as a function of the peptide's concentration yielded binding curves that suggest a level of cooperativity in the interactions (Fig. 2 and S7 †) for all the peptides except for P9 (Fig. S7 †). The data were ftted to the Hill equation Dd ¼ [peptide free ] n /(K D n + [peptide free ]), where K D is the dissociation constant and n the Hill coefficient that describes peptide cooperativity (Fig. 2). The K D values that were obtained from the ftting of the peptideinduced chemical shift perturbation curves (Fig. 2a, S7 † and Table 2) suggest the formation of a supramolecular-binding complex between Fmoc-FpY and the peptides P3, P7 and P29 but with different K D as shown in Table 2. P7 was revealed to have the highest affinity towards Fmoc-FpY with a K D of 74 AE 3 mM, which is in the same order of magnitude compared to the values obtained for the peptides with affinity for previously reported self-assembling structures. 33,34 The data obtained for P3 and P29 showed comparable cooperative binding profles as for P7 but with K D values that were an order magnitude higher than P7 (Fig. 2, S7 † and Table 2), revealing lower affinity of the P3 and P29 for the assembled target. In contrast, the P9 concentration range used was not sufficient to reach saturation of the complex, and therefore a value of K D was not possible to be determined (Fig. S8 †). These results are in accordance with the panning results. Effectively, the peptide with the highest binding affinity towards the supramolecular Fmoc-FpY aggregates was P7, which corresponds to the lead phage clone binder. To test whether the peptides P3, P7 and P29 for Fmoc-FpY differentiate between the self-assembled target and the free molecules, the titration was repeated at a concentration where Fmoc-FpY is not assembled (1 mM) (Fig. 2 and S9 †). At this concentration, no chemical shift perturbations were observed, confrming that P3, P7 and P29 displays supramolecular target selectivity (Fig. 2). The shape of the curves for all peptides also indicates a mechanism that is similar to positive allosteric cooperativity (n > 1) (Fig. 2 and S7 †), suggesting that binding of one peptide molecule facilitates the binding of additional molecules. For the three peptides, the CH and aromatic protons of the Fmoc group (4.1 and 7.2-7.9 ppm, respectively) showed the highest chemical shift perturbation, followed by the Ha of the phenylalanine (3.5 ppm), and the Hd of phosphorylated tyrosine (7 ppm) (Fig. S5 and S7 †). These observations reveal that the Fmoc group experiences a more dramatic change in its chemical environment upon binding when compared with the more exposed aromatic protons of the tyrosine (Fig. S7 †), suggesting that peptide binding interferes with the Fmoc stacking within the micellar aggregates. This structural disruption is expected to play a role in the observed allosteric cooperativity. To understand the impact of the peptides' conformation on the mode of interaction, circular dichroism (CD) studies were performed for all the peptide sequences (Fig. 2d). The CD spectra revealed that all peptides appear to adopt a random coil conformation with a minimum around 200 nm, except for P7 that revealed a b-hairpin-like conformation, 44 with a minimum at 196 nm, a small shoulder at 210 nm, and a weak positive band around 230 nm (Fig. 2d). P7 peptide is the lead binder with a b-hairpin-like conformation, which can indicate that the binding interaction can beneft from a defned conformation. Inspection of the P7 sequence KVYFSIPWRPM-NH 2 reveals that the optimal combination between the proline positioning, aromaticity, and the N-terminal sequence (KVYF) that contains both cationic and aromatic residue can provide a complement of the Fmoc-FpY structure. Therefore, the respective mode of interaction of P7 with Fmoc-FpY was then studied in more detail. To understand the importance of the phosphate-moiety in mediating the binding mechanism towards Fmoc-FpY, a control experiment was performed to assess binding between P7 and Fmoc-FY. Considering the propensity of Fmoc-FY to undergo gelation, 39 the samples of Fmoc-FY were prepared at lower concentrations, specifcally at below and above the critical assembly concentration. Concentrations were chosen according to the emission spectra of Fmoc-FY where fbers form without gelation (1 mM), and in absence of the fbers formed (0.1 mM) (Fig. S10a †). Fiber formation was also confrmed by TEM in Fig. S10. † Considering these results, the study of the binding of P7 towards Fmoc-FY was conducted by NMR below (0.1 mM) and above (1 mM) the critical assembly concentration. 1D 1 H NMR chemical shift changes of the most downfeld Fmoc-FYp aromatic proton signal were monitored to understand the mode of interactions of the lead peptide P7 and Fmoc-FY (Fig. S11 †). The results shown in Fig. S12 † revealed that in the absence of the phosphate-moiety, the binding event between P7 and Fmoc-FpY does not occur for both Fmoc-FY concentrations tested, which implies a pivotal role for the phosphate-moiety at mediating the binding with P7, possibly through the positively charged residues R and K. The proposed hairpin-like P7 conformation in solution was also observed in MD simulations. As may be expected for a short peptide, P7 samples multiple conformations throughout the simulations (3 250 ns). However, for a signifcant (30%) part of the simulations the peptide adopts a hairpin-like conformation (Fig. 3a) which corroborates the CD spectra (Fig. 2b). This P7 conformation is stabilized by a cooperative exchange of backbone hydrogen bonds (Fig. 3b, S13 and S14 †), and the key interactions are summarized in (Fig. S13 and S15 †). Root mean square fluctuations (RMSF) of the C a atoms from each residue shows that the core of the hairpin (residues 4-9, FSIPWR) is less dynamic compared to the termini (Fig. S16 †). Mutational analysis of the leading P7 peptide sequence (Fig. S15 †) were then studied by 1D 1 H NMR to assess the contribution of each residue to the dynamic hairpin conformation of P7 (Fig. 3c) and to understand in more detail the mechanism of interaction of the supramolecular complex Fmoc-FpY/P7 (Fig. 3d). Due to the precipitation of certain mutants at high concentrations (K1A, P7A, R9A and, P11A), this comparison was carried out at 75 mM where all mutants were soluble. Lower chemical shift differences are related to a decrease in the binding efficiency, and the results indicate that all residues contribute to binding, except for R9 (Fig. 3d). Overall, the KVYF terminus (in particular V and F) of the P7 has a stronger impact on binding compared to the C-terminal of P7, further confrming a role of aforementioned complementarity with Fmoc-FpY. The core of the hairpin (4-8, FSIPW) is also critical for binding, thought to provide structural stability to the hairpin. According to the CD spectra shown in Fig. 3c, the mutations that mostly affect P7 conformation are V2A, Y3A, I6A, P7A, and P11A, revealing their importance in maintaining the stability of the hairpin framework, V2, Y3, and I6 are involved in hydrogen bonding (Fig. S13 and S14 †) and CH a interactions (Fig. S15 †) stabilizing the hairpin-like conformation. Moreover, these residues present lower RMSF values, indicating their contribution to lower flexibility of the hairpin conformation. Also, mutants P7A and P11A revealed a complete loss of hairpin-like structure, confrming the role of these Pro residues in stabilizing the conformation (Fig. 3c). The NMR results, together with the evaluation of the mutants' conformation by CD, suggest that P7's hairpin-like conformation is essential for the formation of the supramolecular Fmoc-FpY/P7 complex. We propose a mechanism involving selective binding towards the assembled state through interaction with the KVYF terminus which is facilitated by the hairpin-conformation and leads to partial disruption of the supramolecular aggregates. A similar induced-binding mechanism has been previously shown for particle-bound UTP-ligands 14 and has been suggested for a heptapeptide bound to self-assembled peptide nanofbers. 34 Finally, the catalytic activity that was observed during screening, and in the phage clone PC7 was investigated for the free oligopeptide P7 using the model substrate pNPP (Fig. 4a). The dependence of the initial rate of the reaction on substrate concentration follows Michaelis-Menten model, with k cat ¼ 10 5 s 1 , K M ¼ 14 AE 4 mM, and catalytic efficiency of k cat /K M ¼ 4 AE 0.3 10 4 M 1 s 1 . We note that the catalytic efficiency of P7 compares favourably to previously reported systems for phosphate ester hydrolysis, including a designed b-hairpin loop based on peptide-nucleic acid conjugate, 45 a 42-residue helixloop-helix peptide motif 46 and peptide amphiphiles 47 that showed catalytic activity when assembled into peptide nano-fbers. A common feature between all is the presence of defned structural elements, reinforcing the importance of folding for chemical catalysis. An additional observation was that P7 concentration influences its catalytic rate, with a dramatic decrease in the catalytic rates observed as the concentration of P7 increases (Fig. 4b). Coinciding with this reduction in catalytic activity is a conformational change of the peptide (Fig. S18 †). The CD signature of the hairpin-like peptide changes as P7 concentration increases, with the minimum band blue-shifted to 202 nm and the weak positive band slightly red-shifted showing the maximum of the positive band at 230 nm and the disappearance of the shoulder at 210 nm (Fig. S16 †). At concentrations higher than 100 mM, P7 revealed a CD signature of PP-II conformation, which is characterized by a negative band around 200-205 nm and by a weaker positive band around 225 nm (ref. 48 and 49) and expected to be enabled by the two proline residues. Also, at higher concentrations, P7 selfassembles into spherical aggregates, which were further confrmed using diffusion-ordered spectroscopy (DOSY) and imaged by TEM and AFM, shown in S19. † This P7 self-assembly is related to a loss of b-hairpin-like structure and a conformational change of P7 to PP-II conformation, acting also as a conformation switch in the catalytic mechanism. The results suggest that the secondary framework of P7 of hairpin-like is the catalytic active conformation, as reported also in other peptide-based catalysts. 45,46,50 We note that the hairpin to PP-II conformational switch observed at higher concentration was an unexpected result that was unrelated to the panning conditions, but an unintended consequence of the proline-rich sequence that was selected. The catalytic mechanism of P7 is currently not understood and will be the subject of future studies. We note that the catalytic activity for the supramolecular target Fmoc-FpY (instead of the para-nitrophenol substrate) could not be conclusively demonstrated because the k cat of P7 was not sufficient to hydrolyse Fmoc-FpY. While molecular recognition is the main driving during selection, the phosphatase activity was also used during the screening process. Although the catalytic activity was not sufficient to hydrolyse the Fmoc-FpY phosphate ester bond, as initially intended and previously observed for an amide condensation target, 15 the modest activity observed, and supramolecular regulation of this activity are interesting features to be exploited for further applications. The peptide combines a remarkable combination of features in one sequence: binding to a supramolecular target, modest catalytic activity, and concentration-dependent supramolecular reconfguration. ## Conclusions A biopanning strategy based on supramolecular recognition led to the discovery of oligopeptides' sequences with chemical diversity and with sequence-dependent patterns. In particular, there is a clear preference of charged residues (R/K) located at sequence terminals, the presence of proline residues in the middle of the sequence, and aromatic residues (Y/F), preferentially positioned in the frst half of the peptide. Although the chemical diversity was key to generate binders with different properties, it is envisaged that in future studies, a consensus sequence can be obtained by carrying out more rounds of biopanning. The detailed binding mechanism of the lead oligopeptidebased sequences revealed micromolar range K Ds (74-520 mM). This indicates reversible binding, and an ability to differentiate tyrosine phosphorylation between monomeric and aggregation states. The P3, P7 and P29 oligopeptides revealed selectivity for supramolecular over non-supramolecular states of phosphate molecules, enabling the creation of micromolar-affinity supramolecular complexes. The lead peptide P7 presented an optimal combination of chemical composition with a defned hairpinlike structure that allowed for an adaptive behavior towards supramolecular structures, via destabilization of the tyrosinephosphorylated micellar aggregates. In future, we envision that the panning strategy developed in this work can also be applied to fnd binders to other dynamic targets such as the phase-separated biomolecular condensates. Although there is no counterpart of the phosphorylated selfassembled Fmoc-FpY in biology, high-density tyrosine phosphorylation is often seen in different diseases (e.g., metabolic, neurodegenerative diseases). Therefore, the development of peptide modalities and functional materials that selectively target and perturb high-density phosphorylation can be incorporated in future diagnostic tools and therapeutics. In addition, we also foresee that the chemical information dictating the ability of dodecapeptides at catalyzing phosphate ester hydrolysis phosphate expands the peptide feld in aqueous media.

chemsum

{"title": "Discovery of phosphotyrosine-binding oligopeptides with supramolecular target selectivity", "journal": "Royal Society of Chemistry (RSC)"}

stimuli-responsive_metal–organic_frameworks_gated_by_pillar[5]arene_supramolecular_switches

## Abstract: Spurred on by recent advances in materials chemistry and drug delivery, a new stimuli-responsive theranostic hybrid platform, based on mechanized monodisperse nano metal-organic frameworks (NMOFs) gated by carboxylatopillar[5]arene (CP5) switches with bio-friendly pH-triggered cargo release capabilities, has been constructed for the first time. This nanoscale smart cargo delivery system showed pH-and/or competitive binding agent-triggered controlled cargo release with negligible premature release, large pore sizes for drug encapsulation, low cytotoxicity, good biodegradability and biocompatibility, and potential application in cell imaging, which offers a new tool in targeted drug delivery and the controlled release of therapeutic agents. Stimuli-responsive metal-organic frameworks gated by pillar arene supramolecular switches † ## Introduction The past decade has witnessed the revolutionary impact of nanotechnology on modern nanomedicine, since advanced nanoscale systems can both alleviate many of the pitfalls associated with free drug therapeutics and improve the efficacy of conventional drugs. In particular, stimuli-responsive functional nanocarriers 5,6 for drug delivery and controlled release hold great promise in achieving revolutionary advances in virtually all aspects of medicine, including in vitro diagnostics, 7,8 bioimaging, 9 targeted and internally/externally triggered therapy, image-guided surgery, and regenerative medicine. Among advanced nanomaterials, metal-organic frameworks (MOFs) 17 have emerged as powerful platforms for gas storage, catalysis, 21 separation, 22 drug delivery, imaging, 26 sensors 27 and detection 28,29 owing to their tunable structural features and pore sizes, high surface areas, chemical/ thermal stability, and versatile functionality. In particular, the biomedical application of MOFs 30 is gaining tremendous attention, especially in the feld of nanomedicine, and this emerging class of porous materials is likely to replace traditional nanoporous materials in drug delivery and storage in the future, due to their unique properties such as exceptionally high surface areas and large pore sizes for drug encapsulation, good biodegradability and biocompatibility, versatile functionality, and potential application in cell imaging. However, to the best of our knowledge, no mechanized MOFs, which are composed of MOFs as scaffolds and supramolecular switches as gating entities to prevent premature cargo leakage and enable cargo release in a fne-tuned, targeted and controlled fashion, have so far demonstrated on-demand release of drugs. In the human body, zinc plays "ubiquitous biological roles", such as maintaining the immune function, sterilization, and treatment of cancer, and is essential for the structural stability of a variety of proteins involved in transcription, protein traf-fcking, neurosecretory products or cofactors, and enzyme catalytic activity. 31 UMCM-1-NH 2 (UMCM ¼ University of Michigan Crystalline Material) is a MOF 'co-polymer' constructed by combining zinc with both a two-fold symmetric (2-amino-1,4-benzenedicarboxylate, BDC-NH 2 ) and a three-fold symmetric (1,3,5-benzene-tri-p-benzoate, BTB) organic linker, which produces a hexagonal mesopore with a 1D hexagonal channel of 27 32 surrounded by six smaller polygonal micropores each with a dimension of 14 17 . 32 As the BDC-NH 2 linkers are located at the junctions of the microporous cages, the introduction of small functional groups as stalks to the linkers through post-synthetic modifcation (PSM) would mainly affect the pore environment of the cages, without affecting the porosity of the large channel. 33,34 Their large pore size and previous successes with PSM reactions make them promising candidates as nanocarriers for controlled drug delivery. Pillar[n]arenes (pillarenes for short) as a relatively new class of synthetic macrocycles have seen a tremendous boost in their use for host-guest chemistry in the last six years since they integrate the advantages of other existing macrocyclic host compounds and possess their own unique characteristics, which facilitate their applications in artifcial transmembrane channels, 40,41 controlled release systems, gas sorption, 47,48 MOFs, 49 sensing and detection, 50,51 stabilization of nanoparticles, and other typical biological applications, 52,53 etc. Herein, monodisperse mechanized nanoMOFs (NMOFs) with pillarene-based supramolecular switches as gatekeepers have been constructed by combining PSM of MOFs with stimuliresponsive host-guest chemistry on MOF surfaces for the frst time (Fig. 1). This system has shown pH-and/or competitive binding-triggered controlled cargo release with negligible premature release, large pore sizes for drug encapsulation, very low cytotoxicity, good biodegradability and biocompatibility, and potential application in cell imaging, which will open a new avenue in targeted drug delivery and controlled release of therapeutic agents, especially in the treatment of degenerative diseases. ## Results and discussion The fabrication of the carboxylatopillar arene (CP5)-based mechanized UMCM-1-NH 2 nanocarrier systems is depicted in Fig. 1. The scaffold UMCM-1-NH 2 was synthesized according to the literature procedure (ESI †). 32 Positively-charged pyridinium (Py) stalks were successfully tethered onto the UMCM-1-NH 2 surfaces via PSM, followed by the loading of luminescent rhodamine 6G (Rh6G, 1 mM) as a model drug in their nanopores at room temperature. Finally, the negatively-charged CP5 macrocycles were introduced to encircle the Py stalks via hostguest complexation to form pseudorotaxanes as the movable elements of the mechanized nanocarriers, thereby realizing the drug encapsulation. The peaks of Py in the 1 H NMR spectra (Fig. S7 †) and the peak of BDC-NH-Py in the electrospray ionization mass spectrometry (ESI-MS, Fig. S8 †) indicated that the Py stalks were successfully anchored to UMCM-1-NH 2 . The conversion of the modifcation of UMCM-1-NH 2 was calculated to be 50%. Fourier transform infrared (FTIR) spectra (Fig. S9 †) were also used to verify and monitor the functionalization of UMCM-1-NH 2 . Compared to UMCM-1-NH 2 , the presence of the peaks of -CH 2 -($2859 and $2928 cm 1 ), -NH-($1283 cm 1 ) and -C]N-($1708 cm 1 ) was indicative of the fact that UMCM-1-NH-Py was formed. Shift and intensity changes in peaks of the Rh6Gloaded, CP5-capped UMCM-1-NH-Py curve appeared at around 2900 cm 1 and peaks from 1700 cm 1 to 1000 cm 1 were caused by the encasing of CP5 and the loading of Rh6G. Meanwhile, thermogravimetric analysis (TGA) of UMCM-1-NH-Py showed a rapid weight loss (4%) frst (Fig. S10 †) which corresponds to the liberation of DMF molecules entrapped inside the cavity, followed by a plateau region until 325 C, where the materials began decomposing, ascribed to the loss of Py. This certifed UMCM-1-NH-Py to have enough thermal stability for constructing drug delivery systems. To further test the microcrystallinity of our newly synthesized functional materials, highresolution transmission electron microscopy (HRTEM) images and electron-diffraction patterns were taken (Fig. 2 and S14 †). HRTEM images indicated that both UMCM-1-NH 2 (Fig. 2b) and Rh6G-loaded, CP5-capped UMCM-1-NH-Py (Fig. 2e) have well-defned crystalline planes with interplanar d-spacing of 0.304 nm, corresponding to the lattice spacing of the (103) planes. Electron-diffraction patterns are a vivid demonstration of the single crystalline nature of these nanoparticles. Furthermore, the pore size distribution (Fig. S11 and S12 †) was obtained using the non-localized DFT (NLDFT) method from N 2 sorption isotherms at 77 K. UMCM-1-NH-Py showed two main sharp peaks at approximately 1.7 nm and 4.6 nm consistent with UMCM-1-NH 2 and the crystal structure according to the literature 32 (Fig. S15 †) which means that the pores of the nanoparticles were not changed after grafting pyridine units onto the openings of the pores by covalent bonds and the nanoparticles can be used as a drug carrier and further incorporated with supramolecular switches to construct nanovalves. The morphology, size, monodispersity, and surface texture of these nanoparticles were investigated by scanning electron S1, † showing that UMCM-1-NH-Py has positive surface charges and Rh6G-loaded, CP5-capped UMCM-1-NH-Py has negative surface charges, which further validated the successful modi-fcation and capping. The zeta potential of Rh6G-loaded, CP5capped UMCM-1-NH-Py was measured to be 16.4 mV which clearly revealed that the newly synthesized drug delivery system can maintain certain stability and is strong enough to transport drugs successfully in biological media. DLS was used to calculate the average diameter of the NMOFs in water. In particular, the average particle size (Table S1 †) in solution became progressively smaller after modifcation and capping due to the good solubility of Py and CP5 in water. The average particle diameter of Rh6G-loaded, CP5-capped UMCM-1-NH-Py was calculated to be 102.9 nm, which is within the size range 61 for easy uptake by cells. Interestingly, the nanoparticles became monodisperse with homogeneous particle sizes (ca. 102 nm), evidenced by SEM (Fig. 2d), which makes the mechanical nanocarriers constructed from UMCM-1-NH-Py and CP5 promising candidates for drug storage and drug delivery. Upon addition of methyl viologen salts, which have a higher binding affinity toward CP5 (K a z 8 10 4 M 1 ), 62 into a cuvette containing Rh6G-loaded, CP5-capped UMCM-1-NH-Py, an immediate release of the cargo molecules was observed (Fig. 3a) as a result of the methyl viologen-induced dethreading of the CP5 rings from the Py stalk components (K a ¼ (2.5 AE 0.7) 10 3 M 1 ). 61 The release rate of Rh6G depends on the amount of methyl viologen salts added, which indicated the important role of the CP5 supramolecular switches. Neutralization of the CP5 sodium salts upon lowering the pH of the solution results in the weakening of the noncovalent bonding interactions between the ring and stalk components of the CP5 pseudorotaxanes, leading to the unblocking of the nanopores. 61 So, Rh6G-loaded, CP5-capped UMCM-1-NH-Py nanoparticles were able to contain Rh6G molecules at neutral pH and basic pH (Fig. 3c), but release them under acidic pH and the release rate of Rh6G depends on the pH level like a pH clock 63 (Fig. 3d). A flat baseline shows (Fig. 3c) that the molecules of Rh6G are held frmly within the nanopores at neutral pH and basic pH: there is no premature release, which is a breakthrough for MOF-based drug delivery systems. When the pH of the solution is lowered to 5, the supramolecular nanovalves are opened and the cargo of Rh6G molecules is released. The lower the pH (e.g., 2), the faster the release of the cargo. Meanwhile, the anticancer drug, doxorubicin hydrochloride (DOX), was loaded into the nanopores: a smooth release profle was observed (Fig. 3b) upon lowering the pH. As pH in areas of tumor tissues is known to be more acidic than in blood and normal tissues (pH 7.4) and in view of the fact that their lysosomal pH levels are somewhat lower than in healthy human cells, a pH responsive drug delivery system can reduce undesired drug release during drug transportation in blood circulation and improve the effective release of the antitumor drug in the tumor tissue or within tumor cells. 63 What's more, since the drug is expected to be released much faster at the tumor site than the surrounding normal tissues maintaining a physiological pH of 7.4, it is expected that the delivery of chemotherapeutic drugs via these systems may also reduce their adverse effects which in some cases can be severely debilitating. So, with no doubt, pH-sensitive carriers are particularly promising candidates for anticancer drug delivery. A series of control experiments have also been done to prove the functionalization of CP5 supramolecular switches in the UMCM-1-NH-Py drug delivery system by comparing the difference of the release performance of Rh6G-loaded, CP5-capped UMCM-1-NH-Py and Rh6G-loaded UMCM-1-NH 2 without CP5 capping by pH activation (Fig. S5 †). Premature leakage without the CP5 rings attached was more obvious than that with the CP5 rings and the encapsulation efficiency of the MOFs without the CP5 rings attached (5 mmol g 1 ) was signifcantly lower than that with the CP5 rings (61 mmol g 1 ). These reveal the important role of CP5 supramolecular switches in our system for tuning the drug loading capacity. As shown in Fig. S16, † with the increase in concentration of the two nanomaterials, the cell viabilities showed a declining trend. However, they only had slight cytotoxicity to normal human cells, which was deduced from the fact that the cell viabilities were higher than 60%, even though their concentration was as high as 50 mg mL 1 . Overall, the nanomaterials, before and after capping with CP5, possess negligible cell cytotoxicity at low concentrations, allowing them to be used as nanocontainers for controlled drug delivery in disease therapy. ## Conclusions In conclusion, monodisperse mechanized nanocarriers based on NMOFs and pillarene-based supramolecular switches as gatekeepers have been constructed for the frst time by combining PSM with stimuli-responsive host-guest chemistry. UMCM-1-NH 2 with mesopores (27 32 ) and micropores (14 17 ) was selected as a nanocontainer and positivelycharged Py stalks were successfully attached to the UMCM-1-NH 2 via PSM without affecting the porosity of the large channel, which was confrmed by 1 H NMR, ESI-MS, FTIR spectra and pore size distribution. Then, Rh6G and DOX were loaded, negatively charged CP5 macrocycles were introduced to encircle the Py stalks via host-guest complexation to form pseudorotaxanes as the movable elements of the nanocarriers, thereby realizing the drug encapsulation. HRTEM, SEM, zeta potential, and DLS showed that this mechanical nanocontainer is mainly in monodisperse microcrystalline form, certainly stable and within the size range that can be easily taken up by cells (102 nm). MTT cytotoxicity assay of 293 cells treated with UMCM-1-NH-Py and CP5-capped UMCM-1-NH-Py at various concentrations conveyed that the new functional materials, before and after CP5 capping, possess negligible cytotoxicity. This unique MOF-based nanovalve system showed pH-and/or competitive binding agent-triggered controlled cargo release with negligible premature release, large pore sizes for drug encapsulation, noncytotoxicity, good biodegradability and biocompatibility, and potential application in cell imaging, which will open new avenues in targeted drug delivery and controlled release of therapeutic agents, especially in the treatment of cancer diseases. Future investigations will employ this integrated nanosystem to carry anticancer drugs, perform pH-responsive drug release in vivo, take advantage of the properties of the metal and organic ligands it contains, and could one day fnd application in excellent treatment of human cancers.

chemsum

{"title": "Stimuli-responsive metal\u2013organic frameworks gated by pillar[5]arene supramolecular switches", "journal": "Royal Society of Chemistry (RSC)"}

synthesis_and_styrene_copolymerization_of_novel_bromo,_chloro,_fluoro,_and_iodo_ring-substituted_oct

## Abstract: Halogen ring-substituted octyl phenylcyanoacrylates, RPhCH=C(CN)CO2CH2(CH2)6CH3 (where R is 4-bromo, 2-chloro, 3-chloro, 4-chloro, 2-fluoro, 3-fluoro, 4-fluoro, 2-iodo, 3iodo, 4-iodo) were prepared and copolymerized with styrene. The acrylates were synthesized by the piperidine catalyzed Knoevenagel condensation of ring-substituted benzaldehydes and octyl cyanoacetate, and characterized by CHN analysis, IR, 1 H and 13 C NMR. All the acrylates were copolymerized with styrene in solution with radical initiation (ABCN) at 70C. The compositions of the copolymers were calculated from nitrogen analysis. ## Introduction 4-Bromo ring-substituted phenylcyanoacrylate (PCA) is reported in synthesis of novel selenophenes from activated acetylenes, ethyl 2-cyano-3-arylacrylate and potassium selenocyanate ; in catalysis using MWCNT-polyamine hybrids ; in tandem deacetalization-Knoevenagel condensation reaction using mesoporous carbon ; in catalysis by selenotungstates incorporating organophosphonate ligands and metal ions of condensation reaction ; in synthesis and study of antimicrobial and antioxidant activities of fused uracils: pyrimidodiazepines, lumazines, triazolouracil and xanthines ; in studies related to controlling the emergence and shift direction of mechanochromic luminescence color of a pyridine-terminated compound ; in synthesis of thiazacridine derivatives as anticancer agents against breast and hematopoietic neoplastic cells ; in design, synthesis and characterization of aurone based α,β-unsaturated carbonyl-amino ligands and their application in microwave assisted Suzuki, Heck and Buchwald reactions ; in catalyst-free [3+3] annulation/oxidation of cyclic amidines with activated olefins: when the substrate olefin is also an oxidant ; in Cu-catalyzed remote transarylation of amines via unstrained C-C functionalization ; in silica bonded N-(propylcarbamoyl)sulfamic acid as a highly efficient and recyclable solid catalyst for the synthesis of benzylidene acrylate derivatives: docking and reverse docking integrated approach of network pharmacology ; and in ultrasonication-assisted synthesis of α,β-unsaturated compounds catalyzed by aminofunctionalized FDU-12 catalyst . 4-Fluoro ring-substituted PCA is involved in synthesis of 3-cyano-2-pyridones derivatives catalyzed by Au-Co/TiO2 ; in synthesis, anticancer, antimicrobial, anti-tuberculosis and molecular docking of heterocyclic N-ethyl-N-methylbenzenesulfonamide derivatives , and in synthesis of thiazacridine derivatives as anticancer agents against breast and hematopoietic neoplastic cells . 4-Iodo ringsubstituted PCA are reported in selective hydrolysis of 1-cyanocyclopropane-1carboxylates: concise preparation of 1-carbamoylcyclopropane-1-carboxylates , and Reaction products of indandione with ethyl α-cyano-β-arylacrylates . In this work we have prepared octyl halogen ring-substituted cyanoacrylates, RPhCH=C(CN)CO2CH2(CH2)6CH3, where R is 4-bromo, 2-chloro, 3-chloro, 4-chloro, 2fluoro, 3-fluoro, 4-fluoro, 2-iodo, 3-iodo, 4-iodo, and explored the feasibility of their copolymerization with styrene. To the best of our knowledge there have been no reports on either synthesis of these compounds, nor their copolymerization with styrene . iodobenzaldehydes, octyl cyanoacetate (≥98.0%), piperidine (99%), styrene (≥99%), 1,1'azobis(cyclohexanecarbonitrile) (98%), (ABCN), and toluene (98%) supplied from Sigma-Aldrich Co., were used as received. Instrumentation is reported in . ## Synthesis and characterization of octyl phenylcyanoacrylates All octyl phenylcyanoacrylates (OPCA) compounds were synthesized by Knoevenagel condensation CO2CH2), 1.8-1.7 (q, 2H, OCH2CH2), 1.5-1.4 (m, 6H, OCH2CH2(CH2)3), 1.4-1.2 (m, 4H, ABCN at an overall monomer concentration 2.44 mol/L in 10 mL of toluene. The copolymerization was conducted at 70ºC. After a predetermined time, the mixture was cooled to room temperature, and precipitated dropwise in methanol. The composition of the copolymers was determined based on the nitrogen content (cyano group in OPCA). The novel synthesized OPCA compounds copolymerized readily with ST under freeradical conditions (Scheme 2) forming white flaky precipitates when their solutions were poured into methanol. The conversion of the copolymers was kept between 10 and 20% to minimize compositional drift (Table 1). Nitrogen elemental analysis showed that between 23.3 and 41.1 mol% of OPCA is present in the copolymers prepared at ST/OPCA = 3 (mol), which is indicative of relatively high reactivity of the OPCA monomers towards ST radical which is typical of halogen ring-substituted OPCA. Since OPCA monomers do not homopolymerize, the most likely structure of the copolymers would be isolated OPCA monomer units alternating with short ST sequences (Scheme 2). The copolymers prepared in the present work are all soluble in ethyl acetate, THF, DMF and CHCl3 and insoluble in methanol, ethyl ether, and petroleum ether. ## Conclusions Novel trisubstituted ethylenes, octyl halogen ring-substituted phenylcyanoacrylates, RPhCH=C(CN)CO2CH2(CH2)6CH3 (where R is 4-bromo, 2-chloro, 3-chloro, 4-chloro, 2fluoro, 3-fluoro, 4-fluoro, 2-iodo, 3-iodo, 4-iodo) were prepared and copolymerized with styrene.

chemsum

{"title": "Synthesis and styrene copolymerization of novel bromo, chloro, fluoro, and iodo ring-substituted octyl phenylcyanoacrylates", "journal": "ChemRxiv"}

a_polymer_prodrug_strategy_to_switch_from_intravenous_to_subcutaneous_cancer_therapy_for_irritant/ve

## Abstract: Chemotherapy is almost exclusively administered via the intravenous (IV) route, which has serious limitations (e.g., patient discomfort, long hospital stays, need for trained staff, high cost, catheter failures, infections). Therefore, the development of effective and less costly chemotherapy that is more comfortable for the patient would revolutionize cancer therapy.While subcutaneous (SC) administration has the potential to meet these criteria, it is extremely restrictive as it cannot be applied to most anticancer drugs, such as irritant or vesicant ones, for local toxicity reasons. Herein, we report a facile, general and scalable approach for the SC administration of anticancer drugs through the design of well-defined hydrophilic polymer prodrugs. This was applied to the anticancer drug paclitaxel (Ptx) as a worst-case scenario due to its high hydrophobicity and vesicant properties (two factors promoting necrosis at the injection site), whereas polyacrylamide (PAAm) was chosen as a hydrophilic polymer for its biocompatibility and stealth properties. A small library of Ptx-based polymer prodrugs was designed by adjusting the nature of the linker (ester, diglycolate and carbonate), and then evaluated in terms of rheological/viscosity properties in aqueous solutions, drug release kinetics in PBS and in murine plasma, cytotoxicity on two different cancer cell lines, acute local and systemic toxicity, pharmacokinetics and biodistribution, and finally their anticancer efficacy.We demonstrated that Ptx-PAAm polymer prodrugs could be safely injected subcutaneously without inducing local toxicity while outperforming Taxol, the commercial formulation of Ptx, thus opening the door to the safe transposition from IV to SC chemotherapy. ## Introduction Due to population growth and aging, the number of new cancer cases is expected to increase by approximately 70% over the next 20 years. 1,2 As a result, not only will more and more patients have to deal with cancer, but hospital organization will be strained while patients and health care systems will face an increasing financial burden. 3,4 In addition, since chemotherapy is mostly administered intravenously (IV), 5 it is usually accompanied by severe limitations that are directly responsible for patient discomfort and the high cost of cancer treatments: (i) injectable formulations must be prepared in chemotherapy reconstitution units; (ii) administration must be performed by qualified workers at the hospital, often via a central IV route that requires an implantable chamber; (iii) the patient must stay at the hospital during treatment to be monitored for an early detection of infusion-related toxicities and (iv) catheter failures and life-threatening infections often occur. 6,7 Therefore, the development of effective chemotherapy that is more comfortable and less dangerous for the patient and also less costly, to significantly decrease the financial burden on patients and health care systems, represents an urgent and unmet clinical need. To address this challenge, one can turn to the area of subcutaneous (SC) injectables. SC administration is indeed much more comfortable for the patient than IV administration, as it is less invasive and easy to implement. 8 Also, no hospital stay is required, making home chemotherapy and even self-administration possible. 9 Compared to the oral route, SC administration offers superior bioavailability (>80%), faster and better controlled absorption of the drug, drastically reduced compliance problems and less variability between patients. 10 The technologies currently developed for the SC administration of small drugs/therapeutic proteins are mainly based on either their direct administration, 10,11 with strategies to increase their aqueous stability (e.g., cyclodextrins, Biochaperone) 12,13 or SC injection volume (e.g., hyaluronidase), 14 or on the injection of drug-loaded nanoscale systems (e.g., hydrogels, nanoparticles, liposomes, lipid prodrugs). However, these approaches cannot be applied to the vast majority of anticancer drugs. The field of SC injectables for cancer therapy is indeed extremely restricted, 10 because most anticancer drugs (including very effective ones such as taxanes, vinca alkaloids, doxorubicine, etc.) are irritant or vesicant. They induce prohibitive local toxicity such as severe irritation and necrosis, 18 which are triggered by their prolonged retention in SC tissue due to their high lipophilicity. Anticancer drugs are thus repeatedly internalized by SC cells, causing their death and preventing the healing process. Herein, we report the first preclinical development of a general strategy for the SC administration of irritant/vesicant, anticancer drugs. Our idea is based on the design of watersoluble polymer prodrugs comprising one anticancer drug molecule attached at the extremity of a well-defined polyacrylamide (PAAm) chain (Figure 1a). PAAm is an uncharged, highly water-soluble and biocompatible polymer used in nanomedicine, 22,23 with stealth properties and also employed as permanent dermal filler (Aquamid®). 24 It thus fully meets the criteria for SC administration as recommended by Mrsny. 25 To demonstrate the proof of concept, we chose paclitaxel (Ptx), a representative hydrophobic irritant/vesicant anticancer drug widely used in the clinic. It was bound to PAAm via a cleavable linker positioned on its C2' hydroxyl group, 26 resulting in inactive Ptx-based prodrugs (Figure 1b). The prodrugs' characteristics thus: (i) prevent early release of the drug into the SC tissue; (ii) promote their diffusion throughout the SC tissue and absorption into blood/lymph capillaries to yield high bioavailability and (iii) allow the drug to be released into the bloodstream where it can exert its therapeutic activity (Figure 1c). We showed that our strategy is safe as no local toxicity was observed. Precise tuning of the prodrug structure also allowed us to greatly decrease the peak drug concentration (Cmax), responsible for systemic toxicity, 27 while achieving sustained drug exposure. Importantly, our approach enabled a 3-fold increase of the maximum tolerated dose (MTD) and therefore a greater anticancer efficacy when benchmarked against IV-administered Taxol, the most common commercial formulation of Ptx. ## Materials Acrylamide (AAm) (≥ 99%, Sigma-Aldrich) was recrystallized from chloroform. (3 Ci.mmol -1 , 1 mCi) was purchased from Moravek both used as received. Ptx-diglycolate-CDP was synthesized as described elsewhere. 28 Deuterated chloroform (CDCl3) and dimethyl sulfoxide (d6-DMSO) were obtained from Eurisotop. Taxol was purchased from Fresenius Kabi France. All solvents were purchased from Sigma-Aldrich at the highest grade. ## Analytical methods Nuclear magnetic resonance (NMR) spectroscopy. 1 H NMR and 13 C NMR spectroscopy of small molecules was performed in 5 mm diameter tubes in deuterated chloroform (CDCl3) on a Bruker Avance 300 spectrometer operating at 300 MHz ( 1 H) or 75 MHz ( 13 C) at room temperature. For 1 H NMR spectroscopy of polymers, acquisition was performed in 5 mm diameter tubes in d6-DMSO at 70 °C (128 scans) on a Bruker Avance 3 HD 400 spectrometer operating at 400 MHz. NMR determination of the number-average molar mass (Mn,NMR) of the prodrugs was performed by comparing the integration of the doublet at 8 ppm, corresponding to 2 aromatic protons from one of the Ptx aromatic groups (noted 3 and 7 in Supplementary Information, Figure S1) and the integration of the broad peak between 1.29 and 1.80 ppm corresponding to methylene protons of AAm repeat units. ## Size exclusion chromatography (SEC) . SEC was performed on a set-up from Viscotek (TDAmax) composed of a TDA 305 Triple Detector Array containing a differential viscometer, a right-angle laser-light scattering (90°, RALLS) detector, low-angle laser-light scattering (7°, LALLS) detector and refractive index (RI) detector. The chromatographic column set consisted of a guard column (PL, 50 × 7.5 mm) followed by two columns (PSS Gram, 300 × 8 mm; bead diameter 10 µm; molar mass range 500-10 6 g.mol -1 ). The system was equipped with a triple detection system (Viscotek TDA/GPCmax from Malvern) comprising a differential refractive index detector, low and right-angle light scattering detectors, a differential viscometer detector and a UV detector. The GPCmax was composed of an on-line degasser and a dual piston pump set at a flow rate of 0.7 mL.min -1 with DMSO as the eluent, previously filtered through a 0.2 µm filter. The TDAmax was thermostated at 50°C. The system was calibrated using a narrow pullulan standard and each polymer sample was injected at 5 different injection volumes to determine the refractive index increment (dn/dc = 0.057 mL.g -1 ). Before the injection (100 µL), the samples were filtered through a polytetrafluoroethylene (PTFE) membrane with 0.2 µm pore. This allowed the molar mass (Mn,SEC) and the dispersity (Ð = Mw/Mn) of the polymers to be determined by triple detection using the OmniSEC software version 4.6.1.354. Rheological measurements. All rheological measurements were carried out on a rotational rheometer ARG2 (TA instruments, New Castle, USA). The geometry was an aluminum plate/plate (diameter 20 mm) equipped with a solvent trap. The TRIOS software was used for data analysis. Flow properties of the prodrugs were determined at 20 °C by a stress sweep. After a 2-min equilibration time, the shear rate was increased gradually from 10 to 1000 s -1 . Injectability. Injectability tests were carried out using a custom-built device described previously. 29 This device was coupled to a texture analyzer TAXT2 (Stable MicroSystems, Godalming, UK) in compression mode which was equipped with a force transducer calibrated with a 30 kg sensor. 400 µL of solution are taken in a 1-mL syringe (MeritMedical, Medaillon® Syringe, USA) which is then fitted with a 26 G x ½'' needle (0.45 × 12 mm, Terumo Neolus, Japan) before injection at a 1 mm.s -1 rate. Liquid chromatography-tandem mass spectrometry (LC-MS/MS). Liquid chromatography conditions were as follows: C18 (HILIC) column (Nucleodur, EC 125/2, 100-5-C18, Macherey-Nagel, Hoerdt, France). Mobile phase: acetonitrile/water (50/50) with formic acid 0.1 %; run time: 8 min; flow rate: 0.3 mL.mL -1 . ESI-MS/MS Analyses were performed on a triple quadrupole mass spectrometer detector (TQD) with electrospray ionization (ESI) interface (Quattro Ultima, Waters, Guyancourt, France). Electrospray and mass parameters were optimized by direct infusion of pure analytes into the system. ESI parameters: capillary voltage 3.5 kV, cone voltage 35 V, source temperature 120 °C desolvation temperature 350 °C, with a nitrogen flow of 506 L.h -1 . Mass parameters: transitions were monitored as follows Ptx 854/286; Ptx-d5 859/291. Calibration: Calibration curve was linear in the range 5-1000 ng.mL -1 (y = 0.0047.x + 0.0838; R² = 0.9936 in PBS and y = 0.0052.x -0.0131; R² = 0.9949 in mouse plasma). ## Synthesis Synthesis of Ptx-ester-CDSPA and [ 3 H]-Ptx-ester-CDSPA. CDSPA (121 mg, 0.30 mmol), DMAP (40 mg, 0.33 mmol) and EDC.HCl (67 mg, 0.35 mmol) were dissolved in 2 mL anhydrous CH2Cl2 and mixed in a reaction flask under argon at room temperature. After 15 min, a solution of Ptx (100 mg, 0.12 mmol) in DMF (0.5 mL) was added dropwise into the flask. After stirring at 30 °C for 4 h, an additional 20 mg (0.10 mmol) of EDC.HCl solution in 200 µL anhydrous dichloromethane (DCM) was added. The reaction was stirred at 30 °C for another 22 h and was poured into 20 mL of ethyl acetate (EtOAc). The organic phase was washed with aqueous NaHCO3 and brine before being dried over MgSO4. The residue was concentrated under reduced pressure and purified by flash chromatography (SiO2, from DCM/EtOAc = 5/1 to DCM/EtOAc = 4/1, v/v) to give 88 mg (0.071 mmol) of Ptx-ester-CDSPA as a yellow, sticky solid. Yield = 61%. 1 For [ 3 H]-Ptx-ester-CDSPA, the procedure was the same except that [ 3 H]-Ptx was added in the reaction mixture as follows. Ethanol was carefully evaporated under vacuum from the initial stock solution of [ 3 H]-Ptx. [ 3 H]-Ptx (1 mCi) was then solubilized in 100 µL of DMF prior to addition to the reaction mixture containing non-radiolabeled Ptx (100 mg, 0.12 mmol) and the other reagents in 200 µL of DMF. The vial containing the initial [ 3 H]-Ptx was further rinsed twice with 100 µL of DMF and these volumes were added to the reaction mixture. The following steps were identical to those described for the synthesis of Ptx-ester-CDSPA and a mixture of Ptx-ester-CDSPA/[ 3 H]-Ptx-ester-CDSPA with a total activity of 258 µCi was obtained as a yellow sticky solid. Yield = 19 %. ## Synthesis of Ptx-carbonate-CDP. To a solution of Ptx (194 mg, 0.227 mmol) in dry DCM (4 mL) under an argon atmosphere was added 4 drops of pyridine. Then 4-nitrophenyl chloroformate (273 mg, 1.362 mmol) in dry DCM was added at -50 °C, the reaction mixture was stirred at -50 °C and after 4 h, 4-nitrophenyl chloroformate (183 mg, 0.908 mmol) was added again. After 1 h the mixture was diluted with DCM and washed with sodium bicarbonate (NaHCO3, 0.5 N) and brine and dried over anhydrous sodium sulfate. The organic layer was separated and evaporated under vacuum. After evaporation of the solvents the crude was purified by column chromatography (ethyl acetate-cyclohexane, 1:1), to yield activated paclitaxel. Yield 45 %. 1 Hz, 1H), 1.81 (s, 1H), 1.71 (s, 3H), 1.67 (s, 2H), 1.28 (s, 3H), 1.17 (s, 3H). 13 Activated Ptx (220 mg, 0.215 mmol) and CDP (83 mg, 0.215 mmol) in dry DCM (12 mL) were treated at room temperature with DMAP (31 mg, 0.258 mmol). The reaction mixture was stirred in the dark for 48 h and was then diluted with DCM. The organic layer was washed with saturated NaHCO3 and dried over anhydrous sodium sulfate. The organic layers were concentrated and the crude was purified with column chromatography using cyclohexane/ethyl acetate as eluant (using a gradient from 80/20 to 50/50). The compound was isolated as yellow powder. Yield 74 %. 1 H NMR (300 MHz, CDCl3): δ = 8.17 (m, 26H), 0.90 (t, J = 6.5 Hz, 3H). 13 In a 7-mL glass vial were added AIBN (0.8 mg, 0.005 mmol), the Ptx-functionalized RAFT agents [Ptx-ester-CDSPA (30 mg, 0.024 mmol, for P3e) or Ptx-diglycolate-CDP (30.18 mg, 0.022 mmol, for P3d) or Ptx-carbonate-CDP (30.18 mg, 0.024 mmol, for P3c)], AAm (454 mg, 6.39 mmol) and DMSO (1.6 mL). The mixture was degassed with argon for 15 min under vigorous stirring before being placed in a 70 °C-preheated oil bath for 24 h under stirring. After the reaction, the polymer was precipitated twice in methanol (MeOH). The polymer was further solubilized in DMSO and placed in a 3.5 kDa Spectra/Por 3 dialysis bag for dialysis against deionized water for 3 days, with dialysis water changed twice per day. The dialysate was then freeze-dried to yield Ptx-ester-PAAm (P3e), Ptx-diglycolate-PAAm (P3d) or Ptx-carbonate-PAAm (P3c), respectively, as a white-to-yellow, spongy solid. Another two polymerizations were carried out with [AAm]0/[PTX-ester-CDP]0 = 53 (P1e) and 123 (P2e). ## Synthesis of [ 3 H]-Ptx-ester-PAAm. The radiolabeled [ 3 H]-Ptx-ester-PAAm was obtained following the same procedure as for P3e except that the previously synthesized mixture of Ptxester-CDSPA/[ 3 H]-Ptx-ester-CDSPA was used as the RAFT agent and the purification only consisted in two precipitations in MeOH. The obtained polymer was thoroughly dried under vacuum before being dissolved directly in PBS. This solution was then mixed with a solution of non-radiolabeled Ptx-ester-PAAm P3e in PBS to the desired Ptx equivalent concentration and radioactivity for further in vivo studies. Multi-gram scale synthesis of Ptx-ester-PAAm. Synthesis was performed as described previously with some modifications. Briefly, in a round bottom flask, CDSPA + (3.86 g, 0.0095 mol), DMAP + (0.84 g, 0.0068 mol) and EDC.HCl + (1.78 g, 0.0093 mol) were dissolved in 20 mL anhydrous CH2Cl2 and 15 drops of anhydrous DMF ( + these reagents were added portionwise over 20 h), and mixed in a reaction flask under argon at room temperature. After 15 min, a solution of Ptx (4 g, 0.0046 mol) in DCM (20 mL) was added dropwise into the flask. After stirring at 30 °C for 29 h. The organic phase was washed with aqueous NaHCO3 and brine before being dried over MgSO4. The residue was concentrated under reduced pressure and purified by flash chromatography (SiO2, from DCM-EtOAc 8:2) to give Ptx-ester-CDSPA as a yellow solid. Yield = 85%. In a 250 mL round bottom flask were added AIBN (36 mg, 0.2 mmol), Ptx-ester-CDSPA (1.363 g,1.089 mmol), AAm (2.613 g, 290 mmol) and DMSO (72.5 mL). The mixture was degassed with argon for 15 min under vigorous stirring before being placed in a 70 °C-oil bath for 3 h under stirring. After the reaction, the polymer was precipitated twice in MeOH. The polymer was further solubilized in DMSO and placed in a 3.5 kDa Spectra/Por 3 dialysis bag for dialysis against de-ionized water for 5 days, with dialysis water changed twice per day. The dialysate was then freeze-dried to yield 14 g of Ptx-ester-PAAm (Mn,NMR = 24 000 g.mol -1 , Mn,SEC = 24 780 g.mol -1 , Đ = 1.17) as a white-to-yellow spongy solid. Yield = 70 %. Determination of residual acrylamide. Analyses were achieved by HPLC via isocratic runs (phosphate buffer mobile phase, 0.6 mL.min -1 flow rate) on a RP-C18 column, 5 µm particle size (250 × 4.6 mm) and a guard column (5 × 3.9 mm) at a wavelength detection of 208 nm and a temperature of 40 °C. Run time was 10 min. Isocratic analyses were performed with a phosphate buffer mobile phase (0.84 g KH2PO4 in 960 mL of H2O and 40 mL of MeOH). Concentrations of 0.1, 0.5, 1, 2, 10, 30, 50 and 100 μg.mL -1 of AAm in deionized water were used to build the calibration curve. Each concentration was injected 4 times. Samples from P3e at 25, 50 and 100 mg.mL -1 in deionized water were used to determine the residual amount of AAm. Column washing between each run was performed by 1 wash with distilled-deionized water and 1 wash with MeOH. ## In vitro evaluation Drug release. Ptx release experiments were performed in PBS (1X, pH 7.4 with 1 wt.% Tween 80) and in mouse plasma. Free Ptx, P3e, P3d and P3c (Table 1) were incubated in PBS and plasma at 37 °C at the same equivalent Ptx concentration (1 µg.mL -1 eq. Ptx). 200 µL samples were taken at 0, 2, 4, 6, 24 and 48 h, for quantification. The samples were mixed with 600 µL of acetonitrile and 20 μL of a solution of deuterated Ptx (Ptx-d5) at 1 µg.mL -1 (internal standard). Samples were shaken during 15 min and centrifuged at 3000 g for 10 min before analysis by LC-MS/MS. Cell culture and cytotoxicity. The cytotoxicity of the different prodrugs was evaluated on two human breast cancer cell lines (MCF-7 and SK-BR-3), obtained from ATCC (USA). SK-BR-3 cells were cultured in DMEM F-12 HAM supplemented with penicillin (50 U.mL -1 ), streptomycin (50 μg.mL -1 ), 20% heat inactivated FBS and 0.01 mg.mL -1 bovine insulin. MCF-7 cells were grown in EMEM supplemented penicillin (50 U mL -1 ), streptomycin (50 μg.mL -1 ), 10% heat-inactivated FBS, 1% non-essential amino acids (NEAA) and 5 mL glutamine. Both types of cells were maintained at 37 °C and 5% CO2 in a humidified atmosphere and were split twice weekly. The cell viability was evaluated using the 3-[4,5-dimethylthiazol-2-yl]-3,5diphenyltetrazolium bromide (MTT) assay. Cells were seeded in 100 μL of culture medium (8 × 10 3 cells/well for SK-BR-3 cells and 5 × 10 3 cells/well for MCF-7 cells) in 96 well plates (TPP, Switzerland) and pre-incubated for 24 h. 100 μL of a serial dilution of prodrug solution was then added to the medium. After 72 h of incubation, 20 μL of MTT solution (5 mg.mL -1 in PBS) was added to each well. After 4 h of incubation, the culture medium was gently aspirated and replaced by 200 μL DMSO to dissolve the formazan crystals. The absorbance of the solubilized dye, which correlates with the number of living cells, was measured with a microplate reader (LAB Systems Original Multiscan MS, Finland) at 570 nm. The percentage of viable cells in each well was calculated as the absorbance ratio between prodrug-treated and untreated control cells. Data was fitted to a Hill slope model with four parameters using GraphPad Prism (version 8.0.2) to determine the IC50. The different IC50 values were determined using a one-way ANOVA test with GraphPad Prism (version 8.0.2). Acute toxicity and histology. Groups of 3 mice were injected subcutaneously at day 0 in the inter-scapular region. The different groups are: (i) Taxol at 60 and 90 mg.kg -1 (positive control); ## In vivo evaluation (ii) PAAm at 700, 1400, 2100, 2800, 3500 and 4200 mg.kg -1 (polymer alone, negative control); (iii) Ptx-ester-PAAm P3e at 90, 120, 150 and 180 mg.kg -1 eq. Ptx; (iv) Ptx-carbonate-PAAm P3c at 90, 120, 150 and 180 mg.kg -1 eq. Ptx and (v) Ptx-diglycolate-PAAm P3d at 90, 120, 150 and 180 mg.kg -1 eq. Ptx. Taxol was also injected intravenously in the tail vain at 10, 20, 30 and 60 mg.kg -1 . Visual toxicities at the injection site and body weight were monitored daily to follow local and systemic toxicities. After 7 days, mice were euthanized by cervical dislocation and injection site were removed and fixed in PFA 4% (overnight). They were then transfer in ethanol 70% for maximum 1 week before paraffin-embedding (System Logos One, Micro France). After paraffin embedding, 4 microns thick tissue sections were made using a microtome (Autosection, Sakura). The slides were then stained (Austostainer XL, Leica) by Hematoxylin-Eosine-Saffron (HES) histopathological examination. A semi-quantitative scoring system, ranging from 0 (no change) to 3 (marked change), was applied. Pharmacokinetics of Ptx by mass spectrometry. Seven-week old female BALB/c OlaHsd mice (~22 g; Envigo, France) were divided into four different groups: (i) Taxol injected intravenously (7 mg.kg -1 ); (ii) Ptx-ester-PAAm injected subcutaneously (7 mg.kg -1 eq. Ptx); (iii) Ptxdiglycolate-PAAm injected subcutaneously (7 mg.kg -1 eq. Ptx) and (iv) Ptx-carbonate-PAAm injected subcutaneously (7 mg.kg -1 eq. Ptx). Each group was composed of 36 mice divided in 9 different time points (0.25, 0.5, 1, 2, 4, 7, 24, 48 and 72 h) leading to 4 mice per group. At each endpoint, mice were euthanized with pentobarbital and blood was sampled by cardiac puncture before plasma was recovered by centrifugation (5 min; 3000 g). After centrifugation, sample was prepared following the protocol bellow. Aliquots of 200 µL were mixed with 600 µL of acetonitrile and 20 μL of deuterated Paclitaxel (Ptx-d5) at 1 µg.mL -1 (internal standard). Samples were shaken during 15 min and centrifuged for 10 min before analysis by LC-MS/MS. Pharmacokinetics and biodistribution of radiolabeled Ptx. Seven-week old female BALB/cOlaHsd mice (∼22 g; Envigo, France) were used. Radiolabeled Taxol and radiolabeled (ii) Ptx-ester-PAAm SC at 15 mg.kg -1 (Taxol equivalent dose); (iii) Ptx-ester-PAAm SC at 60 mg.kg -1 (Taxol equivalent dose) which corresponds to Ptx-PAAm maximal tolerated dose and (iv) Taxol IV at 15 mg.kg -1 (Taxol maximal tolerated dose). Animal viability and behavior were observed daily, and body weights were measured twice a week. Tumor volume was measured twice a week with a caliper and estimated with the following formula: Volume = (length × width²) / 2. Mice were euthanized by overdosage on gas anesthesia (isoflurane) followed by cervical dislocation when Humane endpoints were reached. 31,32 Statistics. Statistics were performed using GraphPad Prism (version 8.0.2). Comparison of tumor growth results between groups were analyzed for statistical significance, using two-way ANOVA, with Tukey multiple comparisons post-hoc. ## Synthesis and characterization The polymer prodrugs were synthesized by the "drug-initiated" method, 33 which relies on the controlled growth of a polymer chain from a drug derivatized by a polymerization initiator/controlling agent to perform controlled polymerization. This strategy has been selected to facilitate clinical translation because it of its simplicity and scalability, since only a few highyielding synthesis steps are necessary. It is also very robust and flexible since it is easily applicable to different drugs/linkers/polymers, 28, leading to a broad range of different polymer prodrugs with tunable drug delivery properties. A small library of well-defined Ptx-PAAm prodrugs was synthesized by reversible addition-fragmentation chain transfer (RAFT) polymerization of AAm using Ptx-based, trithiocarbonate chain transfer agents (Figures 2 and S1). Three different linkers were investigated (ester, carbonate and diglycolate) to find the optimal balance in terms of linker stability vs. lability to prevent early drug release into the SC tissue, while ensuring its release into the blood before prodrug excretion (Figure 1b and 2). These linkers were chosen for their sensitivity to circulating enzymes with esterase activity and with moderate expression variability in humans, thus ensuring comparable interpatient drug release patterns. 38,39 Ptx-ester-PAAm was obtained by coupling Ptx to CDSPA as a chain transfer agent (Figures 2 and S1a), followed by RAFT polymerization at 70°C in DMSO using AIBN as initiator (Figures 2 and S2). By adjusting the [AAm]0/[Ptx-ester-CDSPA]0 ratio from 53 to 266, the PAAm chain length was varied to determine the minimal Mn that allows for complete solubilization of the prodrug in water, which is a prerequisite to prevent SC toxicity and warrant high SC bioavailability. 1 H NMR spectroscopy of the purified prodrugs showed all expected signals, especially amide, methylene and methine protons from the PAAm backbone together with aromatic and characteristic protons from Ptx (Figure S2). The prodrugs exhibited Mn,NMR ranging from 6 200 to 21 600 g.mol -1 in rather good agreement with Mn,SEC values (P1e-P3e, Table 1 and Figure S3) and low dispersities (Ð = 1.07-1.28), thus accounting for a controlled polymerization process. By tuning the PAAm chain length, the drug loading varied from ~14 to ~4 wt.%. Whereas P1e and P2e were only partially soluble in water at 3 mg.mL -1 eq. Ptx because of the too short PAAm chains, P3e (Mn,NMR = 21 600 g.mol -1 ) was fully water-soluble at this equiv. Ptx concentration, which represents a 104-fold increase in solubility compared with free Ptx. The structure of the RAFT agent was then modified to change the nature of the linker. Previous reports have shown that diglycolate-based linkers are highly labile in plasma with faster release kinetics than the ester counterparts, 28,36,40 whereas carbonate linkers have shown slower release kinetics. 41 Therefore, well-defined Ptx-carbonate-PAAm (P3c) and Ptx-diglycolate-PAAm (P3d) of similar Mn to that of P3e were synthesized (Figures S1-S3, Table 1). They were obtained by following an identical polymerization procedure to that of the Ptx-carbonate-CDP and Ptx-diglycolate-CDP functional RAFT agents, respectively (Figure 2). Those were synthesized by activation of Ptx by 4-nitrophenyl chloroformate followed by reaction with CDP, or by coupling Ptx to diglycolate-CDP. Successful clinical translation requires simple and robust manufacturing methods that ensure the preparation of newly developed materials in large scales and with a high level of purity. 42 In this context, we also performed a multi-gram scale synthesis of P3e where 4.8 g of Ptx-ester-CDSPA and 17.6 g of the corresponding polymer prodrug (Mn,NMR = 24 000 g.mol -1 , Mn,SEC = 24 780 g.mol -1 , Đ = 1.17) where obtained, with an overall yield of 60%. The high purity of P3e was assessed by HPLC, leading to residual amounts of free AAm and Ptx both below 1 ppm, much lower than the average dietary intake of AAm (1 μg.kg -1 body weight.day -1 ) 43 and below the threshold established by the European Medicines Agency for AAm in cosmetics. 44 ## Physicochemical characteristics and in vitro evaluation Prior to performing biological evaluations, key physico-chemical characteristics were investigated: (i) the viscosity and injectability of the prodrugs in aqueous solution, to ensure they can be injected under standard conditions used for SC administration and (ii) the release kinetics of Ptx from the prodrugs in different media, to assess its fine tuning depending on the prodrug's structure. Measuring the viscosity and injectability (i.e., force required for injection) of the prodrugs in aqueous solution is of crucial importance as the maximum volume generally accepted for a SC injection is ~2 mL, thus requiring administration of the relatively concentrated solutions to reach the same dose regimens as the IV-administered counterparts. Whereas the viscosity of PAAm (Mn,SEC = 37 000 g.mol -1 , Đ = 1.10) synthesized by the same procedure was close to that of water (< 10 cP) at 50 mg.mL -1 , viscosity of P3e was ~200 cP at 50 mg.mL -1 and increased to ∼1×10 4 cP at 200 mg.mL -1 (Figure S4). This is due to the presence of strongly hydrophobic Ptx moieties that induce the formation of hydrophobic domains, via Ptx-Ptx and likely Ptx-C12 alkyl interactions, decreasing the mobility of the polymer chains. The injectability of aqueous solutions of P3e, P3d and P3c was measured as the function of the concentration with a 26 G × ½'' needle, as the preferred needle size for humans is ∼25-27 G. Up to 50 mg.mL -1 , injection of the polymer prodrugs required a very low force of ~1 N, which was comparable to that of PAAm (Figure S5). Despite an increase in viscosity with the polymer prodrug concentrations, a concentration as high as ~130 mg.mL -1 was achieved (corresponding to ~6 mg.mL -1 in Ptx) at 30 N, which is the maximum acceptable injection force for SC administration. 45 The release of Ptx from the prodrugs P3c, P3d and P3e was then monitored in PBS and in murine plasma at 37 °C to investigate the influence of both the nature of the linker and of the medium (i.e., hydrolytic vs. hydrolytic + enzymatic cleavage) on the release kinetics. The diglycolate moiety of P3d showed a dual hydrolytic/enzymatic susceptibility resulting in the fastest release of Ptx in both media (~50% in PBS after 20 h and ~90% in plasma after 5 h) (Figure 3a and 3b). By comparison, P3e and P3c were both stable in PBS up to at least 70 h and gave comparable Ptx release kinetics in plasma (~40% after 24 h). Release kinetics were not monitored beyond 24 h in plasma due to the documented degradation of Ptx under these conditions. 46,47 To assess whether the drug release profiles observed in plasma correlate with the cytotoxicity of the prodrugs, cell viability experiments were performed by measuring the mitochondrial activity via MTT assay on two breast cancer cell lines (MCF-7 and SK-BR-3), corresponding to clinically relevant cancer models for Ptx. Importantly, all prodrugs led to significant cytotoxicity on both cell lines and their IC50 values were in the following order: P3d < P3e < P3c. While PAAm was not cytotoxic (> 75% cell viability) up to 500 nM on both cell lines, free Ptx gave an IC50 as low as 5 nM (Figure 3c-3f). Since Ptx must be released from the prodrug before passively diffusing through the cell membranes to reach the microtubules, slow release in plasma might be correlated with a high IC50. It is also interesting to note that: (i) due to the high lability of the diglycolate linker, P3d has the same IC50 as that of free Ptx and (ii) despite similar drug release profiles for P3c and P3e in PBS and plasma, P3e led to much lower IC50 than that P3c, possibly due to differences in the enzymatic composition of murine plasma and cell culture medium. ## Systemic and acute local toxicity The systemic toxicity of the prodrugs was then examined in mice to evaluate the MTD (i.e. the threshold at which all animals survived with a body weight loss lower than 10%) to find optimized treatments, followed by evaluation of the acute local toxicity at the injection site (Figure 4). Increasing concentrations of free PAAm and prodrugs P3e, P3d and P3c were SC injected (PAAm SC , P3e SC , P3d SC and P3c SC , respectively) to healthy mice (single injection), followed by monitoring of their body weight and their behavior for 7 days (Figure 4a). The same protocol was applied to SC and IV injections of Taxol (Taxol SC and Taxol IV , respectively). Whereas Taxol IV led to a MTD of 60 mg.kg -1 , Taxol SC allowed to reach 90 mg.kg -1 , probably due to a decrease in Cmax compared with IV administration and thus a dose-limiting reduction in Cmax-related. 48 Mice treated with free PAAm SC showed no sign of systemic toxicity up to a concentration as high as 6000 mg.kg -1 , in good agreement with its well-documented biocompatibility/safety. Importantly, all prodrugs were successfully SC injected up to at least 180 mg.kg -1 equiv. Ptx without exceeding a body weight loss of 10%. Neither mortality nor noticeable modification in terms of feeding and behavior were observed, thus suggesting absence of systemic toxicity. Notably, the MTD was increased at least by a factor 3 and 2 compared to Taxol IV and Taxol SC , respectively. 3) 7 days after injection of Taxol IV at 60 mg.kg -1 , Taxol SC at 60 mg.kg -1 , PAAm SC at 4.2 g.kg -1 , and P3e SC , P3d SC and P3c SC at 180 mg.kg -1 (equiv. Ptx). The black arrows indicate necrotic areas. (c) Representative HES-stained sections of skin samples from mice removed at the injection site after injection of Taxol SC at 90 mg.kg -1 , PBS, PAAm SC at 4.2 g.kg -1 , and P3e SC , P3d SC and P3c SC at 180 mg.kg -1 (equiv. Ptx). The black/white arrows indicate the severe cutaneous necrosis, only observed after SC injection of Taxol at 90 mg.kg -1 . (d) Histopathological scoring (H-Score) of degenerative/necrotic changes and tissular inflammation in mice after injection of Taxol SC at 90 mg.kg -1 , PAAm SC up to 4.2 g.kg -1 , and P3e SC , P3d SC and P3c SC up to 180 mg.kg -1 (equiv. Ptx). The values are expressed as the means ± SD. Unpaired two-tailed t test between Taxol SC group and PAAm SC , P3e SC , P3d SC or P3c SC group; * (p < 0.05). See all pictures and individual scores in Figure S6 and Table S1. Similarly to free PAAm SC , none of the prodrugs showed local toxicity at and near the injection site up to 180 mg.kg -1 equiv. Ptx (Figure 4b). This observation likely ruled out early Ptx release in the SC tissue from the prodrugs even from P3d SC that contains the most labile linker. Conversely, Taxol IV and Taxol SC led to significant ulceration and necrosis of the mice skin tissue at 60 mg.kg -1 (see black arrows in Figure 4b), in agreement with the literature. 49 Histopathological examination of HES-stained sections of skin samples removed at the injection site confirmed the above-mentioned macroscopic observations (Figure 4c). SC administration of the different polymer prodrugs evidenced a preserved architectural structure of the skin/SC tissue, with only focal small granulomatous lesion along needle tract. Neither significant degenerative or necrotic tegumentary changes were observed, nor inflammatory reaction, associated with the polymer prodrugs injection. On the contrary, Taxol IV and especially Taxol SC induced marked to severe ulcerative dermatitis with epidermal changes including hyperplasia and hyperkeratosis or severe epidermal-dermal necrosis replaced by a sero-cellular crust. Deep dermal and hypodermal inflammation was observed, granulomatous and/or granulocytic, associated with pannicular cytosteatonecrosis. Altogether, these results establish for the first time the possibility to safely administer a vesicant/irritant anticancer drug by SC injection. ## Pharmacokinetics and biodistribution The biological fate of the prodrugs was then evaluated in terms of pharmacokinetics and biodistribution in mice. A first pharmacokinetic study based on LC-MS/MS allowed to follow the evolution in time of the Ptx concentration coming from Taxol IV or released from the prodrugs at 7 mg.kg -1 equiv. Ptx after SC administration. Taxol IV exhibited a high Cmax of 4 660 ng.mL -1 15 min post-administration (tmax) followed by rapid clearance with undetectable amounts in plasma after 24 h (Figure 5a), in good agreement with previous pharmacokinetic studies of Taxol. 50 Conversely, the Cmax values of P3d SC , P3e SC and P3c SC were lowered by at least an order of magnitude, to reach 310, 105 and 41 ng.mL -1 , respectively (Table 2). These results are in agreement with the MTD of the prodrugs from the toxicity study (Figure 4), as lower Cmax values led to decreased toxicity and thus enabled a higher MTD than Taxol. 48,51,52 Interestingly, the Cmax values were observed at ~1-2 h (tmax) for all prodrugs. This delayed tmax compared to that of Taxol IV is attributed to the time required for the prodrugs to be absorbed into the blood or lymphatic capillaries, combined with the prolonged release of Ptx from the prodrugs once they reach the bloodstream. Notably, P3e SC showed a very different PK profile to the other prodrugs and Taxol IV . Whereas the elimination half-lives (t1/2) of P3d SC , P3c SC and Taxol IV were in the range of 1.5-1.7 h, t1/2 of P3e SC approached 14 h and it was detectable for more than 3 days. The mean residence time (MRT) was also much higher for P3e SC (22.2 h vs. 0.9-3.3 h). The values are expressed as the means ± SD (n = 4). The horizontal dashed line represents the limit of quantification (0.14%) a Determined according to AUC0→∞ / AUC0→∞ IV. The apparent bioavailability of Ptx for P3d SC , P3e SC and P3c SC amounted to 21%, 28%, and 4% relative to Taxol IV, respectively (Table 2). This makes P3e SC the best candidate as it possessed both the most suitable PK profile and the highest apparent bioavailability. Despite similar apparent bioavailability for P3e SC and P3d SC , P3d SC exhibited a lower MRT and rapid release of Ptx once in the blood, leading to a too rapid clearance of the drug. For P3c SC , Ptx was released too slowly and the prodrug was therefore excreted before it could effectively release its payload. The optimal performance of P3e SC could be explained by its intermediate Ptx release profile in vivo (probably due to the presence of specific enzymes such as esterases), combined with the stealth properties provided by PAAm. 22,23,53 This resulted in a long circulating prodrug acting as a slow-release reservoir of Ptx. These results are important not only because they confirm that the nature of the linker plays a key role in the pharmacokinetics of Ptx, but also because they show that bioavailability does not correlate linearly with the drug release pattern and thus screening each prodrug in vivo was necessary. P3e was then selected for further study. A radiolabeled counterpart (P3e*) was synthesized from [H 3 ]-Ptx and used in a second pharmacokinetic study at the same dose to monitor the whole amount of Ptx in comparison to that of radiolabeled Taxol* (Figure 5b). Since quantification is performed by radioactivity counting, free [H 3 ]-Ptx, P3e* and their metabolites were dosed all together, which allows the fate of the prodrug to be followed. Free [H 3 ]-Ptx administered intravenously (Taxol* IV ) was rapidly cleared from the blood compartment (<1% of the injected dose still circulating at 30 min post-injection, Figure 5b) and exhibited most of the pharmacokinetic parameters similar to those previously observed by LC-MS/MS (Table 2). In comparison, Taxol* SC showed a delayed entrance into the blood circulation, as shown by its very low Cmax (< 1% of the injected dose) and bioavailability of 25%. Ptx from IV-injected P3e* (P3e* IV ) has a prolonged circulation time with t1/2 10 times and an AUC 100 times greater than Taxol* IV (Table S2). Remarkably, Ptx from SC-injected P3e* (P3e* SC ) exhibited a high bioavailability (84% relative to P3e* IV ) and a total dose slowly increasing over time, from 3% of the injected dose 15 min post-injection up to 46% after 4 h. Once in the blood compartment, the prodrug remained in circulation for a prolonged period of time (MRT ~36 h), with a final Ptx blood concentration of still ~1% of the injected dose 6 days after injection, similarly to that of P3e* IV . It is worth noting that P3e* SC and P3e* IV exhibited the same t1/2 value of ~25 h, revealing that absorption rate is not significant after 24 h, suggesting quantitative absorption of P3e* SC into the blood within this period of time. From the biodistribution study into key organs, both P3e* SC and P3e* IV showed very limited accumulation in the liver (< 10% of the injected dose) 48 h post-injection, compared to 30% of the injected dose for Taxol* IV after 30 min, presumably as a result of the stealth properties of the prodrugs (Figure 5c). For other organs (lungs, spleen, kidneys), the total concentrations of Ptx from P3e* SC , P3e* IV and Taxol* IV were low and in the same range (except a modest accumulation of P3e* IV in the spleen), revealing no noticeable acute toxicity. The total amount of Ptx from P3e* SC was also monitored in the SC tissue. It decreased sharply over time, in parallel with an increase in the bloodstream, as shown from the pharmacokinetic profile (Figure 5c). The SC data further prove the rapid blood passage of the hydrophilic prodrug from the SC tissue. Overall, taking into account the PK/BD data and the toxicity study, these results argue for efficacy studies of P3e SC in mouse tumor models. ## Anticancer efficacy An efficacy study was then designed to address two important points. Will P3e SC be as efficient as Taxol IV at the same dose? And if yes, can P3e SC outperform Taxol IV at a higher dose thanks to its higher MTD? In this context, mice bearing MCF-7 xenografts were treated with: (i) PAAm SC at 1520 mg.kg -1 , which would correspond to 60 mg.kg -1 equiv. Ptx for the prodrug counterpart; (ii) Taxol IV at 15 mg.kg -1 , determined to be the MTD for a weekly injection repeated over three weeks and (iii) P3e SC at two different doses; either 15 mg.kg -1 equiv. Ptx (to have the same dose as for Taxol IV ) or at a four-time higher dose of 60 mg.kg -1 (determined to be the MTD in equivalent Ptx of P3e SC for such a dose regimen). The antitumor efficacy of the different treatments was evaluated by following the tumor growth (Figure 6a), from which two key metrics used to characterize the antitumor activity were extracted: 54 the tumor volume over control volume (T/C) (Figure 6b) and the tumor growth inhibition (TGI) (Figure 6c). The overall survival of mice (Figure 6d) during the study and the mice body weight evolution (Figure 6e) were also monitored. and P3e SC at 60 mg.kg -1 (P3e SC-60 ) equiv. Ptx, on days 0, 7 and 14 (black arrows). PAAm SC -treated mice exhibited rapid tumor growth with an average tumor volume exceeding 1500 cm 3 ~40 days post-tumor induction (Figure 6a and S7). Conversely, Taxol IV and P3e SC at 15 mg.kg -1 both showed similar anticancer activity as attested by reduction on tumor growth compared to PAAm SC (Figure 6a and S7), together with similar T/C (59-64%) and TGI (40-46%) values ten days after treatment termination (Figures 6b and 6c, Tables S2 and S3). This first result is of crucial importance as, despite the lower apparent bioavailability of Ptx from P3e SC (Table 2), the efficacy study revealed that it had a similar antitumoral activity as Taxol IV at the same dose. In combination with the toxicity data, this suggested a successful and safe transposition from IV-injected Taxol to SC-injected Ptx in the form of a water-soluble polymer prodrug. Remarkably, when P3e SC was administered at a higher dose of 60 mg.kg -1 equiv. Ptx, it displayed a dose-dependent anticancer activity and outperformed Taxol IV (Figure 6a and S7) with a T/C as low as 35% and a much higher TGI value of 73% ten days after treatment termination (Figures 6b and 6c, Tables S2 and S3). Consequently, not only was SC administration of P3e successful, but it could also induce greater anticancer activity than Taxol IV thanks to its higher MTD. In terms of overall survival of mice (Figure 6d), P3e SC administered at 60 mg.kg -1 equiv. Ptx led to the highest survival rate of 78%, 25 days after treatment termination, whereas it was 44% for P3e SC at 15 mg.kg -1 , only 33% for Taxol IV and 0% for the control group (PAAm SC ). As a result, P3e SC more than doubled the survival rate compared to Taxol IV . The evolution of the relative body-weight loss in P3e SC -treated mice also revealed that the treatment was well tolerated as mice lost no more than 10 % of their body weight throughout the efficacy study (Figure 6e). ## Conclusion In this work, we presented a novel and general approach for the SC administration of irritant/vesicant anticancer drugs via the design of well-defined hydrophilic polymer prodrugs constructed by the "drug-initiated" method. To validate our strategy, we applied it to the anticancer drug Ptx as a worst-case scenario due to its high hydrophobicity and vesicant nature, while PAAm was chosen due to its high hydrophilicity and stealth properties. We first synthesized a small library of Ptx-based polymer prodrugs by screening different linkers (ester, diglycolate and carbonate) and choosing the appropriate chain length (Mn ~20 kg.mol -1 ) to obtain fully water-soluble polymer prodrugs. We then performed a comprehensive preclinical development of these polymer prodrugs by studying their physicochemical properties, drug release kinetics on two different cancer cell lines and acute local and systemic toxicity, as well as their pharmacokinetic and biodistribution profiles, and anticancer efficacy in tumor-bearing mice of the most promising candidate (i.e., Ptx-ester-PAAm). We demonstrated that SC injection of hydrophilic polymer prodrugs based on Ptx as a representative vesicant/irritant anticancer drug allowed sustained release of Ptx in the bloodstream and outperformed the anticancer efficacy of Taxol, the commercial formulation of Ptx, without inducing local toxicity. Given the flexibility of the synthetic approach, these achievements pave the way for SC administration of a wide range of anticancer drugs, including irritant and vesicant ones, and make it possible to safely consider the translation of many IV chemotherapies to SC chemotherapies. From a more general perspective, this new drug-delivery platform could also represent an important step towards self-administration and chemotherapy at home, which would greatly increase patient comfort and reduce the high cost of cancer treatment; the latter being crucial for low-and middle-income countries.

chemsum

{"title": "A Polymer Prodrug Strategy to Switch from Intravenous to Subcutaneous Cancer Therapy for Irritant/Vesicant Drugs", "journal": "ChemRxiv"}

dna-templated_control_of_chirality_and_efficient_energy_transport_in_supramolecular_dna_architecture

## Abstract: Two conjugates of tetraphenylethylene with D-2 0 -deoxyuridine (1D) and L-2 0 -deoxyuridine (1L) were synthesized to construct new supramolecular DNA-architectures by self-assembly. The non-templated assemblies of 1D and 1L show strong aggregation-induced emission and their chirality is exclusively controlled by the configuration of their sugar part. In contrast, the chirality of the DNA-templated assemblies is governed by the configuration of the DNA, and there is no configuration-selective binding of 1D to D-A 20 and 1L to L-A 20 . The quantum yield of the assembly of 1D along the single-stranded DNA A 20 is 0.40; approximately every second available binding site on the DNA template is occupied by 1D.The strong aggregation-induced emission of these DNA architectures can be efficiently quenched and the excitation energy can be transported to Atto dyes at the 5 0 -terminus. A multistep energy transport "hopping" precedes the final energy transfer to the terminal acceptor. The building block 1D promotes this energy transport as stepping stones. This was elucidated by reference DNA double strands in which 1D was covalently incorporated at two distinct sites in the sequences, one near the Atto dye, and one farther away. This new type of completely self-assembled supramolecular DNA architecture is hierarchically ordered and the DNA template controls not only the binding but also the energy transport properties. The high intensity of the aggregation-induced emission and the excellent energy transport properties make these DNA-based materials promising candidates for optoelectronic applications. ## Introduction Conventional chromophores suffer from aggregation-caused quenching (ACQ), a phenomenon describing the quenched emission at high concentrations. For technical applications, particularly OLEDs, chromophores with strong fluorescence in the aggregated or condensed phase are crucial. The term "aggregation-induced emission" (AIE) was introduced by Tang's group in 2001 describing the emissive behavior of 1,2,3,4,5pentaphenylsilole in ethanol. 1 Since then, the number of organic chromophores that show AIE is constantly increasing; 2,3 in particular, tetraphenylethylene (Tpe) has been identifed as a lead structure with AIE. 4 Similar to the aforementioned silol, the propeller-shaped and non-planar Tpe is weakly emissive in dilute solution, but becomes highly emissive in the aggregated or solid state. It is not a distinct type of stacking interaction that increases the emission but rather the restriction of intramolecular rotations blocking the non-radiative decay pathways from the photoexcited state. 5 The supramolecular organization of chromophores gives access to nanostructured materials with unique optical properties. 6 The bottom-up approach is a powerful concept to achieve this goal, 7 and DNA is a unique template with sequence specifcity through the recognition by hydrogen bonding to arrange chromophores in a precise way, 8 and this DNA-like structure also persists in the solid state for optoelectronic applications. 9 The most precise, covalent incorporation of chromophores into DNA is limited to 5-10 units in a row, until the yields drop. 10,11 The non-covalent self-assembly of nucleosides along single-stranded DNA is an important alternative, for instance with naphthalenes 12 and porphyrins 13,14 through binding to T 20 or T 40 . We recently described the sequenceselective assembly of pyrenes, perylenes 18 and nile red 16,17 as modifed 2 0 -deoxynucleosides along single-stranded DNA templates and mainly ACQ was observed. However, AIE would increase the value of DNA-based materials for optoelectronics. This makes Tpe a promising candidate for DNA-based architectures with AIE. 19 The fluorescence intensity increase of typical DNA staining agents, such as SYBR Green, is a phenomenon related to AIE since rigidifcation of the chromophore has an impact on the fluorescence in both cases. 19 However, it should be noted here that, molecules, particularly derivatives of Tpe, showing AIE in assemblies are promising candidates for new materials. Achiral Tpe derivatives promote sensing of DNA by their emission and circularly polarized luminescence. 23,24 The covalent conjugation of Tpe with DNA 25 helps in the visualization of cellular RNA 26 and gives DNA-grafted nanosheets, 27 and the AIE can be controlled by DNA hybridization. 28 To construct new supramolecular DNA-architectures with strong AIE by self-assembly, we present two chiral conjugates of Tpe, 1D and 1L (Fig. 1). The confguration of their 2 0 -deoxyribofuranosides differs to probe the chirality of both the non-templated assembly and the assembly along D-and L-DNA templates by means of optical spectroscopy. These supramolecular architectures are hierarchically ordered and the DNA template controls the binding selectivity, the chirality and the light harvesting through efficient energy transport to the Atto dye attached to the 5 0 -terminus. ## Non-templated supramolecular assemblies Both conjugates 1D and 1L were synthesized by Sonogashiracoupling as described in the ESI (Schemes S1, S2-S7). † Firstly, the formation of assemblies with 1D and 1L, respectively, was investigated in aqueous solution without any DNA template. Stock solutions of 1D and 1L in THF or DMSO (4.3 mM) were prepared to ensure full solubility of the modifed 2 0 -deoxyuridines. The assembly of 1D and 1L was induced by diluting samples (8.8 mL) of the stock solutions with H 2 O (total volume of 1 mL, 37.5 mM in H 2 O with 0.9% THF or DMSO). The frst samples of 1D and 1L were simply stored for 1 h at room temperature and characterized by optical spectroscopy (Fig. 2). A second set of samples was treated under the typical conditions of DNA annealing, including incubation at 90 C for 5 min and slow cooling to room temperature (approximately 1 C min 1 ). Interestingly, these different preparations yield different types of water-soluble assemblies. Representatively, we discuss the optical properties of the samples that were prepared with THF as cosolvent; the corresponding results with DMSO as cosolvent are similar (see Fig. S22 †). The fluorescence of the assemblies of 1D and 1L prepared at room temperature show maxima at 478 nm, whereas the annealed assemblies show maxima at 442 nm and reduced intensities. The frst type of assembly is dominated by Tpe interactions, whereas the second, the annealed type of assembly is, at least partially, controlled by stacking of the 5ethynyl-2 0 -deoxyuridines. Accordingly, the electronic decoupling of Tpe from the 5-ethynyl 2 0 -deoxyuridine part by a rotational twist along the phenylene bridge causes a blue-shift of fluorescence. Chirality is an essential feature of these supramolecular structures and therefore the CD spectra were recorded. Remarkably, the CD spectra of assemblies of 1D and 1L from both types of preparation show strong mirror bisignate signals in the absorption range between 305 nm and 450 nm of the Tpe chromophore. Interestingly, the D-confgured conjugate 1D yields an assembly with left-handed helicity according to this bisignate signal, 29 whereas 1L yields assemblies with righthanded chirality. The CD signals of the annealed assemblies of 1D and 1L look similar in the range between 305 nm and 400 nm and their crossing points are shifted from 350 nm to 370 nm. The absorbance range between 200 nm and 305 nm, which is the typical absorption range for the 5-ethynyl-uracil part of the conjugates, shows another bisignate signal. Here, the annealed assembly formed with 1D shows left-handed helicity whereas the annealed assembly formed with 1L shows right-handed chirality. Their crossing point is at 270 nm, which is near the absorbance maximum of 5-ethynyl-2 0 -deoxyuridine. This supports our aforementioned hypothesis that the annealed assemblies are controlled by the nucleoside stacking. The chirality order may be unexpected, but, in fact, it is not. For instance, the D-confgured nile red-modifed 2 0 -deoxyuridines yielded assemblies with left-handed chirality, too. 30 The CD spectra show very clearly that the chirality of the assemblies of 1D and 1L is exclusively controlled by the confguration of the attached 2 0 -deoxyribofuranosides. The fluorescence detected circular dichroism (FDCD) combines conventional CD with the high sensitivity of fluorescence spectroscopy. It allows investigating the chiral behavior of fluorescent chromophores in a non-emissive chiral architecture, such as DNA. 31 The detected FDCD signals are mirror bisignate signals and again show that the chirality of the assemblies of 1D and 1L is controlled by the different confgurations of the attached chiral 2 0deoxyribofuranosides. ## DNA-templated supramolecular architectures The next level of hierarchically ordered supramolecular architectures is the structural control of the self-assembly of 1D and 1L in aqueous solution by DNA templates. Therefore, we followed our established protocol to prepare such DNA-based supramolecular architectures. 32,33 We used 1D and 1L as building blocks and probed their assembly along D-and L-confgured DNA templates with the sequences A 20 and T 20 by means of optical spectroscopy. Small volumes of 2 0 -deoxynucleoside stock solutions in DMSO were added to an aqueous solution of the DNA templates (1 mL, 1.25 mM). The concentrations of both stock solutions were prepared sufficiently high that not more than 0.9% DMSO as cosolvent was added to the fnal samples in water. This amount of cosolvent is typically tolerated by the DNA conformation. Both, 1D and 1L, were added in a 1.5-fold excess to promote the occupancy of the available binding sites along the templates. 250 mM NaCl is known to stabilize double-helical DNA and thus expected to also support the formation of the noncovalent supramolecular DNA architectures with 1D and 1L. Both conjugates are nearly insoluble in this high salt aqueous solution unless they are bound to single stranded DNA and therefore remain in the solution. The excess and unbound 1D or 1L was removed by short centrifugation (3 min @ 16 000g). The supernatant contained the DNA-templated assembly and was investigated by optical spectroscopy (Fig. 3). The absorbances in the range between 300 nm and 400 nm of the assemblies of 1D with D-A 20 and 1L with L-A 20 as templates are signifcantly higher than those with D-T 20 and L-T 20 . This result reveals preferred binding of 1D and 1L to D-A 20 and L-A 20 , respectively, which matches the rules of canonical base-pairing, because both modifed 2 0 -deoxuridines are complementary to 2 0 -deoxyadenosines in the DNA template A 20 , but not complementary to T 20 . This selectivity is not as pronounced as for the Watson-Crick base pairing in unmodifed double-stranded DNA because "mismatched" binding of 1D and 1L to the wrong templates D-T 20 and L-T 20 occurs to a certain extent. There is, however, no confguration-selective recognition by the DNA templates. The absorbances of the assemblies of 1D with D-A 20 and L-A 20 are equally high (Fig. 3), which is also the case for the assemblies of 1L with both D-A 20 and L-A 20 (Fig. S23 †). The absorbances can be used to calculate the occupancy fraction f which is the number of occupied binding sites (by 1D or 1L) divided by the number of available binding sites on the template. For the assemblies of 1D and 1L along D-A 20 and L-A 20 , f lies in the range of 0.55 AE 0.1; approximately 11 out of 20 available binding sites of the DNA templates are occupied. Obviously, the four phenyl groups of Tpe are sterically hindering and prevent a complete occupancy of all binding sites, as previously observed for planar chromophores, such as nile red. 32 The fluorescence gives additional support for the selective binding of the chromophore-nucleosides. Since Tpe shows AIE and unbound chromophore conjugates were removed by centrifugation, the fluorescence can be clearly attributed to the DNA-templated assemblies of 1D and 1L. Although confguration selectivity was not observed for the binding of 1D and 1L to the DNA templates, the chirality of the formed helical assemblies was probed by CD spectroscopy. The CD spectra of the DNA-templated assemblies of 1D and 1L show generally weaker signals than those of the non-templated assemblies, and there are only signals of 2 0 -deoxynucleosides in the absorbance range between 200 nm and 300 nm (dU in the Tpe-conjugates and A or T in the templates). In contrast to the non-templated assemblies of 1D and 1L, as discussed above, the chirality in the templated assemblies is not controlled by the confguration of 2 0 -deoxyribofuranosides in 1D and 1L, but instead by the confguration of the DNA templates. The CD signal of all four DNA templates shows the slightly different helical preorganized conformations of the single strands (Fig. S24 †) which are not changed by the binding of 1D (Fig. 3) or 1L (Fig. S23 †). According to the characteristic bisignate signals with a zero crossing at approximately 280 nm, assembling 1D with D-A 20 and D-T 20 yields right-handed helicity, whereas with L-A 20 and L-T 20 it gives left-handed helicity in accordance with the expected chirality for such DNA-like helical structures. There is no measurable CD above 305 nm and thus no information on the ordered chirality of the Tpe chromophores along the DNA templates. Only approximately every second binding site of the DNA templates is occupied by 1D or 1L, preventing the chromophores from CD-active excitonic interactions. Based on the determined general helicity of the DNA-like architectures as determined by their CD signals between 200 nm and 300 nm, we assume that the Tpe chromophores are also helically arranged. The next step towards DNA-based light harvesting systems is to transfer the excitation energy from the aggregation-emissive Tpe to an appropriate energy acceptor. For the preparation of such energy transfer systems, we focused solely on the components with D-confguration (1D and D-A 20 ) due to the commercial availability of appropriate dye-DNA conjugates. But in principle, we assume that the following experiments would also work with the components with L-confguration. The DNA template D-A 20 bears the Atto565 and the Atto633 dye conjugated to the 5 0 -termini that serve as energy acceptors in the supramolecular assemblies of 1D. The preparation was carried out following the previous protocol including centrifugation to remove unbound 1D by precipitation from the solution. The fluorescence of the Tpe chromophores in the assemblies along the A 20 -Atto565 and A 20 -Atto633 templates at 492 nm is strongly reduced compared to that in the assembly of 1D along When control experiments with the "wrong" templates T 20 were performed, T 20 -Atto565 and T 20 -Atto633 show signifcantly less energy transfer since very few molecules of 1D are bound to these "wrong" DNA templates according to the low absorbance at 345 nm after removal of unbound 1D by centrifugation (Fig. S26 †). As mentioned above, the assembly of 1D along D-A 20 (without any attached Atto dye) has a quantum yield of F F ¼ 0.40. Accordingly, the emission of the 1D assemblies along A 20 -Atto565 and A 20 -Atto633 as templates is quenched by 86% and 80%, respectively. The excitation spectra clearly show that the emission of the Atto dyes originates from the excited 1D units. If we assume an occupancy fraction of f ¼ 0.55 for the 20 binding sites available on these templates, the emission of approximately 9 out of 11 DNA-bound molecules of 1D is quenched. Based on a regular stacking distance of 3.4 , the farthest 1D engaged in energy transfer would be approximately 58 away from the Atto dye. The efficient quenching over such long distances is likely not the result of one-step energy transfer processes from the individual Tpe that is assembled as a donor at different distances to the 5 0 -terminal Atto dye as the acceptor. In contrast, the efficient quenching indicates a step-wise energy transport between the Tpe molecules taking place before the fnal energy transfer to the Atto acceptor dyes occurs. In order to probe the distance dependence of the energy transfer between 1D and the Atto dyes, we incorporated 1D as a modifed nucleotide into synthetic single-stranded DNA. The optical properties of this covalent modifcation were compared with the non-covalent supramolecular assemblies (Fig. 5). The phosphoramidite as a DNA building block was synthesized by standard procedures and used for automated oligonucleotide synthesis on a solid phase as described in the ESI (Fig. S8-S12). † The synthetic oligouncleotides TPE1a, TPE1b, TPE2a and TPE2b were purifed by semi-preparative HPLC and were checked by MALDI-TOF mass spectrometry. TPE1a bears a single 1D modifcation in the middle of the DNA sequence, and TPE1b, nearer to the 5 0 -terminus. The fluorescence of these single-stranded oligonucleotides show maxima at 489 nm. The fluorescence intensity is signifcantly reduced and the maximum is shifted to 430 nm in the double stranded hybrid TPE1a-DNA1, but not in TPE1b-DNA1, when the modifed single strands were annealed with the complementary and unmodifed counterstrand DNA1. This was similarly observed in the annealed assemblies of 1D (see above). We explain this observation again by stacking the 5-ethynyluracil part in a DNA-like double-helical assembly which induces a rotational twist to the phenylene group and decouples the Tpe chromophore from the ethynyl-nucleoside part. This stacking of the core part of the DNA architectures is induced by annealing with complementary counter strand DNA1 with strong effects on the fluorescenceless fluorescence intensity and blue-shift. The conformation of the Tpe chromphore is obviously influenced also by the base on the 3 0 -side of the 1D modifcation, since the fluorescence of the DNA hybrid TPE1b-DNA1 is only quenched, but not blue-shifted. TPE1a and TPE1b were subsequently annealed with DNA2 bearing the Atto565 dye or with DNA3 bearing the Atto633 dye as acceptor dyes. The TPE emission of the hybrids TPE1a-DNA2 and TPE1a-DNA3 is not quenched at all and there is no observable energy transfer. Obviously, the distance of 34 is too long for energy transfer. In comparison, the hybrids TPE1b-DNA2 and TPE1b-DNA3 show 48% and 46% quenching of the TPE emission and a detectable energy transfer to the Atto dyes over a distance of 17 . In comparison, in the non-covalent DNA architectures 1D 20 -D-A 20 -Atto565 and 1D 20 -D-A 20 -Atto, the pronounced fluorescence quenching involves an energy transfer distance of up to 58 , as discussed above. The typical sigmoidal Förster dependence of the energy transfer efficiency on the distance with a characteristic Förster radius cannot be applied to DNA architectures with chromophores that are attached with short and rigid linkers. Taken together, it implies that a multistep homoenergy transport between the assembled Tpe chromophores precedes the fnal energy transfer to the Atto dyes as acceptors in the supramolecular structures. Such homo-energy transport cannot be promoted by the unmodifed A-T pairs, separating the Tpe from the Atto dyes in the hybrids TPE1a-DNA2 and TPE1a-DNA3. This result underscores the signifcant role of the DNA templates that control not only the direction of the energy transport but also signifcantly improve the energy transport properties by their building blocks within the supramolecular DNA architectures, such that the light is harvested much better. The strands TPE2a and TPE2b were modifed twice with the nucleotide 1D. The hybridization with the counter strand DNA1 affects the fluorescence in the same manner like in case of TPE1b. TPE2a or TPE2b was annealed with DNA2 modifed with the Atto565 dye or with DNA3 modifed with the Atto633 dye. In both cases, there was no energy transfer observable between 1D and the Atto dyes in these double-stranded DNA hybrids. ## Conclusions The two new conjugates 1D and 1L with 2 0 -deoxyuridine as the recognition unit for 2 0 -deoxyadenosine in single-stranded DNA templates and with Tpe as a fluorescent chromophore can be applied to prepare a new type of supramolecular DNA architecture by self-assembly. Most importantly, all assemblies, including the non-templated and the DNA-templated ones, show signifcant AIE. The quantum yield of the assembly of 1D along A 20 is 0.40; approximately 55% of the available binding sites on the DNA template are occupied by 1D. The two conjugates 1D and 1L differ by the confgurations of 2 0 -deoxyribofuranosides to probe the chirality of both their non-templated assembly and their templated assembly along D-and L-confgured DNA templates by means of optical spectroscopy. The chirality of the non-templated assemblies of 1D and 1L is exclusively controlled by the confguration of their sugar parts. In contrast, the chirality of the templated assemblies is governed by the confguration of the sugar part of the DNA templates, and there is no confguration-selective binding of 1D to D-A 20 and 1L to L-A 20 . These supramolecular DNA architectures are hierarchically ordered and the DNA template controls not only the chirality, but also the binding selectivity and the energy transport properties. Moreover, the strong AIE of these DNA architectures with 1D and 1L can be efficiently quenched and the excitation energy can be transported to Atto dyes that were attached to the 5 0 -termini of the DNA templates. The building blocks of the self-assembled DNA architectures participate themselves in the energy transport along the DNA template; a multistep energy "hopping" transport precedes the fnal energy transfer to the terminal Atto dyes. This was elucidated by the reference DNA double strands in which we incorporated 1D covalently at two distinct sites in the sequences, one near the Atto dye (4 base pairs, ca. 17 ), and one farther away (9 base pairs, 34 ). Only the frst DNA construct showed energy transfer properties. The high intensity of the AIE of this new type of supramolecular DNA architecture and the excellent energy transport properties efficiently harvest light and make these DNA-based materials promising candidates for optoelectronic applications.

chemsum

{"title": "DNA-templated control of chirality and efficient energy transport in supramolecular DNA architectures with aggregation-induced emission", "journal": "Royal Society of Chemistry (RSC)"}

tuning_activity_and_selectivity_during_alkyne_activation_by_gold(i)/platinum(0)_frustrated_lewis_pai

## Abstract: Introducing transition metals into frustrated Lewis pair systems has attracted considerable attention in recent years. Here we report a selection of three metal-only frustrated systems based on Au(I)/Pt(0) combinations and their reactivity towards alkynes. We have inspected the activation of the simplest alkyne, namely acetylene, as well as of other internal and terminal triply bonded hydrocarbons. The gold(I) fragments are stabilized by three bulky phosphines bearing terphenyl groups. We have observed that subtle modifications on the substituents of these ligands proved critical to control the regioselectivity of acetylene activation and the product distribution resulting from C(sp)-H cleavage of phenylacetylene. A mechanistic picture based on experimental observations and computational analysis is provided. As a result of the cooperative action of the two metals of the frustrated pairs, several uncommon heterobimetallic structures have been fully characterized. ## Introduction Frustrated Lewis Pairs (FLPs) emerged more than a decade ago as a paradigmatic example of chemical cooperativity, permitting bond activation and catalysis in the absence of transition metals. 1 However, the incorporation of the latter as core components of FLP systems has also attracted considerable attention in the last years. 2 , 3 Introducing transition metals into frustrated designs largely increases the amount and structural diversity of Lewis acid/base combinations available. Besides, it provides an array of elementary reactions accessible for transition metals that is foreseen to extend the catalytic usefulness of main group FLPs beyond their current status. However, fundamental knowledge on transition metal FLPs regarding mechanistic aspects, solution dynamics, acid-base interactions or selectivity effects are rather underexplored compared to main group FLPs, despite the fact that this information is vital for expeditious catalyst development. To this regard, the group of Wass soon demonstrated that subtle ligand modifications have a strong impact on the ability of zirconocene-based FLPs towards dihydrogen splitting. 4 This type of systems was also examined to clarify the nature of Lewis acidbased interactions by DOSY NMR spectroscopy. 5 Going beyond monometallic systems, our group focused on gaining fundamental knowledge on FLPs in which the two Lewis components are based on transition metals. Thus, we recently reported the first example of its kind by combining Au(I) and Pt(0) species as the acidic and basic sites, respectively. 6 To achieve frustration we targeted sterically hindered phosphine ligands for both gold and platinum monometallic complexes. Our experimental/computational investigations regarding the heterolytic splitting of dihydrogen mediated by these pairs led us to propose a genuine bimetallic FLP-type pathway 7 analogous to the models assumed for main group counterparts. 8 Moreover, we could analyze the strong influence that Au•••Pt interactions have on the activation capacity of the bimetallic pairs, as well as the solution dynamic equilibria between the metal-only Lewis pairs and the individual monometallic fragments. This is a particularly important aspect in the field of FLPs that has been widely investigated for metal-free systems, where the term 'thermally induced FLPs' 9 was coined to refer those pairs in which the Lewis adduct is the resting state. Despite this fact, many of these pairs exhibit a rich FLP reactivity 10 and in some cases superior catalytic performance to fully frustrated counterparts. 11 In this study, we extend our fundamental knowledge on transition metal-only FLPs (TMOFLPs) by exploring regioselectivity effects derived from ligand modification during the activation of alkynes, also model substrates widely investigated in the field of frustrated systems. 1 , 12 We have focused on the effects derived from varying the degree of frustration, for which we have used terphenyl phosphine ligands PMe2Ar Xyl2 (a), PMe2Ar Dipp2 (b) and PCyp2Ar Xyl2 (c) (Cyp = cyclopentyl), of different steric profiles, to stabilize electrophilic gold fragments 1 (Figure 1). These acidic complexes combined with the basic Pt(0) compound [Pt(P t Bu3)2] (2) promote the cooperative activation of terminal alkynes. Our studies demonstrate the potential of using transition metal Lewis acids (i.e. [PR3Au(I)] + fragments) to control the selectivity in the activation of small molecules by tuning the steric properties of the ancillary ligands. This is particularly appealing in view of the challenging available protocols to synthesize the typically used acidic boranes and their scant stability towards moisture. ## Results and Discussion To investigate regioselectivity effects during alkyne activation we first examined the reactivity of the three Au/Pt bimetallic pairs depicted in Figure 1 towards acetylene, a reaction that we had previously reported with the 1b:2 combination. 6a When a dichloromethane or benzene solution of the latter pair is exposed to acetylene (0.5 bar, 25 °C) a rapid color change from bright yellow to intense orange takes place. Multinuclear spectroscopic analysis revealed the formation of a clean mixture of two structurally different isomers, namely a bridging σ,π-acetylide (3b) and a rather unusual heterobimetallic vinylene (-CH=CH-) (4b), which are produced in a 4:1 ratio (Scheme 1). These metallic species are highly reminiscent of the organic products derived from the reactivity of traditional phosphine/borane FLPs with alkynes, where the prevalence of one or the other isomer typically depends on the basicity of the phosphine. 14 We now tested the analogous reactivity using gold precursors [(PMe2Ar Xyl2 )Au(NTf2)] 1a and [(PCyp2Ar Xyl2 )Au(NTf2)] 1c in our search for regioselectivity effects while keeping unaltered the basicity of the metallic base (2). Moreover, the acidity of the gold precursors 1a-c barely differs from one another, 15 thus any anticipated outcomes mostly build on steric grounds. In fact, we found a drastic change in product distribution from the less hindered system (1a, PMe2Ar Xyl2 ) to the more congested one (1c, PCyp2Ar Xyl2 ), as determined by NMR spectroscopy. While the former yields around 95% of the bridging Au/Pt acetylide 3a and only a residual amount of the vinylene (4a, <5%), the more hindered pair comprising the [(PCyp2Ar Xyl2 )Au] + fragment (1c) fully reversed the selectivity towards the exclusive formation of the corresponding vinylene 4c (Scheme 1). Attempts to isolate 3c by the reaction of independently prepared compounds [(PCyp2Ar Xyl2 )Au(C≡CH)] (5c) and [Pt(P t Bu3)2H][NTf2] 16 (6) proved unsuccessful and resulted in intractable mixtures. ## Scheme 1. Regioselectivity in the activation of acetylene by TMOFLPs 1:2. As aforesaid, this dramatic shift in regioselectivity seems to be dominated by steric effects, which contrasts with prior strategies to modulate alkyne activation by FLPs that mostly rely on phosphine basicity. More importantly, it evinces the potential of FLP systems that incorporate transition metal Lewis acids to easily tune the selectivity during bond activation processes and, as such, in subsequent catalytic applications that incorporate those activation events. As pointed out earlier, this could be seen as a key advantage compared to traditional FLP designs that usually involve fluorinated boranes, since accessing these moieties already entails substantial synthetic challenges and limitations, not to mention their limited stability towards moisture and air. 13 In stark contrast, the preparation of terphenyl phosphines PR2Ar' is straightforward and highly versatile, 17 while the resulting gold precursors 1 are readily obtained in high yields and exhibit stability towards water or under moderate oxidizing conditions. 18 The nature of the new heterobimetallic compounds 3a and 4c was ascertained by comparison of their 1 H and 31 P{ 1 H} NMR signals with those derived from their analogous species based on PMe2Ar Dipp2 , that is, 3b and 4b, respectively. 6 a The heterobimetallic nature of compounds 4 is evinced by the 195 Pt satellites that flank the 31 P{ 1 H} resonances associated to terphenyl phosphines, which appear at 2.1 (4b, 4 JPPt = 282 Hz) and 51.7 (4c, 4 JPPt = 277 Hz) ppm. The bridging vinylene (-CH=CH-) moiety displays a distinctive pair of 1 H NMR signals in the region between 4.0 and 4.5 ppm that reveal scalar coupling to the 195 Pt center in the range 120 -200 Hz (see Experimental Section for details). By analogy, we attribute a 31 P{ 1 H} NMR resonance at 3.67 ppm to the minor species (ca. 5%) in the PMe2Ar Xyl2 system (4a), with an identical 1 H NMR pattern comprised of signals at 4.54 and 4.37 ppm, though their corresponding 195 Pt satellites could not be observed due to the low concentration of isomer 4a. Corresponding 13 C NMR resonances for compounds 4b and 4c emerge at ca. 155 ( 1 JCH ≈ 175 Hz) and 115 ( 1 JCH ≈ 190 Hz) ppm, supporting the proposed formulation and the sp 2 hybridization of the carbon atoms. The molecular structure of compounds 3a and 4c was further corroborated by X-ray diffraction studies (Figure 2). The presence of the σ,π-acetylide or vinylene linker distorts the linearity around the platinum center, having P-Pt-P angles of around 165° that shifted from the ideal 180° due to the steric pressure exerted by the bulky gold fragment. The Pt1-C1 (2.016(6) ) and Au1-C1 (2.311(5) ) bond distances in 3a appear slightly shortened compared to 3b (dPt-C1 = 2.044(7) ; dAu-C1 = 2.360(7) ). The average C=C bond distances of the vinylene linkers in the two crystallographically independent molecules of 4c accounts for 1.278(13) , comparable to 4b (1.287(11) ) and another related species. 19 Figure 2. ORTEP diagrams of compounds 3a and 4c; for the sake of clarity most hydrogen atoms and triflimide anions are excluded and some substituents have been represented in wireframe format, while thermal ellipsoids are set at 50 % probability. We had previously observed that the three investigated terphenyl phosphines permit to control the equilibrium between complete frustration and bimetallic adduct formation. 7 Thus, while a dative PtAu bond is immediately formed between [PMe2Ar Xyl2 )Au(NTf2)] (1a) and [Pt(P t Bu3)2] (2), the formation of an identical adduct based on (PCyp2Ar Xyl2 ) is endergonic and could not be experimentally detected. An intermediate situation is reached for the medium-sized phosphine (PMe2Ar Dipp2 ), where the prevalence of the monometallic fragments or the bimetallic adduct depends upon experimental conditions. In this context, we have observed that TMOFLPs (1:2) based on gold precursors [(PMe2Ar Dipp2 )Au(NTf2)] (1b) and [(PCyp2Ar Xyl2 )Au(NTf2)] (1c) are considerably more active towards alkyne activation than the one built on [(PMe2Ar Xyl2 )Au(NTf2)] (1a). While full conversion towards compounds 3 and 4 was recorded by the time of placing the sample in the NMR probe (< 5min) in the case of using 1b:2 or 1c:2, the analogous transformation essayed with 1a required up to 24 hours to reach completion under otherwise identical conditions (C2H2, 0.5 bar, 25 °C, toluene or C6D6). This fact speaks in favor of a genuine FLP mechanism that imposes an energetic demand to overcome the PtAu bond cleavage prior to acetylene activation, a requirement that only applies to the less hindered gold precursor 1a. In addition, it is key to highlight that the cooperative reactivity depicted in Scheme 1 contrasts with that of the individual Au(I) or Pt(0) fragments (Scheme 2). For instance, [Pt(P t Bu3)2] (2) readily catalyzes acetylene polymerization, evinced by the rapid formation of a purple-black solid accompanied by the disappearance of a 1 H NMR resonance at 1.34 ppm due to C2H2, while signals due to 2 remained unchanged. At variance, no indication of polyacetylene formation is apparent when gold is also present in solution. In the case of the individual gold compounds 1 there is no sign of chemical transformation in the short term (ca. 30 min), while at longer reaction times the gold triflimide precursors evolve to bridging σ,π-acetylide compounds 7 (Scheme 2A), 20 albeit only in moderate yields and accompanied by other unidentified gold-containing species. ## Scheme 2. (A) Reactivity of individual gold (1) and platinum (2) compounds towards acetylene; (B) Reaction of σ,π-acetylide 7b and T-shaped platinum hydride 6. Having the previous experimental findings on hand, we were interested in further understanding the cooperative action of TMOFLPs 1:2 to gain fundamental knowledge of significance for future catalytic applications. As such, we initially wondered about the possible role of acetylide-bridged digold compounds like 7 as precursors towards complexes 3 and 4. In fact, the reaction between 7b and [Pt(P t Bu3)2H][NTf2] (6) immediately yielded the corresponding heterobimetallic σ,π-acetylide compound 3b, where the unsaturated linker is now σ-bonded to the platinum center instead of the gold nucleus. Nevertheless, the complete absence of the vinylene isomer 4b during the latter reaction would require an additional competing route to provide access to this unusual bimetallic motif, which actually is the exclusive isolated isomer for the bulkier PCyp2Ar Xyl2 -based system (Scheme 1). Moreover, formation of compounds 7 requires several hours to proceed to appreciable conversions, while the activation of acetylene by 1:2 pairs is immediate (<5 min), except for the gold precursor 1a bearing PMe2Ar Xyl2 , which as noted earlier takes around 24 hours to convert into 3a in the presence of [Pt(P t Bu3)2] (2). To further investigate the mechanism and the reasons for the drastically different regioselectivity observed we carried out computational studies (DFT, ωB97XD/6-31G(d,p) + SDD). Focusing on the system based on PMe2Ar Dipp2 , we began searching for initial acetylene activation steps by approaching the acetylene molecule to the individual Au(I) (1b) and Pt(0) (2) fragments, assuming that Au-Pt dissociation is a prerequisite for alkyne activation, as deduced from the reduced reactivity of the pair 1a:2 compared to 1b:2 and 1c:2, and previous computational results. 7 Interestingly, formation of 3 and 4 seems to share a common intermediate, namely a Au(I) acetylene adduct of formula [(PR2Ar')Au(C2H2)] + (9). The formation of this type of π-complexes has been previously proposed in the context of alkyne 12 , 14 a and alkene 21 activation by P/B pairs, but no experimental proofs of their existence have been reported. In an attempt to spectroscopically identify such an intermediate, we recorded the formation of a new gold-containing species by low temperature NMR (-80 °C) as the major species (ca. 80%) upon exposing a CD2Cl2 solution of 1b to acetylene atmosphere. This species exhibits a distinctive 1 H NMR signal at 3.43 ppm that correlates with a 31 P NMR resonance at -0.3 ppm (compared to δP = -9.9 ppm for 1b) and presents dynamic exchange with free acetylene (δH = 2.11 ppm) as seen by EXSY NMR experiments, while other 1 H NMR resonances are comparable to those of 1b. This finding constitutes an additional benefit of TMFLPs for mechanistic investigations in frustrated systems, since they provide additional modes of stabilizing otherwise fleeting intermediates. Starting from acetylene adduct 9b, our calculations indicate that the attack of the platinum compound 2 over 9b leads to either of the two bimetallic isomers 3b and 4b depending on the trajectory followed by 2 while approaching 9b (Figure 3). Thus, if 2 approaches the acetylene adduct along the Au-C2H2 direction the corresponding vinylene 4b forms (ΔG ‡ = 23.4 kcal•mol -1 from the 1b:2 Lewis pair + acetylene), in a process somehow reminiscent of the gold-mediated nucleophilic attack over activated alkynes (e. g. gold-catalyzed hydroamination). 22 In contrast, the alignment of the basic 2 in an orthogonal disposition with respect to the Au-C2H2 bond results in deprotonation of the activated acetylene (15.6 kcal•mol -1 for the pair 1b:2) to yield the corresponding Au(I) terminal acetylide [(PMe2Ar Dipp2 )Au(C≡CH)] (5b) and [Pt(P t Bu3)2H][NTf2] (6). These two fragments readily rearrange to intermediates [(PR2Ar')Au(μ-η 1 :η 2 -C≡CH)Pt(H)(P t Bu3)2] + (A) that subsequently evolve to compounds 3 by rapid σ,πisomerization of the bridging μ-C≡CR unit. In experimental agreement, reaction of independently synthesized [(PMe2Ar Dipp2 )Au(C≡CH)] (5b) and [Pt(P t Bu3)2H][NTf2] (6) rapidly yields complex 3b. As mentioned above, this does not apply to compound 5c, whose reaction with 6 resulted in a complex mixture of products that include decomposition into black gold. Nevertheless, the corresponding [(PCyp2Ar Xyl2 )Au(μη 2 :η 1 -C≡CH)Pt(H)(P t Bu3)2] + (3c) has not been detected during the bimetallic activation of acetylene. An alternative orthogonal mechanism involving the initial oxidative addition of acetylene over 2, followed by cis-trans isomerization and coupling with 1 was ruled out on the basis of significantly higher overall activation barriers (see Figure S2). Information on the activation of the simplest alkyne (C2H2) by FLPs is rather scarce. 12 a For the sake of completeness and to better compare our results with prior studies on main group FLPs, we decided to test other more commonly employed triply bonded hydrocarbons. Regarding internal alkynes, all our attempts to access bimetallic vinylenes were unsuccessful. Reaction of 1:2 pairs with diphenylacetylene, 2-butyne and 1,4-Diphenylbutadiyne did not result in the formation of any new species even under more forcing experimental conditions. At variance, addition of phenylacetylene to equimolar mixtures of 1 and 2 provided phosphine-dependent divergent outcomes derived from C(sp)-H bond cleavage (Scheme 3). Paralleling acetylene activation with the more congested pairs based on 1b and 1c, reaction of these systems with PhC≡CH was also immediate. On the contrary, while acetylene activation took up to 24 hours for the non-frustrated 1a:2 pair, the reaction was complete after around 15 min in the case of phenylacetylene. The appearance of a distinctive low-frequency 1 H NMR resonance at around -10 ppm flanked by 195 Pt satellites ( 1 JHPt = 608 Hz, 1a:2; 1 JHPt = 533 Hz, 1b:2 and 1c:2) prompted us to believe that corresponding heterobimetallic σ,π-acetylide complexes were formed in all cases. However, a more careful analysis revealed unexpected differences in product distribution for the less bulky terphenyl phosphine (PMe2Ar Xyl2 ) compared to the more hindered systems (Scheme 3). Extracting the reaction crudes with pentane permitted isolation of the same platinum containing compound 10 for the pairs 1b:2 and 1c:2, while the less hindered gold fragment 1a did not led to any metallic species soluble in non-polar hydrocarbon solvents. An infrared band at 2090 cm -1 was recorded for a Pt-hydride ligand in 10, while characteristic 13 C NMR resonances at 118.8 and 117.6 ppm accounted for a σ-bonded acetylide ligand. Based on these spectroscopic features and its high solubility we proposed a molecular formulation for 10 as [Pt(P t Bu3)2(H)(C≡CPh)], that is, the formal oxidative addition of phenylacetylene over Pt(0) compound 2. This assumption was further corroborated by X-ray diffraction analysis (see Figure S3). Nevertheless, it is important to remark that compound 2 does not react with phenylacetylene even after longer reaction times (48 h) or at elevated temperatures (80 ºC), which suggest a cooperative action between platinum and gold to account for alkyne C-H activation. The only detectable gold-containing species in these reactions were assigned to the corresponding bridging σ,π-acetylide digold complexes (11). Those where characterized by 31 P{ 1 H} NMR signals at 0.4 (11b) and 53.9 (11c), shifted at higher frequencies with respect to their precursors 1 (c.f. -11.9, 1b; 48.8 ppm, 1c), in agreement to other related examples. 23 Also similar to those, the presence of a single 31 P resonance for each compound suggests rapid exchange of the σ,π-coordination in solution. 18 The process is frozen though in the solid-state. An ORTEP representation of the molecular structure of 11b is shown in Figure 4a. The σ,π-coordination of the acetylide is reflected by a nonsymmetric arrangement characterized by bond distances of 2.021(10) and 2.209(9) for the Au2-C1 and Au1-C1 bond distances, respectively. The Au1 center is also connected to C2 by a slightly longer bond distance of 2.310(9) . The presence of an aurophilic interaction can be inferred from a Au1-Au2 distance of 3.366(1) , which is faintly elongated compared to its related acetylide analogue ([Au2{µ-C≡CH}], dAu1-Au2 = 3.31 ), likely as a result of the higher steric pressure exerted by phenylacetylide. Digold complexes 11b and 11c were accompanied by equimolar amounts of [Pt(P t Bu3)2(H)] + (6) that could not be washed out with pentane, as evinced by a broad low-frequency 1 H NMR resonance at 1.16 ppm ( 1 JPPt = 6.5Hz). As noted earlier, the reaction of the less hindered 1a:2 pair with phenylacetylene yielded a divergent result. C(sp)-H activation became evident by the presence of a distinctive hydridic 1 H NMR resonance at -10.4 ppm ( 2 JHP = 14 Hz, 1 JHPt = 608 Hz). However, a single platinum complex was formed in this case, which resonates at 82.2 ppm ( 1 JPPt = 2810 Hz) in its 31 P{ 1 H} NMR spectrum and could not be washed out using non-polar hydrocarbon solvents. The corresponding C≡C stretching frequency rendered a band shifted to lower wavenumbers (νC≡C = 1982 cm -1 ; c.f. 2048, 11b; 2029, 11c; 2090 cm -1 , 10), while sp-hybridized carbon atoms resonate at lower frequencies (91.1 and 85.9 ppm) compared to compounds 11b, 11c and 10 (116 -125 ppm). These observations expose the divergent product distribution derived from using phosphines with different steric profiles. Since the recorded parameters equate with heterobimetallic σ,π-acetylide compounds 3, we assumed an analogous structure for this complex (12a), a premise that we could substantiate by X-ray diffraction analysis (Figure 4b). The bulkier nature of the phenylacetylide moiety with respect to the unsubstituted acetylide in 3a is likely the cause of a more intense distortion of the T-shape platinum fragment, where a P-Pt-P angle of 156.71(8)º is recorded 162.76(5)º for 3a). The close proximity between the acetylide phenyl fragment and the ortho substituents of one of the flanking aryl rings of the terphenyl moiety may be responsible for the dissimilar product distribution found between PMe2Ar Xyl2 vs PMe2Ar Dipp2 /PCyp2Ar Xyl2 . Based on our experimental and computational studies on acetylene activation, initial formation of a gold-alkyne adduct alike 9 seems most plausible. The slightly higher acidity of phenylacetylene, as well as its higher size, may account for the prevalence of the deprotonation pathway in detriment of the 1,2-addition route towards vinylene structures, which were not detected. In turn, the dissimilar reaction products depicted in Scheme 3 might be understood according to steric grounds. Thus, selective formation of compounds 11 for the bulkier PMe2Ar Dipp2 /PCyp2Ar Xyl2 -based systems could be the result of a higher steric clash between the tert-butyl substituents on the platinum fragment and the terphenyl moiety of the gold-bound phosphine. In contrast, the reduced steric pressure introduced by PMe2Ar Xyl2 may permit easier access to the heterobimetallic compound 12a, analogous to its unsubstituted acetylide version (3a). We thought of interest to carry out several further experiments to shed some light into the operating cooperative mechanism. In the same manner as we observed for acetylene activation, the reactivity of the individual metallic fragments starkly contrasts with that of the bimetallic pairs. Accordingly, platinum compound 2 does not exhibit any reactivity towards phenylacetylene even after heating at 80 ºC for 48 hours (Scheme 4a). In turn, compounds 1 promote C(sp)-H cleavage of the alkyne to form the corresponding digold σ,π-acetylide complexes 11, but at a considerably slower pace (24 h at 25 ºC: 1a, 9% of 11a; 1b, 70% of 11b; 1c, 12% of 11c; NMR spectroscopic yields). To account for the origin of 10 we performed the reaction of 2 with equivalent amounts of phenylacetylene and catalytic quantities of 1. In addition, we also tested the same transformation but using 6 as the catalyst, since this presumably forms in situ after protonation of 2 by small amounts of HNTf2 derived from the reaction of 1 and phenylacetylene towards 11. Catalytic amounts of compound 6 may also derive from the cooperative Au/Pt C(sp)-H activation of phenylacetylene by a deprotonation mechanism. As anticipated, immediate conversion of 2 into 10 was observed at room temperature using either 1 or 6 in catalytic amounts as low as 2 mol% (Scheme 4b). On its part, reaction of the independently synthesized neutral σ-acetylide compound [(PMe2Ar Dipp2 )Au(C≡CPh)] (13b; ORTEP diagram in Figure S4) with 1b yielded the expected digold σ,π-acetylide 11b under mild conditions (Scheme 4c). This result parallels the reactivity previously described for NHC-based gold complexes, where the latter transfor-mation proceeds smoothly, 23f as also occurs with the parent unsubstituted acetylide (C≡CH) fragment. 18 Thus terminal acetylides of type 13 may be regarded as key intermediates towards compounds 11 during the activation of PhC≡CH. Considering all the information discussed above, Scheme 5 contains an overall mechanistic picture to account for the phosphine-dependent product distribution during the activation of alkynes. A common gold acetylene adduct 9 is proposed to be a key intermediate. While both deprotonation and 1,2-addition mechanisms (blue and red in Scheme 5, respectively) are viable for acetylene, only the deprotonation pathway seems to be operative in the case of the more acidic phenyl-substituted alkyne. Once the latter is deprotonated to form an equimolar mixture of 13 and 6, steric factors appear to dominate the final product distribution. The combination of the more hindered terphenyl phosphines with the bulkier phenylacetylene pre-vents formation of the corresponding Au/Pt heterobimetallic adducts likely due to steric clash, while the latter is the only observed complex in the PMe2Ar Xyl2 system. In turn, terminal gold acetylide 13, although unable to react with [Pt(P t Bu3)2H] + (6), rapidly yield the corresponding digold σ,π-acetylides 11 upon combination with still unreacted triflimides 1. We can also infer from our experimental observations that several of the transformations depicted in Scheme 5 constitute dynamic equilibria that are dependent on reaction conditions. Scheme 5. Overall representation of the proposed pathways to account for product distribution during alkyne activation. ## Conclusions In summary, we provide evidence for the potential of frustrated Lewis pairs entirely based on transition metals, more precisely on Au(I)/Pt(0) combinations, for the cooperative activation of small molecules. We have demonstrated that subtle modifications of the phosphine ligands bound to gold have a strong effect on the regioselectivity of acetylene activation. Thus, simply by slightly adjusting the steric hindrance of those phosphines, we have been able to select the operating mechanism (deprotonation vs 1,2-addition) during acetylene activation with full specificity. This is possible without the need of altering the basicity of the platinum(0) moiety, thus contrasting with main group FLP systems. Moreover, fine ligand tuning also permitted us to diverge the product distribution that results from C-H bond cleavage in phenylacetylene. While the less congested system based on PMe2Ar Xyl2 yield heterobimetallic σ,π-acetylide compound, the more hindered pairs constructed around PMe2Ar Dipp2 and PCyp2Ar Xyl2 resulted in the formation of equimolar mixtures of digold σ,π-acetylde compounds, [Pt(P t Bu3)2(H)] + and [Pt(P t Bu3)2(H)(C≡CPh)]. These results highlight one of the advantages of incorporating transition metals into frustrated designs, namely the ability to easily tune the stereoelectronic properties of the acidic site, in turn a challenging task in traditional FLPs that rely on the electrophilicity of fluorinated boranes. The straightforward control of these properties in TMFLPs may be exploited for the development of more selective catalytic transformations drawn on the concept of frustration. ## Experimental Section General considerations. All preparations and manipulations were carried out using standard Schlenk and glove-box techniques, under an atmosphere of argon and of high purity nitrogen, respectively. All solvents were dried, stored over 4 molecular sieves, and degassed prior to use. Toluene (C7H8) and n-pentane (C5H12) were distilled under nitrogen over sodium. Tetrahydrofuran (THF) and diethyl ether were distilled under nitrogen over sodium/benzophenone. [D6]Benzene was dried over molecular sieves (4 ) and CD2Cl2 over CaH2 and distilled under argon. [AuCl(THT)] (THT = tetrahydrothiophene), 24 compounds 1a, 7 1b, 18 1c 7 , 2 25 and 6 16 were prepared as described previously. Other chemicals were commercially available and used as received. Solution NMR spectra were recorded on Bruker AMX-300, DRX-400 and DRX-500 spectrometers. Spectra were referenced to external SiMe4 (δ: 0 ppm) using the residual proton solvent peaks as internal standards ( 1 H NMR experiments), or the characteristic resonances of the solvent nuclei ( 13 C NMR experiments), while 31 P was referenced to H3PO4. Spectral assignments were made by routine one-and two-dimensional NMR experiments where appropriate (Figure 5). Infrared spectra were recorded on a Bruker Vector 22 spectrometer and sampling preparation was made in Nujol. For elemental analyses a LECO TruSpec CHN elementary analyzer, was utilized. 1965525, 1965526 and 1986501-1986504 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre. Figure 5. General labeling scheme used for 1 H and 13 C{ 1 H} NMR assignments. [(PMe2Ar Xyl2 )Au(μ-η 2 :η 1 -C≡CH)Pt(H)(P t Bu3)2]NTf2 (3a). A solid mixture of 1a (100 mg, 0.121 mmol) and 2 (73 mg, 0.121 mmol) was dissolved in 5 mL of toluene and stirred at room temperature for 12 hours under C2H2 atmosphere (0.5 bar). The solution was layered with pentane (10 mL) and stored at -30 °C overnight to yield compound 3a as colorless crystals (115 mg, 66%). Anal. Calcd. for C52H83AuF6NO4P3PtS2: C, 43.1; H, 5.8; N, 1.0; S, 4.4. Found: C, 43.1; H, 5.5; N, 1.1; S, 4.4. 1 H NMR (400 MHz, C6D6, 25 °C) δ: 7.26 (t, 2 H, 3 JHH = 7.6 Hz, Hb), 7.18 (m, 1 H, Hd), 7.15 (d, 4 H, 3 JHH = 7.6 Hz, Ha), 6.71 (dd, 2 H, 3 JHH = 7.5 Hz, 4 JHP = 3.5 Hz, Hc), 2.98 (m, 1 H, C≡CH), 2.15 (s, 12 H, MeXyl), 1.52 (d, 6 H, 2 JHP = 9.5 Hz, PMe2), 1.49 (vt, 54 H, 3 JHP = 6.4 Hz, t Bu), -10.27 (m, 1H, 2 JHP = 14 Hz, 1 JHPt = 571 Hz, Pt-H). 13 Hz, PMe2). 31 P{ 1 H} NMR (160 MHz, C6D6, 25 °C) δ: 82.9 ( 1 JPPt = 2768 Hz, P( t Bu)3), 0.8 (Au-P). IR (Nujol): ν(≡C-H) 3171, ν(C≡C) 1843 cm -1 . [(PCyp2Ar Xyl2 )Au(μ-η 1 :η 1 -HC=CH)Pt(P t Bu3)2]NTf2 (4c). A mixture of compounds 1c (100 mg, 0.107 mmol) and 2 (64 mg, 0.107 mmol) was dissolved in toluene (5 mL) and the argon atmosphere was replaced by C2H2 (0.5 bar), upon which the bright yellow solution changed to an intense orange color. The solution was then filtered, layered with pentane and stored at -20 °C overnight to yield compound 4c as orange crystals (68 mg, 41%). Anal. Calcd. for C60H95AuF6NO4P3PtS2: C, 46.3; H, 6.1; N, 1.0; S, 4.1. Found: C, 45.9; H, 5.7; N, 1.0; S, 3.8. 1 H NMR (400 MHz, C6D6, 25 °C) δ: 7.19-7.08 (m, 3 H, Hb, Hd), 7.01 (d, 4 H, 3 JHH = 7.6 Hz, Ha), 6.66 (dd, 2 H, 3 JHH = 7.7 Hz, 4 JHP = 2.7 Hz, Hc), 4.51 (dt, 1 H, 3 JHH = 7.6 Hz, 3 JHP = 3.4 Hz, 2 JHPt = 110 Hz, Hβ), 4.20 (dd, 1 H, 3 JHH = 7.6 Hz, 3 JHP = 6.4 Hz, 3 JHPt = 194 Hz, Hα), 2.22-2.09 (m, 2 H, CH), 1.97 (s, 24 H, MeXyl), 1.71-1.47 (m, 12 H, CH2), 1.36 (vt, 54 H, 3 JHP = 6.5 Hz, t Bu), 1.24-1.09 (m, 4H, CH2). 13 [(PCyp2Ar Xyl2 )Au(C≡CH)] (5c). A suspension of (PCyp2Ar Xyl2 )AuCl 7 (200 mg, 0.29 mmol) in toluene (10 mL) was cooled to -40 ºC and a toluene solution containing a small excess (1.2 eq.) of Mg(C≡CH)Br was added dropwise. The mixture was stirred for additional 2 hours at -40 ºC. The volatiles were removed in vacuo and the residue extracted with pentane. Evaporation of the solvent led to compound 5c as a white powder (35 mg, 18 %). Anal. Calcd. for C34H40AuP: C, 60.4; H, 6.0. Found: C, 60.4; H, 5.8. 1 H NMR (400 MHz, C6D6, 25 ºC) δ: 7.29 (t, 2 H, 3 JHH = 7.6 Hz, Hb), 7.09 (d, 4 H, 3 JHH = 7.56 Hz, Ha), 6.94 (td, 1 H, 3 JHH = 7.6 Hz, 5 JHP = 1.6 Hz, Hd), 6.64 (dd, 2 H, 3 JHH = 7.6 Hz, 4 JHP = 2.7 Hz, Hc), 2.17-2.05 (m, 2 H, PCH), 1.97 (s, 12 H, MeXyl), 1.78 (d, 1H, 4 JHP = 5.5 Hz, AuC≡CH), 1.7-1.1 (m, 16 H, CH2). 13 [Pt(P t Bu3)2(H)(C≡CPh)] (10). A toluene (5 mL) solution of 2 (64 mg, 0.107 mmol), phenylacetylene (11 µL, 0.107 mmol) and 1b or 1c (2 mg, 0.002 mmol) was stirred at room temperature for 15 minutes. The volatiles were removed in vacuo and the residue extracted with pentane (3x5 mL). Evaporation of the solvent yield compound 10 as colorless oil (26 mg, 35%). Suitable crystals for X-ray diffraction studies can be obtained by slow pentane evaporation at room temperature. Anal. Calcd. for C32H60P2Pt: C, 54.8; H, 8.6. Found: C, 54.9; H, 8.4. 1 H NMR (400 MHz, C6D6, 25 ºC) : 7.64 (d, 2 H, 3 JHH = 7.3 Hz, o-C6H5), 7.21 (t, 2 H, 3 JHH = 8.2 Hz, m-C6H5), 7.02 (t, 1 H, 3 JHH = 7.3 Hz, p-C6H5), 1.58 (vt, 54 H, 3 JHP = 6.3 Hz, t Bu), -9.46 (t, 1 H, 2 JHP = 15.4, 1 JHPt = 532.9 Hz, Pt-H). 13 [(PR2Ar)2Au2(μ-η 1 :η 2 -C≡CPh)]NTf2 (11). To a solution of compounds 1 (100 mg, 0.107 mmol) and 2 (64 mg, 0.107 mmol) in toluene (5 mL) was added one equivalent of phenylacetylene (11 µL, 0.107 mmol) and the mixture stirred at room temperature for 15 minutes. The volatiles were removed in vacuo and the residue washed with pentane (3x5 mL). The resulting fine white powder (60 mg, 11b; 65 mg,11c) contain a mixture of compounds 11 and 6, the latter which could not be separated by common methods but whose spectroscopic features did not hamper full characterization of compounds 11. Single crystal of 11b suitable for X-ray diffraction studies were grown by slow diffusion of pentane into toluene solution (2:1 by vol.) at -30 ºC. Alternatively, compound 11b could be synthesized free of 6: to a solution of 1b (100 mg, 0.107 mmol) in toluene (5 mL) was added one equivalent of phenylacetylene (11 µL, 0.107 mmol) and the mixture stirred at room temperature for 18 hours. The volatiles were removed in vacuo and the residue was washed with pentane (3x5 mL) to yield 11b as a white powder (90 mg, 50 %). Compound 11b. 1 31 P{ 1 H} NMR (160 MHz, C6D6, 25 ºC) δ: 53.9. IR (Nujol): ν(C≡C) 2029 cm -1 [(PMe2Ar Xyl2 )Au(μ-η 2 :η 1 -C≡CPh)Pt(H)(P t Bu3)2]NTf2 (12a). To a solution of 1a (100 mg, 0.121 mmol) and 2 (73 mg, 0.121 mmol) in toluene (5 mL) was added one equivalent of phenylacetylene (13 µL, 0.121 mmol) and the mixture stirred at room temperature for 15 minutes. The volatiles were removed in vacuo and the residue washed with pentane (3x5 mL) to yield compound 12a as a white powder (70 mg, 38 %). Anal. Calcd. for C58H87AuF6NO4P3PtS2: C, 45.7; H, 5.8; N, 0.9; S, 4.2. Found: C, 45.9; H, 5.6; N, 1.1; S, 4. [(PMe2Ar Dipp2 )Au(C≡CPh)] (13b). Following a previously reported method, 26 to a solution of phenylacetylene (36 µL, 0.325 mmol) and KOH (18 mg, 0.325 mmol) in 15 mL of methanol was added a suspension of (PMe2Ar Dipp2 )AuCl (150 mg, 0.216 mmol). The solution was stirred for 20 hours at room temperature, and the solid filtered and washed with diethyl ether (2 x 5 ml) to yield compound 13b as a white solid (139 mg, 85%). Anal. Calcd. for C40H48AuP: C, 63.5; H, 6.4. Found: C, 63.2; H, 6.7. 1 H NMR (400 MHz, C6D6, 25 ºC) δ, ppm: 7.74 (d, 2 H, 3 JHH = 7.3 Hz, o-C6H5), 7.36 (t, 2 H, 3 JHH = 7.8 Hz, m-C6H5), 7.17 (d, 4 H, 3 JHH = 7.8 Hz, Ha), 7.06 (t, 1 H, 3 JHH = 8.3 Hz, p-C6H5), 6.99 (m, 5 H, Hb, Hc, Hd), 2.64 (sept, 4 H, 3 JHH = 6.5 Hz, i Pr(CH)), 1.34 (d, 12 H, 3 JHH = 6.5 Hz, i Pr(CH3)), 0.92 (d, 12 H, 3 JHH = 6.5 Hz, i Pr(CH3)), 0.90 (d, 6 H, 2 JHP = 10 Hz, PMe2). 13 Computational details. Geometry optimization of minima and transition states was carried out with the Gaussian software package. 27 Optimizations were carried out without symmetry restrictions using the ωB97xD functional 28 that includes empirical dispersion corrections. 29 The 6-31g(d,p) basis set 30 was used for non-metal atoms, Au and Pt atoms were described with the SDD basis and associated electron core potential (ECP). 31 Bulk solvent effects (dichloromethane) were included during optimization with the SMD continuum model. 32 Vibrational analysis was carried out on the stationary points to characterize them as minima or transition states as well as to calculate the zero-point corrections, and thermal corrections to enthalpy and free energy. Free energies were corrected (ΔGqh) to count for errors associated with the harmonic oscillator approximation. Thus, according to Truhlars's quasi harmonic approximation, all vibrational frequencies below 100 cm -1 were set to this value so that the entropy contribution is not overestimated. 33 These anharmonic corrections were calculated with the Goodvibes code. 34 The CYLview visualization software has been used to create some of the figures. 35

chemsum

{"title": "Tuning Activity and Selectivity during Alkyne Activation by Gold(I)/Platinum(0) Frustrated Lewis Pairs", "journal": "ChemRxiv"}

photodeposited_polyamorphous_cuo_x_hole-transport_layers_in_organic_photovoltaics

## Abstract: Hole-selective charge transport layers are an important part of modern thin-film electronics, serving to direct electron flow and prevent leakage current. Crystalline metal-oxide hole-transport layers (HTLs) such as NiO and CuO x exhibit high performance and stability. However they are traditionally not amenable to scalable and sustainable solution-processing techniques. Conversely, amorphous metal oxides are much more readily prepared by lowtemperature solution processing methods but often lack the charge transport properties of crystalline semiconductors. Herein we report the fabrication of amorphous a-CuO x thin films from commercially available starting material using a simple UV-based thin-film deposition method. Subsequent thermal annealing of the a-CuO x induces an amorphousto-amorphous phase transition resulting in p-type semiconducting behavior. The resulting thin films were used as HTLs in organic photovoltaic devices with power conversion efficiencies comparable to those fabricated with PEDOT:PSS. ## Introduction Many types of modern thin-film electronic devices such as organic light-emitting diodes, organic photovoltaic devices (OPVs), and perovskite solar cells require charge-selective interlayers to optimize their function. In these devices hole-transport layers (HTLs) serve to promote the movement of holes, while blocking the back-flow of electrons. Common hole-transport layers include doped organic semiconductors such as PEDOT:PSS or Spiro-OMeTAD, 2,4 and inorganic semiconductors such as MoO x , BiOCl, 8 CuO x , and NiO. 3, This latter class of HTLs, inorganic semiconductors, are sought after for their durability and stability in highperformance devices such as perovskite solar cells. Both NiO and mixed-valence CuO x are wide band gap p-type semiconductors, their energy levels are ideal for the transport of holes from the active layer to the transparent electrode while their high conduction band levels prevent leakage current of electrons. Typically these materials are used in their crystalline form to optimize charge carrier mobility. Traditionally, sputter-coating has been the method of choice for growing optical quality metal-oxide thin films; however the high vacuum and voltage required for sputter coating are energy intensive and not amenable to high throughput; a solution-based deposition method would be the ideal future to enable low-cost roll-to-roll processing. Unfortunately, solution processing methods common for crystalline metal-oxide semiconductors rely on high-temperature annealing steps in the case of sol-gel fabrication, or synthesis of metastable precursors in the case of nanoparticles. An ideal metal-oxide semiconductor HTL would be solution processable using inexpensive precursors, be amenable to large-scale processing techniques such as slot-die coating or spray coating, and require minimal post-deposition processing steps. One such deposition technique is the UV light-assisted decomposition of a precursor coordination complex to produce metal-oxide films. 21,22 In this case, a photosensitive precursor is deposited as a thin film on a substrate and irradiated with short-wave UV light. Upon excitation of the ligand-to-metal charge transfer transition the metal centre is reduced to an M(0) state while the resulting ligand-based radical decomposes into volatile side products that are eliminated from the film. When this process is performed in air the thin metal film is rapidly converted to an amorphous metal (oxy)hydroxide material. For simplicity these layers are generally (and will be in this report) referred to as metal oxides. UV decomposition-derived metal-oxide thin films are completely amorphous, and have been shown to be effective catalysts for the oxygen evolution reaction in aqueous electrocatalysis. 21,23,24 Recently, such amorphous thin films have been used as both the electrode and active material in solid-state electrochromic windows. 25,26 These devices demonstrated the use of photodeposited amorphous NiO x and WO x , paving the way for their use in other optoelectronic devices. In this study we will outline the fabrication of amorphous copper oxide (a-CuO x ) using UVdecomposition of solution-processed commercially available precursor transition metal complexes. The resulting amorphous a-CuO x (by XRD) thin films were structurally characterized using x-ray absorption spectroscopy, x-ray photoelectron spectroscopy (XPS) and atomic force microscopy(AFM); additionally, their electronic properties were studied by UV/visible spectroscopy and UV photoelectron spectroscopy (UPS). It was discovered that annealing at temperatures greater than 200 °C induced an amorphous-to-amorphous phase transition without yielding a crystalline material. Finally, the a-CuO x thin films were incorporated into OPVs as the HTL. Films of the high-temperature polyamorph of a-CuO x (those annealed above 200 °C) performed comparably to control devices fabricated with PEDOT:PSS-based HTLs. ## Photodeposition of a-CuO x thin films To create smooth, continuous films of a-CuO x we use a precursor containing large poorly crystalline aliphatic carboxylic acid ligands with an unresolved chiral centre, copper(II) bis(2-ethylhexanoate) (Cu(2-ethex) 2 ). 27 When deposited via spin-or slot-die coating this complex forms a continuous film less prone to cracking that those of more crystalline precursors such as Cu(II) acetylacetanoate 2 . 22 Cu(2-ethex) 2 is also commercially available, relatively inexpensive, and is dark blue in solution and thin film, which makes it simpler to qualitatively assess printed or cast layers before UV treatment which renders the layer light brown to colourless (see Fig. S1). In this case the precursor was amenable to spin casting -as well as roll-to-roll compatible, industrially relevant, slot-die coating -from a range of solvents; n-butyl acetate was determined to provide optimum layer quality while allowing us to avoid highly toxic halogenated solvents. The UV-induced decomposition of precursor films on IR-transparent CaF 2 substrates can be followed by transmission FTIR by tracking the disappearance of signals corresponding to the 2-ethylhexanoate ligand (Fig. 1). The decay of the major peaks in Fig. 1 at 2959 and 1584 cm −1 fit well to first-order reaction kinetics with half-lives of 8.25 and 4.25 min, respectively. For a thin film deposited from a 15 w/w% solution onto a CaF 2 window, the decomposition process was largely completed after 32 minutes; in this work substrates were left under the lamp for 2 hours, representing 14.5 half-lives for the slowest decay process, to ensure complete conversion of the precursor complex to the corresponding a-CuO x . In order to assess the structural nature of our a-CuO x thin films, XRD patterns were collected on the as-prepared samples, and at annealing temperatures up to 250 °C. Fig. 2a demonstrates that the as-prepared films are completely amorphous with no evidence of diffraction peaks, this amorphous nature is retained up to the maximum annealing temperature of 250 °C that is tolerated by the ITO-coated substrates. 28 To study any short-range order that may be present in the a-CuO x structure we probed the films using X-ray absorption near-edge spectroscopy (XANES) over a series of annealing temperatures from as-prepared (AP) to 250 °C, denoted Cu-AP, Cu-100, Cu-150, Cu-200, and Cu-250 (Fig. 2b). 1 It is immediately clear that thermal annealing has a dramatic effect on the material's structure, despite the lack of diffraction-inducing long-range order. Both the Cu-AP and Cu-100 samples exhibit prominent pre-edge peaks at 8985 eV reminiscent of that seen in the crystalline Cu 2 O standard (Fig. S2a), corresponding to a 1s-4p transition. At higher annealing temperatures (200 and 250 °C) the XANES spectra more strongly resemble CuO (Fig. S2b). In these cases a small pre-edge peak is observed at 8982 eV corresponding to the 1s to 3d transition (see Fig. S2b). Additionally, the rising edge 4p←1s transition is replaced by a less pronounced shoulder slightly higher in energy (8989-8990 eV) consistent with a shake-down transition (a 4p←1s transition coupled to a ligand to metal charge transfer). The spectra of the Cu-200 ## Structural Characterization of a-CuO x and Cu-250 samples are almost identical, suggesting that the structural change occurring with annealing happens at or just below 200 °C. We characterized the a-CuO x surface using XPS and atomic force microscopy on films prepared from a 5 w/w% Cu(2-ethex) 2 solution. The presence of satellite peaks in the Cu 2p region of the XPS spectrum is characteristic of Cu 2+ species (Figs. 2c and S3); peak fitting revealed a mix of Cu(OH) 2 and CuO (Table S1). Fitting of the Cu 2p 3/2 peak (Fig. 2c) indicates that the as-prepared sample's surface consists of 97% Cu(OH) 2 and 3% CuO; upon annealing the relative amount of CuO increased to a maximum of 52% CuO and 48% Cu(OH) 2 for Cu-250 (full fitting parameters are documented in Table S1). This trend is corroborated by the O 1s peak fitting which show the hydorxide (Cu(OH) 2 ) peak decreasing compared to the oxide (CuO) peak (Fig. 2d). The lack of a sharp peak at lower energy in the Cu 2p 3/2 region precludes significant amount of surface Cu + or Cu 0 species. Furthermore the ratio of the 2p 3/2 peak and its satellite peaks remains constant with increasing annealing temperature, indicating that the ratio of Cu 2+ to Cu + or Cu 0 is constant and that the presence of Cu + remains insignificant. 34 The modified Auger parameter was determiend for cu. Due to the lower-resolution data of the survey scan the Cu LMM peak location is harder to determine with great precision, especially in the case of the high-temperature samples which are combination of seven peaks. 35 Despite this, clear signs of the shift from Cu(OH) 2 to CuO are present. The main peak shift toward higher kinetic energy, consistent with increased CuO content, and the appearance of a shoulder at 914 eV is also indicative of greater CuO presence. A small, high-energy peak in the O 1s spectra was attributed to adsorbed water 36 and is present at all annealing temperatures. AFM imaging of the sample deposited onto ITO from 5 w/w% Cu(2-ethex) 2 solutions is presented in Fig. S4. Fig. S4b,c present the surface coated with Cu(2-ethex) 2 , before and after UV irradiation. The formation of pinholes in the film is apparent, these have a depth of around 10 nm, compared to the film surface, surrounded by protrusions that can reach up to 30 nm. This results in an increase in root-mean-squared roughness (R rms ) from 5.8 nm in the films before irradiation to 7.9 nm for Cu-AP. Increasing the annealing temperature gradually reduces the prominence of these pinholes and the roughness of the film decreases to 6.2 nm for Cu-250. In all cases, the a-CuO x planarizes the ITO surface (R ITO rms = 10.9 nm, Fig. S4a). AFM was also used to estimate the thickness of the films by partially masking the sample from UV irradiation. It was found that the film thickness gradually decrease with annealing temperature. Cu-AP is around 20 nm thick, this thickness is reduced to 10 nm for Cu-100, and 5 nm for Cu-200 and ## Cu-250. A similar densification of a-FeO x with annealing was previously observed. 21 ## Electronic Characterization To assess the work function and valence band energy maximum of the a-CuO x over the annealing series UPS was performed on a-CuO x layers deposited on gold-coated Si wafers (Fig. 3). The Cu-250 sample exhibits the lowest work function at -4.9 eV (Fig. S5), this is ideally posi- The optical band gap and transmissivity of our a-CuO x materials was measured by UV/visible spectroscopy. Using Tauc plot fitting of the UV/visible transmission spectra (Fig. S7) it was determined that the a-CuO x thin films were indirect band gap semiconductors. 38,39 The transmission UV/visible spectra reveal that the a-CuO x thin films have greater than 80% transmissivity up to 340 nm for all annealing temperatures (Fig. 4). UV/visible absorbance spectra show that ## Organic photovoltaic devices Having characterized the electronic and structural properties of the a-CuO x materials, we demonstrate their utility as HTLs in OPV applications. Devices were constructed in a normal architecture configuration of ITO/HTL/Active layer/bathocuproine/Ag.The well-studied pairing of a wide band-gap polymer (PBDB-T) and non-fullerene acceptor (ITIC-Cl) were used as the active 3 Discussion ## Structural evolution of a-CuO x The photoinduced decomposition of Cu(2-ethex) 2 to form a-CuO x is driven primarily by an LMCT transition resulting in a photoreduction of the Cu metal centre and subsequent removal of the resulting radical-containing ligand fragments. When monitored by IR spectroscopy (Fig. 1) the decomposition product exhibits different decay constants for both of the most prominent stretches at 2959 and 1584 cm −1 . This suggests that intermediates are likely formed during the reaction. This is consistent with previous investigations into similar UV-based decomposition reactions. 22 After decomposition is complete the as-prepared films are highly amorphous as assessed by x-ray diffraction in which no signal is detectable, as seen in Fig. 2a. Subsequent annealing of the films in air failed to produce any evidence of crystallization up to a maximum temperature of 250 °C. Higher annealing temperatures were not tested as the intended application of these films was as hole-transport layers deposited on ITO substrates, and ITO is known to begin to thermally degrade at temperatures higher than 250 °C. 28 The completely amorphous nature of our a-CuO x materials required the use of synchrotronbased XANES spectroscopy in order to study any short-range ordering present in the structures. XANES spectra of a-CuO x annealed at increasing temperatures revealed fascinating trends in the short-range structure of the materials. Both the Cu-AP and Cu-100 samples exhibit a sharp features in the rising edge of the spectrum at 8985 eV (Fig. S2a). This is a well-known feature of Cu K-edge XANES spectra and is typically seen in Cu + species with low valence coordination modes such as 2-coordinate linear and 3-coordinate trigonal geometries, such as crystalline Cu 2 O. In both of these spectra the pre-edge peak corresponding to a 3d←1s transition is conspicuously absent, indicating that the levels of Cu 2+ in these species is below the detection limit. Upon annealing to 200 °C there is significant change in the Cu K edge spectrum (Figs. 2b and S2c). For both Cu-200 and Cu-250 the peak at 8985 eV has disappeared and been replaced by a shoulder at 8989-8990 eV ancillary to the main peak (S2b). This higher energy shoulderpeak is typical of octahedral Cu 2+ coordination environments, and is assigned to a "shakedown" peak made up of the 4p←1s transition coupled to an LMCT transition. Additionally, a small preedge peak, corresponding to the 3d←1s transitions is observable in these spectra at 8982 eV. Both of these observations are consistent with a CuO-like short-range order while maintaining long-range amorphous nature. All of these changes in the Cu K-edge XANES spectra taken together point to the presence of an amorphous-to-amorphous phase transition in our a-CuO x thin films; in other words, we are dealing with (broadly speaking) two amorphous polymorphs of a-CuO x . During this transition we move from linear low-valence coordination modes in a predominantly Cu + material to a primarily octahedral coordination environment with a much larger Cu 2+ presence. Amorphous-to-amorphous phase transitions are much rarer than their counterparts in crystalline systems but have been observed in semiconductor materials such as α-Si; 44 we recently observed similar behaviour in amorphous Ni and Fe oxyhydroxides used for electrocatalytic water splitting. 45,46 Given that we have observed changes in electrocatalytic behavior induced by amorphous phase transitions in other materials we were immediately interested in whether these similar transitions would lead to changes in the electronic properties of our a-CuO x systems. Scheme 1: The as-prepared films, whose bulk composition is best described as an amorphous Cu 2 O, undergoes drastic modification upon annealing while retaining its amorphous nature, rendering useful amorphous CuO hole-transport layers. In addition to X-ray absorption spectroscopy, XPS and AFM yield interesting information about the surface structure of our a-CuO x films. XPS is estimated to only probe the first 2-3 nm of the film in a CuO/Cu(OH) 2 film. This, combined with film thickness determination using AFM, inform us on the difference between surface and bulk film (Scheme 1). Cu-AP and Cu-100 XPS reveal a surface consisting almost entirely of Cu(OH) 2 , this is at odds with XANES results, which show a sample predominantly Cu I O 2 ). This is most likely a result of the surface contact with air/moisture, which allows for a more complete oxidation and formation of a hydroxide. Despite the similarity of Cu-AP and Cu-100 by both XANES and XPS, AFM thickness measurements reveal that the films height halved, from around 20 nm to 10 nm. The film thickness obtained for Cu-200 and Cu-250 are similar ( 5 nm) and their XANES spectra also show little difference. However, a distinct change is observed in the XPS spectra (Fig. 2), where the relative amount of Cu(OH) 2 decreases and the amount of CuO increases upon annealing from 200 to 250 °C. Since the thickness does not change significantly, and the XANES spectra match quite well, this can be assumed to be a surface effect, as more hydroxy-groups are lost in favor of an oxygen-terminated CuO surface during annealing. This may also explain the difference in work function between the Cu-200 and Cu-250 samples as well as the slight improvement in device performance. ## Electronic properties and device performance of polyamorphous a-CuO x Using a combination of ultraviolet photoelectron spectroscopy and UV/visible spectroscopy we were able to asses the Fermi energy (E F ), valence band edge (E VB ), and optical band gap (E opt g ) of the a-CuO x , the results of which are summarized in an energy level diagram (Fig. 7). The E F , as determined by secondary electron cutoff, are -4.34, -3.73, -4.12, and -4.90 eV for photocurrents and J (V ) curves with a convex shape, which is indicative of an imbalance in charge transport. 47,48 As this is not seen with control PEDOT:PSS-based devices, it can be safely inferred that the imbalance is due to a lack of hole extraction through the low-temperature a-CuO x polymorph HTL. This charge imbalance changes with increased annealing temperature of the a-CuO x . When the annealing temperature is increased from 100 to 200 °C the downward inflection in the J (V ) curve disappears and the solar cell performance drastically improves (PCE increases from 0.16% to 5.4%) displaying a diode-like J (V ) curve. Upon further annealing device performance reaches a maximum between 200 and 250 °C annealing with a highest PCE to date of 5.8%. ## Cu Taken together, the change in valence band energy with increased annealing temperature along with the improvement of hole transport in the photovoltaic devices implies that the nature of the a-CuO x is changing from that of an insulator or intrinsic semiconductor to a p-type semiconductor with increased density of holes increasing the layer's selective conductivity of positive charge. This change in electronic properties correlates very well with the change in structure observed in the XANES spectra of the a-CuO x annealing series. As the film converts from low coordinate Cu + to a four-coordinate Cu + 2 environment the doubling of positive charge must be matched by an increase in oxygen ligand content. If this charge is not fully compensated it may result in oxygen vacancies leading to a p-doped semiconductor. This preponderance of both electronic and structural evidence leads us to the conclusion that our a-CuO x thin films are poorly conductive with low charge carrier densities as prepared. Annealing of the a-CuO x layers induces structural change and oxidation state changes resulting in p-doping of the films and an increase in hole mobility. ## Conclusions To summarize, we have presented the preparation of amorphous a-CuO x semiconductor thin films using mild fabrication methods from inexpensive commercially available starting materials using UV-assisted deposition techniques. As prepared, these films are poorly conducting with low charge carrier densities. Annealing at temperatures in excess of 200 °C induces an amorphous-to-amorphous phase transition which is accompanied by structural and electronic changes in the metal-oxide film that result in p-type semiconducting behavior. These p-type semiconductor thin films have been applied as the hole transport layer in thin-film organic photovoltaic devices. In these devices the a-CuO x layers function comparably to control devices using PEDOT:PSS. In conclusion, we have demonstrated that polyamorphism exhibited by amorphous metal oxide thin films can result in material phases with drastically different electronic properties, leading to dramatic improvements in thin-film electronic device performance depending on the phase used. This demonstration of the utility of amorphous transition metal oxides has the potential for a wide variety of applications in organic electronic devices. in XPS analysis of first row transition metals, oxides and hydroxides: Sc, Ti, V, Cu and Zn. s. Silver counter electrodes were deposited using a thermal evaporator at a maximum pressure of 1×10 −6 mbar, at a ramped rate of 0.2 to 1 s −1 . The active device area was 0.14 cm 2 . ## Appl Current-voltage measurements were performed in air using a Keithley 2400 Source-Measure Unit and a 100 mW cm 2 light source using a simulated AM1.5G filter (Newport 92251A-1000). Power intensity was calibrated using a reference Si solar cell (Newport 91150V). EQE spectra were obtained using a QEX7 Solar Cell Spectral Response/QE/IPCE Measurement System (PV Measurement, Model QEX7) with an optically focused spot size of 0.04 cm 2 . Measurements were calibrated from 300-1100 nm using a Si photodiode. The decay of the absorbance was followed at two distinct wavenumbers νi each associated with well-resolved absorbance bands associated with the ligand, namely 2959 and 1584 cm −1 . ## S1.4 FTIR studies The absorbance was fit to a single-order decay: from which a half-life t 1/2 could be determined using ## S1.5 X-ray diffraction (XRD) XRD was performed with a Bruker D8 Advance Eco with a Cu K α tube (λ = 1.5406 ) in the 2θ = 3 -70 °range (step width 0.02 °, 1 s averaging per step). ## S1.6 Atomic force microscopy (AFM) AFM was performed with an Agilent 5500 atomic force microscope with silicon probe (tap300DLC, Ted Pella Inc.) in tapping mode. Samples scanned for imaging were deposited onto bare ITO; samples used in layer thickness determination were deposited onto Si(100). Samples used for height determination were made by masking a portion of the precursor Cu(2-ethex) 2 film during the UV irradiation process to prevent the formation of a-CuO x . The samples were then rinsed in n-butyl acetate to selectively remove unreacted Cu(2-ethex) 2 , yielding an area of bare Si substrate adjacent to a fully formed a-CuO x film, allowing height determination by AFM. Data analysis was performed in Gwyddion (http://gwyddion.net/). 49 ## S1.7 X-ray absorption spectroscopy (XAS) X-ray absorption spectra at the Cu K-edge were collected at the BioXAS beamline (main station) of the Canadian Light Source. X-ray fluorescence spectra were recorded with a 32-channel energy-discriminating Canberra Ge fluorescence detector. Spectra were acquired from 150 eV below the absorption edge to k = 14 −1 . All spectra were analyzed using the Demeter software package version 0.9.26, 50 after normalization and flattening. The energy was calibrated using the absolute energy list by Kraft, 51 using a Cu foil. The following ranges were used to flatten and S4 normalize the XANES spectra: pre-edge range -150 to -50 eV, normalization range 200 to 700 eV. ## S1.8 X-ray photoelectron spectroscopy (XPS) X-ray photoelectron spectroscopy (XPS) was performed using a Kratos Analytical XPS spectrometer (nanoFAB Fabrication and Characterization Centre, University of Alberta) with a monochromatic Al K α source (λ = 1486.6 eV) and a 2 mm 2 probing area. Sample were deposited onto goldcoated Si wafer using the same procedure as for bare Si substrates. All spectra were analyzed using CasaXPS software (http://www.casaxps.com/). Spectra were corrected by calibrating all peaks to the Au 4f 7/2 peak signal at 83.95 eV. A Shirley-type background was used, and curve fitting was performed using a combination of Gaussian and Lorentzian [GL (30)] profiles. The analysis of the O 1s and Cu 2p 3/2 region followed a methodology presented elsewhere. 35 Auger parameters were obtained by fitting the Cu LMM peak obtained from the survey scan. ## S1.9 Ultra-violet photoelectron spectroscopy (UPS) UPS measurements were performed using a Kratos Axis spectrometer with a source-analyser angle of 54.735600 °(nanoFAB Fabrication and Characterization Centre, University of Alberta), using a He(I) UV emission line with an energy hν of 21.21 eV. Work functions (φ) were determined by linear fitting of the secondary electron cutoff (E se ) and using the following equation: The valence band edge (E VB ) was determined using the minimum binding energy (E bind ) (determined by a linear fit of the onset of binding energy signal) and the work function according to the following equation: The conduction band edge (E CB was determined by adding the optical band gap E g (see ## S5 Section S1.10) to E VB : ## S1.10 UV-vis spectroscopy UV-vis measurements were carried out on a UV-vis-NIR Cary 5e using thin films coated on glass substrates as per Section S1.2. Optical band gaps were determined using the Tauc equation: (αhν) where n = 1/2 for a direct band gap transition. 38,39 S6 S2 Supplemental results

chemsum

{"title": "Photodeposited polyamorphous CuO x hole-transport layers in organic photovoltaics", "journal": "ChemRxiv"}

theoretically_probing_the_possible_degradation_mechanisms_of_an_fenc_catalyst_during_the_oxygen_redu

## Abstract: For the FeNC catalyst widely used in the oxygen reduction reaction (ORR), its instability under fuel cell (FC) operating conditions has become the biggest obstacle during its practical application. The complexity of the degradation process of the FeNC catalyst in FCs poses a huge challenge when it comes to revealing the underlying degradation mechanism that directly leads to the decay in ORR activity. Herein, using density functional theory (DFT) and ab initio molecular dynamics (AIMD) approaches and the FeN 4 moiety as an active site, we find that during catalyzing the ORR, Fe site oxidation in the form of *Fe(OH) 2 , in which 2OH* species are adsorbed on Fe on the same side of the FeN 4 plane, results in the successive protonation of N and then permanent damage to the FeN 4 moiety, which causes the leaching of the Fe site in the form of Fe(OH) 2 species and a sharp irreversible decline in the ORR activity. However, other types of OH* adsorption on Fe in the form of HO*FeOH and *FeOH intermediates cannot cause the protonation of N or any breaking of Fe-N bonds in the FeN 4 moiety, only inducing the blocking of the Fe site. Meanwhile, based on the competitive relationship between catalyzing the ORR and Fe site oxidation, we propose a trade-off potential (U RHE TMOR ) to describe the anti-oxidation abilities of the TM site in the TMN X moiety during the ORR. ## Introduction The oxygen reduction reaction (ORR) at the cathode of Fuel Cells (FCs), as the most critical step heavily limits the total output power density of FCs, because of the sluggish kinetics of ORR. While Pt and Pt-based metals as the best catalysts for ORR are extremely defcient in the Earth's resources. 1 To meet the demands for the widespread adoption of FCs, it is necessary to explore non-noble-metal-based catalysts with relatively low cost as alternative electrocatalysts for ORR. In the past ten years, meaningful progress has been achieved, especially for transitional metal-nitrogen-carbon (TMNC) catalysts, such as FeNC, 2 ZnNC, 3 and CoNC, 4 which stand out from other carbonbased catalysts as the most promising substitutable Pt-based ORR catalysts, not only in alkaline media but also in acidic media. At present, the power density of TMNC catalysts is approaching about half that of commercial Pt-based catalysts in membrane electrode assembly (MEA) tests. 5 However, numerous challenges still remain before TMNC catalysts will become viable for Proton Exchange Membrane FCs (PEMFCs), of which catalyst instability seems to be the greatest. More specifcally, the performance of a PEMFC with an FeNC based cathode typically degrades by 40-80% in the frst 100 h of durability testing showing fast degradation of activity. 5 Therefore, understanding the degradation mechanisms of TMNC catalysts becomes critically important, especially in acidic media. Four types of mechanisms have been proposed for the deactivation of an FeNC catalyst in an acidic medium: (1) Nitrogen species protonation: 9 the protonation of highly basic nitrogen groups neighboring the active site, followed by anion adsorption. Herranz et al. 9 proposed that the turnover frequency (TOF) value on FeNC active sites should be high when the basic nitrogen groups next to the FeN X site are protonated, but without anion adsorption. But once the anion adsorption took place on an Fe/C site, the TOF value would become low. Dodelet et al. 8 speculated that fluorination (which could be regarded as direct anion adsorption) of FeN 4 active sites leads to the formation of C-F bonds on some carbon sites and Fe-F bonds on all metal sites, and these bonds induce instability in the FeNC catalyst during FC processing. However, their subsequent research demonstrated that the anion adsorption on FeN 4 active sites has no effect on the initial decay in ORR activity, and the hypothesis of the anionic neutralization of the FeN 4 /NH + site should be abandoned. 10 Similarly, Mukerjee et al. 11,12 also confrmed that the deactivation of FeNC catalysts does not depend on the anions, and FeN 4 exhibits immunity to surface poisoning by some anions. (2) Carbon/metal oxidation: 6 the carbon/metal active sites on the catalyst surface can be attacked by H 2 O 2 -derived radicals, which then proceed to induce either a Fenton reaction (Fe 2+ + H 2 O 2 + H + / Fe 3+ + $OH + H 2 O) or the formation of surface oxidation intermediates. However, this type of deactivation might be partially reversible upon reduction of the catalyst surface. Jaouen et al. 7 demonstrated that the FeNC catalyst is structurally stable, but in an acidic medium, it is electrochemically unstable. This is owing to the existence of H 2 O 2 resulting in some untouched Fe-based catalytic sites and oxidation of the carbon surface by which TOF should be decreased. The TOF is then recovered upon electrochemical reduction of the carbon surface. Their results also confrmed that it is the peroxidederived reactive oxygen species (such as OH, OOH and O radicals), rather than molecular H 2 O 2 , that cause the degradation of the FeNC catalyst. (3) Demetalation of the Fe atom or FeN X center: 13 the demetalation of Fe atoms or FeN X moieties on the catalyst surface by chemical/electrochemical corrosion or by degradation of the carbon framework. Mayrhofer et al. 13 demonstrated that carbon oxidation and demetalation of Fe species happen at high (>0.9 V) and low (<0.7 V) potentials, respectively. Even though Fe demetalation can potentially cause damage to the FC system, they did not think that Fe demetalation could lead to decay in ORR activity. Conversely, Chenitz et al. 14 proposed that the demetalation of the FeN X center is independent of FC potential. Based on the Le Chatelier principle, the FeNC/Fe 2+ thermodynamic equilibrium would shift toward the formation of more Fe 2+ while water runs into the micropores, which should be responsible for the fast decay of FC performance. (4) Water flooding: 10 water runs into the pores of the catalysts and impedes the transport of oxygen into the cathode. Dodelet et al. 10 found that the most active FeN X sites, that occupy the micropores of catalysts, are involved in the initial rapid decay of catalytic activity in FCs due to the water flooding of catalyst microspores. However, Banham et al. 15 considered that micropore flooding cannot be associated with the observed destruction by investigating the micropore flooding in situ before and after stability testing. This result suggested that the 'deactivation' and/or 'loss' of key active sites lead to kinetic losses, and the oxidation of active sites/carbon should be the most probable cause. In particular, more than one of these deactivation mechanisms may occur in parallel during stability or durability testing of FCs, which signifcantly enhances the complexity of the degradation process and makes it a big challenge to reveal the underlying dominant one. Presently, there are a lot of unanswered questions: for example, which degradation mechanism is primarily responsible for the fast decay of ORR catalytic activity? What causes the permanent damage to the FeNC catalyst, such as the loss of active sites and demetalation of Fe sites? How do the degradation mechanisms interact with each other or do they combine together to destroy the FeNC active site and cause the loss of activity? Thus, this necessitates a comprehensive recognition of these possible degradation mechanisms and their contribution to the decay of ORR catalytic activity for an in-depth understanding of the stability of TMNC catalysts. Here, we have combined density functional theory (DFT) with ab initio molecular dynamics (AIMD) to probe the thermodynamic description of each possible degradation mechanism and the dynamic structural evolution of the active sites of FeNC during ORR, and we try to illuminate their interaction and contribution to the destruction of active sites and the decay of ORR activity. We chose the FeN 4 moiety as the active site model at the atomic level and constructed an FeNC periodic slab model without defects or edges to undertake the following study. This is because FeN 4 is the dominant existing form in FeNC catalysts produced by a pyrolysis method, 5, and is the most-studied active site showing high ORR activity. 6,9,11,19 Though the defects or edges might enhance the ORR activity of FeNC catalysts by increasing the density of active sites or changing the electronic structure, 20 the ambiguous interaction between defects/edges and active sites makes the atomically active site structure more complex and varied, and then it is hard to explore the degradation mechanism directly caused by the FeN 4 moiety. In this work, frstly, we calculated each proposed degradation mechanism, including protonation of N sites and oxidation of Fe sites (FeOR) by ORR intermediates, to clarify the possibility that each mechanism happened during ORR in terms of thermodynamics. Then, we comprehensively investigated the dynamic structural evolution of the FeN 4 moiety under the effect of different degradation mechanisms to reveal their interactions and influences on triggering the demetalation of the FeN 4 moiety. Meanwhile, we unveiled the crucial intermediate bridging the competitive relationship between ORR and FeOR, and defned a trade-off potential (U RHE TMOR ) to describe the anti-oxidation ability of metal active sites. ## Methods Density functional theory (DFT) was implemented in Vienna Ab Initio Simulation Package (VASP) code. 21 In order to simulate an H 2 O solvent environment, the thermodynamic zero-point energy was calculated with the implicit solvation model-VAS-Psol. 22 The Perdew-Burke-Ernzerhof (PBE) functional within the generalized gradient approximation (GGA) was adopted to describe electronic exchange-correlation energy. The ionic cores were described with the projector augmented wave (PAW) method. The 6 6 supercell doped-graphene structure was separated by a vacuum of 15 height from its neighbours. An energy cutoff of 500 eV was used for the plane wave basis, and the Kpoints were set to 2 2 1. It is generally recognized that the localized 3d electron correlation for a transition metal in the fourth period can be described by the DFT + U method. 23 Here we applied DFT + U through a rotationally invariant approach with the corresponding U-J values ((U-J) Fe ¼ 3.29 and (U-J) Zn ¼ 4.12). 24 All of the calculations were continued until the force and energy had converged to less than 0.02 eV 1 and 10 5 eV, respectively. 25 The AIMD simulations were performed using the canonical ensemble (NVT) and Nosé-Hoover thermostat method at 300-500 K (from room temperature to FC operating temperature and even higher) and lasted for 10-15 ps, in which the timestep is set as 1 fs, by increasing the hydrogen mass to 2 atomic mass units (H/D-exchange) to reduce the computational cost. 3. Results and discussion ## Nitrogen species protonation The protonation of N sites in the FeN 4 moiety was calculated by combined DFT and AIMD simulations to elucidate the details of the change in Fe-N bonds (Fig. 1). And a set of atomic coordinates obtained as a result of the AIMD simulations at 500 K (Fig. S1-S3 †), were used to calculate radial distribution functions 31 (RDF, Fig. 1d). In Fig. 1a, for an FeNC slab with one adsorbed H atom on the N site (Fig. 1ai, FeN 4 H), *H would spontaneously shift from the N to the Fe site after DFT optimization (Fig. 1aii). Meanwhile, this structure shows dynamic stability under a 10 ps AIMD process in implicit solvation (Fig. 1aiii and S4a †) and explicit solvation (Fig. S5 and S6, Video S1a †). And both implicit and explicit solvation show similar effects on the dynamic stability of the FeNC structure. Once the number of adsorbed H increases to 2, as shown in Fig. 1b and S2 † (FeN 4 H 2 ), *H not only bonds with the Fe site, but also forms an H-H bond, and then diffuses to the vacuum layer as molecular H 2 within 0.1 ps which is shown in Fig. S4b. † When the third H is introduced into the system (Fig. 1ci, FeN 4 H 3 ), the Fe site binds two *H atoms and the N site adsorbs one *H (Fig. 1cii), showing thermodynamic stability. While, in AIMD, *H 2 is desorbed from the Fe site after 0.5 ps and then *H on the N site migrates to the Fe site at the same time, indicating the unstable protonation of N (Fig. S3 and S4c †). In addition, the Fe-N distance distribution (Fig. 1d) shows that the adsorbed proton results in a negative shift of about 0.01-0.02 for the Fe-N peak value compared to that of the bare FeNC model (black dotted line, 1.90 ), meaning the Fe-N bond is strengthened slightly. Therefore, on a bare FeNC surface, protonation of N in the FeN 4 moiety is unlikely to occur, and simultaneously a proton prefers to bind with the Fe site which has almost no effect on the structure of the FeN 4 moiety. ## Thermodynamic mechanism of Fe oxidation by ORR intermediates During the ORR process, the Fe active sites on the FeNC catalyst surface would be attacked or oxidized by ORR key intermediates (OOH, O, OH and H 2 O 32 ), forming surface oxidation intermediates or proceeding in a Fenton reaction, even leading to the permanent loss of Fe sites or FeN X centers. Here, as shown in In Fig. 2a, at U RHE ¼ 0.9 V (about the half-wave potential of the FeNC catalyst when catalyzing ORR 18 ), the formation of *FeOOH is spontaneous, and *FeOOH acts as a bifurcation point, directing the following reactions branched to ORR(2e ), ORR(4e ) and FeOR. Specifcally, the successive step of ORR(2e ) is proton and electron transfer (PET) to form H 2 O 2 , which is the PDS of ORR(2e ) with a DG PDS value of 1.20 eV. The following ORR(4e ) steps include the downhill dissociation of O-OH to generate *FeO, and continuous PET to desorb OH. Among these steps, OH desorption is the PDS of ORR(4e ) with a DG PDS of 0.49 eV. The frst FeOR path starts from the rearrangement of *FeOOH, which subsequently leads to the formation of *FeO(OH) oxidation species. The successive protonation step to form *Fe(OH) 2 , in which 2OH species are adsorbed on the same side of the FeN 4 moiety, is the PDS with a DG PDS of 0.20 eV. Apart from *FeOOH, *FeOH is the second bifurcation point connecting the OH desorption in ORR(4e ) and the further OH adsorption in the second FeOR path. The further OH adsorption on *FeOH to form HO*FeOH is the PDS with a DG PDS of 0.11 eV. Because the DG PDS value of FeOR is much lower than that of ORR(2e ) and ORR(4e ) at U RHE ¼ 0.9 V, the Fe site of the FeN 4 moiety is more likely to be oxidized by ORR intermediates instead of catalyzing ORR. Once the near C site of the FeN 4 moiety binds with OH, in Fig. 2b It can be seen that *Fe(OH), *Fe(OH) 2 and HO*FeOH are the dominant intermediates, and would have a great influence on stability and ORR activity. Among them, the *FeOH intermediate is key, connecting the PDS of ORR(4e ) (*FeOH + e / *Fe + OH , OH desorption) and the PDS of FeOR (*FeOH + OH / OH*FeOH + e , OH further adsorption/attack), corresponding to the recovering and blocking of Fe active sites, respectively. As shown in Fig. 2c, the competitive relationship can be described by comparing the DG values of the two reactions. When the DG PDS of ORR(4e ) is equal to that of FeOR at a certain electrode potential, the OH desorption and the further OH adsorption of *FeOH have the same probability. Then, we defne the certain electrode potential as a trade-off potential (U RHE TMOR , the process of derivation is listed in ESI †) to indicate the competitive relationship between ORR and FeOR and describe the anti-oxidation ability of the Fe site. As shown in Fig. 2d, for the FeN 4 C surface, the DG intersection of OH desorption and further OH adsorption corresponds to the trade-off potential value, which is 0.61 V. On the left-hand side (U RHE < 0.61 V), the DG of OH desorption is lower than that of further OH adsorption; while on the right-hand side (U RHE > 0.61 V), the DG of OH desorption becomes larger than that of further OH adsorption, suggesting that the blocking of the Fe site is more likely to occur than recovery of the Fe site when the electrode potential is higher than 0.61 V. In the case of FeN 4 COH, U RHE TMOR decreases to 0.44 V, showing that OH desorption needs a higher overpotential to compete with further OH adsorption, and the Fe active site is prone to be occupied by OH at a wider potential range. Then, the U RHE TMORindicates the turning point at which the reaction changes from further OH adsorption to OH desorption with decreasing U RHE . And the more negative the U RHE TMOR value (smaller than the ORR equilibrium potential, 1.23 V vs. RHE), meaning the poorer the competitive ability of OH desorption with further OH adsorption on the Fe site, and the lower the anti-oxidation ability. Thoroughly regulating the balance of adsorption/desorption of one or more OH species on the Fe site could shift the U RHE TMOR to positive, beneftting the recovery of the active site and improving the stability and activity simultaneously. Hence, from the thermodynamic point of view, Fe site oxidation by ORR intermediates is thermodynamically feasible. And the blocking of the Fe site could occur more smoothly than ORR(4e ) or ORR(2e ) when U RHE is higher 0.61 V. ## The trigger for the demetalation of the FeN 4 moiety Although the thermodynamic mechanism demonstrates that the formation of *FeOH, HO*FeOH and *Fe(OH) 2 oxidation intermediates during ORR is inevitable, it is still unknown whether the Fe site would leach or not. To probe the possibility of demetalation of the FeN 4 moiety, these key intermediates should be further researched. The structures of *FeOH, HO*FeOH and *Fe(OH) 2 were calculated by DFT calculation and AIMD simulation to unveil the influence of OH species on the FeN 4 moiety. The DFT optimized confgurations in Fig. 3a show that the structure of *Fe(OH) 2 is more distorted than those of *FeOH or HO*FeOH, because the same side adsorbed OH species on the Fe site, damaging the framework plane. And the Fe-N bond lengths of the three intermediates in DFT optimization are in good agreement with the statistic of the Fe-N distance in AIMD. In Fig. 3b, the Fe-N lengths in the *FeOH and HO*FeOH system (1.91 and 1.92 , respectively) are close to that in the bare FeNC model (1.90 , Fig. 1d), while the largest Fe-N length in the *Fe(OH) 2 structure of 2.14 is signifcantly elongated, demonstrating that the same side adsorbing OH species would weaken the interaction of Fe-N. Furthermore, *Fe(OH) 2 with COR (*Fe(OH) 2 -C(OH) X (X ¼ 1 and 2)) similarly shows an elongated Fe-N distance in the range of 1.97 to 2.12 (Fig. S9 †). The structural deformation of the FeN 4 moiety determines that *Fe(OH) 2 must be the crucial intermediate, which can quite possibly induce the further destruction of the FeN 4 moiety. However, AIMD simulation fnds that *Fe(OH) 2 is dynamically stable due to the almost unchanged structure after 10 ps of AIMD simulation, even in explicit solvation (Fig. S10 and S11, Video S1b †). Therefore, other factors, that directly cause the breaking of the Fe-N bond and then leaching of the Fe site, need to be revealed in detail. Deep analysis of the electronic structure about *FeOH, HO*FeOH, *Fe(OH) 2 , including ICOHP (Fig. S12 †) and Bader charge (Table S1 †), indicates that *Fe(OH) 2 with fewer electrons between Fe-N bonds causes N with a likely pyridine N property. This means that protonation on N in *Fe(OH) 2 would take place. DFT optimization demonstrates that the protonation of N in *Fe(OH) 2 is stable, while HO*FeOH with oppositely adsorbed OH species cannot cause the protonation of N, which instead steps up the desorption of OH due to the combination of OH and proton to form H 2 O (Fig. S13 †). In Fig. 4a, the DFT calculated result reveals that the frst protonation of N in *Fe(OH) 2 (*Fe(OH) 2 -NH) is thermodynamically favorable (DG < 0 eV), in which two of the four Fe-N bonds dissociate and *Fe(OH) 2 stands out in the 2D framework plane. The second protonation of N (*Fe(OH) 2 -(NH) 2 ) further destroys the third Fe-N bond leaving only one Fe-N bond. Although the formation of *Fe(OH) 2 -(NH) 2 is an endothermic process, the energy barrier of 0.77 eV is close to 0.75 eV, 34 indicating that the reaction can be surmounted easily by tuning the reaction conditions, such as temperature, species concentration, and electrode potential. AIMD simulation further shows that the formation of *Fe(OH) 2 -NH is dependent on the direction of proton adsorption. As shown in Fig. S14 and Video S2a, † one proton interacts with N on the same side of OH adsorption in the *Fe(OH) 2 system, and H would combine with OH and form H 2 O to beneft ORR(4e ). While, when the proton attacks N from the opposite side of the adsorbed OH species in the *Fe(OH) 2 system (Video S2b †), the protonation of the N atom then breaks two Fe-N bonds spontaneously. As shown in Fig. 4b and S15, † the *Fe(OH) 2 -NH structure exhibits dynamic stability during the AIMD process within 15 ps at 300, 400 or even 500 K (Fig. S16 and S17 †). The Fe-N RDF in *Fe(OH) 2 -NH shows two peak values at 1.96 (Fe-N bond) and 3.04 , respectively, which is consistent with the DFT results. In the AIMD trajectories of Fig. 4c and S18, † *Fe(OH) 2 -(NH) 2 is metastable, and can only survive for $2 ps at 300 K, in which the third Fe-N bond dissociates with one elongated Fe-N bond remaining (Fig. 4d). After 2 ps, the only remaining Fe-N bond in *Fe(OH) 2 -(NH) 2 is gradually broken, and Fe(OH) 2 detaches from the graphene plane and diffuses into the vacuum layer. Furthermore, when the temperature is raised to 400 or even 500 K (Fig. S19 and S20 †), the Fe-N bonds in *Fe(OH) 2 -(NH) 2 dissociate immediately and release Fe(OH) 2 species. The above results reveal that, for the demetalation of an FeNC catalyst in an acidic medium, the formation of *Fe(OH) 2 is an essential prerequisite, which results in N protonation in Considering the degradation mechanism and the decline of activity, it is feasible to identify the anti-oxidation ability of TMNC catalysts by researching the formation of *TM(OH) X (X ¼ 1 and 2) during ORR, and is also even feasible to decrease the leaching of TM sites by eliminating the formation of a *TM(OH) 2 intermediate. To verify the practicability of the above guidelines, we further calculated the oxidation mechanism of ZnNC during ORR, referring to our previous experimental results. 3 In Fig. S22, † the free energy diagram of *Zn(OH) 2 and *ZnOH shows that ZnNC prefers to be occupied by OH instead of leaching Zn due to the rather difficult TMOR of ZnNC is higher than 1.23 V. Then, compared to FeNC, ZnNC exhibits much higher anti-oxidation ability, which is in agreement with our previous experimental results. 3 That is, the ZnNC catalyst largely maintains its ZnN X active sites after an accelerated stress test, and shows only a slight decrease in ZnN X content and a relatively small decrease in pyridinic-N content in an acidic medium, whereas the FeNC catalyst shows a rather large decrease in FeN X and pyridinic-N content. The half-wave potential of ZnNC decayed from 0.746 V to 0.726 V after 1000 CV cycles, whereas a signifcantly larger decay (from 0.743 to 0.712 V) was observed for the FeNC catalyst. 3 However, due to the too weak OH adsorption, ZnNC shows poorer ORR intrinsic activity than FeNC. This means that, to screen TMNC catalysts with high activity and stability, it is not only necessary to consider the volcano relationship between *OH adsorption and ORR activity, 35,36 but the trade-off relationship between ORR activity and anti-oxidation ability should also be considered. ## Conclusions The present work provided a DFT + AIMD approach to investigate the possibility of each degradation mechanism relating to FeN 4 happening during the ORR with the aim of depicting the complete thermodynamic mechanism and the dynamic structural evolution of an FeNC catalyst. The calculation results found that N protonation of bare FeNC is difficult, while Fe/C oxidation by ORR intermediates is inevitable. *FeOH as a key intermediate can describe the competitive relationship between catalyzing the ORR(4e ) and FeOR, and FeOR can surpass the ORR(4e ) when the electrode potential is higher than the trade-off potential U RHE TMOR of 0.61 V. The formation of *Fe(OH) 2 from the FeN 4 moiety leads to favorable N protonation. And the *Fe(OH) 2 intermediate interacts with successive N protonation to further destroy the FeN 4 moiety, thus largely leading to the leaching of the Fe site in the form of Fe(OH) 2 . The change in the intrinsic ORR activity determines that the demetalation of an FeNC catalyst is the underlying dominant reason for activity decay in an acidic medium. Meanwhile, our research has presented a trade-off potential to describe the anti-oxidation abilities of TMNC catalysts.

chemsum

{"title": "Theoretically probing the possible degradation mechanisms of an FeNC catalyst during the oxygen reduction reaction", "journal": "Royal Society of Chemistry (RSC)"}

j_o_u_r_n_a_l_na_me_screening_of_hydrogen_bonding_interactions_by_a_single_layer_graphene

## Abstract: A single layer of graphene when transferred to a solid substrate has the ability to screen or transmit interactions from the underlying substrate, which has direct consequences in applications of this 2D material to flexible electronics and sensors. Previous reports using a multitude of techniques present contradictory views on graphene's ability to screen or transmit van der Waals (vdW) and polar interactions. In the present study, we use interface-sensitive spectroscopy to demonstrate that a single layer graphene is opaque to hydrogen bonding interactions (a subset of acid-base interactions), answering a question that has remained unresolved for a decade. Similar frequency shifts of sapphire hydroxyl (OH) for graphene-coated sapphire in contact with air and polydimethylsiloxane (PDMS) demonstrate the insensitivity of sapphire OH to PDMS. The screening ability of graphene is also evident in the smaller magnitude of this frequency shift for graphene-coated sapphire in comparison to that for bare sapphire. The screening of acid-base interactions by a single layer graphene results in the significant reduction of adhesion hysteresis for PDMS lens on graphene-coated substrates (sapphire and silicon wafer, SiO 2 /Si) than bare substrates. Our results have implications in the use of PDMS stamps to transfer graphene to other substrates eliminating the need for a wet-transfer process. ## Introduction Graphene, a single-atom-thick sheet of carbon atoms arranged in a hexagonal lattice, is a fascinating material for understanding the physics of two-dimensional (2D) materials and for use in many technological applications. 1,2 Its unique 2D structure, high surface area, optical transparency, and unprecedented electrical, mechanical, and thermal properties make it an attractive material for fabricating transparent, flexible electronics and sensors. 3,4 The use of graphene, the thinnest known material (0.34 nm), as a conformal coating opens up interesting pathways for surface modification of materials. For instance, addition of a graphene layer on copper (Cu) is known to prevent its atmospheric oxidation. 5 Additionally, in most practical applications, graphene is typically supported on an underlying substrate. Thus, the question on how the substrate supporting the graphene layer influences the interactions of graphene with other media has been of significant interest in the past decade. A technique that has been widely employed in the literature to determine the transparency of graphene (i.e. the extent to which graphene influences interactions between the underlying substrate and media on other side) is contact angle measurements with water and other liquids. However, results from studies utilizing the same approach present controversial ideas on transparency of graphene. For instance, while the first report by Rafiee et al. suggested complete wetting transparency of graphene for substrates interacting primarily via van der Waals (vdW) interactions, 6 later work by several researchers suggested partial wetting transparency to complete wetting opacity. The transparency of graphene still remains an unresolved topic as Belyaeva et al. recently showed complete transparency of graphene to vdW and polar interactions, 12 while Du et al. reported only 20% transmission of polar interactions. 13 The controversy over graphene's transparency also prevails in studies utilizing other techniques including atomic force microscopy (AFM) and electron backscatter diffraction (EBSD). Tsoi et al. inferred complete screening of vdW interactions by a single layer graphene and molybdenum disulfide. 14 However, Chiou et al. reported different Hamaker constants on free and supported graphene layers (derived using bimodal AFM) signifying an influence of the underlying substrate. 15 Therefore, more direct spectroscopic approaches would aid in determining the ability of graphene to shield vdW and polar interactions. Recently, x-ray techniques including x-ray photoelectron spectroscopy (XPS) and x-ray photon correlation spectroscopy (XPCS) have been utilized to understand how the addition of graphene on an underlying substrate influences adsorption. Presel et al. employed combined experimental XPS and theoretical approach to study adsorption of carbon monoxide (CO) and argon (Ar) and concluded 50% transmission of vdW interactions. 18 In another study, adsorption of water was found to be fully dominated by water-substrate interactions. 19 Additionally, surfacesensitive non-linear spectroscopies, second harmonic generation (SHG) and sum frequency generation (SFG), have been employed to provide insights into graphene's ability to screen or transmit interactions. SHG results with water and SFG results with polystyrene on bare and graphene-coated silica suggest partial to complete opacity of graphene towards intermolecular interactions. 22,24,28 Similar conclusions can be derived from structure of ionic liquids on bare and graphene-coated CaF 2 and BaF 2 substrates, and water structure on bare and graphene-coated sapphire. 21,25,26 However, adsorption of 1-hexanol from cyclohexane on bare and graphene coated alumina, 23 and water structure on bare and graphene-coated CaF 2 and silica substrates indicated negligible effect of the graphene layer. 27 All these studies indirectly analyze the effect of graphene by comparing spectral signatures across bare and graphene-coated substrates. However, the strong dependence of SFG signals on orientation could influence the conclusions derived thereby necessitating the need for a direct molecular probe for resolving this ongoing debate over transparency of graphene. Spectroscopic shifts, originally proposed by Badger and Bauer, offer a unique way to determine the nature and strength of intermolecular interactions. Kurian et al. and Wilson et al. combined this concept with surface-sensitive SFG to examine interfacial interactions between hydroxylated sapphire and various nonpolar/polar liquids and polymers. 36,37 The frequency shift (∆ν) of sapphire surface hydroxyls (OH) provided a direct measure of interaction strength. Weak vdW interactions shifted the sapphire OH peak by only 20-30 cm −1 with respect to its position in air, while strong acid-base interactions (a broad term encompassing hydrogen bonding, donor-acceptor, and electrophile-nucleophile interactions) shifted the sapphire OH by as much as 120 cm −1 . The interaction energy calculated using the observed frequency shifts correlated well with that calculated using Drago-Wayland coefficients signifying the validity of this approach. Further, this concept successfully predicted the preferential surface segregation of strongly interacting component from binary liquid mixtures, polymer blends, and polymer solutions, consistent with experimental and simulation results. The differences in the frequency shift of sapphire OH peak in contact with another media with or without graphene layer could provide direct evidence of graphene's transparency towards vdW and polar interactions. In the present study, we investigate the ability of graphene to screen acid-base interactions between polydimethylsiloxane (PDMS) and hydroxylated sapphire (and silicon wafer, SiO 2 /Si) using a combination of adhesion and interface-sensitive spectroscopy measurements. PDMS-sapphire and PDMS-SiO 2 /Si were chosen because it is well known that acid-base interactions between weakly basic siloxane (Si-O-Si) groups of PDMS and acidic OH groups of sapphire (or SiO 2 /Si) result in a significantly high adhesion hysteresis (i.e. work done in separating the two surfaces from adhesive contact is larger than the work done in bringing the two surfaces in contact). 43,44 A comparison of the differences between bare and graphene-coated substrate in terms of adhesion behavior and sapphire OH peak shifts would provide a clear understanding of graphene's transparency. Our adhesion results indicate a significant reduction in adhesion hysteresis for graphene-coated substrates, indicating the screening of acid-base interactions by a single layer of graphene. Similar sapphire OH frequency shifts for graphene-coated sapphire in contact with both air and PDMS indicate the insensitivity of sapphire OH to PDMS. Additionally, the relatively small frequency shift of sapphire OH in contact with PDMS for graphene-coated sapphire as compared to bare sapphire further corroborates the opaqueness of graphene to acid-base interactions of the underlying substrate. Our results would benefit the scientific community aiming to understand transparency of graphene and also provide an interesting avenue to quantify PDMS-graphene adhesive interactions crucial for realizing large-scale roll-to-roll transfer of graphene using PDMS stamp. 2 Results ## Graphene Characterization We determine the quality of our monolayer graphene transferred onto SiO 2 /Si and sapphire substrates using Raman spectroscopy in order to ensure that the transparency of our graphene is not influenced by defects. 17,50 Figure 1a shows the representative Raman spectra of graphene layer transferred onto SiO 2 /Si and sapphire substrates, which show prominent peaks at 1585 cm −1 and 2695 cm −1 known as the G and 2D peaks, respectively. 51 Absence of D peak at ∼1350 cm −1 indicates that the transferred graphene is defect free. The calculated ratio of intensity for 2D and G peaks (I 2D /I G >1.5) along with a narrow full width half maxima (FWHM) of 2D peak confirms the presence of a good quality monolayer graphene. We further evaluate the conformality of graphene to the underlying substrates by collecting AFM topographical images of bare and graphene-coated substrates (Figure 1b, c). The AFM image of graphene-coated SiO 2 /Si appears very similar to that of the underlying SiO 2 /Si substrate with very few wrinkles caused by the wet transfer technique. 51 The calculated root mean square (RMS) roughness values for graphene(Gr)/SiO 2 /Si (77±15 pm) and SiO 2 /Si (59±6 pm) are similar, providing evidence of conformal contact between graphene monolayer and the underlying silicon wafer. Similar roughness values are observed for bare (63±9 pm) and graphene-coated (82±15 pm) sapphire substrates (Figure S1, †ESI). ## Adhesion Measurements We perform adhesion measurements using a custom-designed JKR setup (Figure 2a), 52,53 where a soft elastomeric PDMS lens is brought in contact with bare (or graphene-coated) substrates. After loading the lens to a maximum preload of 1 mN and equilibrating for 3 min, the PDMS lens is retracted at a constant speed of 60 nm/s until pull-off event occurs before complete detachment. During both loading and unloading, force (F) and contact area are recorded simultaneously. The loading curve is fit with Equation 1to calculate W a from loading. Since it is difficult to obtain a good fit for the unloading data, we use the pull-off force (F pull−o f f ) to calculate W a during unloading using Equation 3. Figure 2b shows the contact radius cube (a 3 ) vs. load (F) plot for 1.9 MPa PDMS lens on Gr/SiO 2 /Si (pink circles) and bare SiO 2 /Si (black triangles) during loading (or approach) and unloading (or retraction). For Gr/SiO 2 /Si, the unloading curve follows a path similar to loading. However, for bare SiO 2 /Si, the unloading curve follows a different path compared to loading indicating significantly larger amount of adhesion hysteresis (i.e., work done in separating the PDMS and SiO 2 /Si surfaces is larger than the work done in bringing them in contact). Additionally, the higher strain energy release rate (G, calculated by solving Equation 1 at each data point (Figure S2, †ESI)) from unloading as compared to loading also highlights this large adhesion hysteresis. The moduli of PDMS lenses has a negligible effect on the adhesion behavior as similar results are observed for 0.7 MPa PDMS lenses (Figure S3, †ESI). The adhesion results obtained for 1.9 MPa PDMS lenses on graphene-coated (red circles) and bare sapphire (blue squares) substrates (Figure 2c) are consistent with trends observed with graphene-coated and bare SiO 2 /Si. The W a values obtained from loading and pull-off data for PDMS lenses tested on bare and graphene-coated SiO 2 /Si and sapphire substrates are summarized in Figure 2d. The W a values obtained from loading for Gr/SiO 2 /Si and SiO 2 /Si are 46±5 mJ/m 2 and 55±2 mJ/m 2 , respectively. Similar W a values are obtained for uncoated and graphene-coated sapphire substrates. Interestingly, the effect of graphene is more evident in the W a values obtained from pull-off measurements. For Gr/SiO 2 /Si, the W a obtained from pulloff is 64±2 mJ/m 2 , a value not very different from that obtained from loading implying a low adhesion hysteresis. The amount of adhesion hysteresis for Gr/SiO 2 /Si (17±4 mJ/m 2 ) is similar to the inherent hysteresis of PDMS lenses (18±6 mJ/m 2 ) measured by testing them on octadecyltrichlorosilane (OTS)-coated silicon wafers (Games-Howell test, p-value=0.94, Figure S4, †ESI). However, for SiO 2 /Si substrate, the W a calculated from pull-off is 360±5 mJ/m 2 (∼6 times the value obtained from loading) indicating a large adhesion hysteresis, consistent with literature. 43,44 Same is true for bare and graphene-coated sapphire, except for a slightly larger adhesion hysteresis for graphene-coated sapphire than graphene-coated SiO 2 /Si. The high adhesion hysteresis observed for bare substrates has been attributed to acid-base interactions between the acidic surface Si-OH (or Al-OH) groups of silicon wafer (or sapphire) and weakly basic siloxane (Si-O-Si) groups of PDMS. 43,44 The low adhesion hysteresis observed for graphenecoated substrates would thus indicate screening of polar acid-base interactions by a single layer graphene. ## Spectroscopic Measurements To provide direct evidence of graphene's ability to screen acid-base interactions, we use SFG to examine the contact interface between PDMS and graphene-coated (or bare) sapphire. Before bringing PDMS lens in contact, we collect SFG scans for air-Gr/sapphire and air-sapphire (Figure 3a). The air-sapphire SFG spectrum (red circles) in PPP polarization shows a peak at ∼3708±3 cm −1 , attributed to the O-H stretch vibration of the sapphire surface OHs not participating in any interactions (or referred to as sapphire free OHs). 36,37,39 In contact with graphene, the sapphire OH peak shifts to 3644±7 cm −1 (i.e. frequency shift of 65±9 cm −1 , which is similar to 56±13 cm −1 observed for benzene in contact with sapphire 37,39 ) due to interactions between sapphire OHs and graphene. The observed frequency shift is comparable to that reported by Ohto et al. for water free OD in contact with graphene at the air-water interface. 54 Part of the sapphire free OH peak persists even when graphene is in contact with sapphire due to either the inability of all sapphire OHs to interact with graphene or incomplete conformal contact. No C-H signatures are observed in the air-Gr/sapphire SFG spectrum confirming a clean graphene surface. When PDMS lens is brought in contact with graphene-coated sapphire (Figure 3b), we observe s-CH 3 (∼2910 cm −1 ) and as-CH 3 (∼2965 cm −1 ) signatures attributed to PDMS in the C-H vibrational region (2750-3100 cm −1 ) confirming that we are indeed probing the contact interface. 55 Interestingly, the shifted sapphire OH peak position for graphene-coated sapphire in air (3643±8 cm −1 ) and in contact with PDMS (3638±4 cm −1 ) are similar (t-test, p-value=0.22), suggesting that the sapphire OH is not influenced by presence of PDMS (Figure S5, †ESI). Additionally, the relative amplitude (A q ) of sapphire free OH peak (∼3708 cm −1 ) to the shifted sapphire OH peak (∼3640 cm −1 ) does not vary (t-test, p-value=0.40) for graphene-coated sapphire in contact with air (0.25±0.06) and PDMS (0.22±0.06) across multiple repeats. The screening effect of graphene is much more evident when comparisons are made to the PDMS-sapphire contact spectrum, which shows a significantly higher shifted sapphire OH peak (3606±17 cm −1 ) due to strong acid-base interactions between sapphire OHs and siloxane groups of PDMS. 37,39 The low frequency shift for graphene-coated sapphire (70±5 cm −1 ) compared to bare sapphire (106±19 cm −1 ) in contact with PDMS illustrates screening of acid-base interactions between sapphire OHs and Si-O-Si groups of PDMS by monolayer graphene. The relative intensity of s-CH 3 and as-CH 3 peaks of PDMS differs across PDMS-Gr/sapphire and PDMS-sapphire, which could be due to differences in orientation of methyl groups at the two interfaces. ## Discussion Our study clearly illustrates that a single layer graphene is sufficient to screen acid-base interactions of the underlying substrate (SiO 2 /Si and sapphire), an unresolved topic in the past decade. The screening of acid-base interactions from the underlying substrate by a single layer graphene is evidenced by comparisons of adhesion behavior, especially during unloading, and spectroscopic shifts for bare and graphene-coated substrates. Our experimentally measured W a between PDMS and graphene deposited on two different substrates (46-59 mJ/m 2 ) from loading is in agreement with PDMS-graphene W a reported in literature (41-44 mJ/m 2 , calculated using polar and dispersive surface energy components of contacting surfaces derived from contact angle measurements). 47,49 However, similarity in the loading W a values for bare and graphene-coated substrates makes it challenging to interpret the transparency of graphene from loading behavior. The screening ability of monolayer graphene, especially to acid-base interactions, is more evident from the unloading behavior, i.e., the significant reduction in the acid-base interactions-driven adhesion hysteresis for graphene-coated substrates relative to bare substrates. For instance, the addition of a single layer graphene onto during approach (open markers) and retraction (solid markers), where a 3 vs. load (F) data is plotted for 1.9 MPa PDMS lenses on Gr/SiO 2 /Si (pink circles) and SiO 2 /Si (black triangles) substrates. (c) Adhesion measurements during approach (open markers) and retraction (solid markers), where a 3 vs. load (F) data is plotted for 1.9 MPa PDMS lenses on Gr/sapphire (red circles) and sapphire (blue squares) substrates. (d) Comparison of mean W a values calculated from loading (by fitting the loading data using Equation 1) and pull-off (calculated using Equation 3) for uncoated and graphene-coated SiO 2 /Si and sapphire substrates. Error bars indicate ±1 standard deviation. 4). Insets show schematics of the experimental geometry used for probing air-Gr/sapphire and PDMS-Gr/sapphire contact interfaces. SiO 2 /Si (or sapphire) substrate reduces adhesion hysteresis from ∼300 mJ/m 2 (or ∼270 mJ/m 2 ) to ∼20 mJ/m 2 (or ∼40 mJ/m 2 ). This small difference in adhesion hysteresis between graphenecoated SiO 2 /Si and sapphire could be due to either a subtle difference between the two underlying substrates (SiO 2 /Si and sapphire) or the quality of the graphene coating (wrinkles/folds) transferred on the two substrates. The large adhesion hysteresis observed for PDMS lenses in contact with oxide surfaces (such as SiO 2 and sapphire) has been reported previously in the literature. 43,44,56,57 Choi et al. also showed that the W a during approach is not different for the adhesion of a PDMS lens to surfaces functionalized with selfassembled monolayers having varying concentrations of polar OH headgroups. 58 However, in their experiments, the adhesion hysteresis increased with an increase in concentration of surface OH groups. PDMS is a hydrophobic polymer with its surface primarily covered by methyl (CH 3 ) groups. 59 It has been proposed that the PDMS chains near the interface rearrange to allow formation of acid-base interactions between the surface hydroxyl groups and the siloxane repeat units, resulting in high adhesion hysteresis or higher values of strain energy release rate (G) from unloading than loading (Figure S2, †ESI). 43 The addition of a single layer of graphene dramatically reduces adhesion hysteresis, thus providing direct evidence of the disruption of acid-base interactions between the surface Si-OH (or Al-OH) groups and siloxane groups of PDMS. These results are consistent with the hypothesis that during approach, the interactions of PDMS with polar substrates are dominated by vdW interactions. The influence of strong acid-base interactions between sapphire and PDMS is also manifested in the high frequency shift of sapphire OH for bare sapphire (106±19 cm −1 ) in contact with PDMS. This shift is significantly higher than the expected frequency shift based on solely vdW interactions (∼20-30 cm −1 ). 36,37 By multiplying the number density of sapphire OH (n) and the enthalpy of PDMS-sapphire interactions (∆H, calculated using the frequency shift via the Badger-Bauer equation), i.e., n*∆H, we calculate the PDMS-sapphire work of adhesion to be 70±12 mJ/m 2 . 29,36,39,40,60 The contribution of acid-base interactions to the work of adhesion between PDMS and silica gel has been previously determined by measuring the heat of adsorption using calorimetry (16 mJ/m 2 ). 58 The W a calculated by adding the loading W a (52 mJ/m 2 , dominated by vdW interactions) and the acid-base W a (16 mJ/m 2 ) is similar to the value estimated from our spectroscopy experiments (70±12 mJ/m 2 ). However, this W a is much smaller than the measured W a value from unloading (∼320 mJ/m 2 ). The experimentally measured high W a (or G) from unloading is due to the energy dissipated both in stretching of polymer chains at the interface (Lake-Thomas effect) and in breaking acid-base interactions (including hydrogen bonds) during crack propagation (unloading). 58 The energy dissipated in stretching of chains has been shown to be a strong function of velocity. 61 Furthermore, even a velocity as low as 60 nm/s (during unloading cycle) is not sufficient to eliminate the energy dissipated in chain stretching in these experiments. 62 Addition of a single layer of graphene eliminates the contribution of the acid-base interactions (as reflected in the lower frequency shift of sapphire OH for graphene-coated sapphire than bare sapphire) to G, resulting in a low adhesion hysteresis. The screening of acid-base interactions by a single layer graphene is also discernible from similar frequency shifts of sapphire OH for graphene-coated sapphire in contact with air and PDMS. It is important to underline that our approach used to determine screening ability of graphene is based on a direct spectro-scopic method while minimizing the influence of adsorbed contaminants and non-conformality of graphene to the underlying substrate. 6,10,12,13 This approach overcomes the limitations inherent to the widely reported contact angle measurements method (to determine the influence of a single graphene layer). Thus, we can conclude that interactions of graphene with other media are not fully governed by the underlying substrate, especially for substrates interacting via acid-base interactions. Our results have ramifications for applications of graphene as a conductive coating for flexible touch screens, chemical sensors and bio-sensors, and supercapacitors. 46, The performance of such devices can now be tuned to the interactions between graphene and other media (ions, gases, liquids, proteins) knowing there is little influence of the underlying substrate. ## Conclusions In this study, we provide molecular-level evidence of graphene's ability to screen hydrogen bonding interactions (a subset of acidbase interactions) from the underlying substrate, settling a debate that has remained unresolved for a decade. Using interfacesensitive spectroscopy, we demonstrate that a single layer of graphene screens acid-base interactions between oxide surfaces and PDMS. The screening of acid-base interactions is also evident in almost complete elimination of adhesion hysteresis for PDMS on graphene-coated glass and sapphire substrates. Our results have important implications in various technological applications of graphene, including flexible electronics and sensors, where graphene is typically supported on an underlying substrate. In addition, our work highlights that a single layer of graphene can be used to alter surface interactions in adhesion science. ## Wet Transfer of Graphene A 60 mm × 40 mm sample of chemically vapor deposited (CVD) monolayer graphene grown on Cu foil with a sacrificial poly(methyl methacrylate) (PMMA) coating was purchased from Graphenea Inc. Both silicon wafer (SiO 2 /Si) and sapphire substrates were cleaned prior to graphene transfer. The silicon wafers were cleaned using a Piranha solution with 3:7 ratio of 30% hydrogen peroxide (H 2 O 2 ) and concentrated sulfuric acid (H 2 SO 4 ). Caution should be exercised while handling Piranha solution as the reaction is highly exothermic. Afterwards, the silicon wafers were rinsed with copious amount of deionized water. The sapphire substrates were cleaned using sequential sonication with toluene, chloroform, acetone, ethanol, and water for at least 1 h each. Just before using, the silicon wafers and sapphire substrates were dried with nitrogen and plasma sterilized for 5 min (Harrick Plasma PDC-32G). Before transferring the monolayer graphene onto the clean silicon wafer (or sapphire plate), the partially removed bottom layer of graphene on Cu foil was etched using a solution of 8:1:1 water, H 2 O 2 , and concentrated hydrochloric acid (HCl) (following recommendation from Graphenea Inc). Afterwards, the Cu foil was etched using a 0.1 M ammonium persulfate solution for 12 h. 50 Once the Cu foil was completely etched out, the PMMA/graphene film was first rinsed thrice with ultrapure water (Millipore filtration system with a resistivity of 18.2 MΩ•cm) and then transferred onto the clean SiO 2 /Si (or sapphire) substrate. The sample was allowed to dry in ambient atmosphere overnight and then vacuum dried at 140 • C for 3 h to improve the adhesion between graphene and underlying substrate, and to facilitate PMMA removal using dissolution with acetone for 12 h. 50 The transferred monolayer graphene on SiO 2 /Si (or sapphire) was characterized using Raman spectroscopy and atomic force microscopy (AFM) to check for defects and conformality to the substrate. ## Graphene Characterization Raman spectrum of the graphene transferred on the SiO 2 /Si (or sapphire) substrate was obtained using the Renishaw InVia Raman microspectrometer with a 514 nm excitation laser and 50× objective lens to evaluate the quality of graphene. Multiple Raman spectra were collected at different spots on the sample to check for spot-by-spot spectral variance. Additionally, topographic images of the transferred CVD graphene on bare and graphene-coated SiO 2 /Si and sapphire substrates were collected using AFM (Bruker Dimension Icon) in the non-contact mode to evaluate the conformality of graphene to the underlying substrate. Multiple spots on different samples were tested in each category to calculate the root mean square (RMS) roughness of bare and graphene-coated substrates. ## Graphene Surface Cleaning Since it has been well established that it is hard to get rid of all of PMMA with acetone, the graphene supported on SiO 2 /Si (or sapphire) substrate was subjected to thermal annealing at 450 • C for at least 1 h under a 10:1 flow of Argon (Ar): Hydrogen (H 2 ), where the Ar flow rate was set to 1 L/min to remove any remaining contaminants. 10,50 The sample was allowed to cool down to room temperature under an inert atmosphere of Ar. To prevent further adsorption of contaminants, the samples were either immediately used for adhesion measurements or stored in ultrapure water and dried with nitrogen just before adhesion measurements. 10,26,66 ## Preparation of PDMS Hemispherical Lenses Optically transparent hemispherical PDMS lenses of radius 1-2 mm diameter with two different moduli (0.7 and 1.9 MPa) were prepared using the protocol described by Dalvi et al. 53 To minimize the influence of uncross-linked PDMS chains on the adhesion results, the PDMS lenses were Soxhlet extracted using toluene for 2 days. Before performing adhesion measurements, the PDMS hemispheres were tested for inherent hysteresis by testing them against a low-surface energy octadecyltrichlorosilane (OTS) monolayer coated silicon wafer (water contact angle ∼110 • ). The adhesion results with OTS-PDMS demonstrated negligible adhesion hysteresis (difference in the works of adhesion calculated from approach and retraction) due to viscoelasticity (Figure S4, †ESI), thus confirming the removal of any uncrosslinked oligomers. ## Adhesion Measurements A custom-built JKR apparatus, described in previous publications, 52,53 was used to measure the work of adhesion (Figure 2a). Briefly, a PDMS elastomeric lens with a height greater than 700 µm (to avoid the effect of underlying substrate) was adhered on a glass arm and then carefully brought in contact with uncoated and graphene-coated SiO 2 /Si (and sapphire) substrates mounted on the load cell with a computer-controlled high resolution Newport picomotor. After confirming contact between the PDMS lens and the sample, the system was loaded to a maximum normal force of 1 mN at a constant loading speed of 60 nm/s. After providing an equilibration time of 3 min, the system was unloaded at a constant velocity of 60 nm/s until the PDMS lens detached from the sample. During loading and unloading, force (F) and contact area were recorded simultaneously with the help of a load cell and an Olympus optical microscope, respectively. A minimum of 3 measurements were performed on each sample and at least 2 different samples were tested to obtain the average work of adhesion from the JKR analysis. ## Analysis of Adhesion Data The experimental force (F) vs. contact radius (a) curve obtained during loading was fit with the JKR equation 67 (Equation 1) using Igor Pro 8, where radius of the PDMS lens (R) is the known parameter (measured using optical microscope) and the thermodynamic work of adhesion (W a ) and the effective elastic modulus (E * ) are unknown parameters. E * can be calculated using both the modulus (E) and Poisson's ratio (ν) of PDMS elastomer and Gr/substrate, respectively (Equation 2). 68 The unloading data was difficult to fit with Equation 1, thus, we use the pull-off force (F pull−o f f ) observed during unloading to calculate the W a using Equation 3. In addition, we plot the strain energy release rate (G) as a function of contact area (Figure S2, †ESI) to express our experimental contact data from the viewpoint of fracture mechanics. The value of G (or W a ) at each point during loading and unloading was obtained by solving the Equation 1 using experimentally measured values of F, R, and a, and value of E * obtained from fit of loading data . ## Cleaning Procedure Equilateral sapphire prisms (Meller Optics Inc.) were first baked at 760 • C in a quartz tube furnace for at least 3-4 h. Afterwards, the prisms were sonicated (Branson 1510 Utrasonic Cleaner) in a series of solvents including toluene, chloroform, acetone, ethanol, and ultrapure water for at least 1 h per solvent to remove any adsorbed contaminants. The prisms were dried with nitrogen and plasma sterilized for 5 min before either assembling onto the SFG contact cell (using a Teflon spacer) or graphene transfer by the wet transfer approach. 50 The graphene-coated sapphire prism was cleaned via thermal annealing under a 10:1 flow of Ar:H 2 at 450 • C for at least 1 h. 10,50 Subsequently, the graphene-coated sapphire prism was cooled to room temperature under Ar atmosphere and immediately assembled onto the contact cell (cleaned by sonication with toluene and chloroform followed by atmospheric plasma treatment for 5 min just before use). ## SFG Procedure SFG spectra were acquired using a picosecond Spectra-Physics laser system, details of which have been described elsewhere. 26,52,55 Briefly, it involves the overlap of a tunable wavelength IR beam (∼3.5 µJ, 1 ps pulse width, and 1 kHz repetition rate) and a fixed 800 nm wavelength visible beam (∼70 µJ, 1 ps pulse width, and 1 kHz repetition rate) at the interface of interest. The resonantly-enhanced SFG signals provide information on the chemical identity and orientation of molecular species at an interface. An incidence angle of 42 • (with respect to the sapphire surface normal) for the IR beam was used to probe the air-sapphire and air-Gr/sapphire interfaces in total internal reflection (TIR) geometry, while an incidence angle of 10 • was used to investigate the PDMS-sapphire and PDMS-Gr/sapphire contact interfaces. The incidence angle of the visible laser beam was ∼1.5 • lower than that of the IR beam. Additional details on probing the contact interface using SFG can be found in previous studies. 52,55 SFG spectra were collected by scanning in C-H (2750 to 3200 cm −1 ) and O-H (3100 to 3800 cm −1 ) vibrational region using PPP polarization (p-polarized SFG, p-polarized visible, and p-polarized IR) because of intense SFG signals in this polarization. The SFG spectra presented in the current study have not been corrected for changes in Fresnel factors as a function of wavenumber 69 and were fit using a Lorentzian function 70 In equation 4, χ NR describes the non-resonant contribution, that does not change with scanning wavenumber (ω IR ). A q , Γ q , and ω q are the amplitude strength, damping constant, and resonant frequency of the qth vibrational resonance, respectively. ## Conflicts of interest There are no conflicts to declare.

chemsum

{"title": "J o u r n a l Na me Screening of Hydrogen Bonding Interactions by a Single Layer Graphene", "journal": "ChemRxiv"}

tailoring_cspbbr_3_growth_via_non-polar_solvent_choice_and_heating_method

## Abstract: This study describes an investigation of the role of non-polar solvents on the growth of cesium lead halide (CsPbX3 X = Br, I) nanoplatelets. We employed two solvents, benzyl ether (BE), and 1-octadecene (ODE), as well as two nucleation and growth mechanisms, one-pot, facilitated by microwave irradiation (MWI) based heating, and hot-injection, using conventional heating. Using BE and MWI, large mesoscale CsPbBr3 nanoplatelets were produced, whereas use of ODE produced thin small crystallites. Differences between the products were observed by optical spectroscopies, which showed first band edge absorptions consistent with thicknesses of ~ 9 nm (~15 monolayer (ML)) for the BE-CsPbBr3, and ~5 nm (~9 ML) for ODE-CsPbBr3. Both products had orthorhombic crystal structure, with the BE-CsPbBr3 revealing significant preferred orientation diffraction signals consistent with the asymmetric and two-dimensional (2D) platelet morphology. The differences in final morphology were also observed for products formed via hotinjection, with BE-CsPbBr3 showing thinner square platelets with thicknesses of ~2 ML, and ODE-CsPbBr3 showing similar morphologies and small crystallite sizes. To understand the role solvent plays in crystal growth, we studied lead plumbate precursor (PbBrn 2-n ) formation in both solvents, as well as solvent plus ligand solutions. The findings suggest that BE dissolves PbBr2 salts to a higher degree than ODE, and that this BE to precursor affinity persists during growth. ## Introduction All inorganic cesium lead halides (CsPbX3 , X = Cl, Br, I) and hybrid methylammonium lead halides (MAPbX3) are important functional materials that can be synthesized as crystals, thinfilms, and nanomaterials with varied morphologies. 5, 14 Two dimensional (2D) plates and platelets are common, with thicknesses of only a few monolayers (ML) often observed, and lengths varied from nanometers to microns, which are formed during nucleation and growth, or self-assembled via solvent and or ligand mediated interactions. 4, Studies have revealed that control over ligand types and stoichiometry, 20 as well as time and temperature of the synthesis can render various morphologies. 3,16,17, Sonication, 32 solvothermal, 17,33,34 mechanochemical, 35 and microwave irradiation (MWI) 27 have also been employed to control CsPbBr3 growth. Studies have shown that Ostwald ripening, 36,37 and long-ranged growth that includes oriented attachment of smaller building blocks (cubes, rods, etc.) renders 2D growth into nanoplatelets. 24,33,34,38 The role that solvent plays in CsPbBr3 growth has been explored, 33, and the role it plays in the dissolution and solvation of PbX2 salt into PbXn 2-n plumbate complexes. This is particularly interesting as the varied lead centered polyhedra formed will have different charges, molecular weights, solubility, as well as concentration and reactivity. The dynamic equilibrium between plumbates is sensitive to the Lewis basicity of the solvents and ligands involved, as well as temperature. To date, studies have focused primarily on polar solvents and their role in plumbate formation, with nitrogen or sulfur bearing solvents (e.g., DMF, DMSO) acting as stabilizers of PbXn 2-n . 43 In this report, we study whether non-polar, high boiling point solvents can tune CsPbBr3 growth at both low and high temperatures, with the latter being used to compare a one-pot growth mechanism, facilitated by microwave irradiation (MWI), and a hot-injection mechanism, via convection. The findings indicate that these solvents do tune PbXn 2-n formation, leading to controlled CsPbBr3 growth and properties. ## Experimental Chemicals: Lead iodide (PbI2, 99%), lead bromide (PbBr2, 99.99%), cesium carbonate (Cs2CO3, 97%), methyl ammonium iodide (MAI, 98%), oleyl amine (OAm, 70%, technical grade), oleic acid (OAc, 90%), 1-octadecene (ODE, 90%), benzyl ether (BE, 98%), N,N-Dimethylformamide (DMF, anhydrous, 99.8%) were purchased from Sigma Aldrich and used as received unless otherwise noted. Precursor Preparation: In a typical one-pot reaction, precursors were first prepared by adding 0.80 g of Cs2CO3 to 30 mL of solvent (ODE or BE), along with 2.4 mL of OAc and the mixture was heated at 120°C under vacuum until all was dissolved (0.15 M Cs-OAc). Likewise, 0.28 g of PbBr2 or 0.34 g of PbI2 in 20 mL solvent was heated at 120°C for 1 h under vacuum. Next, 2 mL of OAc and 2 mL OAm was added to the mixture under Ar and the mixture was heated at 120°C for another hour for a complete dispersion of a 0.03 M PbX2 stock solution. Trials were also run using purified OAm, 44 as were reactions where Cs2CO3 was reacted with OAc at higher ratios. 45 One-Pot MWI Heating and Synthesis of CsPbX3 : In a typical one-pot MWI heated reaction, 3 mL of the as-prepared PbX2 precursor was purged with Ar gas for 5 min at room temperature. Next, 200 μL of the cesium oleate precursor solution was heated to 85 o C and injected to PbX2 solution followed immediately by MWI heating to 160 °C where the trajectory of the heat transfer was different between BE and ODE, and depends on the dielectric constant (ε) of the non-polar solvents. 46 The BE reaction (ε = 3.82) reached 160 ℃ within 100 seconds, while the ODE reaction (ε = 2.25) reached after 240 s. Upon reaching this set point, there was a rapid quenching of temperature by the active cooling of the MWI reactor. These prepared nanoplatelets were purified by centrifuging 1 ml aliquots at 10,000 rpm for 3 min, followed by the discarding of the supernatant and redispersion in toluene aided by sonication. This procedure was repeated at least two times, except in the case of FTIR sample preparation, in which additional purification steps were used. Hot Injection and Conventional Heating Based Synthesis of CsPbX3. First, a 0.15 M Cs-OAc precursor was prepared as described in the precursor preparation section and was stored at 85 o C prior to injection. To prepare the PbBr2 precursor, a mixture of 0.14 g PbBr2 powder and 10 mL BE was heated under vacuum at 120 o C for 1h. Then, the mixture was placed under Ar and 1 mL OAm and 1 mL OAc were injected to dissolve the powder. After 1 h, temperature was raised to 140 o C. Next, 0.8 mL Cs-OAc was injected to the solution, and the reactions were let to anneal for 1, 30, and 60 min before removing the heating mantle to quench the reaction. The products were purified as described above. This method was also used for the room temperature syntheses described, except that all reactants were cooled to room temperature before initiation. ## BE-CsPbI3 synthesis via Halide Exchange of BE-CsPbBr3: Prior to the halide exchange (HE), BE-CsPbBr3 were purified and redispersed in toluene to the approximated concentrations. 47 This solution was then combined with an aliquot from a 0.20M PbI2 stock solution so that the combined [Iˉ]:[Brˉ] = 1, following a method recently developed in our lab. 48 The mixture was allowed to react for 30 min, then PL emission was measured. For XRD measurement, a portion of the mixture was separated, purified and drop-casted on a zero-diffraction quartz substrate. The rest of the mixture was centrifuged and redispersed in toluene to the initial volume to remove the free Brin the solution. In the second step, PbI2 with the same ratio of 1:1 was added to the remaining solution, followed by XRD and PL measurements. Benesi-Hildebrand Analysis: The competitive assay analysis was performed by first preparing a 0.1 M MAI in BE. Next, MAI aliquots were added to a diluted BE-PbI2 at a ratio of [MAI]:[PbI2] = 1-35. The resulting Iodoplumbate complex formation was monitored by UV-vis spectroscopy. ## Instrumentation Microwave irradiation heating (MWI) was performed using a Discover-SP microwave synthesizer (CEM Inc.) with magnetron frequency of 2450 MHz, where temperature, power and time were controlled by Synergy software. Each reaction was stirred in a 10 mL glass vial during which temperature was monitored by a IR sensor. To stop the reaction, samples were cooled down by compressed N2 circulating inside the microwave chamber. The optical characterization was performed on a Cary 50 Bio UV-vis spectrophotometer (Varian Inc.), and photoluminescence spectroscopy was performed on a Cary Eclipse fluorescence spectrophotometer (Varian Inc.). The excitation wavelength was 400 nm. The powder X-ray diffraction (XRD) was performed using D2 PHASER XRD (Bruker Inc.) with a Cu radiation source. Samples were prepared by drop-casting purified products on a zero-diffraction quartz holder. Transmission electron microscopy (TEM) was performed on either a JEM 2100F or JEM 1400 (JEOL Inc.), operated at 200 or 120kV, respectively. Samples were drop cast from toluene dispersions onto carbon coated copper grids. Atomic Force Microscopy (AFM) analysis was performed on an Innova SPM (Bruker Inc.) in tapping mode using samples deposited on a freshly cleaved HOPG grid. Finally, Fourier Transform Infrared Spectroscopy (FTIR) spectra was collected using a Thermo Nicolet 6700 FTIR equipped with a diamond smart iTR attenuated internal reflectance accessory, and a liquid N2 cooled MCT-A detector. Samples were prepared by three rounds of purification in toluene with centrifugation at 4k rpm for 10 minutes in a glass tube. Samples were then drop-casted on the iTR diamond and air-dried prior to the measurement. ## Results and Discussion Figure 1a shows a schematic illustration of the synthetic conditioned tailored to understand CsPbBr3 growth using either benzyl ether (BE) or 1-octadecene (ODE) as non-polar hgh boiling point solvents, cesium carbonate dissolved and complexed with oleic acid (OAc) and lead plumbate (PbBrn 2-n ) 42, formed via dissolving in solvent (S) and ligands (L), as described in the experimental section. The products of these reactions are denoted as BE-CsPbBr3 or ODE-CsPbBr3. Two nucleation and growth mechanisms were studied, so-called one-pot and hotinjection, at either high temperature or low temperature. purification, the resulting optoelectronic properties were measured. Figure 1b shows the UVvisible absorption (UV-vis) results. The BE-CsPbBr3 had a first excitonic absorption peak at 523 nm (i), while the ODE-CsPbBr3 absorbed at 488 nm. Both products exhibited photoluminescence (PL), as shown in Figure 1c, with BE-CsPbBr3 emitting at 527 nm (i), and ODE-CsPbBr3 at 495 nm (ii). The red-shift indicates either larger sizes or thicker platelets for BE-CsPbBr3. The band edge absorption is quantized by the minimum CsPbBr3 dimension, especially when sizes are comparable to the exciton Bohr radius (aB), which for CsPbBr3 is ~ 7 nm. Using the studies reported by others, 19,22,24,52 we estimate a thickness of >15 monolayers (ML), where ML is defined a linear chain of corner sharing PbBr6 4octahedra with thickness of 0.59 nm, for the BE-CsPbBr3, corresponding to approximate thickness of 8 ~ 9 nm. We note that the absorption is broad, indicating a polydisperse sample and distribution in thicknesses. Using the same approach, the ODE-CsPbBr3 would have a minimum feature size of ~ 9 ML, or ~5.3 nm. The dimensions and morphology of the two products were characterized using transmission electron microscopy (TEM). Figure 2a shows a micrograph for the ODE-CsPbBr3 products, which has small square like lateral morphology with an length of ~ 9.7 ± 0.8 nm, which combined with the optical data above is ~5 nm thick, and is consistent with other ODE based CsPbBr3 platelets. In contrast, the BE-CsPbBr3 products had larger 2D platelet like morphology, as shown in Figure 2b-e, and supporting Figures S1. The large platelet morphology of the products was highly reproducible; however, the lateral dimensions were polydisperse. For example, the platelets largest length varies from 20 -500 nm. The platelets were indeed thin, as suggested by the UV-vis, as illustrated in the TEM micrographs where the platelets orient on top of another. The thickness was also probed by atomic force microscopy (AFM). Figure S2 shows a typical AFM image of BE-CsPbBr3 platelets dropcast from a toluene dispersion onto a HOPG grid. At the resolution shown, domains consisting of many platelets are imaged. Importantly, sharp edges can be observed, and cross-section analysis reveals overall thickness profiles of the domains (Fig. S2). The heights measured vary slightly, from 3-5 nm, which are thinner than the ML estimate from the main absorption band above, suggesting that thinner platelets were sampled. We note that some products dispersions showed significant aggregation, and after drop casting, domains revealed stacks or clusters of platelets that revealed larger heights. To understand this, scanning electron microscopy (SEM) was used to image those dropcast substrates. Figure S3 shows the platelets and the 2D morphology, as well as grouped discrete aggregates, which were difficult to separate, and was attributed to either residue BE in the purified product, or the result of excess purification steps, as described below. The crystalline nature of the CsPbBr3 products were studied by powder X-ray diffraction (XRD). Figure 3 shows the XRD analysis of the BE-(i) and ODE-CsPbBr3 (ii), as compared to an orthorhombic CsPbBr3 bulk standard. Both products index with the orthorhombic standard with minimal variation to Bragg angles, but intensity ratios differed, as did the extent of Scherrer broadening. For example, the BE-CsPbBr3 (i) showed pronounced preferred orientation of the planes diffracting at 2 = 30.4 and 30.7 • , which correspond to (004) and (220) of the crystal. Clearly, the intensities do not match the standard, and suggest not only that each platelet grows in the same orientation, but that each platelet is highly crystalline. It's important to note that preferred orientation in XRD can be a result of crystal growth, as well as an artifact produced by the way a sample self-assembles during drying, as well as by substrate type and sample-to-substrate interactions. 53 We suspect that each of these factors influence the XRD shown here. The samples were prepared via drop-casting from a concentrated solution, and we assume they form into the irregular clusters shown by SEM (Fig. S3). Nonetheless, a number of control experiments were performed to better understand the peak intensities and the origin of the preferred orientation. To test whether the intensity ratios could be an artifact of platelet drying on the XRD substrate, samples that were both drop cast from solution, and from dried powders, each of which showed similar intensities. In another control, a concentrated carbon black slurry with colloidal carbon (~20 nm) was added to a toluene solution of purified BE-CsPbBr3, sonicated, and dropcast, with the aim of using the carbon to inhibit platelet stacking during drying. This however resulted in similar XRD signatures (Fig. S4), suggesting that crystal orientation plays at least some role, and that growth occurs in the (004) and ( 220) directions of the platelets, of which (220) can be indexed to the longer dimension of the platelet. The thickness of the platelets, and thus the planes of atoms in that direction, are outweighed in terms of number, and have lower Bragg intensities. The ODE-CsPbBr3 on the other hand (ii), had intensity ratios consistent with that of bulk, as well as broadening consistent with the smaller dimensions. The halide concentration of the platelets could also be fine-tuned, either by introducing iodide (I) into the synthetic solutions, or via halide exchange (HE), resulting in BE-CsPbBr3-xIx. Here, we focused on only the BE solvent, since there are numerous examples of halide control in ODEbased products. 54 Figure S5 shows the PL for BE-CsPbBr3-xIx, synthesized by varying the [Br -]:[I -] feed ratio, with the corresponding XRD shown in Figure S6. The XRD signatures were consistent with those platelet morphologies shown above and increased I-rich content is indicated by shifts in 2 Products formed at low Iconcentrations showed the platelet like preferred orientation, however this was lost at high Icontent. Alternatively, HE could be used to transform the BE-CsPbBr3 via addition of Irich precursors 54 or small organohalide molecules. 55,56 Using a protocol recently developed in our lab, 48 we found that the BE-CsPbBr3 platelets could undergo HE without disrupting the crystal structure, and allowed for a broader control of composition than direct synthesis (Fig. S7). A more detailed study of the synthesis of mixed halides and of halide exchange in BE-CsPbBr3 are beyond the scope of this paper and will be reported elsewhere. The novel component of this study is understanding the role of BE in the formation of CsPbBr3, and we next prepared BE-CsPbBr3 not with MWI heating, but and instead via hot-injection. In contrast to MWI based heating, in which all precursors are in 'one-pot' and growth is facilitated or activated by heating, 'hot-injection' introduces the final precursor at an elevated temperature, inducing burst nucleation and growth of what is typically a smaller and more monodispersed product. Figure 4 shows a set of TEM micrographs for BE-CsPbBr3 products collected after hot injection and ~1 min annealing at 140 o C (a-c), and after annealing for 30 min (d). One observation made after synthesis was that the product had soluble and insoluble fractions after 1 min (Fig. S8a). A TEM of the soluble portion is shown in Figure 4a, with small square crystals with edge lengths of l = 6.8 ± 1.1 nm visible. The insoluble portion (5b-c) shows larger square platelets with edge lengths of l = 11.1 ± 3.5 nm. Both fractions showed smaller clusters or nuclei with diameters of d ~3.5 nm and very uniform inter-cluster distances, see arrow. Based on the optical signature, which showed a band edge absorption at 500 (a-c) and 510 nm (d), these had thicknesses of ~2 ML, respectively (Fig. S8), making them much thinner than the MWI based products. Figure 5 shows the powder XRD for BE-CsPbBr3 soluble (i) and insoluble (ii) products from hotinjection, as compared to ODE-CsPbBr3 (iii). Compared to the MWI BE-CsPbBr3 products (Fig. 4), these showed more cubic crystal characteristics, with the insoluble products (ii) showing some preferred orientation. The ODE-CsPbBr3 similarly showed cubic similarities, but with slight 2 shifts that may suggest some orthorhombic features. We hypothesize that the reason why BE and ODE produce different morphologies is due to the different fractions of precursor types at the time of nucleation, as well the affinity of the solvent to Pb 2+ . Figure 6 shows the electronic absorption of PbBr2 solid dissolved in BE (i) and ODE (iv), and mixtures of BE+OAm+OAc (ii), as well as ODE+OAm+OAc (iii). The absorptions observed are categorized broadly as exfoliated PbBr2 solids, which may be two dimensional in nature, 50 and multiple PbBrn 2-n lead plumbate complexes which are often defined as PbBr6 2-, PbBr3 -, and PbBr4 2 . 49,51 Comparing (i) to (iv) reveals that BE is more effective at dissolving PbBr2 to PbBrn 2-n than ODE. This was also physically observed in the experiment, where BE dissolved the PbBr2 salt with less visible solids than ODE, in which a high percentage of insoluble salt was still observed. Secondly, the addition of OAm and OAc further dissolves the solids in the case of ODE and shifts the absorption wavelengths in BE. Both solvent plus ligand mixtures show high concentrations of PbBrn 2-n , likely the result of ligand coordination to Pb 2+ and substitution of one or more bromides (see below), with ODE+OAm+OAc still showing a considerable percentage of insoluble PbBr2. Whether or not this solid PbBr2 is incorporated into forming the nano CsPbBr3 is likely dependent on the effect of the synthetic temperature on dissolution equilibrium. This insight suggests that BE coordinates and solvates PbBr2 better than ODE, which is understandable considering it's -rich nature. Researchers have studied the solvent effect of perovskite formation previously, especially as it relates to thin film formations of methyl ammonium halides (MAPbX3) using polar solvents. 42 And, we note that BE has been used in nanoparticle synthesis before, especially in the recent synthesis of Qdot heterostructured libraries. 57,58 By using a competitive assay between solvents and halides, a Benesi-Hildebrand analysis can be used to approximate coordination strength by way of estimating equilibrium constant (K) and Guttmann donor number (DN), which measures PbBrn 2-n concentration and type in the presence of excess halides by way of UV-vis. 42,43,49,59 For instance, Loo and co-workers compared solvent dielectric constants and DN with either crystal growth or thin-film growth mechanisms, and showed that while ε values did not predict growth, that DN > 15 consistently resulted in thin film growth, whereas higher numbers consistently showed crystal growth. 42 Thus, higher DN solvents, typical polar and strong Lewis bases, coordinate favorably with Pb 2+ , resulting in PbBrn 2-n with higher n, resulting in more crystal growth (ideal building blocks), whereas DN < 15, coordinate weakly with Pb 2+ , resulting in lower n, and more amorphous or thin-film growth. 42 To date, most analysis is compared to polar solvents, like DMF and DMSO, both of which bind strongly to Pb 2+ and decrease plumbate equilibrium. 42,43,49 Both BE and ODE are considerably less polar that many of these solvents used for perovskite growth. Figure S9 than ODE, but much weaker than DMSO (KDMSO ~10, Fig. S9). The composition of the final organic capping monolayer of BE-CsPbBr3 was also studied via Fourier Transform Infrared (FTIR) spectroscopy, and shown in Figure S10. Vibrations attributed to BE adsorbed to the CsPbBr3 interface were consistently observed, further suggesting coordination. While the DN and K values aid in understanding the PbBr2 dissolution, the cesium oleate precursor is also important, and can have different temperature dependent and stoichiometry related solubility. Control experiments fully solubilizing cesium oleate (Cs + -OAc) at room temperature were performed, which used high OAc-to-Cs molar ratios, following a method recently described. 45 The products of that control synthesis using MWI had more soluble final products, but the platelet morphology and XRD intensity ratios persisted (Fig. S11). This Cs precursor was also used in the hot-injection synthesis described above. Considering the procedural steps employed in this study, and the findings above, Figure 8 idealizes the mechanism for CsPbBr3 growth. The dissolution of PbBr2 salt (a) in a solvent (S = BE or ODE) produces two intermediates, two dimensional, exfoliated (PbBr2)x solid layers solvated by S, as described recently, 50 and the PbBrn 2-n plumbates of various coordination, such as PbBr4 2-, PbBr3 -, etc. 51,60 Here, the PbBrn 2-n may have a Brsubstituted by S, which is not charged (b). Upon addition of ligands (L = OAm, OAc, OAm+OAc), the equilibrium shifts to forming a higher percentage of PbBrn 2-n afforded by strong L-to-Pb 2+ coordination (c), which breaks the PbBr2 into smaller fragments or lower molecular weight polyhedra. Upon the addition of Cs + (d), the PbBrn 2-n polyhedra are electrostatically attracted to one another, forming 2-nCs + PbBrn 2-n complexes, but still under the coordination of excess L and S. In this study, steps a-c (precursor preparation) occurs over the course of an hour, whereas step (d) occurs over a few minutes before heating in the case of MWI heating, or within seconds during hot-injection. Upon heating, the 2-nCs + PbBrn 2n complexes loose coordinating S as well as L and are consumed producing CsPbBr3 perovskite platelets (e). Loss of coordinating solvent during heating is often observed in the formation of perovskite thin films from polar solvents, however in this study, loss of solvent refers to those molecules that were either coordinating to the crystal or separating intermediate plumbates. It is possible that both BE DN numbers likely reside in the thin-film growth regime, as described above, resulting in large mesoscale platelets are observed with prolonged MWI heating and smaller square platelets are formed via quick hot-injection, while ODE produces smaller crystallites of similar sizes for both heating conditions. Also of importance is the temperature used, as it will influence the equilibrium between PbBrn 2-n types and CsPbBr3 crystalization in the presence of S and L, promoting PbBrn 2-n at lower temperatures. Interestingly, if the reaction is held at step d for long periods of time (days) at room temperature (f), then differences between BE and ODE can also be observed. Kinetically, the BE-CsPbBr3 formed slower and resulted in thinner CsPbBr3 (Fig. S12), where smaller crystallites are formed that have well defined inter-crystal distances, which we attribute to repulsion from coatings of charged PbBrn 2-n at the interface. A TEM image of these is shown in Figure S13. The slower kinetics and smaller crystal size (d ~ 3 nm) in the case of BE at room temperature again suggests strong coordination to PbBrn 2-n , the release of which is more sensitive to temperature. ODE on the other hand formed uniform rod like structures with lengths < 15 nm. This final point suggests that judicious selection of both solvent as well as modest temperature changes may allow for a wealth of morphologies to be formed and controlled, which is part of our ongoing work and will be reported elsewhere. ## Conclusion Taken together, a synthesis route for CsPbBr3 nanoplatelets has been described in which choice of non-polar solvents and heating method can be used to control morphology. The findings demonstrate that combining BE, a one-pot mechanism, and MWI heating prove effective at influencing nucleation and growth to the point of forming highly crystalline platelets, with lateral dimensions of 20-500 nm, and relatively thick, ~15 ML, thicknesses. These platelets show preferred orientation in XRD signatures along the (220) and (004) planes. Synthesis via hotinjection with BE also leads to platelets, but a more uniform square shapes, ~17 nm lengths, and ~2 ML thicknesses. On the contrary, use of ODE results in small crystallites, ~10 nm, in both heating approaches. The ability of the solvent, and solvent plus ligand mixtures to dissolve PbBr2 salt into varied PbBrn 2-n plumbates was studied, and showed that BE is more effective, due in large part to its -donating character and coordination to Pb 2+ . The compositions of the BE-CsPbBr3 could be tailored by adding iodine either via synthesis upon addition of PbIn 2-n during synthesis, or via halide exchange. ## Associated Content The Supplemental Information is available free of charge. Experimental details, control studies, and supplemental Figures S1-S13

chemsum

{"title": "Tailoring CsPbBr 3 Growth Via Non-Polar Solvent Choice and Heating Method", "journal": "ChemRxiv"}

atom-economic_generation_of_gold_carbenes:_gold-catalyzed_formal_[3+2]_cycloaddition_between_ynamide

## Abstract: The generation of gold carbenes via the gold-catalyzed intermolecular reaction of nucleophiles containing relatively labile N-O or N-N bonds with alkynes has received considerable attention during recent years.However, this protocol is not atom-economic as the reaction produces a stoichiometric amount of pyridine or quinoline waste, the cleaved part of the N-O or N-N bonds. In this article, we disclose an unprecedented gold-catalyzed formal [3+2] cycloaddition between ynamides and isoxazoles, allowing rapid and practical access to a wide range of synthetically-useful 2-aminopyrroles. Most importantly, mechanistic studies and theoretical calculations revealed that this reaction presumably proceeds via an a-imino gold carbene pathway, thus providing a strategically novel, atom-economic route to the generation of gold carbenes. Other significant features of this approach include the use of readilyavailable starting materials, high flexibility, simple procedure, mild reaction conditions, and in particular, no need to exclude moisture or air ("open flask"). ## Introduction Catalytic transformations involving gold carbenes are arguably the most important aspect of homogeneous gold catalysis. 1 Recently, the possibility of forming an a-oxo gold carbenoid species via gold-catalyzed intramolecular or intermolecular alkyne oxidation by a N-O bond oxidant (initially a sulfoxide), pioneered by Toste and Zhang,2a,b represents a signifcant advance in gold carbene chemistry, and various efficient synthetic methods have been developed based on this strategy. 2 Compared with intramolecular alkyne oxidation, the intermolecular approach offers much greater flexibility as no tethering of the oxidant is required, and therefore it is more synthetically useful. 3 However, this intermolecular approach is obviously not atom-economic as the reaction produces a stoichiometric amount of pyridines or quinolines, the reduced form of the corresponding pyridine N-oxides or quinoline N-oxides, as waste (eqn (1)), 4 which may even deactivate the gold catalyst via coordination. 5 (1) Access to the related a-imino gold carbenes via gold-catalyzed nitrene transfer to alkynes, however, remains a highly challenging task. Here, it should be noted that: (1) the nitrene moiety is delivered via an outer sphere attack and no gold nitrene complex 6 is involved in this case; this mode of nitrene transfer is distinctively different from many well-established nitrene transfer reactions; 7 (2) this protocol would present alkynes as equivalents of a-diazo imines, which are difficult to access as a-diazo imines can readily cyclize into the corresponding 1,2,3-triazoles. To date, only limited success has been achieved in this type of gold-catalyzed nitrene transfer, mainly by the intramolecular reaction of alkyne and azide. 8 For example, Toste and co-workers used azide as an effective nitrene equivalent and realized the frst protocol for the generation of a-imino gold carbenes in 2005. 8a Later, elegant studies on the synthesis of indoles from alkynyl azides were demonstrated by Gagosz 8c and Zhang, 8d independently. Recently, several studies have invoked the intermolecular transfer of nitrene to alkynes by the use of iminopyridinium ylides as nitrene-transfer reagents, as disclosed by the groups of Zhang, 9a Davies, 9b,c and Liu. 9d However, similar to those of the above-mentioned goldcatalyzed intermolecular alkyne oxidations, a stoichiometric amount of pyridine was produced as the waste in these cases. Therefore, the exploration of intermolecular approaches to the generation of a-imino gold carbenes, especially in an atomeconomic way, is very attractive to researchers. We envisioned that the a-imino gold carbene intermediate B might be generated through the gold-catalyzed intermolecular reaction of ynamides 1 10 with isoxazoles 2, which could be obtained in an efficient and modular manner following the synthetic routes shown in eqn ( 3) and ( 4) in Scheme 1. 11 The carbene B, likely highly electrophilic, could then undergo an electrophilic cyclization to yield the fnal 2-aminopyrroles 3, thus constituting a gold-catalyzed formal [3+2] cycloaddition (Scheme 1, eqn (2)). Herein, we report the successful implementation of this mechanistic design to a facile and practical synthesis of a wide range of polysubstituted 2-aminopyrroles, which are common structural motifs found in natural products and pharmacologically active molecules (Fig. 1) 12 and are difficult to access via traditional methods for pyrrole synthesis. 13 Most importantly, an a-imino gold carbene is most likely generated as the key intermediate on the basis of both mechanistic studies and theoretical calculations, thereby providing a strategically-novel, atom-economic route to the generation of gold carbenes. ## Results and discussion At the outset, ynamide 1a and 3,5-dimethylisoxazole 2a were used as the reacting species and a series of experiments were performed in order to validate our approach. To our delight, the expected product 3a was indeed formed in 70% 1 H NMR yield in the presence of 5 mol% IPrAuNTf 2 (Table 1, entry 1). Then, various typical gold catalysts with a range of electronic and steric characteristics were screened (Table 1, entries 2-7), and (ArO) 3 PAuNTf 2 (Ar ¼ 2,4-di-tert-butylphenyl) gave the best yield of the desired product (Table 1, entry 7). Somewhat surprisingly, AgNTf 2 could also catalyze this reaction in 50% yield (Table 1, entry 8). Notably, without a metal catalyst, the reaction failed to give even a trace of 3a, and PtCl 2 and Zn(OTf) 2 were not effective in promoting this reaction (Table 1, entries 9-10). 14 The reaction proved to be less efficient when it was performed at a reduced temperature (Table 1, entry 11). In addition, the use of 2 equiv. of 2a also gave the desired pyrrole 3a in 90% yield (Table 1, entry 12). With the optimized reaction conditions in hand, the scope of the transformation was explored. As seen from the results collected in Table 2, the reaction proceeded smoothly with various ynamide substrates 1, and the yields ranged from 58% to 96%. For example, ynamides with different protecting groups, even the Ns group (Table 2, entries 4-5), readily gave the desired 2-aminopyrroles 3a-f (Table 2, entries 1-6). Of note, an excellent yield could be achieved in the case of an ynamide with an oxazolidinone moiety and no dimerization reaction was observed (Table 2, entry 6). 15 When R 1 is an allyl group, the Scheme 1 Synthetic design for the atom-economic generation of aimino gold carbenes: formation of 2-aminopyrroles 3 through goldcatalyzed formal [3+2] cycloaddition between ynamides 1 and isoxazoles 2. desired 3j could also be formed in 86% yield, and no cyclopropanation product was formed (Table 2, entry 10). 16,5g Other aryl-substituted ynamides were also suitable substrates for this reaction, giving the corresponding functionalized pyrroles 3k-l in excellent yields (Table 2, entries 11-12). Interestingly, for styryl or cyclopropyl-substituted ynamides, this reaction still led to 75% yield and 58% yield, respectively (Table 2, entries 13-14). The molecular structure of 3a was further confrmed by X-ray diffraction (Fig. 2). 17 We next extended the reaction to different 3,5-disubstituted isoxazoles 2. To our delight, the reaction of ynamide 1i with various isoxazole substrates 2 worked well under the above optimized reaction conditions, giving versatile polysubstituted 2-aminopyrroles 3o-z in generally good to excellent yields. As summarized in Table 3, a range of aryl-substituted isoxazoles 2c-g were successful (R 2 ¼ aryl), delivering the desired 3p-t in 72-96% yield (Table 3, entries 2-6). In addition, when R 1 is an aryl group, the reaction also worked well to afford the corresponding pyrroles 3v-w in excellent yields (Table 3, entries 8-9). Pleasingly, methyl 3-pyrrolecarboxylate 3x was formed in 90% yield from the corresponding isoxazole (Table 3, entry 10). It should be mentioned that 3-formylpyrroles 3y-z could also be prepared in serviceable yields (Table 3, entries 11-12). Interestingly, when the scope of the method was extended to fully-substituted isoxazoles 4, the reaction also proceeded well, allowing the convenient synthesis of deacylated polysubstituted 2-aminopyrroles 5. A series of readily-available substituted ynamides was frst examined. The corresponding pyrroles 5a-d were obtained in 72-85% yield (Table 4, entries 1-4). Then, isoxazoles 4 with substituents at the 4-position were also investigated, giving the products 5e-m in mostly good to excellent yields (Table 4, entries 5-13). Notably, methyl 3-pyrrolecarboxylate 5n could also be obtained in 77% yield from the corresponding 4-substituted isoxazole, which is complementary to the above protocol based on the 3,5-disubstituted isoxazoles 2 (Table 4, entry 14 vs. Table 3, entry 10). In particular, the 3,4diphenyl substituted isoxazole also reacted smoothly, delivering 4, entry 15). To further test the practicality of the current catalytic system, a gram-scale reaction of 1.36 g of 1a and 1.07 g of 2a was carried out with a much lower catalyst loading (1 mol%), and 1.72 g of the desired pyrrole 3a was formed in 85% yield, highlighting the synthetic utility of this chemistry (eqn ( 5)). Interestingly, the reaction could also be performed well even in water to afford the desired product 3a in 80% yield and no hydration of the ynamide was observed (eqn ( 6)), 10a-c thus making this protocol more practical and environmentally benign. (5) (6) This chemistry can also be used to construct N-heteropyrrolizines, which are present in a variety of bioactive molecules. 18,12k For example, treatment of ynamide 1p with isoxazole 4a under the optimized reaction conditions gave the pyrrole 5p, which could be converted into fused 2-aminopyrrole 6 in basic conditions in a one-pot process (63% two-step overall yield, eqn ( 7)). Compound 6 might serve as a precursor for the synthesis of lipoxygenase inhibitors (Fig. 1). 12k (7) The sulfonamide could be readily transformed into a free amine (Scheme 2). For example, the reaction of ynamide 1d with isoxazole 2c under the optimized reaction conditions furnished pyrrole 7 in 81% yield. Nitrogen protection of 7 with a methyl group and subsequent removal of the Ns group using the standard conditions (PhSH, K 2 CO 3 ) resulted in the formation of species 7a (53% two-step overall yield). Subsequent deprotection of the benzyl group in 7a could be realized by performing MnO 2 -mediated oxidation followed by hydrolysis to afford 7b in 56% yield. 13e To probe the mechanism of this reaction, we frst synthesized the alkyl-substituted ynamide 1q as the alkyl-substituted gold carbene is well-known in the gold-catalyzed cycloisomerizations of enynes; hydride shift followed by elimination of the gold catalyst was involved as the critical deauration step. 1e,19 Indeed, as depicted in eqn (8), when ynamide 1q reacted with 2a under the standard reaction conditions, none of the desired pyrrole was detected and a,bunsaturated amide 8 was isolated in 25% yield. Amide 8 is supposed to be derived from hydride shift followed by elimination of the gold catalyst and subsequent hydrolysis. This result indicated that a gold carbene is most likely generated as the key intermediate in this process. On the other hand, the low chemoselectivity in the case of n-butyl substituted ynamide shows the importance of aryl substituents on the ynamides to keep a high reactivity for the reactions in Tables 2-4. 20 (8) In addition, it was found that a key intermediate 3H-pyrrole 5l 0 could be detected and isolated in the case of the reaction of ynamide 1i with fully-substituted isoxazole 4i (Table 4, entry 12). To further demonstrate this process, we monitored the reaction by 1 H NMR spectroscopy, as depicted in Fig. 3. Here, the reaction was performed in the presence of 2 mol% (ArO) 3 PAuNTf 2 in CDCl 3 in order to better track the reaction intermediates. At the early stage of the reaction, we could clearly observe the formation of the 3H-pyrrole 5l 0 , which was gradually transformed into the fnal 1H-pyrrole 5l. A plausible mechanism to rationalize the formation of pyrrole In the case of fully-substituted isoxazole substrates 4, D is ultimately transformed into the fnal 1H-pyrrole 5, presumably by a water-assisted deacylative aromatization. 25 ## Conclusions In summary, we have developed a novel gold-catalyzed formal [3+2] cycloaddition between ynamides and isoxazoles, leading to the concise and flexible synthesis of polysubstituted 2-aminopyrroles. This methodology makes it possible to introduce four substituents onto a pyrrole ring very freely with high efficiency. Of particular interest, fully substituted isoxazoles also react under deacylation, closing a further gap in the reaction scope. Moreover, an a-imino gold carbene is the most likely intermediate based on both mechanistic studies and theoretical calculations, thus providing a new strategy for the generation of gold carbenes, especially in an atom-economic way. Studies to elucidate the detailed mechanism and further synthetic applications of the current protocol are in progress in our laboratory.

chemsum

{"title": "Atom-economic generation of gold carbenes: gold-catalyzed formal [3+2] cycloaddition between ynamides and isoxazoles", "journal": "Royal Society of Chemistry (RSC)"}

time_resolved_transient_circular_dichroism_spectroscopy_using_synchrotron_natural_polarisation

## Abstract: Ultraviolet (UV) synchrotron radiation circular dichroism (SRCD) spectroscopy has made an important contribution to the determination and understanding of the structure of biomolecules. In this paper, we report an innovative approach that we term time-resolved SRCD (tr-SRCD), which overcomes the limitations of current broadband UV SRCD setups. This technique allows accessing ultrafast time scales (down to nanoseconds), previously measurable only by other methods, such as infrared (IR), nuclear magnetic resonance (NMR), fluorescence and absorbance spectroscopies and small angle X-ray scattering (SAXS). The tr-SRCD setup takes advantage of the natural polarisation of the synchrotron radiation emitted by a bending magnet to record broadband UV CD faster than any current SRCD setup, improving the acquisition speed from 10 mHz to 130 Hz and the accessible temporal resolution by several orders of magnitude. We illustrate the new approach by following the isomers concentration changes of an azopeptide after a photoisomerisation. This breakthrough in SRCD spectroscopy opens up a wide range of 2 potential applications to the detailed characterisation of biological processes, such as protein folding, protein-ligand binding. ## Introduction Circular dichroism (CD) is an optical property of molecules having chiral structure(s) and/or spatially oriented arrays of chromophores. It manifests itself as a difference in absorption for left-and right-circularly polarised light. In the ultraviolet (UV) range. This feature has been exploited for decades for the characterisation of organic molecules, materials with supramolecular chirality, and in protein conformation determination, where there are distinctive spectral signatures for each secondary structure type, i.e., α-helices and β-sheets 1 . Thus, CD spectroscopy is an important biophysical tool for characterising native and modified proteins. In the biomedical context, protein misfolding can have dramatic consequences on cell physiology, causing serious neurodegenerative diseases, as found in Alzheimer's and Parkinson's 2 . Structural and kinetics studies of protein folding, using time-resolved approaches, are providing crucial insights at the molecular level into the aetiology of these diseases. There are two main ways to measure CD spectra: the ellipsometric method and the direct absorption difference detection method. The former is based on quantifying the variations of the ellipticity and azimuth orientation of a highly elliptically polarised beam passing through a dichroic sample 3 . In 2012, Eom et al. 4 innovatively adapted the ellipsometric method to a heterodyne-detection technique, providing both the CD spectrum and the optical rotation dispersion spectrum, by analysis of the phase and the amplitude of the transmitted orthogonal electric field of the incident light polarisation. More recently, Hiramatsu et al. 5 coupled the heterodyne detection technique to a singular value decomposition analysis. This improvement removes linear dichroism and linear birefringence artifacts, allowing accurate time-resolved CD (tr-CD) measurements in the visible range with sub-picosecond temporal resolution. The second way to acquire a CD spectrum is based on absorption difference measurement. The light is alternately circularly right and left polarised using an optical or acousto-optic modulator. Then the detection system records intensity variations and allows CD determination. This method has been extensively refined, and can provide sub-picosecond resolution for monochromatic measurements 6 . Hache and colleagues used this technique to probe ultra-fast kinetics in biomolecules 7 and achieved 800 fs temporal resolution 8 . Several groups 9,10 The synchrotron radiation circular dichroism (SRCD) technique was developed in 1980 12,13 . Since its first use for protein structure determination 14 , it has been used in a wide range of applications . Indeed, the brilliance of the synchrotron radiation (SR) in the vacuum UV (VUV) range and its stability enables one to measure CD spectra of samples dissolved in buffer down to 160 nm 18 with an acceptable signal to noise ratio for the heterodyne detection. In this paper, we describe the development of a different approach for SRCD measurements, utilising the natural polarisation of the SR emitted by a bending magnet 19,20 . The DISCO beamline provides a SR beam composed of two parts (Supplementary Information Fig. 1), which are above and below the electron's orbital plane in the synchrotron storage ring. These two continuum beams are similarly elliptically polarised but with opposite direction (Supplementary Information 2). Schiller and Hormes 21 have previously demonstrated that the natural elliptical polarisation of the SR can be used for CD spectroscopy. However, they measured the CD signal sequentially, wavelength by wavelength; so the scanning process of the monochromator limited the temporal resolution of their setup significantly. We have developed a spectrograph that measures simultaneously the intensity of both (oppositely polarised) parts of the SR beam, allowing one to determine a whole UV CD spectrum in just a single measurement. While current SRCD setups only take advantage of the wide spectral range and the brilliance of SR, our setup also uses its natural polarisation and its temporal distribution. This broadband single measurement approach gives access to a combination of a temporal and a spectral range previously inaccessible with the current broadband CD setups, enabling new insights into biomolecular dynamics. ## Method The tr-SRCD setup is annexed to the existing SRCD endstation at the DISCO beamline 22 . The optical layout is shown Figure 1. The beamline excitation monochromator is used first order for the wavelength calibration of the detection spectrograph (Supplementary Information Fig. 3). With the zero order, we get on the sample a pulsed white beam containing all the wavelengths between 120 nm to 600 nm; this pulsed broadband source is used for the tr-SRCD measurement, limited by the CMOS spectral response. The beam is spatially defined by a spatial filter (labelled S1 in the figure), then centred by the two plane mirrors (M'1, M'2) before being refocused on the sample. In order to avoid damage from beam irradiation, this optical system is designed to allow variation of the illuminated area on the sample. The spherical mirror M'3 and the cylindrical mirror M'4 refocused the beam 2975 mm after the monochromator slits; the calculated beam diameter at the focal plane is about 300 µm (FWHM). By moving the sample cell by 200 mm along the optical axis, we get a controllable probe beam diameter from 300 µm up to 3 mm. Adjustment of the beam diameter to the exposure time helps to minimize the sample irradiation dose. Beyond the sample, the cylindrical mirror M'5 and the spherical mirror M'6 horizontally focused the beam on a 6 secondary slit (S2). This secondary slit at 3900 mm is used to define the final spectral resolution of the setup. The flat M'7 mirror reflected the beam on a spherical flat field grating with 580 grooves per mm (Horiba Jobin-Yvon) that diffract and focus the incident UV/vis light horizontally onto the 2D detector. The tr-SRCD setup acquisition frequency is defined by the camera frame rate, which can be set from 0.033 Hz to 100 Hz in full resolution mode and can be increased further by reducing the pixel array size. This range of sampling frequency allows one to study reaction kinetics from tens of milliseconds to minutes through real-time recording and sequential measurements. The image intensifier can be synchronized with the camera output clock and with an external function generator providing trigger signals with a higher frequency. When the intensifier is synchronized with the output clock, the temporal resolution of the measurement corresponds to the duration of the amplification gate or the light pulse length if the gate is sufficiently short (Supplementary Information 4). However, if an external device triggers the image intensifier at a higher frequency than the camera frame rate, the exposure time of the camera defines the temporal resolution of the setup. The intensifier gate only amplifies the light coming from one SR pulse. The intensifier allows overcoming the limitation brought by the minimal exposure time of the CMOS (50 µs) and so measures the intensity of one pulse at once (Figure 1 right). We use this protocol for the results presented in Figure 2b. We integrated 500 pulses for each recorded image for this measurement. ## Equation 1 In this case, the absorption difference between left-handed and right-handed circularly polarised light can be associated with the absorption difference between the upper and lower part of the SR (Supplementary Information Note 5). These portions of the image are selected in order to show opposite polarizations. Although such procedure reduces the photon flux at each wavelength, the masks applied remove at most 25% of the images intensity, giving a moderate effect on the photon flux useful for the measurements. ## Results We chose camphorsulfonic acid (CSA), purchased from Sigma-Aldrich, as our calibration sample, as it has two strong peaks in the VUV/deep UV range: the first at 290 nm and the second at 192 nm. It is commonly used as a calibration standard for CD spectrometers 23,24 . To validate that the recorded signal is CD, the two enantiomers must exhibit opposite signed spectra. The resulting spectra for D-CSA and L-CSA, measured between 190 nm and 315 nm for 30 mg/mL solution concentration, are shown in Figure 2a. The CMOS detector has been used without the image intensifier. It integrated the light for 500 µs and the images were acquired at 20 Hz. As expected, we observe two oppositely signed spectra with identical amplitudes. The ratio between the amplitude of the two peaks should be about 2 24 ; we obtained 2.11 +/-0.11 in the tr-SRCD spectra (curve (d) of Figure 2a). The signal-to-noise ratios at 190 nm for the D- In order to establish the highest theoretical temporal resolution of the setup that is attainable experimentally, we added an image intensifier, thereby obviating the temporal resolution limitation of the CMOS detector. We performed this steady state experiment on the CSA with the maximal temporal resolution as a proof-of-principle. The intensifier was triggered with an external trigger and its amplification was reduced to its shortest value, 100 ns. The measurements were made using the 8-bunch mode of the synchrotron SOLEIL which provides 82 ps pulses at 7.14 MHz. The intensifier amplified only one pulse per gate. So its temporal resolution corresponds to the pulse length, which, in this case, is 82 ps. In Figure 2b, each spectrum comes from ten images; 500 separated pulses were integrated for each image. The spectral range was reduced to between 255 nm and 300 nm, in order to optimize the setup for the detection of the peak at 290 nm. The signal-to-noise ratio at 290 nm for the D-CSA spectrum is 10.33. We tested the capacity of the setup to follow reactions in real time by studying a photocontrolled and reversible system, the azobenzene crosslinked peptide FK-11-X 25,26 . It is composed of a 16 amino acid peptide cross-linked to a photoswitchable molecule: azobenzene (Figure 3). This ligand can be isomerised from the trans to cis and cis to trans conformation using a 370 nm and a visible light source (460 nm), respectively. Several groups have studied the dynamics of this photo-induced reaction and the isomerisation of the azobenzene appears to occur within one picosecond . This conformational change constrains the peptide structure and triggers its folding and unfolding processes. Theoretical 25,31 and experimental 32,33 studies agree that, following the azobenzene isomerisation, peptide conformational changes occur on the microsecond scale. The azobenzene cross-linked peptide was synthesized as described previously 25 (Supplementary Information Note 6). The aim of our study was to follow the change in concentration of the unfolded (cis azobenzene) and folded (trans azobenzene, ground 11 state) FK-11-X peptide. We triggered the isomerisation with two continuous diodes, at 370 nm (5W) and at 460 nm (5W). The concentration of FK-11-X solution was 10 mg/mL in phosphate buffer (70 mM, pH 7). The sample was alternately irradiated with the two diodes. The sample was first irradiated 2.5 seconds with the 370 nm diode to trigger the trans to cis isomerisation and then 2.5 seconds with the 460 nm diode to trigger the cis to trans isomerization. The transient CD and absorption spectra of isomers concentration changes were followed, measuring spectra at 130 Hz with a 520 µs temporal resolution; the measurement has been integrated over 39 cycles of alternate irradiations. The variations in the variations of absorption and CD are shown in Figure 4. shows that the trans to cis isomerisation is completed for the main part of the molecules after 1 second of irradiation with the 370 nm diode. This plateau does not correspond to an equilibrium between the diode flux and the lifetime of the cis-isomer. Indeed, the FK-11-X peptide is relatively stable in its cis configuration and it needs tens of minutes to naturally switch from the cis to the trans configuration. It also shows the complete reversibility of the concentration ratio. The intensity reference I 0 for the determination of the absorbance variation and so for the CD calculation corresponds to the average of the intensity measured between 1 and 2.5 seconds. In this time range, almost all the molecules are in their cis configuration. Thus, the absorbance and CD variation measurements shown in the Figure 4 correspond to the variation from the absorbance and CD spectrum of a solution of cis FK-11-X peptides. The CD variation spectra (Figure 4c and S5) are close to the signal of the α-helical structure and the increasing amplitude reflects the change of the species concentration. The first spectrum corresponds to the total difference between the CD spectrum of the unfolded (cis) FK-11-X peptide and its folded (trans) configuration. Then the concentration of the trans configuration decreases progressively during the 370 nm irradiation and so the amplitude of the CD difference decreases as well until it reaches approximately 0 before the first second of irradiation. The CD variation at 204 nm and 190 nm (Figure 4d) also confirms the reversibility of the system, the CD difference returns to its initial value at t0 after the 2.5 seconds of irradiation with the 460 nm diode. The total CD changes from the initial state and final state measured with the tr-SRCD setup (Figure S5) and from steady state measurement (cis and trans CD spectra) in the literature 32 are similar. This comparison shows that being able to calibrate the SR polarisation, we can measure transient CD spectra similar to those measured with steady state setups. The only difference arising from the elliptical polarisation is the amplitude of the CD, which is linearly correlated to the circularity of the polarisation. Our study demonstrates that the new tr-SRCD setup developed at the DISCO beamline 22 of the SOLEIL synchrotron radiation facility offers new opportunities for the investigation of biomolecular structure and dynamics. The setup is based on the use of the natural properties of the synchrotron radiation provided by the DISCO beamline. It uses both spectral and polarisation characteristics to reduce the time required for the measurement of a transient CD UV spectrum. Our tr-SRCD measurements of the CSA show that an integration time of 500 µs is enough to obtain an acceptable signal-to-noise ratio, i.e., greater than 6. We performed a proof-of-principle high temporal resolution steady state experiment on CSA performed using an image intensifier allowing resolution down to a few tens of picoseconds. The FK11 experiment demonstrates the capacity of the tr-SRCD setup to follow dynamics in real time, with a temporal resolution of 500 µs and acquired at 130 Hz. These values do not correspond to the hardware cap and can be easily improved, by opening the spectrograph slits and integrating more photons in shorter time period and/or coupling the system with the image intensifier. This breakthrough in the SRCD spectroscopy will enable the study of many biological and chemical reactions crucial for our understanding of biomolecular phenomena. The development of new CaF2 microlenses windows for sCMOS camera would grant access to direct detection and measurement of spectra on the full DUV range down to 160 nm. In the near future, the setup could easily be coupled to other reaction triggers, such as a T-jump, lasers and microfluidic stop-flow equipment. Thus, this new approach will provide exciting insights into the dynamics of biomolecules, as well as for molecular and materials systems more broadly. This method will enable one to follow the behaviour of molecules through high quality SRCD spectra on a temporal range from the picoseconds to minutes. Indeed, it covers time scales consistent with the fluctuation and domain motions of proteins

chemsum

{"title": "Time resolved transient circular dichroism spectroscopy using synchrotron natural polarisation", "journal": "ChemRxiv"}

the_optical_properties_of_monoclinic_na3alf6_and_na3alf6:mn_4+

## Abstract: Mn 4+ activated luminescent materials have attracted much attention recently. In particular, alkali metal hexafluorometallates, such as K2SiF6:Mn 4+ or K2TiF6:Mn 4+ , emit light in the red spectral region on phosphor converted LEDs (pc-LED). We applied the cation-exchange method in order to synthesize Mn 4+ doped Na3AlF6. Na3AlF6:Mn 4+ exhibits efficient red photoluminescence peaking at 627 nm, which can be assigned to the 2 Eg → 4 A2g intraconfigurational transition of Mn 4+ ([Ar]3d .3 configuration) within the [MnF6] 2octahedra on the aluminum site in the cryolite host structure. Photoluminescence properties, such as temperature dependence of the PL intensity and luminescence lifetime are presented. Additionally, the band structure of the undoped host material has been treated with Density Functional Theory (DFT). The theoretical results have been evaluated experimentally with diffuse UV reflectance spectroscopy. Finally, luminous efficacy and color rendering values of simulated warm white emitting pcLEDs comprising a dichromatic phosphor blend employing Na3AlF6:Mn 4+ as the red emitting component are calculated and compared to the performance of warm white emitting pcLEDs comprising K2SiF6:Mn 4+ as red emitting component. ## Introduction Since the publication of the first high brightness blue light-emitting diode (LED) and the correlating blue laser diode (LD) based on (In,Ga)N by Nakamura et al. in 1993, Solid-state lighting (SSL) has advanced rapidly during this decade, and has recently become competitive to conventional light sources for general lighting. In the center of SSL is the LED, whose manifold history traverses a wide range of semiconducting materials and device architectures. The current standard architecture for SSL is the phosphor-converted light-emitting diode (pc-LED) in which high brightness LEDs based on (In,Ga)N are combined with one or more down-converting phosphors, mostly Ce 3+ doped garnets, to additively produce white light. These light sources provide high luminous efficacy due to intense emission in the green range of the visible spectrum. Nevertheless, due to the lack of red emission this combination shows a high color temperature and low color rendering index (CRI). These properties are often not suitable for domestic lighting or display applications. To increase the CRI, while not decreasing luminous efficacy too strongly, narrow band red emitting phosphors are required. In fluorescent lamps this problem is solved by Eu 3+ activated phosphors, which are not suitable for pcLEDs due to their low absorption cross section in the near UV or blue spectral region. Another approach is the use of Mn 4+ activated phosphors, which have a rather high absorption cross section due to spin allowed transitions in the near UV and blue range. These are attributed to 4 A2 → 4 T1 and 4 A2 → 4 T2 intraconfigurational transitions of Mn 4+ ([Ar]3d .3 configuration). The photoluminescence emission (PL) of Mn 4+ doped materials originates from the spin-forbidden intraconfigurational transition 2 Eg → 4 A2g. This transition is generally capable of generating narrow band red to deep red light -depending on the host material -with high brightness and quantum efficiencies close to unity. Recently, Mn 4+ dopes alkali metal fluorometallates have attracted much attention due their blue shifted PL in comparison to Mn 4+ doped oxides. Particularly, K2SiF6:Mn 4+ or K2TiF6:Mn 4+ have been studied intensively. Furthermore, PL properties of several Mn 4+ doped fluorides with the general formula A2MF6 (A = K, Na, Cs, Rb or NH4; M = Si, Ge, Ti, Zr or Sn) or AESiF6 (AE = Ba or Zn) and different synthesis methods have been published. These types of materials emit bright light in the desired red spectral region and thus meet the efficiency and color quality of "future warm white" pcLEDs. Lately, a synthesis method for Na3AlF6:Mn 4+ and its PL properties have been published by Song et al. However, using a different synthesis route as published by Zhu et al. we were able to synthesize Na3AlF6:Mn 4+ via a one-step cation exchange approach using K2MnF6 and Na3AlF6 and a very small amount of concentrated hydrofluoric acid. The Mn 4+ cation with an ionic radius of 0.53 pm in octahedral coordination fits perfectly on the Al 3+ (ionic radius = 0.535 pm in octahedral coordination) site within the cryolite host structure. Cryolite is chemically very stable, as indicated by its occurrence as a mineral. Moreover, it has good thermal stability and can be expected, as most fluorides, to be a wide band gap material. Both of which are requirements for phosphors. In as he red emitting component are calculated and compared to the performance of warm white emitting pcLEDs comprising K2SiF6:Mn 4+ as the red emitter. 6 ## Crystal structure and method of calculation Cryolite crystallizes in the monoclinic space group P121/c1. The atomic arrangement is characterized by an open framework of [AlF6] 3octahedra lying at the corners and facecenter of a nearly cubic lattice. The crystal structure is represented in Fig. 1. The octahedra are slightly displaced from the highest symmetry. This results in a low Al site symmetry of Ci. This fact impacts the PL of the investigated material. It can cause several zero phonon lines (ZPLs) depending on the stark splitting of the different sublevels. For the ab initio calculations we have used the CASTEP module of the Materials Studio Package (v. 8.0). The linear density approximation (LDA) with CA-PZ functional, based on the Ceperley and Alder data as parameterized by Perdew and Zunger, was applied to treat the exchange-correlation effects. The ultrasoft pseudopotentials were employed for a description of interaction between the ionic cores and the valence electrons. The electronic configurations used for the calculations were as follows: 2s 2 2p 6 3s 1 for Na, 3s 2 3p 1 for Al, and 2s 2 2p 5 for F. The Monkhorst-Pack k-points was set as 10 × 10 × 10. The calculations have been performed for monoclinic cell. ## Structural and electronic properties Optimization of the crystal lattice constants with the aforementioned calculating settings has led to the following lattice constants (in ): a = 5.4139, b = 5.5602, c = 7.7777. The calculated band structure of Na3AlF6 is shown in Fig. 2. The calculated direct band gap is 7.05 eV, which allows to classify this compound as a wide-gap dielectric. The experimental bandgap (see Fig. 3) for this host material was investigated by UV-reflectance spectroscopy, as described by Enseling et al. before . To obtain the optical band gap Kubelka-Munk and Tauc relation have been applied . The experimentally obtained bandgap is 7.06 eV which is in very good agreement with the calculated one. The calculated density of states (DOS) diagrams that are shown in Fig. 4, allow the following assignment of the electronic bands as a wide (about 7 eV) conduction band (CB), which mainly consists of the Al 3s and 3p states plus Na 2s and 2p states, respectively. The valence band (VB) has a width of about 4 eV and consists of many very narrow sub-bands, which exhibit almost no dispersion at all. The dominating contribution to the VB is originating from the F 2p states. The Al 3s and 3p have a significant contribution as well. These states are mainly found at the bottom of the VB. The F 2s states are spread from -21 to -17.6 eV. Al 3s and 3p states have also a minor contribution to these bands. The Na 2p (and 2s with minor contribution) states are peaking at -18.5 eV. ## Experimental Section According to Zhu et al. the investigated Na3AlF6:Mn 4+ samples were synthesized by a onestep cation exchange method, as depicted in Fig. 5. The Mn 4+ source K2MnF6 was synthesized in the following manner: Typically, 60 g of KF (Sigma Aldrich, 99%) have been dissolved in 250 ml hydrofluoric acid (Sigma Aldrich, ≥48%) in a PTFE beaker. The mixture was cooled to 0 °C with an ice bath during the following reaction. 3.75 g of KMnO4 (Sigma Aldrich, ≥99%) were added to the mixture and stirred for about 15 minutes. Undissolved KMnO4 was removed by decantation. Afterwards, an H2O2 solution (Sigma Aldrich, 30%) was added dropwise until the mixture turned from dark violet to dark red and yellow K2MnF6 precipitated. The precipitate was filtered and washed with cold acetone (-5 °C) to remove traces of HF. Finally, the K2MnF6 was dried in a vacuum oven at 40 °C for 4 hours . 0,02 mmol of K2MnF6 was dissolved in a few drops (~ 2 ml) of hydrofluoric acid (Sigma Aldrich, ≥48%) and stirred in a PTFE beaker. 2 mmol of Na3AlF6 (Sigma Aldrich, 99.9%) was added. The mixture was heated up to 70 °C to enhance the ion diffusion process of Mn 4+ into the cryolite. After 20 minutes of stirring the mixture was rapidly cooled down to -10 °C. The resulting Mn 4+ doped cryolite was washed with cold acetone (-5 °C) and dried in a vacuum oven at 40 °C for 4 hours. ICP-MS measurements were made using an ICP Mass Spectrometer by Perkin Elmer in order to quantify the amount of manganese in the material. These revealed that the Mn 4+ content amounts to 0.1 at.% relative to the Al site. Phase purity of Na3AlF6, Na3AlF6:Mn 4+ (0.1%) and K2MnF6 was investigated by powder Xray diffraction (XRD). XRD patterns were recorded on a Rigaku MiniFlex II diffractometer working in Bragg-Brentano geometry using Cu Kα radiation. Photoluminescence (PL) and photoluminescence excitation (PLE) spectra were recorded on an Edinburgh Instruments FSL900 spectrometer equipped with a Xe arc lamp (450 W) and a peltier cooled (-20 °C) single-photon counting photomultiplier (Hamamatsu R2658P). Obtained emission spectra were corrected by applying a correction file gained from a tungsten incandescent lamp certified by the National Physics Laboratory U.K. Lifetime measurements of the investigated sample were performed with a micro-second pulsed Xe lamp (Heraeus µF920H). Temperature dependent PL spectra and specular reflectance measurements from 77 K to 500 K were performed using an Oxford Instruments cryostat MircostatN2. Liquid nitrogen was used as a cooling agent. Temperature stabilization time was set to 30 s with a tolerance of ±3 K. Temperature-dependent measurements in the range 3 to 100 K were performed using a closedcycle cryostat filled with helium (Oxford Instruments Optistat AC-V 12). Scanning electron microscopy (SEM) photographs and energy dispersive X-ray spectra (EDX) were recorded on a Zeiss EVO MA10 equipped with a secondary electron (SE), back-scattered electron (BSE), and a Bruker Quantax EDS detector (XFlash 6), respectively. The SEM was operated in high vacuum mode (10 -7 Pa). ## Results and Discussion Fig. 6 shows the XRD pattern of an as-prepared sample. All the diffraction peaks match to the reference as published by Ross et al. indicating that the obtained sample has a crystal structure identical to cryolite . The average size of the crystallites (obtained by Scherrer equation) is in the sub-micrometer range (~ 0.2 µm). No traces of K2MnF6 residual or any other impurity phases where observed. The SEM image (Fig. 7a) shows very uniform particles with a smooth surface. These are roughly 1 µm in size and mostly agglomerated to form bigger particles with up to 10 µm diameter. The composition according to the formula Na3AlF6 was confimed by EDX spectra (Fig. 7b). Quantification led to the fallowing composition: Na 29.3%, Al 9.7%, and F 61.0%. The sample shows a uniform yellow body color under natural light illumination (Fig. 5b), while under UV excitation (366 nm), a very bright red emission can be observed (Fig. 5a). Fig. 8 shows the PLE and PL spectra of Na3AlF6:Mn 4+ (0.1%) at room temperature. The PLE exhibits three distinct bands at < 250 nm, 360 nm, and 465 nm assigned to F -→ Mn 4+ charge transfer (CT) transition, and spin-allowed 4 A2 → 4 T1 and 4 A2 → 4 T2 interconfigurational transitions of Mn 4+ ion in octahedral coordination. These spectra profiles are similar to those reported before . The PL spectrum, which originates from the spin-forbidden 2 E → 4 A2 transition is composed of several sharp lines in spectral range from 600 to 650 nm with the strongest peak at 627 nm (Fig. 8). The optical transitions of Mn 4+ are sensitive to their local coordination and the local symmetry of the crystallographic site occupied by the Mn 4+ ions. The low Al site symmetry of Ci results in the occurrence of two distinguishable zero phonon lines (ZPL) . Due to crystal field splitting and spin-orbit interaction the occurance of several ZPLs was expected. Additional Stokes and anti-Stokes sidebands, typical for Mn 4+ phosphors, can be observed, too. Crystal field strength Dq and the Racah parameters B and C (listed in Tab. 1) were calculated and compared to those of various other sodium hexafluorometallates. To investigate the number of ZPLs in depth we measured high resolution emission spectra at 3 K (Fig. 9). The 3 K measurement avoids thermal broadening of the spectral bands and allows a more reliable assignment of the energy levels. Na2SnF6 2101 589 3873 16171 It is well known that the PL integrals of most luminescent materials, such as Eu 3+ -doped phosphors, usually decrease sigmoidally with increasing temperature due to the increase of nonradiative transition probability . However, we found an anomalous T-dependent PL behavior for Na3AlF6:Mn 4+ in the temperature range of 80 to 300 K. This phenomenon has been observed for several Mn 4+ doped phosphors before e.g. for Mg14Ge5O24:Mn 4+ and most recently for K2TiF6:Mn 4+ . However, it was suggested by Zhu et al., that the quantum yield remains constant in this temperature region To explain the linear rise up of the PL integrals absorption has to increase. To verify this circumstance, we measured T-dependent reflectance spectra and transformed them with the help of the Kubelka-Munk relation into absorption spectra (see Fig. 12). We found a steady increase of the absorption bands assigned to 4 A2 → 4 T2 interconfigurational transition and a shift to the blue with increasing temperature. The absorption probability gets larger due to a progressive mixing of the 3d Mn 4+ with the 2p Forbitals which weakens the parity rule . The T-dependent emission spectra (see Fig. 10) show a red shift with increasing temperature. This can be explained by increasing covalency of the material which, therefore, supports increased mixing of the orbitals. At temperatures higher than 300 K non-radiative transitions are becoming more dominant and typical T-induced PL quenching sets in. Between 300 and 500 K this behavior can be fitted well with a Fermi Dirac relation (see Fig. 11) . The turning point T1/2 gained from this fit is found at 392 K. The T-dependent decay curves are shown in Fig. 13. With increasing temperature, the PL lifetime of the spin-forbidden 2 E → 4 A2 transition starts to decrease at 100 K in a sigmoidal manner, which has been fitted with Fermi Dirac relation as well. The T1/2 value, gained from this fit, is 285 K. In order to compare the performance of Na3AlF6:Mn 4+ with various red emitting phosphors we simulated pcLED spectra composed of the spectra of a blue 465 nm LED (1000 cm -1 , 22 nm FWHM) and yellow emitting Y3Al5O12:Ce 3+ . Exemplary simulated spectra for Na3AlF6:Mn 4+ and K2SiF6:Mn 4+ as red component are shown in Fig. 14. Furthermore, for the calculated values listed in Tab. 2, was either Na3AlF6:Mn 4+ , K2SiF6:Mn 4+ , Mg14Ge5O24:Mn 4+ , or CaAlSiN3:Eu 2+ used as red component. The individual spectra were combined by summing their intensity values to yield simulated pcLED emission spectra. The ratio of the phosphors was adjusted to result in a spectrum with a correlated colour temperature of 2700 K and 3000 K, respectively. The CIE1931 x/y colour points of the spectra were chosen according to ANSI C78.377 (2700 K: 0.4578/0.4101; 3000 K: 13 0.4338/0.4030). There is but one way to combine three spectra to yield a specific colour point. For each simulated spectrum luminous efficacy (LE) and color rendering index (CRI) were calculated employing Osram Sylvania Color Calculator v4.59. The results are presented in Table 2. Both Mn 4+ activated fluorides exhibit LE and CRI values superior to those of the oxide Mg14Ge5O14:Mn 4+ phosphor and the commercial CaAlSiN3:Eu 2+ phosphor. This is the result of the short peak wavelength and small full-width-at-half-maximum values of Mn 4+ in fluoride hosts. Tab. ## Conclusion In conclusion, the red emitting phosphor Na3AlF6:Mn 4+ , was successfully synthesized via a simple one-step method at moderate temperatures. The optical structure was investigated by DFT calculations and experimentally evaluated with the help of UV-reflectance spectroscopy. The optical band gap of undoped Na3AlF6 was found to be ~ 7 eV. Na3AlF6:Mn 4+ absorbs broad-band blue light and provides bright narrow band red light with an emission maximum at 627 nm. At low temperature (3 K) up to 6 ZPLs can be observed. The unusual increase of emission integrals with increasing temperature originates from a progressive increase of the absorption probability in the low temperature regime (100 -300 K). At higher temperatures, the typical PL quenching behavior can be observed. Decay curve recordings show a strong drop of τ with increasing temperature. Therefore, the activator efficiency or internal quantum efficiency drops much faster than indicated by T-dependent emission measurements. This is the result of increasing absorption counter-acting the decreasing activator efficiency. Furthermore, LE and CRI values of simulated warm white emitting pcLEDs comprising YAG:Ce and Na3AlF6:Mn 4+ were calculated and compared to the performance of pcLEDs comprising YAG:Ce and K2SiF6:Mn 4+ , or Mg14Ge5O24:Mn 4+ , or CaAlSiN3:Eu 2+ . It turns out, that Na3AlF6:Mn 4+ has much better properties for warm-white pc-LEDs in comparison to Mg14Ge5O24:Mn 4+ and CaAlSiN3:Eu 2+ . Its performance is slightly better compared to K2SiF6:Mn 4+ . ## Figures Fig. Fig. 2 The calculated band structure of Na3AlF6. LDA calculated electronic bands are shown. The Fermi level is set zero. The coordinates of the points (reciprocal lattice unit vectors) of the Brillioun zone are: Z (0,0,½); Γ (0,0,0); Y (0,½,0); A (-½,½,0); B (-½,0,0); D (-½,0,½,); E (-½,½,½).

chemsum

{"title": "The Optical Properties of Monoclinic Na3AlF6 and Na3AlF6:Mn 4+", "journal": "ChemRxiv"}

a_unified_strategy_to_reverse-prenylated_indole_alkaloids:_total_syntheses_of_preparaherquamide,_pre

## Abstract: A full account of our studies toward reverse-prenylated indole alkaloids that contain a bicyclo[2.2.2]core is described. A divergent route is reported which has resulted in the synthesis of preparaherquamide, (+)-VM-55599, and premalbrancheamide. An intramolecular Dieckmann cyclization between an enolate and isocyanate was used to forge the bicyclo[2.2.2]diazaoctane core that is characteristic of these molecules.The pentacyclic indole scaffold was constructed through a one-pot Hofmann rearrangement followed by Fischer indole synthesis. The utilization of our previously reported indole peripheral functionalization strategy also led to natural products including malbrancheamides B, C, stephacidin A, notoamides F, I and R, aspergamide B, and waikialoid A. Ultimately, the divergent route that we devised provided access to a wide range of prenylated indole alkaloids that are differently substituted on the cyclic amine core. ## Introduction Secondary metabolites isolated from fungi of the genera Penicillium and Aspergillus continue to attract the interest of the chemical synthesis community because of their structural complexity and diverse biological activity. Among these natural products are a series of reverse-prenylated indole alkaloids that exhibit a wide range of bioactivity including, but not limited to, insecticidal, cytotoxic, anthelmintic, and anti-bacterial properties. 1 Representative of this group are the paraherquamides (e.g., 2, Fig. 1), stephacidins (e.g., 5), and marcfortines (e.g., 7), all of which possess a bicyclo[2.2.2]diazaoctane core ring system (highlighted in red in 1). Structurally, members of this group are comprised of two amino acids, tryptophan (highlighted in blue in 2) and proline or isoleucine (highlighted in maroon in 3 and 6, respectively). These core motifs are in turn reverse-prenylated (highlighted in green in 5). The different substitution and oxidation patterns present on the core framework of these indole alkaloids, along with their intriguing biogenesis, has spurred numerous synthetic campaigns to prepare them. Prior synthetic strategies can be organized around the construction of the bicyclo[2.2.2]diazaoctane core (Fig. 1B). 2 For example, Williams and co-workers employed a biomimetic intramolecular Diels-Alder reaction to construct the bicyclo ring system en route to ()-VM5599, rac- a complementary strategy employed by Williams and coworkers featured an intramolecular S N 2 0 cyclization in the syntheses of brevianamide B, ()-paraherquamide A, (+)-paraherquamide B, stephacidins A and B, notoamide B, and avrainvillamide (forming C 22 -C 6 , Fig. 1B). 3 Inspired by the seminal reports of the Mislow and Saegusa laboratories as well as previous work from their own laboratories, Baran and coworkers utilized an oxidative enolate coupling strategy to achieve the frst total synthesis of stephacidin A (forming C 22 -C 6 , Fig. 1B). 4 Further oxidative elaboration also resulted in syntheses of avrainvillamide and stephacidin B. 4 Other strategies have relied on cationic, radical, acyl radical, and oxidative aza-Prins cyclizations (Fig. 1B). Despite the existing elegant strategies to access specifc congeners within this family of natural products, a unifed approach to access prenylated indole alkaloids that either possess or lack additional substituents on the cyclic amine ring (highlighted in maroon in Fig. 2A and labelled ring A in Fig. 2C for clarity) of the hexacyclic framework remained an outstanding challenge. 9 Biosynthetically (Fig. 2A), natural products like paraherquamide A (6), which contain a methylproline residue (R 0 ¼ Me), are proposed to arise from a 4-electron oxidation of isoleucine to furnish an aldehyde intermediate which undergoes reductive amination. 10,11 Alternatively, congeners like stephacidin A, which feature no substituents on the cyclic amine core, are derived from L-proline. Synthetically, it became evident that adhering to a bioinspired approach would require a series of challenging site selective late-stage oxidations in order to access each member of this reverseprenylated indole alkaloid family (Fig. 2B). 10,11 For example, issues of chemoselectivity were encountered in our own work with previously reported amide 7 and late-stage oxidation in the presence of the indole moiety proved difficult (8 in Fig. 2B). Given the aforementioned challenges and strategic value of divergent total synthesis for accessing related synthetic targets, a route was designed to leverage a versatile common intermediate (9, Fig. 2C) which contains a ketone as a synthetic handle on ring A for diversifcation. Because ketones engage in a plethora of organic reactions, we envisioned using the diverse reactivity of the carbonyl group to perform either late-stage nucleophilic additions to access members within the paraherquamide family, ring expansion to access macfortine natural products, or leverage enolate chemistry to access the mangrovamides. Another strategic design element embedded in the versatile common intermediate ( 9) is the unsubstituted indole motif, which was selected to maximize access to the diverse indole substitution patterns characteristic of the reverseprenylated indole alkaloids through late-stage indole C-H functionalization. From 9, the bicyclo[2.2.2]diazaoctane structural motif could also be constructed using a similar strategy to our previously reported Dieckmann-type cyclization. 13 Pentacyclic indole 10 was envisioned to arise from our previously reported tricyclic intermediate 11 and phenyl hydrazine by Fischer indole synthesis. 13 ## Results and discussion Our initial approach centered around a one-pot protocol for the construction of the bicyclo[2.2.2]diazaoctane core. It was envisioned that a Hofmann rearrangement performed in the absence of any external nucleophiles would generate isocyanate 13 in situ, which could be intercepted by an attendant enol or enolate, through an intramolecular cyclization, to provide the bicyclo[2.2.2]diazaoctane ring (see 9) in a single-pot transformation. 16 In order to explore the feasibility of this one-pot Hofmann/cyclization event, access to pentacyclic indole 12 was required. We commenced our synthetic studies with enone 11, which is available in gram-scale quantities (9 steps, 37% overall yield) from commercially available 1-tert-butyl 2-ethyl 3oxopyrrolidine-1,2-dicarboxylate (Fig. 3B; see S1 in the ESI ‡). 13 Following hydrogenation of enone 11 using Pd/C, treatment with BBr 3 effected cleavage of the benzyl group to provide ketone 14. 13 Treatment of ketone 14 with phenyl hydrazine in aqueous sulfuric acid at 100 C afforded pentacyclic indole 15 in 83% yield. Oxidation of the secondary alcohol group with Dess-Martin periodinane (DMP) gave the desired ketone (not shown) as an inseparable mixture of diastereomers in a 1 : 2 ratio (a : b epimers) that was taken directly to the next reaction because of stability issues. The use of the Ghaffar-Parkins platinum complex failed to provide any desired hydrated product (12) and led to a complex product distribution. 17 An alternative procedure for nitrile hydration reported by Lee and co-workers using Wilkinson's catalyst and acetaldoxime in toluene at reflux led only to decomposition of starting material. 18 Presumably, the enol (not shown) can compete for the binding site of the metal center in these cases leading to decomposition pathways. Given the difficulty in performing the nitrile hydration after oxidation of 15, an alternative sequence was explored. Nitrile 15 was hydrated with the Ghaffar-Parkins complex (16) to provide primary amide 17 in 96% yield. At this stage, traditional oxidation methods were explored, which led to either decomposition, low yields, or recovered starting material (Fig. 3B; see the ESI ‡ for further details). Presumably, the presence of the nucleophilic primary amide in 17 adversely affected the oxidation of the secondary alcohol group. Because of the issues with functional group incompatibility, Hofmann rearrangement prior to oxidation of the secondary alcohol was explored. With alcohol carboxamide 17 in hand, conditions to effect the Hofmann rearrangement were investigated (see Fig. 3D). Treating 17 with phenyliodoso-trifluoromethyl acetate (PIFA) resulted in decomposition of the starting material. Interestingly, treating carboxamide 17 with Pb(OAc) 4 in a mixture of DMF/MeOH at room temperature, resulted in the formation of [3.2.1] bicycle 21 as the sole product of the reaction in 89% yield instead of the expected methyl carbamate (18). Presumably, this [3.2.1] bicyclic system arises from an initial oxidation of the primary carboxamide to generate the N-acyl nitrene (19), which then interacts with the indole C2-C3 double bond forming aziridine 20. The indole nitrogen then facilitates opening of the aziridine at the C2 position, driven by release of ring strain (as shown by the red arrows). A proton transfer then delivers 21. Interestingly, while reminiscent of aspeverin, 19 this unique ring system does not translate to any natural product scaffolds that have been reported to date. On the basis of these results, the Hofmann rearrangement would have to be accomplished prior to the installation of the indole moiety in order to avoid the formation of [3.2.1] bicycle 21. Following from our prior studies, it was determined that Pb(OAc) 4 in the presence of MeOH at room temperature was optimal for mediating the desired transformation to provide the methyl carbamate (23, Fig. 3E) accompanied by varying amounts of recovered starting material. After some optimization, it was established that elevated temperatures (70 C) were required to attain full conversion of the starting material. However, reaction yields dropped dramatically upon scale-up and over 6 equivalents of Pb(OAc) 4 were required to achieve complete consumption of the starting material. Nevertheless, methyl carbamate 23 was advanced through the Fischer indole synthesis to provide indole 24 in 74% yield. Subsequent oxidation with Dess-Martin periodinane (DMP) provided the desired ketone precursor (after oxidation of secondary alcohol highlighted in maroon), which was predisposed for the latestage Dieckmann cyclization to afford the bicyclo[2.2.2]diazaoctane core. Unfortunately, treatment with a variety of bases (NaH, KHMDS, or KO t Bu) or acid (e.g., TFA), were unsuccessful in providing the desired bicyclo[2.2.2] product (9). Presumably, the methyl carbamate is not sufficiently electrophilic for the cyclization step and/or the methoxide nucleofuge may subsequently serve as a better nucleophile in an undesired irreversible direction (cleaving the bond highlighted in blue in 9; Fig. 3E), leading to decomposition. Therefore, the installation of different carbamates was explored; carbamates bearing better leaving groups, such as phenols and polyfluoroalcohols, and thus less nucleophilic nucleofuges, emerged as an attractive option to minimize decomposition pathways. Ultimately, a one-pot protocol was developed to access pentacyclic indole 25 (Scheme 1). It is worth noting that a modifed work up procedure yielded two-step access to primary amide 22 from enone 11 and obviated the need for transition metalcatalyzed nitrile hydration. From 22, Hofmann rearrangement of the carboxamide was effected under mild conditions using (tosylimino)phenyl-l 3 -iodane (PhINTs). 20 Upon treatment with aqueous acid, the resulting isocyanate (23) was converted to the corresponding ammonium intermediate (24), which was directly subjected to phenylhydrazine to effect Fischer indolization, providing pentacyclic indole 25 in a single-pot operation from 22. Chemoselective amine carbamoylation of 25 in the presence of a secondary hydroxy group was achieved in high yield with phenyl chloroformate to afford phenyl carbamate 26. Oxidation of secondary alcohol 26 provided desired cyclization precursor 27. Treatment of 27 with K CO 3 in acetone yielded the bicyclo[2.2.2]diazaoctane core, presumably through Dieckmann cyclization of an intermediate enolate (generated from the ketone group) and the isocyanate group (generated in situ from the phenyl carbamate under basic conditions). 13,21 Notably, the bicyclo[2.2.2]diazaoctane core was constructed in 5 steps from 11, which is an improvement over our prior work (10 steps from 11 in that case) and provides access to a wider range of congeners by leveraging the unsubstituted indole motif (vide infra). 13 The syntheses of preparaherquamide and (+)-VM55599 were accomplished through a four-step sequence that installed the requisite functionality on the fve-membered ring (Scheme 1). Initial efforts toward olefnation of ketone 9 were unsuccessful as Wittig olefnations lead to recovered starting material. A twostep nucleophilic addition followed by alcohol elimination was explored. For example, MeMgBr addition afforded tertiary alcohol 28 in low yields with signifcant recovered starting material. Presumably, side reactions like alpha deprotonation (enolization) result in low conversion. Re-subjecting the crude mixture to another cycle led to 48% yield of the desired tertiary alcohol (28). Efforts to attenuate the basicity of the Grignard reagent were unsuccessful as addition of cerium(III) chloride did not improve conversion and ultimately, it was found that the addition of LiCl led to a 51% yield (85% BRSM) of the desired tertiary alcohol 28. 22 Treatment of 28 with the Burgess reagent gave a mixture of exocyclic (29) and endocyclic (30) alkenes in 71% yield (1 : 2 mixture). Hydrogenation of the mixture of alkenes (i.e., 29 and 30) using Pd/C, followed by chemoselective tertiary amide reduction with DIBAL-H, afforded epimeric natural products preparaherquamide (2) and (+)-VM-55599 (1) in 25% and 27% yield, respectively. Having successfully accessed natural products bearing substituents on the pyrrolidine ring (ring A), we turned our efforts toward congeners derived from L-proline to showcase the utility of our unifed approach (Scheme 1). Notably, from 9, indole alkaloids lacking substituents on ring A can be accessed through sequential reduction processes. For example, Wolff-Kishner reduction of the pyrrolidone ketone group in 9 yielded ketopremalbrancheamide (31, Scheme 1). From ketopremalbrancheamide, premalbrancheamide (32) was synthesized according to the precedent of Williams and co-workers. 23,24 Thus, treating 31 with an excess of DIBAL-H effected chemoselective reduction of the tertiary amide. On the basis of our prior work on indole functionalization, ketomalbrancheamide (31) can be elaborated to malbrancheamides B and C, as well as stephacidins A and B, waikialoid A, aspergamide B, and fnally notoamides F, I, and R. 15 ## Conclusions In conclusion, a unifed approach was developed to access hexacyclic indole alkaloids that either possess or lack substituents on the cyclic amine. Specifcally, studies that culminated in the total syntheses of (+)-VM55599 (1), preparaherquamide (2), and premalbrancheamide (31) as well access to ketomalbrancheamide (32) which can be elaborated to malbrancheamides B, C, stephacidin A, notoamides F, I and R, aspergamide B, and waikialoid A through our previously reported peripheral indole functionalization strategy is reported. This work serves as a blueprint for a unifed approach to the synthesis of reverse-prenylated indole alkaloids possessing a bicyclo[2.2.2]diazaoctane core by addressing a key challenge posed by the diverse range of substitution found on the cyclic amine ring of the core framework. Our modular strategy hinged on leveraging a late-stage intermediate possessing a synthetic handle as well as rapid construction of the pentacyclic indole skeleton through a one-pot Hofmann rearrangement and Fischer indole synthesis. ## Conflicts of interest There are no conflicts to declare.

chemsum

{"title": "A unified strategy to reverse-prenylated indole alkaloids: total syntheses of preparaherquamide, premalbrancheamide, and (+)-VM-55599", "journal": "Royal Society of Chemistry (RSC)"}

role_of_viscosity_in_deviations_from_the_nernst-einstein_relation

## Abstract: Deviations from the Nernst-Einstein relation are commonly attributed to ion-ion correlation and ion-pairing. Despite the fact that these deviations can be quantified by either experimental measurements or molecular dynamics simulations, there is no rule of thumb to tell the extent of deviations. Here, we show that deviations from the Nernst-Einstein relation is proportional to the inverse viscosity by exploring the finite-size effect on the transport properties in periodic boundary conditions. This conclusion is in accord with established experimental results of ionic liquids. ## Introduction "Physical chemistry of ionically conducting solutions" is one of the cornerstones for energy storage applications in super-capacitors and lithium-ion batteries. 1 Historically, most electrolyte were regarded as incompletely dissociated and the dissociation constant is related to the factor α which can be expressed as the conductivity ratio Λ c /Λ 0 according to Arrhenius, where Λ 0 is the value at the infinite dilution. 2 Following the idea of using transport properties to quantify the extend of ion dissociation ("ionicity"), Angell and co-workers proposed to use the classical Walden rule for the purpose of classification. 3 The product of the conductivity Λ and the viscosity η of KCl solution measured at 0.1 m was set as the reference point. Downward deviations from KCl line are usually regarded as the formation of charge neutral ion pairs. The concept of "ionicity" was put forward further by Watanabe and co-workers. They suggested to use the molar conductivity ratio Λ imp /Λ NMR measured from the impedance spectroscopy (imp) and pulse-field gradient NMR to quantify the self-dissociativity of ionic liquids (ILs). Despite of its conceptual simplicity, the nature of "ionicity" is by no means simple. Apart from the electrostatic picture of charge neutral ion pairs, other factors may alter the interpretation of the experimental measured ratio Λ imp /Λ NMR . For instance, Λ NMR was obtained via the Nernst-Einstein relation and the charge transfer effect can lead to the deviation from the formal charge of ions. 7,8 Exceptional case can also be found that the ionicity goes up with the concentration counter-intuitively where the chelate effects become important. 9 A more general point related to "ionicity" is that ion-pairing is a subset of ion-ion correlations. 10,11 In one of a seriers of classic works on dense ionized matter from Hansen and McDonald, 12 they commented that "It is also clear that deviations from the Nernst-Einstein relation are not necessarily the result of a permanent association of ions of opposite charge". However, the remaining question is still what determines the deviation from the Nernst-Eistein relation if not ion-pairing. In this work, we used the finite-size effect in molecular dynamics (MD) simulation of transport properties to investigate the deviation from the Nernst-Einstein relation in case permanent ion-pairing is excluded. It is found that while the Nersnt-Einstein conductivity depends strongly on the system-size, the Green-Kubo conductivity is system-size independent. We showed that these two types of conductivities crossovers at certain simulation box size L min for both NaCl solutions and [BMIM][PF 6 ] IL. Furthermore, this observation suggests that the deviation from the Nernst-Einstein relation i.e. (Λ N−E − Λ G−K ) is inversely proportional to the viscosity η resembling the classical Walden rule, with L min being a system-specific parameter. We verified this relation with published experimental data for a variety of ILs. These results indicate that viscosity is a dominating factor for the deviation from the Nernst-Einstein relation and provide a new avenue to gauge the extent of ion-ion correlations in electrolyte systems. ## Theoretical background of the ionic conductivity At low salt concentration, the ionic conductivity of 1:1 symmetric electrolyte can be described by the Nernst-Einstein (N-E) equation (Eq. 1), in which the ionic conductivity is only linked to the self-diffusion of ions. 13 where β = 1/k b T is the inverse temperature, q is the formal charge of each ion and ρ = N/Ω is the number density of electrolyte (in the formula unit) with Ω as the system volume. σ s + and σ s − are contributions to the ionic conductivity from self-diffusion coefficients D s + and D s − of cations and anions respectively. The Nernst-Einstein relation becomes approximated at high salt concentration, where the effect of ion-ion cross-correlation starts to show up. 11,12, In this case, the ionic conductivity can be formally defined by the Green-Kubo (G-K) formula: 20 = lim where J is the total current density and P is the itinerant polarization in ionic solutions 21 or the Berry phase polarization in solids. 22,23 . 15,18 The name "distinct" emphasizes the nature of crosscorrelation between different ions either in the same species or in different species. Subsequently, this allows to rewrite the Green-Kubo conductivity as where σ d ++ , σ d −− and σ d +− are the distinct ionic conductivities from corresponding distinct diffusion coefficients. Deviations from the Nernst-Einstein relation, i.e. (σ N−E − σ G−K ), can be quantified by either experiments or MD simulations. In experiments, they can be obtained as the difference between pulsed-field gradient NMR and impedance spectroscopy measurements for the same system in the same conditions. 5,10,24,25 In MD simulations, one can compute σ G−K using either Eq. 3 or Eq. 4 and σ N−E with selfdiffusion coefficients obtained from either velocity autocorrelation functions Eq. 7 or mean squared displacement Eq. 8. 20 = lim where t is the time, N is the the number of cations or anions in solution, α ∈ {+, −} v i,α is the velocity of ith cation or anion, r i,α is the corresponding position. It is worth to note that MD results of σ N−E obtained by computing self-diffusion coefficients D s + and D s − contain a significant finitesize error because of the hydrodynamic selfinteraction in periodic systems. 26,27 To obtain the corrected self-diffusion coefficients D s α (L → ∞) which is system-size independent, the following formula can be applied: where D s α (L) is the self-diffusion coefficient obtained from Eq. 8 with the box length L, ξ is about 2.837297 for cubic simulation boxes 27 and η is the shear viscosity. ## Model Systems and Molecular Dynamics Simulations Following the spirit of using the ideal potassium chloride (KCl) line for the Walden plot, 3 here we took sodium chloride (NaCl) electrolyte solution as a prototype system for aqueous electrolytes. Water molecules were described by the simple point charge/extended (SPC/E) model 28 and Na + /Cl − ions were modelled as point charge plus Lennard-Jones potential using the parameters from Joung and Cheatham, 29 which is suitable for highly concentrated solution. Stoichiometry of three different simulation boxes (Large, medium and small) is listed in Section A of the Supporting Information. in The molecular dynamics simulations were performed with the LAMMPS code. 33 The size of cubic simulation box was determined by experimental densities. 34 The long-range electrostatics was computed using the particle-particle particle-mesh (PPPM) solver. 35 Short-range cutoffs for the Van der Waals and Coulomb interaction in direct space are 9.8 . For computing ionic conductivities, N V T (constant number of particles, constant volume and constant tempera-ture) simulations ran for 20 ns with a timestep of 2 fs, trajectories were collected every 0.5 ps. The Bussi-Donadio-Parrinello thermostat 36 was used to keep the given temperature 20 • C. Because [BMIM][PF 6 ] system does not show permanent ion-pairing, 37 so we picked up this model system in our investigations of ILs here. The interaction model and parameters derived from OPLS-based force field 38 for ILs (OPLS-2009IL) 39 are used for the [BMIM][PF 6 ] system. A charge scaling factor of 0.8 e was applied to account for the electronic polarization effects, 40 which was shown to improve the prediction of self-diffusion coefficients. 41 Stoichiometry of three different simulation boxes (Large, medium and small) is listed in Section B of the Supporting Information. Shortrange cutoffs are 13 for [BMIM][PF 6 ] system. N P T (constant number of particles, constant pressure and constant temperature) simulations ran for 100 ns with a timestep of 1 fs and trajectories were collected every 0.5 ps. The Bussi-Donadio-Parrinello thermostat and the Parrinello-Rahman barostat 42,43 was used to to keep the selected temperatures constant and the pressure at 1.0 atm. ## System-size dependence of the Green-Kubo conductivity As shown in Fig. 1, we found that the ionic conductivities computed using the Green-Kubo formula show no system-size dependence. Such characteristic is simlar to that of the viscosity η which is also a system-size independent quantity. 27 Despite that there is no obvious reason why this should be the case, note that both the supercell polarization P used for computing the Green-Kubo conductivity and the pressure tensor p used for computing the viscosity are collective properties of the whole system rather than the average of individual particle's properties. With periodic boundary conditions, point charge density and point force density are modified by the compensating background as q i (δ(r − r i ) − 1/Ω) and F i (δ(r − r i ) − 1/Ω) respectively. This gives the supercell polarization and the virial part of the pressure as Considering the mathematical similarity between these expressions, it may not be a total surprise that the resulting Green-Kubo conductivity and the viscosity from the linear response theory have the same system-size dependence. Another angle to look into this problem may be through the connection between the ionic conductivity σ G−K and the Maxwell-Stefan diffusion coefficient D M−S +− . The Maxwell-Stefan diffusion coefficient D M−S +− describes the mutual diffusion between cations and anions which are independent of the reference frame. In binary systems, it is linked to the Green-Kubo conductivity as Recently, it has been proposed that the system-size dependence of the Maxwell-Stefan diffusion coefficient D M−S in molecular binary mixtures follows the expression: 44 Apart from the familiar expression given in Eq. 9, the new ingredient is the inclusion of the thermodynamic factor Γ as a correction. When Γ is significantly larger than 1 which happens when the two species like to associate with each other, D M−S becomes effectively system-size independent. What we observed in the case of IL [BMIM][PF 6 ] may fall into this category, where cations and anions attract each other naturally. However, Eq. 10 simply does not hold for the case of binary electrolyte solution (cations, anions and solvent molecules). This makes Eq. 11 not applicable to NaCl solutions. Moreover, the thermodynamic factor calculated from the experimental mean activity coefficient is not much larger than 1 over the whole concentration range of NaCl solutions (see Section C in the Supporting Information), which further indicates that Eq. 11 may not be applicable to explain what was seen in Fig. 1. ## System-size dependence of the distinct conductivities As we found that σ G−K is system-size independent (Fig. 1) and we knew that D s + and D s − (therefore σ s + and σ s − ) are system-size dependent (Eq. 9), these in together imply that some if not all of the distinct conductivities in Eq. 6 should be also system-size dependent. To verify this, we calculated the distinct diffusion coefficient for both NaCl solution and [BMIM][PF 6 ] IL with different box sizes and the results are shown in Fig. 2. It is found that the distinct conductivity σ d ++ of cations (or σ d −− of anions) has a very similar and strong systemsize dependence as that of the corresponding σ s + (or σ s − ) coming from the self-diffusion of ions. These system-size dependencies are more apparent in the case of NaCl solutions than the ]. This is likely due to the fact that the viscosity of NaCl solutions are much smaller than that of [BMIM][PF 6 ] ILs, following the relation in Eq. 9. In contrast, the cation-anion distinct conductivity σ d +− shows little or no system-size dependence. Why do these distinct conductivities have different system-size dependencies? A simple argument would be that this is for the sake of the symmetry. Since there are five terms in Eq. 5, σ d ++ (or σ d −− ) should be paired up with σ s + (or σ s − ) and this leaves σ d +− on its own. In fact, it is not just an intuition. By connecting Onsager's phenomenological transport equations with the linear response theory, 45 Schönert showed that the distinct diffusion coefficient and the selfdiffusion coefficient has the following general re-lation for 1:1 electrolytes: where α ∈ {+, −}, β ∈ {+, −}, Ω αβ are barycentric-fixed Onsager coefficients, N A is the Avogadro constant and δ αβ is the Kronecker delta function. Barycentric-fixed reference frame means the velocity of the center of mass of the system is set to zero, which is most suitable reference frame for MD simulations. For ILs, these Ω αβ coefficients are not independent but follows the expression below be-cause of the conservation of momentum. 45 ) where M + and M 2 are molecular weight of cations and anions respectively. Putting Eq. 6, Eq. 12 and Eq. 13 together, one can arrive at the following expression: This means σ d +− has the same system-size dependence as σ G−K in the case of ILs. Since σ G−K is system-size independent, therefore, σ d +− is also system-size independent. This theoretical prediction is exactly what is shown in Fig. 2f. Subsequently, Eq. 14 also indicates that Ω +− , Ω ++ and Ω −− are all system-size independent quantities. Nevertheless, one needs beware that there is no such simple relation as Eq. 13 for a solution made of simple salt and solvent, e.g. NaCl solutions. Therefore, the similar behavior of σ d +− shown for NaCl solutions in Fig. 2c remains as simulation observations. ## Cross-over box length between the Nernst-Einstein conductivity and the Green-Kubo conductivity It is known that the self-diffusion coefficients have strong system-size dependence (Eq. 9), therefore, one would expect the Nernst-Einstein ionic conductivity has the same tendency. Indeed, it is the case for both NaCl solutions and [BMIM][PF 6 ] ILs as shown in Fig. 3. What is the interesting is that for a small enough simulation box, there exists a crossover box length between the Nernst-Einstein ionic conductivity σ N−E and the Green-Kubo ionic conductivity σ G−K . This is clearly seen in both cases of NaCl solutions and Of course, one would immediately argue that the actual cross-over box length depends on the force-field used even for the same type of systems. However, this is not the question that we will dwell on in this work. Instead, the question what matters here is: Can this cross-over between σ N−E and σ G−K be always achieved ? Supposing that all cation and anions in the system are paired up permanently, then σ G−K will be absolutely zero while σ N−E are not. In other words, if the system has permanent ion-pairing, then the cross-over between σ N−E and σ G−K will never happen. Therefore, we restrict our following discussions to the cases where there are no permanent ion-pairing. ## Implication of the cross-over box length for ion transport in ILs The system-size dependence as discussed in previous sections is usually considered as a finite-size error which needs to be corrected. However, here we turn the tables and use it as a tool instead to investigate the role of viscosity in deviation from the Nernst-Einstein relation. The observation of the system-size independence of σ G−K and the cross-over box length L min implies that: Combining Eq. 1 and Eq. 9, one can get: ) Inserting Eq. 15 into Eq. 16, one arrives at the following expression: Since Λ = σ/c with c as the molar concentration and c = ρ/N A , the above equation in terms of the molar conductivity Λ can be expressed as: Eq. 18 suggests that deviations from Nernst-Einstein relation has a linear relation with respect to 1/η. This is reminiscent of the wellknown Walden rule Λη = k. 46 In other words, it states that the system has a high viscosity will have a small deviation from the Nernst-Einstein relation or vice versa. For the prototype systems NaCl and [BMIM][PF 6 ] used here, L min are about 12.3 and 18.0 respectively. We notice that L min /2 for NaCl solutions is about 6.1 , which is close to the Kirkwood correlation length in bulk liquid water. 47 It is curious to know how the boundaries set by these prototype systems looks like according to Eq. 18 when comparing to experiments. For this purpose, we took a few of seminal experimental works in ILs which promoted the idea of "ionicity" 5,24,25 and made the following mapping: Λ N−E (L → ∞) ↔ Λ NMR and Λ G−K ↔ Λ Imp . This leads to results shown in Fig. 4. Fig. 4a contains 13 different types of ILs measured at 30 • C (See Table 1 in Ref. 5 and the list of names in Section D of the Supporting Information) and in Fig. 4b, the temperature dependence of molar conductivities and viscosities of 6 types of ILs were measured experimentally and fitted to Vogel-Fulcher-Tammann (VFT) equations. 24,25 It is interesting to see that most experimental data and lines fall into the boundaries set by NaCl and [BMIM][PF 6 ] and follow Eq. 18. Since we now know that the value of L min depends on the specific system under the investigation, therefore, this agreement is somehow fortuitous. Furthermore, we notice that the corresponding L min /2 follows the order which is in reverse to the alkyl chain length. This excludes the option that L min /2 is related to the hydrodynamics radius of ions. Instead, it suggests that L min /2 should be regarded as an effective ion-ion correlation length that goes down as the size of cation becomes larger, following the strength of electrostatic interactions. Alternatively, this trend may come from the reduction of the dielectric constant of corresponding ILs with the increase of the alkyl chain. Verifying these implications should be the topics for future works. Before closing this section, it is necessary to make a connection to the quantities related "ionicity" in ILs. For example, the deviation in "ionicity" ∆. 10 According to the Stokes-Einstein relation, Λ N−E can be expressed as follows: where r + and r − are the hydrodynamic radii for cations and anions respectively. r is the mean hydrodynamic radius. Combining Eq. 18, Eq. 19 and Eq. 20, one arrives at a succinct expression of ∆. As shown in Eq. 21, ∆ does not explicitly depend on the temperature and the pressure. Therefore, one would expect ∆ is close to a constant for one specific system. This agrees with the experimental observations. 11 When ∆ = 1, this implies Λ G−K = 0 and L min is about 3 times of the mean hydrodynamic radius of ions (See text around Eq. 9 regarding the constant ξ). Note that in this limit, Eq. 9 is no longer applicable. 48 This in turn suggests that ∆ = 1 limit will never be met when permanent ion-pairing is not considered. On the other hand, when ∆ = 0, this means that Λ N−E simply equals to Λ G−K regardless of the system size. For the infinite dilute solution, this means the correlation length will diverge and L min → ∞. ## Conclusions The system-size dependence of the Nersnt-Einstein conductivity σ N−E and the Green-Kubo conductivity σ G−K in NaCl solutions and [BMIM][PF 6 ] IL were investigated using MD simulations. It is found that σ N−E is strongly system-size dependent as expected while σ G−K does not depend on the system-size. By analyzing the contributions from the distinct diffusion coefficient, we further showed that σ d +,− between cations and anions has the the same system-size dependence as σ G−K , which is exact for the case of ILs and effective for electrolyte solutions. Due to different system-size dependences in the Nersnt-Einstein conductivity and the Green-Kubo conductivity, there exist cross-over box lengths where these two types of conductivities become effectively the same. This leads to an expression for the deviation from the Nernst-Einstein relation (Λ G−K −Λ N−E ) showing that a low viscosity leads a strong deviation and a high viscosity leads to a weak deviation (for systems without permanent cation-anion associations), following Eq. 18. This new expression was verified against published experimental data of different types of ILs and the system-specific cross-over box length L min may provide a new avenue to gauge the ion-ion correlation in electrolyte system. Future works should focus on extending the current formulation to the cases which contain permanent ion-pairing and investigating the relationship between the hydrodynamic radius of ions, the L min and nano-scale confinement. Swedish National Strategic e-Science program eSSENCE is also gratefully acknowledged. The simulations were performed on the resources provided by the Swedish National Infrastructure for Computing (SNIC) at UPPMAX and PDC. We thank helpful discussions with M. Hellström.

chemsum

{"title": "Role of Viscosity in Deviations from the Nernst-Einstein Relation", "journal": "ChemRxiv"}

influence_of_a_single_ether_bond_on_assembly,_orientation,_and_miscibility_of_phosphocholine_lipids_

## Abstract: How does a small change in the structure of a phospholipid affect its supramolecular assembly? In aqueous suspensions, the substitution of one ester linkage in DPPC (1,2-dipalmitoyl-sn-glycero-3-phosphocholine) by an ether linkage alters its phase behaviour completely. To unravel the effect of replacing a phospholipid's ester linkage by an ether linkage in lipid monolayers, we characterized pure monolayers of the model lipid DPPC and its sn-2 ether analogue PHPC (1-palmitoyl-2-Ohexadecyl-sn-glycero-3-phosphocholine) as well as mixtures of both by measurements of surface pressure -molecular area (π-Amol) isotherms. In addition, we used infrared reflection absorption spectroscopy (IRRAS) to study lipid condensation, lipid chain orientation, headgroup hydration, and lipid miscibility in all samples. Mixed monolayers consisting of DPPC and PHPC were studied further using epifluorescence microscopy. Our results indicate a strong influence of the sn-2 ether linkage on headgroup hydration and ordering effects in the regions of the apolar chains and the headgroups. Both effects could originate from changes in glycerol conformation. Furthermore, we observed a second plateau in the π-Amol isotherms of DPPC/PHPC mixtures and analysis of the mixed π-Amol isotherms reveals a non-ideal mixing behaviour of both lipids which may be caused by conformational differences in their headgroups. ## Introduction In nature, ether lipids are highly abundant in Archaea. 1 Many representatives of this domain of life are extremophiles, needing extraordinarily stable membranes to withstand their extreme habitat conditions, e.g. at extremely low pH, high temperature, or high ionic strength. 1 To maintain a functional membrane structure, ester lipids are not suitable. In this environment, chemically more stable ether lipids as well as membrane-spanning bipolar tetraether lipids (TELs), also known as bolaamphiphiles, 2 have evolved. 3,4 In biophysics of lipids, it is well known that ether and ester lipids exhibit different phase behaviours. In aqueous suspensions of some ether lipids, a complex thermotropic polymorphism is observed. 5 When comparing 1,2-dipalmitoyl-snglycero-3-phosphocholine (DPPC) with its mono-or di-ether derivatives, namely 1-palmitoyl-2-O-hexadecyl-sn-glycero-3phosphocholine (PHPC) and 1,2-di-O-hexadecyl-sn-glycero-3phosphocholine (DHPC), respectively, significant differences in gel phase structures are found (see the Supporting Information for the molecular structures). While DPPC forms crystalline (Lc), lamellar gel (Lβ'), and rippled gel (Pβ') phases at temperatures below the main transition temperature (Tm) of 41.6 °C, PHPC and DHPC form a lamellar interdigitated gel phase LβI (and DHPC an additional Lβ phase between 34.8 and 43.9 °C) at temperatures below Tm. 5 In PHPC and DHPC, one or two small changes in the region linking the alkyl chain and headgroup, i.e. the glycerol backbone, cause major structural modifications of their supramolecular assemblies in comparison with DPPC. Lewis et al. studied the reasons for the differences between these otherwise structurally identical lipids extensively using infrared (IR) spectroscopy and isotopic labeling. 5 They concluded that the exchange of at least one ester bond with an ether bond induces a conformational change of the involved glycerol backbone, altered hydration of the remaining carbonyl moiety, and conformational changes of the adjacent chain segment. 5 Hence, a change in glycerol orientation caused by as little as the substitution of one ester bond with an ether bond enables the lipid to preferably aggregate in a lipid gel phase showing alkyl chain interdigitation at temperatures below Tm. Consequently, in aqueous suspensions the effects of ether bonds on lipid properties are understood quite well. In contrast, differences between DPPC and PHPC in Langmuir monolayers are far less studied and understood. Surface pressure -molecular area (π-Amol) isotherms of DPPC, PHPC, and DHPC were already measured and characterized by fluorescence microscopy and X-ray diffraction. Brezesinski et al. observed a decreased chain tilt angle of DHPC and PHPC, in comparison to DPPC. Moreover, the lateral lipid density of both ether lipids is increased compared to DPPC monolayers. 6 They predicted a change in glycerol conformation to be responsible for these differences and the hydration of the headgroup to Germany. Electronic Supplementary Information (ESI) available: Synthetic details, further IRRAS data, simulation of IRRA spectra, further PCA results and thermodynamic data. Please do not adjust margins Please do not adjust margins change. 6 However, this remains speculation until now and little is known about structure, conformation, and hydration of the glycerol backbone of ether lipids in monolayers. Here, we first investigate the phase behaviour, lipid conformation, and hydration of DPPC and PHPC monolayers, respectively. We present π-Amol isotherms combined with infrared reflection absorption (IRRA) spectra to detect how monolayers of both lipids differ in their molecular changes caused by compression of the monolayer. We evaluate the chain order parameter (S(CH2)) and the order parameter of the phosphate group (S(PO2 -)) by performing a least square minimization of both methylene stretching vibrational bands and the antisymmetric phosphate stretching vibrational band, respectively. To characterize the hydration of the carbonyl moiety of both lipids, we use principal component analysis (PCA). In the second part of this study, we report the characterization of mixed monolayers of DPPC and PHPC, in particular their miscibility at the air-water interface, by epifluorescence microscopy using a rhodamine-labelled lipid, the surface phase rule, and calculation of the excess Gibbs energy of mixing (ΔGexc) and the Gibbs energy of mixing (ΔGmix). 9,10 In addition, we present IRRA spectra of mixtures of PHPC and DPPC-d62, bearing fully deuterated alkyl chains, to characterize their miscibility spectroscopically. ## Experimental Materials 1,2-Dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) was purchased from Genzyme Pharmaceuticals (Cambridge, MA, USA) and used without further purification. 1-Palmitoyl-2-Ohexadecyl-sn-glycero-3-phosphocholine (PHPC) was synthesized as described in the Supporting Information. DPPC bearing perdeuterated acyl chains (DPPC-d62) was purchased from Avanti Polar Lipids Inc. (Alabaster, AL, USA). The fluorescent dye 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (ammonium salt) (Rh-DPPE) was obtained from Life Technologies GmbH (Darmstadt, Germany). The solvents chloroform and methanol (HPLC-grade) were obtained from Carl Roth (Karlsruhe, Germany). ## Methods Sample preparation. All used lipids were dissolved in chloroform/ methanol (9/1, v/v). Mixtures of phospholipids were prepared by mixing different amounts of appropriate lipid stock solutions using glass syringes (Hamilton Bonaduz, Bonaduz, Switzerland). For storage, solvents were then removed in a gentle stream of nitrogen. Directly prior to experiments, the lipids and lipid mixtures were dissolved again in chloroform/methanol. Monolayer preparation. All π-Amol isotherms were measured on a rectangular Teflon trough (78.3 × 6.8 cm 2 Riegler & Kirstein, Potsdam, Germany) except for compression isotherm measurements with parallel IRRAS detection. The Langmuir trough was equipped with a Wilhelmy sensor and filled with H2O (MilliQ Millipore water with a specific resistance of ρ = 18.2 MΩ cm). After the trough was filled with water, the freshly dissolved lipids or lipid mixtures were carefully spread on the surface with a glass syringe (Hamilton Bonaduz, Bonaduz, Switzerland). The solvents were allowed to evaporate for at least 10 min prior to each measurement. Monolayer compression. All shown π-Amol isotherms were compressed at the air-water interface by Teflon barriers moving at a compression speed of 2 2 (molecule min) −1 . During the measurements, 40 points per seconds were averaged. In all measurements, the temperature of subphase and monolayer was kept constant (accuracy ∆T ± 0.2 K) through a coupled water-cooling system operated at 20 °C. Each isotherm was measured at least three times, from which one representative isotherm is shown. Analysis of Langmuir isotherms. From each measured isotherm, the monolayer compressibility CS was evaluated: to determine the transition surface pressure πplateau from the maximal value of CS. The mixing behaviour of the lipids was studied as follows. Thermodynamically, the Gibbs energy of mixing ΔGmix describes whether the components of a mixture are miscible or not. It can be split in an ideal and an excess, i.e. non-ideal, term: . ΔGid only depends on entropy, specifically on mixture composition, that is the mole fraction xi, and temperature. This value is always negative, which means that, ideally, all components are miscible regardless of composition: However, demixing can occur. Deviations from ideal behaviour are included in ΔGexc. In two-dimensional systems such as Langmuir monolayers, deviations from ideal mixing are associated with an excess areas Aexc. The integration of (4) was conducted from π = 0 mN m -1 , starting in the gaseous/liquid-expanded (LE) transition region. Combination of ΔGid and ΔGexc results in: ## Infrared reflection absorption spectroscopy (IRRAS). According to a procedure described elsewhere, 11,12 IRRAS experiments were performed on a Bruker Vector 70 FT-IR spectrometer equipped with an A511 reflection unit (Bruker Optics, Ettlingen, Please do not adjust margins Please do not adjust margins Germany), a liquid nitrogen-cooled MCT detector, and a Langmuir trough system (Riegler & Kirstein, Potsdam, Germany). The trough system consists of a rectangular sample trough (30 × 6 cm 2 ) and a small circular reference trough (r = 3 cm). Surface pressure in the sample trough was detected with a Wilhelmy sensor using a filter paper as the pressure probe. During the measurements, the filling levels of both troughs were kept equal and constant by means of an automated, laser-reflection-controlled pumping system connected to a reservoir of H2O. Prior to collection of each monolayer IRRA spectrum, a spectrum of the pure subphase was measured with identical conditions to ensure best comparability and effective water vapor compensation. Two general types of experiments were performed in this work: (i) IRRAS measurements at constant angle of incidence and polarization during compression of the monolayer and (ii) angle-and polarization-dependent IRRAS measurements to reveal orientations of different molecular moieties. For (i), IRRA spectra were collected at a constant angle of incidence φ = 60° and with s-polarized IR light. 1000 single interferograms were collected, averaged, and, subsequently, Fourier-transformed with a zero-filling factor of 2 to obtain one spectrum with a nominal spectral resolution of 2 cm -1 . Here, five of these individual spectra were averaged to obtain one final spectrum to ensure excellent signal to noise ratio. For (ii), we varied both, angle of incidence (25 to 70° in increments of 3°) and polarization (s-and p-polarization) of the incoming IR beam. Either 1000 (s-polarized IR light) or 2000 (p-polarized IR light) interferograms were averaged and Fourier-transformed with the same parameters as described above. At least three of the resulting spectra were averaged to acquire the shown spectra. The π-Amol isotherms of DPPC and PHPC, respectively, were halted at four different surface pressures at 3, 10, 20, and 30 mN m -1 ) and sets of polarization-and angle-dependent IRRA spectra were measured at each pressure, which allows comparison of different compression states of both lipids. Simulation and fitting of the experimental angle-and polarization-dependent IRRA spectra was conducted as explained in detail elsewhere. Principal component analysis (PCA). The goal of a principal component analysis (PCA) is to unravel subtle changes or relations in large datasets and to simplify the data. To this end, the intercorrelated variables of the data are transformed into principal components (PCs) which are orthogonal and linear combinations of the original variables. For further mathematical details we refer to literature. 16 In this work, we present PCA of several IR-bands of DPPC and PHPC in monolayers, as measured with IRRAS. All spectra used in the PCA were measured in s-polarization at angles of incidence of the IR beam ranging from 25° to 70°. Since the band shape of s-polarized IRRA spectra is independent of orientation of the absorbing group they can be used directly for PCA. 17 The aim of this analysis is to correlate spectral changes with the phase state of the lipid and to obtain information about differences between both lipid species. We have chosen the spectral range of the carbonyl stretching vibrational band (1700 -1775 cm -1 ) due to its sensitivity to hydration of the polarapolar interfacial region of the lipids. 18 We measured IRRA spectra of the pure H2O subphase directly before collecting the monolayer spectra with identical measurement parameters. These water reference spectra were subtracted from each spectrum to reduce spectral contributions of the water vapor vibrational-rotational bands. The subtraction factor was determined by minimizing the variance of the second derivative of the spectrum in the spectral range of 3500 -4000 cm -1 . Since our aim was PCA of the carbonyl bands which overlap with the water deformation band γ(H2O), we additionally subtracted simulated water absorption bands of the measured monolayer to ensure minimal spectral contributions of the γ(H2O) band. Both subtractions were performed with home-written MATLAB scripts (MathWorks Inc., Natick, MA, USA). Prior to PCA, a linear baseline was subtracted from the pretreated spectra and vector normalization of the carbonyl bands of all spectra were conducted. This ensures exclusion of major intensity differences between the bands from the PCs. The PCA was performed using the princomp function of MATLAB (MathWorks Inc., Natick, MA, USA). Epifluorescence microscopy. As described previously, 11 fluorescence images of monolayers being composed of DPPC, PHPC, and their mixtures, respectively, were recorded with an Axio Scope A1 Vario epifluorescence microscope (Carl Zeiss MicroImaging, Jena, Germany). Imaging was conducted during compression of the monolayer, which was performed as described before, allowing full control of the pressure status when the images were taken. A Teflon-coated trough (26.6 × 9.9 cm 2 ) equipped with a Wilhelmy balance (Riegler & Kirstein, Potsdam, Germany) was mounted below the microscope on an x-y-z stage (Märzhäuser, Wetzlar, Germany) which was controlled by a MAC5000 system (Ludl Electronic Products, Hawthorne, NY, USA). During measurements, a homebuilt Plexiglas hood covered the film balance. The microscope was equipped with a mercury short arc reflector lamp HXP 120 C, a long working distance objective (50x magnification, LD EC Epiplan-NEOFLUAR), and a filter/beam splitter combination which was appropriate for the used Rhodamine dye (all components from Carl Zeiss MicroImaging, Jena, Germany). To measure fluorescence, 0.2 mol% Rh-DPPE was added to the lipid solutions, before spreading the monolayers. This fluorescence dye partitions preferentially in lipid LE phases, which leads to a brightness contrast in phase separated monolayers. 10,19 ## Results and discussion The aim of this study is to compare monolayers of pure DPPC and PHPC, respectively, and to characterize their miscibility. First, we show how both lipids self-organize individually at the Please do not adjust margins Please do not adjust margins air-water interface using IRRAS parallel to measurements of the respective π-Amol isotherms. Secondly, we present insights into mixed monolayers of DPPC and PHPC by measuring π-Amol isotherms as a basis for subsequent thermodynamic analyses, and by performing epifluorescence measurements and IRRAS of the mixed monolayers. ## Monolayers of pure DPPC and pure PHPC To answer the question whether a small change in chemical structure-that is replacing one ester linkage between the glycerol and the sn-2 chain with an ether linkage-leads to measurable changes in lipid behaviour in monolayers, we performed π-Amol isotherm measurements. As can be seen in Figure 1, there are only small differences between π-Amol isotherms of DPPC and PHPC. The only noteworthy differences are the slightly increased phase transition pressure of PHPC (4.42 mN m -1 for DPPC vs. 6.05 mN m -1 for PHPC at 20 °C) and the overall higher compressibility of PHPC in the liquid-condensed (LC) phase compared to DPPC, as deduced from the decreased slope of the isotherm in the LC phase region. An increased transition pressure of PHPC, as measured in this work, is in accordance with literature reports 8 while the difference in LC phase compressibility could be due to different pressure-dependent ordering of either lipid. A spectroscopic IRRA analysis was coupled to the isotherm measurements, with the aim to obtain more detailed insights into the organization of the lipids in the monolayer and to detect structural differences between DPPC and PHPC. Both lipids were previously characterized in aqueous suspensions using IR spectroscopy. 5,20 It was concluded that the main spectroscopic differences between both lipid bilayers are due to different orientations of the glycerol backbones. This effect seems to result in different hydration of the carbonyl moieties as well as conformational changes in adjacent methylene segments. 5,20 Furthermore, from X-ray diffraction data and π-Amol isotherms of both lipids, Brezesinski et al. suggested different glycerol orientations in monolayers as well. 6 In our study, we consequently first evaluated chain ordering and headgroup hydration in monolayers from IRRAS data. For evaluating the order in the apolar lipid chain region, both symmetric and antisymmetric CH2 stretching vibrations were measured and simulated (see Figure S1 and Figure S2). Further details of the simulation procedure can be found elsewhere. The band position of both CH2 stretching vibrations are indicative of the trans/gauche ratio in alkyl chains. 17,21 Therefore, they are widely used to monitor phase transitions connected to chain melting. Simulation of angle-and polarization-dependent IRRA bands allows evaluating the order parameter S(CH2) of the whole all-trans chain if the orientation of the transition dipole moment is known. This order parameter can be translated into hydrocarbon chain tilt angle. The wavenumbers of νs(CH2) and νas(CH2) measured in this study are shown in Figure 1B (νs(CH2) as a function of surface pressure) and, additionally, in Figure S1 and Figure S2 together with the simulation results. The derived order parameters of the lipid chains as well as the resulting chain tilt angles in the LC phase are plotted in Figure S3 and included in Table 1. To interpret changes in νs(CH2) and, hence, changes in the trans/gauche ratio, we plotted νs(CH2) versus π during monolayer compression in Figure 1B (scattered data). We only show the frequency of νs(CH2) because νas(CH2) contains overlapping contributions from CH3 group vibrations and a Fermi resonance band. | 5 Please do not adjust margins Please do not adjust margins Our data suggest that the trans/gauche ratio of both lipids is similar since νs(CH2) are identical at similar π. We also found a comparably disordered LE phase in both lipid monolayers which becomes more ordered after phase transition to the LC phase. Directly after phase transition, at 10 mN m -1 , S(CH2) of both lipids are similar. However, during further compression, differences between DPPC and PHPC arise as PHPC forms more ordered monolayers above 20 mN m -1 . In the LC phase, the chains of both lipids are in all-trans conformation as concluded from the CH2 stretching frequencies (see Figure 1B, scattered data). 17,22 Therefore, we calculated the tilt angle θ of the fully stretched lipid chains, which is also shown in Table 1. PHPC possesses a smaller tilt angle with respect to the surface normal than DPPC, which is in excellent accordance with x-ray data. 6 In addition, θ of PHPC decreases further during LC phase compression as opposed to DPPC. This continuous film reorganization upon PHPC LC phase compression explains the higher compressibility of PHPC as compared to DPPC in the condensed phase as shown in Figure 1B (inset, solid lines). Additional interpretation of deformation bands of the methylene groups is typically conducted to get insights into ordering effects of the chains and to elucidate coupling with other moieties. Evaluation of the CH2 scissoring vibrational band (δ(CH2)) enables us to detect the geometry of the lipid unit cell in the monolayer. 17,21 For the two studied lipids, we found the frequency of the δ(CH2) = 1468.9 cm -1 (at 30 mN m -1 ) for both DPPC and PHPC in the LC phase (see Figure S4) while this band is not visible in the fluid LE phase. Both band positions in the LC phase are indicative of a hexagonal lattice which is in accordance with X-ray diffraction data. 6,17 During LC phase compression, the value slightly decreases, but no difference between both lipids could be detected. Another methylene deformation band, the CH2 wagging band progressions, will be discussed later in this work in combination with the antisymmetric phosphate vibration. Headgroup vibrations of the characterized lipids contain information on ion binding and hydration of their polar moieties. Typical IRRA bands originating from headgroup vibrations of both lipids are the carbonyl stretching vibrational band ν(C=O) and the antisymmetric phosphate stretching vibrational band νas(PO2 -). Both are shown in Figure 2. As can be seen directly, a small frequency decrease of the ν(C=O) band from DPPC to PHPC was found that is significantly less pronounced than the shift found in the bulk system. 5,20 As opposed to aqueous suspensions, where complete subtraction of solvent spectra is possible, IRRA spectra inherently contain subphase contributions. Thus, the negative ν(C=O) overlaps with the positive water deformation band (γ(H2O), see Figure 2A) which could hinder a direct band interpretation and shifts the C=O band minima to higher wavenumbers. To circumvent this issue, we simulated the water absorption bands for each spectrum and subtracted them from the original data to yield spectra that are nearly free from water absorptions (see Experimental). Subsequently, we performed PCA of the corrected carbonyl bands to analyse a large dataset of spectra recorded at various pressures and angles for both lipids. The frequency of ν(C=O) can be interpreted in terms of hydration of the carbonyl group. 5,18,20 In general, increasing hydration shifts the band centre to lower wavenumbers. However, it is not possible to derive the number of bound water molecules directly without knowing the exact absorption coefficients of the dehydrated, monohydrated, and dihydrated species, respectively. In addition, hydrogen bonding by more water molecules and stronger hydrogen bonding are indiscernible from IRRAS frequency shifts. 23 Still, in aqueous suspensions, different subcomponents of the ν(C=O) band are interpreted as different hydration states of the carbonyl group containing distinct numbers of bound water molecules. 18 The spectra shown in Figure 2A are therefore indicative of differences in hydration of the headgroups between both lipids. However, a more precise analysis is necessary due to different influence of the overlapping γ(H2O) band on the C=O bands of both lipids. In contrast to IR measurements of aqueous suspensions, we are not able to detect whether two or three subcomponents 20 are included in the lipids' carbonyl stretching vibrational band because of the overlap of the ν(C=O) band with the water deformation band. Please do not adjust margins Please do not adjust margins Using IRRAS, we evaluated the ν(C=O) bands of both pure lipid monolayers at four distinct surface pressures, namely at 3 mN m -1 in the LE phase and at 10, 20, and 30 mN m -1 in the LC phase. For each surface pressure, we performed angle-and polarization-dependent IRRA measurements that allow fitting the data to unravel conformational differences between both lipids and subtraction of the water absorption bands, which only depend on monolayer thickness, monolayer refractive index, and the quality of the used polarizer. The latter was determined empirically from all measurements with the used polarizer to be  = 0.007. The monolayer refractive index of phospholipids is known from literature (n = 1.41). 24 Monolayer thickness as the only remaining parameter can be derived from fitting the theoretical subphase water absorption bands to the measured data. After subtracting the simulated water bands from all IRRA spectra, we conducted a PCA of the vector-normalized carbonyl bands of all s-polarized spectra of both lipids recorded at various angles of incidence (25 -70°) and surface pressures. The results of the PCA are shown in Figure 3 The interpretation of spectral PCA is not necessarily straightforward. At first, one must interpret the resulting loadings of the principal components (PCs). PC 1 and PC 2 (Figure 3A) account for approximately 90 % of all differences in the ν(C=O) dataset and, hence, we limit our interpretation to these PCs. The loading of PC 1 reflects a spectral shift of the band minima from high to low frequency. Thus, higher scores on PC 1 are indicative for a higher hydration of the interfacial carbonyls, which is not visible as clearly in the averaged spectra without PCA (compare to Figure 2A). The second PC, in contrast, seems to result mainly from atmospheric water vapor and experimental noise. It is an advantage of the PCA that these disturbing contributions are separated from the systematic variations mapped through PC 1. To interpret the differences between DPPC and PHPC, the score of PC 1 versus the surface pressure is shown in Figure 3B as a box plot, were the individual boxes contain contributions of IRRA spectra measured at various angles of incidence. Figure 3B shows a clear separation of the scores of DPPC and PHPC on the PC 1 axis at all examined surface pressures, indicating different hydration of their respective carbonyl moieties. Additionally, within each subset of data, the scores on the PC 1 decrease as well with increasing compression of the monolayer. For both lipids, this shift is most pronounced between 3 and 10 mN m -1 , corresponding to the LE/LC phase transition. Within the LC phase, this shift is smaller; however, it is more pronounced for PHPC than for DPPC. This reflects once more the ongoing ordering of PHPC upon LC phase compression corresponding to the chain order parameter and compressibility, respectively, reported above. The scores of PC 2 versus PC 1 are shown in Figure S5. No systematic changes are visible in the second PC, which matches the assumption of stochastic causes for the PC 2. The angle of incidence did not affect the score of any spectrum on the PC 1 as can be seen in Figure S6. However, the score of PC 2 depends on the angle of incidence, which, in turn, is due to the dependence of surface reflectivity on φ. This means the reflectivity and, hence, the signal-to-noise ratio increases for spolarized IR light with increasing angle of incidence. 17 As can be interpreted from Figure 3, the PCA yields two results: a) at all surface pressures, the frequency of the ν(C=O) band of PHPC is decreased compared to DPPC and b) for both lipids, the frequency of the ν(C=O) band increases upon compression. Since a frequency difference of the carbonyl stretching vibration between DPPC and PHPC reflects different hydration of the carbonyl group, we conclude that the carbonyl moiety of PHPC is either more hydrated (more bound water molecules) or the existing water molecules are bound more tightly to the carbonyls at the interface when compared to DPPC. This finding at first glance seems counterintuitive, as the substitution of an ester linkage between the glycerol and the sn-2 chain with an ether should result in a slightly more apolar headgroup region in comparison to DPPC. However, this seems not to decrease the hydration of the remaining carbonyl, but rather increases it. ## ARTICLE | 7 Please do not adjust margins ## Please do not adjust margins With respect to the literature, this may be interpreted as a change in glycerol orientation from approximately perpendicular to the water surface in DPPC to parallel to the water surface in PHPC-similar to findings in aqueous (bulk) suspensions. 5 An altered glycerol orientation in the gel phase of PHPC in bulk is connected to the formation of an interdigitated gel phase (LβI). In contrast, in monolayers an interdigitated arrangement is obviously not possible. To support this hypothesis, we also evaluated the CH2 wagging band progressions (1260 -1262 and 1243 cm -1 ) of both lipids in monolayers, which overlap with the antisymmetric phosphate stretching vibrational band, νas(PO2 -) (1221 -1226 cm -1 ), of the headgroup. 25 The CH2 wagging band progressions are sensitive to single gauche conformers near the carbonyl groups of the lipid in the LC phase, because the intensity of the wagging band progressions increases significantly with coupling to the carbonyl group. 22 Adjacent kinks within the alkyl chain, i.e. the presence of gauche conformers, prevent this coupling resulting in attenuation of the CH2 wagging band progressions. No band progressions of the CH2 wagging vibration are therefore observed in fluid LE phases. 22 When comparing PHPC with DPPC, this effect is superimposed by the attenuation of these bands due to removal of one carbonyl group in PHPC. However, in DPPC bilayers, the CH2 wagging band progressions almost exclusively originate from the sn-1 chain, since the sn-2 chain includes a kink adjacent to the carbonyl group. 5 Therefore, if PHPC adopts a similar conformation of the glycerol backbone and the sn-1 ester linkage compared to DPPC, no or only little attenuation should occur. However, the CH2 wagging band progressions are significantly attenuated in bulk, which was considered by Lewis et al. as an additional argument for different glycerol orientations when comparing DPPC and PHPC. 5 Our results in monolayers, as shown in Figure 2B, expose an almost complete vanishing of the CH2 wagging band progressions of PHPC as compared to DPPC independent of the lipid phase. This must be caused by the introduction of one or more gauche conformers adjacent to the carbonyl moiety of the sn-1 chain. 22 Since the glycerol orientation in the LC phase of typical diester phosphocholines does not induce this gauche conformer 26 and measured spectra, thus, show significant coupling of carbonyl group and CH2 wagging vibrations, 5 we interpret this spectral difference of PHPC again in terms of a different glycerol orientation. By comparing our monolayer studies with literature-based knowledge from aqueous suspensions, 5 it becomes obvious that the attenuation of the CH2 wagging band progressions is remarkably more pronounced in monolayers than in bilayers. This can be interpreted as the sn-1 carbonyl moieties and the sn-1 chains having different orientations towards each other in bulk and in the monolayer, respectively. Additionally, all molecules in the monolayer are probably more uniformly arranged when compared to the bulk system, i.e., one adjacent gauche conformer is introduced in all molecules in the PHPC monolayer. 22 However, these differences are expected because the typical arrangement of lipid molecules in monolayer LC phases is by no means the same as in interdigitated gel phases in bulk. It is remarkable that a change as small as substitution of one carbonyl with a CH2 group has a similar effect on the glycerol backbone conformation as changing the whole chain position from sn-2 to sn-3 and, thus, forcing the glycerol to be oriented parallel to the bilayer surface, as found in 1,3-dipalmitoylglycerophosphocholine (1,3-DPPC, β-DPPC) 27,28 and 1,3-diamidophospholipids, respectively. 29,30 The absence of CH2 wagging band progressions in the LE phase of both lipid monolayers allows direct comparison of the νas(PO2 -) bands. This band of the polar headgroup is sensitive to ion binding as well as hydrogen bonding. 17,21 Since we used deionized water for all experiments, we may neglect contributions from remaining ions. Consequently, frequency shifts of νas(PO2 -) are caused by hydrogen bonding to either water or other lipid molecules. However, the phosphate groups in phosphocholines are known to be proton acceptors while no acidic protons can be donated from phosphocholines (unlike, e.g., phosphoethanolamines). 31 Thus, hydrogen bonds cannot be formed directly with other lipids and must always include hydrating water molecules. In the IRRA spectra of both lipids at 3 mN m -1 , CH2 wagging band progressions are absent as expected, therefore, the frequency of the νas(PO2 -) band could be derived by simulating the spectra using only a single component. The results are summarized in Table 1. From the data it becomes evident that PHPC exhibits a lower νas(PO2 -) frequency than DPPC. Similar to what we found for the carbonyl stretching vibrational band, this again indicates an increased hydration of the PHPC headgroups, i.e. more or stronger bound water molecules. 31,32 By simulating the νas(PO2 -) region including CH2 wagging band progression contributions in the LC phase, it is possible to calculate the spectral components of these overlapping bands directly (see selected data in Figure 4, all spectra in Figure S7 and Figure S8). For simulation, we used literature values for CH2 wagging band progression frequencies of DPPC in aqueous suspensions as starting parameters 25 and adjusted them in a nonlinear least square fit. However, the frequencies fitted to our measurements did not deviate more than 5 cm -1 from literature values despite being measured in LC phase monolayers instead of gel phase bilayers. Especially the evaluation of the order parameter S(PO2 -) from νas(PO2 -) is promising as it can be correlated with the headgroup ordering. The determined order parameters of the νas(PO2 -) are shown in Figure S9 and Table 1. We simulated S(PO2 -) with respect to the axis defined by the transition dipole moment of νas(PO2 -) (α = 0°) 15 to evaluate only the orientation of the phosphate group itself. S(PO2 -) of both lipids depends on the lipid phase and our data clearly show differences between DPPC and PHPC. The phosphate groups of DPPC monolayers are less ordered, i.e., S(PO2 -) is smaller when compared to PHPC, regardless of the monolayer phase state. While DPPC maintains similar S(PO2 -) values over the whole compression range, PHPC shows a jump during phase transition from a higher magnitude of S(PO2 -) in the LE phase to a lower magnitude of S(PO2 -) in the LC phase. Beyond the LE/LC phase transition, both lipids exhibit a decreasing S(PO2 -) during further compression of the condensed monolayer, with PHPC covering a larger range of ordering. As can be seen in Figure S9, S(PO2 -) of PHPC decreases significantly during compression of the LC Please do not adjust margins Please do not adjust margins phase. However, this effect should be interpreted cautiously as only three order parameters were measured in this lipid phase. All the observed differences between DPPC and PHPC regarding their S(PO2 -) are associated with the different glycerol orientation of both lipids, which must be caused by the comparably small chemical change of the sn-2 linkage. With respect to all the results regarding the phosphate group, we conclude that an increased hydration of the PHPC phosphate in the LE phase goes along with a higher degree of headgroup ordering. These effects do not occur in DPPC monolayers and may therefore be related to different arrangements of glycerol backbones when DPPC and PHPC are compared. Presumably, more or strongly bound water molecules increase the ordering of the phosphate moiety in PHPC monolayers. Furthermore, while in DPPC monolayers significant overlapping of CH2 wagging band progressions with νas(PO2 -) impede interpretation of separate bands, the corresponding bands in PHPC are attenuated to such a degree that they are hardly observable at all. This, firstly, supports our interpretation of a glycerol backbone oriented parallel to the surface and, secondly, allows direct interpretation of the νas(PO2 -) band. In summary, the substitution of the sn-2 ester bond in DPPC by an ether bond in PHPC leads to: • a different ordering of the lipid alkyl/acyl chains while maintaining the overall phase behaviour, • a rearrangement of the glycerol backbone to a presumably parallel orientation with respect to the water surface connected with the introduction of (at least one) gauche conformers in the sn-1 alkyl chain adjacent to the carbonyl moiety, and • a higher degree of headgroup ordering combined with increased hydration of both the carbonyl and the phosphate group. The resulting different molecular orientations of DPPC and PHPC molecules at the air-water interface are schematically depicted in Scheme 1. ## Mixed monolayers of DPPC and PHPC In the second part of this study, we focus on the mixing behaviour of DPPC and PHPC in monolayers at the air-water interface. For this purpose, we measured π-Amol isotherms of mixed monolayers containing both lipids at different mixing ratios at 20 °C and, simultaneously, performed epifluorescence ## ARTICLE | 9 Please do not adjust margins Please do not adjust margins microscopy of these mixed monolayers. Furthermore, we measured IRRA spectra of the corresponding mixtures of PHPC and DPPC-d62 which is the chain-perdeuterated analogue of DPPC, to obtain spectral selectivity between the two lipids. The π-Amol isotherms of mixtures containing 90 to 25 % DPPC and 10 to 75 % PHPC are shown in Figure 5A together with the isotherms of the pure compounds. In the isotherms of the mixtures, two plateaus or kinks are observable, the upper of which does not exist in isotherms of the pure substances. Therefore, the question arises, whether this plateau is connected to a phase transition or not. This issue can only be elucidated using additional techniques and will be discussed later. By evaluating the compressibility maxima (see Figure S10) of all films, the surface pressure values at both plateaus were determined. These values are plotted in Figure 5B in form of a partial phase diagram and are further shown in Table 2. The midpoint of the lower transition is always located at the transition pressure of pure PHPC, whereas the upper plateau pressure increases with increasing PHPC content, which also has to be discussed in terms of phase transitions, below. To further characterize the phase behaviour of the mixed monolayers, epifluorescence microscopy was performed using Rh-DPPE as fluorescent dye. The obtained micrographs are shown in Figure 6 (including pure DPPC (x(PHPC) = 0), top row, and pure PHPC (x(PHPC) = 1), bottom row, monolayers) and additionally in Figure S11-S16. Typically, Rh-DPPE dissolves readily in LE-phase monolayers while it is excluded from LCphase domains. 10,19 As can be seen in column (c) of Figure 6, this is true for a pure DPPC monolayer, where LC domains appear black. 19 In contrast, in PHPC-containing monolayers, Rh-DPPE can partition into LC domains leading to a lower contrast between the LE and LC phase. Moreover, the shape of DPPC and PHPC LC domains is remarkably different. Therefore, changes in LC phase composition can be unravelled by contrast and shape of the observed domains in mixed monolayers of DPPC and PHPC. First, epifluorescence microscopy of the mixed monolayers enables us to determine, whether (partial) demixing occurs in the LE and/or the LC phase of the mixed monolayers. As it is evident from Figure 6 and Figure S12-S15, the LE phases of all measured mixtures are uniformly bright. This leads to the conclusion that no phase separation occurs below the lower plateau of the respective mixture. At surface pressures above this plateau, the micrographs of all DPPC/PHPC mixtures appear uniformly grey indicating the existence of only one homogeneously mixed phase (column (e) in Figure 6). No further changes were detected at the upper plateau of the compression isotherms. Please do not adjust margins Please do not adjust margins ## ARTICLE | 11 Please do not adjust margins Please do not adjust margins Second, the LE/LC transition of the mixtures, which evidently occurs at the lower plateau, can be studied in comparison to the LE/LC transition of both pure substances. The LC domain shape of pure DPPC and PHPC differs significantly (see Figure 6). While DPPC forms characteristic chiral bean or propeller shaped domains, PHPC, which is racemic in this study, forms star-like, fractal grey LC domains which do not exhibit chirality. 33 When both lipids are mixed, nucleation of the LC domains begins with mainly DPPC, as can be deduced from the appearance of compact black domains (see column (b) in Figure 6). Rh-DPPE is excluded from these small DPPC-rich domains/ nuclei at the onset of the LE/LC transition but, subsequently, partitions into the LC phase while PHPC is incorporated. As the LC domains grow in size, PHPC joins at the rim of already formed DPPC domains with its typical star-like LC domain shape leading to demixed LC domains (see column (c) in Figure 6). At the end of the phase transition, the LC phase homogenizes, i.e., both lipids form a mixed LC phase, being embedded in a continuous LE phase (see column (d) in Figure 6). Mixtures with high DPPC content remain in the demixed LC state up to higher surface pressures and their LC phases tend to be darker because the dye is excluded from DPPC-rich domains to a higher extent. Although these results suggest a co-existence of three phases in the lower plateau, these three phases were not stable when the compression was paused (see Figure S17). Likewise, only two phases could be observed in the LC/LE transition upon expansion of a mixed DPPC/PHPC monolayer (x(PHPC) = 0.1) from the LC phase (see Figure S17). We therefore assume that the three phases occurring in the LE/LC transition region upon compression are meta-stable and not in equilibrium, i.e., their appearance is due to a kinetically hindered condensation upon continuous compression. Yet, the origin of the upper plateau remains uncertain. By means of fluorescence microscopy and with the used Rh-DPPE dye it could not be attributed to a phase transition. To unravel further details of lipid miscibility, we interpret our data by applying the Gibbs phase rule and by calculating the Gibbs energy of mixing, 10 both of which are regularly used in miscibility studies in monolayers. Although the surface phase rule is frequently used for miscibility studies in Langmuir monolayers, 9,34 in this work application of the Gibbs phase rule is sufficient as shown in Appendix 1. The Gibbs phase rule is given as 𝐹 = 𝐶 − 𝑃 + 2, where F denotes the degrees of freedom, C is the number of components in the monolayer, and P is the number of phases in the monolayer. It is possible to only discuss the components and phases of the monolayer as long as no lipid exchange between monolayer and bulk phases occurs, which is the case for longchain phospholipids. 35 With two monolayer components, DPPC and PHPC, the Gibbs phase rule simplifies to 𝐹 = 4 − 𝑃 or 𝑃 = 4 − 𝐹,, when solved for the number of phases. Thus, by evaluating the degrees of freedom of the studied systems, a prediction of the number of co-existing phases is possible. Since there must be at least one monolayer phase, maximal three degrees of freedom can exist, which are π, T and x. Consequently, P can vary between one (F = 3) and four (F = 0). As evident from Figure 5, in all phases between the πplateau as well as above and below them, π and x and presumably T (compare Figure S18) are degrees of freedom, as they can be varied independently without changing the state of the system. Hence, only one mixed phase would exist in these states. This is consistent with the results of epifluorescence microscopy which show one uniform phase in the respective surface pressure ranges (Figure 6, columns (a) and (e) and Figure S12-S15) as well as the analysis of ΔGmix discussed below. In contrast, in the plateaus of the π-Amol isotherms, x and π are correlated (Figure 5B), i.e., π depends on x and is therefore not a degree of freedom as opposed to x. Consequently, F is either one (x) or two (T, x) and, hence, P is three or two, respectively. Note that π of the lower plateau seems to be independent of x due to both pure lipids exhibiting nearly the same LE/LC transition pressure. However, if chain-perdeuterated DPPC-d62 is used in these mixtures is, which has an increased phase transition pressure, 36,37 π of the lower plateau also depends on x (see below and Figure S20). Interestingly, the epifluorescence micrographs shown in columns (c) and (d) of Figure 6 and further in Figure S12-S15 display three apparent phases in the lower plateau of the mixtures, which are not stable (see Figure S17), and are probably the consequence of the continuous compression of the monolayer. Hence, they do not represent the equilibrium state within the LE/LC phase transition and, consequently, cannot be interpreted using the phase rule which only is valid in equilibrium. This means that the number of equilibrium phases in the transition is two and consequently both x and T are degrees of freedom, i.e., π depends on x and on T. The latter, π depends on T, was proven exemplarily for one mixture (Figure S18). One can clearly see that the compressibility maxima of the DPPC/PHPC 3:1 monolayer shift linearly to lower surface pressure with decreasing temperature. This shows that the transition exists at different temperatures and, hence, T is a degree of freedom in the phase transition. This allows the conclusion that beyond the plateaus no phase coexistence occurs. To further evaluate the thermodynamics of the miscibility of both lipids below, in between, and above the plateaus, one can calculate the Gibbs energy of mixing ΔGmix and the excess Gibbs energy of mixing ΔGexc, which describes the deviations from ideal miscibility. 10 To this end, we compare the observed π-Amol isotherms with ideal ones, calculated from the pure substances, to yield molecular excess areas Aexc. The Aexc are then integrated and values of ΔGexc are calculated. Using the known ideal Gibbs energy of mixing, ΔGmix is then derived. Further information on calculations are given in the Experimental part of this study. The calculated ΔGexc and ΔGmix are included in Table S1 and plotted in Figure 7. From these data it can be concluded that at all mixing ratios, DPPC and PHPC are miscible. In the LE phase at 3 mN m -1 , nearly ideal mixing or complete demixing was observed. As stated before, a complete immiscibility in the LE is rather unlikely and was not found using epifluorescence microscopy. Therefore, we interpret this result as ideal miscibility of both lipid species in the LE phase. However, positive deviations from an ideal miscibility at 10 mN m -1 (between both plateaus) and 30 mN m -1 Please do not adjust margins Please do not adjust margins (above both plateaus), i.e. repulsive interactions between both lipids, were observed for the DPPC/PHPC mixtures consisting of 10 % and 75 % of PHPC, respectively (Figure 7A). However, ΔGexc is not high enough to induce a complete phase separation. To evaluate the miscibility in more molecular detail, again infrared reflection absorption spectroscopy (IRRAS) was used. IRRAS enables us to discriminate both lipids in mixtures through isotopic labelling, since different nuclear isotopes result in different reduced masses of the vibrating moieties. Hence, a shift is observed in the IR spectra. When mixing PHPC with chain-perdeuterated DPPC-d62 it is, thus, possible to compare the surface pressure-dependent change in frequency of νs(CH2) and νs(CD2), to observe the phase transition pressure of both lipids in their mixtures independently. If both lipids show a condensation (decrease of the methylene stretching vibration wavenumbers) at the same surface pressure during compression, they must be considered miscible. If the LE/LC transition pressure differs between both lipids, they demix at least partially. These measurements can also answer the question whether DPPC and PHPC contribute differently to both transitions observed in the isotherms. Furthermore, it can be deduced whether lipid chain condensation/ordering is involved in the upper plateau, i.e., if this plateau can also be considered as phase transition plateau. In Figure 8, the frequencies of νs(CH2) and νs(CD2) are plotted together with the compressibility of the monolayer for the mixture x(PHPC) = 0.75, which exhibits positive deviation from ideal mixing behaviour (compare to Figure 7). The corresponding plots of all other mixing ratios are shown in Figure S19. Note that the phase transition surface pressure is increased in comparison to the measurements shown in Figure 5, as a result of one mixing component being chain-perdeuterated (see Figure S20). Deuteration of the lipids' acyl chains has a significant effect on the phase transition surface pressure as well as the main phase transition temperature in aqueous suspension. 36,37 In all measured mixed monolayers, the lower transition appears to be shifted to higher π-values, that is, no plateau at the pure PHPC's transition pressure is detectable. This is an indication for at least partial miscibility. Like the band position of the CH2 stretching vibrations, also νs(CD2) and νas(CD2) are indicative of the trans/gauche ratio in the respective alkyl chains and can therefore be used to detect phase transitions involving chain melting. For the presented x(PHPC) = 0.75 mixture (just like for all other compositions), we found that both lipids undergo a common phase transition and that at the plateau at lower π, a significant frequency shift, i.e. condensation of the lipids, occurs simultaneously. The second transition at higher surface pressure is connected to a comparatively small decrease in CH2/CD2 stretching vibrational frequencies. This leads to the conclusion that lipid chain condensations have only minor contribution to this second transition. When chain-deuterated DPPC is used, the two plateaus are only discernible for mixtures of x(PHPC) ≥ 0.33. In the isotherms of mixtures containing less PHPC, both transitions overlay which results in one broadened CS curve (compare to Figure S19). A comparative examination at both methylene stretching vibrational bands (CH2 as well as CD2) and at the phosphate stretching vibrational bands using PCA shows simultaneous transitions of the deuterated and non-deuterated lipid chains but a delayed transition in the phosphate headgroup (Figure S21 and Figure S22). After a concomitant change of the PC 1 scores in all three spectral ranges at the onset of the first plateau, the most pronounced changes in the phosphate vibrations are found at slightly elevated surface pressures, which could correlate with the surface pressure at the second plateau in the isotherms. In addition, the phosphate stretching Please do not adjust margins Please do not adjust margins vibrations shows another, albeit less pronounced, change in the range between 20 and 30 mN m −1 . This might be correlated with a headgroup re-orientation. In any case, it shows that transitions in the headgroup region may exist that are independent of the chain condensation. However, since we do not have enough complementary data about this transition, we refrain from further speculation about its origin. By combining the results of all experiments performed in this study regarding miscibility of DPPC and PHPC, we conclude that there is substantial evidence that the two lipids are miscible in all mixing ratios but might show deviations from ideal miscibility. From IRRAS measurements of mixtures of PHPC and DPPC-d62 it is found that the upper plateau does not involve a significant amount of chain ordering as it would occur in lipid LE/LC phase transitions. By employing epifluorescence microscopy, we observed two stable co-existing monolayer phases during the LE/LC phase transition (at the lower plateau). However, the full nature of the second plateau appearing only in the mixtures of both substances could not be unravelled completely. ## Conclusions In this work, we present studies of pure monolayers of the structurally related phospholipids DPPC and PHPC using π-Amol isotherms and IRRAS. In addition, we provide a detailed characterization of mixed DPPC/PHPC monolayers through their isotherms using epifluorescence microscopy, the surface phase rule, evaluation of thermodynamic mixing parameters, and IRRAS. The isotherms of both pure lipid monolayers are comparable and differ only slightly in their phase transition pressures and LC phase compressibilities. DPPC and PHPC exhibit comparable alkyl chain trans/gauche ratios in their corresponding LE and LC phases and both form hexagonal, ordered LC phases above their LE/LC phase transition. We find that the substitution of the ester linkage at the sn-2 chain by an ether linkage causes several changes in lipid monolayer organization despite both lipids showing similar π-Amol isotherms. PHPC, when compared to DPPC, exhibits: • a smaller chain tilt angle at surface pressures of 20 mN m -1 and above, • stronger hydration of the carbonyl group independent of π, • stronger hydration of the phosphate group in the LE phase, • increased headgroup ordering, and • strong attenuation of the CH2 wagging band progressions independent of π. These findings lead us to the conclusion that the glycerol moiety of PHPC adopts an orientation parallel to the water surface, which is different from the orientation of the glycerol of DPPC or other unsubstituted 1,2-diester phosphocholines. Similar orientational differences have been found in aqueous suspensions of the lipids before. In the second part of this study, we present plateaus in the π-Amol isotherms of mixed monolayers containing DPPC and PHPC, that do not appear in the isotherms of either of the pure lipids. The miscibility studies of mixed monolayers in the full mixing range can be interpreted in terms of non-ideal mixing behaviour but no demixing in the LE or LC phase as detected by epifluorescence microscopy and being confirmed by mixing energy calculations. Further characterization of the mixing behaviour by IRRAS shows that the upper plateau does not involve significant ordering of lipid chains. The cause for existence of the upper plateau remains unclear from our experiments but seems to involve changes in headgroup hydration. It is likely to originate from the geometrical differences (for example headgroup re-orientation) found for both pure lipid monolayers. Author Contributions ## Conflicts of interest There are no conflicts to declare.

chemsum

{"title": "Influence of a Single Ether Bond on Assembly, Orientation, and Miscibility of Phosphocholine Lipids at the Air-Water Interface", "journal": "ChemRxiv"}

synthesis,_structure,_and_mechanical_properties_of_silica_nanocomposite_polyrotaxane_gels

## Abstract: A significantly soft and tough nanocomposite gel was realized by a novel network formed using cyclodextrin-based polyrotaxanes. Covalent bond formation between the cyclic components of polyrotaxanes and the surface of silica nanoparticles (15 nm diameter) resulted in an infinite network structure without direct bonds between the main chain polymer and the silica. Small-angle X-ray scattering revealed that the homogeneous distribution of silica nanoparticles in solution was maintained in the gel state. Such homogeneous nanocomposite gels were obtained with at least 30 wt % silica content, and the Young's modulus increased with silica content. Gelation did not occur without silica. This suggests that the silica nanoparticles behave as cross-linkers. Viscoelastic measurements of the nanocomposite gels showed no stress relaxation regardless of the silica content for <20% compression strain, indicating an infinite stable network without physical cross-links that have finite lifetime. On the other hand, the infinite network exhibited an abnormally low Young's modulus, ~1 kPa, which is not explainable by traditional rubber theory. In addition, the composite gels were tough enough to completely maintain the network structure under 80% compression strain. These toughness and softness properties are attributable to both the characteristic sliding of polymer chains through the immobilized cyclodextrins on the silica nanoparticle and the entropic contribution of the cyclic components to the elasticity of the gels. ## Introduction Nanocomposite materials, in which nanoparticles are distributed via a matrix such as resin or rubber, exhibit various functions that the matrix material cannot achieve by itself. For instance, polyurethane/magnetic nanoparticle nanocomposite elastomers permitted a wide range of elasticity modulation by controlling the magnetic field , owing to the alignment of nanoparticles within the polymer network. In addition to such functionalization, nanocomposition can effectively reinforce materials; a typical example is natural rubbers reinforced by nanoparticles such as carbon black and silica for practical use as tire materials. Because of adhesion of polymer chains to the nanoparticle surfaces, the chain mobility at the interface is considerably suppressed, thereby increasing the elastic modulus of the nanocomposite . Although the nanocomposite strategy is promising for the design of relatively hard functional materials, it is not easy to realize soft nanocomposite materials because of the strong interactions at the interface that reduce the polymer chain mobility. For dielectric elastomers expected to be used as actuators and electric generating systems, nanocomposites should satisfy both a high dielectric constant that is assured by the nanoparticles and a low elastic modulus . Mechanically interlocked supramolecular polymers, such as polyrotaxane , can control the interface between the matrix polymer and the nanoparticles. Polyrotaxane comprising an end-capped backbone polymer and threaded cyclic molecules such as cyclodextrins (CDs) can form a network structure by intermolecular binding of the cyclic components . Since the polymer chains are topologically connected to each other without chemical bonds, the chains can slide through the crosslinks and lead to several unique properties . Polyrotaxanes were also applied for surface modification of substrates by attaching the cyclic components to the surface . These results were based on the indirect connection among different polymers or between polymers and surfaces, so that the mobility of the polymers can be maintained at the cross-linking points or the interface. We demonstrate here that the indirect connection between the backbone polymers and nanoparticles can be applied also for soft nanocomposite materials. As illustrated in Figure 1, formation of chemical bonds between the cyclic components of polyrotaxanes to the surface of a nanoparticle may result in a network structure where the nanoparticle acts as a cross-linker. As the first model system, we employed a silica nanoparticle and a CD-based polyrotaxane whose rings were chemically modified to react with the surface of silica. The inner structure and the mechanical properties of the silica nanocomposite polyrotaxane gels were studied by small-angle X-ray scattering and viscoelastic measurements, respectively. ## Results and Discussion To achieve a covalent bond between the cyclic components (CDs, in this study) and the nanoparticle surfaces (SiO 2 ), hydroxy groups of the cyclic components of a polyrotaxane comprised of α-CD and PEG (PR) was modified as shown in Scheme 1. The polyrotaxane with triethoxysilylated CD (TES-PR) is suitable for the reaction with SiO 2 in the presence of base. Although TES-PR can be isolated just by vacuum drying to evaporate the reactant and solvent, the solid isolated in this way becomes insoluble in any solvent. This is due to the crosslinking induced by reactions between CDs. Thus, DMSO was added to the reaction solution first, followed by drying the solution under vacuum to eliminate acetone and reactant; as a result, a pure DMSO solution of TES-PR was obtained. The solution was used for the following characterization. Figure 2 shows the 1 H NMR spectra of TES-PR, PR, and an intermediate polyrotaxane with acryloyl groups at the CDs (Acryl-PR). All peaks were assigned as shown. From these integral values, the modification degree of Acryl-PR was calculated to almost 100% (corresponding to 18 acryloyl groups in each CD), with only 23% of the acryloyl groups reacted with triethoxysilane to generate TES-PR. GPC traces shown in Figure 3 indicate that the interlocked structure was retained throughout the modification and that the molecular weight of PR was increased with each modification step. The GPC trace of TES-PR also indicated the existence of high-molecular weight TES-PR. This probably multimeric TES-PR is the result of intermolecular reactions between triethoxysilyl groups during hydrosilylation. Since the pure TES-PR DMSO solution was stable at room temperature, the solution was used for the following reaction with silica nanoparticles. A TES-PR solution in DMSO was mixed with a dispersed solution of silica nanoparticles with 15 nm diameter in N,Ndimethylacetamide (DMAc) (20 wt %), followed by the addition of diisopropylethylamine to initiate the reaction between TES-PR and silica. The pre-gel solution was transferred to a glass mold with 3 mm thickness and cured for 16 h at 373 K. As shown in Figure 4, the obtained gel was transparent, indicating no significant aggregation. Five gels were synthesized in the same way with different initial concentrations of silica nanopar- ticles. Notably, in the absence of silica nanoparticles, gelation did not occur; only a slight increase of viscosity was observed. This result clearly shows that the gelation was achieved mainly by the reaction between silica and TES-PR, with a parallel minor reaction occurring between the triethoxysilyl groups of different TES-PRs. To elucidate the dispersity of silica nanoparticles in the gels, small-angle X-ray scattering (SAXS) was carried out. shows the SAXS profiles of silica nanocomposite gels with different silica concentrations. A Bragg's peak around q = 0.01-0.05 −1 was observed, with the peak shifting toward higher q with increasing silica concentration. It is noteworthy that no increase of scattering intensity toward the low q limit, indicating that no aggregation is seen even in that q range. Similar SAXS profiles were observed in silica solutions (Figure 5b). Since the correlation distance becomes shorter with the increase of concentration, the distance is thought to be the separation between silica nanoparticles. When we assume homogeneous distribution of particles, the following relation exists in the distance between particles d and the concentration c: (1) Figure 5c shows the double log arithmic plots of correlation length, which is obtained from the q of the peak top (d = 2π/q), against silica concentration for the nanocomposite gels and silica solutions. In both cases, a similar power dependence, , as in Equation 1, was obtained. Therefore, similarly to the silica solution, the homogeneous dispersion of silica nanoparticles was retained in the nanocomposite gels until at least 30 wt % silica content. Viscoelastic measurements were carried out for the obtained gels. Figure 6a shows the results of stress relaxation tests. At high silica concentration (more than 15%), the Young's modulus is about 1 kPa, whereas a lower silica concentration yielded a lower modulus of about 0.5 kPa. In addition, as mentioned above, no gelation occurred without silica. Thus, the modulus tends to increase with the concentration of silica, though the incremental increase of modulus is not proportional to silica concentration but stepwise. This suggests that the cross-linking density becomes higher with increasing silica concentration because the silica acts as the polyrotaxanes crosslinker. The results of dynamic viscoelastic measurements shown in Figure 6b also suggest the network formation via silica nanoparticles. Dynamic storage Young's modulus E′ was increased with silica content and the modulus was consistent with E(t). On the other hand, the loss modulus E″ hardly changed with increasing of silica concentration. Thus, the ratio of the loss modulus to storage modulus, the so-called loss tangent, decreased with silica concentration. This typical tendency generally observed in chemical gels that is attributed to the decrease of dangling chains, resulting in the formation of a denser network. In addition, there was no stress relaxation with finite equilibrium modulus regardless of the silica content. Figure 6c shows the strain-dependence of the relaxation Young's modulus for the gel with 15% silica. The modulus was almost independent of the strain. These results clearly indicate that the silica nanocomposite polyrotaxane gels form an infinite network structure with negligible physical cross-links that have finite lifetime to exhibit stress relaxation, similar to ideal chemical gels. At the same time, the negligible relaxation indicates that almost all silica nanoparticles were bound to the polymer network. However, the Young's modulus of the nanocomposite gels is abnormally low. From the Young's modulus, E, with an assumption of ideal polymer network, the averaged molecular weight between cross-links, M x , can be estimated by the following equation: (2) where ρ is the density of polymer, R the gas constant, and T the absolute temperature. M x of the representative nanocomposite polyrotaxane gels with E = 1 kPa was calculated to be 7 × 10 5 g/mol. This M x is considerably larger than the molecular weight of the precursor polyrotaxane, TES-PR (~2 × 10 5 g/mol). Thus, this result gives us an unlikely picture for the network where several polyrotaxanes are bound to form a single partial chain. This discrepancy probably arises from the assumption of the validity of rubber elastic theory for these gels. Surely, for most of composite gels or rubbers, the theory is already invalid due to the presence of strong interactions between polymers and embedded particles in addition to the covalent crosslinking points. However, such interactions increase not, decrease, modulus. Thus, the interactions between the polyrotaxane and silica cannot explain the significantly low modulus. Recent research in polyrotaxane gels, which were obtained by the simple cross-linking of polyrotaxanes via intermolecular covalent bonds between CDs, indicated a new origin of the elasticity experimentally and theoretically . Since the polymer chains can slide through the cross-links, the anisotropic orientation of chain segments, which causes the entropic elasticity of rubbers and gels, can be relaxed. Simultaneously, the relaxation of chains results in inhomogeneous distribution of the threaded CDs. As the CDs continuously slide along the backbone polymer, their inhomogeneous distribution can generate entropic stress. Thus, the origin of elasticity is not the entropy of polymer chains but that of the CDs, and the abnormally small modulus of polyrotaxane gels can be explained by the characteristic elasticity. Similarly, the extremely low modulus of the nanocomposite polyrotaxane gels is attributable to the charac-teristic elasticity originated from the entropy of CDs. Since the infinite network was formed mainly by the covalent bonds between CDs and silica surface, the polymer chains are bound to neither CDs nor silica. If the polymer chains are directly attached to the silica surface, the chain mobility were drastically decreased, yielding an increased modulus as observed in conventional composite gels and rubbers. Therefore, the polymer chains slide through the immobilized CDs on the surface of silica, and thus the unique properties observed in polyrotaxane gels may also appear in these nanocomposite polyrotaxane gels. In addition to the characteristic softness, the nanocomposite gels exhibited significant toughness. Figure 7 shows the stress-strain curves of one of the nanocomposite gels under compression. Since the fracture point is not always clear in the stress-compression behavior, the same measurement was repeated using the same gel. The two curves are in complete agreement, indicating that no fracture occurred during the first compression. This data proved that the gel is compressive to at least one-fifth of the original thickness without fracture or network structure recombination. The toughening mechanism may be essentially the same as that in polyrotaxane gels: the stress applied to chains can be distributed by chain sliding through the immobilized CDs on the silica surface, the so-called pulley effect . In this way, the silica nanocomposite polyrotaxane gels exhibited two unique properties: soft and tough, through the novel method for nanocomposites without chemical bonds or significant interactions between polymers and nanoparticles. ## Conclusion Here we demonstrate the synthesis of a novel nanocomposite gel where the polymer chains were not directly bonded but rather mechanically interlocked to the silica surfaces. The silica nanoparticles were homogeneously distributed in the gel and worked as cross-linkers to immobilize the cyclic components of the polyrotaxanes on the silica surfaces. As the backbone polymer chains were not only free from adhesion to the silica surface but can also slide through the immobilized cyclic components, the nanocomposite gel achieved low Young's modulus and high toughness without any detectable fracture or recombination of network structure under 80% compression. These results suggest that the concept of topological crosslinking previously studied with polyrotaxane gels is applicable to other nanocomposite materials, though our model system utilized silica nanocomposites. Functionalization and applications of the nanocomposite polyrotaxane gels with other nanoparticles are now in progress. ## Experimental Materials Crude polyrotaxane consisting of polyethylene glycol (PEG, M n = 32,000) and α-cyclodextrin (CD) were purchased from Advanced Softmaterials, Inc. The crude polyrotaxane was purified by repeated reprecipitation from its DMSO solution into deionized water. The obtained precipitate was freeze-dried and the refined polyrotaxane (PR) obtained as a white powder. The coverage, which is a measure how densely the backbone PEG is covered with CDs, was calculated to be 25% based on the 1 H NMR spectrum. Standard polymers for the calibration of the molecular weights by size-exclusion chromatography (SEC) were purchased from Polymer Source, Inc. Solutions of silica nanoparticle with 15 nm of diameter in N,N-dimethylacetamide was kindly supplied by Nissan Chemical Industries, Ltd. 2-Isocyanatoethyl acrylate was purchased from Showa Denko K.K. Xylene solutions of platinum(0)-1,3-divinyl-1,1,3,3tetramethyldisiloxane (~2%) was purchased from Sigma-Aldrich Corporation. All other chemicals were purchased from Tokyo Chemical Industry Co., Ltd., or Wako Pure Chemical Industries, Ltd., and all reagents were used as received without further purification except for PR. ## Characterization measurements 1 H NMR spectra (400 MHz) were recorded on a JEOL JNM-AL400 spectrometer at 343 K. The chemical shift was calibrated using DMSO (2.50 ppm) as an internal standard. SEC was performed on TOSOH HLC-8220 with two Shodex OH Pack SB-806MHQ columns, with DMSO at 50 °C in the presence of 0.01 M lithium bromide as the eluent using RI detection and PEG standards. The flow rate was 0.4 mL/min. ## Synthesis of acryloyl modified polyrotaxane (Acryl-PR) Three grams of PR previously dried under vacuum was dissolved in anhydrous DMSO (60 mL). 2-Isocyanatoethyl acrylate (9 mL) and dibutyltin dilaurate (90 μL) were added and stirred at room temperature for 5 days. The product was precipitated by pouring the reaction solution into methanol. The precipitate was repeatedly washed with methanol and then dried. The dried product was again dissolved in acetone and then a large amount of methanol was added to precipitate the product. This process was repeated again followed by drying to yield acryloyl modified polyrotaxane (Acryl-PR, 5.41 g) as a white solid: 1 ## Synthesis of triethoxysilyl modified polyrotaxane (TES-PR) Two grams of Acryl-PR previously dried under vacuum was dissolved in anhydrous acetone (40 mL). Triethoxysilane (600 μL) and a xylene solution of platinum(0)-1,3-divinyl-1,1,3,3-tetramethyldisiloxane dibutyltin dilaurate (200 μL) were added followed by refluxing at 65 °C overnight. The reaction solution was then centrifuged to precipitate a minor insoluble component, and the supernatant collected. Because the product became insoluble in any solvent once it was completely dried, the crude acetone solution was stored at room temperature until just before use. For further characterization, DMSO or DMSOd 6 was added to the acetone solution, and then the solution was dried under vacuum to evaporate unreacted reactants and non-DMSO solvent. As a result, pure solutions in DMSO or DMSOd 6 were obtained because DMSO has a sufficiently high boiling point to resist evaporation. 1 H NMR (400 MHz, DMSO-d 6 , 343 K): 7.0 (NH of urethane), 6.3, 5.9 (CH 2 of vinyl), 6.1 (CH of vinyl), 4.9 (C (1) H of CD), 4.1 (C(=O)O-CH 2 ), 3.8 (SiO-CH 2 ), 3.5 (PEG), 3.2 (-CH 2 -NH), 2.3 (OC(=O)-CH 2 ), 1.1 (SiOCH 2 -CH 3 ), 1.0 (Si-CH 2 ). ## Synthesis of silica nanocomposite polyrotaxane gels 1.5 mL of DMSO was added to 3 mL of the crude acetone solution of TES-PR, whose silica content was known to be 5 wt %/vol. The solution was dried under vacuum to eliminate impurities, yielding 1.5 mL of 0.1 g/mL TES-PR solution in DMSO. A DMAc solution of silica nanoparticles was added to the TES-PR solution, and then the mixed solution was dried under vacuum to evaporate DMAc, resulting in 1.5 mL of TES-PR and silica pre-gel solution in DMSO. By changing the added volume of silica solution, five pre-gel solutions with different silica contents were obtained. Diisopropylethylamine (15 μL) was added to the pre-gel solution, and then the solution was transferred into a Teflon/glass mold with a 25 mm × 25 mm × 3 mm void. The reaction was carried out at 100 °C for 16 h to yield a transparent nanocomposite gel. In the same way, thinner gels were also prepared in a mold with a 20 mm × 20 mm × 1 mm void for the following structural analysis. ## Small-angle X-ray scattering measurement Synchrotron small-angle X-ray scattering (SAXS) experiments were carried out at the beamline 6A of Photon Factory, HighEnergy Accelerator Research Organization, KEK (Tsukuba, Japan). The wavelength λ of the incident X-ray beam was 1.50 and the beam size was 0.5 mm (vertical) × 0.5 mm (horizontal). PILATUS 300k (Dectris) was used to record SAXS patterns. The sample-to-detector distance was 2.6 m and the sample thickness was 1 mm. The exposure time for each sample was kept constant at 5 s. The scattering angle θ was calibrated by the diffraction pattern of chicken tendon collagen. The obtained SAXS patterns were converted into 1D profiles, scattering intensity I vs scattering vector q, by circular averaging. The scattering vector is defined as: (3) ## Mechanical measurements Obtained silica nanocomposite polyrotaxane gels were cut into cylindrical shapes of 10 mm diameter and 3 mm thickness. All measurements were conducted with a strain-controlled oscillatory rheometer (RSAIII, TA Instruments) using a parallel plate geometry at room temperature. Frequency sweeps were conducted from 0.01 to 80 Hz, applying 1% of the oscillatory compressive strain amplitude, which was still within the range of linear viscoelasticity. Stress relaxation tests were also performed by the compression mode. Stress-compression curves were obtained at a sufficiently slow and constant rate of strain: 0.2%/s.

chemsum

{"title": "Synthesis, structure, and mechanical properties of silica nanocomposite polyrotaxane gels", "journal": "Beilstein"}

<i>in_situ</i>_real-time_tracing_of_hierarchical_targeting_nanostructures_in_drug_resistant_tumors_u

## Abstract: Nanoparticles that respond to specific endogenous or exogenous stimuli in tumor tissues are actively being developed to address multidrug resistance owing to multiple advantages, including a prolonged circulation time, enhanced permeability and retention effect, and superior cellular uptake. Although some exciting results have been obtained, existing nanoparticles have limited routes to overcome the drug resistance of tumor cells; this limitation results in a failure to ablate resistant tumors via intravenous administration.To resolve this dilemma, we developed a smart theranostic nanoplatform with programmable particle size, activatable target ligands and in vivo multimodal imaging. This nanoplatform, which includes stealth zwitterionic coating, was shown to be quickly trapped in tumor tissue from the blood circulation within 5 min. Subsequently, the targeting moieties were activated in response to the acidic tumor microenvironment by triggering the zwitterionic shell detachment, driving the peeled nanoparticles to penetrate into tumor cells. These smart nanoparticles completely inhibited drug-resistant tumor growth and did not cause any damage to normal organ tissues in live animals. The designed nanoplatforms simultaneously acted as a nanoprobe for fluorescence imaging. Moreover, we also used noninvasive pharmacokinetic diffuse fluorescence tomography (DFT) to dynamically monitor and in situ real-time trace the nanoplatforms' behavior throughout the entire tumor in live animals. The nanoplatforms enabled rapid drug accumulation and deep penetration throughout the entire tumor. The rate of drug accumulation after the administration of nanoplatforms was five-fold higher compared with that after the administration of the free drug, which resulted in increased drug delivery efficiency and improved antitumor efficacy. Collectively, this hierarchical vehicle design provides promising insights for the development of theragnosis for multidrug resistant tumors. Multidrug resistance (MDR) has become a major obstacle to the success of cancer chemotherapy. The major mechanism of MDR involves the P-glycoprotein (P-gp), which is overexpressed in cancer cells and is able to pump various chemotherapeutics out of the cells, resulting in a reduced intracellular drug concentration with a limited therapeutic effect. 1 Tumor-targeting drug delivery nanoparticles have been actively developed to overcome MDR, 2 owing to their preferential accumulation at tumor sites due to enhanced permeability and retention (EPR) and active targeting effects. 3 In contrast to small drug molecules, these nanoparticles are internalized into cells via endocytosis, and thus they are not affected by the overexpression of the P-gp efflux pump on the MDR cell surface. 2d However, the remarkable abnormality and complexity of the tumor microenvironment often pose barriers to the supply, penetration and distribution of nanoparticles. 4 To address these predicaments, taking into consideration the intra-tumor tissue accumulation, retention and penetration of drugs, an emerging delivery strategy was developed wherein vehicles respond to the specifc endogenous or exogenous stimuli within tumor tissues. 5 Although exciting results have been obtained previously, existing nanoparticles have limited routes to inhibit the resistance of tumor cells, and this limitation results in the failure to ablate resistant tumors via intravenous administration. The majority of these self-assembled nanostructures encounter problems with drug-loading stability, which is influenced by the in vivo environment, in which blood components can act as competing drug acceptors, 6 and interactions between polymeric nanoparticles and blood components can lead to drug leakage in the circulation. 7 Therefore, it is vital to develop a nanoscale delivery system for improving delivery efficiency and, eventually, therapeutic efficacy. In addition to the efficient delivery of anticancer therapeutics to tumor cells, the capacity for in vivo imaging to trace drug distribution and accumulation in situ in real-time is also a desired property of theragnostic systems. Although the conventional labeling strategy can track the collective and metabolic behaviors of a nanocarrier in planar hierarchy, it is not currently possible to visualize the real-time trail throughout the entire tumor. Only limited visualization of the surface of the tumor is possible due to the limited spatial resolution and long imaging time. It is indispensable to dynamically trace nanocarriers throughout the entire tumor to understand their behavior at the targeted tissue. In this study, to complement the traditional radioisotopic tracing methods and realize noninvasive in situ nanocarrier monitoring in vivo, pharmacokinetic diffuse fluorescence tomography (DFT) was exploited for the frst time to dynamically monitor and trace the nanocarrier behavior throughout the entire tumor in live animals. Based on the aforementioned considerations, we proposed a smart hierarchical targeting theranostic nanoplatform utilizing dynamic covalent chemistry with a programmable particle size and activatable surface ligands to decrease drug leakage into the circulation, enhance the permeability of cargos and boost penetration into tumors. Boronate esters, which are formed by the reaction of phenylboronic acid (PBA) with diols, are stable under physiological conditions but they easily dissociate in the acidic tumor microenvironment. 8 This chemistry was adopted not only to form a sheddable zwitterionic shell to stabilize the nanovehicles in the blood circulation, but also to act as an acid-responding switch to decorticate the stealth corona in the tumor tissue. When the switch is "on", the zwitterionic shell is pared to release a much smaller core to facilitate tissue penetration. Simultaneously, the targeting ligands on the peeled nanoparticle surface are also activated to enhance cellular internalization. The intratumoral pharmacokinetics of the designed nanocarriers were explored using a real-time, noninvasive and radiation-free imaging technology at the spatial level. Simultaneously, BODIPY was also used to trace the biodistribution of nanocarriers in vivo via fluorescent imaging to supplement this DFT technology. A multidrug resistant tumor model was used to further evaluate the mechanism and treatment efficiency of the designed smart theragnostic system. A nanoplatform for use as an anticancer drug vehicle was fabricated by the self-assembly of sophisticated copolymers, namely boron-dipyrromethene dye (BODIPY)-conjugated poly(carboxybetaine acrylate)-b-poly(2-(acrylamido) glucopyranose) (pCBAA-b-pAGA) and poly(2-lactobionamidoethyl methacrylateb-poly(3-acrylamidophenylboronic acid)) (pLAMA-b-pAAPBA). Two block fluorescent copolymers were prepared by RAFT polymerization (Scheme S1 †) and were subsequently characterized (Fig. S1-8 and Table S1 †). Then, the two copolymers were self-assembled into hierarchical nanostructures via the hydrophilic-hydrophobic interaction and the phenylboronic acid-polysaccharide interaction. Particle size greatly influences the circulation time, tumor accumulation and penetration of nanocarriers. 2e,9 It is generally accepted that nanoparticles with a diameter of more than 400 nm exhibit a superior circulating half-life and less extravasation from the leaky vasculature into the tumor interstitium, while smaller nanoparticles in the range of 10-100 nm can achieve higher tumor penetration owing to the reduced steric and diffusion hindrance. 10 Consequently, an ideal drug delivery system should exhibit size-switching behavior from large to small in response to endogenous or exogenous stimuli within tumor tissues. Such a strategy has been demonstrated to be effective for achieving enhanced drug retention, penetration, cellular internalization and nuclear uptake within tumors. 11 Generally, both asialoglycoprotein receptor (ASGP-R) and sialic acid (SA) are overexpressed on the membrane of cancer cells, which facilitates the active targeting via the specifc LAMA-ASGP-R and PBA-SA interactions. Based on this principle, we designed a programmable size-switching nanoplatform in a fairly simple manner that employs a large size within the circulatory system to decrease drug leakage in circulation to efficiently enhance tumor accumulation, and then switches to a smaller size in the tumor tissue to promote drug penetration and cellular uptake (Scheme 1). The "switch" is realized by the incorporation of dynamic boronate ester bonds. At a pH of 7.4, pCBAA-b-pAGA and pLAMA-b-pAAPBA were shown to self-assemble into nanoparticles, abbreviated as pCA/pLB nanoparticles. These nanoparticles consist of a superhydrophilic zwitterionic pCBAA/pLAMA shell, a cross-linked boronate ester layer and a hydrophobic pAAPBA core. As previously mentioned, each block was designed to endow the nanocarrier with a specifc desired function. The zwitterionic pCBAA stabilizes the nanostructures and increases their halflife in the circulatory system, the ligands pLAMA and pAAPBA facilitate tumor-targeted cell internalization, and the pAGA cross-linked with pAAPBA acts as the acid-responding switch in the acidic tumor microenvironment. The fluorescence emission and UV-vis absorption spectra of BODIPY dye-labeled pCA/pLB nanoparticles were recorded (Fig. S9 †), and the excitation/emission spectra presented a well-defned bandwidth with a visible absorption maximum at 506 nm and emission maximum at 523 nm. To evaluate the effect of the copolymer composition on the particle properties, three nanoparticles with different PBA contents (named pCA/ pLB Np1, pCA/pLB Np2 and pCA/pLB Np3) were prepared (Table S2 †). Dynamic light scattering (DLS) was used to demonstrate that the majority of the particles had a size in the range of 110-170 nm with a narrow polydispersity index (Table S2 and Fig. S10 †). The zeta potential was slightly negative (Table S2 †), which may contribute to enhanced biocompatibility and the avoidance of undesirable clearance by the reticuloendothelial system. 12 The particle size exhibited no signifcant variation in serum within 48 h of incubation, indicating that the nanostructures are stable in complex blood medium (Fig. S11 †). To enable a larger core size to increase the drug loading capacity, pCA/pLB Np3 was chosen for the following study. To demonstrate their size-switching behavior (Fig. 1A), pCA/ pLB Np3 was treated with pH 7.4 and 6.5 buffers before being subjected to measurements. The particle size, as measured by DLS at pH 7.4, was mainly approximately 165 nm (Fig. 1B). The shape was spherical, and the polydispersity was narrow (Fig. 1D). When treated with pH 6.5 buffer, the nanoparticle size decreased to $48 nm (Fig. 1C), confrmed by both DLS and TEM (Fig. 1C and E). The 1 H NMR spectra (Fig. S12 †) showed less pCBAA content in the nanoparticles that were treated in a weakly acidic medium compared with those treated in a medium at pH 7.4, demonstrating that the boronate ester dissociation mediated the zwitterionic shell detachment. To further validate the switching mechanism, pLAMA 30 -b-pAAPBA 20 without fluorescence was synthesized (Scheme S2 †). Then, a new nanoparticle was obtained by the self-assembly of the fluorescent pCBAA-b-pAGA with non-fluorescent pLAMA-b-pAAPBA; this new particle was denoted as pCA/pLB Np-NF. After incubation under different pH conditions (pH 7.4 or 6.5), the fluorescence intensity of the nanoparticles was measured. As shown in Fig. 1F and G, the much weaker fluorescence at pH 6.5 indicates that a considerable amount of pCBAA-b-pAGA was detached from the nanostructures. This detachment was also quantifed; as shown in Fig. 1H, nearly 69% of pCBAA-b-pAGA had dissociated within 24 h in the pH 6.5 medium. The dynamic boronate ester in this nanostructure facilitates the rapid release of drugs in cancer cells because the boronate ester is stable under physiological conditions but easily dissociates in the acidic tumor microenvironment. 13 This property of boronate ester prevents drug leakage during circulation but ensures the rapid cellular internalization of vehicles in tumors. To explore drug loading and release behavior of this system, doxorubicin (DOX) was used as the model drug. The encapsulation efficiency (EE) and loading capacity (LC) of DOX are shown in Table S3. † The LC of pCA/pLB Np3 was higher (11.3%) than that of pCA/pLB Np1 (8.9%) and pCA/pLB Np2 (9.4%), indicating that the chain length of hydrophobic pAAPBA plays an important role in drug loading. Since insufficient intracellular drug release remains the rate-limiting step for reaching a drug concentration level within the therapeutic window, and given that the pH values of endosomes and lysosomes in tumor cells are 5.0-5.5, while those in the blood are $7.4, we investigated the in vitro DOX-release behavior from drug-loaded pCA/ pLB Np3 in different media (pH 5.4 and 7.4, respectively). As shown in Fig. 2A, we determined that the cumulative DOX release at pH 7.4 was only 11% over 60 h, implying that DOX was steadily sequestrated in the nanoparticles. In the pH 5.4 medium, the DOX release profle presented a burst release of $21% after 3 h and 73% after 96 h, possibly due to the dissociation of boronate ester bonds. In general, dynamic boronic ester bonds could be cleaved not only under the acidic conditions but also under the condition where other glucose molecules are neighbors. Since the normal blood glucose level is 0.7-1.1 mg mL 1 in the human body, which may present a hidden danger to the nanocarriers by disrupting the boronate ester bonds, DOX release was further analyzed in a pH 7.4 medium that contained glucose (1 mg mL 1 ). The DOX release was only 18% over 60 h, indicating that the normal blood glucose levels would not trigger DOX leakage during circulation. Zwitterionic polymers prolong the in vivo circulation of nanoparticles by inhibiting nonspecifc protein adsorption and cellular-uptake. 14 However, this property may become an obstacle when nanoparticle cellular internalization is needed inside the tumor tissue. The incorporation of targeted ligands, such as galactose and PBA, onto the nanoparticle surface can enhance cellular internalization. 15 In our nanocarrier design, the weakly acidic tumor microenvironment triggers the detachment of the zwitterionic shell and simultaneously activates the surface ligand PBA to enhance cellular internalization. We compared the nanoparticle cellular uptake in different media with different pH values using multidrug resistant HepG2/ADR tumor cells. As shown in Fig. 2B, much weaker green but stronger red fluorescent signals were observed at pH 6.5 compared with those at pH 7.4. The green signal came from the BODIPY groups, representing the polymer molecules, while the red fluorescence was emitted by DOX. Weak acids triggered the dissociation of dynamic boronate esters and peeled off the pCBAA-b-pAGA polymer shell, resulting in a much weaker green fluorescence. The detachment of the zwitterionic shell greatly enhanced cellular uptake and drug release, and the released drug was primarily detected in the nuclear region. This fnding is attributed to the fact that the PBA and galactose targeting moieties were activated in response to the acidic medium triggering the zwitterionic shell detachment, driving the peeled nanoparticles to penetrate into the tumor cells. It has been validated that a typical PBA-rich nanoplatform can be transported into the nuclei for effective action. 15d DOX is a DNA toxin that induces cell death by targeting nuclear DNA to cause DNA damage and inhibiting topoisomerase II to block DNA replication inside the nucleus. 16 The signifcantly elevated drug nuclear concentration in the nucleus indicates the successful intracellular release and potentially enhanced therapeutic efficacy. The emergence and development of multidrug resistance have limited the success of cancer chemotherapy, and addressing the cellular uptake mechanism of nanocarriers is critical for solving this problem. A competitive inhibition assay was performed to study the cellular internalization pathway of the peeled nanoparticles. As shown in Fig. 2C, the addition of galactose or PBA resulted in decreased nanoparticle cellular uptake, indicating that the nanoparticle surface ligands (galactose and PBA moieties) effectively facilitated internalization via receptor-mediated endocytosis. Fig. 2D shows that there was no obvious difference between the fluorescent signals of the nanoparticles and DOX, indicating that the drug-loaded nanoparticles were taken up by certain pathways but not via the Pglycoprotein. Selective inhibitors and nanocarriers were cocultured with cells to determine the intracellular drug content using flow cytometry to explore the major pathway of cellular internalization. Fig. 2E shows that the inhibition of clathrin and raft/caveolae-mediated endocytosis with sucrose, nystatin and methyl-b-cyclodextrin, respectively remarkably reduced the amount of intracellular DOX and nanoparticles. However, the inhibition of macropinocytosis with macrowortmannin and amiloride did not have a strong influence on cellular uptake, indicating that the clathrin and raft/caveolae-mediated endocytosis is the major pathway of cellular internalization for the obtained nanoparticles. This journal is © The Royal Society of Chemistry 2019 The cellular uptake behavior was further compared between normal HepG2 cells and drug resistant HepG2/ADR cells. We frst used flow cytometry to verify that the DOX-induced HepG2/ ADR cells were resistant to DOX treatment. Fig. 2F shows that signifcantly less DOX accumulated in HepG2/ADR cells owing to the overexpression of the multidrug resistance efflux pump. 2e When DOX-loaded pCA/pLB Np3 was incubated with HepG2 or HepG2/ADR cells, no signifcant difference was observed between the two groups, indicating that the nanocarrier is capable of overcoming the multidrug drug resistance barrier (Fig. 2D and F). For drug carriers, good cytocompatibility is essential for clinical applications. The in vitro cytotoxicity of blank pCA/ pLB Np3 and the anticancer efficacy of DOX-loaded pCA/pLB Np3 in NIH3T3 and HepG2/ADR cells were analyzed by MTT assay. As shown in Fig. S13, † no obvious cytotoxicity was observed after NIH3T3 and HepG2/ADR cells were treated with blank pCA/pLB Np3. However, drug-loaded pCA/pLB Np3 displayed stronger cytotoxicity to HepG2/ADR cells, likely due to the ligand-mediated enhanced cellular uptake. The cellkilling efficacy of DOX-loaded pCA/pLB Np3 in media at different pH levels was evaluated by incubating drug-loaded pCA/pLB Np3 with HepG2/ADR cells for 24 h. Compared with pH 7.4, cell growth was strongly inhibited at pH 5.4 after the treatment with DOX-loaded pCA/pLB Np3 (Fig. 2G), due to the enhanced internalization and drug release of the peeled nanoparticles. The zwitterionic pCBAA coating on the surface of the nanoparticles is expected to resist opsonization, inhibit nonspecifc cell uptake, and prolong the circulation time in vivo. 17 As shown in Fig. S14, † pCA/pLB Np3 exhibited a much longer circulation profle (t 1/2 ¼ $13.59 h) in mice compared with their counterparts, which were assembled solely from the pLAMA 30 -b-pAAPBA 20 copolymer (t 1/2 ¼ $5.57 h). The signifcantly extended vascular residence time of pCA/pLB Np3 is expected to greatly facilitate drug accumulation in the tumor and reduce adverse effects. To evaluate the antitumor efficacy of the obtained nanoparticles, two groups of HepG2/ADR tumor-bearing Balb/c mice were intravenously administered the same (doxorubicin equivalent) dose of free DOX and DOX-loaded nanoparticles, and the control group was administered PBS only. As shown in Fig. 3A, the tumors grew rapidly without treatment, and a 4.16-fold increase in tumor volume was observed in a period of two weeks. Free DOX showed a moderate inhibitory effect, with only a 3.27-fold increase in tumor volume during the same period. In contrast, the nanoparticle-treated group displayed complete tumor inhibition following the administrations. The rapid blood clearance and insufficient tumor accumulation of free DOX limited its therapeutic efficacy, while pCA/pLB Np3 restrained tumor growth more effectively by protecting DOX from premature clearance as well as enhancing tumor accumulation and retention. No obvious body weight loss was observed, suggesting that the drug dosage was safe for mice (Fig. 3B). Administered DOX is generally metabolized via the slow hepatobiliary route, and long-term drug retention in the reticuloendothelial system increases the risk of organ toxicity. 18 To evaluate drug-induced organ damage, a histological analysis was performed on the major organs (heart, liver, spleen, lungs and kidneys). As shown in Fig. 3C, multifocal hepatic necrosis and apparent inflammatory cell infltration were observed in the livers of the mice treated with free DOX, while no apparent cell or tissue injuries were detected in the PBS or DOX-loaded nanoparticle groups. Currently, the integration of diagnosis and therapy has become an inevitable tendency in multifunctional nanoparticlebased drug delivery systems to achieve targeted drug transport, real time tracing and therapy. Herein, a HepG2/ADR tumorbearing mouse model was exploited to study the in vivo fate of the designed nanocarrier, using its integrated extraordinary imaging capacity. As shown in Fig. S15A, † pCA/pLB Np3 exhibited remarkable fluorescence signals in the tumor postinjection, indicating its rapid tumor accumulation. A similar phenomenon was detected in the tumor tissues harvested from tumor acid-sensitive polymeric vectors 2 h postinjection. 5b Importantly, rapid tumor accumulation and a long retention time are useful for cancer therapy, though many previous nanoparticles have been shown to accumulate in other tissues. The decisive step for the rapid tumor accumulation of this nanocarrier is its robust tumor penetration owing to the smaller size resulting from the pH-triggered detachment of its zwitterionic shell at tumor sites. It has been validated that the penetration of nanoparticles into the tumor relies heavily on particle size, with the consensus that smaller particles have enhanced tissue penetration. 5h The tumor fluorescence signals were quantifed and shown to reach 1.8 10 7 ps 1 cm 2 sr 1 even at 24 h post-injection. Since the short excitation/emission wavelength (506/523 nm) of BODIPY limits deep photon tissue penetration, the fluorescence in BODIPY was not very intense. Twenty-four hours after the intravenous administration of the nanocarriers, the mice were sacrifced to isolate the major organs and tumors for ex vivo imaging. Fig. S15B † shows the biodistribution and quantifed BODIPY fluorescence signal intensity. A signifcant fraction of the injected nanoparticles was still observed at the tumor site, while far less accumulation was detected in the heart, suggesting that these nanoparticles may effectively mitigate DOX cardiotoxicity. To thoroughly understand the rapid accumulation and long retention time of the nanoparticles within the tumor, it is indispensable to monitor the nanocarrier pharmacokinetics inside the targeted tumor tissue in situ in real-time with a high spatial resolution. In this work, a combination of DFT technology (Fig. S16 †) and indocyanine green (ICG) as a tracer agent was used to track the fate of the nanocarriers and semiquantitatively monitor the content of the released drug by combining pharmacokinetic principles with mathematical models. A sequence of fluorescence yield images was reconstructed, and then pharmacokinetic images were estimated from the fluorescence concentration-time curve. The relationship between fluorescence yield g(t) and concentration C(t) at time t could be linearly described using eqn (1). Where x and h are the fluorescence extinction coefficient and quantum efficiency, respectively. For ICG, x ¼ 0.013 mm 1 mM 1 , and h ¼ 0.016. The concentration-time curve was obtained by a biexponential-curve-ftting method based on the normally used twocompartment model. concentration-time curve and the pharmacokinetic parameters were obtained by the least squares ftting method, as shown in Fig. 4E and F. The parameters A and B of the ICGloaded nanoparticle group were signifcantly increased compared with those of the free ICG group, indicating the enhanced drug accumulation in the tumor cells. Furthermore, the larger a and smaller b indicate more efficient drug accumulation and longer retention in the tumor, respectively. Each mouse was given a single dose of 20 mg of ICG, and the pCA/ pLB Np3 group exhibited a fve-fold higher (6.03 mg) tumor accumulation than the free ICG group (1.13 mg) within 5 min after administration. The free ICG, however, was then quickly cleared from the tumor, whereas the nanoparticle delivered ICG exhibited much higher tumor retention at 24 h. The intratumor AUC of free and nanoparticle delivered ICG from 5 min to 24 h was extrapolated to 47.78 and 139.05 a.u. min 1 (p < 0.05), respectively. The lack of nanomedicine accumulation throughout the entire tumor may be one of the primary reasons for the compromised therapeutic efficacy, which is caused by both the tumor pathological characteristics and the inappropriate physicochemical properties of nanomedicines. In this study, our nanoparticle system enables its basic physicochemical properties to adaptively change in response to the endogenous stimuli of the tumor microenvironment to achieve improved therapeutic efficacy by successively increasing the blood circulation time, improving tumor accumulation, facilitating cell internalization and accelerating intracellular drug release. We used the acidic extracellular pH as the stimulus to release small particles and activate the targeting moieties at the tumor sites, driving the peeled nanoparticles to penetrate into tumor cells. It should be noted that the post shrinkage size is the pivotal determinant of the therapeutic efficacy of nanoparticles. 11a To the best of our knowledge, this is the frst report of real-time tracing of nanovehicles inside tumors at the spatial level. In conclusion, we have designed a hierarchical nanoassembly for efficient drug delivery to overcome multidrug resistant tumors and to facilitate real-time imaging. This strategy utilized dynamic covalent chemistry to surmount a series of drug delivery obstacles. A zwitterionic polymer shell is retained in the systemic circulation for a long time to achieve rapid accumulation in the tumor and to mitigate side effects. When the designed nanoparticle reaches the tumor tissue, the acidic microenvironment triggers the detachment of the zwitterionic shell and activates the surface-binding ligands. The particle size decreases from 165 to 48 nm, which greatly facilitates tissue penetration and cellular uptake. Collectively, these merits endow the nanocarriers with a superior anti-tumor therapeutic efficacy. In addition, the intrinsic fluorescence from the BODIPY groups makes these nanocarriers a potential theragnostic platform. Finally, the intratumoral pharmacokinetic properties of these nanocarriers were evaluated in situ in real-time using noninvasive and radiation-free 3D imaging technology. Collectively, our fndings advance the understanding of the behavior of nanovehicles in vivo, and provide guidance to improve drug delivery strategies.

chemsum

{"title": "<i>In situ</i> real-time tracing of hierarchical targeting nanostructures in drug resistant tumors using diffuse fluorescence tomography", "journal": "Royal Society of Chemistry (RSC)"}

remote-controlled_exchange_rates_by_photoswitchable_internal_catalysis_of_dynamic_covalent_bonds

## Abstract: The transesterification of boronate esters with diols is tunable over 14 orders of magnitude. Rate acceleration is achieved by internal base catalysis, which lowers the barrier for proton transfer. Here we report a photoswitchable internal catalyst that tunes the rate of boronic ester/diol exchange over 4 orders of magnitude. We employed an acylhydrazone molecular photoswitch, which forms a thermally stable but photoreversible intramolecular H-bond, to gate the activity of the internal base catalyst in 8-quinoline boronic ester. The photoswitch is bidirectional and can be cycled repeatedly. The intramolecular H-bond is found to be essential to the design of this photoswitchable internal catalyst, as protonating the quinoline with external sources of acid has little effect on the exchange rate. Dynamic covalent chemistry (DCC) combines the strength and directionality of covalent bonds with the reversibility of supramolecular interactions. 1 Owing to their tunability and robustness, dynamic covalent bonds have found wide application in library synthesis, 2 bioconjugation, 3 self-assembled macrocycles and cages, 4 covalent organic frameworks, 5 and self-healing polymers. 6 Light is an attractive stimulus to modulate the formation and exchange of these bonds because it can be applied non-invasively with excellent spatial and temporal control. Photoswitches, which can be reversibly switched between two states using different wavelengths of light, offer a unique opportunity to remotely control DCC. 7 Previously, photoswitches have been employed to govern the reactivity of dynamic covalent bonds via two principal strategies: (i) by rendering the dynamic bond active or inactive through lightdriven valence bond tautomerization; and (ii) by tuning the reactivity of the dynamic bond with an adjacent photoswitch. Hecht used azobenzene and spiropyran photoswitches to mask/unmask an activating hydroxyl group ortho to an aldehyde, which tunes the kinetics of imine formation by 2.4-and 3.1-fold, respectively. 15 Our group recently showed that azobenzene photoswitches can control the equilibrium of the boronic acid-diol condensation, due in part to the formation of intramolecular Hbonds (Figure 1A). 16 This change in thermodynamics was translated into hydrogels with reversibly photocontrolled stiffness. 17 However, the azobenzene boronic ester design was not effective for photocontrolling kinetics, displaying only 2.5-fold change in transesterification rate upon photoisomerization (see Supporting Information (SI), Figure S6). Here, we present a strategy to remotely control the kinetics of dynamic covalent reactions by designing a photoswitch that modulates the reactivity of an internal catalyst. We have termed this approach photoswitchable internal catalysis (PIC). This is distinct from photoswitchable "external" catalysis, in which exogenous photoswitchable catalysts are introduced to modulate a structurally separate exchange reaction. 18,19 Internal catalysis, in contrast, exploits proximity effects (neighboring-group participation) to alter the kinetics of a dynamic covalent reaction. While internal catalysis involves a 1:1 molar ratio of the neighboring group and dynamic covalent bond, the group can be considered catalytic because multiple exchanges (turnovers) occur at the dynamic covalent site without its consumption. 25 PIC can be applied to associative, degenerate exchange, whereas valence bond tautomerization has only been demonstrated for dissociative reactions such as Diels-Alder cycloaddition. We report a PIC design that is capable of switching the exchange rate between boronic ester and free diol over 4 orders of magnitude (Figure 1B). Boronic ester transesterification represents an ideal chemistry to demonstrate PIC due to its wide dynamic range. In 1984, Wulff reported that the rate of exchange of boronic ester with diols spans a remarkable 14 orders of magnitude depending on the structure. 26 Wulff proposed that the exchange reaction proceeds via three fundamental steps: (I) addition, (II) proton transfer, and (III) elimination (Figure 1C). Small changes in structure can alter the identity of the rate-limiting step, resulting in a dramatic change in rates (Figure 2B). In simple phenylboronic esters, proton transfer was determined to be rate limiting (II>III>I). B-N coordination or steric hindrance slow the exchange by increasing the barrier for addition (I>II>III). In contrast, installing a proximal basic group significantly decreases the barrier for proton transfer through internal catalysis, making elimination rate limiting (III>II>I). While the commonly used 2-aminomethyl "Wulff-type" phenylboronic esters exhibit this effect, the rate enhancement is even greater in 8-QBE, thanks to the proximity of the quinoline nitrogen lone pair to the boronic ester 27,28 and the rigid aromatic structure. 29 We imagined that the rate-limiting step of boronic ester exchange could be remotely tuned with an appropriate photoswitch, thus altering the rate of exchange. Our strategy was to deactivate internal catalysis in 8-QBE by engaging the quinoline lone pair in an intramolecular H-bond. Our attention was drawn to a class of acylhydrazone photoswitches that bear an acidic amide N-H bond. Acylhydrazones derived from 2-pyridinecarboxyaldehyde or 2-quinolinecarboxaldehyde are thermally stable in the Z isomer thanks to the formation of a six-membered intramolecular H-bond. 34 We envisioned that the intramolecular H-bond, in addition to stabilizing the Z-isomer, could deactivate internal catalysis in 8-QBE. This design yielded PIC (Figure 1B), wherein exchange is accelerated by internal catalysis when the hydrazone adopts the E configuration and slowed in the Z isomer. We synthesized the boronic acid (E)-PICacid in 6 steps from 2bromoaniline and crotonaldehyde (see SI for details). First, we investigated its photoisomerization by monitoring the conversion from E to Z by 1 H NMR in DMSO-d6. Initially, a sharp singlet appears at 12.23 ppm in the E isomer, corresponding to the free acylhydrazone N-H bond, and at 9.47 ppm, corresponding to the Hbonded boronic acid (Figure 2A and S25). Irradiation at 300 nm promotes conversion from (E)-PICacid→(Z)-PICacid (Figure 2B). After isomerization, the N-H peak shifts downfield to 14.33 ppm, indicating the formation of a strong intramolecular H-bond, and the O-H peak shifts upfield to 8.37 ppm. In the boronic acid form, the Z isomer is relatively unstable, with a thermal half-life of 7.3 hours at 25 ºC (Figure S2). This relatively short half-life is ascribed to the acidity of the boronic acid, which accelerates thermal isomerization. 35 Z→E isomerization is promoted by 350 nm irradiation. After condensation with excess neopentyl glycol, bidirectional switching between (E)-and (Z)-PICester with 300 (93% Z PSS) and 350 (56% E PSS) nm light is observed in DMSO-d6 (Figure S3). Furthermore, (Z)-PICester displays significantly improved thermal stability, with an extrapolated thermal half-life of 102 days at 25 ºC (Figure S2). We anticipate that derivatization of the acylhydrazone will enable further optimization of the photochemical properties. 33 The fatigue resistance of other hydrazone photoswitches is quite high (up to 300 cycles). 33,36 Using UV-Vis, we monitored E→Z isomerization of PICacid over time during irradiation at 300 and 350 nm and found that the Z and E PSS are reached within 1 and 2 minutes, respectively, in acetonitrile at 1.56 × 10 −5 M (Figure S5). Monitoring the UV-Vis absorption of PICacid at 322 nm during alternating irradiation at 300 and 350 nm, we observe no loss in efficiency after 10 cycles under ambient conditions (Figure 2C). Therefore, the presence of a boronic acid does not affect the robustness of the hydrazone photoswitch. We tested the effect of photoisomerization on the degenerate exchange between neopentyl glycol and the corresponding boronic ester (E)-PICester (1:1, 100 mM). Toluene was used as the solvent, and a small amount of acetone was added to fully dissolve the diol (9:1 toluene-acetone). At 25 ºC, 1 H NMR of (E)-PICester shows a broad peak around 1.0 ppm, signifying that the dynamic exchange between diol and ester is occurring faster than the NMR timescale at this temperature. The rate of exchange in (E)-PICester could be determined through coalescence between -CH3 resonances in bound and unbound neopentyl glycol by variable-temperature 1 H NMR (VT-NMR). Upon cooling the mixture, we observe the coalescence temperature of the bimolecular degenerate exchange to be -15 ± 5 ºC (Figure 3). The rate of exchange for (E)-PICester was thus determined to be 4.1 ± 1.4 x 10 3 s -1 at 25 ºC with an activation energy of 12.4 ± 0.2 kcal/mol (see SI for details). A second coalescence temperature is observed at lower temperatures due to a unimolecular fluxional ring flip of the neopentylglycol boronate (Figure S8). 37 After irradiation with 300 nm light to form (Z)-PICester, the bimolecular exchange is significantly slowed, and a coalescence temperature could not be observed even at elevated temperatures (Figure S10). Therefore, we monitored the transesterification using 1 H-1 H 2D exchange spectroscopy (EXSY) NMR at 45 ºC, revealing an exchange rate of 0.19 ± 0.01 s -1 (Figure 3). Using the VT-NMR data, we calculate that the exchange rate for (E)-PICester at 45 ºC is 1.5 x 10 4 s -1 . Therefore, the relative rate of exchange kex,E/kex,E is 8.0 x 10 4 . This krel value is 3-4 orders of magnitude larger than those achieved by previous photoswitch designs 15,17,38 and suggests the potential for PIC as a strategy to remotely control the rates of associative exchange reactions. In addition to PIC, it is possible that the effect of photoswitch conformation on the Lewis acidity of the boronic ester contributes to rate differences. 39 The effect of the photoswitch was qualitatively observed for other diols (pinacol, 1,2propanediol) but krel could not be measured for both isomers (Figures S20-21). Exchange is slowed in more polar solvents such as DMF-d7, but a significant difference in E and Z exchange rates is maintained (Figures S16-S19). To show that chemical stimuli cannot regulate internal catalysis to the same extent, we synthesized the neopentyl glycol ester of (2methylquinolin-8-yl)boronic acid (Me-QBE, SI). VT-NMR revealed that Me-QBE undergoes exchange roughly 10 times faster than to (E)-PICester (~4.9 x 10 4 s -1 at 25 ºC, Figure S6). We hypothesize that the electron-withdrawing hydrazone in (E)-PICester reduces the basicity of the quinoline, slowing exchange relative to Me-QBE. When Me-QBE is exposed to 1.0 equivalent of trifluoroacetic acid (TFA), an acid capable of fully protonating the quinoline, only a moderate decrease in exchange rates was observed (~3.0 x 10 4 s -1 at 25 ºC, Figure S7). This observation is consistent with the fact that esterification of boronic acids can be catalyzed by both acid and base, 40,41 so external proton sources cannot deactivate internal catalysis. These experiments further highlight the importance the intramolecular H-bond in our design to deactivate internal catalysis. Additionally, Letsinger 27 and Wulff 26 have shown that the presence of exogeneous quinoline does not increase the transesterification rate for simple phenylboronic esters, indicating that the high effective molarity provided by internal catalysis is crucial for accelerating exchange. Guan has shown that internal catalysis of boronic ester 20 and silyl ether 21 exchange provides five and three orders of magnitude rate acceleration, respectively. These differences in small-molecule exchange rates were translated into significant and measurable differences in the physical properties of polymer networks (selfhealing ability and stress relaxation), suggesting that the observed krel of 8.0 x 10 4 is sufficient to mediate macroscopic changes. We prepared a viscoelastic polymer network by condensation of 4-arm poly(caprolactone) with (E)-PICacid at elevated temperature. Using oscillatory shear rheology at 100 ºC, we observed a decrease in the crossover frequency corresponding to slower exchange kinetics when the network was photoswitched at 300 nm (Figures S23-24). No change in storage modulus was observed, consistent with an associative mechanism. 6 We have demonstrated the use of a bidirectional hydrazone photoswitch to control the rate of exchange between boronate ester and diol over 4 orders of magnitude. The dramatic change in rates afforded by reversible deactivation of internal catalysis lays the foundation for photocontrolling kinetics in different dynamic covalent reactions. The ability to remotely and reversibly control a dynamic covalent exchange rate can be translated to turn on and off assembly and reconfiguration in smart materials. The design of photoswitchable internal catalysts for dynamic reactions that can be incorporated into photocontrolled hydrogels is ongoing in our laboratory. 45 Figure 3. 1 H VT-NMR of the diol-ester exchange of (E)-PICester (left, Tc highlighted) and 1 H-1 H 2D EXSY NMR of the exchange of (Z)-PICester (right, crosspeaks highlighted) (1:1 diol:ester, 0.1 M in 9:1 toluene-d8-acetone). The proposed effect of the photoswitch on the proton transfer step is illustrated in the center.

chemsum

{"title": "Remote-controlled exchange rates by photoswitchable internal catalysis of dynamic covalent bonds", "journal": "ChemRxiv"}

molecular_dynamics_based_descriptors_for_predicting_supramolecular_gelation

## Abstract: Whilst the field of supramolecular gels is rapidly moving towards complex materials and applications, their design is still an effortful and laborious trial-and-error process. Herein, we introduce four new descriptors that can be derived from all-atom molecular dynamics simulations and which are able to predict supramolecular gelation in both water and organic solvents. Their predictive ability was demonstrated via two separate machine learning techniques, a decision tree and an artificial neural network, with a dataset composed of urea-based gelators. Owing to the physical relevance of these descriptors to the supramolecular gelation process, their use could be conceptualized to other classes of supramolecular gelators and hence steer their design. ## Introduction In recent years, Low Molecular Weight Gelators (LMWGs) have attracted signifcant attention. Currently, the feld is focussed on developing efficient supramolecular gels for various specialized applications, ranging from drug delivery systems to catalyst templates or even optoelectronic applications. Despite two decades of intense research on supramolecular gelators, their discovery remains surprisingly reliant on serendipity, due to the sensitivity of the supramolecular gelation process towards small molecular changes of the LMWG. In this context, multiscale computational methodologies have the potential to provide a better understanding of the underlying relationship between the molecular structure and the gelation ability. 7 However, besides post-rationalization, the ability to predict supramolecular gelation by means of computational methods has been scarcely explored. The frst predictive tool for organogel formation was reported in 2011 by Raynal and Bouteiller. 8 In their pioneering work, Hansen Solubility Parameters (HSPs) were employed to defne solubility and gelation spheres in Hansen space by means of an elaborate assessment of the behaviour of a known Low Molecular Weight Gelator (LMWG) in several solvents. Based on these spheres, the gelation performance of the LMWG in an untested solvent could then be predicted using the HSPs of this new solvent. If the HSP values fall inside the solubility sphere or the gelation sphere, it is likely that the LMWG will, respectively, be soluble or form a gel in the solvent. Follow-up works from the group of Bouteiller further improved the quality and scope of this method to determine the gelation domain from the solubility data of LMWGs. 9,10 This approach, however, still requires the synthesis of the molecule and an extensive gelation screening beforehand, as for each new LMWG, solubility data needs to be gathered to defne the gelation domain. Furthermore, the reliability of the prediction depends on the quality of the initial solubility data set. 8 The combined effort of the Tuttle and Ulijn groups resulted in the development of a predictive method for the self-assembly properties of tripeptides in water based solely on computations. 11,12 Using high throughput coarse-grained molecular dynamics simulations, a hydrophobicity-corrected aggregation propensity score (AP H ) could be obtained, which originates from the solvent accessible surface area of the aggregate as well as the partitioning coefficient (log P) of the gelator. 13 By screening the AP H score of 8000 tripeptides in water, they were able to bring forth a set of design rules to promote aggregation and supramolecular hydrogelation. As their method focuses on the hydrogelation of tripeptide gelators, the applicability to non-peptide gelators requires a specialized coarse-grained model. Recently, Adams and Berry developed a machine learning model to successfully predict the hydrogelation propensity of functionalised amino acids and dipeptides using physicochemical properties and molecular fngerprints. 14 Descriptors such as the number of rings, polar surface area, solvent accessible surface area and log P emerged as key parameters for predicting gelation, together with a number of molecular fngerprint descriptors, which are abstract and difficult to interpret. An added value of such approach is that it can be offered via an online interface, since only a SMILES code is needed as input to obtain a prediction of the hydrogelation properties of the molecule. Moreover, next to the prediction itself, the relevancy of the prediction is indicated. The predictions are accurate as long as the molecule falls within the applicability domain of the model, defned by the properties of the amino acids and dipeptides present in the training set. To recreate a similar predictive model for non-peptide gelators that currently fall outside the applicability domain of their prediction model, a signifcant amount of new data is required. Among the different types of LMWGs, peptide-based hydrogelators belong to one of the best represented classes of gelators due to their potency in several biomedical applications. 15,16 While they beneft from biocompatibility, their synthesis is often costly and time consuming. In contrast, the class of ureabased gelators enjoy a cheap and straightforward synthesis allowing easy derivatization. Recently, we reported on a thixotropic and cytocompatible bis-urea derivative with potential in biomedical applications. 17 In this work, we set out to develop a conceptual molecular dynamics guided predictive model for urea-based supramolecular gelation to improve their current empirical design strategy. In essence, we aim for an approach fulflling the following four criteria: (i) the method should be able to correctly predict supramolecular gelation of urea-based molecules by means of computations exclusively. (ii) Instead of a binary yes/no answer, the outcome of the predictions should be a three-level categorical response: gel, precipitate and solution. (iii) Only descriptors with a physical relevance towards supramolecular gelation should be used to build the model, in order to provide chemical insights into the structure-gelation relationships. With suitable descriptors, the approach could be conceptualized to other supramolecular gelator classes. (iv) And fnally, the predictions should be accurate in both water (hydrogelation) and organic solvents (organogelation) (Fig. 1). ## Methodology The dataset To construct the model and test the predictive ability of our approach, a library of urea-based LMWGs was generated, which is shown in Fig. 2. Some LMWGs were previously reported by other research groups (1, 2, 3, 4 and 5), while other compounds were recently synthesized by our group (6, 7, 8, 9, 10 and 11). 7, The compounds were selected on the basis of two criteria. First, the molecule should contain at least one urea moiety. Second, the experimental protocol to assess the gelation performance in a certain solvent should be exactly the same for all compounds. It is crucial to fulfl the latter criteria in order for the data to be consistent, since the gelation procedure signifcantly affects the gelation propensity. 22 This imposes serious restrictions on the data available in literature on urea-based LMWGs that can be used in this study, as gelation procedures often vary or are ill-defned. Herein, all compounds have been screened for their gelation ability based on the same procedure, involving a heating and cooling cycle to obtain the gel at a concentration of LMWG smaller or equal than 1.0% w/v. The full library contains non-gelators, organogelators as well as hydrogelators. A total of 65 data points has been obtained by screening the gelation performance in different solvents. For compound 1-5, gelator-solvent combinations were selected based on available literature data keeping diversity in gelation properties in mind. Whilst data on compound 6-11 was generated by our own experiments, with solvents being selected based on their difference in polarity and hydrogen bonding capabilities. Although the size of this dataset is modest, previous studies in different felds have shown to deliver a promising predictive model with similar dataset sizes, by using models with a low complexity or by means of chemically relevant descriptors. 14,23,24 ## Molecular dynamics In our previous work on the rationalization of supramolecular hydrogelation through a multiscale computational approach, we have showed that all-atom molecular dynamics simulations, emerge as a unique tool to provide insight into the aggregation phase during supramolecular gelation. 7 Similarly, in the pioneering work of Tuttle and Ulijn, 11 coarse grained molecular dynamics was applied to obtain their AP H score and predict the hydrogelation performance of tripeptides. Herein, molecular dynamics simulations are performed to obtain a set of descriptors, which we envisioned to have predictive ability value in supramolecular gelation. The MD simulations were performed by placing 5 gelator molecules in a periodic cubic box flled with solvent molecules to reach a concentration of 1.0% w/ v. The same concentration is used in the experimental determination of the gelation performance. A trajectory of 50 ns with a timestep of 0.5 fs was gathered using the CHARMM27 force feld, with the parameters retrieved from the Swissparam service. 25,26 This force feld was already shown to accurately describe urea based interactions, which are key during the supramolecular gelation of urea containing compounds. 7,27 Before production, an energy minimization and temperature/ pressure equilibration step was performed to ensure steric clashes or inadequate equilibration of the NPT-ensemble would not affect the production simulation. The V-rescale thermostat and Parrinello-Rahman barostat were used with the Fig. 1 Comparison of reported supramolecular gelation predictive models with the approach followed in this work. temperature set at 300 K and the pressure at 1.0 bar. 28 All molecules were randomly placed inside the simulation box with the gelator molecules separately being dispersed in the solvent, using the Packmol software, while simulations were run with the Gromacs software version 2018.3. 29,30 All descriptors were obtained as a time average over the 50 ns simulation, with a snapshot taken every 2.5 ps. A detailed description of the methodology can be found in the ESI (S3 †). ## Descriptors for supramolecular gelation For a molecule to act as an efficient low molecular weight gelator, a number of criteria needs to be met: (i) the molecule should have a suitable solvophobic balance, (ii) it should contain sites for non-covalent interactions allowing the formation of a reversible network and (iii) these non-covalent interactions should promote the anisotropic growth of a Self-Assembled Fibrillar Network (SAFiN) that immobilizes the solvent and causes the distinct supramolecular gel features. 31 Four MD generated descriptors are introduced in this work that quantify one or more of these criteria and thus could have the ability to predict supramolecular gelation. Relative solvent accessible surface area (rSASA). The Solvent Accessible Surface Area (SASA) is defned as the total area of a molecule that is accessible to the solvent. If unfavourable interactions are present between the molecule and its solvent, the molecule will tend to decrease the contact area with the solvent by aggregation and as a result a small SASA is observed (Fig. 3B). As such, the SASA is associated with the solvophobic balance of the molecule under investigation and its intermolecular aggregation. Tuttle and Ulijn use the SASA to determine their aggregation propensity score, as previously mentioned. 11 In this work, a relative Solvent Accessible Surface Area (rSASA) is introduced, which is computed by dividing the time-average of the combined SASA SASA Á of the gelator molecules during the 50 ns simulations with a maximum SASA (SASA max ). The latter is obtained by multiplying the SASA of a single fully extended gelator molecule (i.e. all dihedral angles of the backbone are set to 180 ) with the number of gelator molecules present in the simulation box. By taking the time-averaged value of the SASA instead of the SASA of the fnal frame of the simulation, a robust score is obtained important for small scale all-atom simulations. The SASA and SASA max are computed by the double cubic lattice method with the radius of the solvent probe set at 1.4 . 32 Values close to 1 of the rSASA descriptor indicate absence of aggregation, while values signifcantly smaller than 1 indicate solvophobic interactions triggering aggregation. Relative end-to-end distance (rH). In our previous work, we showed how the end-to-end distance is a valuable descriptor to defne the shape of a single gelator molecule. 7 To assess the effect that the solvent environment, other gelator molecules and intramolecular interactions have on the shape of the gelator molecule, a relative end-to-end distance (rH) is defned, which is computed by measuring the average distance between the most distant atoms of the backbone over time in all gelator molecules present in the simulation ( R) and dividing this value with the maximum end-to-end distance, obtained by measuring the distance between the respective atoms of a corresponding fully extended molecule (R max ) (Fig. 3C). Similarly, it is rationalized that, if sufficient rotatable bonds are present in the molecule, the rH descriptor evaluates the interactions between the gelator molecule and the environment. If the molecule has pronounced solvophobic interactions, it will decrease its contact with the solvent resulting in a collapsed shape and a decreased value of rH. Therefore, rH can be regarded as a measure for intramolecular aggregation. Nevertheless, it is important to note that, next to gelator/solvent interactions, intramolecular and intermolecular gelator/gelator interactions could also have a pronounced effect on the value of rH. Indeed, upon self-assembly of the gelator molecules to a nanofber, one can assume that in the centre of the fber, the gelator molecules are mainly surrounded by other gelator molecules. While at the surface, the gelator molecules are in contact with solvent molecules and here solvent-gelator interactions need to be taken into consideration. As molecules are constantly moving during the simulation, it is not straightforward to make a clear differentiation between molecules located in the centre of the aggregate or at the surface. For this reason, the average is taken over all gelator molecules in the simulation. Additionally, the flexibility of the gelator molecule will have an effect on the value of this descriptor, as molecules with no or little rotatable bonds in their structure will have an almost constant value of rH irrespective of the solvent. Hydrogen bonding percentage (HB%). The urea moiety has unique hydrogen bonding characteristics, being able to act as a hydrogen bond donor as well as a hydrogen bond acceptor. Hydrogen bonding between urea moieties resulting in a urea atape motif is an important factor in the anisotropic fbre formation and gelation process of urea-based supramolecular gels. 33 However, previous work indicates other types of hydrogen bonding interactions, such as hydrogen bonding between a urea and a pyridyl moiety, to influence the gelation performance in these gelators as well. 7,34,35 With this in mind, the hydrogen bonding percentage descriptor (HB%) is introduced to quantify the non-covalent intermolecular interactions that connect the gelator molecules (Fig. 3E). To calculate the HB%, frst all classical hydrogen-bond donors (NH, OH, SH, .) and acceptors (O, N, F, .) in the molecule are identifed. Next, the sum is taken over every intermolecular connection between a hydrogen bond donor atom (i) and a hydrogen bond acceptor atom (j) over every time step (N) of the simulation. A connection is deemed present when the distance between the donor and acceptor atom is below 3 . This distance was selected based on hydrogen bond distances observed in an interaction library, containing a variety of urea-based hydrogen bonding interactions. 7 This sum is divided by the total number of time steps in the simulation and multiplied by 100% to get the fnal HB% value. As such, the HB% descriptor is closely linked to the native contact analysis applied to study protein folding. 36 Note that bifurcated hydrogen bonds are counted as a single connection during the analysis, as either one acceptor or one donor atom is involved (i.e. for a certain time step N and a certain value of i or j, the value of t ij cannot exceed 1). ## HB% the distance between atom i and j t ij ¼ 04no connection; the distance between atom i and j $ 3 Shape factor (F). While the above descriptors provide information concerning the solvophobic balance and non-covalent interactions of the gelator molecules, it is essential to be able to quantify the shape of the aggregate. For this reason, we present a shape factor (F) that can be calculated by taking the ratio of a time averaged computed radius of gyration ðR g Þ to a pseudo hydrodynamic radius ðR 0 h Þ, similar to the particle shape factor used in the feld of proteins and polymers (Fig. 3D). 37 R g is calculated as the square root of the mass averaged distance of all gelator atoms (A) to the centre of mass of all gelator atoms in the simulation. Classically, the hydrodynamic radius R h of a particle is measured by dynamic light scattering experiments and is defned as the radius of a hypothetical hard sphere that will diffuse with the same speed as the solvated particle under investigation. 37 Based on this, a computable pseudo hydrodynamic radius ðR 0 h Þ is suggested as the radius of a hypothetical hard sphere that has the same volume as the combined molecular volume of all gelator molecules (V gel ). The latter can be approximated by calculating the volume of a single extended gelator molecule using the double cubic lattice method with the probe radius set to 1. 4 and multiplying this value with the total number of gelator molecules present in the simulation. The purpose of the shape factor F lies in the description of the shape of the aggregate that is observed during the molecular dynamics simulation. When the aggregate has a spherical shape a low value of F is computed, when the aggregate adopts a more fbrous shape, F increases. The descriptors defned above can all be calculated based on data generated in a molecular dynamics simulation and by employing the open-source GROMACS software (version 2018.3) together with its implementations. 29 A detailed explanation on the practical aspects to obtain the descriptors is provided in the ESI (S4-S7 †). ## Gelation results A thorough gelation screening in multiple solvents was already performed on some of the urea-based compounds in our library (1, 2, 3, 4 and 5), making the classifcation of data associated with these compounds straightforward (Fig. 2). In order to get a meaningful and consistent three-level classifcation of the gelator-solvent combinations as precipitate (P), solution (S) or gel (G), a fxed gelation procedure and minimum gelation concentration (MGC) need to be defned. The gelation procedure consisted of heating the sample to dissolve the gelator followed by cooling to room temperature. The sample is deemed to be a gel if the material does not flow upon vial inversion. When the material does flow upon inversion, it is classifed as a precipitate if solid particles are observed, or as soluble when the sample is a clear solution. All gels that are formed have an MGC no higher than 1.0% w/v. From the results gathered in Table 1, it is clear that our dataset comprises ureabased non-gelators (5, 7, 8, 9, 10), organogelators (1, 2, 3, 4) and hydrogelators (6, 11) under the experimental conditions speci-fed above. ## Molecular dynamics derived descriptors For each gelator-solvent combination, a 50 ns molecular dynamics simulation was performed to obtain a time averaged value for the descriptors (Fig. 3). All possible 2D plots between the four molecular descriptors (rSASA, HB%, rH and F) are presented in Fig. 4, together with their respective linear regression R 2 value and Pearson correlation coefficient (r). From these graphs, it can be seen that the molecular descriptors rSASA, HB% and F have an R 2 value between themselves ranging from 0.747 to 0.866. Additionally, their corresponding correlation coefficient is either higher than 0.85 or lower than 0.85 suggesting a linear trend between these descriptors. The negative linear trend between rSASA and HB% and between HB% and F might be expected. Indeed, as the gelator molecules in the simulation tend to aggregate in a solvent, the values of rSASA and F decrease as they quantify respectively the aggregation tendency and shape of the aggregate being formed by the gelator molecules. The HB%, on the other hand, will increase as more intermolecular hydrogen bonds are being formed between the gelator molecules in the aggregated state compared to the soluble state. With the same reasoning, the positive linear trend between rSASA and F can be rationalized. One might argue that because of the apparent linear relationship between rSASA, HB% and F, the aforementioned descriptors provide the same information and hence could be reduced to a single property. However, the following thought experiment demonstrates their unique intrinsic value and their independency from one another. Imagine a class of gelators consisting solely out of carbon and hydrogen atoms. The HB% descriptor will be equal to 0, regardless of the solvent and the gelator molecule as there are no hydrogen bond donor or acceptor atoms present. However, in this case rSASA and F will still vary upon changing the solvent as the solubility of the molecule and the aggregates shape of the aggregates can be influenced by other gelatorsolvent interactions such as van der Waals interactions. In this case, no linear trend would be observed between HB% and rSASA or between HB% and F. Note, that this statement is supported by the 2-dimensional plots of rSASA vs. HB% and HB% vs. F in Fig. 4. In these plots, systems that are characterized by a value of HB% close to 0 still differ signifcantly in their values of rSASA (ranging from approximately 0.85 to 1.00) and F (ranging from approximately 2.00 to 3.00). Otherwise, the independency between rSASA and F can be demonstrated by imagining a set of gelator molecules that do not self-aggregate in a range of solvents. Here the rSASA will always be valued close to 1. Nevertheless, the value of F can still vary depending on the placement of the gelator molecules in the solvent. For example, a 1-dimensional alignment of the gelator molecules connected to each other through a solvent molecule will have a substantially larger value of F compared to a disperse placement of the gelator molecules, while both cases have an rSASA value close to 1. Although to a lesser extent, this is observable in the 2-dimensional plot of rSASA vs. F for values of rSASA close to 1. As such, while the descriptors rSASA, HB% and F might seem to be linearly correlated for the systems under investigation, they are independent from each other and provide their own unique information. rH shows no apparent linear trend with any of the molecular descriptors. This is because rH is highly dependent on the structure and flexibility of the gelator molecule and only mildly dependent on the aggregation behaviour. Compound 6, for example, has only two free rotatable bonds in its backbone, resulting in an rH value larger than 0.9, independent of the solvent. In contrast, compound 11 contains a large amount of rotatable bonds in its backbone, resulting in an rH value ranging from 0.55 to 0.78 (Table S2 †). With this in mind, rH is a descriptor of the flexibility of the molecule under study. When allocating the data points in the graph to the respective experimental result of the gelation test (green ¼ soluble, blue ¼ gel, red ¼ precipitate), it is clear that soluble samples are characterized by a high value of rSASA and F and a low value of HB%, whereas a gel material is characterized by intermediate values of rSASA, F and HB%. Following this reasoning, samples resulting in a precipitate should be characterized by a low value of rSASA and F and a high value of HB% (Fig. 5). While a majority of the data agrees with this trend, there remain several outliers, especially when the sample forms a precipitate. For example, compound 8 in 1-octanol is experimentally clas-sifed as a precipitate (Table 1). Nevertheless, from the respective molecular dynamics simulations, relatively high values of rSASA (0.8818) and F (2.74) and a low value of HB% (162.40%) were obtained (Table S2 †). This makes prediction of supramolecular gelation by visually inspecting the 2D-plots of the molecular descriptors challenging. However, we believe that the latter can be achieved by increasing the size of the simulation box, increasing the total simulation time and/or using more accurate sampling techniques such as ab initio molecular dynamics. At present, these methods would render the simulation computationally intractable. ## Prediction using machine learning methods In recent years, several machine learning (ML) methods have established themselves in different areas of chemical research. 38 For example in targeted drug discovery, where ML can be used to model quantitative structure-activity relationships (QSAR). 39,40 Also in theoretical chemistry these methods have shown to assist the interpretation of complex calculations, replace otherwise computationally demanding methods, develop new accurate density functionals or force felds and even predict the electronic charge density and density of states within the framework of density functional theory (DFT). Moreover, in the feld of materials discovery, ML methods have shown their usefulness as evidenced by the earlier referred work of Gupta et al. 14,46,47 In this work, two separate ML methods, a decision tree and an artifcial neural network (ANN), are used to showcase the ability to predict supramolecular gelation by the proposed molecular descriptors. 48 To achieve this challenging goal, the data is partitioned as follows: all data points coming from compound 11 will be used for testing, while the rest is used for training and validating the models (Table S3 †). As such, the classifcation ability of the models will be tested on an unseen compound that can show any of the three responses (gel, soluble, precipitate) depending on the solvent. The models were constructed using JMP Pro version 14. 49 Decision tree. Decision trees (DT) are intuitive flowchartlike diagrams, where nodes create branches that partition the data based on a selected descriptor. 48 More nodes in the tree translate to a more branched, complex DT that has the tendency to produce an over ftted model. One of the major assets of DT over other machine learning methods is their transparent nature, making the prediction process easily understandable. The optimized DT model together with the response probabilities of each leaf is presented in Fig. 6. As is shown, the DT model contains fve nodes resulting in a total of six leaves. The optimization procedure for the DT model is described in detail in the ESI (S20 †). Upon closer investigation, leaf 2 shows the highest probability for the sample to be soluble (81.20%). When inspecting the flowchart, we can see that this leaf is characterized by a low value for HB% (<43.878) and a high value for rSASA (>0.958). Furthermore, leaf 3 is characterized by a high value for HB% (>43.877) and a low value for rSASA (<0.622) and shows the highest probability for a precipitate response (94.65%). This is in close agreement with our earlier observed trend for these descriptors (Fig. 5). The compound-solvent combination is predicted to be a gel if it is categorized in leaf 6, with the certainty of the prediction being 69.37%. This is considerably lower compared to the predictions made from leaf 2 and leaf 3 for soluble and precipitation respectively. A possible reason for this difference is most likely that cases where gelation is observed are rare in comparison to soluble and precipitated samples. Notably, the descriptor rH is never selected as a node in the optimized DT model. The node selection is based on the split that results in the statistically best performing model for the training data. Hence, this suggests that the rH descriptor does not provide the same predictive value in a partitioning model compared to the other three descriptors HB%, rSASA and F. Again, this is in line with our previous observations (Fig. 5). A subset of 14 data points, which was randomly stratifed according to the gelation response, was used as validation to assess the quality of the DT model and avoid over ftting. The remaining data was used for training (Table S3 † specifes which data is used during training, validation and fnal testing). Multiple measures of ft, such as the balanced accuracy (BA), entropy R 2 , the misclassifcation rate (MR), Cohen's kappa (K) and the area under the receiver operating characteristics curve for the three responses (AUROC (S), AUROC (G), AUROC (P)) are summarized for the training set (T), validation set (V) and a hypothetical perfect model (P) and random model (R) in Table 2. A defnition for each of these statistical evaluation metrics is provided in the ESI (S23 †). From Table 2, we observe that the DT model is adequately ftted as the training and validation data show similar values for all measures. Models that are over ftted are characterized by large discrepancies between the measures of ft obtained from training data and validation data (i.e. excellent measures of ft are obtained on the training set, but poor measures of ft are obtained on the validation set). Additionally, the DT exhibits substantial predictive behaviour, when comparing the measures of ft of the validation set to a perfect and random model. Artifcial neural network. Artifcial neural networks (ANN) are built out of a number of neurons, with each neuron accepting inputs, applying a weighted function to the inputs and forwarding the new information till eventually an output is reached. 48 The flexibility of ANN is evidenced by the myriad of hyperparameters that are adjustable, such as: the number of neurons, the transformation function used, number of hidden layers (if more than 3 hidden layers are used, the network is referred to as a deep neural network) and the method of optimizing the weight coefficients. 48,50 Due to this flexibility, highly accurate non-linear predictive models can be constructed. As such, in contrast to a decision tree, an ANN generally provides a less understandable "black box" predictive model. With this in mind, one needs to be extra wary for over ftting when architecting an ANN. This is usually accomplished by training and validating the model on hundreds to millions of data points. In this study, the amount of data is on the lower side of the spectrum. For this reason, two precautions were taken during the construction of the neural network to mitigate overftting issues. First, the neural networks architecture is kept relatively simple, with a maximum of 5 neurons being Table 2 Performance statistics of the decision tree model for the training set (T) and validation set (V). For comparative reasons, the measures of fit for a hypothetical perfect model (P) and random model (R) are also provided considered during the hyperparameter optimization (S21 †). As a neural network consists out of more neurons, the associated number of weights that need to be optimized during the training grows, which subsequently increases the complexity of the network and the possibility of overftting. Second, a 5-fold cross validation was employed instead of a percentage holdback validation. In a 5-fold cross validation, or in general a k-fold cross validation, the data is randomly partitioned in 5 (or k) subsets. Next, for each set a neural network is trained with 4 sets as training data and the remaining set to validate the model. In total 5 different models are built with each set being used for validation once. Using this approach, data usage is maximized as all data points (of molecule 1-10) are employed during training equally. Importantly, signs of overftting can be detected by discrepancies between the measures of ft obtained on the training and validation data. 51 The optimized ANN (Fig. 7) is built out of 1 hidden layer that consists of 5 hyperbolic tangent neurons of which the weight coefficients are determined by a weight decay procedure (S21 †). All measures of ft indicate that the ANN has excellent predictive abilities and outperforms the decision tree-based model (Table 3). Indeed, the measures of ft obtained on the training and validation data from the ANN model are closer to a hypothetical perfect model compared to the metrics obtained with the DT model. Most likely this can be explained by the superior flexibility of artifcial neural networks over classical decision tree models. Predicting supramolecular gelation of an unknown ureabased molecule. As mentioned earlier, none of the data associated with the gelation ability of compound 11 in different solvents was used during the development of the decision tree and the artifcial neural network. To test the predictive ability of both models on an unseen urea-based gelator, the DT and ANN models were applied to predict the outcome of the gelation tests of compound 11. Both models give satisfactory results, as 5 out of 6 cases for the gelation outcomes were predicted correctly (Table 4). Importantly, the two models successfully predicted that a supramolecular gel was formed in water. This confrms the ability of the proposed molecular descriptors to predict supramolecular gelation of urea-based compounds in a variety of solvents, with the prediction being a three-level classifcation of the sample being a solution, a precipitate or a gel. Upon closer inspection of the gelator molecules that are present in the dataset, it can be observed that compound 6-11 have a more similar molecular structure compared to compound 1-5 (Fig. 2). Therefore, the quality of prediction of the gelation performance of compound 11 with models where compound 1-10 were used during training and validation, might be explained by the high resemblance of 11 with 6-10. To further investigate this, several neural networks were optimized, with the same architecture as the neural network described above, i.e. 1 hidden layer and 5 hyperbolic tangent neurons with the weight of each neuron optimized through a weight decay procedure. But in each neural network all data associated with a certain gelator molecule from the data set was subsequently left out during the training of the model and used for validation, similar to a leave-one-out cross validation approach. Validation statistics of each molecule separately together with the average measures of ft across molecules 1-10 are provided in Table 5. The results of this analysis establish that the models provide the best predictive qualities for compound 6-10, as the lowest observed entropy R 2 for these molecules is equal to 0.99 (compound 6). Nevertheless, substantial predictive power is also observed for compound 1-4, with the lowest observed entropy R 2 for this set being equal to 5 are taken over a low amount of data. Additionally, we want to acknowledge that both the decision tree and neural network models validate the predictive ability of the derived molecular descriptors over supramolecular gelation, however for practical purposes, these models would beneft greatly from a substantial increase in training data. ## Outlook While this work showcases that predicting supramolecular gelation based on descriptors derived from molecular dynamics simulations is feasible, there are still a number of factors limiting their applicability. Here, an overview of these factors is given and we discuss how they might be mitigated or overcome in the near future. Robustness of descriptors. The four descriptors defned in this work: rSASA, rH, HB% and F are obtained through molecular dynamics simulations. Naturally, questions arise on the robustness and reproducibility of these descriptors as the results can be influenced by the total simulation time over which the descriptors are calculated, the initial topology of the simulation box and the randomness that is intrinsically related to molecular dynamics. The choice of the simulation time is determined by a balance between accuracy and computational workload. Longer simulations provide more accurate results but require more computer time. To ensure that a 50 ns would provide sufficient sampling to obtain trustworthy average values, a single 1 ms simulation was run for compound 6 in water. To make this simulation computationally tractable, the cubic simulation box edge was set at 40.32 and contained 5 gelator molecules and 2247 water molecules to reach a gelator concentration of 5.0% w/v. All other settings remained identical to the simulations that were performed to obtain the descriptors in this study. From the evolution of the average computed SASA during this simulation, the mean absolute percentage error (MAPE) on the average SASA obtained from a 50 ns is only 13% if the true average SASA is taken as the one obtained from the full ms simulation (Fig. 8). Hence, a total simulation time of 50 ns simulation can be regarded as adequate sampling, while still retaining sufficient computational speed. Especially if one considers that at 200 ns, which requires a computational workload that is 4 times higher as a 50 ns simulation, the MAPE still exceeds 5%. To further increase the reproducibility of the computed values of the descriptors, initial topologies of the simulation were ensured to have the gelator molecules completely dispersed in the solvent, i.e. no atoms of the gelator molecules are closer than 3.0 from each other at the start of the simulation. To demonstrate the robustness of this method, simulations of compound 2, 5 and 6 in DMSO, ethanol and acetonitrile respectively, were performed in triplicate. From Table 6 it is concluded that this method provides highly repeatable values for each descriptor, as the deviation between the three simulations for each descriptor is relatively small for all systems that were considered. Especially when comparing the standard errors on the averages with the full range of values that were obtained in this study, the reliability of this method is shown. Computational limitations. As the descriptors originate from a molecular dynamics simulation, their usefulness is highly dependent on how accurate the real system is modelled. As briefly mentioned above, simulating bigger systems (i.e. more gelator and solvent molecules) for a longer time (i.e. more timesteps) might increase the predictive potential. Furthermore, as computing power is constantly improving, the level of theory employed to run the simulations can become more accurate as well, enhancing the merits of the descriptors even further. 52 Next, it is also important to note that if one would want to apply this method to discover a new LMWG, a screening of hundreds to thousands of compounds might be necessary. This level of throughput is at present not computationally viable at an all-atom scale. The latter can become possible, however, by software benefting from state-of-the art technologies and hardware. 53,54 Applicability domain. It is important to delineate the boundaries of the applicability domain of the predictive methods proposed in this work. 55 While the focus was to predict supramolecular gelation of simple organic urea-based molecules, these boundaries are still somewhat ambiguous. If certain types of atoms or functional groups are not well represented during the training of the predictive model, cases for which the LMWG contains such a functional group might fall outside the applicability domain, even if the LMWG is a ureabased molecule. For traditional QSAR models, various methods exist to determine the applicability domain, such as a Principle Component Analysis (PCA), distance to model (DM) or a K Nearest Neighbours (KNN) approach. These methods are, however, ineffective in this case due to the relatively low number of descriptors and data points. One way to mitigate this problem, is to scan the applicability boundary by calculating the descriptors and implementing a similar method to predict other classes of materials, such as peptide or glycosylated supramolecular gels. Additionally, we underline that the models only predict supramolecular gelation based on a specifc gelation procedure and minimum gelation concentration. This is important because some molecule-solvent combinations might be classifed as a non-gel by the model, while a different gelation trigger or concentration does render them a gel. For example, compound 8 in water is known to form a gel by introducing sonication during the gelation procedure. 17 However, here it is classifed as a precipitate because the gelation procedure, on which the model is based, only uses heating and subsequent cooling as a trigger. In principle, for every different gelation procedure a new predictive model should be built requiring a library of data points obtained following exactly the same protocol for gelation performance. Prediction of material properties. The material properties largely determine the usefulness of a supramolecular gel in certain applications. For example, in drug delivery and 3D bioprinting where the material needs to retain or recover their gel state upon injection. It would be interesting to see if descriptors, similar as proposed in this work, could be used to make a prediction regarding relevant properties such as the yield strain, storage modulus or loss modulus. To achieve this, a uniform dataset containing these properties of several supramolecular gels needs to be gathered, which will be the scope of future works. ## Conclusion Predicting supramolecular gelation on the basis of computations is regarded as a challenging task. In this study, four molecular dynamic based descriptors with physical relevance to supramolecular gelation are introduced: the relative solvent accessible surface area (rSASA) to evaluate aggregation, the relative end-to-end distance (rH) describing the flexibility and conformational preferences of the molecules, the hydrogen bonding percentage (HB%) to quantify the non-covalent linkage of the gelator molecules, and the shape factor (F) which is a measure for the aggregate's shape. Via two separate machine learning techniques, it was demonstrated that these descriptors can accurately differentiate the gelation response of a set of urea-based gelators as a precipitate, a gel or a fully solubilised sample. To the best of our knowledge, this is the frst computational method that addresses the prediction of urea-based supramolecular gelation in both organic solvents as well as in water. We hope that the proposed descriptors can be conceptualized for other types of gelators and will steer the feld to discover potential new low molecular weight gelators in the near future. ## Conflicts of interest There are no conflicts to declare.

chemsum

{"title": "Molecular dynamics based descriptors for predicting supramolecular gelation", "journal": "Royal Society of Chemistry (RSC)"}

caged_bulky_organic_dyes_in_a_polyaromatic_framework_and_their_spectroscopic_peculiarities

## Abstract: Host-guest structures and properties have been widely studied using relatively small dyes (<1 nm) without bulky groups, due to their smooth incorporation, efficient host-guest interactions, and high analytical accessibility. In this report, on the other hand, three types of sterically demanding organic dyes trapped by a polyaromatic cage were investigated by spectroscopic analyses on the basis of supramolecular interactions. Coumarins with two bulky substituents are bound by the cage in aqueous solution. The resultant caged dyes show unusual emission enhancement, depending on the difference of a single heteroatom in their substituents. The color of perylene bisimides with two bulky substituents is remarkably changed from yellow to red upon caging. This peculiarity stems from the twist of the substituents in the cage, revealed by the combination of absorption and theoretical studies. Furthermore, tetrasubstituted, bulky porphyrins are caught by the cage in aqueous solution. The caged bulky dyes also display altered color and absorption properties, which remain intact even under acidic conditions. In contrast to typical covalent functionalization and previous host-guest studies toward small and non-bulky dyes, the unusual, non-covalent spectroscopic modulation of the large and bulky dyes can be accomplished for the first time by the present cage, featuring a prolate polyaromatic framework with four openings. †Electronic Supplementary Information (ESI) available. See ## Introduction Organic dyes displaying characteristic colors are ubiquitous in our daily lives. 1 The majority of these dyes possess relatively bulky substituents and rigid core frameworks bearing extended conjugation systems. 2 Modulation and alteration of their colors and spectroscopic properties have been widely demonstrated by covalent functionalization of the frameworks as a common synthetic approach. Non-covalent functionalization of dyes upon encapsulation by molecular cages is recently regarded as an alternative approach, 3 without multistep and time-consuming synthetic protocols. Coordination-driven self-assembly is one of the most promising methods for the preparation of the desired cages, from the viewpoints of high designability and synthetic accessibility. 4 However, coordination capsules and cages capable of binding such large and bulky dyes have been rarely reported so far (Figure 1a, left), owing to size/shape mismatch as well as poor host-guest interactions in the cavity, 5 whereas the embedding of various dyes (e.g., perylene bisimide and porphyrin) into the coordinative ligands is a wellestablished approach. We considered that a coordination-driven molecular cage featuring both a large cavity and several mediumsized windows would be a useful molecular tool to bind a variety of organic dyes with two and more bulky substituents (Figure 1a, right). Here we report that, as a proof-of-concept, the open large cavity of M 2 L 4 polyaromatic cage 1 (Figure 1b,c) can bind sterically demanding organic dyes (1.5 nm in maximum length) with (i) coumarin, (ii) perylene bisimide, and (iii) porphyrin cores in aqueous solution. Notably, the spectroscopic properties of the caged dyes are facilely Host functions of the polyaromatic cavities of coordination capsules have been highlighted recently. 6,7 For example, the closed polyaromatic cavity of M 2 L 4 capsule 2 (Figure 1b) shows efficient/selective binding abilities toward various synthetic compounds 8 as well as biomolecules in aqueous solution, through efficient hydrophobic effect and multipoint -/CH-/hydrogen-bonding interactions. 6,9 However, like other coordination cages reported previously, 4 capsule 2 exhibits no binding ability toward synthetic dyes with multiple bulky substituents, whose sizes are larger than that of the cavity (d = 1.2 nm), owing to the closed spherical framework (Figure 1d). We thus herein employed the analogous, prolate M 2 L 4 cage 1', with a cavity size of d max. = 1.6 nm and a window size of d max. = 0.7 nm (Figure 1c). 10 As the key molecular design in this study, the replacement of the short methoxyethoxy (MOE) side-chains with long hydrophilic ones (i.e., methoxytriethylene glycol (TEG)) allows cage 1 to be used in aqueous solution as well as to bind bulky, hydrophobic organic dyes efficiently for the first time. Crystallographic insights into the host-guest structures are, on the other hand, inaccessible in this system, due to the flexibility of the side-chains. It should be noted that, in contrast to small organic dyes, relatively large organic dyes with bulky substituents and rigid core-frameworks, studied in the present work, have several disadvantages regarding their host-guest investigations. Their low or no solubility in common solvents reduces analytical accessibility, e.g., for typical Job's plot and binding constant analyses. The product yield and stability are often compromised by inefficient host-guest interactions. Accordingly, the investigations of new host-guest structures and properties using large and bulky dyes have been severely limited so far, especially in aqueous media. ## Synthesis of aqueous cage 1 Bispyridine ligand 3 bearing two TEG groups was synthesized from brominated, bent anthracene dimers with two MOE groups in four steps (Figure S1-10). 11 TEG-substituted cage 1 was facilely formed from Pt(II) ions and ligands 3 in DMSO-d 6 (88% isolated yield), as confirmed by 1 H NMR and ESI-TOF MS analyses (Figure 2a and S11-15). 11 The combination of DOSY, DLS, and molecular modeling studies suggested the core and outer diameters of 1 being ~2 and ~4 nm, respectively (Figure 2c). 12 The multiple, long and flexible side-chains suppress its self-aggregation in solution and ordered packing in the solid state. Whereas MOE-substituted cage 1' is insoluble in water, present cage 1 is slightly soluble in water (~0.04 mM; Figure S17) and well-soluble in 15:1 water/acetonitrile solution at room temperature. A Pd(II) analogue, obtained in the same way, showed lower water-solubility relative to 1. The fullyassignable, sharp proton signals of 1 in CD 3 CN (Figure 2a) were characteristically broadened in D 2 O/CD 3 CN (Figure 2b) in the 1 H NMR spectra. The significant broadening of the NMR signals of 1 in the aqueous solution is caused by the restricted motion of the multiple polyaromatic panels through efficient intramolecular, panel-panel interactions. Such NMR properties prompted us to investigate the host-guest structures mainly through UV-visible and fluorescence analyses. ## Caged bulky coumarin dyes As the first result, the open polyaromatic cavity of cage 1 could bind one molecule of disubstituted coumarin dyes 4 (X = S, O, and NH; Figure 3a, left), which are hardly encapsulated by previous capsule 2 even under various conditions. The resultant caged dyes displayed unusual substituent-dependent emission enhancement, whereas these dyes virtually provide an isostructure and low to moderate emission abilities in aqueous solutions without 1. When cage 1 (0.14 μmol) and benzothiazolyl-based coumarin dye 4a (X = S; 0.41 μmol) were stirred in a 15:1 water/acetonitrile (AN) solution (0.45 mL) at 80 ºC for 2 h, a clear pale yellow solution of 1:1 host-guest complex 1•4a was obtained in 50% yield after filtration (Figure 3a). 11 Rigid and hydrophobic dye 4a itself, with a maximum length of 1.5 nm, is insoluble in the aqueous solution (15:1 water/AN) yet soluble in a 2:1 water/AN solution (up to 8 μM). Whereas the 1 H NMR spectrum of 1•4a was broadened like that of empty 1 (Figure S26 and 2b), the ESI-TOF MS and UV-visible analyses clearly indicated the 1:1 host-guest complex formation (Figure S26 and S19) and the binding of 4a in the hydrophobic cavity of 1, respectively. New absorption bands derived from caged 4a were found in the range of 430 to 530 nm, which are slightly red-shifted ( = +9 nm) relative to the bands of uncaged 4a in 2:1 water/AN (Figure 3b). The emission bands of 4a were also redshifted upon binding ( max = 519 nm,  = +6 nm; Figure 3c). These rather small shifts indicated the absence of host-guest -stacking and charge-transfer (CT) interactions in the polyaromatic cavity. The hostguest structure of 1•4a is stable under ambient dilution conditions (~35 μM; Figure S19), implying a moderate binding constant (~2.5  10 4 M -1 ). 11 Please do not adjust margins ## Please do not adjust margins Isostructural coumarin dyes such as benzoxazolyl derivative 4b (X = O) and benzimidazolyl derivative 4c (X = NH) were also bound by 1 under the same conditions (Figure 3a). New absorption and emission bands of the caged coumarins were observed in their spectra (Figure 3b,c and S19-21), 11 in a manner similar to the spectroscopic properties of 4a. Again, slight band shifts ( = +3-17 nm) for 1•4b and 1•4c suggested the absence of host-guest - and CT interactions. The solution of caged dye 1•4a emitted strong green fluorescence with high quantum yield ( F = 62%) in water/AN solution, upon light irradiation ( ex = 450 nm; Figure 3d), whereas cage 1 itself shows no fluorescence despite possessing multiple anthracene fluorophores. Notably, the emission quantum yield of 4a ( F = 52%) was enhanced by 1.2-fold within 1, in contrast to the majority of previous coordination hosts, which quench guest fluorescence to a large degree, owing to the heavy metal effect and host-guest -stacking/CT interactions. 4,6 The emission properties of 1•4a were maintained for at least 7 d at ambient temperature in the dark (Figure S19). The green emission of coumarin dyes 4b and 4c was further enhanced by 2.1-fold ( F = 20%  42%) yet slightly suppressed ( F = 54%  38%), respectively, within the cage (Figure 3d). 11 The CIE diagram displayed a minor change of the emission colors of the dyes within the cage, due to weak host-guest steric and electronic interactions (Figure 3e). The fluorescence lifetime () of 1•4a was estimated to be 5.9 ns, which is longer than that of free 4a ( = 2.8 ns; Figure 3f), indicating the stabilization of the excited state upon caging. Further prolongation of the guest emission lifetime was observed for 1•4b ( = 7.1 ns) under the same conditions. The opposite emission behavior of 4c is likely derived from a weakening of the intramolecular hydrogen-bonding interaction between the C=O and N-H (Figure S27). The optimized structure of caged dye 1•4a indicated that the rodshaped dye is threaded through the opposite two windows of the host framework (Figure 3a, right). Importantly, no aromatic-aromatic stack between the host and guest frameworks is observed, consistent with small band shifts in the UV-visible and fluorescence spectra and the high quantum yield. The coumarin core can be effectively isolated from aqueous solvent, acting as an emission-quenching medium, by the polyaromatic shell. The bulky aromatic substituent of 4a is situated at the window with close host-guest contact, which most probably alters the guest emission in the cavity, depending on the difference in the heteroatom (i.e., S, O, and NH). ## Caged bulky perylene bisimide dyes Next, perylene bisimide (PBI) dyes 5 with two substituted aromatic rings were employed as longer and more rigid organic dyes (2.2 nm in maximum length) than 4. Although the binding of these bulky dyes by synthetic hosts has been quite uncommon, 13,14 one molecule of 5 was selectively trapped by the open framework of 1 in a threading fashion like 4, accompanying a unique change in color. The sonication (40 kHz, 20 min) of a mixture of cage 1 and 2,6-diisopropylphenylbased PBI dye 5a (0.14 μmol each) in 5:1 water/AN (1.0 mL) gave rise to a red clear solution of 1:1 host-guest complex 1•5a in 48% yield, through removal of suspended free 5a via simple filtration (Figure 4a). 11 The broad 1 H NMR spectrum of the red solution at room temperature partially became sharp at elevated temperature (e.g., 75 ºC). The host aromatic signals were split complicatedly and the guest signal H A was observed at -0.96 ppm with a large upfield shift (δ = -2.2 ppm; Figure 4b-d), due to the desymmetrization and aromatic shielding effect of the host framework, respectively. The 1:1 hostguest complexation was confirmed by the ESI-TOF MS analysis, Please do not adjust margins Please do not adjust margins which showed molecular ion peaks corresponding to the [1•5an•NO 3 -] n+ (n = 4 and 3) species (Figure 4e). Owing to the very broad DOSY spectrum of 1•5a at room temperature, the product size was determined by the DLS analysis (d = 4.3 nm; Figure S32c). 11 In the same way, caged dyes 1•5b and 1•5c were obtained by the treatment of cage 1 with 2,5-di(tert-butyl)phenyl-based PBI 5b and 3,5-dimethylphenyl-based PBI 5c, respectively (Figure S28). 11 Although no host-guest titration studies were performed due to the insolubility of 5a-c in 5:1 water/AN without 1, the concentrationdependent UV-visible analysis of 1•5a revealed the stability of the host-complex at ~3.5 μM (Figure S28), suggesting a binding constant of >10 8 M -1 . 11 Again, no encapsulation of bulky PBIs 5a-c took place with capsule 2 even under various conditions, owing to its closed cavity. Drastic color change of diisopropylphenyl derivative 5a from yellow to red through binding by 1 was observed and the present unusual phenomenon was elucidated by theoretical spectroscopic studies. Monomeric free 5a is yellow in less polar solvent (e.g., CH 2 Cl 2 ; Figure and orange in polar solvent (e.g., DMSO; Figure S28) with similarly shaped absorption bands. 11 In sharp contrast, caged dye 1•5a displayed red color in aqueous solution. The visible difference was qualified by UV-visible and fluorescence analyses. The relatively sharp absorption bands of uncaged 5a were recorded in CH 2 Cl 2 in the range of 420 to 550 nm, which were significantly broadened and red-shifted (up to ~610 nm) within 1 (Figure 5a). The emission band of caged 5a was further broadened and shifted ( max = 625 nm,  = +78 nm) relative to free 5a (Figure 5b). A similar yellow-to-red color change was observed for bulky di(tertbutyl)phenyl derivative 5b. The sharp absorption bands (415-550 nm) of 5b in CH 2 Cl 2 were largely broadened and red-shifted (up to ~600 nm) through caging in aqueous solution (Figure 5c). In contrast, neither color change nor spectral change occurred for less bulky and planar dimethylphenyl derivative 5c, featuring broad absorption bands at 430-600 nm (Figure 5c), through binding by 1 under the same conditions. The absorption bands derived from 1 at 330-430 nm remained also unchanged upon caging. These results suggest that no donor-acceptor -stacking interactions exist between the cage cavity and the PBI framework of 5a-c. The theoretical studies indicated that the observed band shifts and broadening stem from the twisted conformation between the bulky substituents (R) and the PBI core of dye 5 within cage 1. 15 The dihedral angle () about the phenyl and imide rings of 5 corresponds to its electronic absorption features. For instance, the most stable conformer 5a with  = 90º theoretically provides an intense, single electronic absorption band at 500 nm (Figure 5d). The band is redshifted by +6 nm through its transformation into one of the metastable conformers with  = 70º. The estimated energy is +75 kJ mol -1 higher than that of the most stable conformer. In the optimized host-guest structure of 1•5a, dye 5a displayed  = 70º within 1 (Figure 4a, right). Accordingly, the observed, unusual color changes of caged 5a and 5b were induced through non-covalent twisting of the bulky substituents in the open cavity. ## Caged bulky porphyrin dyes Finally, to our surprise, the open polyaromatic cavity enabled cage 1 to bind one molecule of porphyrin dyes 6 bearing four or eight substituents. The encapsulation of porphyrins and metalloporphyrins by large proteins has been intensively studied so far 16 yet a nonbiological approach using coordination cages and capsules is complicated by the large and bulky dye frameworks. 5 In a manner similar to 5, tetraphenylporphyrin 6a (R = Ph; 1.8 nm in maximum length) in purple was bound by 1 via sonication (40 kHz) of the mixture in 8:1 water/AN for 20 min (Figure 6a). 11 The resultant brown solution including 1:1 host-guest complex 1•6a (70% yield) exhibited new absorption bands, assignable to caged 6a in the UV-visible spectrum (Figure 6b). The characteristic Soret and the first Q bands of 6a were red-shifted by +3 and +4 nm, respectively, as compared with those of free 6a in CH 2 Cl 2 , through the binding. These bands remained unchanged even after 7 d under ambient conditions, suggesting adequate stability of the host-guest structure (Figure S37). The insolubility of dye 6a in aqueous solution prevented titration, but concentration-dependent UV-visible analysis of 1•6a in 8:1 water/AN indicated high stability (K a >10 8 M -1 ) against high dilution (up to ~3.5 μM; Figure S39). 11 In the 1 H NMR and ESI-TOF MS spectra of the resultant solution, the guest and host-guest signals were scarcely observed owing to significant signal broadening and instability under MS conditions (e.g., extremely low concentration and high vacuum), respectively (Figure S39). 11 The DLS analysis supported the product size with a single peak at d = 4.7 nm, slightly larger than that of empty 1 (d = 3.6 nm, Figure 6c). Unfortunately, no crystals of the host-guest complex were obtained even under various conditions. The structure of obtained, caged dye 1•6a was further supported by the theoretical calculation, where the four bulky substituents of 6a are located at the four openings of 1, whereas the large -conjugated core of 6a is effectively surrounded by the polyaromatic cage-like framework of 1 without direct contacts (Figure 6a Please do not adjust margins Please do not adjust margins caged 6a are most probably derived from the stabilization of its planar conformation over its butterfly-shaped one 16 within 1. Other bulky porphyrin dyes were also trapped by cage 1 in aqueous solution with selectivity 17 and the caged dyes were protected by the polyaromatic frameworks against acid. Further color and absorption changes were found through caging of Zn(II)-tetraphenylporphyrin 6b by 1 through the same sonication protocol. Although 6b itself shows no solvatochromic properties, the resultant solution of 1•6b is bluishpurple in contrast to a pink solution of free 6b in CH 2 Cl 2 (Figure 6d). In addition, distinct shifts of the Soret and first Q bands of 6b by approximately +15 nm in each case were observed in the UV-visible spectrum upon trapping. These unusual shifts are most likely derived from the cage-directed, non-covalent planarization of the metalloporphyrin framework. Caged octaethylporphyrin dye 1•6c and tetrakis(pentafluorophenyl)porphyrin dye 1•6d also displayed modulated color and altered absorption bands, relative to the corresponding free dyes (Figure 6d and S37). Furthermore, the polyaromatic cage-like framework of 1 could prevent the conjugated core of 6a from protonation under acidic conditions. Uncaged 6a was fully protonated upon addition of 20 equivalent of HNO 3 , accompanying its color and spectral changes (Figure 6e). 18 On the other hand, the color and spectrum of caged 6a remained intact, except for a slight intensity change, even after the addition of excess HNO 3 (up to 100 equivalent; Figure 6f and S41). 11 The present protection effect for the porphyrin dye, which was also found for 1•6c (Figure S41), has not been reported with the previous synthetic hosts, except for a covalent cage from the Stoddart group. 5,19 The theoretical studies indicated that the host-guest complexation of 1 and diprotonated dye 6a' is unfavorable (E > 800 kJ mol -1 based on PM6 calculation; Table S2), owing to its sterically demanding, largely distorted framework, rather than long-range, host-guest cation-cation repulsion (d = 0.7 nm; Figure S43). 11 ## Conclusions We have demonstrated the binding and spectroscopic modulation of three types of bulky organic dyes (1.5 nm in maximum length) by a polyaromatic cage in aqueous solution. Binding dyes with several bulky substitutes has faced many difficulties so far, because neither too large nor too small cavities/windows are suitable for the desired host compounds. We herein employed an M 2 L 4 cage with an open polyaromatic cavity, after simple modification of its hydrophilic sidechains, to successfully bind sterically demanding dyes with coumarin, perylene bisimide, and porphyrin cores. Thanks to size and shape host-guest matching as well as effective host-guest interactions (i.e., hydrophobic effect and multiple CH- interactions), these dyes were straightforwardly caught by the cage in a threading fashion. Notably, the caged dyes displayed unusual fluorescence/color and spectroscopic features due to the polyaromatic cage effect, without stacking interactions, which was revealed by detailed theoretical and spectroscopic studies. As compared with typical covalent functionalization and previous host-guest systems targeting small and less bulky dyes, the present non-covalent modification of large and bulky organic dyes provides various advantages, since the coordination framework can for example be easily modulated via doping with heteroatoms and attachment of interactive side chains. Biological and materials applications of the present and other caged bulky dyes will be our next research targets.

chemsum

{"title": "Caged Bulky Organic Dyes in a Polyaromatic Framework and Their Spectroscopic Peculiarities", "journal": "Royal Society of Chemistry (RSC)"}

anomalous_structure_transition_in_undercooled_melt_regulates_polymorphic_selection_in_barium_titanat

## Abstract: The crystallization processes of titanates are central to the fabrication of optical and electrical crystals and glasses, but their rich polymorphism is not fully understood. Here, we show when and how polymorphic selection occurs during the crystallization of barium titanate (BaTiO 3 , BT) using in situ high energy synchrotron X-ray diffraction and ab initio molecular dynamic simulation. An anomalous structure transition is found in molten BT during cooling across the cubic-hexagonal transition temperature, which enables nucleation selection of BT by manipulating the undercooling: a cubic phase is preferred if nucleation is triggered at large undercooling, whereas a hexagonal phase is promoted at small undercooling. We further reveal that the nucleation selection between the cubic and the hexagonal phase is regulated by the intrinsic structure property of the melt, in particular, the degree of polymerization between Ti-O polyhedra. These findings provide an innovative perspective to link the polymorphic crystallization to the non-isomorphic structure transition of the melt beyond the conventional cognition of structural heredity. T he crystallization process of titanates from their melt is critical to fabrication of many functional materials such as optical/electrical crystals , sprayed coatings 4,5 , ferrous metallurgy 6,7 , and high-refractive optical glass 8,9 . There is a universal and interesting phenomenon that titanates exhibit polymorphic phase selection when they nucleate from undercooled melt. For instance, anatase and rutile phase selection occurs during synthesis of TiO 2 nanoparticles 10 ; hexagonal and cubic phase selection exists in the quenching process of BaTiO 3 (BT) melt 11,12 . To obtain a desired phase, a certain condition at which precipitation of one phase is preferential to other possible polymorphs is required 13 . However, searching for such appropriate condition can be difficult as the polymorphic selection mechanism of titanates is still unclear and therefore requires an in-depth investigation. As polymorphic selection is widely observed in a large variety of materials , lot of efforts have been devoted to understanding the underlying mechanism(s). Early studies attempt to understand the polymorphic selection from the perspective of free energy following the Ostwald's step rule 19 , and propose that the difference in entropy/enthalpy of fusion between isomers being the reason for polymorphic selection 20,21 . However, such interpretation from the energy viewpoint encounters difficulty in explaining the complicated polymorphic selection behaviors in atomic scale such as the preferential formation of metastable bcc crystallites in the Lennard-Jones system and random stacking or cross-nucleation of hcp/fcc in the hard-sphere system 22,23 . For complex systems (e.g., metals, polymers, oxides, or solutions), understanding the polymorphic selection mechanism is more challenging as it may be influenced by many factors. Taking aluminum as an example, its nucleation is proceeded with random packing of fcc and hcp crystallites in contrast to the prediction that formation of bcc-like phase is preferred. Such discrepancy is discussed from the strong cohesive interaction in Al 24 . Very recently, An et al. proposed that the stability of isomers also encodes the crystallization pathway 17 . Furthermore, nucleation in complex systems is usually proceeded as multi-step scenarios 16,25,26 , which makes the polymorphic selection process a particularly challenging issue for investigation. To clarify the polymorphic selection mechanism, the following two critical questions should be addressed: (i) When does polymorphic selection take place? From the viewpoint of stepwise energy 19,27 , it should occur during the nucleation stage, but computational simulations on controlling polymorphism 15,17 suggest that it may occur during the growth stage. (ii) What is the atomic/molecular-scale structural explanation for polymorphic selection? Recent studies by advanced TEM technique reveal that the crystallization process is not a single channel but a crossover of multiple intermediate states 28,29 , which raises the question whether the structural characteristics of these intermediate states can be the origin for polymorphic selection. Addressing the above questions relies on in situ tracking of the structural evolution of the melt during the nucleation/crystallization process. For titanates, this can be particularly difficult due to high temperature and highly corrosive melt. Fortunately, the advances of containerless processing and its combination with high-energy X-ray diffraction (HEXRD) in synchrotron radiation facilities provide a useful approach to probe the structure of melt . Such technique has been successfully applied to many molten oxides to probe their structural features and to investigate the solidification process . Results from previous studies have illuminated that the structural evolution of melt has a dramatic impact on the subsequent nucleation behavior. Mechanistic research on the correlation between melt structure and nucleation follows two major mainstreams: (1) Cross-link among structural evolution, macroscopic properties of melt and nucleation. It is proposed that change of the locally order structure of cation-oxygen polyhedra with temperature or composition projects into the variation of melt density or viscosity, and consequently manipulates the nucleation process . (2) Heredity of structural features from melt to crystalline phases. It is demonstrated that the topologically order structures (cations-oxygen polyhedra or the chains/rings connected by these polyhedra) act as prototypes of crystal nucleus . In view of these preceding work, it is convincing that elaborative analyses on the structural features of cation-oxygen polyhedra and their variation with temperature and composition will contribute to unveiling the mechanism of polymorphic selection during the crystallization process of titanates from atomic/molecular scale. BT is a very important functional material from the titanate family. It has several polymorphs and undergoes a sequence of phase transitions during cooling: hexagonal (>~1698 K) → cubic (1698−403 K) → tetragonal (403−273 K) → orthorhombic and rhombohedral (<273 K) 40 . Polymorphic selection of BT has been found during the crystallization process from its undercooled melt and amorphous film, which plays a critical role in preparation of ferroelectric materials, polarized quasi-amorphous films, and colossal permittivity materials 2,41,42 . Here we focus on identifying when and how polymorphic selection takes place during the crystallization process of undercooled BT melt using aerodynamic levitation (ADL) facilities combined with in situ time-resolved HEXRD. We track the structure evolution of BT melt from above the liquidus to supercooled state and reveal an anomalous structure transition in molten BT during freezing across the cubic-hexagonal transition temperature (T c→h ), which accounts for the multi-path crystallization behavior of BT and enables nucleation selection by manipulating the undercooling. The coupling relation between structural transition in melt and nucleation path selection revealed in this work provides a new sight on the elusive crystallization behavior of titanates beyond the traditional cognition of structural heredity, and can be used as a potential strategy to prepare functional oxides with desired phase to meet the property requirement from different applications. ## Results Polymorphic nucleation selection manipulated by undercooling. Undercooled BT melt was triggered nucleation at different undercoolings based on ADL facilities. Room-temperature XRD patterns of solidified BT suggest that polymorphic selection occurs during the crystallization process. As shown in Fig. 1, BT crystallizes to pure tetragonal (t) phase (t-BT, room temperature configuration of cubic phase, c-BT) if triggered nucleation at large undercoolings (ΔT ≥ 316.1 K), but to a mixture of hexagonal (h-BT) and tetragonal phases at small undercoolings (ΔT ≤ 177.9 K). The peaks for h-BT decrease intensity with increasing undercooling and disappear at the undercooling interval between 177.9 and 316.1 K, which is across the transition undercooling ΔT c!h (defined as ΔT c!h ¼ T m T c!h , where T m is the melting point and T c!h is the cubic-hexagonal phase transition temperature, 193 K). Cross-sectional scanning electron microscopy (SEM) images combined with Raman spectroscopy mapping reveal that h-BT and c-BT show a regional distribution in the sample with hybrid phases composition (triggered nucleation at ΔT = 71 K), as presented in Supplementary Note 1 and Figs. S1 and S2. h-BT forms near the top, and c-BT generates mainly around the chill contact region, suggesting that nucleation of h-BT and c-BT requires notably different thermal conditions. Detailed descriptions of the microstructural features at three selected areas (top, middle, and bottom) of the sample are given in Supplementary Note 1. Combining microstructural analysis and XRD patterns, it is plausibly elucidated that the solidified phases selection in BT crystallization is regulated by a polymorphic nucleation sequence manipulated by undercoolings based on the following considerations: (i) h-BT and c-BT nucleate from the undercooled melt directly rather than from a solid-state phase transition due to the clear trajectory of crystal growth from melt as presented in Fig. S1; (ii) formation of h-BT is not cross-nucleation based on c-BT crystal nucleus, although this heterogeneous nucleation between cubic and hexagonal phases is common in several condensed matter systems . Under the heterogeneous nucleation condition, the amount of h-BT should increase with increasing undercooling, which conflicts with the XRD results that the peak intensity of h-BT decreases with increasing undercooling (Fig. 1); (iii) precipitation of c-BT as prominent phase in supercooled liquid is not due to a higher crystal growth rate, as the crystal growth of both h-BT and c-BT presents typical facet growth mode (Fig. S3); in addition, the dynamic factors usually affect the selection of phases with different chemical compositions 43,44 . Furthermore, we track the structural evolution of the BT melt during its crystallization process by in situ time-resolved HEXRD. X-ray photon beam is incident on the top-side of the sample and high-angle diffraction signals are collected, as presented in Supplementary Note 2 and Figs. S4-S6. For the supercooled liquid (ΔT = 644 K), its temperature-time profile (Fig. 2a) shows a significant recalescence associated with crystallization 45 . Diffraction patterns during the crystallization process of the supercooled liquid obtained from integrating the twodimensional diffraction patterns (Fig. S5) are presented in Fig. 2b. It is clear that the primary phase solidified from the supercooled liquid is c-BT (t3), which grows up rapidly without transition to h-BT as confirmed by the unvaried diffraction patterns from t4 to t10. For the crystallization at a small undercooling (ΔT = 71 K), the temperature-time profile and the integrated diffraction patterns are shown in Fig. 2c and d, respectively (see Fig. S6 for two-dimensional diffraction patterns). It is intriguing to notice that the primary phase shows up at t3 is c-BT, suggesting that c-BT crystallizes first near the triggering point, which is consistent with the microstructural analysis (Fig. S1 and S2). Subsequently, the characteristic peaks of h-BT at 26.28°and 41.22°emerge (t4), indicating that h-BT has nucleated from small undercooling melt at the upper side of the droplet. The result also provides evidence that h-BT is formed from the melt directly rather than from a solid-phase transition from c-BT due to the sluggish kinetics for c to h phase transition which usually takes hours 46 . It is noticed that the diffraction peaks are slightly weakened at t5, suggesting the occurrence of re-melting due to release of latent heat. Subsequently (t6 to t12), the characteristic peaks of h-BT are intensified and a fine peak at ~49.10°appears, indicating further growth of h-BT. In situ XRD patterns illustrate how BT melt with a small undercooling crystallizes to a mixture of h and c phases. Combining the information from microstructural observations and in situ time-resolved HEXRD results, it can be concluded that the polymorphic crystallization of BT stems from the nucleation selection manipulated by undercooling. Structural evolution of BT melt. To explore the structural origin of this nucleation path selection manipulated by undercooling, the melt-BT structure from superheating to supercooling state was tracked by HEXRD. The Faber-Ziman structure factor, S(Q), of melt-BT at various temperatures is presented in Fig. 3a. S(Q) of molten BT shows a principal sharp peak at ~1.9 −1 , which reflects the structural information of cation-cation periodicity in the real space (~3.3 ). The first peak position, Q 1 , is commonly used to probe the structural evolution or dynamic transition . Here we extracted the Q 1 value and its associated uncertainty by direct reading or Lorentzian curve fitting (inset figure in Fig. 3b). It is intriguing to note that Q 1 shows a discontinuous variation with decreasing temperature that an unanticipated drop is observed (see Fig. 3b) between the phase transition temperature (T c→h = 1705 K) and the melting point (T m = 1898 K). Such anomalous transition in supercooled BT melt is rarely observed in analogous titanate systems, such as BaTi 2 O 5 (BT2) 39,47 and TiO 2 melt (our data), where the Q 1 increases monotonically with decreasing temperature, as also shown in Fig. 3b for comparison. This Q 1 anomalous transition is strongly correlated with the variation of rheological properties, which may encode the freezing fate of BT melt. According to the Ehrenfest relation Q 1 = 1.23 (2 × π/a), Q 1 is proportional to the reciprocal of the mean atomic spacing 'a', which is directly related to the free volume in liquid. Based on Doolittle's theory 50 , the drop of Q 1 suggests a decrease in the viscosity (η) of the melt-BT. Nucleation in cooling BT melt will be therefore greatly promoted due to an enhanced nucleation driving force and a low viscosity, which may be the reason why BT is difficult to vitrify 41 . The first principal peak embodies intermediate range order (IRO) cation-cation periodicity, which is constructed by short range order (SRO) cation-oxygen polyhedra. The anomalous structural transition of Q 1 prompts us to investigate the structural evolution features of the second diffraction peak that reflects SRO polyhedra in real space (estimated by periodicity 2π/Q 2 ). The second diffraction peak, as shown by the expanded view in Fig. 3c, consists of two shoulder peaks, the approximate positions of which are indicated by the short-dash arrows. We extracted and plotted the intensity of the two shoulder peaks as a function of temperature, as shown in the inset figure in Fig. 3c. It is demonstrated that the intensity of both peaks increases with decreasing temperature but the enhancement of two shoulder peaks are not consistent. There is a relatively larger intensity difference between two shoulders in high-temperature BT melt than that in deep-undercooled melt. Near the structural transition temperature (1783 K) of Q 1 , the intensity discrepancy between two shoulder peaks has taken a turning, which strongly indicates that the structure of fundamental motifs (cation-oxygen polyhedra) in BT melt have also changed in cooling process. 4b. It can be seen that the Ti-O bond length first decreases with decreasing temperature, but this trend reverses in the temperature range across T c→h that it increases from 1.83 (1843 K) to 1.85 (1563 K). Such discontinuous variation of bond length is rarely observed in molten oxides including titanates such as molten TiO 2 (ref. 55 ) and BT2 (refs. 39,47 ), in which the bond length of cation-oxygen pairs varies monotonically with decreasing temperature. Furthermore, when liquid BT is undercooled into cubic region (T < T c→h ), the Ti-O bond length is significantly larger than that extrapolated from the high-temperature data. Based on the bond valence theory 56 , a large Ti-O bond length indicates that Ti cations are bonded with more anions (O 2− ), and thus an enhanced connectivity of Ti-O polyhedra in the BT melt. The coordination number, C n Ti ðOÞ, of the Ti-O pair is calculated from T n (r) using the same method reported by Hennet 57 . where r 1 and r 2 are the minimum and maximum cutoff of the Ti-O pair, which are set as 1.50 and 2.75 , respectvely; c o is the concentration of oxygen atoms and W Ti;O is the weighting factor of Ti-O pairs (see details in Supplementary Note 4 and Fig. S9). The calculated C n Ti ðOÞ is plotted as a function of temperature, as shown in Fig. 4c. It is observed that the Ti-O coordination number initially decreases with decreasing temperature, but this temperature dependence reverses in the temperature range between 1843 and 1563 K (across T c→h ). Such a transition is extraordinary in comparison with the continuous variation of coordination number in TiO 2 melt 55 and BT2 melt 39,47 . This discontinuous structural evolution of Ti-O motifs is rare and poorly understood in previous structural studies of molten titanates. In order to further unveil the detailed scenario of the structural changes, the pair distribution functions (PDFs) within the temperature range of structural transition and the corresponding difference functions are extracted and compared in Fig. 4d. It is found that the Ti-O peak shifts to higher r during transition, being concomitant with an increased intensity and FWHM, which indicates an increase in the percentage of Ti-O polyhedra with high oxygen coordination number. This conclusion is also consistent with the C n Ti ðOÞ. Estimating from the electrostatic bond strength theory, the increased oxygen 4a). Additionally, it is intriguing to note that the correlation length of Ti-Ti shrinks with the structural transition proceeding, as shown by the difference functions Fig. 4d, which suggests that the connection mode between some Ti-O polyhedra has perhaps changed from long connection (corner sharing) to short connection (edge sharing or even face sharing). In view of the Ti-O polyhedra transition behavior in Fig. 4b and c, it is elucidated that the structural transition occurring in undercooled BT melt may be driven by temperature, namely, gradual transformation through incremental metastable states within a certain temperature range. This is analogous with several first-order phase transition behaviors (e.g. high-/low-density amorphous transition in glassy water 35 ). Given the controversial and complicated nature of liquid-liquid structural (phase) transition, we perform ab initio molecular dynamics (AIMD) simulations to give further evidence to support the experimental observations. The comparison of structural correlation functions determined by AIMD simulation and measured by experiment, at three typical temperatures (superheating, near melting point, and supercooling melt), is shown in Fig. 5. It is intuitive that the total S(Q) functions obtained by AIMD reflect all the characteristic peaks in corresponding with experimental patterns (see Fig. 5a); however, strong finite size effect of AIMD boxes may lead to the deviation of S(Q) patterns in low-Q side and the first principal peak. The comparison of T(r) functions in real space is presented in Fig. 5b. The first Ti-O peak shows a good accordance between simulation and experiment, but the second peak and the third peak in simulation patterns are relatively lower than that of experimental patterns, which may correspond to a lower FSDP in simulated S(Q) patterns, because this peak is usually regarded to be related to IRO cation-cation periodicity (the third peak in T(r) patterns). The contribution from different atomic pairs on total T(r) functions was also presented by Fig. 5b, it is clear that the first peak in T(r) pattern is dominant by Ti-O pair, and the second peak is mainly contributed by Ba-O pair. The overlap between Ti-O peak and Ba-O peak is not serious, benefiting from a large difference of ionic radius between Ti 4+ and Ba 2+ , which indicates the peak-splitting and fitting in Fig. 4b and c is reasonable. Structural statistics given by AIMD are presented in Fig. 6. In the rapid quenching of BT melt, it is illustrated that both the bond length and the coordination number of Ti-O pair show a discontinuous transition in a narrow temperature range (1523-1583 K), as presented in Fig. 6a. The temperature dependence of Ti-O polyhedra structural parameters calculated by AIMD is consistent with our diffraction experimental results (Fig. 4b and c) apart from a difference in the transition temperature. One possible reason we considered is that this anomalous structural transition is strongly dynamic-correlated, and it will be influenced by thermal history just like most phase transitions in materials. The typical partial radial distribution functions (RDFs) of Ti-O pair are presented in Fig. 6b, the intensity of Ti-O RDF patterns in high-r side increases after structural transition (inset graph in Fig. 6b), which corresponds to a larger coordination number and is consistent with our experimental results (Fig. 4d). Intriguingly, the Ti-O RDF patterns shown asymmetric feature, which can be fitted by asymmetric Extreme function appropriately, but is somewhat against to before Gaussian peak fitting (see Fig. 4b). The speculation that enhanced connection between Ti-O polyhedra and the change of connection modes are also confirmed, as presented in Fig. 6c and d. After structural transition, the folds of Ti-Ti clusters are notably increased, indicating the polymerization between Ti-O polyhedra is strengthened. Additionally, after transition, the percentage of corner sharing mode decreases, in contrast, the edge sharing mode increases, which accounts for the contracted Ti-Ti correlation length in Fig. 4d. The change of connection between Ti-O polyhedra also corresponds to an essential fluctuation of average energy per atom (E 0 ), as shown in Fig. 6d. Due to the percentage of unstable edge sharing increases after structural transition, a small positive deviation of E 0 is observed, which also provide firm evidence of structural transition in BT supercooled liquid from perspective of energy. Given the asymmetric feature of Ti-O RDF peaks illustrated by AIMD, we compared the difference of Ti-O structural parameters obtained by Gaussian or Extreme curves fitting, as shown in Fig. 7. In addition, the size of structural transition determined by AIMD simulation and diffraction experiment were also compared in this figure. The bond length and coordination number of Ti-O pair obtained by different peak fitting method are quite accordant 47 ), and TiO 2 (ref. 55 )) melt with temperature. The error bar corresponds to the mean absolute error of that calculated from the real Ti-O bond length (read directly from T(r) pattern) and that obtained from Gaussian fitting. The Ti-O bond length of c-BT and h-BT calculated by using weighted average are also presented for comparison. The inset graph gives the Ti-O bond length distribution in c-BT and h-BT unit cell, respectively. c The coordination number of Ti-O pair in BT and other titanate melts (BT2 (ref. 47 ), TiO 2 (ref. 55 )) at different temperatures. The error bar represents the mean absolute error of two Ti-O coordination number results estimated by integrating the original T(r) curves from 1.4 to the first minima and by integrating the Gaussian fitting curves. Ti-O octahedra motifs (CN = 6) in crystalline phases are also included for demonstration. Open squares represent the measurement by Alderman et al. 77 for BT superheated melt. d The change of pair distribution functions across the structural transition process (upper), and the corresponding difference functions (lower). The vertical dash lines are used to indicate the approximate bond length of each atomic pair. These values are referred from ref. 77 and our AIMD results. with each other, which manifest the structural transition observed in Fig. 4b and c should not be so-called uncertainties introduced by Gaussian peak fitting. Although the structural transition determined by AIMD simulation is smaller than that observed in experiment, the transition size is in the same order magnitude. In addition, the size of structural transition determined by AIMD is clearly exceeding the standard error caused by configurational fluctuation, therefore, a convinced structural transition in supercooled BT liquid is expected from AIMD evidence. Furthermore, we also test the discontinuous transition of Ti-O coordination number determined by AIMD, using a different starting configuration or even a completely random starting configuration (see details in Supplementary Note 5 and Fig. S10). The anomalous structural transition existing in undercooled BT liquid are seemingly independent of starting configuration. However, it should be emphasized that the magnitude of structural transition determined by AIMD may be sensitive to the total energy state of starting configurations, a relatively stable starting configurations with lower total energy may be helpful to investigate this liquid-liquid structural transition. ## Discussion Relationship between polymorphic nucleation selection and structure heredity. The structural tracking reveals an anomalous structure transition of undercooled BT melt during the freezing process, which may promote nucleation. In the following sections, we would like to discuss the question whether the polymorphism of BT melt accounts for the subsequent polymorphic crystallization selection. Both the classical nucleation theory (CNT) and the more widely accepted multistep nucleation scenario suggest that a crystal with structure analogous with the parent (liquid) phase possess a low nucleation barrier 58,59 , which is known as structural heredity. In previous studies, evolution of the SRO cation-oxygen polyhedra in molten oxides is regarded as an essential structural probe to describe liquid-solid transition 38,60 , glass forming 32 , and liquid-liquid phase transition 48,49 because connection of these SRO structural units constructs larger order structure, such as IRO or critical nucleus. Therefore, we first take the SRO cation-oxygen polyhedra into consideration. In titanates, Ti-O polyhedra are regarded as the structural units because Ti-O pair has stronger bonding strength than Ba-O. Thus, the structural parameters of Ti-O polyhedra in melt-BT with those in crystalline c-and h-BT phases are compared. The mean Ti-O bond length in crystalline phases is calculated based on published crystal configurations (c-BT 61 , h-BT 62 ) by extrapolating the available data to high temperature range assuming the same thermal expansition coefficient. Coordination number of the Ti-O pair in c-and h-BT is 6 and remains unvaried with temperature. As shown in Fig. 4b and c, both bond length and coordination number of Ti-O pair in the BT melt are significantly lower than those in the crystalline phases. The striking difference in the structral parameters between BT melt and crystal phases demonstrates that the crystal structure is severely destroyed during the melting process, and suggests that BT liquid is very fragile 32 . It is also noticed that the Ti-O bond length in BT melt is closer to that in h-BT than in c-BT through the freezing process, indicating that the Ti-O polyhedra in BT melt have a higher degree of similarity with those in h-BT than in c-BT, which cannot explain the preferential nucleation of c-BT in supercooled liquid. On the other hand, the bond length of Ba-O pair is also extracted by peak-differentiating and plotted as a function of temperature, as illustrated in Fig. 8. For comparison, the calculated Ba-O bond lengths from AIMD simulations are also presented. Different from Ti-O pair, Ba-O pair does not show abrupt variation across T c→h . In addition, the Ba-O bond length in BT melt is closer to c-BT rather than h-BT. These conflicting information from Ti-O and Ba-O suggests that SRO of cation-oxygen polyhedra is inadequate to describe the polymorphic nucleation selection of BT. On a larger scale, we investigate the tolerance factor of the undercooled BT melt. The tolerance factor, defined as t ¼ d AO = ffiffi ffi 2 p d BO where d represents the bond length of cation-oxygen 63 , describes the structure stability of ABO 3 -type perovskites; t has been successfully used to explain the stabilization of metastable phases and nucleation behavior in RFeO 3 (ref. 64 ) and RMnO 3 (ref. 65 ) (R = rare-earth element) prepared by containerless processing under low oxygen partial pressure. For BT, t = 1 suggests a cubic structure and t = 1.03 suggests a tetragonal or hexagonal phase. The calculated t of the undercooled BT melt shows a fluctuation around 1.03 (inset figure in Fig. 8), which agrees with the AIMD simulation results and suggests that the melt structure has a high similarity with the high-temperature h-BT phase. From the traditional view of structural similarity, BT melt should crystallize to h-BT, which fails to explain the polymorphic nucleation of undercooled BT melt. This drives us to explore more basic generality on the polymorphic nucleation selection of BT. New insight of BT polymorphic nucleation selection based on polyhedra polymerization. Previous studies suggest that bridging The variation of Ba-O mean bond length in BT melt, obtained from peakdifferentiation of T(r) or calculated based on AIMD ensembles, as a function of temperature. The error bars correspond to the standard deviation derived from Gaussian fitting. Open square is the result from ref. 77 . Data for BT2 melt 47 and two crystalline BT phases 61,62 between Ti-O polyhedra plays an important role in the choice of crystallization path. For instance, weakening the bridging by oxygen vacancies 66 is responsible for the retention of metastable h-BT in the sintering process, whereas restricting the arrangement of Ti-O polyhedra (strengthening the bridging) directly regulates the polymorphic selection during the crystallization in amorphous BT film 42 . These facts inspire us to explore the possible interrelationship between bridging of Ti-O polyhedra and polymorphic nucleation selection of BT. Bridging of cation-oxygen polyhedra can be described by a polymerization, which is defined as the ratio of non-bridging oxygen per cation and is approximately delineated by the mean oxygen coordination number around cations in here. Polymerization has been employed to understand the dynamic behavior 67 , glass forming ability 36 , and crystallization 52 of strong molten oxides such as silicates, aluminates, and borates. However, for extremely fragile melt such as zirconates and titanates, polymerization is less attended due to distinct structural units in melt and corresponding crystals 32,34,55 . In the BaO-TiO 2 system associated with this work, polymerization has exhibited a potential influence on polymorphic nucleation: with strengthening Ti-O polyhedra bridging (excess of TiO 2 ), the primary phase of BT changes from h-BT to c-BT 46 . Here we have a quantitative analysis of the possible effect of polymerization on the primary phases crystallized from melt. The evolution profile of Ti-O coordination number with temperature and composition in (1−x)BaO−xTiO 2 (0.5 ≤ x ≤ 0.6667) melt is calculated by interpolation: and illustrated as the color map in Fig. 9 about the BaO-TiO 2 binary phase diagram 46 . Along the liquidus, the Ti-O coordination number increases with increasing TiO 2 content (x), indicating that the polymerization among Ti-O polyhedra units is strengthened. At the critical composition point for h to c transition (x = 0.596), the Ti-O coordination number is about 4.50, which is comparable to the value of cubic region (T<T c!h ) after structural transition (Fig. 4c), suggesting that the variation of polymerization in undercooled BT melt resulted from anomalous structure transition across T c→h is parallel to that induced by increasing TiO 2 content. Therefore, the polymorphic nucleation behavior may be 'self-regulated' by the dynamically polymerized and depolymerized tendency between Ti-O polyhedra. Although the scenario of polymerization change induced by varying TiO 2 content and by manipulating undercooling are different (linear variation with TiO 2 content whereas non-linear variation with undercooling), they both result in an enhancement in polymerization. It is therefore concluded that the strongly ordered [TiO m ] n groups developed in highly polymerized BT melt (largeundercooling or rich-TiO 2 ) may initiate the c-BT nucleation, in contrast, weak bonded units in liquid (small-undercooling or poor-TiO 2 ) will contribute to the preferential crystallization of h-BT. ## Conclusion By in situ HEXRD, we reveal an anomalous structure transition in molten BT during cooling across the cubic-hexagonal transition temperature (T c→h ), which accounts for the multi-path crystallization behavior of BT. Such abnormal transition is rarely observed in other molten oxides and enables nucleation selection of BT by manipulating the undercooling: c-BT is preferred if nucleation is triggered at a large undercooling (T < T c!h ), whereas h-BT precipitation is promoted at a small undercooling (T c!h < T < T m ). We find that the nucleation selection between h-BT and c-BT breaks the traditional cognition of structural heredity. The atomic-scale mechanism behind this polymorphic nucleation selection manipulated by undercooling was investigated by tracking the structural evolution of BT melt from overheating to supercooling state. A discontinuous structure transition mirroring both in reciprocal space correlation function S(Q) and in the structural parameters of cation-oxygen (Ti-O) polyhedra is observed across T c→h during cooling of a BT melt. We unveil that the nucleation selection between c-and h-BT is self-regulated by the intrinsic structure property of the melt, in particular, the degree of polymerization between Ti-O polyhedra: a highly polymerized liquid promotes the cubic phase nucleation, on contrary, the hexagonal phase primarily precipitates from more depolymerized liquid. Our findings shed light on the atomic-scale scenario of liquid-BT structural evolution, and provide a new perspective (polymerization effect) for linking structural transition occurring in molten oxides with alternation of crystallization path, which will spur further investigation in physical mechanism and industrial application. ## Methods Sample preparation and containerless processing. Highly pure BaCO 3 (99.95%, Aladdin, China) and TiO 2 (99.99%, rutile, Aladdin, China) powders were used as starting materials to synthesize BT. Equimolar BaCO 3 and TiO 2 were homogeneously mixed by wet ball milling in ethanol for 36 h. After drying, the resultant mixture was pressed into tablets (diameter 20 mm, thickness 2 mm) at 20 MPa by uniaxial press, and then sintered in a muffle furnace at 1273 K for 12 h. The sintered tablets were crushed to small pieces of ~15 mg and were used for subsequent containerless processing. The containerless melting was conducted by ADL equipped with a 100 W CO 2 laser. Samples were suspended in the furnace by purging highly pure oxygen (99.999%) with controlled flow rate to provide drag force and buoyancy to balance the gravity. A CCD camera was used to monitor the suspension status of sample and an infrared pyrometer with a wavelength of 1.255 μm was used to record the temperature profile. Nucleation was triggered by decreasing oxygen flow, which destroyed the stable levitation of the molten droplet and induced the nucleation chilling effect (contact with the wall of nozzle). Once recalescence was observed on the temperature-time profile, laser was turned off immediately to preserve the primary phase by quenching. Characterization of phase composition and microstructure. The phases of solidified BT were identified by Cu Kα XRD (Rigaku Ultima IV) on finely ground powder. Spacial distribution of each phase and sample morphology were characterized by Raman imaging combined with SEM (Raman-SEM, MAIA3 GMU model 2016) on the polished and chemically etched cross-section. Raman spectrum was obtained by a 532-nm Ar laser with a spot size of ~0.3 μm. In situ time-resolved HEXRD and data processing. The diffraction experiments were conducted in beam line station BL13W1 of Shanghai Synchrotron Radiation Facility (SSRF). ADL facilities were assembled on a two-axis translation stage to ensure precise alignment between the specimen and the synchrotron radiation Xray beam. Schematic of the experimental setup is shown in Fig. 10. The energy of high-energy X-ray was calibrated by a tungsten target as 69.525 KeV (a wavelength of 0.17835 ), which provides a sufficient scattering range Q and reduces the influence of self-absorption and multiple scattering. A PE flat-panel X-ray plate (XRD 1621 AN3 ES) was used to collect the scattering photons during the step cooling process. Data collection time for time-resolved HEXRD experiment (tracking crystallization) was 0.5 s to ensure signal quality. To characterize the structure of molten BT, data collection time was extended to 60 s (4 × 15 s) to ensure temperature stability and to avoid overexposure of the PE plate. The X-ray beam (1.3 mm × 0.6 mm) was partially incident by the top of the specimen, as shown in Fig. S4, to obtain larger count rates at larger 2θ values 68 . Background diffraction was measured on empty chamber without sample. The initial diffraction intensity, extracted by integrating the image plate pattern through software fit2D, was corrected for polarization, absorption, Compton scattering, and fluorescence. After that, the normalized scattering intensity, Is(Q) was used to obtain the total X-ray structural factor S(Q). Subsequently, the distribution functions, such as T(r) and g(r), in real space can be derived from S(Q) through Fourier transition corrected by Lorch function to reduce the truncation effect 69 . Density measurement of molten BT droplet. The melt density of BT was measured by combining ADL with ultraviolet-based imaging technique, as described by Langstaff 70 . A schematic diagram of the facility is shown in Fig. S7. A high-resolution, black and white high-speed camera (Vision Research Inc, VEO 640) was used to acquire the magnified image of the sample, which was illuminated by a UV lamp (LC8, L9588-02) from the opposite direction. To eliminate the thermal radiation noise (mainly visible and infrared light) at high temperatures, which may blur the boundary of the droplet, a high-pass filter (Thorlabs, FESH0450) was mounted in front of the high-speed camera lens. The molten droplet is approximately ellipsoidal, and its geometric size was obtained by fitting the boundary of the backlighted image by an elliptic equation. The actual size of a molten droplet was converted by a scale factor, which was determined through fitting the boundary of the backlighted imaging of a SiC bead with known diameter (; ¼ 3:175 mm). AIMD simulations. AIMD calculations were performed by using the density functional theory (DFT) with Vienna AB-initio Simulation Package (VASP) 71 . Interaction between the ionic core and the valence electrons is treated by the projector augmented wave (PAW) method 72,73 . The exchange-correlation interactions were described by the generalized gradient approximation (GGA) in the Perdew Burke Ernzerhof (PBE) formalism 74,75 . A canonical ensemble (constant atomic number, volume, and temperature, NVT) was employed. The cutoff energy was set as 400 eV. The 2 × 2× 2 grid was used for Brillouin zone integrations. An energy tolerance of 10 −5 eV and a force tolerance of 0.01 eV −1 were adopted to ensure accuracy. The initial liquid model, containing 200 atoms (40 Ba, 40 Ti, and 120 O), was generated by fitting the experimental structural functions (S(Q) and g (r)) based on Reverse Monte Carlo method. This initial starting configuration was relaxed in 5000 K for 30 ps with a timestep of 3 fs, such a high temperature was used to guarantee a non-crystalline configuration. Then, the high-temperature configuration was cooled to the desired liquid temperatures, after that, a simulation time of 15 ps with a timestep of 3 fs was performed to ensure energy stability. At each temperature, the model was equilibrated for another 5 ps with a timestep of 1 fs. The configurations corresponding to the final 5 ps were collected to calculate the partial PDF, the folds of Ti-Ti clusters, the connectivity between Ti-O polyhedra, and other structural parameters. In all the AIMD simulations, temperature was controlled by a Nosé thermostat 76 . The total pressure was obtained from the sum of ideal gas part (ρkBT) and the external pressure was provided by VASP. Under each temperature, the supercell length was tested carefully to ensure that the absolute value of the total pressure is lower than 200 MPa.

chemsum

{"title": "Anomalous structure transition in undercooled melt regulates polymorphic selection in barium titanate crystallization", "journal": "Nature Communications Chemistry"}

digital_chromatography:_separating_amino_acids_into_spatially_discrete_containers

## Abstract: Separation of amino acids (AAs) in a mixture has been conventionally done with chromatography or electrophoresis; separation usually occurs over some continuous space. The present communication proposes a digital method with a 20-stage pipeline for separating (and counting) single AA molecules in a mixture, with each of the 20 proteinogenic AAs ending up in its own discrete and spatially distinct container. Presently the method is designed for samples with a few molecules on up to the atto-mole level and can be used with samples collected from single cells. It is based on the superspecificity property of transfer RNAs (tRNAs): an AA can bind only with a cognate tRNA and not with any other; the binding error rate is about 1 in 350. Four necessary conditions for accurate separation are noted; it is shown informally that they can be satisfied. The method can also be used for peptide sequencing by feeding terminal residues cleaved from a peptide into the first stage of the pipeline. Separation of components of a mixture is currently dominated by two technologies: chromatography (CH) and electrophoresis (EP) . CH is broadly divided into partition chromatography and column chromatography. An example of the former is paper chromatography, the latter has several variants, including gel, ion exchange, affinity, and high performance liquid (HPL) chromatography. The last is the method of choice when separation is followed by a detection step to identify the separated component. Currently the most widely used technique for identification is mass spectrometry (MS) , with HPLC commonly coupled to MS as a precursor. Amino acid (AA) separation has been the subject of extensive research aimed at highly selective collection of the 20 proteinogenic AAs. Detection of specific AAs is often done by labeling. One way to do this is derivatization , which modifies a part of the analyte (such as the amino group in an AA) to render it detectable by some means such as optical. Sorting of single molecules (including the 20 AAs) represents the finest granularity in analyte separation. Nanochannel-based separation methods have been developed for this purpose. In a nanofluidic network is used to separate and sort DNA strands by length. In three classes of AAs are derivatized for detection with a nanopore. With two types of derivatives, 9 of the 20 AAs can be distinguished. In AAs cleaved from a peptide are delivered through a glass capillary into the input vacuum manifold of a mass spectrometer. In CH, EP, and nanopore-based methods, the separation space is usually a continuum. Because of this all these methods are basically analog methods. The work reported here represents a departure from this analog approach; it proposes a digital method for AA separation based on the 'superspecificity' property of tRNAs that can be used with sample sizes in the atto-to zepto-mole level and in the analysis of single cells . At its center is the 'superspecificity' property of transfer RNAs (tRNAs), which is used by living cells to translate mRNA to protein . Thus for every one of the 20 proteinogenic AAs there is a tRNA that is charged with (that is, binds to) that AA and to no other; the in vitro error rate is about 1 in 350 . The following equations summarize the charging process : In Equation 2, AARS catalyzes the binding of AA to tRNA. The separation process can be viewed as a form of column chromatography in which the column is a discrete stack of 20 levels A1, ... , A20, with level i consisting of a separation unit that separates AAi from the mixture at that level into a discrete container. Each unit goes through a sequence of four steps: 1) confinement of reactants (AA, tRNA unique to the level, its cognate amino-acyl tRNA synthetase (AARS), and adenosine triphosphate (ATP)) in a cavity; if cognate the tRNA is charged with AA and adenosine monophosphate (AMP) is released; 2) filtering or separation of tRNA (charged or not) and AARS from small molecules (ATP, AMP, and unattached or excess AAs); 3) deacylation of tRNA, resulting in release of the bound AA if tRNA is charged; and 4) separation of the freed AA at that level into a discrete container for that AA type. Figure 1 shows the column as a pipeline with 20 stages, where a stage corresponds to a level in the column. Each stage consists of four steps, represented diagrammatically in Figure 2. Step 1: Charging of tRNA The reactants are input from a reservoir and restricted to a small space to avoid the dispersive effects of diffusion. Such restricted spaces may take the form of low micro-to nano-volume containers or cavities . No measurements are made on the reservoir-cavity structure so the dimensions do not have to be exact. The dimensions of the reservoir, passage, and cavity in Figure 2a are about the same order as those in . (The reservoir and passage, which are adjoined to the cavity from above, are not shown.) Step 2: Filtration of small molecules (AMP, ATP, non-cognate AAs and excess cognate AAs) After a suitable incubation time in the container (determined experimentally) the products may include charged cognate tRNA, uncharged tRNA, free AA, AARS, ATP, and AMP. The last will be present only if tRNA is cognate and at least one tRNA molecule has been charged. tRNA, charged or not, and AARS are separated from AA, ATP, and AMP (if any) (Figure 2b). The filter sizes required are calculated from the axes of enclosing ellipsoids for each of the analytes involved. Table 1 shows the axis sizes of the covering ellipsoids for each type of reactant. Referring to Table 1, a 3-6 nm diameter pore can separate tRNA and AARS from AA, AMP, and ATP; and a 9-10 nm diameter pore can separate tRNA from AARS (if desired). At the end of this step the small molecules will have filtered through the nanofilter and may be removed if desired (Figure 2b, bottom). 2 1.10 0.7 0.34 AMP 3 1.67 0.95 0.4 ATP 4 2.34 1.09 0.44 Phe-tRNA 5 10.56 6.54 3.64 Ala-AARS 6 13.94 9.41 8.46 1,2,3,4,6 Atomic coordinate data from files.rcsb.org/ligands/view/{GLY, TRP, AMP, ATP}_ideal.sdf, www.rcsb.org/structure/1YFS. (The last is for Ala-AARS.) 5 Atomic coordinate data from . ## Step 3: Addition of NaOH for deacylation The tRNA, if charged, can be deacylated to release the bound AA, which can then be detected electrically, optically, or by other means. Deacylation can be done enzymatically or without an enzyme. Non-enzymatic deacylation is much simpler. It may be done with NaOH or by controlling the solvent's pH level . The products of the deacylation step are either tRNA and the dissociated AA (if tRNA was charged) or only tRNA. After an empirically determined time following filtration of the smaller molecules, NaOH is added to the filter input (Figure 2c, top) for deacylation of the tRNA-AA complex. Step 4: Deacylation of charged tRNA and release of freed AA into discrete container After an empirically determined time during which deacylation can take place, the freed AA filters through the nanofilter into the adjoining container under hydraulic pressure (Figure 2d). In the pipeline of Figure 1, the exiting filtrate at each level consists of smaller molecules (ATP, excess AAi (that is, those that remain unbound to tRNAi), non-matching AAs (that is, AAj≠i), and AMP (the byproduct of tRNA charging, see Equation 2) and passes to the next level. The number of AA molecules separated at any given level i is limited by the number of tRNAi molecules in the i-th stage in the pipeline. Therefore the filtrate from level 20 is recycled to level 1. This cycle is repeated until there are no incoming AAs detected in a container at any level. The end of the separation process is signaled when every one of the 20 counters has stopped counting. The efficacy of this approach can be judged by how well it satisfies four conditions; the following is an informal analysis. Condition 1. The reactants necessary for charging of a tRNA (Step 1) must be in close proximity until successful charging occurs. It is shown in by simulation that confinement at the bottom of the cavity is facilitated and ensured by hydraulic pressure. Condition 2. tRNA molecules are perfectly separated from small molecules (AA, ATP, AMP) during filtration. Table 1 shows that this condition is satisfied by filter pore diameters in the range 3-6 nm, which will allow AMP, ATP, and any of the 20 AAs to filter through but not Phe-tRNA or Phenylalanyl-tRNA synthetase. Condition 3. Deacylation of a charged tRNA always occurs and is complete. In the living cell deacylation is done by an enzyme to correct charging errors . Here non-enzymatic deacylation with NaOH is used; see for a description of its use in the deacylation of cysteine-tRNA. Condition 4. Filtration times must be practical. As noted in Condition 2, filter pores in Figure 2a and 2b have a diameter in the range 3-6 nm. For simplicity consider a cubical cavity of side 5 μm (volume = 125 × 10 -18 m 3 = 125 attoliters). Consider a filter pore with a diameter of 5 nm and a length of 5 nm. A solid-state pore with these dimensions is possible. With Poiseuille flow the time required to evacuate the cavity is about 3 minutes. With an array of 6 × 6 pores it is 5 seconds. Electrolytic cells (e-cells) with nanopores offer a simple way to count the AAs released into a container. The presence of an analyte can be inferred from the current blockade caused when the analyte translocates through the pore under electrophoresis, this is used to update a counter. This is a binary measurement and is independent of the AA (the method's digital character is owed to this in part). Thus what is normally a high-precision analog measurement in nanopore studies is reduced to a simple threshold-based binary signal. Such an e-cell can be designed as an extension of the container; the latter becomes the cis chamber of an e-cell (Figure 2e). One of the disadvantages with nanopores is the high translocation speed of the analyte through the pore, which usually exceeds the bandwidth of the detector. A label-free method to overcome this problem is described in ; it is fairly easy to implement and consists of an e-cell with a three layer membrane of the form (solid-state membrane, conducting layer, solid-state membrane) containing a stack of three pores. A bi-level voltage profile with a positive voltage between cis and the conducting layer and a negative voltage between the conducting layer and trans is used. As shown in , in conjunction with pH tuning such a voltage profile leads to translocation times of ~1 ms or more for any of the 20 proteinogenic AAs so that a low bandwidth (1-10 Khz) detector can make the required binary detection without difficulty. Alternatively the AA can be labeled with an optical tag prior to entering the pipeline (this is equivalent to precolumn derivatization in column chromatography). The AA can be detected as it emerges on the trans side of the nanopore of an e-cell . Labeling usually adds a complex step to the workflow, most protein sequencing and identification methods use multiple optical labels to discriminate among monomers (4 in DNA; ~3 or 4 with proteins at the present time ). The labeling requirement in the present case is minimal because there is no need to use different colors for different AAs, a single color is sufficient. ## Notes 1) The outputs of the 20 counters N = [N(AA1), ..., N(AA20)] in Figure 1 effectively constitute a quantitating digital chromatogram for the method. 2) The efficiency of the pipeline can be increased in different ways. For example, when a counter at some level j stops counting there are no more AAj molecules in the sample; this counter state can be used to bypass that stage in the pipeline. 3) Almost all of the reactants can be reused. Thus there is no need to replace tRNA and AARS molecules in any of the separator units, unused ATP is continually being recycled through the pipeline, and AMP, the product of tRNA charging, can be fully recovered at the end of the separation process. 4) The pipeline of Figure 1 can be used for peptide sequencing by feeding successive terminal residues cleaved from a peptide to Stage 1 of the pipeline. Exactly one of the counters will output a value ≥ 1, all others will remain at 0 after each such terminal residue has been processed. A single pass through the pipeline is sufficient. See for a discussion and a parallel architecture for peptide sequencing. 5) The e-cell can be skipped if optical detection is used. Thus a fluorescent tag can be attached to an AA before it enters the pipeline and all 20 containers examined optically for a freed AA. Exactly one of the containers will have an optical output, all the others will be dark. Note that the tag has to be attached to the amino group of the AA as the carboxyl group is used in the tRNA charging process by a cognate AARS to bind the AA to a cognate tRNA. 6) The separation method given here shifts the conventional processing and measurement paradigm in chemical analysis from analog to digital and appears to have no precursor in the literature. It opens up a new avenue for the use of single-molecule methods in separation science .

chemsum

{"title": "Digital chromatography: separating amino acids into spatially discrete containers", "journal": "ChemRxiv"}

synthesis_of_spongistatin_2_employing_a_new_route_to_the_ef_fragment

## Abstract: An improved route to the EF fragment of the spongistatins has been developed and employed in a synthesis of spongistatin 2. The C48-C51 diene side chain, which lacks the chlorine substituent present in spongistatin 1, presented some compatibility issues during target assembly. These were overcome by implementing a late stage Stille cross coupling to construct the diene portion of the natural product. ## Introduction According to recent WHO fgures, cancer is the 4 th largest cause of death worldwide, accounting for 13% of total mortality in 2008. 1 With an increasingly aging population, the incidence of cancer is predicted to follow an upward trend. 2 In the urgent battle against this disease, natural products have (directly or indirectly) delivered many important therapeutic advances and remain an invaluable resource in this respect. 3 Since they were isolated independently by Pettit, 4 Kitagawa 5 and Fusetani 6 in 1993, the spongistatins have attracted considerable scientifc interest, owing principally to their exceptional antitumour activity against a variety of cell types. Indeed, all nine members of the spongistatin family have displayed remarkable growth inhibition characteristics against the US National Cancer Institute's panel of 60 human cancer cell lines. 7 Of the family, spongistatin 1 is the most active, having GI 50 values between 0.025-0.035 nM with particular efficacy against cell lines derived from human melanoma. 7 Unfortunately, due to the very low abundance of these compounds, as well as the difficulties associated with isolation and purifcation, obtaining signifcant quantities of natural material for further research is neither practical nor economic. Consequently, signifcant research efforts have been made towards their total synthesis. The absolute stereochemistry of these complex 42 membered macrolides remained unknown until the frst syntheses of spongistatin 2 by Evans 8 in 1997 and spongistatin 1 by Kishi a year later. 9 Since then, several additional total syntheses have been established by Crimmins,10 Heathcock, 11 Paterson, 12 Smith 13 and ourselves. 14 For our total synthesis of spongistatin 1 we developed an efficient and scalable route to the ABCD fragment which profited from the pseudo-C2 symmetry of this fragment. 14d It was clear, however, that despite successfully completing the total synthesis of spongistatin 1, access to biologically signifcant quantities of these natural products would require an improved synthetic route to the EF fragment; one which was more scalable, required fewer synthetic steps and provided a higher overall yield. We report herein our second generation synthesis of the EF fragment and its subsequent employment in a total synthesis of spongistatin 2. 14e ## Retrosynthesis In line with previous syntheses of the spongistatins, 8-14 our initial retrosynthetic analysis begins with disconnection of the C41(O)-C1 ester and the C28-C29 olefn to afford the ABCD aldehyde fragment, along with the corresponding EF phosphonium salt (3) (Fig. 1). As we already had a satisfactory route to the ABCD fragment (with material in hand), 14d as well as a proven endgame strategy, we sought to use a similar protecting group regime for this new synthesis. However, whilst our earlier synthesis of spongistatin 1 used an EF fragment containing a fully assembled C48-C51 chlorodiene appendage, the norchlorodiene required for spongistatin 2 alerted us to potential compatibility concerns during the subsequent removal of the PMB protecting groups. 11d Conjugated dienes are prone to undergo side reactions when exposed to DDQ, 15 the standard reagent used for the oxidative cleavage of PMB groups, and it seems the chlorine substituent in spongistatin 1 mitigates this. 12a,14d These concerns notwithstanding, we elected to continue with PMB protecting groups for the EF fragment. Experiments (including model studies) confrmed that the diene would likely not survive the conditions used to remove the PMB groups. For this reason we chose to append a vinyl iodide substituent which we hoped to transform at a late stage (after PMB deprotection) into the diene using a Stille coupling with tributyl(vinyl)tin. 16 Further disconnection of EF fragment 3 via two aldol reactions leads to trans vinyl iodide 4, F ring-containing fragment 5 and aldehyde 6 as potential coupling partners. Fragment 5 would then derive from the intramolecular oxy-Michael cyclisation of substrate 7. Development of a sequence capable of furnishing multigram quantities of fragment 5 was therefore key to a scalable route to the EF fragment and thereafter the completion of spongistatin 2 (2) (Fig. 1). ## Results and discussion Our synthesis began from the commercially available (S)-Roche ester 8. This was protected as its p-methoxybenzyl ether, reduced to the corresponding aldehyde and reacted with Ph 3 PCHCO 2 Et to furnish enoate 9 in excellent yield as a single (E)-isomer. Dihydroxylation using the Sharpless 17 AD-mix-b provided 10 with excellent diastereoselectivity (dr ¼ 94 : 6). Fortunately, this diol was highly crystalline and recrystallisation furnished a diastereomerically pure product in 92% yield. Protection as the acetonide followed by addition of MeLi$LiBr to the ester and Wittig reaction of the resulting ketone afforded 11. This 7 step sequence was carried out on large scale (100 mmol batches), allowing for the generation of 11 in substantial amounts. Removal of the PMB group, oxidation and Wittig-Horner reaction with known phosphonate 13 18 yielded 14, which underwent selective dihydroxylation (dr ¼ 86 : 14, single isomer after chromatography) to afford diol 15 (Scheme 1). We next focused on the construction of the F ring by means of an oxy-Michael cyclisation reaction. Encouraged by studies undertaken by Paterson and coworkers, 12d our initial efforts centred on removal of the acetonide protecting group in 15 under acidic conditions, which we hoped would be accompanied by spontaneous cyclisation to form the F ring in a one-pot procedure. However, although the acetonide could cleanly be removed (using MP-TsOH), the resulting tetraol 7 was resistant to subsequent cyclisation, despite assaying a range of conditions for this transformation (inc. Amberlyst A-15, TFA, Tf 2 NH, TfOH, AcOH, TMSCl : EtOH (0.3 M)). This being the case, we focussed our attention on the cyclisation process itself. The use of a variety of different bases (TBAF, NaH, KHMDS, t-BuOK, DBU, Et 3 N, NaHCO 3 ) under different conditions (solvent and temperature) either failed to promote any reaction or afforded the desired product only in low yield (#50%). 19 In many instances the desired thermodynamic pyran 16 was formed in varying ratio with products resulting from epimerisation at C42 or C43, the latter of which tended to form the 6,5-cis fused bicyclic lactone 17. Having achieved little success with bases, we turned instead to the use of Lewis acids in order to affect this crucial transformation. Although such methods to construct densely functionalised tetrahydropyran rings are not widely reported, 20 it seemed reasonable that a Lewis acid might activate the enoate towards nucleophilic attack. A selection of the additives which were screened are shown in Table 1. The presence of a variety of Lewis acids (20 mol%) at 80 C in ethanol either led to no reaction or to decomposition (entries 1-5). The results for Pt(IV) and Au(III) were disappointing as we speculated that the propensity of these metals to activate ynones toward intramolecular reactions would extend to enones. 21 In order to investigate the use of higher temperatures with the same solvent, a microwave reactor was used to conveniently allow the solvent to be superheated under pressure (ethanol b.p. ¼ 78.4 C at 1 atm) (entries 6-11). 22 Extensive decomposition was observed in several cases (entries 6-8). However, ZnCl 2 , which had no effect at 80 C, afforded at 180 C the desired product 16 in 65% yield, in addition to lactone 17 (15%). Whilst the use of other zinc(II) halogenide additives (ZnF 2 , ZnBr 2 , ZnI 2 ) were ineffective (entry 10), Zn(OTf) 2 proved to be the optimal choice, with the desired product produced in 75% yield (entry 11). Fortunately, this reaction proved to be both higher yielding (88%) and more scalable when performed in a sealed Schlenk tube, using conventional heating (oil bath) at 100 C. Negating the need for a microwave reactor allowed multigram batches of material to be processed (entry 12). The origins of the superiority of Zn(OTf) 2 as a catalyst for this reaction are not clear, although it seems unlikely that any residual TfOH is acting as a catalyst (entry 13). Having developed an effective procedure to ensure scalable and efficient access to the F-ring scaffold we required a method that would allow for the protection of all three hydroxyl groups present in 16. PMB protection of hydroxyl groups is typically achieved using one of two sets of conditions: (a) base in conjunction with PMB-Cl/Br and a catalytic amount of TBAI/ NaI 23 or (b) PMB-trichloroacetimidate (TCA) combined with a Brønsted or Lewis acid catalyst. 24 The latter method was investigated initially since experience gained from the preceding oxy-Michael cyclisation had revealed the C43 stereocentre to be highly prone to epimerisation. However, despite trying a variety of different acid catalysts (TfOH, La(OTf) 3 , BF 3 $OEt 2 , Ph 3 CBF 4 ) these efforts were met with decomposition or only low yields of the desired tris-PMB ether 18. Little improvement was observed using tried-and-tested basic conditions. Wishing to avoid the daunting prospect of reworking our protecting group strategy, we persevered. After investigating a plethora of reagents and workarounds, we were pleased to fnd that a respectable 72% yield of 18 could be obtained using PMBlepidine in the presence of MeOTs. 25 Use of this reagent offered several advantages compared to PMB-TCA: (a) it was stable at room temperature and could be easily prepared, 25 (b) the reactive intermediate could be formed under non-acidic conditions and (c) the lepidine byproduct could be easily removed by silica gel chromatography. Hydrolysis of the ester functionality and formation of the corresponding Weinreb amide 26 proceeded smoothly. There-after dihydroxylation of the alkene and Pb(OAc) 4 -mediated cleavage of the resulting diols furnished ketone 5; allowing for interception of our frst generation synthesis of spongistatin 1 (Scheme 2). 14e With ketone 5 in hand, our efforts shifted to the construction of the E ring. The boron mediated aldol coupling of ketone 5 and aldehyde 19 27 was achieved in high yield (82%) and in excellent diastereoselectivity (dr > 95 : 5), giving the Felkin-Anh product (20). 28 Presumably the 1,2-stereochemical induction predicted by the Felkin-Anh model overrides any 1,3-asymmetric induction favoured by the chiral aldehyde. 29 Treatment of 20 with PPTS in the presence of trimethyl orthoformate and MeOH effected C33(OH) silyl deprotection and triggered spontaneous E ring cyclisation with concomitant formation of the C37 methyl acetal. The C35(OH) was then protected as its TBS ether, and the Weinreb amide functionality converted to the corresponding ketone (21) in quantitative yield by reaction with a methylcerium reagent prepared in situ from methyllithium and anhydrous ceric chloride. 30 Completion of EF fragment 21 via this new route resulted in a signifcantly shorter reaction sequence (22 linear steps) and a much higher overall yield of 20% compared to our frst generation synthesis (27 linear steps and 5.6% overall yield). 14e This enabled the production of 6.4 g of highly functionalised compound 21, and compares favourably with other routes reported towards similar EF fragments. 12,13 At this stage the route diverges from our original spongistatin 1 synthesis. Aldol coupling of 21 with an excess 31 of 3-iodoacrolein 22 32 furnished compound 23 in high diastereoselectivity (dr > 95 : 5) and good yield (65%) (Scheme 3). 33 This was then TBS-protected, and subjected to the Takai-Lombardo methylenation 34 conditions followed by Finkelstein reaction 35 in a highly efficient 3 step sequence in 74% overall yield. Following considerable experimentation, a Wittig olefnation reaction between the corresponding phosphonium salt 3 (1.0 equiv.) and ABCD aldehyde 24 (1.3 equiv.) 14d in the presence of LiHMDS (3.0 equiv.) at 78 C resulted in a 56% yield of the (Z)-alkene. The three PMB protecting groups in this compound were then oxidatively cleaved in 71% yield using DDQ. At this stage the C48-C51 diene side chain could be completed. Pleasingly, the necessary Stille cross coupling reaction was not perturbed by the C1 allyl ester protecting group. A one-pot Stille-coupling/deprotection sequence 36 was effected using Pd 2 dba 3 (10 mol%), AsPh 3 (40 mol%), tributyl(vinyl)tin (5.0 equiv.) and morpholine (10.0 equiv.), affording 26 in an excellent yield of 84%. By installing the diene unit at this late stage, we effectively avoided the side reactions (oxidation, cycloadduct formation 11d ), that we and others had previously observed using DDQ on a similar diene substrate. 11d Yamaguchi macrolactonisation 37 of seco-acid 26 using the conditions developed by Heathcock 11 led to a mixture of the desired product as well as some byproducts. From the 1 H NMR of the crude product mixture, it appeared that partial hydrolysis of the methyl acetal at C37 had occurred, somewhat hampering full 1 H NMR analysis. However, treatment of the crude mixture with 47-51% aq. HF at 20 C resulted in global silyl deprotection and methyl acetal hydrolysis to deliver spongistatin 2 (2) in 42% over the two steps. The synthetic material was found to be identical to the natural product in all respects ( 1 H NMR, 13 C NMR, IR, mass spectrometry, optical rotation). 4b,5c ## Conclusions In summary we have developed a highly scalable and efficient preparation of the EF fragment of the spongistatins. This was made possible by the design of a route that gave access to multigram quantities of tetraol 7. Satisfactory conditions for the F ring cyclisation initially proved elusive but this key transformation was eventually achieved in a yield of 88% using Zn(OTf) 2 . Further synthetic manipulation brought about interception with our frst generation synthesis, leading to compound 21 in 22 steps (longest linear sequence) and 20% overall yield. With 5 fewer steps and a greater than threefold increase in yield, this clearly represents a signifcant gain in efficiency. A boron mediated aldol reaction using 3-iodo-acrolein (22) allowed us to install a truncated vinyl iodide side chain. By extending this to the diene at a late stage via a Stille cross coupling with tributyl(vinyl)tin, we managed to avoid serious incompatibility issues related to PMB deprotection and were able to complete a total synthesis of spongistatin 2 (2) with an overall yield of 1.3% (from Roche ester) over 32 steps (longest linear sequence). To date, 6.4 g of 21 have been prepared. With scalable routes to both the EF and ABCD fragments at our disposal, we are now in a position to synthesise spongistatin 2 (and other members of the family) in quantities useful for further biological investigation.

chemsum

{"title": "Synthesis of spongistatin 2 employing a new route to the EF fragment", "journal": "Royal Society of Chemistry (RSC)"}

face_and_edge_directed_self-assembly_of_pd₁₂_tetrahedral_nano-cages_and_their_self-sortin

## Abstract: Reactions of a cis-blocked Pd(II) 90 acceptor [cis-(tmeda)Pd(NO 3 ) 2 ] (M) with 1,4-di(1H-tetrazol-5-yl) benzene (H 2 L 1 ) and [1,3,5-tri(1H-tetrazol-5-yl)benzene] (H 3 L 2 ) in 1 : 1 and 3 : 2 molar ratios respectively, yielded soft metallogels G1 and G2 [tmeda ¼ N,N,N 0 ,N 0 -tetramethylethane-1,2-diamine]. Post-metalation of the gels G1 and G2 with M yielded highly water-soluble edge and face directed self-assembled Pd 12 tetrahedral nano-cages T1 and T2, respectively. Such facile conversion of Pd(II) gels to discrete tetrahedral metallocages is unprecedented. Moreover, distinct self-sorting of these two tetrahedral cages of similar sizes was observed in the self-assembly of M with a mixture of H 2 L 1 and H 3 L 2 in aqueous medium. The edge directed tetrahedral cage (T1) was successfully used to perform Michael reactions of a series of water insoluble nitro-olefins assisted by encapsulation into the cage in aqueous medium. ## Introduction Nature, especially in biological systems, has an extraordinary ability to develop complex and functional molecular assemblies employing reversible non-covalent interactions. 1 These natural aesthetic examples have enticed synthetic chemists over the past several years to develop potential synthetic protocols to produce myriad complex assemblies employing non-covalent interactions like H-bonding, p-p interactions and stronger metal-ligand coordination. In principle, by tuning these directional non-covalent driving forces one can construct supramolecular polymers as well as discrete molecular assemblies. Such ordered and self-organized polymers may go through gelation in the presence of strong inter-molecular interactions (network formation). 5 The three-dimensional fbrous network structure generated by the self-assembly of the gelator molecules may impart solid-like properties in the gels, which make them worthy candidates for substrate recognition, catalysis and biomedical applications. 6 Meanwhile, discrete molecular architectures are also potential candidates for stabilizing reactive intermediates, sensing, and host-guest chemistry, as well as cavity-induced 'organic transformations'. 7 Introducing multiple interactions in a single system and tuning them by varying different parameters such as solvent type, temperature and stoichiometry to construct desired molecular architectures (discrete or polymeric), including their transformation, is interesting and provides diverse functional materials. Discrete supramolecular metallocages have been designed, mainly employing pyridyl, imidazole and carboxylate linkers. 8 Polytetrazoles are not preferred linkers in designing molecular cages due to difficulty in predicting the fnal structure because of the multiple nitrogen atoms. The N-H moiety in the tetrazole ring promotes many metal complexes of polytetrazoles to form metallogels by intermolecular H-bonding. 9 The coordination-driven self-assembly strategy has inspired chemists in the past two decades to design basic 3D structures of high symmetry and well defned shapes and sizes. 7b,10 A tetrahedron is one of the common 3D geometries. Several tetrahedral molecular cages have been reported using mainly octahedral metal ions with a few examples of analogous cages of lanthanides with a higher coordination number, though their solubility is restricted to organic solvents in the majority of cases. 11,12 Due to geometric restriction, square planar metal ions like Pd(II) have not been explored much for the design of regular tetrahedral cages. Herein, we report the formation of a supramolecular Pd(II) metallogel (G1) upon 1 : 1 treatment of H 2 L 1 [1,4-di(1H-tetrazol-5-yl)benzene] with cis-(tmeda)Pd(NO 3 ) 2 (M) in water or DMSO (Scheme 1). Post-metalation of G1 with M in water/or DMSO caused deprotonation of the N-H moieties and transformed the gel into a highly water-soluble unusual tetrahedral M 12 L 1 6 (T1) cage where the donors (L 1 ) occupy the six edges of the tetrahedron. Such post-metalation was expected to form an open 2D square M 8 L 1 4 as observed using pyrazole linkers instead of a closed 3D tetrahedral cage. 13 This unusual outcome enticed our attention towards a face directed tetrahedron to examine the generality of designing tetrahedral cages of square planar metal ions employing polytetrazole donors. Replacement of H 2 L 1 with H 3 L 2 [1,3,5-tri(1H-tetrazol-5-yl)benzene] in the above-mentioned two-step reaction afforded a Pd(II) gel (G 2 ) followed by a water soluble chiral tetrahedral cage M 12 L 2 4 (T 2 ) (Scheme 1), where the four triangular faces of the tetrahedron were occupied by L 2 . T1 and T2 are unusual from a symmetry point of view. 11b,14 The spatial orientations of the linkers L 1 and L 2 enable the fnal assemblies (T1 and T2) to adopt a specifc geometry with a loss of symmetry to become chiral cages of square planar Pd(II) using achiral building blocks. ## Synthesis and characterization of metallogels Both the ligands (H 2 L 1 and H 3 L 2 ) were prepared from their cyano derivatives following the reported procedure. 9 Supramolecular hydrogel G1 was prepared by adding an aqueous yellow solution of the acceptor M to H 2 L 1 in a 1 : 1 molar ratio with a total weight percentage of 2. The mixture was heated with stirring at 60 C for 2 h to give a transparent solution, and subsequent cooling to room temperature yielded the gel G1 (Scheme 1). A similar gel was also obtained when DMSO was used as a solvent in the above reaction. The viscoelastic nature of the gel was characterized by two types of dynamic rheology 15 experiments: (a) frequency sweep at a constant stress of 1.0 Pa and (b) stress sweep at a constant frequency of 1.0 Hz. The stiffness of the gel (G 0 /G 00 ) can be measured from the frst experiment whereas fragility (yield stress) can be measured from the second one. For G1 and G2 these experimental values are quite small (ESI †), which suggests the formation of a weak gel (Fig. 1). Finally, feld emission scanning electron microscopy of the xerogels revealed that the formation of a fbrous nanostructure (tertiary structure) is responsible for gelation (ESI †). ## Synthesis and characterization of nano-cages Treatment of the metallogel G1 with an aqueous solution of one molar equivalent of M at 60 C for 3 h yielded T1. To investigate the extent of the deprotonation of the ligand H 2 L 1 during the self-assembly process, a pH monitored self-assembly reaction was performed in water with 2 molar equivalents of M and one Scheme 1 Schematic representation of the synthesis of metallogels and their facile conversion to 3D tetrahedral nano-cages. molar equivalent of H 2 L 1 . The change in pH of the reaction mixture after the consumption of all the ligand H 2 L 1 (ESI †) clearly suggests deprotonation of both of the N-H protons of H 2 L 1 . Evaporation of the solvent under vacuum and subsequent washing of the resulting solid with acetone yielded the fnal product in pure form. In the case of the DMSO gel, a similar procedure was followed with a DMSO solution of M, and fnally, T1 was obtained as a white solid upon treating the solution with ethyl acetate. 1 H NMR analysis of the product (T1) in D 2 O displayed a single peak at 9.1 ppm, which is signifcantly downfeld shifted compared to the peak for free H 2 L 1 (ESI †). ESI-MS analysis in water indicated a [12 + 6] composition of M and L 1 in the fnal product (T1) by the appearance of two major peaks at m/z ¼ 1110 and 1500 with isotropic distribution patterns corresponding to the fragments [T1(NO 3 ) 8 ] 4+ and [T1(NO 3 ) 9 ] 3+ , respectively (ESI †). T2 was synthesized following the above-mentioned procedure using H 3 L 2 instead of H 2 L 1 . The metallogel G2 was obtained by treating H 3 L 2 and M in a 2 : 3 molar ratio in DMSO or water. G2 was converted to T2 by post-metalation treatment with M. The presence of a singlet at 10.05 ppm in 1 Finally, both the cages were successfully crystallized by diffusion of acetone vapour into the aqueous solutions of the cages. Single crystal XRD analysis of both T1 and T2 unequivocally confrmed the formation of 3D tetrahedral cages (Fig. 2). T1 was crystallized in the C2/c space group and each vertex corner of the tetrahedron contains 3 Pd(II) acceptors connected by the nitrogens of the tetrazole moieties of the linkers to form a Pd 3 triangle and the ligands L 1 are fastened by two corners along the edges. Whereas T2 was crystallized in the I222 space group and each vertex corner of the tetrahedron contains three acceptors and the linker L 2 is tied by three corners along the face. So, in the case of T1 the ligands occupy the edges of the tetrahedron while in T2 the donors occupy the triangular faces of the tetrahedron. L 1 and L 2 bind the Pd(II) centers through the nitrogens of the 1 and 3 positions of the tetrazole rings (Fig. 3). Such binding modes of the tetrazole ligands (for L 1 and L 2 ) make them special in terms of "symmetry breaking", 11b because L 1 and L 2 have no plane of symmetry when they are coordinated to metal centers, except in the molecular plane, which no longer exists in the 3D cage. Now if we consider the 1,3 binding mode, the tetrazole moiety is residing on the right with respect to the phenyl ring (Fig. 3). If we consider this geometry as a D confguration; the 2,5 binding mode brings the tetrazole moiety over to the left side, which may be represented as a L confguration. Crystal structures of T1 and T2 (CCDC 1471523 and 1471434 †) clearly displayed that, in a particular nano-cage, all the linkers have a similar type of coordination mode with the Pd(II) acceptors as shown in Fig. 4. Hence, the handedness of all the Pd 3 triangles (vertices of each tetrahedron) is the same, either DDDD or LLLL. Such spatial arrangements of the donors and acceptors make the resulting cages chiral even without using any chiral building blocks. As the cage T1 was crystallized in a centrosymmetric space group (C2/c), the crystal packing exposed the presence of both enantiomers along the c glide plane in the unit cell. Although T2 was crystallized in a chiral (I222) space group, the Flack parameter was evaluated to be 0.5, which is due to the formation of an inversion twin i.e. a racemic mixture. Nonetheless, linker L 3 forms a prismatic structure M 6 L 3 3 (P) (CCDC 1471478 †).This was synthesized in a similar way as followed for T1 and T2. The binding mode of L 3 is shown in Fig. 3. It forms two Pd 3 triangles with opposing handedness yielding an achiral geometry (Fig. 4). So, the achiral or chiral nature of the fnal assemblies is controlled by the handedness of the Pd 3 triangular units. ## Variable-temperature 1 H NMR study of T1 Interestingly, the 1 H NMR study of T1 at room temperature does not support the asymmetric nature of the solid-state XRD structure. The single peak at 9.1 ppm is due to the possibility of rotation of the phenyl ring with a frequency higher than the frequency of the NMR technique at room temperature. Variable temperature NMR spectra at low temperature were needed to confrm this phenomenon. Although the cage was synthesized in water or DMSO, it is soluble in MeOH. This allowed us to study 1 H NMR of the cage at 50 C (Fig. 5), which displayed two different peaks at 8.5 and 9.8 ppm. Moreover, a DOSY NMR study at that temperature showed the presence of a single product. Upon increasing the temperature, the peaks started to , 2016, 7, 5893-5899 This journal is © The Royal Society of Chemistry 2016 merge together and at around 20 C these peaks disappeared and the original peak at 9.1 ppm appeared. ## Self-sorting experiment of T1 and T2 As both the linkers L 1 and L 2 form tetrahedral cages of almost equivalent shape and size with complementary building units (edge directed and face directed respectively), we were curious to know whether the presence of both the donors (L 1 and L 2 ) in a single reaction mixture would result in these two individual tetrahedral cages by a self-sorting process or a complicated multicomponent product. To investigate this, a self-assembly experiment was carried out in water taking M, H 2 L 1 and H 3 L 2 in a 24 : 6 : 4 molar ratio. The fnal product was isolated and characterized by 1 H NMR and mass spectroscopy. The presence of two sharp singlets at 10.07 and 9.13 ppm in the 1 H NMR spectra clearly indicates the formation of T2 and T1 in pure form without the presence of any byproduct (Fig. 6). ## Catalytic Michael addition reaction The donors (L 1 ) in the edge directed cage T1 occupy the edges of the tetrahedron by joining two Pd 3 triangles, keeping four triangular faces of the tetrahedron open, having dimensions of 6.8 6.9 , which may allow one or more aromatic guests to enter. Cage T2 has complementary geometry, where all four faces of the tetrahedron are occupied by the linker L 2 leaving no open windows for aromatic guest encapsulation. The edges of this cage are also blocked by the methyl groups of the acceptor leaving no pathway for the entry of guests. This structural nature of the cages motivated us to study the possibility of encapsulating a water insoluble aromatic guest. To verify our observation the aqueous solutions of the cages were treated with 1-(2-nitrovinyl)naphthalene (1). As anticipated, the aromatic guest (1) was encapsulated in T1 which was identifed by a change in the colour of the cage solution (colourless to light yellow) and was fnally confrmed by 1 H NMR spectroscopy (Fig. 7) where signifcant up-feld shifts of the guest peaks (1) were noticed. The stoichiometry of cage vs. guest was evaluated to be 1 : 2 from the 1 H NMR spectra (ESI †). The broad 1 H NMR peaks for the encapsulated cage can be explained in terms of the rotational movement of the phenyl ring of the cage T1, which is restricted by the guest molecules present inside the cage. This host-guest binding was further confrmed by 1 H DOSY NMR (ESI †), which showed identical diffusion coefficients for the guest (1) and the host cage. Several attempts to crystallize the guest encapsulated cage 1 3 T1 have so far been unsuccessful. The host-guest (1 : 2) complex was modeled and the structure was optimized by a semi-empirical method with a PM6 basis set. As the nitro group is more polar than the naphthyl group, it may lean outside the cavity. However, the naphthyl moieties are stabilized inside the hydrophobic cavity of the cage by p-p stacking interactions (Fig. 8). This successful encapsulation of the aromatic nitro-alkene enabled it to be soluble in an aqueous medium and it was possible to carry out further organic transformation utilizing the cage T1 in a phase transfer catalytic manner. Several reactions were studied using 1,3-dimethylbarbituric acid (3) and different nitro-alkenes (2) (Table 1). Although similar reactions were studied by us using a urea decorated functional selfassembled molecular prism in a heterogeneous manner, 7i the present study was done in a homogeneous manner. The initial results presented in Table 1 indicate an enhancement of the yield in the presence of the cage T1. Cage T2 didn't show any such encapsulation of nitro-olefns due to the absence of large open windows. Control reactions were carried out with T2, which indicated almost no catalytic effect on these reactions. ## Conclusions In conclusion, self-assembly of a Pd(II) acceptor M with polytetrazole donors H 2 L 1 and H 3 L 2 in 1 : 1 and 3 : 2 molar ratios yielded supramolecular metallogels G1 and G2, respectively. Post-metalation reactions of the gels with M transformed them to edge-and face-directed self-assembled water-soluble tetrahedral cages T1 and T2. Reversible transformation of the cages to their corresponding parent gels was also achieved upon treating them with respective tetrazoles. To the best of our knowledge, such facile and reversible transformation of metallogels to discrete metallocages is unusual. Although pyridine, imidazole and carboxylates have been widely employed to obtain metallosupramolecular discrete architectures, T1-T2 represent a unique class of tetrahedral architectures of square planar metal ions employing polytetrazoles as linkers. Interestingly, the coordination modes of the tetrazole units promote 'symmetry breaking' in these assemblies. Single crystal structures showed that the handedness of the Pd 3 vertices formed by the tetrazole units is responsible for the 'symmetry breaking' which led to the formation of chiral nano-cages T symmetry. However, the binding mode of L 3 imparts a mirror symmetry leading to an achiral prismatic structure P. The mixed-ligand self-assembly with M produced self-sorted cages T1 and T2 without any by-product. This clean self-sorting between the iso-structural tetrahedral cages of similar sizes in a complex mixture of H 2 L 1 , H 3 L 2 and M is reminiscent of the selectivity observed in nature. The open triangular windows in the edge directed tetrahedron (T1) enabled it to encapsulate water insoluble aromatic nitro-olefns in aqueous medium followed by Michael reactions of the nitro-olefns with 1,3-dimethybarbituric acid using phase transfer-type catalysis. The freshly prepared water-soluble tetrahedral cages provide a platform for the development of a new generation of functional molecular nanovessels employing polytetrazole donors for chemical reactions and encapsulation of various guests.

chemsum

{"title": "Face and edge directed self-assembly of Pd₁₂ tetrahedral nano-cages and their self-sorting", "journal": "Royal Society of Chemistry (RSC)"}

photoinduced_charge_transfer_from_quantum_dots_measured_by_cyclic_voltammetry

## Abstract: Measuring and modulating charge-transfer processes at quantum dot interfaces are crucial steps in developing quantum dots as photocatalysts. In this work, cyclic voltammetry under illumination is demonstrated to measure the rate of photoinduced charge transfer from CdS quantum dots by directly probing the changing oxidation states of a library of molecular charge acceptors, including both hole and electron acceptors. The voltammetry data demonstrates the presence of long-lived charge donor states generated by native photodoping of the quantum dots as well as a positive correlation between driving force and rate of charge transfer. Changes to the voltammograms under illumination follow mechanistic predictions from classic zone diagrams and electrochemical modeling allows for measurement of the rate of productive electron transfer. Observed rates for photoinduced charge transfer on the order of 0.1 s -1 are calculated, which are distinct from the picosecond dynamics measured by conventional transient optical spectroscopy methods and are more closely connected to the quantum yield of light mediated chemical transformations. ## Introduction Photoinduced charge separation is a key step in artificial photosynthesis for the conversion of solar energy to high-value chemical compounds. 1 Quantum dots (QDs) have long been promoted as ideal photosensitizers for photocatalysis due to their high extinction coefficients, electronic tunability, and solution processability, 2 but efficient extraction of high energy charge carriers from QDs remains a design challenge. 3 Photoinduced charge transfer from QD donors requires transfer of charges across a complex interface between the inorganic QD core and a molecular cocatalyst or substrate in solution. 4,5 This complicated interface comprises a high prevalence of defect electronic states in the QD, 6,7 and the covalent and non-covalent interactions between the QD, the insulating ligand shell, and the charge acceptor. 8 Conventional models of charge transfer in molecular systems (e.g. the two state system described by the Marcus formalism) are therefore insufficient to predict the rate of useful charge extraction from QDs, prompting experimental exploration. 9 Photoluminescence spectroscopy and transient absorbance spectroscopy 13,14 are frequently employed to determine rates of photoinduced charge transfer in QD systems. In these experiments, the charge transfer process measured is pseudo-unimolecular with a first-order rate constant. This rate presumes preadsorption of the charge acceptor to the QD and does not consider freely diffusing charge acceptors nor the dynamic noncovalent chemical interactions between the QD and acceptor. 10,13,15,16 While determination of the first order rate has utility, especially when compared with other unimolecular photophysical processes such as electron/hole recombination, there is a large disconnect in the literature between the time scale for this fundamental process (picoseconds) and the time scale of photocatalytic reactions (minutes) 17,18 . It may then be counterintuitive that several reports have found that the ratelimiting step of photocatalysis is charge transfer from nanocrystal photosensitizers to substrate or cocatalyst. This disconnect begs us to consider that the spectroscopic first order rate of charge transfer does not accurately report on the rate of production of charge separated states, and instead a new method is needed to understand processes taking place on the same time scales as chemical reactions. 3 Alternatively, charge transfer can be rationalized as a bimolecular reaction that is first order with respect to both the charge donor (excited QD) and acceptor (substrate). 22 The two species must first collide before charge can be extracted from the QD, and the rate of observed charge extraction will depend on the frequency of collisions, the rate of the fundamental photophysical process observed by time-resolved spectroscopies, and the fraction of collisions that allow strong electronic coupling between the QD and charge acceptor. To this end, we turned to cyclic voltammetry (CV), a measurement tool that directly probes the changing oxidation state of a redox active small molecule. CV has been employed in homogeneous electrocatalysis literature as a probe for the changing oxidation states of a molecular electrocatalyst, 23 and has been theorized to be a tool for evaluating molecular photoelectrocatalysis. 24 We hypothesized that CV could be extended to systems involving photoinduced charge transfer from QDs. In the electrocatalysis literature, one of the simplest and most well-understood systems is described by two reactions: the oxidation and reduction of the electrocatalyst at the electrode, and the catalytic reaction in which the electrocatalyst transfers charge to substrate. This mechanism is termed ErCi'. In such a system, the CV is modulated as compared to CVs in the absence of substrate, and this modulation can be quantified to obtain the rate constant for the catalytic reaction. For a thorough review of this technique, see Rountree et al. 23 In this work we aim to analogously measure the rate of productive charge extraction from QDs using CV (Scheme 1). We believe that the rates obtained through this measurement (kPCT) will accurately reflect the extraction of charge from QDs and will bridge the gap in time scales between photophysics and chemical transformations. Scheme 1. The ErCi' mechanism employed in the electrocatalysis literature (left), and the extension of this mechanism to photoinduced charge transfer from an excited QD (QD*) to a molecular acceptor (M). In this work, kPCT represents the intrinsic rate constant of photoinduced charge transfer. ## Photoelectrochemistry cell design A traditional three-electrode electrochemical cell was modified for in situ illumination. A 448 nm LED (Luxeon Star, equipped with a 12° beam optic, FWHM 20nm) was positioned under a quartz cuvette with a polished bottom and open top (Figure 1). The cuvette was placed on top of the LED. The LED was powered by a DC power supply (Nice-Power). The driving current was 0.2-0.8A, corresponding to approximately 0.3-1.1W of illumination. Holes were drilled in a cuvette cap for the three electrodes and the glassy carbon disc working electrode (BASi) was epoxied to the cap, ensuring the light had a constant and known pathlength (0.67 mm) through the solution to the active area of the working electrode. The pathlength is small to minimize undesired convection effects on the voltammogram from photoirradiation, 25 as well as to decrease the amount of light that is attenuated by the highly absorbent QDs in solution before reaching species near the working electrode surface. The counter electrode was a platinum wire, and the pseudo-reference electrode was a silver wire in a ceramic-fritted glass tube (Pine) filled with 0. ## Solvent and electrolyte design for photoelectrochemistry The selection of solvent and supporting electrolyte is critical to obtaining electrochemical measurements suitable for quantitatively monitoring photoinduced charge transfer. The solvent reorganizes to facilitate charge transfer, both from the working electrode to the redox probe and between the QD and the redox probe, so it must be polar to minimize internal resistance. The solvent must also allow high electrolyte concentration and have a wide electrochemical window to screen a wide range of redox probes. These electrochemical considerations are general, but for photoelectrochemistry, the solvent must additionally not undergo any photodecomposition nor reactivity with excited QDs. Previously, our group has found that a mixture of 9:1 THF:MeCN was able to suspend oleic acid capped QDs with low internal resistance. 26 However, when THF was used in this work, the CV exhibited current crossover (Figure S1), an unusual observation that indicates that the product of Faradaic oxidation on the forward scan of the CV has been chemically converted to another species that is more easily oxidized and observed on the backward segment. 27 Given prior observations that THF degrades under illumination to form reactive radicals, 28 THF is not a suitable solvent for this study. Dichloromethane was another attractive solvent due to its modest polarity and ability to disperse assynthesized QDs. Unfortunately, CVs under illumination displayed oscillations in the current, especially in the diffusion limited regime (Figure S2). These oscillations were the result of gas bubbles evolving and reaching the surface of the working electrode, which was observed visually during illumination of the sample. Headspace analysis detected production of methane after illumination (Figure S3). With these observations, as well as prior observation of dehalogenation of CH2Cl2 with QD photocatalysts 29 , we conclude that the system photocatalytically dehalogenates CH2Cl2 to methane, so CH2Cl2 is not a suitable choice for photoelectrochemical measurement. Another limitation in solvent choice is the solubility of the QDs, as QDs are often natively capped with aliphatic ligands that prevent dispersion in polar solvent at high electrolyte concentration. Ligand exchange was performed on QDs to replace the native oleic acid ligand shell with 2-[2-(2-Methoxyethoxy)ethoxy]acetic acid (MEEAA), which is known to be an amphiphilic ligand that has dissolved nanocrystals in solvents ranging from toluene to water. 3031 In our hands, 3.8 nm CdS QDs capped with this ligand are readily soluble in a variety of polar solvents, including water, acetone, and ethanol, but cannot be dispersed in some polar, aprotic solvents suitable for electrochemistry such as acetonitrile and propylene carbonate. Ultimately, benzonitrile (PhCN) was selected for this study because of the good colloidal stability of QDs in electrolyte solutions prepared using this solvent. MEEAA capped QDs in benzonitrile solution remain suspended for at least several months even in the presence of electrolyte. Finally, the solvent and electrolyte should allow reversible CVs for all the redox probes in the absence of QDs and illumination. Using the more common tetrabutylammonium salt of the [PF6]anion prevented reversible redox behavior of ferrocenecarboxylic acid (FcCOOH), presumably due to the high electrophilicity of the [FcCOOH] + cation. Instead, the tetrabutylammonium salt of the weakly coordinating anion [B(C6F5)4]was used. This completely fluorinated phenyl borate is known to stabilize organometallic cations, such that the only allowed processes in the CVs were oxidation and reduction of the metal center. 32 When this anion was used in the supporting electrolyte, FcCOOH displayed nearly ideal electrochemical reversibility. 33 ## Photodoping and slow electron trapping observed by CV By illuminating the sample, the chemical reaction in the ErCi' mechanism is turned on, and we observe distortion of the CV shape. Classically, the shape of the CV in this mechanism can be described by a zone diagram (Figure 2a), where the zone observed will depend on the concentrations of charge donor and acceptor, as well as the scan rate and the intrinsic rate of charge transfer. Generally, the solution in the electrochemical cell was 1.1×10 -5 M QDs and approximately 130 equivalents of the redox probe. After beginning illumination of a solution of CdS QDs with ferrocene a representative redox probe, successive CV scans continue to distort as compared to the dark trace for several minutes (Figure 2b). The CVs move to the right across the ErCi' zone diagram, from zone D to zone KD to zone KS, which by analogy to electrocatalysis literature 23 demonstrates an increase in the concentration of charge donor states (herein represented as [QD*]) (Figure 2a,b). This distortion occurs over ca. 20 minutes of illumination and then stabilizes, corresponding to a stabilization of [QD*]. This extremely long time scale until equilibration of [QD*] as compared to the speed of photoexcitation (femtoseconds) suggests that the charge donor state is not simply an exciton, but rather the product of a slow chemical process following excitation. Some excitons may directly act as charge donors, but exciton dissociation directly to the molecular probe is not the only process observed. Previous studies have reported native n-type photodoping in cadmium chalcogenide QDs over the same timescale observed in this study, wherein after excitation a valence band hole is extracted without any added reductant, leaving behind a long-lived conduction band electron. 34,35 To further investigate the nature of the charge donor state, we monitored the solution with successive CV scans after illumination was stopped. Over the course of ca. 20 minutes, the CV recovers back to its original dark trace as QD* is slowly depleted to zero, thus tracking to the left along the ErCi' zone diagram (Figure 2c). Others have also reported that negatively photodoped QDs live for many minutes due to extremely slow conduction band electron trapping. The long-lived electron donor state herein may be long-lived conduction band electrons and/or electrons trapped as reduced surface Cd, but this technique alone cannot deconvolute the two. While this work deals with QDs that natively photodope, the technique is agnostic to the specific nature of the electron donor state. The changing oxidation state of the redox probe is being measured rather than changing photophysics of the QD, so the measurement is general regardless of the identity of the charge donor state. While [QD*] stabilizes for a given light intensity after many minutes, the stable CVs of a representative redox probe, FcCOOH are not the same when the light intensity is varied. As the power of illumination is increased from 0.33 W to 1.14 W, the stable CV is distorted further from the dark CV, again well matched to traversing to the right across the zone diagram (Figure 2d). This observation indicates that although at any given light intensity [QD*] reaches an equilibrium, a maximum concentration of charge donors has not been reached. It is expected that as light intensity is further increased, the CV would eventually stop distorting, but this light-saturated regime is not observed due to the limited power output of the LED light source. When QDs are added to solutions of Co(Cp)(dppe) (Cp = cyclopentadienyl, dppe = 1,2-Bis(diphenylphosphino)ethane), aminoferrocene (FcNH2), ferrocene (Fc), FcCOOH, or acetylferrocene (FcCOCH3) the CV remains unchanged for traces without illumination. This observation, alongside no observed change in the dark open circuit potential, demonstrates that none of these probes exhibit charge transfer reactions with the QDs in the dark. Furthermore, the magnitude of the current does not change upon addition of QDs to the probes in the dark, indicating no adsorption to the QDs. If indeed there was adsorption, the effective diffusion coefficient of the redox probes would decrease due to the much larger QD, decreasing the current measured in CV. Previously, FcCOOH was observed to bind to oleate-capped CdSe QDs using CV through carboxylate-carboxylate exchange with the native ligand shell. 26 In contrast, FcCOOH does not undergo similar exchange with MEEAA-capped CdS QDs. The lack of exchange is rationalized by the lower pKa of MEEAA (pKa = 3.61) 41 compared to oleic acid (pKa= 9.85) 42 . ## Mathematical Determination of ErCi' Rate Constant for Co(Cp)(dppe), FcNH2, Fc, FcCOOH, FcCOCH3 The rate constant for the photoinduced charge transfer reaction (Ci' in the ErCi' mechanism) can be determined mathematically from voltammograms when in zone KD or KS, which are the zones observed in this work. In these experiments, the observed rate in the experiment (kobs) is related to the scan-rate independent plateau current (ic) observed in zone KS and zone KD by Equation 1, where n is the number of electrons transferred at the electrode, and ip and ν are the peak current and scan rate for a reversible, dark experiment. Notably, this equation does not require any knowledge of the diffusion coefficient or concentration of the redox probe because the currents are taken as a ratio. A plot of ic/ip against the inverse square root of scan rate for several dark scans yielded a straight line with a slope related to kobs and constants only (Figure S4). The forward rate, kobs, is a direct reporter on the rate of effective charge extraction and is distinct from values obtained spectroscopically. kobs is plotted against the redox potential of the charge-accepting probes in Figure 3b. For a plot against the estimated driving force for electron transfer, see SI Figure S5. ## Uncertainty in [QD*] results in uncertain intrinsic rate The intrinsic rate constant, kPCT, is related to kobs by Equation 2. It is experimentally challenging to determine the concentration of charge donors in the system, [QD*], especially given that these charge donors may be electrons from excitons, conduction band electrons in photodoped QDs, or reduced surface traps. The simplest starting hypothesis is that [QD*] is approximately equal to the analytical concentration of QDs, [QD]0. In this assumption, each QD has one conduction band electron that is available for charge transfer. Others have shown that while multiexcitation is possible, 43 the maximum average number of excess conduction band electrons is about one per QD. 34,36,44,45 If we estimate that each QD has exactly one conduction band electron ready for electron transfer, then [QD*] =[QD]0 = 1.1 × 10 -5 M and kPCT is on the order of 10 4 M -1 s -1 . Though estimating [QD*] = [QD]0 has solid conceptual backing, this value cannot explain the data with a simple ErCi' mechanism. The CVs taken during illumination pass from zone D to KD to KS (Figure 2a, b). In electrocatalysis literature, zones KS and KD are observed when operating under conditions of no substrate consumption due to large excess of substrate compared to the concentration of catalyst. By analogy, this implies that zones KS and KD should only be observed when QD* is not consumed by the charge transfer reaction. This could occur when either QD* is in excess compared to the molecular probe M or when QD* is regenerated once an electron is transferred from QD* to M + , effectively making [QD*] constant despite being small. The amount QD* is in excess compared to M is quantified by the dimensionless parameter γ, defined in Equation 3. With only the two reactions in the ErCi' mechanism, zone KS should only be observed when log(γ) >1. 2gives observed rates on the order of 0.1 s -1 (Figure S7). During illuminated studies, regeneration of QD* makes good sense; after electron transfer the QD can be re-excited. However, when illumination is stopped, there cannot be any photoinduced regeneration of QD* at the electrode, but the CVs are still in zone KD (Figure 2c). To explain the experimental data then, [QD*] could be several orders of magnitude larger than [QD]0, so that [QD*] is not greatly changed after charge is transferred to M + . For example, if we let log 4). In this method, the modeled intrinsic rate constants kPCT are on the order of 1 M -1 s -1 and multiplying by [QD*] again gives observed rates on the order of 0.1 s -1 . These observed rates are comparable to those quantified by the direct mathematical calculation from the plateau and peak currents (Figure 3b). The two methods of modeling the data give nearly the same simulated CVs in addition to well-matching the experiment (Figure 4). We are pleased to report that electrochemical modeling was an effective method of determination of the observed rate because it adds generality to our method. In these experiments, only zones D, KD, and KS were observed, but in other systems reaching these zones may be experimentally constrained, precluding the use of the direct mathematical determination of the rate. Using both mathematical determination of charge transfer as well as electrochemical modeling, kobs was determined for the range of electron accepting probes. When comparing the mathematical determination and the modeling results (Figure 3), kobs was generally comparable. Unsurprisingly, with larger driving force, kobs monotonically increases in both methods of determination. This observation is well supported by existing QD literature, wherein the Marcus inverted region is never observed and photoinduced charge transfer from quantum dots is better explained by other rationalizations. 9,11,46,47 While others have demonstrated a similar relationship between driving force and rate of charge transfer, 9,11,48,49 we were uniquely able to measure this through CV. We have demonstrated that the driving force for photoinduced charge transfer is the critical factor controlling kobs rather than chemical identity. FcCOOH and FcCOCH3 have nearly the same E 0 but have different chemical interactions with solvent, electrolyte, and the QD ligand shell. Despite these differences, the kobs values for these two redox probes are nearly identical. Therefore, the differences between these redox probes are due to different rates of the pseudo-unimolecular photoinduced charge transfer elementary step (which is directly controlled by the driving force) rather than chemical interactions with the QD. This observation contrasts with studies where the charge acceptor was bound to the quantum dot through a head group, and the identity of this head group controlled the rate of photoinduced charge transfer by controlling the binding equilibrium to the QD surface. 15 The estimated kPCT values are on the order of 1M -1 s -1 for the model with high [QD*] and without regeneration and are on the order of 10 4 M -1 s -1 for the model with low [QD*] and regeneration. As a benchmark, the diffusion-controlled rate constant (kdiff, the rate assuming every collision results in a charge transferred) is estimated by the Smoluchowski equation (Eq. 4), where RQD and RM are the radii of the QD and molecular charge acceptor, respectively, and DQD and DM are the diffusion coefficients (see SI for details). 50 Importantly, kdiff can be directly compared to the result from this work, as both describe bimolecular processes with the same units. Then, kdiff ~ 10 10 M -1 s -1 is at least six orders of magnitude larger than kPCT determined in this work. This implies that productive photoinduced charge transfer is a rare event in these experiments: for one million collisions, less thanone charge is effectively transferred to the charge acceptor. We believe the low kPCT helps explain common observations that photocatalytic reactions suffer from extremely poor quantum yield. 18 We attribute the small kPCT to the extremely weak electronic coupling between the inorganic QD core and M in solution. Either charges must tunnel through the ligand shell to reach M in solution or M must bury itself in the ligand shell to get better electronic overlap. 51 𝑘 𝑑𝑖𝑓𝑓 = 4𝜋𝑁 1000 (𝑅 𝑄𝐷 + 𝑅 𝑀 )(𝐷 𝑄𝐷 + 𝐷 𝑀 ) = 2 × 10 10 (𝑀 s) −1 Eq. 4 Further, we can compare the observed rate constant (kobs) to reported turn over frequencies (TOF) for homogeneous catalysts. 23 In this context, kobs describes the moles of electrons transferred from QD to redox probe, per unit time per mole of the oxidized redox probe in the diffusion layer. Then, the maximum TOF for the electron acceptors in this work is just the observed rate and is on the order of 0.1 s -1 . In comparison, the well-known nitrogenase enzyme, which reduces N2 to NH3, was measured electrochemically to have an electron transfer TOF of 14 s -1 . 52 Similarly, we can compare to photocatalytic systems. In an iridium photocatalytic system tuned for CO2 reduction, the highest observed TOF was 0.006 s -1 . 53 In a CdSe QD photocatalytic system tuned for C-O bond cleavage, the TOF was 1.7 s -1 . 17 These benchmarks place observed photoinduced electron transfer from QDs faster than reductive photocatalysis in a molecular system, slower than an enzymatic reduction, and about on par with a QD photocatalysis system. ## Net hole transfer to CoCp2 To expand the utility of this method, we considered a probe with lower E 0 : cobaltocenium (CoCp2 + ). In illuminated CV experiments with this redox probe, the oxidative current decreases and the reductive current increases in a consistent with the ErCi' mechanism, indicating that there is effective photoinduced hole extraction from the QD to the probe (Figure 5). We are particularly excited by this result because it demonstrates that our method for measuring charge transfer can be generalized to hole transfer as well as electron transfer. This is in contrast with spectroscopic characterization, where electron and hole dynamics are difficult to isolate. In the CoCp2 + solution with QDs, after illumination is begun the CV distorts over several minutes as described above, then the CVs stop changing (Figure 5a). Similarly, when illumination is stopped, the CVs take several minutes before overlaying with the trace before illumination (Figure 5b). This indicates that, as in the case of electron transfer, the hole-donating species forms over several minutes under illumination before equilibration, and some of these hole-donating species are long-lived. We propose that this long-lived hole-donating species is the hole trap that is populated during the n-type photodoping process and that is slowly depopulated when a conduction band electron recombines with localized holes. Trap mediated hole transfer to molecules has previously been demonstrated in similar QD systems. 11,48 In the same manner as the electron acceptor series, the rate of photoinduced hole transfer to CoCp2 was determined mathematically and through electrochemical modeling. Both methods require knowledge of the concentration of hole-donors, which we estimate is equal to the concentration of the QDs. The mathematical method gives kPCT of 1.38×10 4 M -1 s -1 and the modeling method with [QD*] = 0.11M gives 1.17×10 4 M -1 s -1 . Both results are slower than the slowest kPCT in the electron transfer series. This is in good agreement with prior observations that in reductive photocatalysis, hole quenching rather than electron transfer to cocatalyst is rate limiting. 20,55 Uniquely, we are able to easily disentangle hole transfer dynamics from electron transfer by directly monitoring either oxidation or reduction of the molecular probe. ## Conclusion In this work, cyclic voltammetry has been used for the first time to quantify the rate of photoinduced charge transfer in solution. By carefully designing the photoelectrochemical cell and solvent/electrolyte combination, we were able to simultaneously irradiate and take CV data, generating dynamics that could be readily described by a two-reaction ErCi' mechanism. This technique is a powerful tool for screening photocatalytic systems by directly measuring the effective rate of charge extraction from a photosensitizer. By varying the redox potential of molecular charge acceptors, both net electron and hole transfer from photodoped colloidal quantum dots were observed. Using this technique, we were able to reproduce spectroscopic observation that the rate of photoinduced electron transfer from QDs increases monotonically with driving force. This method is especially compelling because it directly probes the changing oxidation state of the charge acceptor, in contrast with many other techniques that focus on the photophysics of the photosensitizer. The resulting observed rates of charge transfer, on the order of 0.1s -1 , are distinct from the spectroscopically measured picosecond dynamics, and report on the rate of generation of charge separated states relevant to photocatalysis.

chemsum

{"title": "Photoinduced charge transfer from quantum dots measured by cyclic voltammetry", "journal": "ChemRxiv"}

photoredox_ketone_catalysis_for_the_direct_c–h_imidation_and_acyloxylation_of_arenes

## Abstract: The photoexcited aryl ketone-catalyzed C-H imidation of arenes and heteroarenes is reported. Using 3,6dimethoxy-9H-thioxanthen-9-one as a catalyst in combination with a bench-stable imidating reagent, C-N bond formation proceeds with high efficiency and a broad substrate scope. A key part of this method is that the thioxanthone catalyst acts as an excited-state reductant, thus establishing an oxidative quenching cycle for radical aromatic substitution. The synthetic potential of this photoexcited ketone catalysis is further demonstrated by application to the direct C-H acyloxylation of arenes. Scheme 1 (a) Modes of photoexcited aryl ketone catalysis. (b) Photoexcited aryl ketone-catalyzed C-H imidation of arenes (Phth ¼ phthaloyl). (c) Proposed catalytic cycle. ## Introduction Since its inception, the photochemistry of carbonyl compounds, especially ketones, has been studied extensively, and the electronically excited state of ketones is known to undergo different types of bond scission and reformation depending on the reaction conditions. 1 In addition to their own structural reorganizations and transformations, a series of aryl ketones, such as benzophenone and its derivatives, act as effective photosensitizers. 2 Upon exposure to light, they are excited to a singlet state, and subsequent rapid transition to a triplet state through intersystem crossing (ISC) proceeds almost quantitatively. Owing to their relatively long lifetimes, aryl ketone triplets have long been appreciated for their capability to facilitate photochemical reactions. However, their actual usage as catalysts in selective organic synthesis has been rather limited (Scheme 1a). In particular, while the ability of photoexcited aryl ketones to mediate triplet energy transfer (EnT) 3 and hydrogen atom transfer (HAT) 4 has been exploited in several reaction systems, the utility of their photoinduced electron transfer (PET) reactivity in catalysis remains largely underexplored. 5 This is rather intriguing as the simple aryl ketones offer a unique opportunity to tune the redox properties for a given transformation through elaboration and ready modifcation of the primary ketone frameworks. In this context, and in consideration of the prevailing mode of photocatalysis with the currently available organic chromophores, 6 we became interested in exploring the potential of aryl ketones as photoredox catalysts, specifcally as excited-state reductants, in synthetically valuable bond-forming reactions. As an initial step, we disclose herein the efficient catalysis of appropriately modifed thioxanthones under photoirradiation for the direct C-H imidation of arenes and heteroarenes (Scheme 1b). The applicability of thioxanthone catalysis to the C-H acyloxylation of arenes is also demonstrated. Aromatic and heteroaromatic amines constitute the core structural components of a wide array of functional organic molecules. 7 Accordingly, the development of reliable methods for the assembly of arylamines has been a subject of central importance in synthetic chemistry, and direct arene C-H aminations have emerged as powerful means for this pursuit. 8,9 Among the various strategies developed to date, the photocatalytic system reported by Sanford is unique, 9a wherein a key nitrogen-based radical was generated from N-acyloxyphthalimide through one-electron reduction by an iridiumcentred photosensitizer under visible light irradiation. This mechanistic proposal, in addition to the inherent synthetic value of C-N bond formation in its own right, inspired us to choose this class of C-H amination as a testing ground for photoredox ketone catalysis. We envisaged that if the triplet excited state of an aryl ketone could donate an electron to the aminating reagent, the corresponding anion radical would form with concomitant generation of a ketone cation radical. The anion radical then fragments to generate a requisite phthalimidyl radical that participates in the radical aromatic substitution process (Scheme 1c). At this stage, we recognized that aryl ketones are generally poor reductants, and distinguishing between the energy and electron transfer pathways may also be challenging. 10 Nevertheless, we reasoned that the use of aryl ketones with appropriate structural features in combination with an electronically modulated N-acyloxyphthalimide would enable the establishment of an oxidative quenching cycle, thereby allowing the photoredox ketone-catalyzed C-H imidation of arenes. ## Results and discussion At the outset of our investigation to assess the validity of this hypothesis, we selected benzotrifluoride (1a) as a model substrate with the expectation that if a sufficient level of reactivity was attained with this generally less reactive arene, we could demonstrate the advantages of our approach through the reaction development (Table 1). An initial experiment was thus conducted by stirring a mixture of 3,5-bis(trifluoromethyl)phenylacyloxyphthalimide (2), 11 1a (10 equiv.) and a catalytic amount of benzophenone (I) (5 mol%) in acetonitrile (CH 3 CN) under 365 nm LED light irradiation (1500 W m 2 ) at ambient temperature for 15 h. However, 1 H-NMR analysis of the crude material showed very low conversion. Subsequent attempts with benzophenone derivatives, such as II and III, revealed that an electron-rich catalyst exhibited better efficacy, while changing the ketone skeleton to fluorenone (IV) turned out to be ineffective. To further evaluate the relationship between the structure and activity of aryl ketone catalysts, we examined the reaction in the presence of thioxanthone (V), which is known to have a long-lived triplet excited state, and observed an improved reactivity profle. 12 On the other hand, the use of structurally related xanthone (VI) and 10-benzylacridin-9(10H)-one (VII) resulted in lower conversions. We next pursued the structural modifcation of the thioxanthone framework by introducing an electron-donating group to the 3-position, which had a notable yet benefcial impact on the catalytic activity. Under the influence of 3-methyl and 3-dimethylamino-substituted VIII and IX as catalysts, C-N bond formation occurred with signifcantly higher efficiency and the imidated product 3a was isolated in good yields. Interestingly, 3-methoxy derivative X exerted even higher catalytic activity. These observations led us to prepare 3,6-dimethoxy-9H-thioxanthen-9-one (XI) and we found that it delivered a critical improvement in the reactivity. Eventually, by increasing the loading of XI to 10 mol%, this imidation of the electron-defcient arene 1a proceeded smoothly to afford 3a in near quantitative yield (98%). Meanwhile, we screened other representative N-acyloxyphthalimides with different leaving abilities of the carboxylate anion as imidating agents; however, 2 remained optimal. 13 It is also worth adding that the C-H imidation relied on the intensity of 365 nm LED, as the reaction under irradiation with an intensity of 500 W m 2 exhibited lower conversion (75%), whereas full conversion was observed with an intensity of 1000 W m 2 and 1500 W m 2 . 13 The optimal catalyst and reaction conditions were applied to probe the scope of this photoexcited ketone-catalyzed C-H imidation protocol (Table 2). As summarized in Table 2, a broad range of arenes and heteroarenes underwent imidations in good to high yields under the catalysis of XI. The reactivity profle depended on the electronic nature of the arenes. The present system accommodated simple arenes, heteroarenes and electron-rich arenes, and the use of 5 mol% of XI was sufficient for smooth reactions. The imidations of electron-defcient arenes were generally challenging; however, a satisfactory level of reactivity could be attained by increasing the loading of XI to 10 mol%. It should be noted that the observed site selectivity is analogous to that anticipated for a radical aromatic substitution reaction. 9b,14 Moreover, reactions with the arene as the limiting reagent also appeared feasible under slightly modifed Having grasped the general applicability, we then studied the reaction mechanism with the primary objective of distinguishing the presumed electron transfer (ET) pathway from the possible alternative that involves energy transfer (EnT) from the triplet excited state of XI to the imidating agent 2, followed by homolytic cleavage of the N-O bond. This mechanistic study was initiated by measuring the UV-visible spectra of the representative catalysts, V and XI, and 2 in CH 3 CN, which revealed that only the catalyst has an absorption in the range of 365 nm. We then performed a reaction with benzo-trifluoride (1a) under the optimized conditions but with light irradiation at fxed intervals, and observed that the reaction proceeded only when irradiated. 13 We also detected a low quantum yield (F ¼ 0.036) for the imidation. 13 These results not only confrmed that photoexcitation was essential but also suggested the limited intervention of a radical chain process. 15 Another useful piece of information to ascertain the involvement of the triplet excited state of the ketone catalyst was that the reaction was signifcantly suppressed by triplet quenchers (O 2 , pyridazine and 2,5-dimethylhexa-2,4-diene) 16a (Scheme 2a). Unlike reactions that proceed through EnT, this imidation reaction depended heavily on the solvent, and substantial product formation was observed only in CH 3 CN, a general characteristic of reactions involving ET processes (Scheme 2b). 16 Furthermore, the DG et for XI was calculated to be 4.61 kcal mol 1 by the Rehm-Weller equation, indicating the feasibility of ET from XI to 2. 13 At the same time, however, we recognized that these observations were still circumstantial, and thus, more compelling evidence was obtained by determining the triplet excited state energies and redox potentials of ketone catalysts XI, I and V, and the imidating agent 2, by measuring phosphorescence spectra and using cyclic voltammetry as well as theoretical calculations (Scheme 3). 13 As illustrated in Scheme 3a and b, the triplet excited state of 2, 3 *, has an energy of 74.26 kcal mol 1 , whereas those of the ketones, 3 [ketone]*, lie at much lower energy levels ( 3[XI]* ¼ 69.74 kcal mol 1 , 3 [I]* ¼ 69.24 kcal mol 1 and 3 [V]* ¼ 64.53 kcal mol 1 ). This energy gap between 3 [ketone]* and 3 * (4.5 kcal mol 1 even for XI) is signifcant enough to preclude the possibility of the EnT pathway. In addition, even if the triplet excited state energy is a critical element for the ketone catalyst to be able to mediate the imidation, the nearly equal energy levels of 3 [XI]* and 3 [I]* could not rationalize the experimentally observed considerable difference in reactivity between I and XI. On the other hand, comparison of the triplet excited state oxidation potentials (E T ox ) of the ketones and the reduction potential (E 1/2 ) of 2 strongly supported the operation of the ET pathway. Among XI, V and I with E T ox values of 1.32 V, 1.12 V and 0.61 V, respectively, versus SCE, XI should be the most competent electron donor to 2 (E red 1/2 ¼ 1.12 V vs. SCE), followed by V and I (Scheme 3b and c), which is in accordance with the experimental results. The outcomes of these investigations prove that the photoexcited ketone-catalyzed direct arene imidation proceeds through an ET pathway, meaning that aryl ketones with suitable electronic properties, such as the optimal catalyst XI, act as excited-state reductants to establish an oxidative quenching cycle for radical aromatic substitution, as we initially postulated (Scheme 1b). The catalytic cycle commences with photoexcitation and subsequent ISC to afford 3 [ketone]*. The ketone triplet with an appropriate oxidation potential donates an electron to the imidating agent, 2, to form a ketone cation radical ([ketone]c + ) 17 and an anion radical of 2 ( c ) that undergoes fragmentation to generate a phthalimidyl radical (PhthNc) and a 3,5-bis(trifluoromethyl)benzoate anion. The PhthNc adds to the arene to bring forth a neutral radical species that can be oxidized by [ketone]c + to provide a Wheland intermediate and regenerate the ketone catalyst. Deprotonation of the Wheland intermediate by the 3,5-bis(trifluoromethyl)benzoate anion yields 3 and the corresponding carboxylic acid. 19 It is noteworthy that the absence of the kinetic isotope effect rules out the possibility of C-H abstraction as a rate-determining step (Scheme 2c). After establishing the C-H imidation of arenes, we decided to further explore the synthetic potential of photoexcited ketone catalysis and found it to be applicable to the C-H acyloxylation of simple arenes. 20 For instance, light irradiation (325 W m 2 ) over a mixture of pentafluorobenzoyl peroxide (4) and benzene (10 equiv.) in CH 3 CN/DCE (1 : 1) in the presence of 3-methoxy-9H-thioxanthen-9-one (X) (10 mol%) at room temperature for 15 h resulted in the formation of the acyloxylated product 5a in good yield (Table 3). 21 Other selected examples listed in Table 3 show the tolerance of the present system to the electronic property of arenes. ## Conclusions In conclusion, we have developed a photoexcited ketonecatalyzed C-H imidation of arenes. Under simple and mild conditions, direct C-N bond formation proceeds efficiently with a broad range of arenes and heteroarenes. A distinct feature of this novel photocatalytic system is that the thioxanthonederived catalyst behaves as an excited-state one-electron reductant and thus establishes an oxidative quenching cycle, as verifed unambiguously through mechanistic investigations based on experimental and theoretical approaches. The utility of photoexcited ketone catalysis has also been demonstrated by application to the direct C-H acyloxylation of arenes. We believe that the present study indicates the possibility of designing and structurally manipulating simple aryl ketones to explore their potential utility as photoredox catalysts, which would be

chemsum

{"title": "Photoredox ketone catalysis for the direct C\u2013H imidation and acyloxylation of arenes", "journal": "Royal Society of Chemistry (RSC)"}

carbene_catalyzed_umpolung_of_α,β-enals:_a_reactivity_study_of_diamino_dienols_vs._azolium_enolates,

## Abstract: Since their discovery by Bode and Glorius in 2004, N-heterocyclic carbene catalyzed conjugate umpolung reactions of a,b-enals have been postulated to involve the formation of diamino dienols ("homoenolates") and/or azolium enolates ("enolates"), typically followed by addition to electrophiles, e.g. Michael-acceptors.In this article, we provide evidence, for the first time, for the postulated individual and specific reactivity patterns of diamino dienols (g-C-C-bond formation) vs. azolium enolates (b-C-C-bond formation). Our study is based on the pre-formation of well defined diamino dienols and azolium enolates, and the in situ NMR monitoring of their reactivities towards enone electrophiles. Additionally, reaction intermediates were isolated and characterized, inter alia by X-ray crystallography.Scheme 1 Early intermediates in the NHC-catalyzed umpolung of a,b-unsaturated aldehydes. ## Introduction In N-heterocyclic carbene (NHC) organocatalysis, 1 the "conjugate umpolung" of a,b-unsaturated aldehydes is a most thriving and proliferative feld. As schematically shown in Scheme 1, interaction of an a,b-enal (a 3 ) with an NHC frst generates a Breslow-type 2 intermediate, the diamino dienol I. A subsequent proton shift from the diamino dienol's -OH to Cg leads to the azolium enolate II. The diamino dienol I carries a partial negative charge on Cg, and therefore represents a homoenolate equivalent (d 3 ). On the other hand, the azolium enolate II is nucleophilic at Cb, and therefore behaves as an enolate equivalent (d 2 ). Numerous experimental studies have revealed that the homoenolate vs. enolate behaviour of a,b-enals, when exposed to NHCs, can be influenced by the type of catalyst employed, and by the reaction conditions. 3,4 For example, homoenolate chemistry is favoured by imidazolium precatalysts, in combination with strong bases. 3,4 Reactions proceeding via the homoenolate pathway have been used to provide g-lactones, 5 spiro-lactones, 6 spiro-bis-lactones, 7 bicyclic lactones, 8 g-lactams, 9 bicyclic b-lactams, 10 cyclopentenes, 5c,11 and saturated esters. 12 Enolate chemistry, on the other hand, is favoured by triazolium precatalysts in combination with weak bases. 3,4 Azolium enolates have been generated by the combination of NHCs with ketenes, 13 aldehydes, 3a,14 and esters. 15 Reactions proceeding via the azolium enolate pathway have been used to provide b-lactams, 13b,c b-lactones, 13d,e unsaturated d-lactams, 14b,f,15b,c and unsaturated d-lactones. 14a,14e-g,15c As outlined in Scheme 1, it is generally believed that diamino dienols I and the tautomeric azolium enolates II are the starting points of divergent reaction pathways, leading to different (isomeric) products when exposed to one and the same electrophilic reaction partner. This divergent reactivity is interpreted in the sense that diamino dienols I add electrophiles at Cg, whereas the tautomeric azolium enolates II react at Cb. In stark contrast to their pivotal importance in a,b-enal umpolung, no investigations of the reaction modes of pre-formed diamino dienols I and azolium enolates II (i.e. C-C bond formation with C-electrophiles at Cb vs. Cg) appear to have been reported to date. 16 Several azolium enolates II are described in the literature. However, they were accessed by addition of carbenes to ketenes, 16,17 and not by reaction of a,b-unsaturated aldehydes with N-heterocyclic carbenes (NHCs). With this in mind, we set out to investigate the reactivity patterns of pre-formed diamino dienols I and azolium enolates II with enone Michael acceptors. The frst successful generation of both diamino dienols I and azolium enolates II from a,b-unsaturated aldehydes and carbenes, and their characterization by NMR and X-ray, was recently reported by our group. 18 ## Results and discussion Reactivity studies of diamino dienols Cyclopentene formation with enones. In 2006, Nair et al. reported that the NHC-catalyzed reaction of cinnamic aldehydes with enones affords 1,3,4-trisubstituted cyclopentenes. 11a As schematically shown in Scheme 2, this transformation was interpreted by homoenolate addition to the Michael acceptor, giving rise to the intermediate III. 19 Aldol ring closure leads to intermediate IV. From there, the b-lactone V is formed, with concomitant regeneration of the NHC catalyst. Decarboxylation of the b-lactone V fnally gives the cyclopentene product VI. We had reported earlier 18 that under strictly oxygen-free conditions, the saturated imidazolidinylidene SIPr (1,3-bis[2,6di-(2-propylphenyl)]imidazolidin-2-ylidene) reacts smoothly with E-cinnamic aldehyde in THF at room temperature to the diamino dienol 1 (Scheme 3). Protonation of the latter exclusively gives the Cg-protonation product 2 (an azolium enol), and thus nicely proves Cg-nucleophilicity (Scheme 3, top). When the pre-formed and stable diamino dienol 1 was exposed to an equimolar amount of methyl-E-4-oxo-2-pentenoate 3a (Scheme 3, middle) under 1 In the same manner, we exposed the diamino dienol 1 to an equimolar amount of ethyl E-3-benzoylacrylate (3b-Et). Again, NMR monitoring revealed the instantaneous disappearance of diamino dienol 1, with concomitant formation of the corresponding Michael product 4b-Et (Scheme 3, middle; see ESI † for the full 1D and 2D NMR characterization of 4b-Et). In addition, crystallization of this Michael product 4b-Et and of its methyl analogue, 4b-Me [obtained from methyl 3-benzoylacrylate (3b-Me)], was achieved from benzene and THF solution, respectively, by slow addition of n-hexane at room temperature, and under strictly anaerobic conditions. The X-ray crystal structures of the azolium enolates 4b-Et and 4b-Me are shown in Fig. 2. First of all, the X-ray structures provide unambiguous proof for the formation and the constitution of the Michael addition products 4b-Et/Me. Furthermore, they nicely reveal the almost orthogonal arrangement of the imidazolium ring and the enolate moiety, as evidenced by the dihedral angles O-C5-C2-N1 ¼ 44.5( 4 When the diamino dienol 1 was exposed to E-chalcone (3c) in an analogous manner, the slow formation of the Michael addition product 4c was observed (Scheme 3, middle; ca. 80% conversion at room temperature after ca. 12 h; see ESI † for full NMR characterization of 4c). In summary, in all four cases studied (diamino dienol 1 + enones 3a, 3b-Et/Me, 3c), C-C bond formation had indeed occured at C-g, of the diamino dienol and gave the azolium enolate intermediates 4a, 4b-Et/Me and 4c postulated for cyclopentene formation. 11 The further conversion of the azolium enolate intermediates such as 4a, 4b-Et/Me, and 4c is typically formulated as an aldol addition of the enolate to the ketone moiety, followed by b-lactone formation and decarboxylation (vide supra, Scheme 2). Note that intermediate azolium enolates such as 4a, 4b-Et/Me and 4c en route to b-lactones/ cyclopentenes had not been observed before. By employing the saturated NHC SIPr, we achieved sufficient stabilization of these intermediates such that the subsequent intramolecular aldol addition to the 5-membered carbocycles does not occur spontaneously at room temperature. However, as studied exemplarily with the Michael addition adducts 4b-Et and 4c, heating to 80 C for 12 h in THF or toluene indeed resulted in the formation of the expected cyclopentene derivatives 5b-Et and 5c, along with the disappearance of the starting azolium enolates 4b-Et,c (Scheme 4; see ESI † for NMR spectra). g-Butyrolactone formation with aldehydes. Diamino dienols I have been postulated as intermediates in g-butyrolactone (VII) formation from enals and aldehydes (Scheme 5, top). 5 Fig. 2 Top: X-ray crystal structure of the Michael product 4b-Et, obtained from the addition of ethyl E-3-benzoylacrylate (3b-Et) to the diamino dienol 1; bottom: X-ray crystal structure of the Michael product 4b-Me obtained from diamino dienol 1 and methyl E-3-benzoylacrylate (3b-Me). Exposition of the diamino dienol 1 to benzaldehyde ( 6) in THF at 70 C indeed resulted in a slow conversion (ca. 50% after 24 h) to the saturated lactone 7 (trans : cis 3.3 : 1; Scheme 5, bottom). The most characteristic 1 H NMR signals of 7 are a doublet at d ¼ 5.44 ppm [ 3 J HH ¼ 9.0 Hz, 1H, H4 (trans)] and a doublet at d ¼ 5.85 ppm [ 3 J HH ¼ 6.6 Hz, 1H, H4 (cis)]. In line with our earlier experience, 18 the liberated NHC SIPr reacted with benzaldehyde to cleanly afford the diamino enol 8 (see ESI † for the NMR identifcation of lactone 7 and diamino enol 8). ## Reactivity studies of azolium enolates Formation of g,d-unsaturated d-lactones with enones. As discussed above, the conversion of a,b-unsaturated aldehydes to cyclopentenes VI proceeds via initial diamino dienol formation and subsequent reaction of the latter with an enone electrophile (Scheme 2). In contrast, the conversion of a,b-unsaturated aldehydes with enones to g,d-unsaturated dlactones VIII (i.e. same starting materials, but different products) is assumed to involve additional tautomerization of the diamino dienol I to an azolium enolate II (see Scheme 1). The latter then reacts with the enone Michael acceptor, ultimately affording the g,d-unsaturated d-lactone VIII (Scheme 6). For studying the reactivity of preformed azolium enolates, we chose the two stable representatives 11a and 11b shown in Fig. 3 (top). Upon addition of n-hexenal (9a) to SIPr in THF-[D 8 ] at room temperature, we observed the instantaneous disappearance of the aldehyde signal characteristic of 9a, and the appearance of diamino dienol 10a, as evidenced by a doublet at d ¼ 5.32 ppm ( 3 J HH ¼ 12.0 Hz, 1H, H18), a multiplet at d ¼ 4.71-4.66 ppm (1H, H19) and singlet at d ¼ 3.40 ppm (OH). At room temperature, the diamino dienol 10a tautomerized to the azolium enolate 11a within ca. 20 min. 20 The latter shows a characteristic 1 H NMR triplet at d ¼ 3.46 ppm ( 3 J HH ¼ 7.0 Hz, 1H, H18), and a multiplet at d ¼ 1.82-1.78 ppm, (2H, H19). Indicative 13 C NMR resonances are those of C2 and C18, appearing at d ¼ 172.5 ppm and 100.5 ppm, respectively (see ESI † for further NMR data of 11a). In a similar manner, when we exposed E-5-phenylpent-2enal (9b) to SIPr, 1 H NMR monitoring frst revealed the instantaneous formation of the diamino dienol 10b, characterized by a doublet at d ¼ 5.39 ppm ( 3 J HH ¼ 14.9 Hz, H18), a multiplet at d ¼ 4.80-4.75 ppm (H19), and a singlet at d ¼ 3.42 ppm (OH) (see ESI † for further NMR data of 11b). After 10 min, the formation of the azolium enolate 11b was noticeable, and its concentration increased over time (Fig. 3, bottom). The azolium enolate 11b is characterized by a 1 H NMR triplet at d ¼ 3.55 ppm ( 3 J HH ¼ 7.1 Hz, 1H, H18), and a multiplet at d ¼ 1.89-1.86 ppm (2H, H19). In the 13 C NMR spectrum, the formation of 11b is evidenced by the characteristic signals of C2 and C18, appearing at d ¼ 172.4, 99.4 ppm respectively (see ESI † for further NMR data of 11b). Note that in an earlier report from our laboratory, we had observed that diamino dienols derived from enals with additional conjugation (e.g. E-cinnamic aldehyde, sorbic aldehyde) do not undergo tautomerization to azolium enolates. 18b Tautomerization occurs only in the absence of this additional conjugative stabilization of the diamino dienol state, for example with E-hexenal (9a) and E-5-phenylpent-2-enal (9b) as reported here, or with E-crotonic aldehyde as substrate aldehyde. 18b When we added E-chalcone (3c) to the pre-formed azolium enolate 11b, the concentrations of both 11b and 3c decreased simultaneously over time (Fig. 4), along with the appearance of the unsaturated d-lactone 12b (trans : cis 5. ## Conclusion We have reported (i) the selective generation and characterization of a number of hitherto postulated diamino dienol and azolium enolate reaction intermediates, by interaction of the Nheterocyclic carbene SIPr with various a,b-unsaturated aldehydes. (ii) The homoenolate and enolate equivalents thus prepared were stable enough for NMR-spectroscopic characterization, but still reactive enough for further transformations when exposed to electrophilic reaction partners: exposure of diamino dienols to Michael acceptors gave hitherto postulated addition products stable enough for NMR and even X-ray crystallographic characterization. Heating of the latter completed the reaction cycle, affording trisubstituted cyclopentenes. (iii) In the same manner, the postulated reaction of diamino dienol intermediates with aldehydes to g-butyrolactones could be verifed experimentally. (iv) The tautomerization of primarily formed diamino dienols to azolium enolates, the postulated precursors of g,d-unsaturated d-lactones, was monitored by 1 H NMR in two cases. Subsequent exposure of the azolium enolates to E-chalcone as Michael acceptor indeed gave the corresponding g,d-unsaturated d-lactones, thus proving the postulated C-C bond formation at Cb of the azolium enolate intermediate. We are convinced that the mechanistic information disclosed herein will promote the understanding of other existing NHC-catalyzed transformations, and the design of novel ones.

chemsum

{"title": "Carbene catalyzed umpolung of \u03b1,\u03b2-enals: a reactivity study of diamino dienols vs. azolium enolates, and the characterization of advanced reaction intermediates", "journal": "Royal Society of Chemistry (RSC)"}

three-component_stereoselective_enzymatic_synthesis_of_amino_diols_and_amino-polyols

## Abstract: Amino polyols represent attractive chemical building blocks but can be challenging to synthesize because of the high density of asymmetric functionalities and the need for extensive protection group strategies. Here we present a three-component strategy for the stereoselective enzymatic synthesis of amino diols and amino polyols using a diverse set of prochiral aldehydes, hydroxy ketones and amines as starting materials. We were able to combine biocatalytic aldol reactions, using variants of D-fructose-6-phosphate aldolase (FSA), with reductive aminations catalyzed by IRED-259, identified from a metagenomic library. A two-step process, without the need for intermediate isolation, was developed to avoid cross-reactivity of the carbonyl components. Stereoselective formation of the (R),(R),(R) enantiomers of amino polyols was observed and confirmed by x-ray crystallography. ## Main Chiral polyhydroxylated amines are important pharmacophores that occupy a distinct chemical space in terms of LogP, molecular weight, hydrogen bonding and other physicochemical properties . Amino diols and amino polyols are present in the backbone of many bioactive compounds including Pactamycin (antitumor), Myriocin (antibiotic), the proteasome inhibitor TMC-95A and iminosugars , many of which are potent glycosidase inhibitors, including Miglustat and Miglitol which are in clinic development. These iminosugars have been the targets for many elegant chemical syntheses but require lengthy routes due to protection strategies, all of which limits scale-up and wider industrial application. Biocatalysis has proven to be an effective alternative to traditional synthetic techniques for the preparation of chiral amines due to the mild conditions under which enzymes typically operate, whilst providing excellent regio-, chemo-and stereoselectivity . Transaminases (TAs) , amine dehydrogenases (AmDHs) , imine reductases (IREDs) and amine oxidases (MAOs) have all been utilized for the preparation of enantiopure amines,which prompted us to investigate their use for amino polyol formation. Retrosynthetic analysis suggested using IREDs for the final step on 2,3 dihydroxy ketones, which in turn can be generated by aldolasecatalyzed biotransformations from aldehydes and hydroxy-ketones (Figure 1). This strategy would present a number of challenges to overcome. Firstly, any unreacted aldehyde or ketone from the initial aldol-catalyzed reaction is also a potential substrate for the IRED, resulting in side-product formation. Secondly, the aldolase catalyzed reaction is reversible, which precluded a simple overall 'one-pot' process . Thirdly, there was no precedent for using IREDs on multifunctional polar dihydroxy ketones. A further concern was that the dihydroxy ketone intermediates might be prone to epimerization. Finally, the diol intermediates and products are highly polar and hence extraction from water would be challenging. Keeping these issues in mind, for the first step in the proposed cascade the D-fructose-6-phosphate-aldolase (FSA) from Escherichia coli was chosen, since it readily accepts dihydroxyacetone (DHA) analogues as donors whilst maintaining complete control over the stereoselectivity of the reaction [12, . Wildtype FSA (WT) and variants A129S, A165G and A129S/A165 were screened for initial aldol addition using aldehydes 1 -6 with donor molecules hydroxyacetone (HA, a), dihydroxyacetone (DHA, b) or hydroxybutanone (HB, c) to form aldol adducts 1a -6c (Table 1). The initial screen revealed conversion for all FSA variants with all aldehyde substrates and variants that gave the highest conversion with each aldehyde and donor combination were chosen for preparative scale reactions (total reaction volume = 60 mL) with isolated yields for compounds 1a -6c ranging from 14 -86%. (Supporting information). Analysis of the purified products by 1 H NMR and chromatography showed formation of a single diastereomer, assigned as the (3S,4R) configuration with a and b and the (4S,5R) configuration with c. For products 1a -2c optical rotation values agreed with literature data (Supporting information). ## Table 1. Screening of FSA-catalyzed aldol additions Reaction conditions: 100 mM aldehyde, 100 mM HA/ 100 mM DHA/ 100 mM HB, 2 mg ml -1 FSA variant, 20% vol/vol DMSO, 100 mM TEA buffer (pH 8), reaction volume 500 μL, 25 o C, 250 rpm, 24 h. For the next reductive amination step, a metagenomic library of 384 IREDs was screened using aldol adducts 1a, 1b and 1c and cyclopropylamine i as substrates. Conversion was determined by HPLC using chemically synthesized amine products 1ai, 1bi and 1ci, as mixtures of diastereomers for standards. The positive hits obtained from these IRED plates are shown in Table 2. Pleasingly, 12 positive hits were obtained for substrate 1a with up Table 3. Screening of amine donor (i-vi) scope for IR-259 catalyzed reductive amination of aldol adducts 1a -6c. Reaction conditions: 10 mM aldol adduct, 200 mM cyclopropylamine, 10 mg ml -1 IRED lysate, 0.25 mg ml -1 CDX-901 GDH, 50 mM glucose, 0.5 mM NADP + , 10% DMSO, 100 mM TEA buffer (pH 8), 50 μL Reaction volume, 30 o C, 700 rpm, 24 h. Reaction conditions were then optimized for the IR-259 catalyzed transformation of 1a to 1ai (Supporting Information, Table S4, Figure S2). Amine loading, IRED concentration and substrate loading were varied whilst GDH, glucose and NADP + concentrations were kept constant. High amine concentrations yielded the highest conversions with > 90% conversion when 200 mM cyclopropylamine was used with 5 mM substrate and 2 mg IRED mL -1 . Good yields were observed at 25 mM 1a concentration (conversion dropping from 75% to 32% with 50 mM 1a and 500 mM cyclopropylamine). Overall conversion improved with increasing IR-259 enzyme concentration, with the highest level achieved at 10 mg mL -1 . Conversions were found to be optimal at pH 8. Reaction conditions were also optimized for the conversion of 1a to 1aiii using allylamine (iii) as the amine donor (Supporting information, Table S5 and Figure S3) leading to 39% conversion. To test the feasibility of the cascade in one-pot, conversions of 1 to 1ai, 1bi and 1ci respectively were investigated. Aldol adduct formation reached steady state within 1 -2 h and was then consumed by either the reverse aldol reaction or reductive amination with cyclopropylamine. The subsequent reductive amination of the aldol adducts (1a, 1b, 1c) appeared slower and was further limited by unwanted reductive amination of 1 forming side product 7 (Supporting Information, Figure S4). To overcome this side product formation, while minimizing isolation of intermediates, a two-step approach was investigated. In the step 1, the FSA components were added to the reaction mixture and after steady state formation of the aldol product (6 h) the FSA was removed by filtering with centrifugation using a vivaspin with a 10,000 MW cut-off, thereby preventing the reverse aldol reaction occurring. In step 2, the crude filtrate was treated with cyclopropylamine, IR-259, CDX-901 GDH, glucose and NADP + . This two-step approach improved the formation of the desired products (1ai -6ci), over side product 7, whilst still avoiding the need to purify intermediates (1a -6c). This two-step, three-component protocol was then applied to aldehydes 1 -6 (Table 4). Aldehyde concentration was kept constant (10 mM) while cyclopropylamine concentrations were varied (100 mM, 200 mM, and 500 mM). In general, conversions ranged from (30 -88%) over both steps. Using the optimized conditions, preparative scale 50 mL reactions for the synthesis of 1ai, 1bi, 1ci, 2ai, 2bi and 2ci were carried out using the previously established two-step approach (Supporting information). High conversions were recorded for all reactions (85 -94%). Product 1bi was recrystallized from MeOH and single X-ray crystallography data was recorded (CCDC Deposition number: 2154768, Supporting information) demonstrating an absolute stereochemistry of (R,R,R). This suggested that the stereochemistry of the aldol product had been conserved and that the previously uncharacterized IR-259 is R selective. Reaction conditions: 20 mM aldehyde, 20 mM HA/ DHA/ HB, 2 mg ml -1 FSA variant lysate, 20 % vol/vol DMSO, 100 mM TEA buffer (pH 8), 500 μL reaction volume, 250 rpm, 30 o C, 6 h. The FSA was then removed by centrifugation (13,000 rpm, 10 min) and the supernatant filtered by Vivaspin (MWCO 10,000) by centrifugation (4,000 rpm, 10 min). The supernatant was then diluted with the second step components to final concentration of ~10 mM aldol adducts, 100 mM/ 200 mM/ 500 mM cyclopropylamine, 10 mg ml -1 IR-259 lysate, 0.25 mg ml -1 CDX-901 GDH, 1 mM NADP + , 50 mM glucose, 100 mM TEA buffer (pH 8), 10% vol/vol DMSO, 1 mL reaction volume, 30 o C, 200 rpm, 24 h. All products (1ai, 1bi, 1ci, 2ai, 2bi and 2ci) were fully characterized by NMR spectroscopy. UPLC chromatograms of products were compared to diastereomeric mixtures generated by chemical reduction (rather than enzymatic) of the intermediate dihydroxy ketones and confirmed that only one diastereomer was formed by each IRED-catalyzed which was assumed to be the (R,R,R) product, based on the previous assignment of 1bi. Given the unusual substrate scope of IR-259, molecular docking studies were performed to better understand substrate specificity. An IR-259 homology model was created based on the reported AspRedAm (PDB 5G6R). Product 1bi was used as the ligand for the docking simulation (Supporting Information, Figure S5). The active site of the model was highly conserved with AspRedAm containing residues D174, Y182, W212 previously thought to have a role in catalysis. S237 and S99 are non-conserved and therefore may be responsible for assisting substrate binding through formation of hydrogen bonds. The IR-259 active site contains a hydrophobic pocket consisting of I178, F185, M181, F223, W212, A219 and I127 which presumably assists in the binding of the phenyl ring of the substrate. Moreover, the docking pose positions the electrophilic carbon of the amine product 4.3 Ã from the C4 of the NADP(H) cofactor for optimal hydride transfer. In conclusion, we have established a two-step three component reaction that converts simple precursors to dihydroxy amine products containing three asymmetric centres with high stereoselectivity. Key to the success was the identification of IREDs by screening of a metagenomic panel, which identified an imine reductase (IRED-259) able to tolerate hydroxylated substrates and provide high (R)-selectivity. ## Materials & Equipment Commercially available chemicals used were purchased from Sigma-Aldrich (Poole, Dorset, UK), Acros Organics (Loughborough, UK), Alfa Aesar (Heysham, Lancashire, UK), Fluorochem (Hadfield, Derbyshire, UK) and used without further purification. Commercially available 384 metagenomic IRED panel was purchased from Prozomix (Building 4, West End Ind. Estate, Haltwhistle, UK) A Bruker Avance 400 spectrometer was used to record NMR spectra with chemical shifts reported in ppm relative to tetramethylsilane (TMS). Coupling constants (J) are reported in Hz. High-resolution mass spectrometry (HRMS) was recorded using a Waters LCT time-of-flight mass spectrometer, connected to a Waters Alliance LC (Waters, Milford, MA, USA). Reverse Phase HPLC was carried out on an Agilent 1260 Infinity II Series system equipped with a G1379A degasser, G1312A binary pump, G1367A well plate auto sampler, G1316A temperature-controlled compartment and a diode array detector. A Waters C18, 5 μ m, 4.6 x 250 mm column was used as a stationary phase (Massachusetts, USA). Reverse phase UPLC was recorded using a Waters Acquity UPLC H-class-QDA System is equipped with a quaternary solvent manager (QSM), autosampler, photodiode array (PDA) detector and single quadrupole QDa mass detector. A Waters Acquity UPLC HSS C18, 1.8 μm, 2.1 x 100 mm column was used as a stationary phase. Reverse Phase colorimetric IREDY2GO screen was carried out using previously reported methodology. Analytical Methods ## Biocatalyst Production Chemically competent E. coli were transformed with DNA plasmid vectors containing the genes for the relevant enzyme and grown on LB agar plates containing 30 mg mL -1 antibiotic ( ## Lysate Preparation Cell pellets were re-suspended in KPi (100mM, pH 7), submerged in an ice bath and lysed by ultrasonication (60 sec ON, 90 sec OFF, 40% amplitude, 4 -6 cycles using a Soniprep 150 (MSE UK Ltd.). The solution was clarified by ultracentrifugation at 18,000 rpm at 4 o C for 60 min. The clarified supernatant was filtered through a cellulose membrane (0.45 μm) into a 50 mL falcon tube and snap froze using liquid N2 before freeze drying on a Heto-PowerDry LL1500 (ThermoScientific). The resulting lyophilized powder was stored at -20 o C until further use. ## Purification Lyophilized powder of the desired enzyme was resuspended in 10% buffer B (100 mM KPi, 300 mM NaCl, 30 mM imidazole, pH 7) and loaded onto a His-Trap Crude FF column (GE Healthcare) charged with 0.1 M nickel sulphate, preequilibrated with 10% buffer B. The column was washed with 15 -25 mL of 10 % buffer B and then 15 -25 mL of 20 % buffer B (100 mM KPi, 300 mM NaCl, 60 mM imidazole, pH 7). The His-tagged protein was then eluted with 100 % buffer B (100 mM KPi, 300 mM NaCl, 300 mM imidazole, pH 7) and 1 -2 mL fractions were collected. Protein concentration was determined using a thermofisher NanoDrop TM microvolume spectrophotometer and the fractions containing pure protein were combined. The pure purified was added to a membrane 30,000 MWCO centrifugal concentrator (Sartoius, UK) to remove excess imidazole. Buffer A (100 mM NaPi buffer, 300 mM NaCl, pH 7.5) was added to the samples until desired concentration was achieved. The pure protein was divided into 1 mL aliquots, snap frozen and stored at -80 o C until further use. ## Investigation into the substrate scope of FSA to perform aldol addition on aldehyde substrates A panel of six aldehydes were chosen for testing with FSA variants (WT, A129S, A165G and A129S/A165G) to determine their ability to form aldol products with three separate donors: hydroxyacetone (HA, a), dihydroxyacetone (DHA, b) and hydroxybutanone (HB, c). Aldehyde 1 and 2 is an already known substrate and 3 -6 were chosen based on structure similarity. ## General Procedure for analytical scale biocatalytic aldol addition of donor molecules with aldehydes Analytical scale reactions were carried out in 1.5 mL Eppendorf tubes with a reaction volume of 500 μL. Each reaction contained 100 mM aldehyde substrate, 100 mM aldol donor (a, b or c), 2 mg ml -1 FSA variant, 100 mM triethanolamine (TEA) buffer pH 7 and 20% v/v DMSO. Reactions were shaken at 250 rpm at 25 o C for 24 hours before a 3 μL biotransformation sample was taken and diluted with 297 μL MeOH/0.1% HCl (50:50) to achieve a final concentration of approximately 1 mM. The sample was centrifuged for 10 min at 14,000 rpm and added to a Thompson filter vial for UPLC analysis. General Procedure for preparative scale biocatalytic aldol addition of donor molecules with aldehydes. To a 50 mL falcon tube the components for the FSA reactions were added (same conditions used in previous section) with a total reaction volume of 20 mL. Each reaction was setup in triplicate. For each unique aldehyde/donor combination the FSA which gave the highest conversion in the initial screen was chosen. The reactions were incubated at 25 o C at 250 rpm for 24 hr. The reactions were filtered through Celite® to remove protein and then extracted with EtOAc (3 x 80 mL). The organic layer was washed with Brine (10 mL) dried over anhydrous MgSO4 and concentrated by rotary evaporation. The crude residue was resuspended in minimal EtOAc and purified by flash chromatography (hexane/ethyl acetate 95:5 à 5:95 20 min). Isolated yields are presented in Table S1. ## Identification of IRED to perform the reductive amination of aldol products IREDy-to-go (Reverse-direction screening) A previously described colorimetric screen was used to identify an IRED capable of mediating reductive amination with aldol product 1a and cyclopropylamine i to form product 1ai. The assay is based on running the reaction in the reverse direction using the amine product of interest as a substrate which is oxidized to the corresponding imine. This is coupled with a reaction forming a red compound formazan which is formed from the NADPH dependent oxidation of diaphorase. The formation of formazan can be monitored by a plate reader at 490 nm which gives an indication of activity of the reductive amination process. ## Protocol: A 384 well plate was used to carry out the assay. Each well contained lyophilized IRED lysate and IREDy-to-go components (NADP + and Diaphorase). 50 μL of the assay master mix (0.125 mg ml -1 INT, 10 mM amine in Tris-HCl buffer pH 9) was added to each well, changing tips in between each addition to minimize contamination between enzymes. The plate was centrifuged at 1,500 rpm for 1 minute. A plate reading measurement was taken as a 0 hr measurement and then the plate was covered in foil and incubated at 30 o C for 24 hr. Another plate reading was taken at 24 hr. A blank plate was also run under the same conditions, but without the amine to eliminate false positives. (Figure S1) The IREDy-to-go yielded a lot of results for amine 1ai, but when tested in the forward direction many of the hits from the assay showed no activity. We hypothesized that the hydroxyl groups present on our substrate may interfere with the screen as there may be endogenous alcohol dehydrogenases within the IRED lysate. : Top: picture of diaphorase reaction coupled to reductive aminase oxidation, resulting in production of red formazan dye. Bottom: Results of IREDY-2GO screen for compound 1ai after 24 h incubation. False Positives often reported at B02, D08, F12, H21, I02, I07, K16, O01, O08 and P17. ## General procedure for screening IRED panel (Forward direction) As an alternative, forward direction reductive aminations were setup for substrates 1a, 1b and 1c with cyclopropylamine as an amine donor. To a 96 well plate, each well contained 10 mM aldol substrate, 200 mM cyclopropylamine, 0.25 mg ml -1 CDX-801 GDH, 50 mM glucose, 0.5 mM NADP + , 5 mg ml -1 IRED, 10 % vol/vol DMSO and 100 mM TEA buffer pH 8 to a total reaction volume of 100 μL. Samples were placed on a plate shaker at 30 o C, 700 rpm for 24 hr. 20 μL of the biotransformations filtered with a 96-well filter plate and centrifuged at 4,000 rpm for 10 min. The filtered biotransformation was then diluted with 180 μL MeOH: 0.1 % HCl (50:50) and run on reverse phase HPLC directly from the plate. Results presented below in Table S2. ## Intensification of reductive amination of 1a with cyclopropylamine and IR-259 Taking the biotransformation of 1a to 1ai and 1a to 1aiii as a model reaction, components were varied to investigate which components have the largest influence over conversion. ## Investigation into a one-pot FSA-IRED cascade for the reductive amination of aldol compounds The transformation of aldehyde 1 to desired amine product 1ai was tested first to determine the feasibility of the cascade in one-pot. Reactions were performed in a 500 μL reaction mixture containing 10mM aldehyde, 10mM aldol donor (a, b or c), 2mg ml -1 FSA variant lysate, 200mM amine, 2mgml ## Investigation into a two-step sequential addition FSA-IRED cascade for the reductive amination of aldol compounds The first part of the reactions was performed in a 500 μL reaction mixture containing 20mM aldehyde, 20mM aldol donor (a, b or c), 2mg ml -1 FSA variant lysate and 20% vol/vol DMSO in TEA pH 8 buffer. Reactions were incubated at 30 o C at 200rpm for 6 -8hr. Reactions were then filtered at 4,000rpm for 10min using a 10,000 MWco Vivaspin pre-washed with water. The flow through was then diluted up to a total reaction volume of 1mL so that the components from part one was diluted by a factor of 2 and now also containing 100/200/500mM amine, 10mgml ## Investigation into the potential of GDH catalyzed reductive amination of aldehyde or aldol substrates To investigate the possible interference in the reductive amination step of the cascades of possible imine reduction activity of the CDX-901 GDH enzyme used to recycle glucose, control reactions were set up in which aldehyde or aldol substrates were incubated with GDH lysate and stoichiometric NADPH. Reactions contained 10mM Aldehyde or Aldol, 200mM amine, 10mM NADPH, 0.25 mg mL -1 GDH, 10% (v/v) DMSO, 100mM pH 8 TEA buffer, with a reaction volume of 500 μL that were incubated at 30 o C with shaking at 200rpm for 24hr. The first part of the reactions was performed in a 500 μL reaction mixture containing 20mM aldehyde, 20mM aldol donor (a, b or c), 2mg ml -1 FSA variant lysate and 20% vol/vol DMSO in TEA pH 8 buffer. Reactions were incubated at 30 o C at 200rpm for 6 -8hr. Reactions were then filtered at 4,000rpm for 10min using a 10,000 Mwco Vivaspin pre-washed with water. The flow through was then diluted up to a total reaction volume of 1mL so that the components from part one was diluted by a factor of 2 and now also containing 100/200/500mM amine, 10mgml -1 IR-259, 1mM NADP + , 0.25mg ml -1 CDX-801 GDH, 50mM glucose, 10% vol/vol dmso and 100mM TEA pH 8 buffer in a 1.5mL Eppendorf tube. Reactions were incubated at 30 o C with shaking at 200rpm for 24h. 30μL of reaction sample was quenched with MeOH:0.1%HCl (50:50), centrifuged at 14,000rpm for 10 min and added to a Thompson filter vial for UPLC analysis. ## Preparative scale sequential FSA-IR-259 cascade 20 mM of aldehyde, 20 mM HA/ DHA/ HB, 2 mg ml -1 FSA variant lysate, 20 % vol/vol DMSO, 100 mM TEA buffer (pH 8) was added to a 50 mL Falcon tube (25 mL reaction volume) and placed in an orbital shaker at 250 rpm, 30 o C for 6 h. The FSA were then removed by centrifugation (13,000 rpm, 10 min) and the supernatant filtered by Vivaspin (MWCO 10,000) by centrifugation (4,000 rpm, 10 min). The supernatant was then diluted with the second step components to final concentration of ~10 mM Aldol product, 100 mM/ 200 mM/ 500 mM cyclopropylamine, 10 mg ml -1 IR-259 lysate, 0.25 mg ml -1 CDX-901 GDH, 1 mM NADP + , 50 mM glucose, 100 mM TEA buffer (pH 8), 10% vol/vol DMSO, 50 mL reaction volume, 30 o C, 200 rpm, 24 h. ## Procedures & Characterization data of preparative biocatalytic formation of Aldol products from aldehydes with donor molecules (3S,4R)-5-(benzyloxy)-3,4-dihydroxypentan-2-one (1a) To a 100mL conical flask, 2-(benzyloxy)acetaldehyde (6 mmol), hydroxyacetone (6 mmol), DMSO (12 mL) and 100 mM triethanolamine buffer pH 8 (47 mL) was added. 120 mg of FSA WT lysate was added, and the flask was sealed with a foam bung and incubated at 30 o C in an orbital shaker at 250rpm for 24 hr. The enzyme was precipitated by the addition of EtOAc (20 mL) and removed by filtration through Celite®. The reaction mixture was extracted with ethyl acetate (3 x 30 mL), dried over anhydrous MgSO4 and the solvent evaporated under vacuum. The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a colorless gel (769mg, 57%). The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 2-phenylacetaldehyde (6mmol), hydroxyacetone (6 mmol), DMSO (12 mL) and 120mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a white solid (439mg, 37%). ## OH The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 2-phenylpropanal (6mmol), hydroxyacetone (6 mmol), DMSO (12 mL) and 120mg of FSA A165G/A129S lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (449mg, 36%) with minor impurities. ## (3S,4R)-3,4-dihydroxy-6-phenylhexan-2-one (4a) The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 3-phenylpropanal (6mmol), hydroxyacetone (6 mmol), DMSO (12 mL) and 120mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (825mg, 66%). ## (3S,4R)-6-(benzyloxy)-3,4-dihydroxyhexan-2-one (5a) The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 3-benzoxypropanal (6mmol), hydroxyacetone (6 mmol), DMSO (12 mL) and 120mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (1229mg, 86%). 1H NMR (400 MHz, MeOD) δ 7.41 -7.24 (m, 5H), 4.57 -4.46 (m, 2H), 4.21 (td, J = 6.8, 2.3 Hz, 1H), 4.14 -4.03 (m, 1H), 3.73 -3.58 (m, 2H), 2.23 (s, 3H), 1.92 (q, J = 6.4 Hz, 2H). The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: phenoxyacetaldehyde (4 mmol), hydroxyacetone (4 mmol), DMSO (8 mL) and 120mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (32 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (235mg, 28%). (3S,4R)-5-(benzyloxy)-1,3,4-trihydroxypentan-2-one (1b) The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 2-(benzyloxy)acetaldehyde (6 mmol), dihydroxyacetone (6 mmol), DMSO (12 mL) and 120mg of FSA A129S lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (302mg, 21%). ## (3S,4R)-1,3,4-trihydroxy-5-phenylpentan-2-one (2b) The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: phenylacetaldehyde (6 mmol), dihydroxyacetone (6 mmol), DMSO (12 mL) and 120mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (264mg, 21%). ## (3S,4R)-1,3,4-trihydroxy-5-phenylhexan-2-one (3b) The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 2-2-phenylpropanal (6 mmol), dihydroxyacetone (6 mmol), DMSO (12 mL) and 120mg of FSA A165G lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (228mg, 17%). The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 2-phenylpropanal (6 mmol), dihydroxyacetone (6 mmol), DMSO (12 mL) and 120mg of FSA A165G lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (484mg, 36%). ## (3S,4R)-6-(benzyloxy)-1,3,4-trihydroxyhexan-2-one (5b) The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 3-benzoxypropanal (6 mmol), dihydroxyacetone (6 mmol), DMSO (12 mL) and 120mg of FSA A129S lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (579mg, 38%). ## (3S,4R)-1,3,4-trihydroxy-5-phenoxypentan-2-one (6b) The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: phenoxyacetaldehyde (6 mmol), dihydroxyacetone (6 mmol), DMSO (12 mL) and 120mg of FSA A129S lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (651mg, 48%). The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 2-benzoxyacetaldehyde(6 mmol), hydroxybutanone (6 mmol), DMSO (12 mL) and 120mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (629mg, 44%). ## (4S,5R)-4,5-dihydroxy-6-phenylhexan-3-one (2c) The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: phenylacetaldehyde (6 mmol), hydroxybutanone (6 mmol), DMSO (12 mL) and 120mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (312mg, 25%). ## (4S,5R)-4,5-dihydroxy-6-phenylheptan-3-one (3c) The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 2-phenylpropanal (6 mmol), hydroxybutanone (6 mmol), DMSO (12 mL) and 120mg of FSA A165G lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (280mg, 21%). The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 3-phenylpropanal (6 mmol), hydroxybutanone (6 mmol), DMSO (12 mL) and 120mg of FSA A165G lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (586 mg, 44%). ## (4S,5R)-7-(benzyloxy)-4,5-dihydroxyheptan-3-one (5c) The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: 3-benzoxyacetaldehyde (6 mmol), hydroxybutanone (6 mmol), DMSO (12 mL) and 120mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (680mg, 45%). ## (4S,5R)-4,5-dihydroxy-6-phenoxyhexan-3-one (6c) The title compound was obtained by the methodology described above for 1a. The amounts of reactants used were the following: phenoxyacetaldehyde (6 mmol), hydroxybutanone (6 mmol), DMSO (12 mL) and 120mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (47 mL). The crude product was purified by flash chromatography on silica gel (EtOAc/Cyclohexane from 1:4 to 1:0) to yield a clear oil (752mg, 56%). ## Procedure & Characterization Data for Preparative biocatalytic reductive amination of amines with aldol products To a 50 mL falcon tube, 20mM aldehyde, 20mM aldol donor (a, b or c), 2mg ml -1 FSA variant lysate and 20% vol/vol DMSO in TEA pH 8 buffer was added so that the total reaction volume was 25 mL. Reactions were incubated at 30 o C at 200rpm for 6 -8hr. Reactions were then filtered at 4,000rpm for 10min using a 10,000 MWCO Vivaspin pre-washed with water. The flow through was then diluted up to a total reaction volume of 50 mL so that the components from part one were diluted by a factor of 2 and now also containing 200 or 500mM amine, 10mgml -1 IR-259, 1mM NADP + , 0.25mg ml -1 CDX-801 GDH, 50mM glucose, 10% vol/vol DMSO and 100mM TEA pH 8 buffer. Reactions were incubated at 30 o C with shaking at 200rpm for 24h. The reactions were quenched with 50 mL MeOH and centrifuged at 4,000 rpm for 5 min and the supernatant was filtered through Celite®. The solution was concentrated by rotary evaporation and purified by reverse-phase flash column using C18 cartridge (H2O:MeOH 95:5 to 20:80) and concentrated under reduced pressure to yield the desired amine product. ## (2R,3R,4R)-1-(benzyloxy)-4-(cyclopropylamino)pentane-2,3-diol (1ai) The title compound was obtained by the methodology described above for 1ai. The amounts of reactants used were the following: 2-benzoxyacetaldehyde (0.5 mmol), hydroxyacetone (0.5 mmol), DMSO (5 mL) and 50mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (20 mL). After filtration the supernatant was diluted to 50 mL with 200mM cyclopropylamine, 10mgml -1 IR-259, 1mM NADP + , 0.25mg ml -1 CDX-801 GDH, 50mM glucose and 100mM TEA pH 8 buffer. The crude product was purified by reverse-phase flash column using C18 cartridge (H2O:MeOH 95:5 to 20:80) to yield an orange/brown oil (31.2mg, 24%). ## (2R,3R,4R)-5-(benzyloxy)-2-(cyclopropylamino)pentane-1,3,4-triol (1bi) The title compound was obtained by the methodology described above for 1ai. The amounts of reactants used were the following: 2-benzoxyacetaldehyde (0.5 mmol), dihydroxyacetone (0.5 mmol), DMSO (5 mL) and 50mg of FSA A129S lysate dissolved in 100 mM triethanolamine buffer pH 8 (20 mL). After filtration the supernatant was diluted to 50 mL with 200mM cyclopropylamine, 10mgml -1 IR-259, 1mM NADP + , 0.25mg ml -1 CDX-801 GDH, 50mM glucose and 100mM TEA pH 8 buffer. The crude product was purified by reverse-phase flash column using C18 cartridge (H2O:MeOH 95:5 to 20:80) to yield an amorphous brown solid was then recrystallized in MeOH to form pale brown single crystals (22.8mg, 16%). ## (2R,3R,4R)-1-(benzyloxy)-4-(cyclopropylamino)hexane-2,3-diol (1ci) The title compound was obtained by the methodology described above for 1ai. The amounts of reactants used were the following: 2-benzoxyacetaldehyde (0.5 mmol), hydroxybutanone (0.5 mmol), DMSO (5 mL) and 50mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (20 mL). After filtration the supernatant was diluted to 50 mL with 200mM cyclopropylamine, 10mgml -1 IR-259, 1mM NADP + , 0.25mg ml -1 CDX-801 GDH, 50mM glucose and 100mM TEA pH 8 buffer. The crude product was purified by reverse-phase flash column using C18 cartridge (H2O:MeOH 95:5 to 20:80) to yield a Brown oil (139mg, 30%). ## (2R,3R,4R)-4-(cyclopropylamino)-1-phenylpentane-2,3-diol (2ai) The title compound was obtained by the methodology described above for 1ai. The amounts of reactants used were the following: 2-phenylacetaldehyde (0.5 mmol), hydroxyacetone (0.5 mmol), DMSO (5 mL) and 50mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (20 mL). After filtration the supernatant was diluted to 50 mL with 200mM cyclopropylamine, 10mgml -1 IR-259, 1mM NADP + , 0.25mg ml -1 CDX-801 GDH, 50mM glucose and 100mM TEA pH 8 buffer. The crude product was purified by reverse-phase flash column using C18 cartridge (H2O:MeOH 95:5 to 20:80) to yield a brown oil (117.6mg, 12%). ## (2R,3R,4R)-2-(cyclopropylamino)-5-phenylpentane-1,3,4-triol (2bi) The title compound was obtained by the methodology described above for 1ai. The amounts of reactants used were the following: 2-phenylacetaldehyde (0.5 mmol), dihydroxyacetone (0.5 mmol), DMSO (5 mL) and 50mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (20 mL). After filtration the supernatant was diluted to 50 mL with 200mM cyclopropylamine, 10mgml -1 IR-259, 1mM NADP + , 0.25mg ml -1 CDX-801 GDH, 50mM glucose and 100mM TEA pH 8 buffer. The crude product was purified by reverse-phase flash column using C18 cartridge (H2O: The title compound was obtained by the methodology described above for 1ai. The amounts of reactants used were the following: 2-phenylacetaldehyde (0.5 mmol), hydroxybutanone (0.5 mmol), DMSO (5 mL) and 50mg of FSA WT lysate dissolved in 100 mM triethanolamine buffer pH 8 (20 mL). After filtration the supernatant was diluted to 50 mL with 200mM cyclopropylamine, 10mgml -1 IR-259, 1mM NADP + , 0.25mg ml -1 CDX-801 GDH, 50mM glucose and 100mM TEA pH 8 buffer. The crude product was purified by reverse-phase flash column using C18 cartridge (H2O:MeOH 95:5 to 20:80) to yield an orange/brown oil (17mg, 14%). ## 2-phenoxyacetaldehyde(6) Title compound was prepared using a method previously reported in literature. Sodium periodate (0.65 M in water, 20 mL) was added to a vigorously stirred suspension of silica gel (20 g) in dichloromethane (160 mL), followed by a solution of 3-phenoxy-l,2-propanediol (1.68 g, 10.0 mmol) in dichloromethane (20 mL). After stirring for 10 min the mixture was filtered and the filtrate was concentrated to give the title compound as a clear liquid (1.36 g,100 %). ## General procedure for reductive amination for the synthesis of amine standards To a stirring solution of THF (20mL) at room temperature, aldehyde (2.0 mmol) and amine (2.5mmol) was added. Glacial acetic acid ( 114 μL, 2.0mmol) and sodium triacetoxyborohydride (0.636 g, 3.0 mmol) were added sequentially and the solution was stirred for 16 hours. The reaction was quenched with saturadted NaHCO3 solution (15mL) and extracted into EtOAc (2 x 10 mL). The organic extracts were washed with HCl solution (1M 3 x 15mL) and the aqueous extracts were then basified to pH 12 with NaOH solution (10M). The combined aqueous layers were then extracted with EtOAc (2 x 20 mL) and then dried over MgSO4. The solvent was removed under rotary evaporation to yield the title compound. The title compound was obtained by the methodology described above. Amount of reactants are as follows: 2benzoxyaldehyde (2 mmol) and cyclopropylamine (2.5 mmol) followed by the general procedure to yield title compound as a brown oil (64 mg, 17% yield). ## N-phenethylcyclopropanamine (8) The title compound was obtained by the methodology described above. Amount of reactants are as follows: phenylacetaldehyde (2 mmol) and cyclopropylamine (2.5 mmol) followed by the general procedure to yield title compound as a brown oil (116 mg, 36% yield) with minor impurities (contaminated with starting material). ## N-(2-phenylpropyl)cyclopropanamine (9) The title compound was obtained by the methodology described above. Amount of reactants are as follows: 2phenylpropanal (2 mmol) and cyclopropylamine (2.5 mmol) followed by the general procedure to yield title compound as a yellow liquid (134 mg, 38% yield). N-(3-phenylpropyl)cyclopropanamine (10) The title compound was obtained by the methodology described above. Amount of reactants are as follows: 3phenylpropanal (2 mmol) and cyclopropylamine (2.5 mmol) followed by the general procedure to yield title compound as a yellow liquid (160 mg, 45% yield). ## N H O N H N H ## N-(3-(benzyloxy)propyl)cyclopropanamine (11) The title compound was obtained by the methodology described above. Amount of reactants are as follows: 3-(benzyloxy)propanal (2 mmol) and cyclopropylamine (2.5 mmol) followed by the general procedure to yield title compound as a yellow liquid (218 mg, 53% yield). N-(2-phenoxyethyl)cyclopropanamine (12) The title compound was obtained by the methodology described above. Amount of reactants are as follows: 2phenoxyacetaldehyde (2 mmol) and cyclopropylamine (2.5 mmol) followed by the general procedure to yield title compound as a yellow liquid (241 mg, 68% yield). ## Alternate procedure for reductive amination for the synthesis of amine standards Aldol product formed and purified from scaled up FSA reactions (2 mmol) was dissolved in methanol (20 mL) followed by the addition of glacial acetic acid (2mmol), amine (4 mmol) and sodium triacetoxyborohydride (3 mmol) added portion wise. The mixture was stirred for 16 h and the crude solution was loaded onto an SCX cartridge. The cartridge was washed with methanol (3 x 20 mL) followed by elution with 7N ammonina in methanol. The solvent was removed by rotary evaporation to yield the title compound. ## (3R,4R)-1-(benzyloxy)-4-(cyclopropylamino)pentane-2,3-diol (1ai) The title compound was obtained as a mixture of diastereomers by the methodology described above. Amount of reactants are as follows: Aldol product 1a (1 mmol) and cyclopropylamine (1.5 mmol) followed by the general procedure to yield title compound as a brown solid (140mg, 53% yield). ## HN The title compound was obtained as a mixture of diastereomers by the methodology described above. Amount of reactants are as follows: Aldol product 1b (1 mmol) and cyclopropylamine (1.5 mmol) followed by the general procedure to yield title compound as a brown solid (109mg, 28% yield). ## (3R,4R)-1-(benzyloxy)-4-(cyclopropylamino)hexane-2,3-diol (1ci) The title compound was obtained as a mixture of diastereomers by the methodology described above. Amount of reactants are as follows: Aldol product 1c (1 mmol) and cyclopropylamine (1.5 mmol) followed by the general procedure to yield title compound as a brown solid (50mg, 18% yield). Data for major diasteromer formed labelled. ## (3R,4R)-4-(cyclopropylamino)-1-phenylpentane-2,3-diol (2ai) The title compound was obtained by the methodology described above. Amount of reactants are as follows: Aldol product 2a (1 mmol) and cyclopropylamine (1.5 mmol) followed by the general procedure to yield title compound as a brown solid (48mg, 20% yield) with minor impurities. ## (3R,4R)-2-(cyclopropylamino)-5-phenylpentane-1,3,4-triol (2bi) The title compound was obtained by the methodology described above. Amount of reactants are as follows: Aldol product 2b (1 mmol) and cyclopropylamine (1.5 mmol) followed by the general procedure to yield title compound as a brown solid (40 mg, 16% yield). ## (3R,4R)-4-(cyclopropylamino)-1-phenylhexane-2,3-diol (2ci) The title compound was obtained by the methodology described above. Amount of reactants are as follows: Aldol product 2c (1 mmol) and cyclopropylamine (1.5 mmol) followed by the general procedure to yield title compound as a brown solid (44 mg, 14% yield). ## Determination of absolute stereochemistry of 1bi by X-Ray Crystallography Crystal growth of C15H23NO4: ## Docking Studies Docking was performed using VINA default parameters in the YASARA molecular modelling software program. The AspRedAm crystal structure (5G6R) was used to generate a homology model of IR-259 using SWISS-MODEL with model building paramters set to automated mode. The IR-259 model was then used as a receptor for simulations. The model was prepared for simulations using the YASARA 'clean' script. To find the correct geometry of the receptor in aqueous environment, a unit cell was defined around the whole model which was filled with water molecules and then the structure was energy-minimizedl by running the 'Energy Minimization' script. Product 2ai was considered as the ligand in the simulation experiment using their terminal amine functional group in the unprotonated form. For the docking simulation, a 7 Ã simulation cell around the NADPH co-factor was defined for which reasonable functional binding poses were calculated. From the 20 calculated binding poses for each docking run, the most catalytically reasonable, with the highest docking score (strongest binding) was selected based on the requirement for the C-N functional group to be position close to the C4 of cofactor NADPH. The ligand-receptor complex structure was further processed using PyMOL software, including identification of hydrophobic residues responsible for binding the phenyl ring as well as polar residues can aid in substrate binding as well as catalysis. ## Determination of diastereomeric ratio of preparative scale reactions Following the determination of absolute stereochemistry of product 2ai, it was assumed that the IRED mediated reaction would follow the same mechanism for the other substrates, resulting in (R), (R), (R) stereocenters for all products. Chemical standards synthesized for the preparative scale reaction could form a range of diastereomers. Inspecting the UPLC traces for the chemical preparation, multiple product peaks are shown, suggesting multiple diastereomers have formed, compared to the crude biocatalytic routes which one resulted in one product peak. ## Possible chemical route: Biocatalytic route:

chemsum

{"title": "Three-component stereoselective enzymatic synthesis of amino diols and amino-polyols", "journal": "ChemRxiv"}

carbon_dioxide_utilisation_for_production_of_transport_fuels:_process_and_economic_analysis

## Abstract: Utilising CO 2 as a feedstock for chemicals and fuels could help mitigate climate change and reduce dependence on fossil fuels. For this reason, there is an increasing world-wide interest in carbon capture and utilisation (CCU). As part of a broader project to identify key technical advances required for sustainable CCU, this work considers different process designs, each at a high level of technology readiness and suitable for large-scale conversion of CO 2 into liquid hydrocarbon fuels, using biogas from sewage sludge as a source of CO 2 . The main objective of the paper is to estimate fuel production yields and costs of different CCU process configurations in order to establish whether the production of hydrocarbon fuels from commercially proven technologies is economically viable. Four process concepts are examined, developed and modelled using the process simulation software Aspen Plus s to determine raw materials, energy and utility requirements. Three design cases are based on typical biogas applications: (1) biogas upgrading using a monoethanolamine (MEA) unit to remove CO 2 , (2) combustion of raw biogas in a combined heat and power (CHP) plant and (3) combustion of upgraded biogas in a CHP plant which represents a combination of the first two options. The fourth case examines a post-combustion CO 2 capture and utilisation system where the CO 2 removal unit is placed right after the CHP plant to remove the excess air with the aim of improving the energy efficiency of the plant. All four concepts include conversion of CO 2 to CO via a reverse water-gas-shift reaction process and subsequent conversion to diesel and gasoline via Fischer-Tropsch synthesis. The studied CCU options are compared in terms of liquid fuel yields, energy requirements, energy efficiencies, capital investment and production costs. The overall plant energy efficiency and production costs range from 12-17% and d15.8-29.6 per litre of liquid fuels, respectively. A sensitivity analysis is also carried out to examine the effect of different economic and technical parameters on the production costs of liquid fuels. The results indicate that the production of liquid hydrocarbon fuels using the existing CCU technology is not economically feasible mainly because of the low CO 2 separation and conversion efficiencies as well as the high energy requirements. Therefore, future research in this area should aim at developing novel CCU technologies which should primarily focus on optimising the CO 2 conversion rate and minimising the energy consumption of the plant. Broader contextCarbon capture and utilisation (CCU) has recently become the focus of large scale international attention not only because it has the potential to reduce anthropogenic CO 2 emissions which contribute to climate change but also because it could generate value from waste CO 2 through the synthesis of fuels and chemicals. Globally, consumption of fuels is two orders of magnitude higher than that of chemicals; therefore, CO 2 utilisation technologies should focus primarily on fuel synthesis to create significant economic value and to make a substantial contribution to the reduction of CO 2 emissions. Successful marketentry of CO 2 -to-fuels technologies strongly depends on their economic competitiveness. In this article, a techno-economic assessment of the manufacture of transport hydrocarbon fuels from waste CO 2 is performed through process simulation, cost modelling and sensitivity analysis. Unlike other studies, the present techno-economic assessment only employs the best currently available and proven CCU technologies. The aim is to support policy makers and businesses in their decision-making by establishing whether the production of liquid transport fuels from CO 2 using current technology is economically feasible and identifying the modifications required to improve the economic competitiveness of CCU processes. ## Introduction Carbon capture and storage (CCS) has attracted significant attention in recent years as a possible technology to help mitigate climate change by reducing CO 2 emissions into the atmosphere. 1 However, CCS requires high investment costs and poses risks associated with the need for long-term storage and potential leakage of CO 2 . 2,3 Carbon capture and utilisation (CCU) is being proposed as a complementary technology to CCS with the aim of both reducing CO 2 emissions and consumption of fossil resources by utilising CO 2 as a feedstock for the production of chemicals and fuels. Global CO 2 emissions from fossil fuel combustion were approximately 31 Gt in 2011 9 and are likely to rise to 57 Gt in 2050. 10 Currently, CO 2 is only used for the production of chemicals, such as urea, salicylic acid and polycarbonates. However, in order to make a significant contribution to reducing CO 2 emissions, its utilisation should focus primarily on the conversion to fuels since the market for chemicals is two orders of magnitude lower than that for fuels. 5 Oxygenates and hydrocarbons can be produced via hydrogenation of CO 2 and could offer feasible alternatives for the transportation sector, reducing its dependency on fossil fuels. CO 2 hydrogenation for oxygenate production is at present the most intensively investigated area of CO 2 utilisation with methanol synthesis from CO 2 and H 2 already being demonstrated at bench-and pilot-scale plants in Asia 11,12 and Europe. 13 Conversely, the production of hydrocarbon fuels from syngas (H 2 and CO) produced from CO 2 and H 2 (e.g. via reverse water-gasshift reaction) is yet to be demonstrated. This is mainly due to the fact that the production of hydrocarbons from CO 2 and H 2 requires a higher amount of hydrogen and energy than oxygenates. 5 However, there is a noticeable lack of published techno-economic feasibility studies in this area that could potentially support this argument. It should also be noted that hydrocarbons produced from syngas via Fischer-Tropsch synthesis (the established industrial process for converting syngas to liquid fuels) are specifically attractive because of their unlimited compatibility with conventional fuels in any proportion and thus, unlike alcohols and ethers, can readily be incorporated and integrated with conventional markets and supply chains. As highlighted above, the high hydrogen and energy requirements associated with CO 2 -to-fuels pathways is one of the main issues for the application of these technologies. In order to make such a process economically and environmentally sustainable, hydrogen should be made from a non-fossil resource or produced within the process itself. The latter would also decrease the overall operating costs, especially since fossilderived hydrogen is still significantly cheaper than that produced from renewable technologies. 14 One non-fossil source of hydrogen could be biogas produced by anaerobic digestion of wet waste, such as sewage sludge. Such biogas contains mainly CO 2 and methane, the latter of which can be utilised to produce hydrogen (e.g. via steam reforming), 15 thus avoiding the depletion of natural gas, currently used for hydrogen production. The CO 2 from the biogas can be separated and utilised for the production of fuels. Within the EU alone, around 10 million tonnes (dry basis) of sewage sludge are generated per year and the digestion of each tonne could produce approximately 590 m 3 of methane. 16 Other wastes could result in even higher methane yields; for example, food waste can generate up to 3.5 times more methane per tonne than sewage sludge. 17 Therefore, anaerobic digestion of waste could be an important source of hydrogen for a CO 2 hydrogenation-tofuels process. It could also be a suitable target process for CO 2 utilisation technologies since it requires moderate capital investment. This is the topic of this paper which examines different CO 2 capture and utilisation process concepts for the conversion of sewage sludge to liquid hydrocarbon fuels. The aim of the study is to identify the most promising process configurations in terms of conversion efficiencies and costs. For these purposes, a comprehensive techno-economic assessment has been carried out to examine the technical and economic performance of four conceptual designs, considering only the best available and proven technologies: amine CO 2 capture, steam reforming, reverse watergas-shift (RWGS) process and Fischer-Tropsch synthesis. An overview of the CCU process concepts developed in this work is presented in the next section. Section 3 outlines the methodology for process modelling and the economic assessment followed by the results in Section 4. A sensitivity analysis is carried out in Section 5 which examines the effect of key economic and technical parameters on the production costs of liquid fuels, including capital and energy costs as well as the CO 2 conversion rate in the RWGS reactor. ## Process overview The general CO 2 utilisation system considered in this study allows the production of liquid hydrocarbon fuels from sewage sludge, as shown in Fig. 1. It consists of six sections: anaerobic digestion of sewage sludge, CO 2 capture, heat and power generation, syngas production, conversion of CO 2 to CO and fuel synthesis. This general concept is realised in four different design configurations which are based on typical biogas utilisation applications. These are summarised in Table 1 along with the main process steps. The different process sections and designs are depicted in more detail in Fig. 2-5. The first process design (PD-MEA in Table 1 and Fig. 2) incorporates a monoethanolamine (MEA) gas treatment unit which is often used to upgrade biogas to the same standards as natural gas by removing CO 2 and other contaminants. 18 The second case (PD-CHP1, Table 1 and Fig. 3) is based on another biogas application: combustion of untreated biogas in a combined heat and power (CHP) unit to produce electricity and heat. The third design (PD-CHP2, Fig. 3) comprises an MEA CO 2 capture system placed before the CHP plant which in this case is fed with the upgraded biogas (i.e. more concentrated in CH 4 ) rather than untreated biogas as in the second case. Thus, this is a pre-combustion CO 2 capture system. The final configuration (PD-CHP3, Fig. 5) is similar to PD-CHP2 but the MEA unit is now placed after the CHP plant so that this process design is based on post-combustion CO 2 capture. 7 In this case, the MEA unit also allows the removal of the excess air used in the CHP plant which acts as an inert diluent, decreasing the efficiency of the downstream processes and necessitating higher power ## Paper Energy & Environmental Science consumption for the subsequent syngas compression. The feedstock for all the process concepts is sewage sludge, a by-product of wastewater treatment. In all four cases, the plant produces liquid hydrocarbon fuels (gasoline and diesel). The different process sections and designs are described in more detail below. ## Anaerobic digestion Anaerobic digestion (AD) is a process in which biodegradable organic material is broken down by microorganisms in the absence of air. The produced biogas, consisting of around 65 vol% CH 4 , 35 vol% CO 2 and trace gases such as H 2 S, H 2 , N and NH 3 , is commonly utilised in heat and electricity co-generation plants. 16 In this study, the digestion conditions are mesophilic (35 1C, 1 atm) which are associated with lower equipment costs and energy requirements compared to thermophilic systems (450 1C, 1 atm). 19 This corresponds to 11 111 kg h 1 of sludge fed into the plant. Fig. 1 An overview of the CO 2 utilisation system for production of synthetic fuels [MEA: monoethanolamine. Dashed lines represent steps which are not present in all design options; see Table 1 for details. In some design cases the heat and electricity generation plant is placed before CO 2 capture]. ## Energy & Environmental Science Paper This journal is © The Royal Society of Chemistry 2015 ## CO 2 capture The flow diagram of the CO 2 capture unit is shown in Fig. 6. To convert CO 2 into a liquid fuel, a concentrated stream of CO 2 needs to be generated by isolating it from biogas. Among the available technologies to capture CO 2 from a gas stream, aminebased regenerative systems have been identified as the most suitable technology that has achieved commercial success. 7 In the packed absorption column, the biogas is fed countercurrently with an MEA aqueous solution (usually 15-35 wt%) which reacts with and absorbs CO 2 in the biogas to form an MEA carbamate soluble salt. The gas stream lean in CO 2 is released from the top of the absorber while the MEA solution rich in CO 2 is pumped to a heat exchanger in which the solution is heated to about 120 1C and then fed into the stripping column. MEA is regenerated in the stripper and recycled to the absorber for re-use (lean MEA solution, Fig. 6). The regeneration conditions are maintained by the reboiler which uses low-pressure steam. Steam which acts as stripping gas in the column, is recovered in the condenser and fed back to the stripper, while the concentrated CO 2 stream is released from the top of the stripper for downstream processing. Process conditions in the capture plant are summarised in Table 2. ## Heat and power generation An outline of the combined heat and power (CHP) plant is shown in Fig. 7. Firstly, biogas is compressed from 1 bar to 8 bar. It is then mixed with steam (8 bar) and compressed air and then burned in the combustor to produce hot gas at 583 1C. Steam is used to lower the combustion temperature (below 750 1C) to minimise NO X formation. The hot gas is first passed through a gas turbine for electricity generation and then to the steam generation area to recover heat. In the steam generation area, the gas passes through five heat exchangers and is cooled down by water or steam. Consequently, three different grade steams are generated: low-pressure (LP) steam at 1.013 bar, medium-pressure (MP) steam at 5 bar and high-pressure (HP) steam at 24 bar. ## Syngas production The main process for producing syngas currently used in Fischer-Tropsch synthesis is steam reforming of methane which is a well-understood and proven technology. 22 In this study, the upgraded biogas from the CO 2 removal section is utilised either in a methane steam reformer (PD-MEA concept, Fig. 2) or the co-generation unit (PD-CHP2 concept, Fig. 4). The CH 4 -rich gas stream leaving the MEA absorption column is mixed with steam (2.6 MPa) and the resulting mixture is preheated to 500 1C and introduced to the catalytic reforming reactor. 15 The steam/methane mixture is passed through a set of externally heated reformer tubes filled with nickel catalyst, where it is converted to CO and H 2 at 900 1C and 25 bar according to the following reaction: Although the theoretical molar ratio of steam to methane is 1 : 1, an excess of steam (H 2 O : CH 4 = 1.2 : 1) is used to prevent deactivation of the catalyst from carbon deposition. 15 2.6 CO 2 conversion 2.6.1 Reverse water gas shift (RWGS) process. The CO 2 conversion technology evaluated in this study is a RWGS reaction process based on the CAMERE pilot plant operated by the Korean Institute of Energy and Research (KIER) and Korea Gas Corporation (KOGAS). 12 The CAMERE process produces methanol from CO 2 in two steps: (1) conversion of CO 2 to CO and water in a RWGS reactor and (2) methanol synthesis after an intermediate water removal. Similar to the CAMERE process, the shift reactor in this study is operated over a ZnAl 2 O 4 catalyst at 650 1C and atmospheric pressure with a feed gas mixture of CO and H 2 preheated before the reactor. The basic reaction is shown in eqn (2): An excess of hydrogen (H 2 : CO 2 = 3 : 1) is used to prevent carbon (coke) deposition on the catalyst surface. The feed CO 2 -rich gas is produced by the MEA plant in the process concepts PD-MEA, PD-CHP2 and PD-CHP3, while in PD-CHP1, CO 2 is generated in the CHP plant. 2.6.2 Hydrogen recovery. A pressure swing adsorption (PSA) system is used after the reformer (PD-MEA) or the RWGS reactor (PD-CHP1, PD-CHP2, PD-CHP3) to recover the excess H 2 which is recycled to the RWGS reactor for re-use. PSA is an established industrial process used extensively for gas or liquid separation. In this study, adsorption is operated at 30 1C and high pressure (40 bar) similarly to the Linde PSA technology. 23 The adsorbent is regenerated by lowering the pressure to slightly above atmospheric. An 85% hydrogen recovery is achieved in the PSA unit 24 which produces high purity hydrogen (99.999%). 23 ## Fuel synthesis Fischer-Tropsch (FT) converts a mixture of CO and H 2 (syngas) in the presence of a catalyst to a variety of organic compounds, mainly hydrocarbon products of variable chain length. The FT reactions are highly exothermic and can be represented by the following basic reaction equation: Eqn (3) describes the formation of paraffins which are the main products of FT synthesis. Olefins, oxygenates and aromatic compounds are also produced although in much lower quantities. 25 The water-gas-shift (WGS) reaction also takes place during FT synthesis but it can be reduced to a minimum when using a cobalt catalyst. Conversely, iron catalysts show a significant WGS activity and generally result in lower liquid selectivity than cobalt catalysts. 26 Low-temperature Fischer-Tropsch (LTFT) synthesis is operated at temperatures between 200 and 250 1C which favour the production of liquid fuels up until middle distillates. 27,28 Contrary to high-temperature Fischer-Tropsch synthesis (HTFT) which is operated at 300-350 1C, LTFT synthesis results in lower gas yields to the advantage of higher diesel yields. 28 The FT process is generally operated at pressures ranging from 20-40 bar. 29 Generally, the FT synthesis process should be operated at relatively low temperatures, high operating pressures and H 2 : CO molar ratios of around 2 in order to achieve a high FT liquid production. 30 In this study, the FT reactor is operated at 30 bar and 220 1C and is assumed to be similar to the Sasol Slurry phase distillate reactor which is designed for LTFT synthesis. 26,28 The reactor's gas effluent is passed to a three phase separator to remove water and heavy hydrocarbons from the residual vapour. The FT off-gas which mainly consists of light hydrocarbons (C 1 -C 4 ) and unconverted syngas is combusted to generate low pressure steam for the anaerobic digesters, whereas the liquid fuels are sent to a central refinery plant for further upgrading. As mentioned earlier, a wide range of products are obtained from the FT synthesis, therefore a quantitative approximation of product distribution is necessary. The most widely used approach to tackle this problem is the Anderson-Schulz-Flory (ASF) product distribution. 31 According to this method, the adsorbed carbon chain can either undergo further addition of a -CH 2 -group or the chain can terminate. 31 The ASF-product distribution model is represented by the following equation: where C n is the molar fraction of a hydrocarbon product consisting of n carbon atoms and a the chain growth probability which determines the hydrocarbon product distribution. The chain growth probability is influenced by a number of factors, such as the type and age of catalyst, the H 2 : CO ratio in the feed gas, reactor type and operating conditions. FT synthesis results in the production of various products, thus it is not a highly selective process. However, it offers the possibility to cover the entire range of petrochemical products so that gasoline, jet fuel and diesel can be produced with adequate process control. FT products are high quality and ultra clean fuels, free of sulphur and aromatic compounds and, unlike other fuels such as dimethyl ether and alcohols, they can be easily assimilated in the existing transport infrastructure, concerning both vehicle engines and distribution channels. ## Modelling methodology The process flowsheets of the four evaluated CCU cases were developed using the process simulation software Aspen Plus 32 to estimate material balances, energy and utility requirements as the inputs for the techno-economic analysis. Various thermodynamic methods have been used to model the different unit operations considered in this study. The Aspen Physical Property System guide 33 was used to ensure that the property methods are tailored to the different classes of compounds and operating conditions. The property method used for most unit operations is the Peng-Robinson with Boston-Mathias modifications (PR-BM) which is recommended for gas processing and refinery applications and provides accurate results for hydrocarbon mixtures and light gases, such as H 2 and CO 2 . 33 The non-randomtwo-liquid (NRTL) method with the Redlich-Kwong (RK) equation of state is used to simulate the anaerobic digestion process. The MEA gas treating unit is modelled using the electrolyte-NRTL based property method ENRTL-RK which is suitable for mixed electrolyte systems up to medium pressures. This method uses the RK equation of state for estimating the vapour phase properties. The anaerobic digester is modelled using the Aspen Plus yield reactor block (RYield). The mass yields of CH 4 , CO 2 and digestate were calculated separately considering a biogas production of 0.6 m 3 per kg of volatile solids loading. 19,34 It was assumed that neither NH 3 nor H 2 S are present in biogas since the former is not produced when sewage sludge is used as feedstock and the latter is present in very low concentrations. 35 Table 3 shows the component mass yields calculated for the anaerobic digestion process. In the CO 2 capture plant, the absorber and stripper columns are simulated using the RadFrac block which is suitable for modelling all types of multistage vapourliquid fractionation operations. As mentioned previously, the thermodynamic and physical properties are estimated using the ENRTL-RK method coupled with an electrolyte calculation option which models the electrolyte solution chemistry and consists of five equilibrium reactions. 36 Design specifications are used to obtain the desired molar split fractions in both the absorber and the stripper. In the absorber, a design specification measured the CO 2 flow rate in the stack stream and adjusted the lean MEA flow rate to ensure that a target recovery of 70% is achieved. In the stripper, a design specification measured the CO 2 molar concentration in the CO 2 product stream and adjusted the reflux ratio to achieve a 98 vol% purity target. 21 The number of minimum equilibrium stages was 5 for the absorber and 10 for the stripper. In the CHP unit, compressed biogas is mixed with steam and air (10% excess) and fed into the combustor. The combustor is simulated using a Gibbs reactor block (RGibbs) which models single-phase chemical equilibrium by minimizing the Gibbs free energy, subject to atom balance constraints. Steam is generated inside a network of heat exchangers while electricity is produced in steam and gas turbines by assuming common isentropic and mechanical efficiencies. 37 The CHP plant simulation was based on a natural gas CHP model previously developed by AspenTech. 38 The main modification made to this model was to replace the natural gas feed stream with the biogas outlet stream of the anaerobic digestion plant. Both the steam reformer and the RWGS reactor are modelled using a stoichiometric reactor block (RStoic). Reaction stoichiometry was specified for each of them as well as the fractional conversion of relevant components (80% for CH 4 and 65% for CO 2 , respectively). 12,39 FT synthesis is modelled using a yield reactor block (RYield). The mass yields of the produced hydrocarbons were calculated in a separate spreadsheet using the ASF distribution model described in Section 2.6 with a chain growth probability of 0.85 which favours the production of middle distillates. The single-pass CO conversion was set to 80%. 40,41 Even though such a CO conversion value is relatively high, it can be achieved in slurry phase reactors employing cobalt catalysts. 26,42 The ASF hydrocarbon distribution was taken up to a carbon number of 100. Production of aromatics, oxygenates and olefins is assumed to be negligible in this study since the presence of these compounds is typically small for LTFT synthesis. 25,43 ## Cost estimation methodology The capital and operating costs were estimated for each CCU process concept using the software Aspen Process Economic Analyzer (APEA) which, like Aspen Plus, is licenced by Aspen Technology. APEA was linked to Aspen Plus to estimate costs by utilising the output results of the Aspen Plus simulations. The UK was set as the default country which defines several economic parameters in APEA, such as currency, salary rates, equipment costs and construction materials. 44 Table 4 summarises the assumptions as well as the prices of raw materials and utilities specified in APEA. These prices were converted to GBP (d) and updated to 2013 where necessary. The same economic assumptions were used for all four process designs to allow fair comparisons between them. As discussed previously, the feedstock for all the evaluated CCU cases is sewage sludge. The CCU facility is considered to be part of a large wastewater treatment part and thus the sludge costs are zero since sludge is a by-product of wastewater treatment. The capital investment is comprised of installed equipment costs, indirect costs (e.g. contingency), tax and working capital. 44 The capital investment required to establish the project is considered to be borrowed and repaid over the lifetime of the project (20 years) at a loan interest rate of 10% per annum. To estimate the fuel production costs of the CCU system, the annual amount required to pay back the loan on capital needs to be determined first: where A is the annuity of the capital investment, TCI the total capital investment as calculated in APEA, r the interest rate and N the lifetime of the project. The total annual costs consist of capital annuities as well as operating costs: raw material, utilities, labour and maintenance costs. The fuel production costs are calculated by dividing the total annual costs by the amount of FT fuels (gasoline and diesel) produced in a year. The price inflation of equipment and raw materials is not considered for the ease of comparison between the evaluated CCU concepts. For the same reason, government subsidies, CO 2 credits and by-product revenues are excluded from the economic analysis. ## Mass and energy balances Table 5 presents the mass and energy balances for all four CCU process concepts evaluated in this study. For all considered cases, the sewage sludge input is the same in terms of mass flow and energy content so that results are directly comparable. The energy balances in Table 5 show that sewage sludge is the major source of energy input in the system. The PD-MEA concept, which includes separation of CO 2 from the biogas stream and subsequent steam methane reforming, produces 43 kg h 1 of hydrocarbon fuels which corresponds to 513 kW (LHV) of fuel energy oput. The second CCU concept, PD-CHP1, which involves biogas combustion in a CHP plant without upstream CO 2 scrubbing, results in a lower fuel production of 33.9 kg h 1 (405 kW). Table 5 also shows that the PD-CHP2 concept produces almost the same fuel output (33.7 kg h 1 or 403 kW) as PD-CHP1. The final CCU design produces 28.8 kg h 1 or 344.5 kW of hydrocarbon fuels which is lower than the other three cases since only 70% of the CO 2 in the CHP product gas is captured in the MEA unit (see also Section 2.3), while the rest exits the top of the absorber and is discarded through a stack. However, it should be noted that PD-CHP3 is the only design that produces electricity as a by-product, whereas the amount of electricity produced in the other three designs is insufficient to cover their electricity requirements (for these concepts, the cost of purchasing the extra electricity has been taken into account in the calculations). The PD-MEA concept produces more fuels than the other three cases because of the higher amount of syngas processed in FT synthesis as a result of its upstream methane steam reforming unit which converts methane to syngas instead of simply burning it in a CHP plant. This also significantly affects the hydrogen requirements of the individual concepts. In the case of PD-MEA, the amount of hydrogen produced from the methane steam reformer is higher than the hydrogen requirements of the RWGS reactor; therefore, some hydrogen is produced as by-product (see also Fig. 2). This is not the case for the other three concepts where additional hydrogen is required from an external source. The impact of the hydrogen price on the fuel production costs of PD-CHP1 and PD-CHP2 is examined in Section 5.4. ## Energy efficiencies Table 5 also shows the energy efficiencies of the four studied CCU designs. The fuel energy efficiency measures the fraction of the energy originally in the sewage sludge that ends up in the hydrocarbon fuel product. It is calculated by dividing the energy in the fuel output by the energy content of sewage sludge. Increased hydrocarbon production leads to higher energy efficiencies and thus the greatest efficiency is achieved by PD-MEA (26.4%). The next best options are PD-CHP1 and PD-CHP2 (B20.8%) while PD-CHP3 is the least fuel-efficient (17.7%). The plant energy efficiency Z plant takes into account the total energy input (sludge, hydrogen, natural gas and electricity) and total energy output (fuels, hydrogen and electricity) and is calculated according to: where : M i is the mass flow (kg h 1 ) and LHV i is the lower heating value (MJ kg 1 ) of the product (fuels, hydrogen) or raw material (sludge, hydrogen, natural gas). Since all the terms in eqn ( 6) are expressed in kWth except for the electricity consumed by the plant, the latter is divided by the thermal efficiency of the power cycle assumed to be 39%. 50 As for the fuel efficiency, PD-MEA also has the highest plant energy efficiency, estimated at 17.1%; this is due to the higher fuel output as well as the excess hydrogen production from which the other process designs do not benefit. PD-CHP3 shows the highest efficiency (14%) of all three CHP-based cases despite the fact that the fuel production is approximately 18% lower than that of the other concepts. The primary reason for this is that PD-CHP3 produces more than enough electricity to cover all the power requirements of the plant so there is no need to provide electricity externally. PD-CHP1 and PD-CHP2 achieve similar efficiencies (11.7% and 11.9%, respectively) which suggests that combustion of upgraded biogas in a CHP plant does not significantly benefit the overall CCU plant performance. ## Energy requirements The energy consumption calculated through the Aspen Plus simulations is used to assess the heat and power consumption in the different parts of the plant; these are given in Table 6. As mentioned in Section 2, steam produced from the combustion of FT off-gas and the LP steam from the CHP plant are used to supply heat to the anaerobic digesters which require approximately 640 kW of LP steam. The amount of LP steam produced by the CCU plant is sufficient to cover the energy requirements of the anaerobic digesters for PD-CHP1 and PD-CHP2, whereas this is not the case for PD-MEA and PD-CHP3. The most energyintensive processing steps in terms of heat requirements are the RWGS CO 2 conversion and steam reforming because they are both endothermic processes and thus require a significant amount of heat to operate. It should be noted that the CHP plant is fully integrated and does not require any external energy inputs (heat and electricity). 38 Note that the FT synthesis is an exothermic process and thus requires cooling (water) rather than heating to maintain the operating temperature in the reactor. However, it is the most power consuming section for the PD-CHP1 and PD-CHP2 designs due to the compression of a large volume of processed syngas which contains excess air from the CHP plant. This is not an issue for PD-MEA which does not include a CHP unit. Similar is true for PD-CHP3 which employs an MEA unit right after the co-generation plant. Generally, it can be seen that none of the four CCU designs is energy self-sufficient with the exception of PD-CHP3 which produces surplus electricity but still requires additional heat. In this study, natural gas is used Energy & Environmental Science Paper to cover the additional heating requirements of the plant, while electricity is bought from the grid, if needed. ## Economic analysis Fig. 8 shows the breakdown of capital costs for different parts of the CCU plant and the resulting total capital investment (TCI) for the four CCU process designs. The calculated TCI is expressed in 2013 million British Pounds and ranges from d30 (PD-MEA) to d36 million (PD-CHP3). The CHP-based cases are associated with higher capital costs than PD-MEA because of the CHP unit which increases the equipment costs and thus the TCI. As indicated in Fig. 8, anaerobic digestion represents 41-45% of the overall costs of the CCU plant. This is due to the two digestion reactors which handle a large volume of sewage sludge, as discussed in Section 2.2. The CHP unit is the second most expensive process section, contributing 19-22% to the TCI. The steam reforming, RWGS and FT synthesis sections necessitate similar capital investment ranging from 13-19% of the overall plant costs. Note that the capital costs of the PSA unit and the FT off-gas combustion are also included in the estimated capital investment of the RWGS and FT synthesis section, respectively. Finally, the capital costs of the MEA unit constitute the smallest fraction, requiring about 10% of the total capital investment. The annual operating and maintenance (O&M) costs are shown in Fig. 9. The operating costs include expenditure for materials (e.g., catalysts, hydrogen), utilities (e.g., electricity, natural gas), labour, maintenance and other costs (e.g., overheads, insurance). The O&M costs range from d4.9-6.4 million with PD-MEA resulting in the lowest expenditure owing to the lower equipment and utilities requirements associated with this process configuration. Labour and maintenance costs are the largest contributor to O&M costs and represent 37-45% of the total O&M expenditure. Other costs represent 31-39%, including the expense for MEA and PSA packing. Hydrogen contributes 12-17% to the total operating costs of the CHPbased designs. Catalyst and electricity costs are higher for PD-CHP1 and PD-CHP2 because of the higher volume of processed gas compared to the other two cases. Finally, cooling and heating utilities represent a small fraction of the total operating costs (2-5%). The production costs per litre of gasoline and diesel are presented in Table 7 for the four CCU configurations, along with the contribution of capital costs (as capital annuity) and O&M expenditure. The calculated production costs do not include tax, duties, producer and retailer profits, marketing expenditure and distribution costs. As can be seen in Table 7, O&M costs are a more important contributor to the production costs than the capital investment as they represent 58-62% of the total production costs. PD-MEA has the lowest production costs at d15.8 per litre because of its lower capital and operating costs as well as higher fuel production compared to the other three cases. The next best option is PD-CHP at d23.2 per litre which has the lowest production costs among the three CHPbased designs. PD-CHP3 is associated with the highest fuel production costs at d29.6 per litre which is approximately 87% higher than for PD-MEA. The main reason for this is that this concept produces a significantly lower amount of liquid fuels than PD-MEA, as discussed in Section 5.1. From these it is clear that the amount of fuel produced (and thus conversion efficiencies) is a very important element of the production costs; thus, its effect is investigated in the sensitivity analysis later in the paper (Section 5.5). 5.4.1 Comparison with costs of conventional transport fuels. Retail prices of liquid fuels produced via the proposed CCU technologies are uncertain at present as several factors, such as profit margins, applicable taxes as well as government subsidies and CO 2 credits, still need to be determined. Another factor which significantly affects the fuel production costs of CCU is the economies of scale: the bigger the plant, the lower the production costs. For this reason, the effect of the economies of scale is investigated here to assess the potential economic competiveness of CCU against conventional, fossil fuel technologies. Only the PD-MEA concept is considered as this process design has the lowest production costs. Twelve plant capacities Fig. 8 Total capital investment for the evaluated CCU process designs. ## Paper Energy & Environmental Science are evaluated, ranging from 1 tonne (base case) to 1670 tonnes of liquid fuels produced per day. The latter capacity corresponds to the Bintulu gas-to-liquids (GTL) plant in Malaysia, one of the largest FT plants in the world, owned by Shell. 51 The six-tenths factor rule 52 was applied to estimate the investment costs of the scaled-up CCU plants as follows: where C 1 and C 2 are the costs of the PD-MEA base case and the larger plant, respectively, S 1 and S 2 are the capacities of the PD-MEA base case and the larger plant, respectively, and 0.6 is the scaling factor. For PD-MEA, the operating costs are approximately 16% of the capital costs; therefore, the same percentage contribution was assumed in calculating the operating costs of the scaled-up plants. Using the above approach, the capital investment for the PD-MEA plant of the largest capacity considered here (1670 tonnes per day) is estimated at d2.6 billion. This is three times higher than the capital investment of the Bintulu plant, which cost $1.3 billion 51 or d831 million (2013 exchange rate: d1 = $1.56 53 ). Therefore, it is highly unlikely that the industry would invest in a CCU plant when they could instead build a conventional fuel production plant of the same capacity at a much lower cost, while also reducing financial and other risks by relying on a commercially proven, rather than a new technology. Fig. 10 shows the effect of scale on the costs of CCU fuels. For the largest plant capacity, the fuel production costs are almost 16 times lower than for the PD-MEA base case (d15.80 vs. d1.00 per litre). However, the effect of economy of scale levels off for capacities above 620 tonnes per day, with much smaller cost reductions thereafter. By comparison, the cost of producing conventional diesel in 2013 was d0.51 per litre and d0.47 per litre for gasoline 54 (gate costs, excluding tax, duty, profits, marketing and distribution costs). This is around two times lower than the CCU fuel costs. Therefore, unless significant improvements are achieved in the conversion efficiencies of CCU technologies, along with introduction of government subsidies and incentives, it is highly unlikely that CCU fuels will be able to compete against conventional transport fuels, which despite their contribution to climate change, are still relatively cheap to produce. The effect of economies of scale on the fuel production costs of the evaluated CCU process designs was also investigated using Aspen Plus and APEA. This also allows for comparisons with the costs estimated with the six-tenths rule above. Two scaled-up models of the PD-MEA concept were considered: a medium scale and a large scale plant at 850 and 1670 tonnes of fuel per day, respectively. Table 8 shows the capital investment and production costs of the two scaled-up designs calculated by APEA. For the medium plant capacity, the production cost drops to d2 per litre which is approximately eight times lower than that of the PD-MEA base case. As expected, the large scale plant with a capacity Energy & Environmental Science Paper equal to the Bintulu plant has an even lower production cost at d1.2 per litre which is about 20% higher than the cost estimated with the six-tenths rule. The capital investment of the large scale process design estimated by APEA is d1.34 billion which is almost half of the equivalent six-tenths rule cost (d2.6 billion); however this is still 61% higher than the capital investment of the GTL Bintulu plant. Therefore, using either cost estimating method, it is clear that CCU fuels are currently significantly more expensive than conventional transport fuels. ## Sensitivity analysis This section considers the effect of several key parameters on the overall fuel production costs: CO 2 removal efficiency of the MEA CO 2 capture plant, percentage of CO 2 conversion in the RWGS reactor, capital costs, loan interest rate, plant life, operating hours, electricity and hydrogen prices. Fig. 11 shows the dependence of fuel production costs on the CO 2 separation efficiency of the MEA CO 2 unit for PD-MEA, PD-CHP2 and PD-CHP3. For 90% CO 2 removal efficiency, production costs decrease to d13.8 per litre and d24.1 per litre for PD-MEA and PD-CHP3, respectively since more CO 2 is converted to CO in the RWGS reactor which results in an increased fuel production (and thus lower production costs). This is not the case for PD-CHP2 where production costs are 32-33% higher than the base case cost with an increase in the CO 2 removal efficiency. Generally, as the CO 2 removal efficiency increases, the volume of gas which is passed to the stripper in the CO 2 capture unit also increases and this results in higher energy requirements as well as higher equipment and operating costs. For PD-MEA and PD-CHP3, this increase in costs is outweighed by the increase in fuel production. However, in PD-CHP2, the same amount of CO 2 in biogas will pass to the RWGS reactor regardless of the absorption efficiency of the MEA unit since the flue gas from the CHP plant and the concentrated CO 2 gas stream are mixed just before the RWGS section (see Fig. 4). Therefore, in case of PD-CHP2 production costs will rise as there is no increase in production which could potentially outbalance the higher capital and operating costs of the MEA CO 2 capture unit as the CO 2 removal efficiency increases. For the other technical and economic parameters, the sensitivity analysis was carried out by changing each parameter in turn by AE30% of its base-case value (see Table 4) with the exception of the plant operating hours which were changed by AE10% since they cannot exceed the maximum hours per year. The results for the four design concepts are shown in Fig. 12. The bars show deviations from the original values of the model parameters with longer bars indicating a higher degree of sensitivity to a particular parameter. In the case of PD-MEA, production costs are most sensitive to the capital investment costs. If the total capital is decreased by 30%, the fuel production cost drops to d13.81 per litre or 14.6% below the base case cost. However, errors of AE30% for capital investment estimates are typical 40,50 and increased accuracy can only be achieved through very detailed and expensive analysis of a real case. Another important factor that increases the inherent uncertainties in projecting CCU capital costs is the different level of development of some of the technologies considered in this study. For example, the RWGS process has only been proven on a small scale and therefore is still under development as opposed to steam reforming which is a mature and well understood technology. The high sensitivity to variations in capital costs also emphasises the importance of economies of scale, which the studied CCU options do not benefit from yet as opposed to conventional fuel production plants. For all CHP based models, production costs are most sensitive to changes in the CO 2 conversion rate in the RWGS reactor. This suggests that the lower the fuel output, the more sensitive production costs are to variations of the CO 2 conversion efficiency. The production costs of the CHP based models can be reduced by 24-29% for a CO 2 conversion rate of 84.5% (30% higher than the base case). Therefore, improving the performance of the CO 2 hydrogenation technology should be an early priority. The loan interest rate is the second most sensitive parameter for PD-MEA and third for all CHP cases. Production costs can be decreased by 8-9% when the interest rate is reduced from 10% to 7%. Interest rates influence the capital annuities and can be controlled by agreeing fixed rates with the lender throughout the life of the project, significantly reducing the uncertainty associated with this economic parameter. Finally, production costs are less sensitive to the project's lifetime, electricity and hydrogen prices as well as operating hours. ## Energy & Environmental Science Paper This journal is © The Royal Society of Chemistry 2015 ## Conclusions This work has examined the technical and economic feasibility of four CCU process configurations for the production of liquid transport fuels. The process designs considered here incorporate existing CCU technologies for the conversion of a carbon source, in this case sewage sludge, to fuels. The first process configuration examined comprises separation of CO 2 from biogas by MEA and steam reforming for conversion of methane to syngas, while the other three cases incorporate a CHP plant for co-generation of heat and power from biogas and the conversion of methane to CO 2 . All four cases include a RWGS reactor for converting CO 2 to syngas and its subsequent conversion to fuels via Fischer-Tropsch synthesis. Detailed designs were developed in Aspen Plus to determine the technical and economic potential of the selected process configurations and identify the concept that has the lowest overall costs. The overall plant energy efficiency and production costs of the evaluated designs range from 12-17% LHV and d15.8-29.6 per litre of produced fuels, respectively. The process configurations which incorporate a CHP plant result in significantly lower efficiencies and higher costs than the process design with MEA CO 2 capture and steam reforming. The primary reasons for this are the higher syngas production in the steam reforming process and the high capital and operating costs of CHP. The sensitivity analysis reveals that the fuel production costs are mainly influenced by variations in capital costs, the CO 2 removal efficiency of the CO 2 capture plant and the rate of CO 2 conversion. This emphasises the importance of optimising current CCU technology, as well as the significance of economies of scale which greatly benefit commercial plants. For example, for the best design case, the costs of fuel production for a larger capacity plant (1670 tonnes per day), are d1-1.2 per litre, down from d16 per litre for a plant producing 1 tonne per day. However, this is still twice as high as the cost of conventional transport fuels. Therefore, fuel production with current CCU technologies is not yet economically viable primarily due to the: (i) low CO 2 conversion in the RWGS process, (ii) low selectivity of the Fischer-Tropsch synthesis and (iii) relatively low CO 2 separation efficiency in the MEA absorber. This highlights the need for new CCU technologies, some of which are currently being developed (e.g. ionic liquids for CO 2 capture, co-electrolysis of CO 2 and water, dry methane reforming). Further research will be carried out to complement the analysis presented here, including the assessment of other less developed technologies (e.g. dry methane reforming) and products (e.g. methanol, formic acid), life cycle assessment to examine the environmental impacts of the studied CCU designs and the possibility of additional revenue from sales of byproducts and avoided greenhouse gas emissions.

chemsum

{"title": "Carbon dioxide utilisation for production of transport fuels: process and economic analysis", "journal": "Royal Society of Chemistry (RSC)"}

cellular_uptake_and_targeting_of_low_dispersity,_dual_emissive,_segmented_block_copolymer_nanofibers

## Abstract: Cellular uptake and targeting of low dispersity, dual emissive, segmented block copolymer nanofi bers Crystallization-driven self-assembly (CDSA) was used to prepare low dispersity segmented 1D nanoparticles from an amphiphilic block copolymer, poly(dihexylfl uorene)b-poly(ethyleneglycol). The cellular uptake of 85-95 nm segmented triblock and pentablock 1D nanofi bers bearing folic acid and a BODIPY dye was studied, revealing that nanofi bers interact with the cell membrane end-on, and localize to the perinuclear region. The presence of folic acid was essential for cell uptake to occur. This fundamental study uncovers insights into the cellular uptake of low dispersity 1D polymer nanoparticles, suggesting their suitability for applications in nanomedicine. rsc.li/chemical-scienceCellular uptake and targeting of low dispersity, dual emissive, segmented block copolymer nanofibers † ## Introduction Nanoparticle-mediated therapeutics show considerable promise in the diagnosis and treatment of a plethora of diseases that affect human health, especially cancer. 1,2 The delivery of cargo such as drugs, proteins, imaging agents and nucleic acids to specifc locations inside cells in target tissue in the human body however remains a major challenge, 3 despite their considerable potential. The ideal nanoparticle delivery system therefore has several requirements such as biocompatibility, 4 high specifcity, 5 and a high loading capacity, 6 maximizing efficacy whilst remaining as cost effective as possible. In practical terms, this means maintaining a modular, versatile design whilst simultaneously exhibiting precise control over nanoparticle size, shape, rigidity, and surface chemistry. To this end, 1D nanomaterials have attracted substantial recent attention, with a wide range of potential advantages evidenced over considerably more well-studied spherical systems such as improved circulation, 12 retention, 13,14 adhesion, 15,16 specifcity, 17 and cell uptake. 18 The 1D shape has also been shown to enable unique endocytosis mechanisms 19 involving improved membrane wrapping 20 and reduced macrophage uptake (which is length dependent) 21,22 thereby offering the promise of enhanced active targeting capabilities. One of the most well-studied active targeting agents is Folic Acid (FA), with several FA conjugates in clinical trials. 23 FA is the substrate for folate receptors such as FRa, which are overexpressed in numerous types of cancer, and have represented a signifcant target for the delivery of tailored therapeutics. A variety of anisotropic nanoparticles has been functionalized with folic acid, such as polyacrylic acid-b-polystyrene spherical and cylindrical micelles, 28 gold nanorods, 29 and coordinationcomplex nanotubes, 30 with anisotropic particles displaying features such as increased uptake over spheres, 28 and disassembly upon cellular internalization. 30 Other polymeric systems that have used FA as a targeting agent include spherical micelles and star-shaped polymers based on PLA 31,32 and spherical micelles based on PCL. 33,34 Recently, a seeded-growth approach termed 'living' crystallization-driven self-assembly (CDSA), has been developed which allows access to a wide range of morphologically pure, low dispersity 1D (and also 2D) core-shell nanomaterials. Briefly, amphiphilic block copolymers ('unimers') with a crystallizable core-forming block are dissolved in a 'common' solvent which is compatible with both blocks and the resulting solution is then mixed with a 'selective' solvent which solvates only one block (the corona-forming block). These conditions yield polydisperse fber-like micelles with an insoluble, solvophobic crystalline-core and a solubilizing solvophilic corona, via a self-nucleation mechanism. Sonication of the resulting polydisperse fber-like micelles causes fragmentation, yielding small 'seed' micelles. Further addition of unimer to the seed micelles leads to epitaxial growth and low dispersity micelles with a length controlled by the unimer to seed ratio in a manner analogous to a living covalent polymerization of molecular monomers. 37 This process is uniquely suited towards the generation of nanoparticles that are otherwise hard to access, such as uniform 1D nanofbers, (and also 2D platelets) 45,46,48 as well as more complex assemblies. 50 For example, 'living' CDSA can also be used to generate hierarchical nanomaterials, such as segmented nanofbers with spatially-defned functionalizable regions, 36,38,39, as well as random-and gradient comicelles by the sequential or simultaneous seeded growth of different block copolymers with distinct coronal chemistries. 52 Signifcantly, the ability to tailor surface chemistry in individual regions allows for modular functionalization. 53,54 Despite the substantial recent progress made with selfassembled nanomaterials formed via 'living' CDSA, the majority of systems described so far have involved the use of organic rather than aqueous media, limiting their potential for biomedical applications. Only a few examples exist of the use of 'living' CDSA to prepare low dispersity fbers of controlled length which can be dispersed in water. Several are based on a crystallizable polyferrocenylsilane (PFS) core. 38,42 The Dove and O 0 Reilly groups have reported the frst example of 'living' CDSA in aqueous media based on a biocompatible and biodegradable polycaprolactone (PCL) core-forming block, thereby accessing fbers with lengths of up to 800 nm. 44 In addition, we recently reported the 'living' CDSA of a biocompatible polycarbonate core-forming block, with morphologically pure 1D block co-micelles accessed with lengths up to $1.6 mm and which are colloidally stable in aqueous media. 39 Despite these advances, the application of functional nanoparticles produced via CDSA to biomedicine remains a nascent feld, requiring much further development in areas such as scalability, incorporation of functionality, and biological activity. Block copolymers with a crystallizable p-conjugated core have also been shown to undergo 'living' CDSA. One such class of p-conjugated materials is polyfluorenes (PFs), 59 which exhibit strong luminescence, making them excellent candidates for chemo/biosensors, diagnostics and imaging agents. Whilst most work has focused on optoelectronic properties, some studies have explored biological applications. The majority of studies have focused on the use of a conjugated PF backbone, with charged side chains that provide aqueous stability. The resultant polymers self-assemble into spherical nanoparticles, with FA either covalently linked to the polymer, leading to selective cell uptake 67 or electrostatically bound, leading to FA dependent fluorescence quenching of PF. 68 Numerous examples also exist of PF containing cationic pconjugated polymers for nucleic acid binding and the detection of pathogens 65 whilst p-conjugated polythiophenes have also been used for the delivery of nucleic acids to cells. 69 Associated with the limited development of 1D fber-like micelles prepared by 'living' CDSA that are dispersible in aqueous media, 38,39,42,44 a paucity of biological data currently exists, with currently available studies largely limited to cytotoxicity experiments. Whilst polydisperse worm-like micelles with amorphous cores have been investigated, 70 fundamental questions remain regarding the in vivo and in vitro effects of low dispersity polymer nanofbers in which the core is crystalline and more rigid. It is noteworthy that, to date, there have only been limited studies on the effect of 1D fber length on cell uptake, 18, with no reports on the behavior of polymerbased nanofbers. Furthermore, there are very few examples of easily functionalizable fber-like micelles that are dispersible in water. Fiber-like micelles with a p-conjugated PF core offer bright fluorescence of potential interest for imaging, tracking nanoparticles inside cells, and sensing. Herein we describe the preparation of length-controlled, low dispersity 1D PDHF-b-PEG (PDHF ¼ poly(di(n-hexyl)fluorene), PEG ¼ poly(ethyleneglycol)) nanofbers in aqueous media, and studies of their functionalization, cytotoxicity, cellular targeting, uptake, and localization. ## Preparation of colloidally stable dual emissive PDHF triblock and pentablock nanobers in water The PF block copolymer PDHF-b-PEG was selected in cellular uptake studies because PEG is known as a 'stealth' polymer and can provide biocompatibility as well as aqueous colloidal stability. First, alkyne-terminated PDHF 13 homopolymer was prepared (M n ¼ 4400 g mol 1 , DP ¼ 13, determined via MALDI-TOF MS, Đ M ¼ 1.22, determined by GPC, Fig. S1 †) via Grignard metathesis (GRIM) polymerization using a previously described procedure (Scheme S1 †). 74 Heterobifunctional PEG was synthesized by mono-tosylation of HO-PEG 249 -OH (Fig. S2 †) and, after substitution of the tosylate for azide, the resulting mixture of HO-PEG 249 -OH and HO-PEG 249 -N 3 was used without further purifcation in the Huisgen 1,3-dipolar cycloaddition 'click' coupling with the alkyne-terminated PDHF 13 according to the previously reported method. 74 Excess HO-PEG 249 -OH was removed via precipitation to yield PDHF 13 -b-PEG 227 (Fig. 1A, M n ¼ 29 900 g mol 1 , Đ M ¼ 1.12 determined by GPC, block ratio determined by 1 H-NMR, Fig. S3 and S4 †). While the p-conjugated PDHF core of the PDHF 13 -b-PEG 227 micelles is inherently fluorescent, in biological systems the blue emission is subject to fluorescence quenching upon interaction with a variety of species, 68,75 and will also compete with background cell autofluorescence. Thus, to supplement the results from PDHF fluorescence, we also introduced a far-red fluorophore to allow for an additional tracking capability. The far-red BODIPY 630/650-X (BD) fluorophore (l ex ¼ 630 nm, l em ¼ 650 nm, excitation/emission in superscript) was selected to attach to the PEG terminus (Fig. 1A). This was achieved by (iii) Triblock nanofibers in MeOH/THF (1 : 1, L n ¼ 56 nm, L w /L n ¼ 1.09, s L ¼ 18 nm), and (iv) pentablock nanofibers in H 2 O (L n ¼ 95 nm, L w /L n ¼ 1.17, s L ¼ 39 nm, W n ¼ 13 nm, W w /W n ¼ 1.02, s W ¼ 2 nm). Samples were prepared at 0.5 mg mL 1 . using the terminal hydroxyl group of the PEG block for further chemical functionalization (Scheme S1 †). Condensation of PDHF 13 -b-PEG 227 with Boc-b-alanine, followed by Boc deprotection yielded a terminal amine residue, which was then further modifed with the BODIPY 630/650-X N-hydroxysuccinimide (NHS) ester to yield an end-functionalized, dualemissive PDHF 13 -b-PEG 227 polymer, termed PDHF 13 -b-PEG 227 -BD (Fig. 1A). The ability to functionalize the corona chain end with an amine also allowed us to employ amide coupling chemistry to attach targeting moieties for the active uptake of nanofbers into a target environment or cell type via receptormediated endocytosis. To this end, we frst selected FA as our targeting group of choice as it is well established that folate receptors are over-expressed in many different types of cancer, with several treatments involving folate targeting undergoing clinical evaluation such as vintafolide. 23 Thus, we adapted the chemistry developed for attaching the BD dye to the PEG chain terminus, instead attaching an N-hydroxysuccinimide activated FA derivative (Scheme S1 †). This yielded FA functionalized PDHF 13 -b-PEG 227 , termed PDHF 13 -b-PEG 227 -FA (Fig. 1A). According to previous studies, dimensions of ca. 10-100 nm represent the most desirable size range for nanoparticles to be used as drug delivery vectors, as this leads to optimum circulation in the bloodstream. Objects within this size regime are sufficiently large to avoid renal and lymphatic clearance, yet sufficiently small to avoid opsonization. 3,76 Therefore, we aimed to prepare low dispersity 1D fber-like micelles of PDHF-b-PEG with lengths of #ca. 100 nm in this study. By comparison, assuming a chain-extended structure for the PDHF segment as previously found, 74 the core cross-section of the PDHF 14 -b-PEG 227 1D fbers is 65 nm 2 (W n H n ¼ 13 5 nm, where W n and H n are the number average width and height respectively). The ability to prepare segmented nanofbers via living CDSA should also yield advantages for the optimal presentation of targeting groups (such as FA) and cargo (such as BODIPY 630/650-X ) as well as facilitating modular nanoparticle construction. To avoid complications with the self-assembly process, unfunctionalized PDHF 13 -b-PEG 227 was used to form the initial seed micelles, after which the PDHF 13 -b-PEG 227 -BD and/or PDHF 13b-PEG 227 -FA unimers were added sequentially to create segmented nanofbers. To facilitate optimum cellular uptake, the segmented nanofbers were designed to possess terminal PDHF 13 -b-PEG 227 -FA blocks, as previous work has revealed that receptor mediated endocytosis of 1D nanomaterials occurs primarily through association of the nanoparticle tip with the cell membrane. 77 The nanofbers were prepared via the seeded-growth method (Fig. 1B). Briefly, polydisperse PDHF 13 -b-PEG 227 nanofbers were formed by the addition of MeOH to a solution of PDHF 13 -b-PEG 227 in THF and aged for 24 h (Fig. S5A †). Seed nanofbers (L n ¼ 21 nm, L w /L n ¼ 1.07, s ¼ 8 nm) were prepared by sonication of the resultant polydisperse nanofbers for 3 h at 22 C To assess the effects of both FA and BD on cell uptake, triblock comicelles containing solely BD or FA functionalization were required as controls. To ensure that results were comparable to those obtained for pentablock nanofbers with both BD and FA decorated segments, and to ensure that nanofber length was not a variable affecting results, we aimed to produce nanofbers with lengths comparable to the FA-BD-PEG-BD-FA pentablock nanofbers prepared previously. Addition of PDHF 13 S9A †), with an average length of 105 nm (L w /L n ¼ 1.05, s L ¼ 24 nm) before dialysis in THF/MeOH (1 : 1, Fig. 2D and E) and an average length of 90 nm (L w /L n ¼ 1.11, s L ¼ 30 nm, W n ¼ 12 nm, W w /W n ¼ 1.02, s W ¼ 2 nm) after dialysis into water (Fig. 2F, S7C and S9B-E †). UV/vis absorption and fluorescence profles of PDHF 13 -b-PEG 227 nanofbers The absorption and fluorescence profles of BD-PEG-BD and FA-PEG-FA triblock nanofbers, and FA-BD-PEG-BD-FA pentablock nanofbers in water and PBS were investigated prior to cellular experiments (Fig. S10-S14 †). The absorbance and fluorescence excitation profle for all nanofbers exhibited a l max of 375 nm, closely matching previously reported spectra for PDHF in organic solvents. 74,78 The emission profles of BD-PEG-BD triblock nanofbers and FA-BD-PEG-BD-FA pentablock nanofbers both exhibited a peak at 650 nm, which corresponds to the BD dye (Fig. S10B and C †). The excitation profle for the emission of BD at 650 nm matched the excitation of PDHF, indicative of Förster resonance energy transfer (FRET) between the p-conjugated PDHF core and the BD dye. As FRET interactions are very sensitive to distance, 79 the results indicate that the BD dye is in close proximity (within ca. 10 nm) to the PDHF core in water and is presumably located near the corecorona interface. Confocal Laser Scanning Microscopy (CLSM) and fluorescence measurements in PBS and cell media revealed that both PDHF and BD can be tracked in complex media for use in cell uptake studies (Fig. 3, S13 and S14 †). ## Cellular uptake of BD-PEG-BD triblock nanobers We sought to utilize the dual-emissive nature of BD-PEG-BD triblock nanofbers to investigate if untargeted PDHF nano-fbers were capable of cellular uptake. Investigations began by incubating the same low dispersity 85 nm BD-PEG-BD triblock nanofbers (L w /L n ¼ 1.19, s L ¼ 38 nm) with HeLa cells for 1 h at a concentration of 50 mg mL 1 . After incubation, the cells were fxed, and the nucleus was stained with DAPI (4 0 ,6-diamidino-2phenylindole), the F-actin was stained with Alexa Fluor 488 Phalloidin, and the cells were imaged using CLSM. The results (Fig. S15 †) revealed that limited intracellular fluorescence was observed upon excitation for the BD fluorophore (l ex ¼ 633 nm, l em ¼ 640-700 nm) for cells incubated with BD labeled nano-fbers over the fluorescence arising from control cells which had not been exposed to any nanofbers. These results, which implied that BD-PEG-BD triblock nanofbers are not internalized by cells, were confrmed by live cell imaging (Fig. S16A-F †), where incubation of the BD-PEG-BD triblock nanofbers with HeLa cells for 45 minutes at a concentration of 50 mg mL 1 also led to similar results, with no observable emission from either PDHF or BD. Finally, similar live cell experiments were conducted with BD-PEG-BD triblock nanofbers where the supernatant was left in suspension over the cells for 1 h before imaging. The results from this experiment (Fig. S16G-I †) revealed that the fluorescence from the BD fluorophore was located extracellularly, confrming the successful visualization of the BD labeled nanofbers in the presence of cells. Taken together, these results indicate that BD-PEG-BD triblock nanofbers with a neutral PEG corona alone are incapable of being internalized by the cells studied, and that the introduction of active targeting (in the form of FA) is required to enable cellular internalization to take place. ## Cellular uptake of folic acid-decorated dual-emissive PDHF-b-PEG nanobers In order to investigate whether the addition of FA to PDHF 13 -b-PEG 227 nanofbers facilitates cellular uptake, dual-emissive FA-BD-PEG-BD-FA pentablock nanofbers (10 mg mL 1 , L n ¼ 95 nm, L w /L n ¼ 1.17, s L ¼ 39 nm) were incubated with HeLa cells, and imaged via live cell CLSM. After 30 minutes incubation, signifcant uptake of FA-BD-PEG-BD-FA pentablock nanofbers was observed (Fig. 4A-D). While negligible fluorescence was observed in the blue channel for PDHF, there was signifcant fluorescence observed from BD. The punctate fluorescence appeared to be within the cell throughout the cytosol, concentrated around the perinuclear region, whilst little uptake was observed in a central region, presumably the nucleus. The observed fluorescence around the perinuclear region may correspond to nanofbers that are located around the nuclear membrane. Further experiments with cells where the nucleus was labelled with DAPI, and the F-actin labelled with Alexa Fluor 488-Phalloidin (Fig. 4E-H) confrmed that little fluorescence is found within the nucleus, implying that FA-BD-PEG-BD-FA pentablock nanofbers are unable to localize in that region. Examination of z-stack data of both fxed and live cells (Fig. S17 †) revealed that the punctate fluorescence was located within the cell, rather than on the surface, indicating that the nanofbers are internalized inside the cell and are not attached to the exterior of the plasma membrane. Fluorescence quenching of the PDHF core of the nanofbers was also observed (see ESI Page S3 †). Next, to quantitatively probe the cellular uptake of FA-BD-PEG-BD-FA pentablock nanofbers and compare this to BD-PEG-BD triblock nanofbers lacking FA, we undertook flow cytometry experiments with HeLa cells (Fig. 5 and Table S1 †). After 45 minutes of incubation with BD labelled nanofbers either bearing FA (L n ¼ 95 nm, L w /L n ¼ 1.17, s L ¼ 39 nm) or lacking FA (L n ¼ 85 nm, L w /L n ¼ 1.19, s L ¼ 38 nm), cells were detached with Accutase® and counted via flow cytometry. After gating sequentially for cells, single cells, and live cells, the remaining cells were gated for either BD fluorescence (l ex ¼ 633 nm, l em ¼ 660/20 nm) or PDHF fluorescence (l ex ¼ 405 nm, l em ¼ 450/50 nm). Results for 85 nm BD-PEG-BD triblock nanofbers indicated that they were not uptaken by the cells, as BD fluorescence was equal to that of control HeLa cells that had not been exposed to any nanofbers. In contrast, results for 95 nm FA-BD-PEG-BD-FA pentablock S1, Fig. S18 and S19. † nanofbers revealed a shift in BD fluorescence, with >99% of the cells counted (>10 000) in every repeat displaying a signifcant increase in BD fluorescence (Fig. 5H). The increase in BD fluorescence over control HeLa cells observed in $100% of the cells counted implies that the addition of FA to the periphery of the terminal segments of the corona of the nanofbers successfully facilitates cellular uptake. Furthermore, because nanofbers without FA do not undergo internalization, the uptake should be entirely dependent on the presence (or absence) of folate receptors on the target cell, opening up the possibility of using PDHF nanofbers for the targeted imaging or delivery of therapeutics for diseases such as cancer. Analysis of the median fluorescence intensity of each cell (Fig. 5G) revealed that for fbers without FA, median fluorescence was comparable to control HeLa cells (149% of control) whereas fbers with FA exhibited a ca. 1660% increase in fluorescence intensity. The large increase in median fluorescence intensity for fbers with FA is further evidence for their uptake into cells. Nanoparticle uptake can also be measured by changes in side scattering from flow cytometry, 80 however no signifcant differences in side scattering were observed for any of the experiments conducted on these nanofbers (Fig. 5I). ## Investigations into uptake pathway and intracellular localization Intrigued by the differences between PDHF and BD fluorescence, we attempted to further investigate the cellular uptake pathway of FA-BD-PEG-BD-FA pentablock nanofbers via correlated CLSM and electron microscopy (CLEM) on cells cooled to 4 C. At this temperature, active transport mechanisms are considerably slowed down. Thus, if internalization still occurs it is likely to proceed through temperatureindependent invagination, whereas if the nanofbers are only bound to the outer cell membrane then uptake is likely to occur through one of the many active transport mechanisms. Considering that the FA-BD-PEG-BD-FA pentablock nano-fbers discussed here are decorated with FA, one might assume that uptake occurs through receptor-mediated endocytosis, as reported for other FA decorated nanoparticles. 65,67, Our initial experiments involved the addition of FA-BD-PEG-BD-FA pentablock nanofbers (50 mg mL 1 , L n ¼ 95 nm, L w /L n ¼ 1.17, s L ¼ 39 nm) to HeLa cells expressing GRASP65-GFP (that contain GFP labelled Golgi apparatus as a reference) 87 on ice. After 10 minutes of incubation (which should allow for association between FA residues and folate receptors on the cell surface), cells were imaged via CLSM. Results (Fig. 6A-D) indicated that BD fluorescence was observed partially inside cells as well as around the cell membrane (e.g. Fig. 6C), which was interesting as uptake of FA-BD-PEG-BD-FA pentablock nanofbers at 4 C was unexpected, potentially pointing to two different uptake mechanisms operating. TEM analysis of 70 nm slices of the cells revealed individual nanofbers interacting with the cell membrane, as well as smaller electron-dense anisotropic particles proximal to the cell membrane (Fig. 6E and H, S20 and S21 †). Analysis of the lengths and widths of these electron-dense anisotropic particles (Fig. S22 †) revealed a L n of 19 nm (L w /L n ¼ 1.11, s L ¼ 6 nm) and a W n of 10 nm (L w /L n ¼ 1.14, s W ¼ 4 nm), values that are comparable to the dimensions expected for the PDHF core. 74 We hypothesize that the anisotropic electron-dense particles observed may correspond to fragments of PDHF 13 -b-PEG 227 nanofbers, and they will be referred to as 'fragments' henceforth. It is important to note that whilst the fragments observed closely match the BD fluorescence in CLSM data, their small size and shape (10-20 nm) also closely match those of other natural cellular structures such as ribosomes, glycogen granules, and nucleosomes, preventing defnitive assignment via TEM. Further evidence for the fragmentation of nanofbers upon cellular internalization was provided by TEM micrographs where a lower contrast 'corona' (Fig. S21, † red circle) was observed around the nanofbers, which appeared to be associated with cleavage (Fig. S21, † green circle). Further analysis of cell slices of a single cell imaged via CLEM revealed both intact FA-BD-PEG-BD-FA pentablock nanofbers and fragments throughout the cell, concentrated around the perinuclear region (Fig. 7 and S23 †). Intact nano-fbers were observed inside endosomal-like vesicles (Fig. 7, circled yellow and Fig. S23 †), alongside fragments (Fig. 7, circled green), which may correspond to late endosomes. 88 Free intact nanofbers (Fig. 7, circled red) and free fragments (Fig. 7, circled purple) were also both observed inside the cytosol. The presence of intact nanofbers inside endosomes is consistent with receptor-mediated endocytosis being an active uptake pathway for these FA decorated nanofbers. The presence of intact nanofbers and fragments in the cytosol, as well as the enrichment of fragments inside endosomes raises questions about the endosomal escape of the materials, as well as other potentially active endocytosis mechanisms that may be operating. Transmembrane penetration by passive diffusion of nanoparticles has been reported, 89 and may be a second internalization pathway in operation for these nanofbers, given the uptake detected at 4 C via CLSM. Statistical analysis of the lengths of the intact nanofbers that were observable inside a single cell revealed 446 individual nanofber-like objects (excluding fragments), with lengths averaging 115 nm (L w /L n ¼ 1.12, s L ¼ 40 nm, Fig. S24 and Table S2 †). This length is very similar to that for FA-BD-PEG-BD-FA pentablock nanofbers before cellular experiments (95 nm, L w /L n ¼ 1.17, s L ¼ 39 nm), providing further evidence for their identity. The fbers and fragments observed via TEM correlate with the intracellular localization of BD fluorescence observed via CLSM for CLEM experiments (Fig. 7B), and suggest that FA-BD-PEG-BD-FA pentablock nanofbers interact with the cell membrane, leading to receptor-mediated endocytosis, and nanofber fragmentation. Dalhaimer et al. 90 also observed fragmentation of multimicrometer long polymeric worm-like micelles upon cellular uptake, although the resulting fragments were still up to ca. 500 nm in length. The number of intact anisotropic particles observed (446) also provides a rough indication for how many nanofbers may be uptaken by an individual cell. If nanofber fragmentation is taken into consideration, this number is likely to be higher. Observations of membrane-bound FA-BD-PEG-BD-FA pentablock nanofbers during experiments at 4 C revealed a larger than expected fraction of nanofbers interacting with the cell membrane through end-on interactions (several examples are circled in blue in Fig. 6E and H, and further cases in Fig. S20 †). This end on interaction mode should be statistically of low frequency if particle anisotropy had no effect on membrane binding. Thus, these images suggest that FA-BD-PEG-BD-FA pentablock nanofbers favor interaction with the cell membrane through an end-on binding mode. Analysis of the entry angle observed between FA-BD-PEG-BD-FA pentablock nanofbers and the cell membrane observed via TEM revealed two distributions, centered around 90 and 165 respectively (Fig. 8 and S25 †). These correspond to nanofbers which are either 'end-on' ($90 , Fig. S25A †) or 'side-on' ($165 , Fig. S25C †). Such modes have been investigated previously, both theoretically (for generic rod-like nanoparticles), 19,20 and experimentally (for carbon nanotubes), 77 and have been reported to be facilitated through increased membrane wrapping of 1D materials, owing to the effects of stiffness, length, and aspect ratio on the uptake mechanism. Analogous CLEM experiments where HeLa cells expressing GRASP65-GFP were incubated with FA-BD-PEG-BD-FA pentablock nanofbers for 90 minutes at 22 C (after association at 4 C) revealed electron rich fragments and intact nanofbers located around the perinuclear region via TEM micrographs of the resulting cell slices (Fig. S26 and S27 †), which correlated with the fluorescence observed via CLSM when data is overlaid (Fig. S26B and C †). The localization of these intact nanofbers and objects via TEM correlates with the fluorescence observed from BD around the perinuclear region, and supports the hypothesis that the anisotropic particles correspond to FA-BD-PEG-BD-FA pentablock nanofber fragments. Finally, to confrm that these results were not due to interference from the presence of the BD dye, FA-PEG-FA triblock nanofbers (100 mg mL 1 or 500 mg mL 1 , L n ¼ 90 nm, L w /L n ¼ 1.11, s L ¼ 30 nm) were incubated with HeLa cells for 5 minutes and 75 minutes respectively, prepared as before, and imaged via TEM (Fig. S28 and S29 †). Whilst some intact nanofbers were observed in TEM micrographs of HeLa cells exposed to 100 mg mL 1 of FA-PEG-FA triblock nanofbers for 5 minutes (Fig. S28, † circled red), fragments and clusters were also observed throughout the cell (Fig. S28, † circled blue), including inside endosomes/lysosomes (Fig. S28, † circled green). Results from HeLa cells exposed to 500 mg mL 1 of FA-PEG-FA triblock nanofbers for 75 minutes also revealed intact nanofbers and fragments throughout the cell, as well as inside endosomes/ lysosomes (Fig. S29 †). There appeared to be an increase in the number of fragments present, which is in accordance with the higher concentration of FA-PEG-FA triblock nanofbers leading to a higher number of fragments inside the cell. Analysis of the small electron-rich fragments present in HeLa cells exposed to 100 mg mL 1 of FA-PEG-FA triblock nanofbers for 5 minutes revealed a length of 26 nm (L n ¼ 26 nm, L w /L n ¼ 1.09, s L ¼ 8 nm) and a width of 12 nm (W n ¼ 12 nm, W w /W n ¼ 1.04, s W ¼ 3 nm, Fig. S30 †), in close agreement with the measured width of FA-PEG-FA triblock nanofbers (W n ¼ 12 nm, W w /W n ¼ 1.02, s W ¼ 2 nm). In summary, our results indicate that FA functionalized PDHF 13 -b-PEG 227 nanofbers appear to interact with the cell membrane at either a 90 (perpendicular) or 165 (parallel) angle of contact, with perpendicular fbers appearing to undergo cellular internalization whilst parallel fbers are either not internalized, or shift to a perpendicular orientation before entering the cell. 77 Upon cellular internalization, we hypothesize that some of the nanofbers fragment into $20 nm long particles. Both intact nanofbers and fragments were observed inside cells, as well as intact nanofbers inside endosomes, indicating receptormediated endocytosis is one active uptake mechanism, but passive diffusion may also be operational. Localization primarily occurs to the perinuclear region however nanofbers and fragments were observed throughout the cell. ## Discussion Precision functional, modular PDHF nanofbers for biomedical applications In this work we have demonstrated the formation of 95 nm pentablock co-micelles with an average segment length of only 19 nm, which is close to the lower limit for the lengths of 1D nanomaterials produced via living CDSA to date. This represents the highest density of segments produced in a polymer nanofber to date. In principle, the highly modular nature of the synthetic route to end-group modifcation of the PDHF-b-PEG polymer allows for a diverse range of targeting groups, imaging agents, and cargo such as drugs to be incorporated into the nanofbers. As a proof of concept, we have taken FA; one of the most well-studied ligands for targeted drug delivery to cancer cells. We have produced two sets of PDHF 13 -b-PEG 227 nano-fbers: those with FA, and those lacking FA. Both nanofbers have a neutral PEG corona which should confer the nanofbers with 'stealth' properties. Overall, these results show that complex nanomaterials can be prepared using living CDSA on a length-scale appropriate for biological applications and provide a method for precisely tailoring surface chemistry in small nanoparticles. ## Cellular uptake of untargeted vs. targeted PDHF-b-PEG nanobers Initial CLSM experiments on PDHF 13 -b-PEG 227 nanofbers lacking FA revealed that, while no discernable cytotoxicity was detected (see ESI Page S2, Fig. S31, S32 and Tables S3-S6, see ESI † for results and discussion), no cellular uptake was observed over a 1 h period either. This result was reinforced by flow cytometry experiments, which revealed basal levels of BD emission, on a par with untreated control cells. One plausible explanation is that a longer time period is required before signifcant uptake will be observed, as the internalization of neutral PEG-coated gold nanorods was observed to occur over a 24 h period, 95 though uptake after 24 h was only 2% of the total added. CLSM indicated that FA-mediated nanofber uptake occurs within 30 minutes, leading primarily to localization in the perinuclear region. Flow cytometry allowed for a comparison of the uptake efficiency of targeted vs. untargeted nano-fbers, with >99% of HeLa cells exhibiting uptake of FA decorated nanofbers, versus <1% of HeLa cells for those lacking FA. ## Intracellular fate of folic acid decorated PDHF-b-PEG nanobers Analysis of TEM images of FA-BD-PEG-BD-FA pentablock nanofbers in the region of the cellular membrane of HeLa cells at 4 C revealed a larger than expected number of nanofbers that interact with the cell membrane in an 'end-on' (perpendicular) and 'side-on' (parallel) fashion. The experimental results obtained here are consistent with theoretical and experimental studies by Shi, 77 and Möller 21 et al., where cellular internalization of rod-like nanoparticles appears to occur frstly via association with the tip of the fber, followed by rotation to a 90 (perpendicular) angle of contact that is driven by a relaxation in elastic energy in the cell membrane. Our results concur with those of Shi et al., where only nanofbers with high angles of contact are observed undergoing cellular internalization. The observation of nanofbers with 'end-on' membrane interactions and curved, flexible tails (Fig. S20 †) supports the proposed transition from 'side-on' to 'end-on' before internalization. 'End-on' internalization is presumably further driven by the segmented block-like structure of the nanofbers, where the FA targeting group is located solely at the fber ends, facilitating cellular uptake. CLEM studies of HeLa cells incubated with FA-BD-PEG-BD-FA pentablock nanofbers at 4 C revealed intact nano-fbers as well as small, high contrast 'fragments' in the immediate region around the cellular membrane and throughout the cell via TEM, which correlated with BD fluorescence observed in CLSM. Some of the nanofbers were found within endosomes, indicating that receptor-mediated endocytosis is an active uptake mechanism, consistent with other FA containing nanomaterials. 23,25,26 The observation of intact nanofbers and fragments outside of endosomes via TEM, and the presence of intracellular BD fluorescence at 4 C via CLSM indicates that another uptake mechanism may also be present, such as passive diffusion. CLEM and TEM experiments involving nanofbers both with and without the BD dye reveal similar intact nanofbers and high contrast fragments observed within the cell and around the nuclear membrane. Taken together, we hypothesize that fragmentation/ disassembly of the nanofbers occurs upon cellular internalization, with subsequent localization primarily in the perinuclear region. Nanofber fragmentation would also ft with the observed fluorescence quenching of the PDHF core (Fig. S33, see ESI † for results and discussion), as it could be imagined that the forces driving nanofber cleavage might involve interaction of the p-conjugated core with species capable of causing fluorescence quenching. Fragmentation of the PDHF nanofbers upon cellular internalization would also be consistent with the behavior of FA-functionalized coordination complex nanotubes observed by Wang et al., 30 and of PEG-b-PCL flomicelles by Geng et al. 12 raising the possibility that this may be a more general consequence of the cellular internalization of 1D nanomaterials with specifc properties. 96 Nanofber fragmentation currently remains a hypothesis, however, as further studies are required to probe and confrm this phenomenon. As many questions remain regarding the cellular internalization and localization of FA targeted PDHFb-PEG nanofbers, future work will focus on probing this process in more detail. ## Summary Using the living CDSA approach, we have developed colloidally stable, hydrophilic segmented 1D nanofbers with a crystalline pconjugated PDHF core, a 'stealth' PEG corona, and spatially confned functionality. Segmented pentablock nanofbers of length 95 nm were prepared through a seeded-growth process, which possess the highest density of different corona-forming blocks in a segmented nanofber to date. The development of nanofbers with length control, and the ability to easily present different functional groups in a modular, controlled fashion over length scales relevant to biomedical applications represents a potentially signifcant advance. In the absence of targeting groups, the nanofbers were not capable of being internalized by HeLa cells after 1 h, however cell uptake was detected by CLSM and flow cytometry within 30 minutes for nanofbers functionalized in the terminal segment with FA, which binds to folate receptors that are overexpressed in cancer cells such as the HeLa cell line examined here. Nanofbers without FA were uptaken into <1% of HeLa cells, in contrast with > 97% uptake of FA decorated nanofbers. The lack of cellular uptake for nanofbers without FA implies that nanofbers bearing this moiety may act as a targeted diagnostic, preferentially undergoing internalization into cells that express folate receptors, such as those in tumors. FA decorated nanofbers were observed to undergo internalization into HeLa cells at 4 C, with some observed in the cellular membrane and others inside the cell. A signifcant number of membranebound nanofbers were observed to interact with the cell membrane in either an 'end-on' or 'side-on' fashion. Only 'endon' fbers were observed to undergo internalization, providing experimental evidence for the unique uptake mechanism of high aspect ratio 1D nanomaterials. Small, high contrast, anisotropic particles ($10 20 nm) were observed inside HeLa cells proximal to the cell membrane, leading us to hypothesize that FA-BD-PEG-BD-FA pentablock nanofbers may undergo fragmentation upon cellular internalization at 4 C. Analogous experiments at room temperature revealed similar particles throughout the cell, but concentrated around the perinuclear region. If the small, high contrast particles observed do correspond to nano-fber fragments, this would point towards a unique uptake and disassembly mechanism for this type of 1D material. Intact nanofbers were also observed throughout the cell, with examples of both nanofbers and fragments free in the cytosol, as well as inside endosomes. Examples of intact nanofbers inside endosomes indicate that receptor-mediated endocytosis is an active uptake mechanism for the FA decorated nanofbers. Overall, these results indicate that the nanofbers are capable of active targeting towards different cell lines, with minimal cellular uptake observed for those lacking an active targeting group. If nanofber fragmentation upon cellular internalization is confrmed, this could allow for the benefts of targeted 1D nanomaterials in vivo, whilst releasing smaller particles after cellular internalization that could have other, additional bene-fts (e.g. nuclear localization if fragment size can be tuned to <5 nm). These results also provide new insights into the cellular uptake of low dispersity 1D nanoparticles, revealing the potential for p-conjugated PDHF nanofbers to act as fluorescence turn-off sensors for cells rich in folate receptors. This work also provides valuable information on the uptake mechanism for anisotropic 1D polymer nanoparticles. Finally, the study indicates that the ability of 'living' CDSA to generate anisotropic polymer nanoparticles with near uniform dimensions and a segmented structure should facilitate further investigations of nanoparticle uptake into cells, and where features such as fber length, width, stiffness, and the spatial location and choice of targeting groups are varied. Analogous 1D nanoparticles also have the potential to deliver therapeutic cargoes, with relevant studies currently in progress.

chemsum

{"title": "Cellular uptake and targeting of low dispersity, dual emissive, segmented block copolymer nanofibers", "journal": "Royal Society of Chemistry (RSC)"}

equilibrium_constants_and_protonation_site_for_<i>n</i>-methylbenzenesulfonamides

## Abstract: The protonation equilibria of four substituted N-methylbenzenesulfonamides, X-MBS: X = 4-MeO (3a), 4-Me (3b), 4-Cl (3c) and 4-NO 2 (3d), in aqueous sulfuric acid were studied at 25 °C by UV-vis spectroscopy. As expected, the values for the acidity constants are highly dependent on the electron-donor character of the substituent (the pK BH+ values are −3.5 ± 0.2, −4.2 ± 0.2, −5.2 ± 0.3 and −6.0 ± 0.3 for 3a, 3b, 3c and 3d, respectively). The solvation parameter m* is always higher than 0.5 and points to a decrease in the importance of solvation on the cation stabilization as the electron-donor character of the substituent increases. Hammett plots of the equilibrium constants showed a better correlation with the σ + substituent parameter than with σ, which indicates that the initial protonation site is the oxygen atom of the sulfonyl group. ## Introduction Having a knowledge of the protonation equilibrium constants for N-methylbenzenesulfonamides 3 is fundamental to achieve a correct understanding of their reactivity, that is to say that the referred constants can be used to estimate the values of the protonation constants for N-methyl-N-nitrosobenzenesulfonamides 1. This information, not yet experimentally available, is of crucial importance in the studies of the nitroso-group transfer mechanism from 1. Such compounds react with a variety of nucleophiles: In the presence of HO − or EtO − , which attack their SO 2 group, decomposition to afford diazomethane occurs . In acidic medium, they undergo denitrosation to the corresponding N-methylbenzenesulfonamides 3 , as is common with other N-nitrosamines. However, unlike with nitrosamines and nitrosoureas, nucleophilic attack by amines at the N=O group affords nitrosamines 4 (Scheme 1). They are also known to be capable of nitroso-group transfer to form nitrosyl complexes . Increasing attention is being paid to the chemistry of nitrosamines owing to the toxicity and carcinogenic , mutagenic , and teratogenic properties of these compounds. The acidity of organic molecules is one of the most relevant factors determining their reactivity. Nevertheless, the values of the protonation and deprotonation equilibrium constants are generally difficult to obtain. This is due to the difficulties in the definition of the acidity scales and in the interpretation of the experimental data. In diluted acid, pK BH+ can be easily evaluated by measuring the ionization ratio I = [BH + ]/[B] and the proton concentration in the medium. However, in strongly acidic solutions, the ability of the medium to protonate a weak base largely exceeds the formal concentration of hydronium ions, due to the mediuminduced effects in the activity coefficients of the different species involved in the equilibrium. Historically, there were two approaches to the analysis of such effects in strongly acidic media. The first approach emphasizes the acidity of the medium and is derived from Hammett's approach, proposed in 1932 in order to achieve an acidity measure contiguous to the pH scale, defined for dilute aqueous solutions. With this purpose, Hammett defined the so-called "Hammett acidity function", H 0 , which is no more than a measure of the deviation, relative to ideality, provoqued by the changes in the medium as the acid concentration increases. Time has proved that Hammett's methodology is only applicable to similar classes of compounds . In reality, during the 1950's, a variety of acidity constants for different kinds of bases, such as tertiary amines (H 0 ''') , amides (H a ) , carbinoles (H R+ ) , and indoles (H I ) , among others , were developed. The second approach to the problem considers that variations in the equilibrium or rate constants in aqueous acidic mixtures may be described by a free-energy linear correlation. This approach was developed by Bunnett and Olsen according to the suggestion of Grunwald and Kresge , has been broadly used , and was reviewed by Bagno, Scorrano and More O'Ferrall in 1987 . In order to use Grunwald's formalism to account for the effects of the medium on acid-base equilibria, a reference equilibrium must be chosen, to which the dependence on the acidity of any other equilibrium is compared. In Equation 1, K and K* are, respectively, the equilibrium constants of the reaction under study and of the reference reaction, and δ M accounts for the effects of changes in the medium (i.e., in the concentration of the strong acid). (1) Equation 1 may be rewritten in the more familiar form of Equation 2, where K 0 and K 0 * are the equilibrium concentration ratios in a reference solvent, which in the case of reactions in aqueous acidic media is normally water. ( If we denote log K*/K 0 * as -X, this equation becomes (3) where K c is the experimental classical ionization constant and K BH+ the thermodynamic ionization constant in water. According to the interpretation of Bagno and Scorrano , m* is a measure of the cation (the protonated base, BH + ) solvation, that is, a solvation coefficient. So, the strength of a weak base is determined by its pK BH+ in the reference solvent, usually water, and by its solvation coefficient in acidic medium. These parameters are the intercept and the slope of Equation 3. The choice of water as the reference solvent and of 4-nitroaniline as reference base, renders m* = 0 for the pair H 3 O + /H 2 O and m* = 1 for pairs formed by anilinium ions and the respective aniline. ## Results and Discussion The determination of the classical equilibrium constant, K c , requires knowledge of the ionization ratio Usually this is obtained by UV-vis spectroscopic measurements, as I relates to the absorbance according to Equation 4(4) where A, A B and A BH+ are the absorbances of the solution, of the free base and of its conjugated acid, respectively. Figure 1 presents the spectra of the four benzenesulfonamides (3a-d) under study, in which a visible change occurs as the substrates protonate. The most striking observation related to the above spectra is the absolute lack of isosbestic points, which arises from the shift in the n → π* absorption bands of the sulfonamides as the acid concentration increases. In order to eliminate this effect, the spectra must be treated by the characteristic vectors analysis (CVA) method . This analysis requires the construction of a matrix of absorbances at different wavelengths and different acid concentrations, from which an average absorbance matrix and a number of characteristic vectors that allegedly contain all the information of the original data are obtained (Equation 5). (5) In most cases, the original data are reproduced with 99% accuracy from two vectors only, in which the first accounts for 94-96% of the variation and the second for the remaining 3-6%. Based on our chemical intuition, we associate the first to the protonation process and the second to the medium effect . Figure 2 shows the spectra obtained after application of the CVA method (considering that the protonation effect is given by the ν 1 vectors). The data was treated according to Simonds original algorithm implemented on Mathcad . The values for the ionization ratio are determined from Equation 4. The composition of the sulfuric acid solution when I = 1 that corresponds to a degree of protonation of 50% can be easily calculated and is namely 65.2, 68.2, 74.0 and 80.6% sulfuric acid (w/w) for compounds 3a, 3b, 3c and 3d, respectively. [H + ] and X values for each sulfuric acid concentration were calculated by interpolation of values from reference . Since 3). From the results presented in Table 1 it is evident, as expected, that there is an increase in the acidity constant with the electronwithdrawing character of the substituents. The solvation parameter m* is higher than 0.5 in all cases and also increases with the electron-withdrawing character of the substituents in the ring, which indicates a decrease in the solvation degree . These results allow us to make some conjectures about the protonation site. Considering that the SO 2 group prevents resonance between the nitrogen atom and the ring, the dependence of the acidity constant on the electronic character of the substituents seems too overwhelming to support protonation on the nitrogen atom. Being so, it is more likely that the protonation occurs on the sulfonyl oxygen atom, as such a structure may present resonance with the electron-donor substituents (Scheme 2). The fact that pK BH+ correlates better with σ + (R = 0.9913) than with σ (R = 0.9681) also indicates protonation on the oxygen atom (Figure 4). Nevertheless, the curvature of the σ Hammet plot could be ascribed to a change in the protonation site from oxygen, on the compounds carrying the more electron-donating substituents, to nitrogen, for those with the more electron-withdrawing substituents. However, if this were the case, the curvature in the correlation with σ + would be more pronounced. Moreover, the solvation parameter m* values found also seem to be compatible with oxygen protonation, since for oxygen bases these values range from 0 to 0.7 but for nitrogen bases lie around unity . In fact, although Menger and Mandell concluded that N-methyl-5-chloro-1,2-benzisothiazoline 1,1-dioxide in fluorosulfonic acid protonated on the nitrogen atom, Chardin and co-workers showed that protonation of sulfonamides occurred on the oxygen atom. Still, a possibility that should not be discarded is the existence of a tautomeric equilibrium between the N-and O-protonated structures, the latter having a greater relevance for the sulfonamides with electron-donor groups (Scheme 3). Scheme 3: Tautomeric equilibrium between N-and O-protonated forms of N-methylbenzenesulfonamides, 3. ## Conclusion The protonation equilibrium constants (pK BH+ ) for the parasubstituted N-methylbenzenesulfonamides 3a-d in aqueous sulfuric acid were obtained from spectrophotometric measurements. Treatment of the spectra by the characteristic vectors analysis (CVA) method, in order to compensate for the shift in the n → π* absorption bands of the sulfonamides as the acid concentration increases, was necessary. The values obtained were seen to increase with the electron-withdrawing character of the substituents. The solvation parameter (m*) values point to a decrease in the degree of solvation as the electron-withdrawing character of the substituents increases and to protonation on the oxygen atom. The correlation between pK BH+ and σ + also indicates oxygen protonation, although the existence of a tautomeric equilibrium between the N-and O-protonated forms cannot be ruled out. ## Experimental Synthesis of N-methylbenzenesulfonamides The N-methylbenzenesulfonamides 3a-d were prepared from the reaction of the parent benzenesulfonyl chlorides with methylamine . ## Preparation of acid solutions Acid solutions were always prepared by weighing the appropriated amount of commercial H 2 SO 4 (98%, Aldrich), which was then carefully diluted in water, and small aliquots of the mixture were then titrated with NaOH solution. The resulting molarities were converted to weight percents by using the conversion table published in the CRC Handbook of Chemistry and Physics . The concentrations of the acid solutions were double checked by measuring the densities of the solutions. All dilutions were made in an ice bath, with careful mixing to prevent the risk of a sudden temperature rise. The solution was then allowed to stand in a water bath at 20 °C and the final volume in the volumetric flask was adjusted. ## Spectroscopic measurements Solutions of 3a-d (5.0 × 10 −5 M) were prepared by adding a small amount, typically 30 µL, of a stock solution to 10 mL of the sulfuric acid solution. UV spectra were recorded in a Varian Cary 100 equipped with a thermostated cell holder. All measurements were made in quartz cells with a 1 cm light path, at 25 °C, and the spectra were run against a solution with the same concentration of sulfuric acid as that of the N-methylbenzenesulfonamide solution.

chemsum

{"title": "Equilibrium constants and protonation site for <i>N</i>-methylbenzenesulfonamides", "journal": "Beilstein"}

organocatalytic_cascade_aza-michael/hemiacetal_reaction_between_disubstituted_hydrazines_and_α,β-uns

## Abstract: The catalytic synthesis of nitrogen-containing heterocycles is of great importance to medicinal and synthetic chemists, and also a challenge for modern chemical methodology. In this paper, we report the synthesis of pyrazolidine derivatives through a domino aza-Michael/hemiacetal sequence with chiral or achiral secondary amines as organocatalysts. Thus, a series of achiral pyrazolidine derivatives were obtained with good yields (up to 90%) and high diastereoselectivities (>20:1) with pyrrolidine as an organocatalyst, and enantioenriched pyrazolidines are also achieved with good results (up to 86% yield, >10/1 regioselectivity, >20:1 dr, 99% ee) in the presence of (S)-diphenylprolinol trimethylsilyl ether catalyst. ## Introduction Pyrazolidines are privileged and valuable heterocyclic compounds, which are of great importance in biological and medicinal chemistry (Figure 1) . Besides, pyrazolidines are also important synthetic intermediates in organic chemistry. For instance, the N-N bond of pyrazolidines can be cleaved under reductive conditions to afford useful 1,3-diamines , and moreover, pyrazolidines can also be oxidized to afford pyrazolines and pyrazoles . The pyrazolidine structural unit is commonly constructed by [3 + 2] cycloaddition reactions using hydrazones or azomethine amines as dipoles. Recently, the Ma group and Toste et al. have reported efficient methods for the synthesis of pyrazolidine derivatives by metal-catalyzed aminations of allenes . Meanwhile, Lewis acid catalyzed carboamination reactions have also been reported as efficient methods for the synthesis of pyrazolidine derivatives by Wolfe et al. . In the past decade, the research area of organocatalysis has grown rapidly and become a third brand of catalysis besides the well-established biocatalysis and metal catalysis . Particularly, organocatalytic domino/cascade reactions have come into focus and become a powerful synthetic approach that allows the construction of structurally diverse and complex molecules, minimizes the number of manual operations, and saves time, effort, and production costs . Thus, many nitrogen-containing heterocyclic compounds have been efficiently generated by means of organocatalytic domino reactions . In 2009 and 2010, List et al. and the Brière group both reported, separately, the enantioselective synthesis of 2-pyrazolines starting from α,β-unsaturated ketones and phenylhydrazine or N-tert-butyloxycarbonylhydrazine in the presence of a chiral Brønsted acid or a phase-transfer catalyst . Compared with monosubstituted hydrazines in organocatalytic asymmetric synthesis, disubstituted hydrazines were also explored by several groups . In 2007, Jørgensen et al. reported that the organocatalyzed asymmetric aza-Michael addition of hydrazones to cyclic enones had been achieved in good yield and stereoselectivity . In 2011, the Deng group developed a highly enantioselective organocatalytic synthesis of 2-pyrazolines using disubstituted hydrazines through an asymmetric conjugate addition followed by a deprotection-cyclization sequence . Due to the importance of pyrazolidine derivatives in both organic and medicinal chemistry, we have become interested in developing an efficient stereoselective cascade reaction for the synthesis of the pyrazolidine compounds through organocatalysis. In this paper, we present a convenient access to racemic and enantioenriched 5-hydroxypyrazolidines through a domino aza-Michael/hemiacetal organocatalytic sequence of disubstituted hydrazines to α,β-unsaturated aldehydes. While proceeding with the submission of our results, we noticed that a related excellent work has been reported by Vicario and co-workers . When comparing Vicario's work with this manuscript, both works are complementary in scope since Vicario's work makes use of unsaturated aldehydes containing only linear alkyl chains, whereas our work provides better results when unsaturated aldehydes bearing an aromatic moiety are employed. ## Results and Discussion First, the cascade aza-Michael/hemiacetal reactions between disubstituted hydrazines 2a-c and 4-nitrocinnamaldehyde (3a) were investigated in the presence of several common secondary amines 1a-f as organocatalysts in chloroform. Pyrrolidine (1c) turned out to be an effective catalyst and di-tert-butyl hydrazine-1,2-dicarboxylate (2c) was a potent donor (see Table S1 in the Supporting Information File 1). When di-tert-butyl hydrazine-1,2-dicarboxylate (2c) was used as the donor and 20 mol % pyrrolidine (1c) as the catalyst, 5-hydroxypyrazolidine 4a could be obtained in 68% yield with over 20:1 dr after five days (Table 1, entry 1). Thus, the tandem aza-Michael/ hemiacetal reaction between di-tert-butyl hydrazine-1,2-dicarboxylate (2c) and 4-nitrocinnamaldehyde (3a) was chosen as the model reaction to further optimize the reaction conditions. The catalytic results were summarized in Table 1. In order to improve the yield, a variety of common solvents were screened (Table 1, entries 2-6). Dichloromethane was finally found to be the best medium for this reaction (74% yield with excellent diastereoselectivity was achieved, Table 1, entry 2). Subsequently, the effects of some basic/and acidic additives were examined. Inorganic bases seemed to be ineffective for the further improvement of the yields (Table 1, entries 7-10). Soluble organic bases were also tested and failed to increase the yields (Table 1, entries 11 and 12). When benzoic acid (5a) was used as an additive, the yield was slightly improved to 77% (Table 1, entry 13). With the increase of the acidity, the yield was noticeably decreased (Table 1, entry 13 versus entries 14 and 15). Finally, when increasing the amount of di-tert-butyl hydrazine-1,2-dicarboxylate (2c) to 2 equiv, the desired product 4a was obtained in 81% yield without any additive (Table 1, entry 16). Having established the optimized reaction conditions, we investigated the scope of substrates for the cascade aza-Michael/hemiacetal reaction with pyrrolidine (1c, 20 mol %) as the catalyst, without any additives, in methylene chloride (Table 2). We found that the nature of the substituents on the phenyl group of α,β-unsaturated aldehydes dramatically affected the reactivity. For example, with disubstituted hydrazine 2c as the donor, the presence of a stronger electron-deficient substituent (-NO 2 , -CN) on the phenyl ring of the α,βunsaturated aldehydes 3a, 3b and 3c promoted the cascade aza-Michael/hemiacetal reaction readily to provide the desired products in good yields (75-81%, Table 2, entries 1-3). The presence of less-electron-deficient substituents (-Cl and -Br) on the phenyl ring of the α,β-unsaturated aldehydes, such as 4-chloroor 4-bromocinnamaldehyde derivatives 3d and 3e, afforded the corresponding products in moderate yields (64 and 74%) even when 5 equiv of disubstituted hydrazine 2c was used (Table 2, entries 4 and 5). When cinnamaldehyde derivatives 3f and 3g, bearing electron-donating substituents (-Me, -OMe) on phenyl rings, were used as the substrate, the reactions became very sluggish (Table 2, entries 6 and 7). On the other hand, more symmetric and asymmetric hydrazine derivatives 2d-h were synthesized and investigated for the tandem aza-Michael/hemiacetal reaction. Generally, all the reactions between 2d-h and α,β-unsaturated aldehydes 3a and 3b proceeded smoothly with sequential catalytic actions of 1c, affording the corresponding desired products 4h-o in moderate to good yields (Table 2, entries 8-11 and 13-16). Notably, the reaction between 2e and 4-methoxycinnamaldehyde (3f) afforded the desired product in 52% yield (Table 2, entry 12). For the asymmetric disubstituted hydrazines 2g and 2h as substrates, regioselective results could be observed (Table 2, entry 14). However, with increasing catalyst loading from catalytic to stoichiometric amounts, the corresponding cascade reactions could provide the major products in 72 and 66% yields (Table 2, entries 15 and 16). In all the reactions, high diastereoselectivity could be obtained (>20:1 dr). Finally, we were able to obtain single crystals of compounds 4a and 4n, which allowed for an unambiguous assignment of the trans configuration of C3 and C5 by X-ray crystallographic analysis (Figure 2 and Figure 3). were found to be ineffective for the reaction, because they afford only trace products after one day (Table 3, entries 1, 2, 5 and 6). Although a moderate yield was obtained with organocatalyst 1i bearing a sulfone functional group, the stereochemical induction was very poor (Table 3, entry 3). MacMillan's cata- lyst 1j was proven to be inefficient for this transformation as only 13% ee was obtained (Table 3, entry 4). Subsequently, three diarylprolinol silyl ether catalysts 1m-o were investigated for this tandem reaction . Gratifyingly, 82% ee was achieved when 1m was used as the catalyst. For catalysts 1n and 1o, slightly higher yields could be obtained, but the enantioselectivities became lower (Table 3, entry 7 versus entries 8 and 9). Relatively speaking, (S)-diphenylprolinol trimethylsilyl ether 1m turned out to be the optimal catalyst in terms of both enantioselectivity and reactivity. Having identified the readily available catalyst 1m as the optimal catalyst for the tandem aza-Michael/hemiacetal reaction of 2c and 3a, we summarize the results for the optimization of the other reaction parameters, including reaction solvents and additives, in Table 4. When the reaction was carried out in a protonic solvent, i.e., methanol, at room temperature for two days, the desired product was furnished in a The reaction was run with 2c (0.5 mmol), 3a (0.25 mmol), 1m (0.05 mmol) and the specified additive (0.05 mmol) in the given solvent (0.5 mL) at room temperature. b Isolated yield. c Determined by HPLC analysis on a chiral stationary phase (Chiralcel OD-H), >20:1 dr. 56% yield with only 22% ee (Table 4, entry 1). After screening several aprotic solvents for this reaction, we were pleased to find that the enantioselectivity of the desired product was improved to 92% ee with toluene or THF as solvent after two days (Table 4, entries 4 and 5). Considering both yield and enantioselectivity, toluene was the optimal reaction medium (Table 4, entry 5). When the time was prolonged to four days, the yield of the product was increased to 80% and the enantioselectivity was retained (Table 4, entry 6). Thereafter, several Brønsted acids 5b, 5d-j were tested as additives for this transformation. Although enantioselectivity was somewhat improved from 92 to 95% ee, the reactivity dramatically decreased as evidenced by the prolonged reaction time and lower yields (Table 4, entry 6 versus entries . It seemed that the present catalytic system could be inhibited by acidic additives. Therefore, we considered whether the reaction could be accelerated by basic additives without loss of enantioselectivity and reactivity. Subsequently, several common inorganic and organic bases were a Unless noted, the reaction was run with 2 (0.5 mmol), 3 (0.25 mmol), and 1m (0.05 mmol) in toluene (0.5 mL) at room temperature. b Isolated yield of pure isomer 4 (the data in parentheses is related to the isolated yield of the 4'). c The ratio based on isolated yield of pure 4 and 4'. d Determined by HPLC analysis on a chiral stationary phase (Chiralcel OD-H, AD-H or AS-H), >20:1 dr. e The ratio of 2/3 is 1.2:1. f The reaction was run with 2 (0.25 mmol), 3 (0.38 mmol), and 1m (0.05 mmol) in toluene (0.5 mL) at room temperature. g Due to the difficulty of separation of the product 4r' from starting material 2h. investigated . Unfortunately, the catalytic results showed that with LiOAc, DMAP, DABCO, Et 3 N, TMEDA as additives, the yield and enantioselectivity were only marginally influenced (Table 4, entries . When DBU was used as an additive, only 11% ee was obtained with moderate yield (Table 4, entry 20). Thus, 1m (20 mol %) as the catalyst and toluene as the reaction medium without any additive at room temperature proved to be the optimal reaction conditions for the asymmetric cascade aza-Michael/hemiacetal reaction. With the optimized reaction conditions in hand, the substrate scope of the organocatalyzed asymmetric domino aza-Michael/ hemiacetal sequence was subsequently explored. Firstly, with symmetric di-tert-butyl hydrazine-1,2-dicarboxylate (2c) as nucleophilic reagent, aromatic α,β-unsaturated aldehydes 3a-g were examined to study the effects of electronic properties and steric hindrance on both enantioselectivity and reactivity. For the substrates 3a, 3b and 3c, bearing substituents of -NO 2 and -CN at the paraor meta-position of the phenyl group, the reactions proceeded smoothly and led to the desired products 4a, 4b and 4c in 80-86% yields with 89-92% ee's (Table 5, entries 1-3). With 3d and 3e bearing -Cl or -Br substituents at the para-position of the phenyl group as substrates, the desired products 4d and 4e were obtained in 61 and 62% yields with 74 and 77% ee, respectively (Table 5, entries 4 and 5). For substrates 3f and 3g bearing electron-donating groups (-Me, -OMe) on the phenyl rings, only a trace amount of the desired products could be observed under otherwise identical reaction conditions (Table 5, entries 6 and 7). These experimental results indicated that chemical yields and enantioselectivities were dramatically affected by the electronic properties and steric hindrance of the aryl group on the α,β-unsaturated aldehydes. High yield and good enantioselectivity could be obtained with strong electron-withdrawing substituents on the phenyl ring of cinnamaldehydes. When diisopropyl hydrazine-1,2-dicarboxylate (2d) as nucleophilic reagent reacted with 4-nitro cinnamaldehyde (3a), the product 4h was obtained in 60% yield and 72% ee (Table 5, entry 8). The result showed that the smallsized substituent on hydrazines was unfavorable on the reaction (Table 5, entry 8 versus entry 1). Subsequently, asymmetric disubstituted hydrazines 2g-j were investigated for the domino aza-Michael/hemiacetal sequence. Due to nucleophilic competition of the two nitrogens in the asymmetric disubstituted hydrazines, regioselective results were observed for these reactions. For asymmetric disubstituted hydrazines 2g, the reaction gave the 1.8:1 molar ratio of the regioselective products 4n to 4n'. The major product 4n was obtained in 58% yield and 88% ee. The enantioselectivity of the minor product 4n' was 55% (Table 5, entry 9). For asymmetric disubstituted hydrazines 2h-i, the molar ratios of the regioselective products ranged from 3.2:1 to 6.4:1. The major products were obtained in moderate to good yields and good enantioselectivities (Table 5, entries 10-15). When disubstituted hydrazine 2j with an electron-donating group on the aromatic ring was used as the nucleophilic donor, the major, reversely regioselective product 4u' was obtained in 75% yield, but with very low enantioselectivity (11% ee, Table 5, entry 16). To our delight, (E)-but-2-enal (3i) was a suitable substrate for this transformation. The reactions of asymmetric disubstituted hydrazines 2h and 2i with (E)-but-2-enal (3i) proceeded smoothly and provided the desired products in 78 and 83% yields with 72 and 74% ee, respectively (Table 5, entries 17 and 18). However, when the cascade aza-Michael/hemiacetal reaction of (E)-pent-2-enal (3j) and 2h was carried out, only a trace amount of the expected products could be detected (Table 5, entry 19). Fortunately, single crystals of compound 4s were obtained by recrystallization from petroleum ether/acetyl acetate, and the absolute configuration was determined by X-ray analyses (Figure 4) . ## Conclusion In summary, we have developed an organocatalytic approach for the synthesis of pyrazolidine derivatives through the cascade aza-Michael/hemiacetal reaction between disubstituted hydrazines and α,β-unsaturated aldehydes. The asymmetric version of this one-pot cascade reaction has also been realized with (S)-diphenylprolinol trimethylsilyl ether 1m as a secondary amine organocatalyst, and a series of enantiomerically enriched pyrazolidine derivatives were obtained in moderate to good chemical yields with moderate to excellent enantioselectivities. The application of the products and further investigation of the reaction are ongoing in our laboratory.

chemsum

{"title": "Organocatalytic cascade aza-Michael/hemiacetal reaction between disubstituted hydrazines and \u03b1,\u03b2-unsaturated aldehydes: Highly diastereo- and enantioselective synthesis of pyrazolidine derivatives", "journal": "Beilstein"}

on_the_electro-oxidation_of_small_organic_molecules:_towards_a_fuel_cell_catalyst_testing_platform_b

## Abstract: The electrocatalytic oxidation of small organic compounds such as methanol or formic acid has been the subject of numerous investigations in the last decades. The motivation for these studies is often their use as fuel in so-called direct methanol or direct formic acid fuel cells, promising alternatives to hydrogen-fueled proton exchange membrane fuel cells. The fundamental research spans from screening studies to identify the best performing catalyst materials to detailed mechanistic investigations of the reaction pathway. These investigations are commonly performed at conditions quite different to fuel cell devices, where no liquid electrolyte will be present. We previously developed a gas diffusion electrode setup to mimic "real-life" reaction conditions and study electrocatalysts for oxygen gas reduction or water splitting. It is here demonstrated that the setup is also suitable to investigate the properties of catalysts for the electro-oxidation of small organic molecules simulating conditions of low temperature proton exchange membrane fuel cells. Methanol oxidation; formic acid oxidation; direct methanol fuel cell; direct formic acid fuel cell; gas diffusion electrode setup; ## Introduction Energy conversion and storage are the most prominent applications of electrochemistry. Electrochemical energy conversion is required to use electric energy for producing fuels and chemicals, e.g., in water electrolysis or carbon dioxide electroreduction, but also to transform fuels to electric energy. The most common fuel thereby is gaseous hydrogen, which can be converted into electric energy using proton exchange fuel cells (PEMFCs). Hydrogen powered PEMFCs are for example used in automotive applications where an extremely high-power density is required . The main challenges for the technology however are the lack of a hydrogen distribution system and the high costs . Liquid fuels such as methanol or formic acid by comparison are easier to distribute and store. However, operating PEMFCs with liquid fuels in so-called direct liquid fuel cells, i.e., direct methanol and direct formic acid fuel cells (DMFC and DFAFCs) , requires substantially larger amounts of the precious and rare catalyst materials to reach a given kW peak power demand due to the sluggish anode reaction in addition to the limited cathode performance. The sluggish anode reaction is mostly associated with the a limited complete oxidation of the fuel to CO2 . For example, methanol oxidation reaction is known to proceed via a CO intermediate which adsorbs to the catalyst surface and thus blocks (poisons) adsorption sites . As a consequence, the CO oxidation reaction becomes the limiting factor for methanol oxidation. By comparison, formic acid should facilitate the direct oxidation to CO2 via the removal of two the hydrogen atoms. Nevertheless, also for formic acid oxidation, CO poisoning has been reported . Despite these limitations, direct liquid fuel cells are an interesting alternative for applications where lower power densities are required. The first steps in catalyst development for PEMFCs are usually performed in rotating disk electrode (RDE) measurements in three-compartment electrochemical cells with liquid electrolyte . Such experiments are straightforward to set up, but for cathode catalysts promoting the oxygen reduction reaction (ORR), it has been observed that the extrapolation of the results to real applications or even measurements in single cell membrane electrode assemblies (MEA) is challenging . This is mostly due to different mass transport conditions, but the electrolyte environment in PEMFCs is also considerably different than in RDE measurements. For this reason, several research groups developed experimental approaches for catalyst testing that aim to establish a second testing platform bridging RDE and MEAs , one of these platforms is the gas diffusion electrode (GDE) setup. As pointed out, the main motivation in developing GDE setups is to establish high reactant mass transport which is crucial for gaseous reactants such as oxygen . However, also the electrolyte type, i.e., liquid or membrane, is expected to have a significant influence on the catalyst reactivity. It is well known for example that the determined ORR activity of Pt based catalysts is substantially different in sulfuric and in perchloric acid based aqueous electrolytes . This phenomenon is related to structure sensitive, specific anion adsorption blocking active catalyst sites and the term spectator species was introduced for anions in the liquid electrolyte as they have no active part in the ORR. Nafion, by comparison, which is often used as membrane electrolyte, does not exhibit structure specific adsorption, although it blocks active catalyst sites as well . Although it has been shown that even non-specifically absorbed cations can influence catalytic reactions , to the best of our knowledge so far it has not been investigated to which degree the electrolyte environment, i.e., aqueous or membrane electrolyte, influences the oxidation of small organic molecules, e.g., methanol and formic acid. In the present study, we therefore compare the performance of two standard catalysts for the methanol and formic acid oxidation, i.e., Pt and Pd nanoparticles supported on high surface area carbon (Pt/C and Pd/C) and performed investigations in a conventional electrochemical (RDE) setup as well as in a GDE setup. We thereby simulate the conditions in a low temperature proton exchange fuel cell (LT-PEMFC), i.e., a Nafion/catalyst interface. ## Materials and chemicals The following materials and chemicals were used: Commercial Pt/C (Tanaka Kikinzoku Group, TEC10E20A, 19.4%) and Pd/C (FC Catalyst, 3151611, 20% Palladium on Vulcan), 37% hydrochloric acid (HCl, Suprapur, Merck), 65% nitric acid (HNO3, Suprapur, Merck), methanol (CH3OH, VWR Chemicals, 98.5%), formic acid (HCOOH, ≥95%, Sigma Aldrich), isopropanol (IPA, 99.7+%, Alfa Aesar), 70% perchloric acid (HClO4, ACS reagent, 70%, Sigma-Aldrich), KOH (pellets for analysis EMSURE®,Merck), Nafion™ D1021 Dispersion (Water based 1100 EW at 10 wt%), Ultrapure water (resistivity>18.2 MΩ•cm, total organic carbon (TOC) < 5 ppb) from a Milli-Q system (Millipore). The following gases from Carbagas AG were used for electrochemical measurements: Ar (99.999%), and CO (99.97%). Gas Diffusion Layer (GDL) without a Microporous Layer (MPL) (Freudenberg H23), Gas Diffusion Layer (GDL) with a Microporous Layer (MPL) (Freudenberg H23C8) and Nafion membrane (Nafion 117, Fuel Cell Store) or anion exchange membrane (Sustainion® X37-50 Grade RT, Fuel Cell Store) were used for the catalyst layer fabrication. ## Electrochemical measurements All electrochemical measurements were performed in a three-electrode system controlled by a potentiostat (ECi 200, Nordic Electrochemistry). If not specifically noted, the same electrochemical procedures (catalyst loadings, treatment, electrolyte, measurement protocol, etc.) were applied for the Pt/C and Pd/C catalyst. All measurements have been conducted at room temperature. ## Measurements in conventional electrochemical cell A glassy carbon electrode (5 mm diameter) was used as the working electrode, a platinum wire was used as counter electrode and a reversible hydrogen electrode (RHE) as reference electrode. The electrolyte was either 1.0 M HClO4 or 1.0 M KOH aqueous solution. The effective solution resistance was determined online with the help of a superposed AC signal (5mV, 5kHz) and was compensated to a value below 5 Ω via an analogue positive feedback scheme of the potentiostat . The catalyst ink for the RDE measurements was prepared by ultrasonically dispersing (ultrasound cleaning bath VWR, USC-THD, 45 kHz) the catalyst powder (amount corresponding to 0.276 mg metal, e.g., 1.380 mg of 20 wt. % Pd/C ) for ca. 8 min in 1.266 mL of a mixed solution containing isopropanol and water (1:3; v:v) and 11.04 μL of 10 wt% Nafion solution to form a homogeneous catalyst ink with a metal concentration of 0.218 mgmetal mL -1 . The RDE working electrodes were fabricated by pipetting 9.0 μL of the electrocatalyst ink onto a glassy carbon electrode leading a nominal metal loading of 10 µgmetal cm -2 geo, followed by drying in air. The electrolyte was deaerated by purging with Ar. Prior to the measurements the catalysts were cleaned by potential cycling between 0.15 VRHE and 1.20 VRHE at a scan rate of 500 mV s -1 until a stable cyclic voltammogram (CV) could be observed. The electrochemically active surface area (ECSA) was determined via the CO oxidation charge in CO monolayer stripping experiments . For this, the electrode was held at 0.15 VRHE in a CO-saturated electrolyte for 2 min. Thereafter the electrolyte was saturated for 10 min with Ar gas to replace the excess CO in the electrolyte. Finally, the adsorbed CO monolayer was oxidized to CO2 by scanning the electrode potential from 0.15 to 1.10 VRHE or 1.20 VRHE at a scan rate of 50 mV s -1 . The ECSA was estimated from the recorded oxidation charge by using a reference oxidation charge value for polycrystalline Pt of 420 µC cm -2 Pt and Pd of 405 µC cm -2 Pd , respectively. The cyclic voltammetry measurements were recorded in the same potential window and at the same scan rate of 50 mV s −1 , but in Ar-saturated electrolyte solution. The electrochemical oxidation of formic acid and methanol, respectively, was performed by collecting cyclic voltammetry curves in a Ar-saturated electrolyte solution containing 1.0 M HClO4 and 0.5 M HCOOH, 1.0 M HClO4 and 0.5 M CH3OH, or 1.0 M KOH and 0.5 M CH3OH at a scan rate of 50 mV s −1 in a potential window between 0.15 and 1.20 VRHE (formic acid oxidation) or between 0.2 and 1.10 VRHE (methanol oxidation). Due to the high reactant concentration, no rotation was applied. Indeed rotation leads to a reduction in the observed reaction rates, see Figure S1. ## Electrochemical measurements in gas diffusion electrode setup In the gas diffusion electrode setup, instead of the GC electrode, a GDE (3 mm diameter) was used as the working electrode (WE), a platinum wire as counter electrode (CE) and a reversible hydrogen electrode (RHE) as reference electrode (RE). The WE is separated from the CE and RE by a Nafion membrane or an anion exchange membrane (for KOH solution) in the upper cell compartment above the membrane 1.0 M HClO4 or 1.0 M KOH aqueous solution was used as electrolyte. All gases purged through the GDE were humidified by first passing through MilliQ water or the reactant solution. It has been previously shown that at room temperature full humidification is reached . WE preparation: The catalyst (amount corresponding to 0.382 mg metal) was ultrasonically (ultrasound cleaning bath VWR, USC-THD, 45 kHz) dispersed for ca. 8 min in 7.645 mL of a mixed solution containing isopropanol, water (3:1; v:v) and 15.29 μL of 10 wt% Nafion (10:1; ul:mgCarbon) solution to form a homogeneous 0.05 mgmetal mL -1 catalyst ink. The catalyst was deposited on the GDL with a MPL by vacuum filtration leading to a nominal loading of 200 µgmetal cm -2 geo . A blank GDL with MPL with a diameter of 2 cm was taken and a 3 mm hole was punched out in its centre. This was placed onto a GDL without MPL and in the hole a 3 mm catalyst coated GDL was placed and everything was pressed, protected by a Teflon sheet, to a Nafion membrane of 1 cm in diameter by a hydraulic press (2 tons pressure, 5 min, room temperature) . For the anion exchange membrane a reduced pressure was applied (1 ton pressure, 5 min, room temperature) to avoid sticking to the Teflon sheet. The system was deaerated by purging the GDE through the bottom cell part with humidified Ar . Prior to the measurements the catalysts were cleaned by potential cycling between 0.15 VRHE and 1.20 VRHE at a scan rate of 500 mV s -1 until a stable CV could be observed, The CO stripping measurements were performed as in the conventional cell, i.e., the electrode was held at 0.15 VRHE with streaming CO through the cell for 2 min, thereafter the CO was replaced by Ar gas to remove all excess CO. The adsorbed CO monolayer was oxidized to CO2 by scanning the electrode potential from 0.15 to 1.20 VRHE (Pd) or 1.10 VRHE (Pt) at a scan rate of 50 mV s -1 . The ECSA was estimated from the recorded oxidation charge using the same reference values as in the RDE measurements. The electrochemical oxidation of formic acid and methanol, respectively, was performed by passing Ar through reactant mixed in aqueous solution instead of pure MilliQ water. Different concentrations of reactants mixed in aqueous solution were tested to exclude reactant mass transport limitations, see Figure S2, and we decided to use the highest tested concentrations of 5.0 M HCOOH or 5.0 M CH3OH where already an inhibiting effect similar to the rotation effect in the RDE was observed to ensure no mass transport limitations. The cyclic voltammetry curves were collected at a scan rate of 50 mV s −1 in the same potential range as noted above. ## Physical characterization of the catalysts The size (diameter) and shape of the Pt and Pd nanocatalysts were evaluated by TEM using a Jeol 2100 transmission electron microscope operated at 200 kV. For the characterization, at least three images in at least three different randomly selected areas of the grids were chosen. The samples were prepared by dropping the catalyst ink (the catalyst was diluted in ethanol) onto carbon coated copper TEM grids. The nanoparticle size was evaluated by measuring the diameter of at least 200 nanoparticles using the software ImageJ. In order to evaluate the nanoparticle size with more statistical power (given that TEM analysis is limited to few hundreds of individual nanoparticles), small angle X-ray scattering were performed as previously described in detail . The Pd/C or Pd/C powders were placed in dedicated holders in between two mica windows. The measurements performed at the Niels Bohr Institute at the University of Copenhagen using a SAXSLab instrument. The data were fitted with polydisperse spheres models described by a volume-weighted log-normal distribution. ## Inductively Coupled Plasma Mass Spectrometry (ICP-MS) The actual metal loadings of WE were evaluated by ICP-MS (NexION 2000 ICP-MS). The ICP-MS was equipped with a cyclonic spray chamber and a PFA-nebulizer. The RF power for the plasma was held at 1300 W with a gas flow of 15 L min -1 . The catalysts on the GDLs were selected 4 different parts and dissolved in aqua regia (volume ratio of HCl:HNO3 = 3:1) and then diluted to 200 mL with milli-Q water. Based on four different measurements average metal loadings of 141 µgPd cm -2 geo and 143 µgPt cm -2 geo for Pd/C and Pt/C, respectively, were determined. To evaluate the real catalyst loading on the RDE tips, 9 μl of the catalyst ink was taken and dissolved in 4 mL aqua regia (volume ratio of HCl:HNO3 = 3:1) and then diluted to 25 mL with milli-Q water. By ICP-MS, metal loadings of 8.0 µgPd cm -2 geo and 7.5 µgPt cm -2 geo on the RDE tips were estimated. ## Results and Discussion As discussed in the introduction, the aim of this work is to establish the suitability of the GDE approach for volatile organic reactants simulating a LT-PEMFC environment. For this purpose, we investigate the performance of two standard catalysts, i.e., Pt/C and Pd/C, for the methanol and formic acid oxidation in a GDE setup with a catalystmembrane electrolyte interface where the reactant is introduced via a humidified gas stream. The experimental setup for the GDE measurements is relatively new and has been previously only used for studying the electro-oxidation of volatile small organic molecules at rather demanding conditions, i.e., hot phosphoric acid simulating conditions in high-temperature PEMFCs . The setup adopted for simulating LT-PEMFC conditions is schematically displayed in Figure 1. Most importantly, in this configuration the catalyst has only contact to the reactant in water and the membrane electrolyte, but no solvated anions such as e.g., perchlorate anions. As benchmark, we compared the GDE measurements to investigations conducted in a conventional electrochemical cell (RDE setup) with aqueous supporting electrolyte to which the reactant, i.e., methanol or formic acid is added. are clearly discernible and well separated by a potential region double layer formation (ca. 0.30 -0.60 VRHE) on Pt and Pd . Furthermore, a clear hysteresis, i.e., a shift in peak potential between positive and negative going scan, is seen in the potential region of oxide formation and reduction (> 0.60 VRHE) . The hysteresis indicates that oxide formation and reduction are rather slow and irreversible processes. This is of importance for the later discussion as surface oxides de-activate the surface by blocking adsorption sides. Also the CO stripping voltammograms for Pt/C and Pd/C in both cell types are similar, but the peak potential and peak width are slightly different. For both catalysts, in the GDE setup the peak potentials of the CO stripping peaks are slightly shifted to lower potentials as compared to the RDE measurements. As the film thickness in the GDE does not influence the position of the CO stripping peak, see Figure S4, these differences are proposed to be the result of a different electrolyte environment, i.e., aqueous electrolyte with mobile perchlorate anions as compared to a membrane electrolyte, influencing the CO stripping voltammograms. That is, the interaction with perchlorate anions slightly shifts the CO oxidation to higher potentials. The ECSA determination based on the CO stripping measurements is required when distinguishing between mass and surface area normalized (specific) currents. Furthermore, it can be used to confirm a full utilization of the catalyst layer in the GDE setup. In RDE measurements typically a full utilization of the catalyst layer is achieved due to the very thin catalyst film (nominal catalyst loading of 10 µgmetal cm -2 geo on the GC electrode). On the other hand, the investigation of realistic film thicknesses is not feasible due to a loss in accessibility of active sites, i.e., a decrease in ECSA with increasing film thickness . By comparison, in the GDE a roughly twenty times thicker catalyst layer is applied without loss in ECSA, see Figure S4. This means that in contrast do RDE measurements , in GDE measurements the ECSA does not depend on the catalyst loading and measurements under more relevant reaction conditions to what is actually used in fuel cell devices are feasible. As the GDE contains no liquid electrolyte in direct contact to the catalyst layer, a proper contact of the nanoparticles with the membrane electrolyte in the catalyst layer is required. This is achieved in the GDE setup by a simple pressing procedure . Integrating the CO stripping voltammograms recorded in the conventional RDE setup leads to ECSA values of ~153 m 2 g -1 Pt and ~130 m 2 g -1 Pd. A correction via the actual metal concentration in the inks by ICP-MS led to even slightly higher values of ~181 m 2 g -1 Pt and 149 m 2 g -1 Pd for Pt/C and Pd/C, respectively. The different ECSAs for Pt/C and Pd/C can be roughly compared with the different average particle size of the two catalysts. Assuming the nanoparticles are perfect spheres and no interface between the metal surface and the carbon support, ECSAs of the "free standing particles" of 165 m 2 g -1 Pt and 119 m 2 g -1 Pd for Pt/C and Pd/C, respectively, are calculated using the average particle size determined from TEM. Typically the surface area of "the free standing particles" is higher than the ECSA measured for "the supported particles" and the difference is assigned to the metal -support interface inaccessible for CO adsorption. In the present work, the correction of the ECSA values by the metal concentration measured in ICP-MS leads to higher ECSA values of "the supported particles". While the comparison is only a rough calculation, this finding might indicate that both catalysts contain many particles which are smaller than the average size determined by TEM. By comparison the ECSA determination in the GDE setup (corrected by the catalyst loading determined by ICP-MS) led to values of 124 m 2 g -1 Pt and 123 m 2 g -1 Pd for Pt/C and Pd/C, respectively. That is, the ECSA measured in the GDE setup tends to be slightly lower than in the RDE setup. This phenomenon might be related to the explained by the presence of Nafion in the GDE layer. Nafion is known to partially block the active surface area of the active catalyst phase and thus reduces the ECSA . Although, we used also Nafion in the conventional cell with liquid electrolyte, the situation is different as the Nafion serves more as a binder to the glassy carbon tip. Due to the liquid electrolyte, no Nafion is required to "electrically contact" the active phase of the catalyst. As a consequence, in the following we normalized the reaction rates to the number of electrochemically active sites determined by CO stripping, i.e., we analyse and compare the specific activities in both setups. After having characterized the catalysts in the supporting electrolyte (aqueous and membrane electrolyte) and having determined the number of electrochemically active sites, we studied the methanol and formic acid electro-oxidation reactions. In the activity measurements, it has been noticed that in the GDE setup for catalyst loadings in the range of what is used in RDE measurements substantially higher activities as compared to RDE measurements, see Figure S5, are obtained. However, it was seen that these high activities were not stable under constant potential cycling. By comparison, at higher catalyst loadings, similar to what is applied in fuel cells, the activity became both stable upon potential cycling as well as independent of the catalyst loading, see Figure S6. Therefore, in the following only measurements obtained at catalyst loadings similar to real fuel cells are discussed. Furthermore, before the activity measurements, the electrocatalysts were activated in Ar atmosphere by cyclic voltammetry until a stable curve shape was obtained and a CO-stripping measurement was performed. For Pd catalysts, different results were obtained, depending on whether a CO stripping voltammogram has been recorded before the electro-oxidation or not. Only after a CO stripping voltammogram was recorded, as in the cases discussed below, the catalyst was fully activated. In Figure S7 in the supporting information the formic acid oxidation on an as prepared Pd/C sample without prior CO stripping is shown. This phenomenon has not been investigated further, but might be related to hydride formation on Pd . In the GDE setup the methanol was supplied by bubbling Ar gas through a 5.0 M methanol aqueous solution. In the RDE setup the electrolyte was 1.0 M HClO4 aqueous solution containing 0.5 M methanol. The CVs were recorded at a scan rate of 50 mV s -1 at room temperature and display the 2 nd scan. In Figure 4, the results from the electro-oxidation of methanol on commercial Pt/C and Pd/C recorded in GDE and RDE are summarized. For Pt/C similar trends are observed on the thicker catalyst films in the GDE and the very thin catalyst films in the RDE. Most prominent, in both systems a clear hysteresis is seen between the peak potential in the positive and negative going scan. The hysteresis of the peak potential of the main oxidation peak, the maximum current density as well as the peak position (potential) where this is achieved are the main characteristics for an evaluation of the electro-oxidation of methanol. The hysteresis in the oxidation of small organic molecules is related to the "oxidation hysteresis" seen in the CVs recorded in inert atmosphere (Figure 3). The hysteresis indicates surface blocking processes that are dependent on the pre-history of the surface. For the small organic molecules it is typically interpreted by the formation of a CO poisoning species at the surface, i.e., coming from low potentials the catalyst surface is in its reduced form and COad accumulates during the scan inhibiting the methanol oxidation . By comparison, when scanning in negative direction the catalyst surface is initially oxidized and becomes active upon the reduction of oxide species blocking the surface . Alternatively, a shift in the rate-determining step of the methanol oxidation from methanol dehydration to OH adsorption by water dissociation has been proposed . Besides, these similarities also differences are observed. In the positive going scan in the GDE setup a pre-peak is seen, which starts around 0.5 VRHE. Also the deactivation due to Pt oxide formation is less pronounced than in the RDE. Last but not least, despite the thicker catalyst layer the surface normalized current densities at the peak potential are more than two times higher than in the RDE, i.e., 0.75 ± 0.03 vs. 0.26 ± 0.02 mA cm -2 Pt, see also Table 1. This observation might be related to the different methanol concentrations at the catalyst surface. However, at the chosen conditions no mass transport limitations are observed in the RDE setup. Therefore, it seems that in the GDE setup the same catalyst can provide higher specific activities than in an aqueous electrolyte despite the thicker catalyst layer. This might be further evidence of a lack of activity inhibition (as in the CO stripping voltammetry) due to (the lack of) specific anion adsorption in the GDE setup. Switching to Pd/C instead of Pt/C, in the RDE Pd/C is completely inactive for the methanol oxidation reaction and basically the same CV is recorded with and without the presence of methanol in the supporting electrolyte. The inactivity of Pd for the methanol oxidation is well established in aqueous acidic electrolyte and in contrast to alkaline electrolyte . Although in the GDE setup minor oxidation currents on Pd/C are recorded, the reached current density high enough to be technologically relevant. In the GDE setup, the acidic Nafion membrane can be easily exchanged by an alkaline anion exchange membrane . In Figure 5, it is demonstrated that in alkaline environment indeed Pd/C is active for the methanol oxidation reaction. The features in both setups, i.e., GDE and RDE are similar, however, the differences in specific activity between the two setups are even more significant. Roughly 25 times higher specific current densities are observed. In addition, in the negative going scan the re-activation of the Pd due to the reduction of the oxide is sharper and more pronounced. In addition to methanol, we also studied the electro-oxidation of formic acid on Pt/C and Pd/C in the two setups. The results are summarized in Figure 6. For the electro-oxidation of formic acid, the peak position (potential) of the main oxidation peak, its current density as well as the observed hysteresis between forward and backward going scan are also the main characteristics for an evaluation of the performance. The RDE measurements confirm that Pd/C is active for formic acid electro-oxidation in acidic environment, while Pt/C is inactive . That is, formic acid oxidation on Pt/C is strongly inhibited and a large hysteresis is seen in the RDE measurements. By comparison, in the GDE the formic acid oxidation on Pt/C exhibits a lower hysteresis, however, the observed current densities are rather low. In contrast to Pt/C, Pd/C exhibits a low peak potential, high current densities, and little hysteresis. It is seen that the performance based on peak position in the positive going scan direction is significantly improved in GDE setup as compared to the RDE, i.e., a peak position of 0.34 vs. 0.44 VRHE is recorded, see Table 1. This shift in peak potential of the main oxidation peak by ~100 mV to lower potentials in the GDE setup can be associated with a reduced overpotential. At the same time peak current density is roughly the same, i.e., 0.97 ± 0.07 vs 0.94 ± 0.06 mA cm -2 Pd. Again, this difference between GDE and RDE might be correlated to specific anion adsorption. The onset of formic acid oxidation is associated with the desorption of Hupd. In an aqueous acid electrolyte there will be always an interplay between Hupd desorption and anion adsorption while in membrane electrolyte no mobile anions exist. However, in the GDE a clear hysteresis and asymmetric oxidation peaks are seen in the negative going scan. It seems that in the GDE, at higher electrode potentials (> 0.4 VRHE), the Pd/C catalyst becomes inhibited by the formation of surface poisoning species, while this seems not to be the case in the RDE measurements. ## Conclusion In the presented work, the application of a GDE setup for the electro-oxidation of volatile small organic molecules is presented. Both methanol and formic acid are potential fuels for fuel cells, i.e., so-called direct methanol/formic acid fuel cells. It is shown that the GDE setup that has been previously used for the investigation of oxygen reduction reaction catalysts at LT-PEMFC conditions, can be easily adopted for the study of these reactants. The reactant is simply introduced via the humidification of the gas stream. In contrast to the conventional RDE technique, catalyst films with realistic thicknesses for applications are studied in the GDE setup. Furthermore, the catalyst is not in direct contact to a liquid electrolyte, but a catalyst -membrane electrolyte interface is formed. Comparing the electrochemical responses in the two systems in a qualitative fashion, similar electrochemical behavior is observed in the RDE and GDE. Nevertheless, differences in specific activity and the peak potential of the main oxidation peak, and the hysteresis between positive and negative going scan be observed, which might be important for an extrapolation of the results from a catalyst screening to their use in direct methanol or direct formic acid fuel cells. The GDE approach therefore provides an important addition to the RDE approach in order to bridge the gap to MEA testing. The data provided can be used as a benchmark for future studies. a scan rate of 50 mV s -1 and display the 2 nd scan. The solid lines are the forward going scans, and the dashed lines the backward going scans. Pd, respectively, are determined demonstrating that the accessibility of the catalyst does not decrease with increasing film thickness. All curves were recorded at room temperature with a scan rate of 50 mV s -1 and display the 1 st , 10 th , 30 th and 50 th scan. The formic acid was supplied to the catalyst by bubbling Ar through 5.0 M formic acid aqueous solution. The solid lines are the forward going scans, and the dashed lines the backward going scans. The measurements demonstrate that the normalized reactivity decreases with the catalyst loading. It is seen that at lower catalyst loading the current decreases rapidly upon constant cycling, whereas at higher loading, the current was stable upon cycling.

chemsum

{"title": "On the electro-oxidation of small organic molecules: towards a fuel cell catalyst testing platform based on gas diffusion electrode setups", "journal": "ChemRxiv"}

differentiating_aβ40_and_aβ42_in_amyloid_plaques_with_a_small_molecule_fluorescence_probe

## Abstract: Differentiating amyloid beta (Ab) subspecies Ab40 and Ab42 has long been considered an impossible mission with small-molecule probes. In this report, based on recently published structures of Ab fibrils, we designed iminocoumarin-thiazole (ICT) fluorescence probes to differentiate Ab40 and Ab42, among which Ab42 has much higher neurotoxicity. We demonstrated that ICTAD-1 robustly responds to Ab fibrils, evidenced by turn-on fluorescence intensity and red-shifting of emission peaks. Remarkably, ICTAD-1 showed different spectra towards Ab40 and Ab42 fibrils. In vitro results demonstrated that ICTAD-1 could be used to differentiate Ab40/42 in solutions. Moreover, our data revealed that ICTAD-1 could be used to separate Ab40/42 components in plaques of AD mouse brain slides. In addition, twophoton imaging suggested that ICTAD-1 was able to cross the BBB and label plaques in vivo.Interestingly, we observed that ICTAD-1 was specific toward plaques, but not cerebral amyloid angiopathy (CAA) on brain blood vessels. Given Ab40 and Ab42 species have significant differences of neurotoxicity, we believe that ICTAD-1 can be used as an important tool for basic studies and has the potential to provide a better diagnosis in the future. ## Introduction Amyloid beta (Ab) plaques, one of the characteristic biomarkers for Alzheimer's disease (AD), have been discovered for more than 100 years in the postmortem brains of AD patients. 1,2 However, the role of Ab plaques in the pathology of AD has still been heavily debated, because the correlation between plaque burdens (numbers and areas) and the severity of AD is poor. In the plaques, Ab40 and Ab42 peptides are major constituents. Nonetheless, unlike the role of plaque, there is nearly no argument that Ab42 has much higher neurotoxicity than Ab40 does. 1,2,4,6,7 Conceivably, differentiating Ab40 and Ab42 can considerably clarify the role of plaque in AD pathology. Unfortunately, small-molecule probes with such capacity are scarce. 7 Due to the small difference in the amino acid sequence of the peptides, discovering small-molecule probes capable of differentiating Ab40 and Ab42 has been considered as an impossible mission. In our previous studies, inspired by the binding principles of antibodies for soluble and insoluble Abs, we designed a series of small fluorescent molecules to selectively detect soluble Abs, 8 the likely biomarker for the early stage of AD pathology. Encouraged by the antibody's capability to specifcally recognize Ab40 or Ab42 peptides, we hypothesized that it could be possible to differentiate Ab40 and Ab42 with a small molecule probe. Our design strategy is different from most previous studies, which are focused on adjusting optical properties to turn on (off) signals or make larger stokes shifts. Unfortunately, few probes have been designed based on the insights from the Ab structures. In this report, we demonstrated that our designed small-molecule fluorescence probe, ICTAD-1, has the capacity to spectrally differentiate Ab40 and Ab42 in vitro and in the biologically relevant environment. We believe that our strategy could provide a new path for designing Ab probes. ## Design of uorescent imaging probes It is obvious that the C-terminal of Ab peptide is the key for designing small-molecule probes to distinguish Ab40 and Ab42. However, this is extremely challenging, due to the difference in only two amino acids (isoleucine-alanine). Nonetheless, it has been routinely performed in numerous laboratories with anti-Ab42 antibodies to determine the contents of Ab42 in cell media and brain tissues. These anti-Ab42 antibodies were designed based on the epitope of the C-terminal of the peptide. This fact has bolstered us to believe that the properties of the C-terminal can be relied on to design our small molecule probes. In the past, X-ray structures of full-length Abs were rare. However, in recent years, several detailed structures of Ab fbrils have been published. Particularly, the advanced cryoEM technology has impressively facilitated Ab structure studies. After having carefully surveyed the published structures of Ab fbrils, we found several potential binding sites for designing probes for Ab fbrils. The sites are A site, which locates in the hydrophilic N-terminal portion, B site, the hydrophobic midregion, and C site, which is formed by a hydrophobic Cterminal from one Ab peptide and a hydrophilic N-terminal from another Ab peptide (Fig. 1a). Numerous small molecules have been reported as ligands of Ab40 and Ab42 species, including several sub-species such as soluble oligomers and insoluble fbrils. However, nearly all of the compounds have been designed to interact with the B site, which is the middle region of Ab peptides. 8,10, Obviously, this is not the suitable region for designing probes to differentiate Ab40 and Ab42. Surprisingly, to the best of our knowledge, no compound has been intentionally designed or validated to target the Cterminal and the C site. Interestingly, we noticed that, for the site C in most cases, multiple pieces of the Ab peptide forms a gulf, in which one side consists of hydrophilic amino acids (K28, D1), and another side is hydrophobic of V39, V40, I41, and A42 (isoleucine-alanine), the last two amino acids in Ab42 (Fig. 1b and S1 †). This interesting feature of the C site requires its binding ligands to meet the following criteria: (1) although Ab40 and Ab42 only have two amino acids difference, their hydrophobicity is very distinct. It is well known that, compared to Ab40, Ab42 is much stickier and considerably easier to aggregate. This is the key feature for designing probes to distinguish them, and this key feature requires the designed probes contain a hydrophobic moiety to interact differently with the hydrophobic side of the C-site; (2) in addition to the hydrophobic moiety, the ligand needs to contain a hydrophilic moiety to interact with the polar side; (3) the designed probe has a different binding affinity to Ab40 and Ab42 fbrils; and (4) the probes should have distinct fluorescence spectra and/or binding strength in the presence of Ab40 and Ab42 fbrils. In terms of the fluorescent properties of small-molecule fluorescent probes, numerous compounds showed blue-shifting (hypsochromic shifting) after interacting with Abs, suggesting they bind to a hydrophobic environment. However, benzothiazole analogues are distinct because they showed fluorescence red-shifting (bathochromic shifting), 36 indicating this moiety can bind to a polar environment. The most typical compound is thioflavin T, which has been used as a gold standard for Ab plaque staining, and its derivative PIB has been widely used in the clinic as a PET tracer for imaging Ab deposits. Thioflavin T provides the classic emission of spectral red-shifting. However, thioflavin T cannot spectrally differentiate the Ab fbrils, likely due to its incapacity to differently interact with the hydrophobic moiety of the C-site. Nonetheless, we consider thioflavin T is a good starting point for probe designing. Based on the above consideration, the designed probe should be tilted between the hydrophilic and hydrophobic moieties with a vertical arrangement (parallel to the fbril axis) (Fig. 1d), while thioflavin T is too short to touch the hydrophobic moiety and this is why it does not have the differentiating capacity. Conceivably, extending the length of the hydrophobic patch of thioflavin T has the potential to achieve this goal (Fig. 1c). In this regard, we designed ICTAD-1 (iminocoumarin-thiazole for AD), in which N,N-dimethyl-phenyl is for touching the hydrophobic moiety and for adjusting the binding affinity. In addition, iminocoumarin was introduced to stretch the hydrophilic moiety, which can enhance the binding through increasing hydrogen bonds (Fig. 1c). Moreover, the NH group in iminocoumarin could be served as a hydrogen donor to form an intramolecular hydrogen bond with benzothiazole moiety, leading to the enhanced planarity of the two heterocycles, which was distinctly different from the previously reported coumarin derivatives. To validate this design strategy, molecular docking between ICTAD-1 and Abs was performed. The molecular docking results ICTAD-1 with Ab42 (PDB: 5OQV) showed that, indeed, the hydrophilic patch of C-site interacted with the stretched hydrophilic moiety of ICTAD-1, and the extended aromatic ring moiety interacted with the hydrophobic patch (Fig. 1d). The hydrophobic interaction between ICTAD-1 with residues ALA2 and ILE41 of Ab42 was observed. The oxygen atom in iminocoumarin formed an intermolecular hydrogen bonding with the NH of GLY38. Moreover, the intramolecular hydrogen bond was also observed. The docking score for ICTAD-1 with Ab42 at site C was 8.8476, which was lower than thioflavin T (6.8242), indicating ICTAD-1 might have better binding affinity to Ab42 than thioflavin T. Similar docking results of ICTAD-1 with other Ab42 fbrillar structures (PDB: 5KK3 and 2NAO) were achieved (Fig. S2a and b †). We also carried out the docking studies of ICTAD-1 with Ab40 fbrillar structures (PDB: 2MVX). The docking results showed ICTAD-1 could bind to the C site of Ab40 through three intramolecular hydrogen bonds with residues HIS6 and GLU11 (Fig. S2c †). And it was observed that there were hydrophobic interactions between ICTAD-1 with residues PHE4 and VAL40. Moreover, the docking score of ICTAD-1 with Ab40 was 8.2130, which was slightly lower than Ab42. Similarly, ICTAD-1 could also dock into the C sites of other Ab40 fbrils (PDB: 2MPZ and 6SHS) (Fig. S2d and e †). Considering the difference in the hydrophobic environment at sites C in Ab40 and Ab42, these results indicated that ICTAD-1 had the potential to bind and discriminate Ab40 and Ab42 fbrils. To investigate whether the iminocoumarin moiety is necessary for binding, we designed ICTAD-2 by replacing it with a naphthalene ring. In addition, we designed ICTAD-3 and -4 to further extend the hydrophobic moiety to investigate whether the compounds can better match with the hydrophobic patch in the C-site. The synthesis of ICTAD-1 is straightforward, and its route is shown in Fig. 1e. The structures and synthetic routes for ICTAD-2, -3, and -4 are shown in Fig. 1f and ESI. † ## Properties of ICTAD-1 With the probe in hand, we frstly performed the photostability experiments in DMF to investigate whether ICTAD-1 is stable under light (1500 joule per minute), and found that there were nearly no changes of fluorescence intensity after irradiating 120 minutes (Fig. 2a and S3a-d †), suggesting that ICTAD-1 has excellent stability to resist photobleaching. To explore ICTAD-1's responses towards different pH media, we titrated it within the range of pH 2-11, and found its absorbance intensity was consistent from pH 6-11. While we found that its fluorescence intensity decreased dramatically from pH 2-5, and its fluorescence is minimal under pH 7 (Fig. 2b and S4 †), suggesting this probe has minimal fluorescence background signal and is suitable for in vivo imaging. With ICTAD-1 in hand, we also investigate its absorbance and emission spectra in different solvents. We found that the absorbance is much less dependent (<30 nm changes) on solvent polarity, while its emission could be drastically changed in different solvents, and about 150 nm difference could be found from non-polar hexane to glycol. As we expected, ICTAD-1 displayed much longer emission in polar solvents such as DMSO and glycol, indicating this probe can interact with a polar environment (Fig. 2c, d and S4 †). Moreover, ICTAD-1 exhibited higher quantum yield in the nonpolar solvents, such as hexane and toluene (Table S1 †). ## Responses of ICTAD-1 to Ab brils in solutions To examine ICTAD-1's emission response to Ab fbrils, we incubated 250 nM of the probe with different concentrations of Ab40 and observed a consistent red-shifting of emission from 1.25 mM to 12.5 mM, suggesting the binding of ICTAD-1 with Abs is not 1 : 1 stoichiometry (Fig. 3a). Meanwhile, apparent fluorescence intensity increases could be observed with different concentrations of Ab40. The largest shift was 47 nm and the largest intensity increase was over 6-fold. We also observed that the fluorescence intensity increased with time and it reached a plateau within 20 minutes (Fig. S5c and d †). The turn-on effect of ICTAD-1 is likely due to twisted intramolecular charge transfer (TICT) upon binding to the fbrils, in which the planar confguration is preferred, and the rotation of the aromatic rings is restricted. 33,44 From this study, we were also able to calculate a binding constant K d ¼ 6.27 mM for Ab40, while K d of Ab42 was 3.78 mM, suggesting that the binding to Ab40 fbrils was weaker than that for Ab42 fbrils (Fig. 3a-d). In addition, we found that ICTAD-1 provided excellent linear correlations with the concentrations of Ab40 and Ab42 fbrils in the range of 0-4 mM (Fig. S5 †). We then performed similar experiments to study the emission changes of ICTAD-1 with Ab42 fbrils and found consistent red-shifts. Remarkably, we observed that the degrees of emission peak shifting were different, and Ab40 provided much larger red-shifting (Fig. 3e and f). This is expected because Ab42 provided tight binding due to its higher hydrophobicity and crowed space, while Ab40 is less crowded and provides loose binding. This is consistent with ICTAD-1's binding constants for Ab40/42. Interestingly, in the solutions of Ab40/42 mixture, we found that the normalized spectral peaks shifted from shorter wavelengths (blue) to longer wavelengths (red) with the increasing ratio of Ab40 component, and the shifted wavelength number was linear to the ratio of Ab40/42 (Fig. S6 †). In addition, we found that ICTAD-1 had excellent selectivity over metal ions and other proteins, such as bovine serum albumin (BSA) and human serum albumin (HSA) (Fig. S7 †). To examine whether a naphthalene ring can be used to replace the iminocoumarin moiety in ICTAD-1, we performed similar solution tests with ICTAD-2, and found that ICTAD-2 provided a slight red-shifting and a slight intensity decrease (Fig. S8a †), suggesting the hydrophilic iminocoumarin moiety is necessary for binding to the hydrophilic patch in Ab fbrils. Different from ICTAD-1, ICTAD-3 showed no apparent red-shifting, while it showed a slight intensity increase (Fig. S8b †). ICTAD-4 showed an observable blue-shifting and a moderate intensity increase (Fig. S8c †). Since ICTAD-2, -3, and -4 didn't have favorable properties for differentiating Ab40/42, we didn't perform further investigation for these compounds. ## Differentiating Ab40 and Ab42 with ICTAD-1 via spectral unmixing imaging To explore whether ICTAD-1 can be used to quantify the amount of Ab40 and Ab42 fbrils, spectral unmixing with mixtures of these fbrils was performed (Fig. S9 †). First, we used pure Ab40 and Ab42 fbrils solutions to establish the spectra on a 96-well plate. The linear relationships could be observed for both fbrils (Fig. S10 †), suggesting that spectral unmixing is feasible to quantify the contribution of each component in the mixture. The detection limitations are 3.3 nM and 28.3 nM for Ab40 fbrils and Ab42 fbrils respectively. Then we conducted librarybased spectral unmixing to separate the ratios of these two fbrils. Indeed, we found that it was feasible to deconvolute the signals with the characteristic spectra of free ICTAD-1 (green), ICTAD-1 with Ab40 fbrils (red), and Ab42 fbrils (blue) (Fig. 4a and b). Next, to investigate whether spectral unmixing is feasible in a biologically relevant environment, we performed the above experiments in the presence of bovine serum albumin (BSA) and mouse brain homogenate (BH). As we expected, spectral unmixing could be used to deconvolute the signals from the free probe (green), binding with BSA or BH (cyan), binding with Ab40 (red) and Ab42 (blue) (Fig. 4c-e), suggesting it is possible to spectrally differentiate the Ab fbrils in biologically relevant environments. ## Tissue staining and spectral unmixing with ICTAD-1 We frst explored whether ICTAD-1 could provide a high quality of plaque staining, and found that plaques on brain slides of 5xFAD mice can be sharply visualized after the slide was incubated with the probe for 15 minutes (Fig. 5a), and the labeled plaque showed excellent colocalization with 6E10 antibody staining (Fig. 5a), which is specifc to Abs. This result suggested that ICTAD-1 was able to specifcally label Ab plaques. The previous results inspired us to explore the capability of ICTAD-1 in differentiating Ab40 and Ab42 fbrils in the plaques. Then we relied on confocal imaging that has been equipped with a spectral unmixing function. Interestingly, the spectra from the core most closely resembled the one for Ab42 in vitro studies, whose peak wavelength was shorter. Meanwhile, the spectra from the periphery were similar to the one from Ab40 fbrils in vitro, whose peak has a longer wavelength (Fig. 5b-e and S11 †). The peak wavelength difference from the two spectra was 12 nm (Fig. 5e). Reportedly, Ab42 is the dominant species in the core of plaques, which is consistent with our spectral unmixing results. Taken together, these results suggested that ICTAD-1 could be used to quantify the contents of Ab42/Ab40 from brain slides. This also indicates that it is possible to quantify the subspecies in human brain tissue, which is very important to clarify the contributions of Ab42 fbrils to the AD pathology. ## In vivo two-photon imaging with ICTAD-1 To validate whether ICTAD-1 can be used for in vivo imaging, we performed two-photon microscopic imaging with 15 month old This journal is © The Royal Society of Chemistry 2020 Chem. Sci., 2020, 11, 5238-5245 | 5241 5xFAD mice. We frst intravenously (iv) injected the mouse with FITC-dextran as the contrast agent for blood vessels. After 5 minutes, we iv injected ICTAD-1 (1 mg kg 1 ), and imaged the mouse at 5 minutes post-injection of ICTAD-1. As expected, the plaques can be easily identifed, due to the excellent contrast (Fig. 6a-c), suggesting that the probe can cross the brain-blood barrier (BBB) and can label plaques in vivo. Interestingly, we experienced some difficulties to identify cerebral amyloid angiopathy (CAA) on the vessels, which can be found from its auto-fluorescence (blue in Fig. 6d and e). These results indicated that ICTAD-1 has weak capacities to label CAAs, in which Ab40 is the dominant species. 49,50 This result could be possibly explained with our in vitro K d data, which indicated that ICTAD-1 had much stronger binding to Ab42, while it showed weaker binding to Ab40. This in vivo data also suggested that differentiating Ab40 and Ab42 in vivo via different binding strength (K d ) was feasible. ## Discussion Ab plaques have been discovered for more than 100 years, however, their roles in AD pathology are still poorly defned. The complicated dynamics and contributions of each component are the likely factors for the perplex. It is well documented that Ab40 and Ab42 are the major components of the plaques. Nonetheless, it is not clear what is the contribution weight of each Ab subspecies to the pathology, and whether the content of Ab42 in plaques has better correlations with the severity of the disease. With our probe, it is possible to partially solve the mysteries, and we may have the capacity to quantify how much Ab42 is in each plaque, each brain slide, and specifc brain areas. Conceivably, it is also possible to quantify Ab42 contents for a whole-brain through serial slide imaging. With such a tool, we will be well equipped to answer those basic mechanism questions and will clarify the questions around postmortem plaques, which will provide more diagnostic information. Thioflavin T reportedly showed slight difference of fluorescence lifetimes in pure solutions for Ab40 and Ab42 fbrils, 51 however, it remains unknown whether the differences can be applied under physiological conditions. Kung et al. 52 reported that radioligand [ 125 I]DMTZ had certain preference for Ab42-positive cerebral amyloid angiopathy (CAA). However, it is not clear whether it has the capacity to differentiate Ab40/42 in the plaques. In this report, we also demonstrated that ICTAD-1 could be used to separate Ab40/42 using spectral properties in vitro and binding strength in vivo. The spectral unmixing is easy to perform on a regular confocal microscope, suggesting our method can be easily adapted by other biological-oriented laboratories. From the slide spectral unmixing, the separation of emission peaks of Ab40/42 was 12 nm, which may indicate that the separation of Ab40/42 is imperfect. However, it is possible to achieve a cleaner separation (a larger difference in the emission peak) if a better algorithm is developed. The difference of K d for Ab40/42 could be utilized for designing PET tracers that have much high specifcity for Ab42, the most toxic species. Our in vivo two-photon imaging indicated that ICTAD-1 had certain binding preference towards Ab42, and this could be considered as clues for designing for Ab42 specifc PET tracers. In addition, it is also tremendously important if we can differentiate CAA and plaques with a PET tracer, which is not currently available. Reportedly, Ab40 is the major component of CAAs. 49,50 Our data from this report suggests that our probe and its analogues may have the potential to differentiate plaques and CAAs. With such a probe, we will be able to dissect the vascular CAA contribution and plaque contribution to AD pathology. Our studies also pointed to several possible directions for future AD research. First, we may have the capacity to monitor therapy that is prone to reduce Ab42. Second, it is possible to design drugs that specifcally target Ab42. Lastly, we will be able to investigate whether the specifc reduction of Ab42 is a validated approach for future drug development. ## Conclusions In summary, we demonstrated that ICTAD-1 was able to spectrally differentiate Ab40/42 in solutions, in plasma and in brain tissues. ICTAD-1 has the potential to dissect the toxicity contributions of Ab40 and Ab42, and could be considered as a lead compound for developing Ab42 specifc PET tracers. ## Ethical statement All animal experiments were performed in strict accordance with the NIH guidelines for the care and use of laboratory animals (NIH Publication No. 85-23 Rev. 1985) and was approved by the Institutional Animal Care and Use Committee at Massachusetts General Hospital. ## Conflicts of interest There are no conflicts to declare.

chemsum

{"title": "Differentiating A\u03b240 and A\u03b242 in amyloid plaques with a small molecule fluorescence probe", "journal": "Royal Society of Chemistry (RSC)"}

building_machine_learning_force_fields_of_proteins_with_fragmentbased_approach_and_data_transfer

## Abstract: We combined our generalized energy-based fragmentation (GEBF) approach and transfer learning technique to construct machine learning force fields (MLFFs) for proteins only from quantum mechanics (QM) calculations of small subsystems. Using a kernel-based model called Gaussian Approximation Potential (GAP), our protocol can automatically generate training sets with high efficiency. To facilitate the construction of training sets for various proteins, a protein's data library is created to store all data of subsystems generated from trained proteins. With this data library, for a new protein only its subsystems with new topological types are required for the construction of the corresponding training set. With two polypeptides, 4ZNN and 1XQ8 segment, as examples, the energies and forces predicted by GEBF-GAP are in good agreement with those from QM calculations, and dihedral angle distributions from GEBF-GAP molecular dynamics (MD) simulations can also well reproduce those from ab initio MD simulations. In addition, with the training set generated from GEBF-GAP, we also demonstrate that GEBF-MLFFs can also be constructed by neural network (NN) methods with full QM quality. Therefore, the present work provides an efficient and systematic way to build force fields for biological systems like proteins with QM accuracy. ## INTRODUCTION Molecular dynamics (MD) simulation has emerged as an important tool to understand how the structure of a protein molecule determines its function in a cell. Currently, MD simulations with the classical force fields have been widely applied for large biomolecules including proteins. 7,8 However, the accuracy of classical force fields is still insufficient for reliable descriptions of some proteins. For example, the α-helical propensity is underestimated by the AMBER99SB force field compared to the corresponding experimental values. 9 The classical force fields cannot accurately describe temperature-dependent folding. 10 Nowadays, the machine learning (ML) method has been increasingly applied to develop more accurate atomistic potentials with very general functional forms than the conventional force fields with physically inspired functional forms. The resulting machine learning potentials, also called ML force fields (MLFFs), have been demonstrated to be quite successful for a variety of different systems. By "learning" from reference data sets obtained from QM calculations for a given system or a type of systems, MLFFs may reach similar accuracy as QM methods at a cost which is orders of magnitude less than that required for QM calculations of the same system. Due to the chemical complexities of proteins and the high computational costs of QM methods for large systems, building MLFFs for proteins remains a great challenge. Energy-based fragmentation (EBF) approaches provide a practical and attractive solution to overcome these two difficulties. With this approach, the ground-state MLFF of a large system can be obtained as the linear combination of MLFF trained from small subsystems, which are representations of different local regions of a large system. In previous studies, a residue-based neural work (NN) approach 46,47 was proposed to construct NN potentials for 20 types of amino acid capped with an acetyl group (ACE) and N-methyl amid group (NME) and 1 type of ACE-NME, as shown in Figure 1. Then, the MLFFs of a protein are expressed as the linear combination of these NN potentials. The resulting ML potentials represent the first step towards ab initio quality protein force fields. However, the local regions on these subsystems are not same as the target system. Thus, these potentials are not yet accurate enough, with the root-mean-square errors (RMSEs) for the energy and forces of (Ala)9 being 0.15 kcal/(mol• atom) and 4.75 kcal/(mol• ), respectively, with respect to reference density functional theory (DFT) data. 47 Based on the generalized energy-based fragmentation (GEBF) approach developed by our group 35 , we also constructed MLFFs for alkanes with the linear combination of MLFFs of small subsystems trained individually in our previous work. 48 Our previous scheme may be suitable for simple biomolecules like cellulose. However, proteins have twenty types of amino acid residues and too many different types of subsystems will be generated. It is difficult to construct MLFFs of all kinds of subsystems individually according to the previous fragment-based ML scheme. In this work, we propose a new protocol to construct MLFFs for proteins with full QM accuracy only from QM calculations on small subsystems. To circumvent the difficulty of MLFFs construction for enormous types of subsystems in previous fragment-based ML schemes, a new strategy is adopted here by fitting the energy (or forces) of a given protein as the summation of atomic contributions from QM calculations of various subsystems. To facilitate the construction of MLFF for various proteins, a protein's data library is created to store all data of subsystems generated from trained proteins. For a new protein, a subset of subsystems with the same topological types that are already in the protein's data library can be directly taken as a part of the training set, together with some newly generated subsystems. To automatically collect the training set, an online active learning 48 is adopted here to generate these new subsystems for studied protein. Then, full-QM quality GEBF-MLFF can be constructed using either kernel model like GAP 12 or NN model like Deep Potential 17 with the training set generated by GEBF-Figure 1. Fragmentation scheme utilized in the construction of MLFFs. In our GEBF method, fragments are capped with their environmental fragments or hydrogen atoms if necessary. In the previous residue-based method, fragments are capped with an acetyl group (ACE) and N-methylamine group (NME). ML protocol. Our protocol is applied on two polypeptides (4ZNN and 1XQ8 segment) to construct the corresponding GEBF-MLFFs and their accuracy and efficiency are validated with reference QM calculations. The results indicate that the GEBF-MLFFs can reproduce QM results very well at speeds several orders of magnitude faster than ab initio calculations. We expect that this protocol will greatly promote the development of fast and accurate MLFFs for various biological systems. The remainder of this paper is organized as follows. In Section 2, we describe the theoretical foundations of GEBF-MLFFs and the data transfer approach. In Section 3, the accuracy and computation costs are demonstrated by applying the GEBF-ML method to two polypeptides. In Section 4, a brief summary is presented. ## METHODOLOGY 2.1. The GEBF-ML Force Fields. To automatically construct the subsystems on the training set, the GEBF approach developed by our group is adopted. The generation of subsystems for a polypeptide 4ZNN is also illustrated in Figure 1, we will generate various subsystems, each of which contains a fragment and its neighboring fragments and capping hydrogen atoms if necessary (in grey oval). Clearly, subsystems constructed in this way are better representations of the local chemical environment of different regions in a protein than those in the residue-based NN approach. As the differences between QM and PM6 methods are generally much smaller than absolute QM values, the errors of ML models can be reduced if the PM6 method is used as the baseline. 49,50 In this work, this strategy is adopted, and an atomic ML model called GAP 12 based on kernel ridge regression with the SOAP kernels 51 (see details in the Sec.1 of the supporting information) is chosen to learn the energy difference of all subsystems for the studied proteins. To illustrate the advantage and disadvantage of this approach, GEBF-MLFFs of the 1XQ8 segment is also constructed by learning the QM energy data of corresponding subsystems directly. The approach to construct GEBF-MLFFs from QM energies is similar to the method for total energy difference prediction explained below, details can be seen in the Sec.1 of the supporting information. In addition, the Deep Potential method 19 is also adopted to construct GEBF-MLFFs with NN methods. In the GAP or Deep Potential method, the energy difference ML m E  of the mth subsystem with m S atoms are described as the summation of atomic energy For GAP, the atomic energy i e can be express as ( , ) Here, B N is the number of representative local atomic environments, B i w is the weight factor. SOAP is used to describe the local atomic environment i X and the kernel K. For Deep Potential, the atomic energy i e can be represented as a neural network. Using a two hidden layer feedforward NN as an example, the atomic energy can be expressed as 0 21 23 12 01 1 2 Where G is the output of a local embedding network to describe the atomic environment. E E E =  + (5) The PM6 energy of the target system with M subsystems are evaluated with the GEBF method by the linear combination of Details of subsystem construction and determination of coefficients are explained in Sec.3 of supporting information. The long-range nonbonded interactions between each subsystem and background charges on distant atoms are treated as the Coulomb interaction. The point charges are obtained from the natural population analysis (NPA) of subsystems, which are generated from the initial structure (extended structure generated from peptide sequence using Amber 16 program 52 ) used in the online training process. After training, the point charges are assumed to be constant like in traditional force fields . A r and A Q denote the coordinate of atom A and the point charge locating on atom A, respectively. ## Data Transfer Approach. Because a subset of subsystems generates from a protein may have the same topological structure in chemical space as those from another protein, we may introduce instance-based transfer learning 53 to avoid redundant QM calculations on these subsystems. The flowchart of the scheme is shown in Figure 2. In our approach, we create a protein's data library, which contains all data of subsystems generated from trained proteins. Starting from a given conformer of a new protein, MD simulation with NVT ensemble is performed based on the GEBF-MLFFs. As GAP can give internal uncer-tainty of Gaussian process regression model, which both consider the sampling density in the conformation space and the actual shape of the potential energy landscape, the ML model is chosen as GAP when the training sets are constructed. During the simulation, subsystems are automatically generated using our GEBF approach. If the subsystem types are already in the data library (the details of subsystem discrimination can be found in the Sec.4 of the supporting information), the corresponding sub-datasets are loaded to the training set. Otherwise, online active learning 48 (see details in Sec.5 of supporting information) is employed to select the representative subsystem conformers. When the training set is updated, the GEBF-ML force fields are also renewed to fit the energies and forces of conformers explored by online training. the online active learning. The total number of subsystem configurations and subsystem types in the training set are 5020 and 65, respectively. The fraction of QM calculations for the 1XQ8 segment is much smaller than that for the 4ZNN, since a large number of subsystems generated from 4ZNN can be reused. Thus, our GEBF-ML scheme shows high efficiency for building the training set. It is also worth mentioning that with a data library for all trained subsystems, QM calculations on many subsystems are avoided when MLFFs on new proteins are constructed. The subsystem data is simply transferred according to their topology structures and the advantage of this approach is robust and intuitive. More advanced data transfer techniques may also be adopted if the aim is to construct MLFFs for a new protein with as few configurations as possible. After the training sets have been constructed, the GEBF-GAP force fields are constructed from the QM energy differences. As GEBF-GAP based MD simulations show small energy drift (less than 0.001 kcal/(mol• atom• ps)) at the microcanonical (NVE) ensemble for both two polypeptides (see details in Sec.6 of the supporting information.), 1-ns GEBF-GAP based MD simulation using a Langevin thermostat 54 are performed at 300 K with a timestep of 1 fs in the canonical (NVT) ensemble. During the MD simulation, no QM calculations are performed and the MLFFs are not renewed online anymore. Figure 3 shows the changes of end-to-end distances between Cα atoms of the first and the last amino acid residues during the MD simulation. Three representative structures at different times are also plotted in Figure 4. As the trajectories show large conformation changes of the polypeptides from the chain-like extended structure to the folded one, 1000 structures for both two target systems are randomly sampled from the trajectories as testing set to evaluate the performance of our MLFFs. The electronic structure calculations on testing sets are carried out at the ωB97XD/6-31G* level with the Gaussian 16 package, 55 and the GEBF-PM6 calculations were performed with MOPAC package 56 and our LSQC program. 57 3.2. Relative Energy Prediction and Structure Optimization. After MLFFs have been constructed, we first show the applicability of the MLFFs on relative energy prediction. The energies of are conformers in testing sets are calculated with GEBF-GAP, PM6, ff14SB and ωB97XD/6-31G*. Here, GEBF-PM6 is used as the baseline for GEBF-GAP and the energy of the first conformer in the test set was taken as zero. For six conformers (the structures of conformers are plotted in Sec.7 of supporting information.) randomly chosen from the testing sets, the absolute deviations of relative energies (relative to the ωB97XD/6-31G* results) are shown in Figure 4a. One can note that the largest deviations are less than 6 kcal/mol for GEBF-GAP results, but are much larger (more than 18 kcal/mol) for PM6 and ff14SB results. Clearly, PM6 and ff14SB methods cannot correctly predict the relative stability of different conformers if these conformers are close in energies. The results indicate that our MLFFs method could be used to search for the low-energy conformers of systems under study. Further, we also test whether our MLFFs are suitable for structure optimization. The conformers with the lowest energy predicted by GEBF-GAP (using GEBF-PM6 method as baseline) in test sets are optimized with the BFGS algorithm 58 (implemented in ASE package 59 ). Figure 4b shows optimized structures obtained with GEBF-GAP and ωB97XD/6-31G* for 4ZNN and 1XQ8 segments. The root-mean-square deviation (RMSD) between DFT and MLFF results is 0.31 and 0.36 on 4ZNN and 1XQ8 segment, respectively. The geometrical parameters obtained with our MLFFs are very close to the corresponding values from the ωB97XD method. In addition, the geometries optimized with PM6 and ff14SB are also calculated for comparison. At respectively optimized structures, the absolute energy deviations predicted by GEBF-GAP, PM6, ff14SB (relative to the ωB97XD/6-31G* results) are 4.14, 13.96, 21.33 kcal/mol, respectively, for 4ZNN, and 0.85, 20.40, 24.60 kcal/mol, respectively, for 1XQ8 segment. Among these three methods, only the relative energies of MLFFs at their optimized structures are in good agreement with those from ωB97XD. Although our MLFFs are not trained from NMR structures of proteins, we also applied GEBF-GAP and ωB97XD/6-31G* method to obtain the optimized structure with the NMR structure of 1XQ8 segment as the initial geometry. The optimized structures are shown in Figure S1. The RMSD between the MLFFs optimized structure and the reference DFT optimized structure is only 0.27 , which suggests that the present MLFFs are also reliable for geometry optimizations outside the MD trajectories. ## Verification of Accuracy for GEBF-MLFFs During the MD Simulation. To investigate the applicability of our MLFFs on MD simulations. We first performed 20-ps MD simulations with GEBF-GAP, ff14SB and PM6 methods, respectively, the GEBF-PM6 method is used as the baseline for GEBF-GAP. MD simulations with ωB97XD/6-31G* are also carried out for comparison. Figure 5 display the dihedral angle distribution calculated with the GEBF-GAP and ωB97XD/6-31G* method. For each backbone dihedral φ, ψ and ꞷ, histograms are accumulated for all amino acid residues except Gly. The results suggest that the distributions obtained from the GEBF-GAP and ωB97XD/6-31G* methods are very close to each other. The distributions predicted by the ff14SB and PM6 methods are plotted in Figure S2 and S3, respectively. The dihedral distributions from these two methods are quite different from the ωB97XD/6-31G* methods. For dihedrals φ and ψ, the shapes of distribution show a great difference when compared with the results from ωB97X-D/6-31G*. For dihedral angle ꞷ, the peak intensity predicted by ff14SB is 20 % larger than the ωB97X-D/6-31G* result, and the deviation of the location of peak predicted by PM6 method from the ωB97X-D/6-31G* one reaches 10°. One can conclude that the dihedral angle distributions from GEBF-GAP are much more accurate than those from the ff14SB and PM6 methods. Table 1 Then, we evaluate the accuracy of our MLFFs on the testing sets, which are randomly sampled from 1-ns GEBF-GAP based trajectories at 300K. As the GEBF-PM6 method is employed as the baseline of the MLFFs, the accuracy of GEBF-PM6 with respect to conventional PM6 is first evaluated. The mean absolute errors (MAEs) of energies between GEBF-PM6 and PM6 on the testing set are only 0.003 kcal/(mol• atom). 10 conformers of the 4ZNN and 1XQ8 segment are randomly chosen from the testing set. The deviations of GEBF-PM6 energies relative to conventional PM6 ones are listed in Table S2. The maximum deviation is only about 0.008 kcal/(mol• atom). Thus, the errors of GEBF-PM6 results with respect to the conventional PM6 ones are negligible for the two polypeptides. In Table 1, the root mean squared errors (RMSEs) of energy and forces obtained with the GEBF-GAP, relative to the conventional ωB97XD/6-31G* are shown. For both two systems, the RMSEs of energy and forces for GEBF-GAP results are about 0.024 kcal/(mol• atom) and 1.5 kcal/(mol• ), respectively. For comparison, the RMSEs of PM6 and ff14SB force field results in energies and forces, relative to the conventional ωB97XD/6-31G* results, are also shown in Table S3. For two polypeptides, the RMSEs with ff14SB are 0.13 kcal/(mol• atom) and 12 kcal/(mol• ), respectively. The RMSEs with PM6 are 0.06 kcal/(mol• atom) and 14 kcal/(mol• ), respectively. These results indicate that our MLFFs are much more accurate than the PM6 or ff14SB method. We also use NN to parametrize subsystems, with the training set generated using the GEBF-ML protocol. DeePMD-kit 19 is adopted to construct the GEBF-NN force field from QM energy differences. Table 1 also shows the accuracy of GEBF-NN on the testing sets. For both two polypeptides, the RMSEs of the energy and forces for GEBF-NN results are about 0.020 kcal/(mol• atom) and 1.3 kcal/(mol• ). Both GEBF-GAP and GEBF-NN could predict the energies and forces with full QM quality. The high accuracy of GEBF-NN also indicates that with the training set generated from GEBF-GAP, our MLFFs can also be constructed by NNs. Although only PM6 calculations on small subsystems are needed for the GEBF-PM6 method and the computation cost of PM6 calculations is smaller than some MLFFs on small molecules, 60 we also construct the GEBF-MLFFs from QM energies directly. Using 1XQ8 segment as an example, the root mean squared errors (RMSEs) of energy and forces obtained with the GEBF-MLFFs, relative to the conventional ωB97XD/6-31G* are shown in Table 2. Using the same training sets (5020 subsystem configurations), the RMSEs of energy and forces for GEBF-GAP on the testing set are about 0.045 kcal/(mol• atom) and 2.5 kcal/(mol• ), respectively. The RMSEs for GEBF-NN are about 0.028 kcal/(mol• atom) and 2.1 kcal/(mol• ), respectively. Both GEBF-GAP and GEBF-NN constructed from QM energies show slightly larger errors than those constructed with energy differences. Fortunately, the accuracy of GEBF-MLFFs constructed from QM energies can further increase by adding more subsystem configurations in training sets. Using 5020 subsystem configurations as preliminary training sets, GEBF-GAP based MD simulation was performed to sampling more subsystem conformers. With 6405 subsystem configurations as the training set, the RMSEs for GEBF-GAP are 0.04 kcal/(mol• atom) and 2.3 kcal/(mol• ), respectively. While the RMSEs for GEBF-NN are 0.026 kcal/(mol• atom) and 1.8 kcal/(mol• ), respectively. The accuracies of both GEBF-MLFFs are increased by adding more data points in the training set. Moreover, the GEBF-NN constructed from QM energies shows similar accuracy as GEBF-MLFFs constructed from en-ergy differences. Thus, accurate GEBF-MLFFs can be constructed either from energy differences between QM and PM6 methods with relatively small training sets or from QM energies of subsystems with more subsystem configurations in the training set. 3.4. MD Simulation with GEBF-MLFFs Constructed from QM Energies. With the GEBF-NN force field constructed from QM energies, 1-ns MD simulations using a Langevin thermostat have been performed for 1XQ8 segment at 300 K with a timestep of 1 fs. To quantitatively describe the conformational changes, the RMSDs with respect to the initial structure of the 1XQ8 segment during the simulation are shown in Figure 6a. The RMSD increases rapidly and reaches the maximum value of 10 during the MD simulation. Thus, the trajectory also shows large conformation changes during the MD simulation. To evaluate the accuracy of the GEBF-NN force field during the simulation, 1000 configurations are evenly sampled from the 1ns trajectory and the mean absolute error (MAE) with respect to traditional ωB97X-D/6-31G* are calculated on each configuration. Figure 6b shows the time evolution of MAE for forces during the 1-ns MD simulations. One can see that the MAEs on almost all configurations are less than 2 kcal/(mol• ). Thus, GEBF-NN force fields can also be constructed from QM energies with full QM quality if enough subsystem configurations are added in training sets. Finally, it is also necessary to demonstrate the computational cost and scalability of our GEBF and GEBF-MLFF approach by comparing them with the conventional QM method and classical force field. Here, we take polypeptides ACE-(Ala)n-NME (n = 50, 100 and 150) as examples. All calculations are carried out on 48-core Intel Xeon Platinum 8163 2.5 GHz CPU. In Table 3, we present the total CPU time required for ACE-(Ala)n-NME (n = 50, 100 and 150) at different theoretical levels (ωB97XD and GEBF-ωB97XD with 6-31G* basis set, ff14SB, GEBF-GAP and GEBF-NN with or without GEBF-PM6 as baseline). First, we compared the scalability of different methods. It can be seen from Table 3 that all the tested methods (except the conventional DFT method) show linear scaling behavior. For instance, the total CPU time required by GEBF-GAP constructed from energy difference is about 1.9 and 2.9 times for ACE-(Ala)100-NME and ACE-(Ala)150-NME than that for ACE-(Ala)50-NME (39.52 seconds). Then, we compare the computational cost of our GEBF method and GEBF-MLFFs with the QM method and traditional force field. As summarized in Table 3, for all three polypeptides, the computational cost of GEBF-DFT is smaller than the conventional DFT method and the acceleration ratio on ACE-(Ala)150-NME is about 20. Thus, the computational costs of QM calculations are highly reduced during the training set construction. For all GEBF-MLFFs, the acceleration ratios are at least three magnitude orders than the full QM calculations. Thus, our GEBF-MLFFs show the low computational cost, with respect to QM methods. However, MLFFs are slower than the ff14SB force field, because much more parameters are needed to describe MLFFs. Further, the computational costs of GEBF-MLFFs constructed from QM energies or energy differences are also compared in Table 3. One can see that the computational costs of two GEBF-GAP force fields are similar. For all tested polypeptides, the total CPU time required by GEBF-GAP constructed from energy difference is about 1.2 times than that from QM energies. However, for cost-effective ML potential like Deep Potential, the computational cost of GEBF-PM6 cannot be neglected, the total CPU time required by GEBF-NN constructed from energy differences is about 3 times than that from QM energies. GEBF-GAP construct from energy differences may be appropriate for the online training process and nanosecondscale MD simulations. If microseconds-scale MD simulations are needed, it is more convenient to construct GEBF-NNs from QM energies directly. ## CONCLUSIONS In summary, we developed a general GEBF-ML protocol to automatically construct MLFFs for proteins with QM accuracy. Using GAP as the ML model, our protocol can automatically generate training sets with high efficiency. Moreover, for a given protein, only QM calculations on small subsystems containing a few residues are required in the construction of training sets. To facilitate the construction of training sets for various proteins, we create a protein's data library, which contains all data of subsystems generated from trained proteins. With this protein's data library, for a new protein only its subsystems with new topological structures are required for the construction of the corresponding training sets. Using two polypeptides 4ZNN and 1XQ8 segment, as examples. The accuracy of the constructed GEBF-GAP for both systems is validated by comparing the conformational energies, optimized structure, and MD simulation results with those from conventional DFT results. Our results show that GEBF-GAP can lead to quite accurate energies and forces similar to those from full QM calculations, and dihedral angle distributions from GEBF-GAP MD simulations are in good agreement with those from ab initio MD simulations. In addition, we also demonstrated that full QM quality GEBF-NN force fields can also be constructed using the training sets generated by GEBF-GAP. Thus, this work provides an efficient and systematic way to build MLFF for proteins, we also expected GEBF-ML protocol could be used for polymer materials and complex biological systems in aqueous solutions in the future. ## ASSOCIATED CONTENT Supporting Information. Computational efficiency, additional ML results, additional MD results, fragmentation scheme, the construction of GEBF subsystems. This material is available free of charge via the Internet at http://pubs.acs.org. ## Table of Contents Molecular dynamic simulation based on quantum mechanics (QM) can give highly accurate results but at high computational costs. Herein, we propose a protocol for the first time to construct machine learning force fields with QM quality at the cost of some QM calculations on subsystems not stored in a data library. This work takes an important step into the practical computational study of biological systems with QM accuracy.

chemsum

{"title": "Building Machine Learning Force Fields of Proteins with Fragmentbased Approach and Data Transfer", "journal": "ChemRxiv"}

diffuson-mediated_thermal_and_ionic_transport_in_superionic_conductors

## Abstract: Ultra-low lattice thermal conductivity as often found in superionic compounds is greatly beneficial for thermoelectric performance, however, a high ionic conductivity can lead to device degradation. Conversely, high ionic conductivities are searched for materials in solid-state battery applications. It is commonly thought that ionic transport induces low thermal conductivity and that ion and thermal transport are not completely independent properties of a material. However, no direct comparison or underlying physical relationship has been shown between the two. Here we establish that ionic transport can be varied independent of thermal transport in Ag + superionic conductors, even though both phenomena arise from atomic vibrations. Thermal conductivity measurements, in conjunction with two-channel lattice dynamics modeling, reveals that the vast majority of Ag + vibrations have non-propagating diffuson-like character, which provides a rational for how these two transport properties can be independent. Our results provide conceptually novel lattice dynamical insights to ionic transport and confirm that ion transport is not a requirement for ultra-low thermal conductivity.Consequently, this work bridges the fields of solid state ionics and thermal transport, thus providing design strategies for functional ionic conducting materials from a vibrational perspective. ## Introduction Superionic conductors are sparking tremendous interest in multiple fields. While fast ionic conductors are searched for solid state batteries, 1 they also often possess low lattice thermal conductivities essential for high thermoelectric efficiencies. In fact, many Ag + and Cu + superionic thermoelectric materials have thermal conductivities below the theoretical "minimum" thermal conductivity for solids, leading some to the suggestion that ion mobility makes these materials more liquid-like. 2,4 However, no physical connections between thermal and ionic transport have thus far been shown. Recently, a foundational misconception within the concept of "phonon liquid electron crystal" 2 was proven by showing that transverse phonon modes, which should not exist in a liquid, do persist above superionic phase transitions in solids. 5,6 Concurrently, a universal theory for heat conduction in solids has been developed that suggests low thermal conductivities can arise solely from (static) atomic disorder, strong anharmonicity and/or complex unit cells, all of which are inherent to fast ion conducting materials. 3,5,6 Lacking a direct comparison between ionic transport data and the thermal and vibrational properties in these systems, the interdependence of both transport processes remains elusive. Fundamentally, thermal and ionic transport are related at the vibrational level as they both arise from fluctuations of the phonon occupation number , which is the instantaneous number of vibrations (phonon quanta) that are in a vibrational mode having frequency . Thermal transport results from the thermodynamic drive to have a constant phonon occupation (thermal energy) throughout the entire material. Ion transport occurs when a thermal fluctuation is energetically capable of moving the ion between adjacent lattice sites. The rate at which phonons move throughout a material depends on the character of the phonon modes. Historically, heat transport by phonons was thought to occur in a propagating manner on a length-scale much larger than interatomic distances (phonon-gas model). In disordered, anharmonic and structurally more complex solids, heat transport can alternatively be conducted by fundamentally different transport mechanisms, one of which is the diffuson. 10,11 In contrast to the phonon-gas model, diffusons correspond to heat transport via local (atomic) scale random walk. 12,13 This diffusive walk of heat 14 means that non-propagating vibrations in a solid transfer thermal energy between adjacent diffuson modes, 11,12 at a much smaller length scale compared to typical propagating modes (shown schematically in Figure 1 a, b). Regardless of the character of the vibrational modes, fluctuations in vibrational energy remain the fundamental origin of thermal transport. 7 On the other hand, ionic transport is restricted by an activation barrier (nominally referred to as EA) that is largely determined by the potential energy landscape, i.e. the coordination environment of the mobile species (Figure 1 c). 15 With that, the magnitude of the fluctuation needed to move an ion is significantly increased, relative to thermal transport, since the kinetic energy of the ion has to surpass this activation barrier. Despite this conceptual difference there are strong underlying physical relationships suggesting a connection between thermal diffusons and ionic diffusion: 1) fast ionic transport is achieved in strongly disordered materials 16,17 and strong disorder promotes the prevalence of thermal transport via diffusons. 12 2) Both transport phenomena operate within an atomic-scale random walk, in which local vibrations carry and transfer heat energy or momentum of ions (Figure 1 b, d). 12,18 3) Anharmonic lattice vibrations enhance diffuson-mediated transport 14 and, although the characteristic vibration of a mobile ion is usually shown using a harmonic potential well, an ion jump is an intrinsically anharmonic process. 19 Motivated by these apparent similarities, this work aims to provide a stepping stone in unifying the concepts of thermal and ionic transport by experimentally accessing both processes and answering the following questions: 1) Are significant diffuson contributions to thermal conductivity present in superionic conductors? 2) Which vibrational modes characterize both transport processes, and 3) how does the magnitude of ion transport influence thermal transport? Especially the latter is of significant relevance, with high ionic conductivities hindering the long-term stability of superionic conductors in thermoelectric devices 20 and with batteries comprised of ionic conductors needing thermal management to decrease charging times and improve safety. ## Results and Discussion Structural features. To answer these questions, a successful isovalent substitution from Ag8SiSe6 to Ag8GeSe6 and Ag8SnSe6 is achieved as shown by the linear increase of the lattice volume (Supplemental Note 1, 2). 21 These materials are superionic conductors and have been intensely investigated for their thermoelectric transport properties due to the low thermal conductivity. 3,22,23 The different compositions have varying room-temperature structures, 24,25 but ultimately undergo a phase transition into the same cubic structure that is characterized by its strongly disordered Ag + sublattice (Figure 2 a and Supplemental Note 2). 3,22 Structurally, there are a large number of Ag + sites that are tetrahedrally coordinated and have face-sharing connectivity. These sites have an average occupation of only 25 %. While the low-temperature phase of Ag8SiSe6 is cubic with an ordered Ag + sublattice, 24 both Ag8GeSe6 and Ag8SnSe6 crystallize in an orthorhombic structure at room-temperature (Supplemental Note 2). In the orthorhombic phase, the Ag + sublattice consists of five distinct lattice sites, each either tetrahedrally or trigonally coordinated by Se 2arranged in a corner-or edge-sharing fashion, respectively. Near the phase transition, the lattice volume increases strongly in a small temperature window before it settles into a volumetric thermal expansion coefficient ranging from 7.7 to 8.7 × 10 -5 K -1 , with an average of 8.2 ± 0.5 × 10 -5 K -1 for all compositions (Supplemental Note 2, 3). While the influence of local structural changes are usually employed as an important metric in the understanding of ionic conduction 26 , the underlying vibrational influences remain elusive. Here, we focus on the vibrational characterization of these superionic conductors to give new insights relating vibrations, ionic jumps and thermal transport. Vibrational frequencies. The vibrational spectrum of a material has significant importance for the fields of ionic conduction and thermal conduction. For ionic transport, the attempt frequency of the mobile species is included in the Arrhenius pre-factor (Supplemental Note 4). 15,27 Often the Debye frequency D is used to approximate the attempt frequency despite not probing the vibrations of the mobile species directly. In fact, especially in disordered and soft materials, the Debye-frequency often fails its original purpose to estimate the maximum vibrationalfrequency, but rather falls into the center of the vibrational density of states. 12,28 With that, it provides a useful experimental insight to the average dynamics of the lattice, but it should not be expected to be an accurate descriptor for the vibrational properties of the mobile ion. 28 A better descriptor for the attempt frequency would be a vibrational frequency E specific for the mobile ion that is conceptually similar to an Einstein oscillator. 29,30 This concept is especially useful for materials with an underlying guest-host structural motif, like the mobile ion within a rigid sub-lattice in superionic conductors 2 or guest atoms and associated "rattler modes" in the skutterudites and clathrates, which have rather dispersion-less phonon branches (Supplemental Note 5). 31 Here, we find that the average Ag + vibrational frequency (1.2 ± 0.2 THz; range = 1.0 -1.5 THz) determined from X-ray diffraction measurements is consistent with the average Ag + frequency found computationally (2 THz), and perfectly aligns with the lowest frequency peak in the vibrational density of states (Figure 2 c). Meanwhile, the average Debye frequency (3.1 ± 0.2 THz; range = 2.5 -3.4 THz) found from speed of sound measurements does capture the average vibrational frequency of the entire density of states (~3 THz), but does not correspond to any particular feature. A detailed discussion of the measurements and further comparison can be found in the Supporting Information (Supplemental Notes 2,3,5). It is important to note that the ability to directly probe vibrations of the mobile ion provides experimental insights to the frequency spectrum relevant for ionic transport. From a thermal transport point-of-view, the Debye frequency is frequently shown to provide a strong and simple approximation even for complex materials, while the determination of Einstein frequencies can provide a useful tool if atom-specific frequencies play a significant role. 31,35 Thus, the characterization of both and provides complementary information in examining thermal and ionic transport together. that, the Ag + on this lattice site has an inherently higher displacement in the direction of the jump, typically believed to be favorable for ionic transport. Ag + ionic transport. The thermoelectric properties of these fast ionic conductors, such as ultralow thermal conductivity, are often inferred to be connected to ion mobility, but the corresponding magnitude of ionic conductivity is rarely (if ever) reported. Here, we evaluate the magnitude and temperature dependence of the ion conduction of Ag8(Si,Ge,Sn)Se6 solid solutions (see Supplemental Note 4). The ionic conductivities vary strongly upon substitution with room-temperature values ranging from 0.088 to 5.0 S/m as determined by impedance spectroscopy. The respective Arrhenius behavior for Ag8SiSe6, Ag8GeSe6 and Ag8SnSe6 reveals significant changes to the activation barrier and, with that, the temperature dependence of ionic transport (Figure 3 a). Here, Ag8GeSe6 exhibits the lowest activation barrier of 0.05 eV while both Ag8SiSe6 and Ag8SnSe6 show larger barriers of ~0.3 eV at temperatures above 298 K. These results are confirmed by nuclear magnetic resonance spectroscopy (Figure 3 b), which indicates that the activation energies are predominantly a bulk property of the solid and not caused by microstructural differences. Furthermore, not only the disordered high-temperature cubic phase but also the ordered low-temperature phases of the Ag + argyrodites have significant ionic conduction. For all investigated materials, the logarithm of the pre-exponential factor scales linearly with the activation barrier (Figure 3 c). This is a common occurrence in ionic materials, known as the Meyer-Neldel rule, and is attributed to the interrelation of migration enthalpy (i.e. Δ ∝ ) and entropy (i.e. Δ ∝ ln ( )). This relation has been derived before using transition state theory and the multi-excitation entropy model. 36,37 Specifically, it is believed that the inverse slope of the Meyer-Neldel plot is related to the energy of the vibrational modes that participate in ionic conduction, as well as a characteristic number of vibrations. However, a consistent definition of these quantities has been lacking. 37 Nevertheless, an inverse slope of ~37 meV is in agreement with those reported for other superionic conductors. 39 In comparison to the energy range of the calculated phonon density of states it is unclear what this value really represents, except that the vibrational states of the Ag + ions do not change significantly through the substitution series, and that the overall vibrational properties are not significantly affected by the composition. By novel considerations of the phonon entropy (Supplemental Note 6), we show that using an average frequency of Ag + vibrational modes (~2 THz) in the evaluation of the Meyer-Neldel slope coincides with a characteristic number of vibrations (~4.6 phonons) nearby the value expected (~3.1 phonons) from Bose-Einstein statistics at room temperature. This result provides confidence that it is, in fact, the Ag + vibrations that are responsible for ion transport within the multi-excitation model. So far, we have used and as bulk isotropic properties of the solid, but it is important to remember the local scale on which each ion jump occurs. Consequently, there are local activation barriers for each possible jump from a given Ag + site, and each Ag + site can have its own vibrational spectrum. Here, we examine one exemplary Ag + jump from both a static (bond valence sum analysis) and lattice dynamical view-point (see Supplemental Note 7 for further discussion). The bond valence sum analysis reveals the potential energy landscape of an ionic jump between two tetrahedrally coordinated Ag3 equilibrium sites (Figure 3 d). The Ag + ion has to jump through the face of the tetrahedron (local potential energy maximum) to a tetrahedrally coordinated interstitial site (local potential energy minimum) from which it can jump to either adjacent Ag + site. The local activation barrier for this jump is expected to be small, since only minor coordination changes (from 4 -3 -4) are necessary and the interstitial site is energetically stablilized. 40 From a lattice dynamics perspective, it is possible to reduce the partial Ag + vibrational density of states (blue shading in Typically, changes in the activation barrier and magnitude of ionic conductivity are discussed from the static perspective, e.g., electronegativity, unit cell and pathway volumes. 15 The sitespecific and directional analysis of phonon modes proposed here allows for identification of prominent vibrational frequencies of the mobile ion from lattice dynamics calculations. The importance of tracking specific vibrational modes within molecular dynamics simulations was recently discussed by Gordiz et al. 41 They showed that the vibrational contributions to ionic conduction are heavily frequency-dependent. Herein, we also assess the relevance and vibrational character (phonon-gas-like or diffuson-like) of phonon modes for thermal transport. With that, we aim to draw conclusions about the interdependence of both phenomena. Diffuson-mediated thermal transport. The ultra-low thermal conductivities of superionic conductors are often attributed to (static and dynamic) atomic disorder, soft bonding, lowfrequency optical phonon modes, anharmonicity and complex crystal structure. 42 All of these parameters are associated with either lowering the phonon group-velocity or increasing the phonon scattering rate, in context of the phonon-gas model. 3,22,23 While these structural and vibrational features are likely to factor into the characteristics of thermal transport, the vicinity to the Ioffe-Regel limit and the associated concept of minimum lattice thermal conductivity suggests the phonon-gas model is likely an incomplete description of thermal transport in these systems. 12 Two-channel modeling based on the calculated lattice dynamics for Ag8GeSe6 is utilized to explain these results. Here, the heat current operator matrix is analyzed regarding its diagonal (phonon-like) and off-diagonal (diffuson-like) elements such that the total lattice thermal conductivity of the two-channel model is = !" + $ %% . 13 Given the computational cost of third and fourth order force constants, an analytical scattering model was implemented to describe the temperature dependence of the phonon lifetimes (details in Supplemental Note 9). 13 The analytical model considers phonon-phonon scattering (& ! '( ) and boundary scattering (& ) '( ) resulting in a total scattering rate of with C1, C2 and A as constants that describe the experimental data. The phonon gas channel contributions, !" , dominate the total lattice thermal conductivity at low-temperatures and well-describe the peak in thermal conductivity (Figure 4 c). With increasing temperature, the phonon gas channel declines while vice-versa the diffuson contributions, $ %% , increase strongly, before plateauing above 100 K. The resulting total lattice thermal conductivity of the two-channel model, = !" + $ %% , captures both the magnitude and temperature-dependence of the experimental data, not only in the phonon (T < 50 K) and diffuson (T > 100 K) mediated regimes but also at temperatures where both channels have similar contributions. The two lines shown for the total lattice thermal conductivity represent the upper-and lower-limit of the calculation, stemming from a slight anisotropy of the orthorhombic structure (Figure 4 b, c, for detail see Supplemental Note 9). This observation is supported by the reported single-crystal thermal conductivity analysis of Ag8SnSe6, which found a comparable anisotropy in thermal conduction at 300 K. 44 It should be emphasized that the conclusion that the diffuson channel dominates is independent of the form of the analytical scattering functions. Corresponding results utilizing a simplified phonon-phonon scattering term are shown in the Supporting Information (Supplemental Note 9). The two-channel model additionally allows for a spectral analysis of the thermal conductivity of both channels (Figure 4 d, e). The contributions from the phonon gas channel arise from the lowest frequency modes below 1 THz, while the spectral thermal conductivity of the diffusonchannel is extended over the whole frequency range. Note that the spectral thermal conductivity is not expected to be zero at zero frequency but is an apparent artifact of uniform q-meshing. 45 Here, the minimal contributions of the high-frequency vibrations, dominated by M = Si, Ge and Sn (see Figure 2 c), explain why no significant changes to the thermal transport are observed in the solid-solutions. Nevertheless, at temperatures relevant for ion transport, the Ag + vibrations have predominantly diffuson-like character. Evaluation of the mode-specific Grüneisen parameters (Supplemental Note 5) lends an explanation for the strong decline of the phonon gas channel (Figure 4 f). Here, large Grüneisen parameters are found in the same frequency range in which the phonon gas channel contributions are the strongest (Figure 4 d, e). Therefore, the extensive anharmonicity at low frequencies is believed to be the driving factor for the strong suppression of the phonon-channel and the subsequent transition to diffuson-dominated thermal transport. This was captured phenomenologically by our analytical scattering function. It is not surprising that such large Grüneisen parameters exist since anharmonic vibrations have been linked to ionic transport. 5,6,46 Having assessed the vibrational character of phonon modes from thermal transport behavior, it is also possible to characterize the vibrational character directly from the mode eigenvectors. The spatial extension and "shape" of phonon modes has been used previously to distinguish vibrational characters. Here, we use the participation ratio for such analysis (Figure 4 g, details in Supplemental Note 10). 9,10 A participation ratio close to unity is indicative of a high spatial extension of a mode and typically found in simple crystalline materials like c-Si that have textbook phonon-gas transport, 10 and so a participation ratio of 1 can be called the phonon limit. Participation ratios below ~0.1 are typical for localized modes that do not contribute to thermal transport, and the localization limit is, by definition, when only 1 atom (out of all the atoms in the unit cell) participates in the vibrational mode. 9 The calculated participation ratios are in the intermediate range, comparable to those of amorphous materials like a-C and a-SiO2 9 , despite the crystalline nature of Ag8GeSe6. While the participation ratio cannot clearly separate phonons and diffusons, 9,47 the similarity to the results of amorphous materials strongly indicates the presence of diffuson-like vibrations complementary to the two-channel model. Again, this indicates that Ag + vibrations in particular are predominantly diffuson-like. These results confirm diffuson-mediated thermal transport in the Ag + argyrodites at temperatures relevant for thermoelectrics and ionic conduction. Accordingly, the phonon-gas channel has only a minor contribution. It can be concluded that a complex crystal structure, resulting in many energetically close phonon modes, and relatively large anharmonicities are sufficient to achieve diffuson-dominated thermal transport in crystalline materials. This is to say that we do not need to explicitly invoke ion transport effects in order to capture the magnitude of thermal conductivity. Thus, significant diffuson-contributions can be expected in other structurally well-defined materials. Connecting ionic and thermal diffusion. The results presented above have the surprising observation that the thermal conductivity of Ag + argyrodites is seemingly independent of ionic transport as the diffuson-based thermal conductivity does not change significantly, while the changes in ion transport (based on Nernst-Einstein diffusion coefficients, Supplemental Note 4) are large across the same temperature range (Figure 5 a). To comprehend the underlying differences of thermal and ionic transport one can examine the number of Ag + ions participating in the respective processes. While all Ag + vibrational modes (and with that all Ag + ions) contribute to thermal transport at high temperatures (e.g., Figure 4 e), the instantaneous fraction of mobile Ag + participating in ionic transport is given by the Boltzmann distribution, where F G) HI is the number density of mobile Ag + ions, F JGJKH is the total number density of Ag + ions, and L is the macroscopic activation barrier (Figure 5 b). Consequently, even with an exceptionally low activation barrier of 0.05 eV, comparable to other superionic conductors like Cu2Se, 48 only ~10% of the ions are thermally mobile (without an applied electric field) at any given time over the investigated temperature range. For higher activation barriers, this fraction stays well below 1%. This observation sheds light on the misconception of a "liquid-like" sublattice, given that temperatures of more than 1000 K would be necessary to even reach a 50% thermally mobile sublattice. While instantaneous mobile ion fractions on the order of several percent could be expected to significantly impact thermal transport within the phonon gas model, the nature of diffuson modes makes them less likely to be largely affected so long as the mode energies do not change drastically as more ions become mobile. This does not seem to be the case, nor does the magnitude of ionic conductivity suggest any significant thermal transport by ion mobility (see Supplemental Note 11). At a deeper level, the difference between ionic and thermal transport is better understood by considering their similarities, that is the microscopic origin of both processes: phonon occupation fluctuations. Thermal transport in solids arises from any deviation ( − ) ≠ 0 from the equilibrium phonon occupation number , with being the instantaneous occupation of mode N (see Supplemental Note 6). Only the drift velocity of the phonon (e.g., phonon-gaslike or diffuson-like) determines the magnitude of its contribution to transport. In contrast, ionic motion is an activated process such that a critical fluctuation has to be reached for the ion to overcome the activation barrier. The magnitude of this critical fluctuation will depend on the vibrational mode(s) involved and the specific jump that is considered (for example see Figure 3 d-f). For a single vibrational mode that oscillates in the jump direction, the critical number of phonons O required for a jump to occur is where is the frequency of the phonon mode and corresponds to the activation barrier of this particular jump (for a mode-specific description see Supplemental Information 6). The ceiling operator is used because phonons are quantized. Intuitively, larger barriers require larger phonon fluctuations. For a Ag + vibrational mode where ℎ ≈ 5 meV, then Y ≈ 10 to 100 phonons using macroscopic values of for in Eq. 3. Similar to the fraction of mobile ions (Eq. 2), only a relatively small fraction of all phonon occupation fluctuations can result in an ionic jump, even if multiple modes participate in the fluctuation. When calculating the probability that a fluctuation is large enough for ionic transport (see Supplemental Information 6) it is significantly lower (blue area in Figure 5 c) than, for example, the probability that a fluctuation is within one standard deviation of the equilibrium phonon occupation number (red area in Figure 5 b). Since the standard deviation of an exponential distribution is equal to its average value, it follows that the majority of phonon fluctuations are within the bounds 0 ≤ ≤ 2 , and ≈ 5.0 phonons for Ag + vibrational modes near room temperature. From this perspective it is not surprising that thermal transport is independent of the activation barrier or ionic transport (Figure 5 c) since thermal transport is dominated primarily by average phonon fluctuations. Discussing ion jumps in the context of phonon fluctuations is a novel concept and may be a first step towards a lattice dynamical theory of ion transport. Generally, one could use the jumpdirection projected density of states for each Ag + site to elucidate jump probabilities for each frequency and each (compare Figure 3 d, e). From our frequency-dependent analysis, we reach an important conclusion that lower frequencies are more beneficial for ionic transport (Supplemental Note 6), independent of the material, temperature or activation barrier. This is a crucial insight for ion conduction research showing that the local jump dynamics need to be tailored rather than the global lattice dynamics. 27,28 The inherently low frequencies of Ag + vibrations in the Ag8MSe6 argyrodites may thus be a contributing factor to their superionic behavior. The prominent contribution of Ag + vibrational modes to both thermal and ionic transport is made clear by comparing the vibrational frequencies of those found experimentally and from the calculated density of states (Figure 5 d). First, the average diffuson frequency found from the spectral two-channel calculation ( $ %% = 2.3 THz) is in excellent agreement with the value estimated experimentally ( $ %% = 2.2 ± 0.2 THz, range 1.9 THz to 2.4 THz) using an analytical diffuson transport model (Supplemental Note 3). 12 Comparatively, an average ionic transport frequency is found by using the average of the partial Ag + density of states ( _ `= 2.0 THz). Meanwhile, the Einstein frequency determined from X-ray diffraction experiments ( = 1.2 THz) coincides with the maximum of the partial Ag + density of states, which consists of modes that contribute to large thermal displacements. The fact that these average frequencies all reside between 1 and 3 THz where Ag + makes up ~70% of the vibrational density of states, and that Ag + vibrations have been shown to be largely diffuson-like in nature means that diffusons are foundational to understanding both thermal and ionic transport in these materials. Consequently, the vibrational character is likely important to understand transport in other ionic conductors, with or without ultralow thermal conductivity. Outlook on engineering transport in ionic conductors. As we have shown, the magnitude of ionic conductivity can be changed independently without altering the thermal transport properties that are governed by diffusons. Therefore, strategies to lower the ionic conductivity, and consequently improve the long-term stability in thermoelectric devices, are not expected to have detrimental effects on the desirable low lattice thermal conductivities. This profound insight opens the possibility to tailor superionic conductors and expand their usage in commercial thermoelectric devices 20,49 by using the known design principles in superionic conductors. 15,40 In addition, our results lead to an important conclusion for the research of solid ionic conductors that are becoming increasingly relevant for the performance optimization of all solid-state batteries. 1,26 Here, the Meyer-Neldel behavior is a bottleneck in achieving higher ionic conductivities since a lowering of the activation barrier has detrimental effects on the conductivity pre-factor. 15,27,36,39 With that, reducing the slope of the Meyer-Neldel plot becomes a main goal. 39 By deriving the multi-excitation entropy using phonon-fluctuation considerations, we show that the Meyer-Neldel slope can be related to the prominent vibrational modes of ionic conduction and a characteristic number of phonons found to be in the range of equilibrium phonon occupations (Supplemental Note 6). 7 This leads to a novel concept and design principle in solid-electrolyte research: determining and manipulating phonon occupations (e.g. targeted phonon excitation 41 , through opto-electronic stimulation 50 ) in conjunction with tailoring the site-specific vibrational frequencies and activation energies. ## Conclusion This work demonstrates that diffuson-mediated thermal transport is dominant in Ag + argyrodite superionic conductors at temperatures relevant for thermoelectrics and ionic transport. An important consequence of this fact is that ionic conductivity can be varied by orders of magnitude without affecting the thermal conductivity. The similarities of the argyrodite structure with lithium and sodium superionic conductors means that diffusons are likely prevalent in battery materials as well. Thus, by understanding the origins of thermal and ionic transport, in particular via vibrational characteristics and phonon occupations, novel design concepts are possible to further improve functional ionic conducting materials.

chemsum

{"title": "Diffuson-mediated thermal and ionic transport in superionic conductors", "journal": "ChemRxiv"}

transition_metal_catalyzed_stereodivergent_synthesis_of_<i>syn</i>-_and_<i>anti</i>-δ-vinyl-lactams:

## Abstract: A stereodivergent and diastereoselective transition-metal-catalyzed intramolecular hydroamidation of allenes and alkynes furnishing d-vinyl-lactams is reported. Employing a rhodium catalyst allowed for the selective synthesis of the syn-d-lactam. Conversely, a palladium catalyst led to the formation of the antid-lactam in high selectivity. The new method shows high functional group compatibility and assorted synthetic transformations were demonstrated as well as its utility for the enantioselective formal total syntheses of (À)-cermizine C and (À)-senepodine G. Transition metal-catalyzed intramolecular hydroamination reactions starting from aminoalkenes, 1 aminoallenes, 2 aminodienes 3 and aminoalkynes 4 have been reported frequently for the synthesis of nitrogen-containing heterocycles. Although the d-lactam moiety is of equally high interest, to date reports of inter-and intramolecular hydroamidations are still rare and stereoselective and stereodivergent variants are unknown. 5 Thus, d-(and g-) lactams are found as pharmacophores in a number of drugs and bioactives, such as (+)-ebmamonin (antiarrhythmic agent), rhynchophylline (treatment of disorder of the central nervous system) and BMD 188 (induces the apoptotic death of prostate cancer cells). Furthermore d-lactams have been used as key intermediates for the synthesis of piperidine type alkaloids such as cermizine C & D and as senepodine G (Fig. 1). 6 Our group recently reported a series of inter-and intramolecular addition reactions of various nucleophiles and pronucleophiles to allenes and alkynes covering regio-, enantioand diastereoselective C-O, C-S, C-N and C-C bond formations. 7,8 Thus, these new methods represent atom-economic alternatives to the transition-metal-catalyzed allylic substitution 9 and allylic oxidation. 10 We herein disclose an unprecedented stereodivergent rhodium-and palladium-catalyzed intramolecular hydroamidation to gain selective access to either syn-or anti-vinyl-d-lactams. Our studies commenced by employing the phenyl-bsubstituted tosyl amide model substrates 1 (Table 1). After a frst successful reactivity test using a Rh/dppf catalytic system (d.r. 85/15), an optimization of the reaction parameters was undertaken (Table 1). Employing [Rh(COD)Cl] 2 /dppf enabled us to selectively obtain the syn-confgurated product albeit in low conversion (entry 2). Fortunately, utilizing chloroacetic acid as a Brønsted acid additive could improve the conversion dramatically to 96% (entry 3). Furthermore, lowering the reaction temperature from 80 C to room temperature led to optimal results (92% isolated yield) and diastereoselectivity of 91 : 9 in favor of the syn-diastereomer (entry 4). Conversely, by altering the metal source to palladium, a complete inversion of the diastereoselectivity in favor of the anti-diastereomer was observed. The combination of [Pd(dba) 2 ], dppf and chloroacetic acid at 80 C led to the best results in terms of diastereoselectivity (6/94) and yield (96%) (entry 8). Further modifying the reaction temperature or examining different additives and ligands showed no improvement. 11 With optimized conditions in hand, the scope and limitations of this reaction were explored (Table 2). Alkyl-, vinyl-and cyclic-functionalized amides behaved well and furnished the Cite this: Chem. Sci., 2019, 10, 3074 All publication charges for this article have been paid for by the Royal Society of Chemistry corresponding syn-and anti-lactams 3a to 6a and 3b to 6b in good to excellent yields and d.r. Next an assorted variety of different functionalized, aromatic b-substituted amides was investigated. Substrates bearing phenyl, biaryl, naphthyl and vinyl groups provided the corresponding lactams in yields up to 95% and d.r. up to 96/4 (2a, 7a, 8a and 2b, 7b, 8b). Compounds 2a and 2b were also synthesized via large scale catalysis without any decline in yield and d.r., respectively. 11 Sensitivity towards steric hindrance was tested employing meta and para substituted derivatives. 13 Functional groups like thioethers and halides attached to the aromatic ring were compatible and afforded the desired products 12a & 13a and 12b & 13b in good to excellent yields. Besides b-substituted amides also asubstituted amides could be cyclized in good yields, though with lower diastereoselectivities (14a & 14b). Overall the Pd-catalyzed reaction towards the anti-product showed slightly better results in terms of d.r. in case of aliphatic-and cyclic-functionalized amides compared to the Rhcatalyzed protocol. The relative confguration of syn-and anti-confgured products were determined by NOE-experiments and confrmed by X-ray diffraction analysis (Fig. 2). 11 To extend the use of this reaction even further, alkyne substrates bearing isopropyl, phenyl and anisole as substituents were subjected to rhodium catalysis conditions (Table 3). The alkyne substrates needed higher reaction temperatures in order to obtain the d-lactams in sufficiently high yields. Thus, yields up to 88% and diastereoselectivities up to 87/14 in favor of the syn-diastereomer were obtained (Table 3). Unfortunately, the palladium catalyst system did not show any reactivity for the terminal alkyne substrates. 11 To demonstrate that the reaction does not just provide access to highly diastereoselective lactams but also enables their stereoselective synthesis, enantiomerically enriched starting material 16 was subjected to the catalysis conditions. 11,13 We were satisfed to observe that the enantiomeric purity was maintained under both the syn-3a-and anti-3b-catalysis conditions (Scheme 1). To investigate the reversibility of the reaction, both products were subjected to the contrary conditions. The anti-lactam subjected to the "syn-conditions" showed no change of relative confguration. Conversely, the syn-lactam exposed to the "anti-conditions" resulted in an inversion of the relative confguration (Scheme 2). 11 To gain insight into the respective stabilities of the syn-and the anti-product, DFT calculations (BP86/def2SVP) were performed. The calculations showed that the anti-is approximately 1.4 kcal mol 1 more stable than the syn-diastereomer. 11 In conclusion, we posit that in presence of rhodium the reaction is driven by kinetic control, whereas in presence of palladium the reaction towards the syn-diastereomer is reversible and the formation of the more stable anti-diastereomer either under thermodynamic control or under product development control. 11 The mechanism of the rhodiumcatalyzed addition of nucleophiles to allenes and alkynes was investigated by our group recently. 14 Based on those results, we propose the following mechanistic rationale for the formation of the syn-product (Scheme 2, left-hand cycle). First, the active rhodium species is generated via oxidative addition, followed by a ligand exchange to form intermediate A. Hydrometalation then gives rise to the key allylrhodium species B. In the confguration-determining step, a reductive elimination via an inner sphere mechanism takes place favoring the syn-product 7a. 15 For the palladium-mediated reaction (Scheme 2, righthand cycle) we suggested a similar mechanism to the Tsuji-Trost reaction. 16 First the active palladium species undergoes hydrometalation to form the p-allyl species C. Then the C-N bond is formed by nucleophilic attack on to the p-allyl intermediate C, via an outer sphere mechanism, favoring the antiproduct. The utility of 2b as scaffold in the synthesis of more complex molecules was demonstrated by performing assorted transformations (Scheme 3). Ozonolysis of the allylic moiety delivered the C 1 -shortened aldehyde 18 in excellent yield (99%). The C 1 -chain-elongated aldehyde 19 was accessed through hydroformylation, employing the self-assembly ligand 6-diphenylphosphinopyridine (6-DPPon) in excellent yield (98%) and regioselectivity. Cleaving the tosyl group under reductive conditions yielded the unprotected lactam 20. Finally, a hydrolysis was performed as a gateway to access diastereomerically enriched d-amino acid 21. Inspired by the variety of functionalization, the newly developed lactam synthesis was applied in the formal total synthesis of cermizine C and senepodine G (Scheme 4). Both natural products were isolated for the frst time in 2004 from the club moss lycopodium carnuum and chinense by KOBAYASHI. 17 This representative from the lycopodium family as well as even more complex alkaloids were often targeted in the total synthesis, due to their high variety, unusual skeletons and biological properties. Some efforts were made to fnd an easy access to these alkaloids and especially to form the main core unit. 21 A frst synthesis was reported by Snider starting from (S)-piperidine ethanol. 22 Even though the synthesis was efficient and elegant, the starting material is expensive and its synthesis not trivial. Other attempts were either based on long reaction sequencesor used auxiliary chemistry to introduce the desired stereochemistry. 23,24 Our interest in cermizine C and senepodine G was initially stimulated by the quinolizidine core, which is accessible in a straightforward fashion by applying the present methodology (Scheme 4). The attempt for the enantioselective formal total synthesis of this compounds was therefore started from (S)methyltosylamide ( 16) which was accessed in two steps from an a,b-unsaturated ester 22. 11 The enantiomerically enriched starting material 16 was subjected to a gram-scale catalytic cyclization and delivered the tosyl-protected lactam 3b in excellent yield (98%), diastereoselectivity (d.r. ¼ 90/10) and enantioselectivity (95% ee). Next was the deprotection of tosyl-lactam 3b followed by an alkylation reaction to obtain 24 in a good yield. With precursor 24 in hand, a Grubbs ring closing metathesis followed by catalytic hydrogenation furnished the quinolizidine core 25, which was previously converted into cermizine C and senepodine G by Snider et. al. 22 Hence, we have realized a highly efficient stereoselective formal total synthesis of these alkaloids (7 steps, 31% overall yield) starting from compound 22. In comparison Snider was able to synthesis compound 25 in 5 steps and an overall yield of 41%, starting from (S)-piperidine ethanol. 25 However, our method compares favorably, in terms of starting material accessibility and costs, to the procedure developed by Snider. ## Conclusions In conclusion, we have established a general and efficient stereodivergent and highly diastereoselective procedure to gain selective access to syn-and anti-vinyl-d-lactams by using either a rhodium or palladium-based catalytic system. The reaction tolerates a wide range of functional groups which enables the synthesis of a variety of different d-lactams. Assorted transformations allowed the functionalization of both the alkene and lactam moiety. Furthermore, we successfully utilized the new developed methodology in a highly stereospecifc and atom economic formal total synthesis of cerminzin C and senepodine G.

chemsum

{"title": "Transition metal catalyzed stereodivergent synthesis of <i>syn</i>- and <i>anti</i>-\u03b4-vinyl-lactams: formal total synthesis of (\u2212)-cermizine C and (\u2212)-senepodine G", "journal": "Royal Society of Chemistry (RSC)"}

total_synthesis_of_himastatin

## Abstract: The synthesis and study of antibiotic natural products with unique structures and mechanisms of action represents a proven strategy to combat the public health crisis posed by antibiotic-resistant bacteria. The natural product himastatin is an antibiotic with an unusual homodimeric structure that presents a significant synthetic challenge. We report the concise total synthesis of himastatin by a newly developed final-stage dimerization strategy that was inspired by a detailed consideration of its biogenesis. Combining our bio-inspired dimerization approach with a modular synthesis enabled expedient access to a number of designed derivatives of himastatin, including synthetic probes that provide insight into its antibiotic activity. identify a new strategy that stands apart from our group's prior radical-based approaches to secure C sp3 -C sp3 linkages and C sp3 -C sp2 linkages between similar (13) and dissimilar fragments (14). We began with a detailed examination of (-)-himastatin's (1) biogenesis from a linear peptide 4 that is cyclized and then subject to oxidative tailoring by three cytochrome p450 enzymes (8). The final step, catalyzed by HmtS, forges the central C5-C5' bond by oxidative dimerization of (+)-himastatin monomer (2). Based on recent theoretical studies of p450-catalyzed C-C bond formation, we envisioned that this enzymatic dimerization may take place via radical-radical coupling of two cyclotryptophan radicals (Fig. S2) (15,16). These radical species are likely generated in rapid succession via indoline N-H hydrogen-atom abstraction at the heme active site, before undergoing combination in its vicinity (16,17). We envisioned that a biogenetically-inspired chemical method for the oxidative dimerization of cyclotryptophans could follow the same radical-radical coupling blueprint. As opposed to hydrogen atom abstraction, we planned to generate an analogous open-shell cyclotryptophan species via singleelectron oxidation of the embedded aniline substructure. Consistent with studies of aniline dimerization via single-electron oxidation (18,19,20), we predicted that the resulting arylamine radical cation would rapidly dimerize at the most accessible position, forming the desired C5-C5' linkage. Late-stage application of this chemistry to dimerization of (+)-himastatin monomer (2) permits a straightforward modular assembly of linear hexadepsipeptide 5 akin to native precursor 4, without the constraints imposed by bidirectional elaboration of a simple dimeric cyclotryptophan (9,11,12). Direct union of complex peptide macrocycles also offers the elusive opportunity to access the first heterodimeric derivatives of (-)-himastatin (1). Our new dimerization method required the identification of a single-electron oxidant that would target the aniline substructure within a complex cyclotryptophan precursor (21). We discovered that stoichiometric silver(I) hexafluoroantimonate, in combination with the non-nucleophilic pyrimidine base TTBP (22) in 1,2-dichloroethane, could effect C5−C5' dimerization of cyclotryptophan, cyclotryptamine, and indoline derivatives (Fig. 2A). In each case, a single regioisomer consistent with a symmetric C5−C5' linked homodimer was isolated. Single crystal X-ray diffraction of dimeric endodiketopiperazine (+)-7h verified the expected connectivity. The use of an aqueous sodium thiosulfate reductive workup was critical for optimal isolation of the products due to their sensitivity toward further oxidation under the reaction conditions (23,24). Notably, we found that exo-configured diketopiperazines 6e and 6g were subject to complete oxidation in approximately half the time of their corresponding endo-derivatives 6f and 6h, respectively. This finding correlates with the increased accessibility of the N1 locus in substrates 6e and 6g, the site of initial oxidation (25). Substitution of N1 with a methyl group in the case of indoline 6k did not inhibit the dimerization, consistent with a radical intermediate as opposed to a closed-shell arenium cation (26). In order to expand the range of reagents that could be utilized in more complex applications of our dimerization method, we also investigated the use of copper(II) salts as single-electron oxidants (20). Cyclotryptophan dimer (-)-7a could be obtained using catalytic copper(II) trifluoromethanesulfonate and silver(I) carbonate as the terminal oxidant, albeit in lower yield (34%, 18% RSM) compared to stoichiometric AgSbF 6 (54%). To investigate the mechanism of this C−C bond forming dimerization reaction, we devised a series of experiments using indoline substrates (Figs. 2B and S3) (23). When an equimolar mixture of C2-methyl and C2-phenyl indolines 6i and 6j, respectively, were subjected to our dimerization conditions, we observed a statistical mixture of homo-and heterodimers that arise from similar rates of single-electron oxidation (Fig. 2B, green; Fig. S3, eq. 1). However, oxidative dimerization of an equal mixture of indolines 6j and 6k gave predominant (90%) homodimer formation, along with a trace (4%) amount of heterodimer 7n (Fig. S3, eq. 2). When a limiting quantity of oxidant was used, we determined that these indoline substrates were consumed sequentially, with N1-methyl indoline 6k dimerizing selectively over NH indoline 6j (Fig. 2B, blue; Fig. S3 eq. 3). Having observed homodimerization of a more readily oxidized monomer in the presence of a similarly nucleophilic but less readily oxidized monomer, we conclude that C5−C5' bond formation preferentially occurs through radical-radical coupling rather than nucleophilic capture. This conclusion is consistent with the absence of adduct formation in the homodimerization of cyclotryptophan 6a despite the presence of external π-nucleophiles (e.g. methallyltrimethylsilane, dimethylketene silyl acetal, N-trimethylsilylindoline), and is reinforced by prior studies demonstrating that radical-radical coupling between aniline radical cations is fast (k = ~10 7 M -1 •s -1 for the dimerization of PhNMe 2 •+ ) (18,19,20). We postulate that the high local concentration of radical species near the surface of the oxidant favors their direct combination over nucleophilic pathways (14,20). In the context of our synthetic efforts, the rapid rate and apparent insensitivity of the radical-radical coupling manifold to nucleophilic interference bode well for the application of this chemistry to complex substrates. These findings highlight a possible underlying parallel between our oxidative dimerization methodology and our mechanistic proposal for the biosynthetic dimerization catalyzed by HmtS (Fig. S2), involving successive generation of radical species in close proximity to each other. For the synthesis of (+)-himastatin monomer (2), we sought to leverage the practical advantages of solid-phase peptide synthesis (27), offering rapid and customizable access to complex peptides by minimizing repetitive purification and isolation steps. In contrast to the reported solution-based approach to intermediates en route to (-)-himastatin (1) (9), we relied on a hybrid solution-solid phase synthetic strategy. The resin-bound D-threonine 9 (Fig. 3) was elaborated with L-leucine (-)-10 and depsitripeptide fragment (+)-8, the latter being prepared in one step from a depsipeptide block (28) and known Nε,O-protected D-5-hydroxypiperazic acid (9). The crude depsipentapeptide acid (+)-11 obtained upon cleavage was then coupled with cyclotryptophan (-)-12 (Fig. S4), affording linear hexadepsipeptide (-)-13 in 55% overall yield from threonine resin 9 (23). The efficient hybrid synthetic strategy we have developed enables convergent assembly of intermediate hexadepsipeptide (-)-13 with only a single chromatographic purification, and compares favorably to linear solution-phase synthesis which requires at least 10 separate steps to access an intermediate of similar complexity (9). Furthermore, our modular strategy allows for conducting difficult couplings in solution (28), and introducing the tryptophan residue as a cyclotryptophan to bypass stereoselectivity concerns that would arise from late-stage oxidation (29). The resulting linear peptide (-)-13 was cyclized to (+)-himastatin monomer (2) in 46% overall yield (Fig. 4), affording the immediate biosynthetic precursor to (-)himastatin (1). All 1 H and 13 C NMR data as well as optical rotation for synthetic monomer (+)-2 were consistent with literature values (8,9). Having accessed (+)-himastatin monomer (2), we focused on the application of our biogenetically inspired oxidative dimerization methodology to complete the total synthesis of (-)-himastatin (1) (Fig. 3). While silver(I) hexafluoroantimonate and copper(II) trifluoromethanesulfonate were effective for the dimerization of simpler cyclotryptophans (Fig. 2), they gave little to no oxidation of the cyclotryptophan incorporated within the more complex (+)-himastatin monomer (2). We hypothesized that aggregation and inactivation of these insoluble oxidants, combined with the lower reactivity of complex macrocyclic peptide substrates, may be responsible for the low conversion, and sought to address the challenge posed by evaluating other single-electron oxidants. Consistent with this hypothesis, insoluble oxidants such as other Ag(I,II) and Cu(II) salts were generally ineffective. However, soluble oxidants including organic radical cations such as magic blue ((4-BrPh) 3 N •+ SbF 6 ), did provide oxidation, but products derived from nucleophilic substitution of the C-Br bond (S N Ar) by the peptide dominated (21). Informed by our prior use of Cu(II) for the dimerization of simpler substrates and in search of an oxidant with both good solubility and low propensity toward nucleophilic capture, we identified copper(II) hexafluoroantimonate. Our isolation of freshly prepared Cu(SbF 6 ) 2 , commonly used as a soluble Lewis acid catalyst (30), provided us with an opportunity to investigate its use as a stoichiometric oxidant. In the event, exposure of (+)-himastatin monomer (2) to Cu(SbF 6 ) 2 and DTBMP in 1,2-dichloroethane, afforded (-)-himastatin (1) in 40% yield, with only trace (<5%) amounts of recovered starting material. All spectroscopic data, as well as optical rotation, for synthetic (-)-himastatin (1) were consistent with literature values (6,9). Our concise and versatile chemical synthesis of himastatin, featuring a biogenetically inspired finalstage dimerization reaction, presented an opportunity both to interrogate structural characteristics that are important for its bioactivity, and to access synthetic probes for chemical biology studies (Fig. 4). We hypothesized that the alternating sequence of D,L-residues present in the macrocyclic rings of (-)himastatin (1) could promote self-assembly (31,32), inspiring our preparation of both the enantiomer (ent-(+)-1) and meso derivative of himastatin (1). These stereochemical probes were prepared from precursors of opposite chirality, and in the case of the heterodimer meso-himastatin (1), by dimerization of an equal mixture of monomer 2 enantiomers and separation of the resulting heterodimer (23). Apart from slight variations in the chemical shifts of aromatic 1 H and 13 C signals, the spectra of mesohimastatin (1) were nearly identical to the corresponding homodimers. We also selected several derivatives with single-residue substitutions to synthesize, each varying a residue that is unique to himastatin amongst related antibiotics. In all cases, our modular hybrid peptide synthesis approach was quickly adapted to introduce the substituted residue, and the resulting monomers were effectively dimerized (21-37% yield) using the conditions developed for the synthesis of (-)-himastatin (1) (Fig. 4A) (23). As an orthogonal mechanistic probe that would permit direct visualization of himastatin's interaction with bacteria, we designed a fluorescent heterodimer that we predicted would retain antibiotic activity (vide infra). TAMRA-himastatin heterodimer (-)-25 was rapidly prepared via the union of himastatin monomer (+)-2 and azidolysine monomer (+)-22 followed by labelling via a reduction-acylation sequence (Fig. 4B). This procedure also provided access to TAMRA-himastatin homodimer (-)-S17 as a useful control (23). We found that synthetic (-)-himastatin (1) showed antibiotic activity against several Gram-positive species, including antibiotic-resistant strains of public health concern (Fig. 4C, Table S10) (1). Our synthetic (-)-himastatin (1) showed similar MIC values (1-2 µg/mL) to those reported for natural (-)himastatin (1) in identical species (4). All monomeric derivatives prepared in this study had MIC values ≥64 µg/mL across all species tested (23), highlighting the critical role of dimerization for antibiotic activity. Our stereochemical probes revealed that the absolute stereochemistry of himastatin has negligible impact on its antibiotic activity; stereoisomers of himastatin (1) were found to have nearly identical MIC values across the B. subtilis, S. aureus, and E. faecalis strains tested. This finding has also been observed amongst enantiomers of certain membrane-targeting cyclic peptides with alternating stereochemistry (33), and is consistent with antibiotic activity depending on achiral as opposed to diastereomeric interactions that would lead to differential activity of each stereoisomer (e.g. with peptides or receptors) (34,35). In contrast, we found that ent-(+)-himastatin (1) was 4-8 fold more active in inhibiting the growth of the producing organism, Streptomyces himastatinicus, compared to (-)himastatin (1). This finding might be explained by the presence of known self-resistance mechanisms that have evolved in other species, such as enzymatic degradation and efflux, which would be expected to show differences between stereoisomers (36). The introduction of a strategically positioned functional handle in (-)-himastatin (1) was a key goal of our derivative design that would permit introduction of a fluorescent tag. We focused on L-leucine substitution, given the natural variation of this site among related antibiotics (Fig. S1). Replacement with an O-methyl serine residue (L-Ser(OMe), dimer (-)-21), which is found in (-)-chloptosin (S1), had minimal impact on antibiotic activity. A similar finding was observed upon substitution with Lazidolysine (L-Lys(N 3 ), dimer (-)-23), which offered the conjugation site exploited in our synthesis of fluorescent probes. However, unlike serine and azidolysine homodimers (-)-21 and (-)-23, respectively, the corresponding TAMRA-himastatin homodimer (-)-S17 was inactive (MIC >64 µg/mL) (Fig. S5, Table S10). In addition to TAMRA, homodimeric himastatin analogues derived from other fluorophores were also found to be inactive (Fig. S5). Consistent with our expectation that minimizing the overall perturbation of himastatin's structure to only one half of the dimer may preserve antibiotic activity, we found that the MIC of TAMRA-heterodimer (-)-25 (Fig. 4B) was indeed comparable to that of (-)himtastatin (1) in Bacilus subtilis (6 vs. 1 µg/mL). Thus, the opportunity for heterodimer formation offered by our biogenetically-inspired late-stage dimerization methodology was instrumental to secure access to a fluorescent himastatin probe (37), as well as other key derivatives including meso-himastatin (1) that would otherwise be significantly challenging to prepare using chemoenzymatic or bidirectional synthesis (9,10). Other structural features specific to (-)-himastatin (1) include a depsipeptide linkage and 5hydroxypiperazic acid residue. Evaluating the derivatives that we prepared to study these particular structural features, we observed a trend of decreasing antibiotic activity when the ester linkage was replaced with either a secondary amide (-)-15 or tertiary amide (-)-17, consistent with the loss of a hydrogen-bond site (38). Furthermore, when the 5-hydroxypiperazic acid residue was replaced with a proline residue, antibiotic activity was completely abolished. While proline residues are known to induce turn formation, especially when the adjacent amino acid is of opposite α-stereochemistry, they do not exhibit a rigidifying effect as pronounced as that seen in N-acyl piperazic acid derivatives (39). Consistent with the predicted loss of rigidity upon proline substitution, NMR spectra of homodimer (-)-19 and monomer (+)-18 in various solvents at 23 °C revealed the presence of minor conformers not observed in the spectra of (-)-himastatin (1) or our other derivatives. Taken together, these results provide evidence that structural rigidity, enforced by hydrogen-bonding and conformational restriction, is important to himastatin's antimicrobial mode of action. Confocal microscopy has been used to observe the biological effects of antibiotics on B. subtilis, including the first approved membrane-disrupting lipopeptide, daptomycin (37). We sought to use our synthetic compounds in conjunction with this experimental approach to further characterize the antibiotic activity of (-)-himastatin (1). Our synthetic heterodimeric probe, TAMRA-himastatin (-)-25, offered an opportunity to directly visualize its interaction with bacteria and monitor cellular localization. When B. subtilis cells were treated with TAMRA-himastatin (-)-25, we observed substantial accumulation in the bacterial envelope (Fig. S6A), with little to no intracellular staining seen at sublethal concentrations. The most intense sites of staining were observed at bacterial septa, in addition to patches of stain along sidewalls. At lethal concentrations (Fig. S6B), defects such as membrane extrusions coincided with lateral accumulation of TAMRA-himastatin (-)-25. These sites of curvature appear to reflect areas where the antibiotic has induced changes to membrane morphology. The staining pattern observed with TAMRA-himastatin (-)-25 was similar to that of the membrane stain FM4-64 with unmodified himastatin (1) (Fig. S7). Untreated B. subtilis cells have smooth membranes and normal septal rings, but cells treated with a sub-lethal concentration of either enantiomer of himastatin (1) display striking membrane defects, notably patches of membrane thickening. Furthermore, the observed similarity in membrane morphology between himastatin (1) enantiomers appears to be consistent with their similar antibiotic activity. In a separate experiment, we evaluated the timescale by which (-)-himastatin (1) acts on bacteria at lethal concentrations (Fig. S8). When treated with (-)-himastatin (1) at a concentration twice the MIC value, bacterial membranes were permeabilized within 30 minutes, as indicated by influx of the viability stain SYTOX Green. The observations of our microscopy studies are comparable to those seen with daptomycin despite a lack of structural similarity to himastatin (37). The membrane defects and localization patterns observed in B. subtilis with unmodified (-)-himastatin (1) and our fluorescent himastatin derivative (-)-25, show resemblance to those seen with unmodified and fluorescent forms of daptomycin, respectively (37). Furthermore, the short timescale of membrane permeabilization following treatment with himastatin (1), like daptomycin, is consistent with a mode of action based on physical perturbations (33,37). This mode of action is distinct from other Gram-positive peptide antibiotics, such as vancomycin and teixobactin, that interfere with cell-wall biosynthesis and have longer kill times (40). Separately, the similarity in MIC values and cellular morphology amongst our series of synthetic himastatin stereoisomers reveals that achiral interactions, for example with the hydrophobic groups of phospholipids (34,35), are largely responsible for the observed antibiotic activity. In summary, our chemical biology studies using our synthetic probes offer findings that are consistent with the hypothesis that (-)-himastatin's (1) antibiotic activity is dependent on interaction with bacterial membranes (7). It is evident that (-)-himastatin (1) is a structurally unique member amongst known membrane-disruptors. These antibiotics target an essential bacterial organelle that can be difficult to alter without severe fitness cost (41). As society continues to battle multidrug-resistant pathogens, membrane-disrupting antibiotics, like the FDA-approved daptomycin, represent an important frontier in the fight. Our bioinspired strategy for the total synthesis of (-)-himastatin (1) provides rapid access to derivatives and is enabling investigations that point to its antibiotic activity via membrane disruption. This effort aims to facilitate further inquiries to advance our understanding and exploit how himastatin's unique molecular structure contributes to its antibiotic activity. exo-(+)-7g, (11R,11R1 ') 23% endo-(+)-7h, (11S,11S1 ') 54% exo-(-)-7e, (2S,3R,2S1 ',3R1 ') 40% endo-(+)-7f, (2R,3S,2R1 ',3S1 ') 63%

chemsum

{"title": "Total Synthesis of Himastatin", "journal": "ChemRxiv"}

synthesis_of_flower-like_magnetite_nanoassembly:_application_in_the_efficient_reduction_of_nitroaren

## Abstract: A facile approach for the synthesis of magnetite microspheres with flower-like morphology is reported that proceeds via the reduction of iron(III) oxide under a hydrogen atmosphere. The ensuing magnetic catalyst is well characterized by XRD, FE-SEM, TEM, N 2 adsorption-desorption isotherm, and Mössbauer spectroscopy and explored for a simple yet efficient transfer hydrogenation reduction of a variety of nitroarenes to respective anilines in good to excellent yields (up to 98%) employing hydrazine hydrate. The catalyst could be easily separated at the end of a reaction using an external magnet and can be recycled up to 10 times without any loss in catalytic activity. The selective reduction of nitroarenes has attracted a great deal of attention as the resulting anilines are important intermediates for the manufacture of pharmaceuticals, dyes, polymers, and fine chemicals . Generally, the synthesis of anilines entails catalytic and non-catalytic methods employing different reducing agents . The non-catalytic processes use either Bechamp or sulphide reduction technology which generates large amounts of undesirable waste that is detrimental to the environment 10 . On the other hand, catalytic process is a well-established technology but relies on mainly expensive precious metal catalysts, namely Pd, Pt, and Ru which lack chemoselectivity in the presence of other common reducible functional groups 6, . Furthermore, when hydrogen is used as the reducing agent, high temperature and pressure are usually needed with requirement of the specialized equipment. These limitations can be circumvented using various hydrogen donors such as formic acid 4,5 , hydrazine hydrate 9, , ammonium salts 20 , and sodium borohydride 21 , among others, in presence of various metal catalysts. Amongst hydrogen donors, hydrazine monohydrate is particularly noteworthy as it produces only harmless by-products, such as nitrogen gas and water, and is relatively safe and easy to handle compared to its anhydrous form. From a sustainability perspective, substitution of precious metals by earth-abundant base metals is a highly desirable pursuit for heterogeneous catalysis. In this aspect, magnetic materials especially iron-based catalysts in organic synthesis have received significant attention, as iron is plentiful, cost effective, and relatively environmentally benign element . Consequently, it is not surprising that the reduction of nitroarenes has been reported utilizing a combination of hydrazine and iron catalysts namely various iron salts, its complexes, and oxide forms and as a supported catalysts 9, 15, 17-19, 28, 29 . The readily available magnetic iron oxide nanoparticles stand out to be very attractive candidates as they are cost-effective nanocatalysts; being recoverable effortlessly with an external magnet due to the paramagnetic behaviour thus avoiding cumbersome filtration/separation processes 12,29, . This strategy could significantly improve the catalytic efficiency and decrease the operational cost which is crucial for practical applications. Herein, we report a simple approach for the synthesis of magnetite via thermally induced solid state reaction of iron (III) oxide under hydrogen a atmosphere (Fig. 1). It is interesting to note that the flower/rod like morphology of the precursor is well preserved even after the hydrogen treatment. The as-prepared magnetite catalyst is characterized by several techniques, namely XRD, FE-SEM, TEM, nitrogen adsorption-desorption isotherm and Mössbauer spectroscopy. The magnetite acts as a catalyst for the transfer hydrogenation of nitroarenes with hydrazine hydrate as the reducing agent in a microwave reactor affording nearly quantitative yields. The salient features of this work are the excellent catalytic performance, simple procedure, easy separation, and the excellent reusability of the catalyst. ## Results and Discussion The preparation of magnetite microspheres by reduction of iron(III) oxide under a hydrogen atmosphere has been well investigated 40,41 . Herein, we report a novel method for the synthesis of magnetite with a unique flower-like morphology from iron(III) oxalate via a simple two-step approach. Firstly, thermally induced solid state decomposition of iron oxalate was used to produce iron(III) oxide (Fe 2 O 3 ) with ultra-small nanostructured particles; and secondly, the subsequent thermally induced reduction of the prepared iron(III) oxide under a hydrogen atmosphere afforded magnetite (Fe 3 O 4 ). Figure 2 depicts the stepwise transformation of iron(III) oxide to magnetite by hydrogen reduction process via in situ monitoring by XRD. The two shoulders (around 40° and 74° of 2θ) are clearly visible in the diffraction patterns up to 210 °C confirming that the material is iron(III) oxide with ultra-small particles. At 220 °C, the diffraction lines belonging to fcc structure of magnetite/maghemite start to emerge and their intensities gradually increased during the 60 min period of isothermal treatment. Thus, we choose temperature 220 °C for 2 hours as the optimum condition for the preparation of magnetite from iron(III) oxide with ultra-small particles using a tube furnace under a hydrogen atmosphere. Figure 3a depicts XRD pattern of magnetite sample. All of the diffraction lines can be clearly ascribed to standard face-centered cubic (fcc) structure of Fe 3 O 4 (space group: Fd3m (227), JCPDS card No. 01-089-3854). Although the isostructural character of magnetite and maghemite cause difficulties in direct and precise identification of these phases by XRD point of view, the cell parameter indicates correct suggestion; it varies from 0.8351 nm for maghemite and 0.8396 nm for stoichiometric magnetite 42,43 . The cell parameter of cubic structure in the prepared sample is a = 0.8394 nm, which is in good agreement with values described for magnetite in the literature 44 . Nevertheless, the Mössbauer spectroscopy is a powerful experimental technique which provides precise identification of valence state of iron atoms and cations distribution and more specifically, for the identification of iron compounds. Consequently, Mössbauer spectroscopy was used for direct identification of iron oxide's state (Fig. 2b). The acquired spectrum is composed of two magnetically split subspectra (i.e., sextets). The first sextet component with an isomer shift (δ) value of 0.27 mm s −1 , quadrupole shift (ε Q ) value of −0.01 mm/s and hyperfine magnetic field (B hf ) value of 49.0 T corresponds to Fe 3+ ions occupying all the tetrahedral positions in the Fe 3 O 4 crystal structure and with a contribution from Fe 3+ ions sitting in the octahedral sites having Fe 3+ ions as the nearest neighbours (i.e. Fe 3+ -O-Fe 3+ pathway). On the other hand, the second sextet with δ = 0.67 mm/s, ε Q = 0.00 mm/s, and B hf = 46.0 T is ascribed to Fe 2+ and Fe 3+ ions occupying the octahedral positions in the Fe 3 O 4 crystal structure among which the electron hopping occurs (i.e., an Fe 2+ ion with a neighbouring Fe 3+ ion and vice versa; Fe 2+ -O-Fe 3+ pathway) with a frequency faster than the characteristic time of the Mössbauer technique and thus manifested as a component with δ value lying in the range typical of an average valence state of + 2.5 45 . Relative spectral area of Fe 3+ and Fe 2.5+ sextet is 41 and 59%, respectively. This difference from ideal spectral area of 33 (Fe 3+ sextet) and 67% (Fe 2.5+ sextet) for stoichiometric magnetite indicates a nonstoichiometry in magnetite. In particular, Fe 3+ ions having Fe 3+ as the nearest neighbours in the octahedral sites, which thus do not participate in the electron hopping process, forms their own subspectrum with hyperfine parameters values very close to those of the subspectrum representing Fe 3+ in the tetrahedral sites. Therefore, these two subspectra with nearly identical parameters are fitted as one and its relative area is increased to the detriment of subspectra representing mixed valence Fe 2.5+ in octahedral sites. The morphology of the prepared samples was obtained using scanning electron microscope (SEM) and transmission electron microscope (TEM). The SEM image (Fig. 4a) of magnetite revealed the retention of rod/flower like pattern as found in the case of iron(III) oxide 46 . From the TEM image (Fig. 4b), it can be seen that the individual nanorods possess an average breadth of size 300 nm while the self-assembled floral pattern has a diameter of about 3 μm. TEM image (Fig. 4b) also showed the porous nature of the rods composed of interconnected microspheres. The sharp diffraction spot due to various planes of Fe 3 O 4 in selected area electron diffraction (SAED) pattern (Fig. 4b inset) of the particles reveals well crystalline nature of materials. The width of the nanorods forming rod/ flower like pattern was found to be uniform along its entire length as evidenced from the TEM image. The pores might be formed during recrystallization process or from the elimination of water during the reduction process. The N 2 adsorption-desorption isotherms show Type II isotherm for the magnetite with a small hysteresis (Fig. 5) which reveal macroporous nature with cylindrical pores. The specific surface area obtained from BET method is 20 m 2 /g. Furthermore to examine the efficiency of the catalyst, we evaluated its reduction prowess for a variety of nitro compounds to their corresponding industrially important amine derivatives in ethanol under microwave (MW) irradiation. Initially, to optimize the reaction conditions, various parameters such as effect of temperature, catalyst loading, solvent, and different hydrogen source including the amount of hydrazine hydrate were studied by choosing nitrobenzene as a model substrate. As expected, no reaction occurred in the absence of magnetite and hydrazine hydrate (Table 1, entries 1-3). Firstly, the reaction was carried out under conventional heating condition using magnetite (30 mg) as a catalyst and 150 μL hydrazine hydrate as hydrogen source in ethanol at 90 °C; complete conversion occurred in 3 h (Table 1, entry 14). Interestingly, when the reaction was performed under MW irradiation condition, the complete conversion could be achieved within 15 min (Table 1, entry 7) and no trace of substrate, intermediates or side product was evident by GC analysis. To explore the optimum amount of needed catalyst, different catalyst loadings (10, 20, and 30 mg) were investigated which revealed that 30 mg catalyst was the optimum loading that afforded > 99% conversion of nitrobenzene (Table 1, entry 5). The quantity of hydrazine hydrate did impact the conversion rate; reaction using 60 μL resulted in only 90% conversion (Table 1, entry 9), while 100 μL delivered quantitative conversion (Table 1, entry 7). Next, the effect of temperature on reduction reactions was determined and at 50 °C, 35% conversion and at 70 °C, 72% conversion was observed (Table 1, entries 10, and 11); increasing the temperature to 90 °C, however, afforded quantitative conversion within 15 min (Table 1, entry 7). Time variation, a crucial factor, was investigated next for complete conversion; 10 minutes delivered 94% conversion and 91% yield (Table 1, entry 12). Notably, with isopropyl alcohol as a hydrogen donor, no reaction occurred (Table 1, entry 13). The catalyst could catalyse the reaction under conventional heating conditions as well; however, an extended reaction time was required up to 240 min (Table 1, entry 14). Further, we observed that the reaction did not occur at room temperature and heating was essential for accomplishing this reaction (Table 1, entry 15). The effect of different solvents on the reduction of nitrobenzene was also investigated and it was discerned that ethanol, 2-propanol, and acetonitrile afforded good conversion and yields (Table 2, entries 1, 3, and 5). For THF, the efficiency of the reduction significantly decreased and only 4% conversion was obtained while a mixture of EtOH:H 2 O (1:1) showed moderate conversion (Table 2, entry 2). Assorted iron catalysts for the reduction of nitrobenzene with hydrazine hydrate were also examined under the optimized conditions. Notably, commercial Fe powder, FeSO 4 .7H 2 O, FeCl 3 .6H 2 O (Table 3, entries 1, 2 and 4) did not show any activity under these conditions, while FeCl 3 .4H 2 O, Fe(acac) 3 and magnetite exhibited 18%, > 99%, and > 99% conversion, respectively (Table 3, entries 3, 5, and 6). Although, as-prepared magnetite and Fe(acac) 3 have the same conversion and selectivity, but due to the homogeneous nature of Fe(acac) 3 , it cannot be recycled which limits its applications. In contrast, magnetite is a heterogeneous catalyst and has shown superiority due to its magnetic separation property and importantly, the ease of recyclability. These optimized reaction conditions were then applied to an array of selected substituted nitroarenes bearing additional reducible groups to ascertain the chemoselectivity aspect and wider scope of the catalyst (Table 4). In most of the cases, quantitative ( > 99%) conversion of the substrates to the desired amine derivatives occurred within 15 min. It was observed that for 5-nitro-1H-indole, sterically hindered 1-methyl-2-nitrobenzene and 6-nitro-2,3-dihydrobenzo dioxine, the reactions were completed in 25 min (Table 4, entries 4, 9, and 15), while 3-fluoro nitrobenzene 4-methoxy nitrobenzene, and 4-methyl nitrobenzene exhibited 99% conversion in 22 min (Table 4, entries 10, 13 and 14). The sole exception was 4-nitrobenzamide which showed 42% conversion in 25 min (Table 4, entry 7) presumably due to the polar nature of the amidic compound. Interestingly, halogenated nitroarenes such as 2-chloro-4-iodo-1-nitrobenzene, 3-fluoro nitrobenzene, 4-bromo nitrobenzene, and 4-chloro nitrobenzene showed excellent conversions (Table 4, entries 1, 10, 11, and 12) without any dehalogenated product being observed. Easily reducible ester groups were well accommodated in this catalytic system (Table 4, entries 5 and 6). The catalytic prowess became apparent in the reduction of 4-nitrobenzonitrile, 6-nitroquinoline, and methyl (4-nitrophenyl)sulfane with excellent yield of the corresponding desired products (Table 4, entries 2, 3, and 8). The catalyst recycling is certainly very essential in heterogeneous catalytic reactions. Therefore, we examined the recyclability of our developed catalyst for reduction reaction by using nitrobenzene as a model substrate under the optimized conditions. After completion of the reaction, the catalyst could be easily separated using an external magnet. The separated spent catalyst was then washed with ethanol and dried before reuse. This process was repeated 10 times successfully without any noticeable decrease in catalytic activity (Fig. 6) suggesting that the catalyst could find application in the practical reduction of nitroarenes on industrial scale. The leaching aspect of any iron after recycling was examined by determining the metal content of reaction solution using AAS (atomic Continued absorption spectroscopy) after removing of the catalyst; metal content was found to be 0.0735 ± 20% mg/L, which shows negligible leaching of iron from the catalyst which bodes well for its robustness and reusability. As mentioned in previous reports, that reduction of nitroarenes can proceed via two common routes 10,47,48 . The first direct route proceeds via nitrosobenzene and N-phenylhydroxylamine intermediates, (Fig. 7a). In contrast, the second route involves the condensation of nitrosobenzene and N-phenylhydroxylamine which advances through the intermediacy of azoxybenzene, azobenzene, and hydrazobenzene. In order to determine the exact route for this reduction of nitrobenzene, a reaction under identical reaction conditions was conducted for azobenzene. At the end of reaction, only hydroazobenzene could be isolated and no trace of aniline was detected which confirmed that reduction of nitroarene proceeded via first direct route. In view of experimental validation of the direct route and on the basis of previous literature reports, a plausible mechanism is proposed . The reaction initiates with the adsorption of hydrazine on the surface of magnetite nanoflowers followed by bond dissociation which produces nitrogen and surface-bound hydrogen as metal hydride. The nitroarenes adsorbed on the surface of magnetite thus get transformed to nitrosoarenes after reaction with surface adsorbed hydrogen. These highly active nitroso moieties further react with hydrogen to form stable hydroxylamine; hydrogenation of hydroxylamine is slow and the rate determining step. In the next step, two proton transfers produce the desired aniline derivatives (Fig. 7). ## Conclusion In summary, we have established a robust, chemoselective and magnetically reusable catalyst for the reduction of industrially valuable nitroarenes substrates in the presence of other sensitive reducible functional groups. A diverse range of amines derivatives could be obtained expeditiously (15 min) in excellent yields under the MW heating conditions at 90 °C using hydrazine hydrate as a hydrogen source which precludes the use of a precious metal catalysts and hydrogen gas in the preparation of amines derivatives. The magnetite with a unique morphology prepared by our method was found to be very stable and could be used ten times successfully with minor decrease in its catalytic activity. The excellent catalytic performance, simple and a safe procedure, easy separation, and the recyclability make this environmentally benign catalytic system a remarkable and useful alternative to other Fe-based catalytic systems. ## Methods Materials. All solvents, hydrazine hydrate (50-60%), iron (II) chloride tetrahydrate (99.99%), oxalic acid (98%), N,N-dimethylacetamide (DMA) ( ≥ 99%), were purchased from Aldrich as analytical grade and were used without further purification. Preparation of the magnetite catalyst. In a typical synthesis protocol, 1 mmol (0.126 g) of oxalic acid (H 2 C 2 O 4 .2H 2 O) was dissolved in 10 mL of DMA under continuous magnetic stirring and admixed with an equal mole ratio (0.198 g) of aqueous iron chloride (FeCl 2 .4H 2 O) followed by addition of 12 mL deionised water. After stirring for 10 min, the as-obtained yellow coloured product (iron oxalate) was separated by centrifugation and washed with ethanol several times and dried at 333 K for 12 h. The as-prepared iron oxalate was thermally treated in air at the conversion temperature of 448 K for 12 h to obtain mesoporous iron(III) oxide (Fe 2 O 3 ) 46 . Further, magnetite (Fe 3 O 4 ) was prepared by thermally induced solid state reaction of iron(III) oxide in hydrogen gas at 220 °C for 2 h. General procedure for the reduction of nitrobenzene. Into a 10 mL microwave vial equipped with a magnetic stir bar, was placed 0.5 mmol of nitro compound in ethanol (1.5 mL), 100 μL of hydrazine hydrate followed by 30 mg catalyst. The vial was sealed with a Teflon-lined septum and irradiated with microwaves in a Monowave 300 single-mode MW reactor (Anton Paar GmbH, Graz, Austria) at 90 °C for 15 min. Progress of the reaction was monitored by TLC (silica gel; hexane/ethyl acetate) and the conversion and yield were determined by GC (gas chromatography) using n-hexadecane as an internal standard. Characterization. XRD patterns of materials were recorded on an X'Pert PRO diffractometer (PANanalytical) in Bragg-Brentano geometry with iron-filtered Co-Kα radiation (λ = 1.7903 ) equipped with fast X' celerator detector. The reaction chamber XRK900 (Anton Paar) mounted to the diffractometer was employed for in situ monitoring of the preparation of the magnetite sample. Data were processed in High Score Plus Software in conjunction with PDF-4 + and ICSD databases. The 57 Fe Mössbauer Spectroscopy measurements were carried out to investigate iron-bearing phase compositions in the studied samples. Mössbauer spectra were recorded with 1024 channels and measured at room temperature employing MS2006 Mössbauer spectrometer based on virtual instrumentation technique 53,54 , operating at a constant acceleration mode and equipped with a 57 Co(Rh) source. The acquired Mössbauer spectra were processed (i.e., noise filtering and fitting) using the MossWinn software program 55 . The isomer shift values were referred against α-Fe foil sample at room temperature. FESEM images were recorded on a Hitachi 6600 FEG microscope operating in the secondary electron mode and using an accelerating voltage of 5 kV. Detailed particle size and morphological studies of solid samples were performed by TEM on a JEOL JEM-2010 instrument equipped by a LaB 6 cathode (accelerating voltage of 160 kV; point-to-point resolution of 0.194 nm). A drop of high-purity ethanol was placed onto a holey carbon film supported by a copper-mesh TEM grid (SPI Supplies, USA) and air-dried at room temperature. The dimensions of the microspheres were measured using ITEM software. Nitrogen adsorption-desorption isotherms at 77.4 K were measured up to the saturation pressure of nitrogen (molecular cross-sectional area 0.162 nm 2 ), and obtained by the static volumetric technique on an Autosorb-iQ-C analyzer (Quantachrome). Prior to the measurements, samples were degassed at room temperature for 12 h to reach pressure below 0.001 torr. Specific surface areas were calculated using the multipoint BET (Brunauer-Emmett-Teller) model. The best fits were obtained using adsorption data in the relative pressures of 0.08/0.25 (P/P 0 ). The analysis and evaluations were performed with the ASiQwin 2.0 software package (Quantachrome). For the reaction, 10 mL glass vial equipped with Teflon-lined cap was irradiated in a Monowave 300 single-mode microwave reactor (Anton Paar GmbH, Graz, Austria) having auto adjusting MW power to maintain the reaction temperature. The nitroarenes reduction products were analyzed using an Agilent 6820 GC equipped with an Agilent DB-5 capillary column (30 m × 0.32 mm, 0.5 m) under the operation parameters: inlet temperature of 100 °C, temperature of flame ionization detector of 250 °C, temperature ramp of the oven from 100 to 250 °C at a rate of 10 °C min −1 .

chemsum

{"title": "Synthesis of flower-like magnetite nanoassembly: Application in the efficient reduction of nitroarenes", "journal": "Scientific Reports - Nature"}

can_halogen_clusters_be_emissive?

## Abstract: Halogen-halogen short contacts, especially halogen bonds (XBs) have been widely utilized in multifarious fields, owing to its bridging function among luminophores as well as well-known heavy atom effect. However, little attention has been paid to the luminescent ability of halogen clusters. It remains unknown whether they are emissive. Herein, inspirited by the clustering-triggered emission of nonconventional luminophores, we report the first examples of emissive halogen clusters with fluorescence-phosphorescence dual emission in aggregated state and even under ambient conditions.Additionally, multi-tunable PL in response to excitation wavelength, temperature, and pressure are noticed. These results shed new lights on the underlying emission mechanism and would inspirit further exploration of nonconventional luminophores involving halogen moieties. Halogen bond (XB) is a kind of burgeoning electron donation-based noncovalent interaction between halogen atoms (XB donor) and electron-rich Lewis base (XB acceptor), thanks to the electrophilic portion termed as σ-hole on halogen atoms at the elongation direction of the R-X covalent bond (R = C, N, halogen, etc.) (Figure 1A). 4 Relating researches were just launched recently after decades of being shelved, 5 wherein XB has participated in multitudinous fields like crystal engineering, molecular recognition, 9,10 as well as supramolecular chemistry, 11 etc. as an effective tool, owing to its moderate strength (5-180 kJ/mol) and superior directionality. 12 In that scenario, XB holds broad applications serving as a universal noncovalent intermolecular interaction in molecular engineering. In addition to broad applications included above, XB has also become an effective pathway in constructing pure organic luminogens, which are playing increasingly significant roles in cell imaging, 13,14 optoelectronics, bio/chemo probes, 19,20 and medical therapy. 21,22 The advantages of XB introduction to photoluminescence (PL) are multifaceted. Not only can XB act as a mighty candidate to multiply intermolecular interactions, similar to other noncovalent interactions (e.g. hydrogen bond, chalcogen bond, ionic and dipole-dipole interactions), with the aim of promoting and regulating photophysical properties, also the heavy atom effect and ample lone pair (n) electrons of halogen atoms will create an enabling environment for spin-orbit coupling (SOC) and subsequent intersystem crossing (ISC) process, resulting in boosted triplet excitons. 26,27 Therefore, XB is widely utilized in the construction of pure organic luminophores, particularly those with efficient room temperature phosphorescence. Thus far, however, to the best of our knowledge, XB or halogen atom in these systems only plays a supporting role in regulating PL emission, whereas no research concerning PL ability of pure halogen clusters (Figure 1A) was reported, presumably on account of lacking evident luminescent chromophores. Nevertheless, the molecular electrostatic potential (MEP) analysis of hexachloroethane (HCE) dimer (with an obvious Cl•••Cl XB) illustrates a well shared electron cloud and expanded delocalization (Figure 1B). Its trimer and tetramer (Supporting Information, Figure S1) show similar extended electron 4 delocalization, which is beneficial to emission. Moreover, besides XB demonstrated in Figure C (type II), simple intra-and intermolecular halogen-halogen short contacts without the involvement of electrophilic σ-hole (Figure 1C, type I and III) are believed to exist, 2 which further facilitate electron communication between HCE molecules. Inspired by such calculated results and the intrinsic PL from various nonconventional luminophores with electron-rich moieties (bearing n and/or π electrons), and regarding the electron-rich character of halogen atoms, herein, photophysical properties of some aliphatic haloalkanes adopting pure halogen clusters have been investigated for the first time. Considering fluoroalkane is poor XB donor 12 while iodoalkanes are mostly chemically unstable, 33 four commercially available chloro-and bromoalkanes, namely HCE, pentaerythrityl tetrachloride (PERTC), pentaerythrityl tetrabromide (PERTB), and tetrabromomethane (TBM) (Figure 1D) were adopted as representatives. For these haloalkanes, diversified halogen clusters with efficient intra-and intermolecular through-space conjugation (TSC) and prominent heavy atom effect are expected. It is found that individual haloalkane molecules are difficult to excite and thus cannot demonstrate notable PL, while their aggregates are capable of emitting, owing to the formation of halogen clusters with effective TSC (Figure 1D). Their single crystals exhibit clearly visible PL and instrument-detectable room temperature phosphorescence (RTP) under 312 and 365 nm UV lights, accompanying evident excitation wavelength (λex)-dependent PL. Meanwhile, an eye-catching multi-tunability of these haloalkanes responding to λex, temperature, and pressure has also been observed. As Figure 2A demonstrated, weak but visible PL from single crystals of these haloalkanes was noticed. With the variation of λex, HCE crystals, as the representative, show different PL intensities with maxima varying from UV-A to visible region at 365/376/434/443/451/530 nm (Figure 2B, upper), which should be ascribed to the coexistence of diverse emissive species. Further time-resolved measurement at ns scale discloses disparate lifetimes (<τ>f) of 2.1 and 2.7 ns at 429 and 454 nm (Figure 2C), respectively, thus confirming the concurrence of heterogeneous emissive species. Similar PL phenomena were also observed from PERTC, PERTB and TBM crystals (Figure S2, S3). These emission behaviors, analogous to most nonconventional luminophores, can be readily rationalized by the cluster-triggered emission (CTE) mechanism. More surprisingly, although couldn't be captured by naked eyes or digital camera due to the extreme weakness, persistent RTP (p-RTP) from these haloalkanes was still detected by spectrometer with 312 nm UV irradiation, as suggested by their lifetimes (<τ>p) of 58. 6 Furthermore, diversity of halogen clusters can also be testified by extensive absorption containing multiple peaks in UV-vis spectra (Figure 2E), wherein obvious bands attributed to XBs due to charge transfer transitions can be found arising in the UV region 5 . Additionally, such wide absorption region from UV-A to near visible region (especially the one for TBM) gives rational interpretation to the behavior that these samples are ready to be excited by even UV-A irradiation. Apart from the diversity of emissive species at the ground state testified by UV-vis spectra, the heterogeneous popularity at excited state can also be rationalized by their excitation spectra (Figure S5), providing a more persuasive argument from another point of view. In general, the dual emission of fluorescence and p-RTP from HCE, PERTC, PERTB, and TBM at room temperature not only broadens new connotations for XB-relating researches, also provides strong fundamental evidence for proving the rationality and generality of the CTE mechanism. To uncover more details of PL from these haloalkanes which are concealed by the weak luminescence due to the frequent molecular motions even under the restraints of lattice, cryogenic experiment was carried out to further rigidify the conformations. Upon cooling down to 77 K, as shown in Figure 3A and S6, besides the extensively boosted intensity of their prompt and delayed emission, highly prolonged afterglows were also acquired, which is up to 8 s for PERTC and 12 s for PERTB crystals (λex = 312 nm, Figure S6), accompanying greatly extended <τ>ps values (Figure 3D, S7). As demonstrated in Figure 3B taking PERTC as the representative, compared with the main peaks in UV-A band at room temperature, visible portion of emission with peaks at ~450 nm was considerably enhanced and became predominant, thus illustrating the increment of visible emission in PL photographs (Figure 3A). Transformation of main peaks and different increasing extents of UV-A and visible emission can be ascribed to the heterogenous population of emissive halogen clusters which have diverse responses to temperature change. Moreover, from PERTC crystals, λex-dependent emission with slight trend of redshift and then blueshift as λex increases was also observed (Figure 3B), which is confirmed by their corresponding Commission International de l'Éclairage (CIE) coordinates (Figure 3C). Notably, such λex-dependence is much more striking in delayed emission, for which color-tunable phosphorescence is obtained. Variable peaks with broad full-width at half-maximum (FWHM) from blue to yellowish-green were observed with λex changing from 280 to 365 nm, with redshift for ~100 nm (Figure 3E), which are consistent with the CIE coordinates from (0.25, 0.28) to (0.36, 0.43) (Figure 3F). Furthermore, for other three haloalkane crystals, similar main peak transformation and λex-dependent PL were also perceived (Figure S8-S10). Specifically, 8 TBM crystals even exhibit broader region of phosphorescence tunability from dark blue to yellowishgreen (Figure S10C) with CIE coordinates varying from (0.17, 0.14) to (0.33, 0.42) (Figure S10D). Miscellaneous emissive halogen clusters are supposed to be stabilized under cryogenic circumstances, leading to immensely promoted and color-tunable PL, especially afterglows, in response to variable λexs. As diverse emissive species adopt various responses to temperature change and due to the better tunability of phosphorescence, delayed emission of these crystals at different temperatures were further studied. As expected, the phosphorescence of HCE shows apparent tunability in response to the ascending of temperature, which is firstly redshifted from 550 to 600 nm and then blueshifted to 560 nm (Figure 4A) together with the CIE coordinates varying from (0.41, 0.49) to (0.46, 0.46) and then to (0.38, 0.43) (Figure 4B). However, for the other three crystals, merely simple redshift trend with the increment of temperature is observed (Figure S11-S13). Furthermore, an extremely broad FWHM of up to 155 nm implies multiple emissive clusters in HCE crystals, whose characteristic peaks fluctuate with the variation in temperature (Figure S14) respectively. Plus, as Figure S14 exhibits, the distribution of main and shoulder peaks is disparate owing to discrepant responses to temperature from characteristic emission of these diverse emissive species, thus leading to different contributions to the outer luminescence and consequently generating temperature-dependent phosphorescence. It is also noted that as temperature rises, phosphorescence intensity of HCE crystals at 550 nm exhibits an obvious dual-linear phase with the intersection point at 218 K (Figure 4C). To acquire more information of that, differential scanning calorimetry (DSC) scans were performed, in which a subtle transition between endothermal and exothermal portion at ~218 K was noticed (Figure 4D). Thus we can speculate that some sort of crystallographic transition may happen at such temperature and the temperature-dependent emission empowers it a feasible method to detect such transitions. When placed in vacuum, the PL, particularly p-RTP, from PERTB crystals can be improved (Figure S15). Nevertheless, the p-RTP was still too weak to be visualized. When compressed into tablets, PERTB, however, depicts remarkably boosted PL with vivid contrast (Figure 5A). Moreover, after removing irradiation, distinct green p-RTP is effectively visualized in PERTB tablets with prolonged <τ>ps (Figure 5E). Furthermore, upon pressurization (750 MPa), remarkably redshifted emission maximum (from 355 to 438 nm) and narrowed FWHM (from 167 to 124 nm) of PERTB are noticed (Figure 5B). Meanwhile, emerging new peaks (398/438/448 nm) are found (Figure 5B). Similar phenomena are also observed for PERTC (Figure 5C, 5D). For HCE and TBM, redshifted trend is also perceived in their prompt emission spectra (Figure S16). However, what varies is that the p-RTP of them is still too weak to be captured by digital cameras. For all these haloalkanes, although without conspicuous peak shifts in transient emission spectra, tablets are perceived to emit stronger phosphorescence at redder region than single crystals 10 (Figure S17A-D), leading to the redshifted triplet emission (Figure S17E-H). These results should be attributed to the enhancement of short contacts and consequently TSCs among halogen atoms, which also induce simultaneously rigidified conformations. Such deduction was confirmed by the contrast of excitation spectra of their single crystals and tablets (Figure S18), where more extended peaks of tablets imply the presence of expanding conjugations. Thus, in summary, the multi-tunable PL in response to λex, temperature, and pressure confirm the coexistence of miscellaneous emissive species in these haloalkanes from disparate aspects, offering new implications to the origin of PL and promising emerging applications of the compounds with halogen clusters. As demonstrated in Figure 6A, CTE mechanism can rationalize the dual emission and multi-tunable PL of these haloalkanes. Although isolated haloalkanes are not capable of emitting, extended conjugations and descending energy gaps of aggregated counterparts can lead to noticeable PL. Meanwhile, the concurrence of diverse emissive halogen clusters sensitive to temperature and pressure results in multitunable PL. To further illustrate the origin of PL from these haloalkanes, their single crystal structures were determined. However, due to extremely active molecular motions at room temperature, the preliminary single crystal determination was failed with no effective results acquired, so that such characterization can only be proceeded under cryogenic circumstance with the aim of restricting fierce molecular motions (Table S1). As shown in Figure 6B and S19, analyzed at 173 K, abundant inter-and intramolecular Cl…Cl short contacts play essential roles in HCE molecules, which remarkably diminish the molecular motions, thus leading to restrained nonradiative dissipation and rigidified conformation. Also, thanks to electron-rich feature of Cl atoms and efficient interactions, 3D TSC network amongst multiple n electron-rich HCE molecules can be effectively facilitated. Therefore, although individual molecules are not qualified to allow efficient excitation and emission, aggregated molecules with halogen clusters are able to generate extensive electron communication and extended delocalizations, consequently bringing about diverse emissive clusters responsible for multi-tunable PL. Such multitunability thus can be well understood by the CTE mechanism. Furthermore, considering the presence of ample n electrons, together with the heavy atom effect of halogens, the phosphorescence emission can be rationalized by the strikingly promoted SOC and ISC process. Moreover, the interactions in TSC network and conformational rigidity would be strengthened under cryogenic circumstances, which could lead to brighter PL with better multi-tunability. For TBM, single crystal analysis has already been performed at 123 K by the previous publication, 33 wherein effective TSC network based on multiple Br…Br XBs is noticed (Figure S20). On the other hand, although severe disorder was found in PERTC and PERTB during their single crystal analyses even at 173 K (Figure S21) due to their extremely active molecular motions, 39 akin structural features concluded from HCE and TBM can still be deduced from them. To acquire further information into the mechanism, theoretical calculations for monomer, dimer, trimer, and tetramer of HCE and TBM were conducted, which again support these judgements by demonstrating efficient intra-and intermolecular TSC. As depicted in Figure 6C, all HOMOs are mainly distributed on Cl atoms in the molecules without any overlap with one another, whereas conspicuous TSC among LUMOs, from monomer to tetramer, are observed, corresponding to the excited states. By forming multifarious emissive halogen clusters, such extensive TSC effectively narrows the energy gap between the ground and excited states and provides ample triplet energy levels (Figure 6C, S22 and Table S2), thus giving rise to the unique multi-tunable luminescence. Similarly, theoretical results for TBM also reveal effective TSCs and reduced energy gaps owing to the formation of Br clusters (Figure S23, S24 and Table S3). These results are consistent with the CTE mechanism and well rationalized the origin of the multi-tunable PL of the haloalkanes.

chemsum

{"title": "Can Halogen Clusters be Emissive?", "journal": "ChemRxiv"}

oxindole_synthesis_<i>via</i>_polar–radical_crossover_of_ketene-derived_amide_enolates_in_a_formal_[

## Abstract: Herein we introduce a simple, efficient and transition-metal free method for the preparation of valuable and sterically hindered 3,3-disubstituted oxindoles via polar-radical crossover of ketene derived amide enolates. Various easily accessible N-alkyl and N-arylanilines are added to disubstituted ketenes and the resulting amide enolates undergo upon single electron transfer oxidation a homolytic aromatic substitution (HAS) to provide 3,3-disubstituted oxindoles in good to excellent yields. A variety of substituted anilines and a 3-amino pyridine engage in this oxidative formal [3 + 2] cycloaddition and cyclic ketenes provide spirooxindoles. Both substrates and reagents are readily available and tolerance to functional groups is broad. Oxindoles, in particular the 3,3-disubstituted congeners, are highly valuable substructures in medicinal chemistry. The oxindole core can be found in various biologically active compounds, that are for example used in the treatment of cancer or as antibacterial agents. 1 In addition, the oxindole moiety also occurs in several complex natural products. 2 The frst oxindole synthesis was reported by Baeyer and Knop in 1866. 3 That time, isatin was converted by sodium amalgam reduction to the corresponding oxindole. Since then, many methods for the preparation of 3,3-disubstituted oxindoles have been developed that proceed via functionalization of a preexisting oxindole core. 4 In addition, methods for the construction of 3,3-disubstituted oxindoles starting from acyclic precursors have also been introduced. 5,6 Along these lines, transition metal-mediated reactions 5,7 or homolytic aromatic substitutions (HAS) have found to be highly efficient for the construction of the oxindole core. Focusing on the latter approach, the intramolecular HAS proceeds via a-carbonyl radicals derived from radical addition to N-arylacrylamides, 8 reduction of a-haloarylamides 9 or oxidation of the corresponding enolates (Scheme 1a). In 2017, the group of Taylor developed a transition metalfree enolate oxidation-HAS-approach towards oxindoles at low temperature using elemental iodine as the oxidant and malonic acid derived N-aryl amides as substrates which are readily deprotonated. 14 The unique reactivity of ketenes 15 has been explored extensively, 16 especially in [2 + 2]-cycloadditions. 17 Moreover, Scheme 1 Selected strategies for the synthesis of oxindoles. Staudinger, 18 Lippman 19 and Taylor 20 showed that ketenes react with aryl nitrones in a tandem [3 + 2]-cycloaddition- sigmatropic-rearrangement cascade 21 followed by hydrolysis to provide oxindoles (Scheme 1b). The use of chiral nitrones leads to chirality transfer and enantiomerically enriched oxindoles can be obtained via this approach. 21,22 In contrast to the examples discussed in Scheme 1a, two s-bonds are formed and the overall sequence can be regarded as a formal [3 + 2] cycloaddition. Despite good yields and high enantiomeric excess, nitrones have to be used as precursors and an aldehyde is formed as the byproduct diminishing reaction economy of these elegant cascades. To address these drawbacks, we decided to use the nucleophilic addition of deprotonated anilines to ketenes for the generation of the corresponding amide enolates that should then be oxidized in a single electron transfer process to a-amide radicals which can undergo a homolytic aromatic substitution providing direct access to sterically challenging 3,3-disubstituted oxindoles in a straightforward one-pot sequence (Scheme 1c). This polar-radical crossover reaction shows high atom economy and as the reaction with the nitrones can also be regarded as a formal [3 + 2] cycloaddition. We initiated the optimization study with N-methylaniline 1a and ethyl phenyl ketene 2a, which was prepared in an easy and scalable one-pot protocol starting from the corresponding carboxylic acid, as model substrates. Deprotonation of 1a with n-BuLi in THF and subsequent addition to the ketene 2a led to desired Li-enolate which was confrmed by protonation with water and isolation of the amide 4aa (56%). Pleasingly, addition of ferrocenium hexafluorophosphate (FcPF 6 , 2.2 equiv.) at room temperature to the Li-enolate afforded the desired oxindole 3aa in 29% yield (Table 1, entry 1). Switching to CuCl 2 (2.2 equiv.) as the oxidant increased the yield to 34% (Table 1, entry 2) and the use of iodine (2.2 equiv.) improved reaction efficiency (41%, Table 1, entry 3). A further increase in yield (44%) was achieved upon I 2 -oxidation of the corresponding Mg-enolate (Table 1, entry 4). Protonated enolate 4aa (23% yield) and the a,bunsaturated amide 5aa (27%) were observed as the major side products in this transformation. In contrast to the Li-enolate discussed earlier, the intermediate Mg-enolate is formed almost quantitatively, which was confrmed by protonation with water and isolation of compound 4aa (91%). Lowering the reaction concentration to 0.02 M and 0.01 M increased the yield signifcantly to 78% and 90%, respectively (Table 1, entries 5 and 6). Decreasing the amount of oxidant to 1.2 equivalents led to a worse result (Table 1, entry 7). The use of a more electrophilic iodine source such as N-iodosuccinimide (NIS, 2.2 equiv.) also resulted in a lower yield of 39% (Table 1, entry 8). Notably, in this case, the a,b-unsaturated amide 5aa was formed as the major product in 60% yield. When the reaction temperature was lowered to 78 C prior to the addition of iodine (1.2 equiv.), the desired oxindole 3aa was formed in 80% yield (Table 1, entry 9). 14 Light does not appear to play a crucial role in this transformation, as performing the reaction in the dark does not have a signifcant effect on the reaction outcome (Table 1, entry 10). Irradiation with a blue LED (467 nm) actually decreased the yield of targeted 3aa to 74% (Table 1, entry 11). The reaction time of step 3 could be signifcantly reduced to two hours when the reaction was carried out in THF under reflux conditions, and the desired oxindole 3aa was formed in 93% yield (Table 1, entry 12). With the optimized reaction conditions in hand, we investigated the scope by frst varying the R 1 -substituent at the Natom using the ketene 2a as the reaction partner (Scheme 2). In general, increasing the steric bulk at the nitrogen leads to diminished yields of the targeted oxindoles. The lower yields go along with the formation of a larger amount of the corresponding a,b-unsaturated amide side product 5. Thus, as compared to the parent N-methyl derivative, all other N-alkyl derivatives were formed in lower yields (49%, 3ab; 35%, 3ac; 49%, 3ad). The N-benzyl protected oxindole 3af and the Nphenyl oxindole 3ae were isolated in 54% and 56% yield, respectively. Next, a diastereoselective oxindole synthesis was attempted using chiral anilines 1g and 1h. Surprisingly, despite the bulkiness of these nucleophiles containing styryl-type Nsubstituents, good yields were obtained for the oxindoles 3ag and 3ah (73-79%). Unfortunately, diastereocontrol was low in both cases (1.9 : 1 d.r. and 1.5 : 1 d.r.). Of note, addition of Mg-1g and Mg-1h to ketene 2a was rather slow under the standard reaction condition and a signifcant amount of unreacted aniline was recovered. That problem could be solved by prolonging the reaction time of both step 1 (deprotonation) and also step 2 (Mg-enolate formation). Next, the substrate scope was investigated by using different anilines in combination with the ketene 2a (Scheme 3). N- a Reactions (0.20 mmol) were conducted under argon atmosphere. b 1 H NMR yield using 1,3,5-trimethoxybenzene as internal standard. c Isolated yield. d Step 1 and 2 were conducted at 0 C. e N-Iodosuccinimide. f Iodine addition at 78 C, then slowly allowed to warm to room temperature. 14 g In the dark. h Irradiation with blue LED (40 W, 467 nm, rt, 8 h). i Refluxing THF for step 3, reaction completed within 2 h. Methyl-p-toluidine 1i and N-methyl-p-haloanilines 1j-m could be successfully transformed to the corresponding oxindoles 3al-am in moderate to good yields (53-87%). Electronwithdrawing and also electron-donating substituents are tolerated and oxindoles derived from p-cyano-(3an, 92%), p-acetyl-(3ao, 30%), p-methoxycarbonyl-(3ap, 82%) and p-methoxy-(3aq, 71%) anilines were isolated in moderate to excellent yields documenting a high functional group tolerance of this reaction. The meta-methyl aniline afforded oxindole 3ar in 76% yield as a 1.8 : 1 mixture of the two regioisomers (only the major isomer drawn). For the pyridyl derivative 3as, a lower yield was obtained (39%), but reaction occurred with complete regiocontrol. Of note, ortho-methyl N-methylaniline provided the corresponding oxindole only in trace amounts (not shown). The ketene component was also varied using N-methylaniline 1a as the reaction partner. The transformation of methyl phenyl ketene 2b provided the oxindole 3ba in 58% yield. p-Bromophenyl ethyl ketene 2c and p-iodophenyl ethyl ketene 2d afforded the oxindoles 3ca and 3da in good yields (70% and 76%). For the ibuprofene-derived ketene 2e a lower yield was obtained (3ea, 40%) and the bulkier phenyl isopropyl congener 3fa was isolated in 27% yield as an inseparable mixture with the protonated enolate 4fa (56% combined yield). In the latter case, increasing the reaction time did neither lead to a higher yield of 3fa nor to a suppression of the formation of 4fa. The lower yield is likely caused by steric effects. Surprisingly, diphenyl ketene 2g delivered the targeted oxindole 3ga in acceptable 55% yield despite the steric demand of the two phenyl groups and the high stability of the corresponding a-amide radical. Spirocyclic oxindoles are of great interest due to their high pharmaceutical potential. 27 We were pleased to fnd that our method also works for the preparation of such spiro compounds as documented by the successful synthesis of 3ha (26%). Mechanistically, we propose initial formation of the enolate A by nucleophilic attack of the deprotonated aniline to the ketene 2, which is then oxidized by elemental iodine to the aamide radical B (pathway b). The radical nature of the transformation is supported by the fact that electronic effects on the arene show no influence on the efficiency of the cyclization, as would be shown by a conceivable polar aromatic substitution. Radical B readily cyclizes onto the aniline ring to generate the cyclohexadienyl radical D which is oxidatively rearomatized via cationic intermediate E to fnally give the oxindole 3 (Scheme 4). Alternatively, enolate A can be iodinated with I 2 to give the unstable iodide C which then undergoes C-I bond homolysis to generate the radical B (pathway a). Indeed, Taylor and coworkers 14 observed under similar reaction conditions the decay of a-iodinated compounds of type C via C-I homolysis 14,28 to give radicals of type B. Usually, we observed a,b-unsaturated amides analogous to 5aa as by-products. However, the corresponding protonated enolates were detected only in tiny amounts in most of these cases. This strongly suggests that those amides are not formed via disproportionation of radical B. HI-elimination seems more likely, pointing towards the presence of the iodinated species C and thus the contribution of pathway b to product formation. In addition, dimerization of radical B was also not observed. To further support pathway b, isolation of the iodinated intermediate C was attempted at low temperature. Upon addition of iodine (1.2 equiv.) to the preformed Mg-enolate A derived from aniline 1a and ketene 2a at 78 C, 14 TLC analysis showed a clean conversion to a single new compound, which was analyzed by rapid ESI-MS analysis and provided evidence for the formation of the iodinated intermediate C (Scheme 5). However, isolation of this highly unstable compound was not possible due to rapid HI-elimination to the amide 5aa. Note that oxindole formation worked well upon I 2 -addition at 78 C and subsequent warming to room temperature (see Table 1, entry 9). In contrast to the unstable tertiary iodide C, the secondary iodide 7 proved to be stable at room temperature as well as in refluxing THF in the absence of light. Due to the stronger C-I bond of 7 as compared to its tertiary congeners, thermal activation of this iodide is not possible. Upon irradiation of 7 with a blue LED (467 nm), the compound decomposed to release iodine without producing the expected oxindole product (Scheme 5). This is consistent with the observation from our optimization studies that irradiation with blue light does not contribute to the yield of oxindole 3aa (Table 1, entry 11). Furthermore, the use of more electrophilic NIS instead of iodine should favor the formation of intermediates of type C over direct SET. In this case, the elimination product 5aa was obtained as the main product (60% yield) and the oxindole 3aa was formed in only 39% yield (Table 1, entry 8). In light of these results, we suggest that both the direct SET of Mg-enolate A to iodine (pathway a) and the C-I bond homolysis of intermediate C (pathway b) might operate in these transformations. ## Conclusions In conclusion, we demonstrated a simple, efficient and transition metal-free procedure for the preparation of sterically challenging valuable 3,3-disubstituted oxindoles via a polarradical crossover of ketene-derived aniline enolates followed by homolytic aromatic substitution at room temperature starting from mostly commercially available anilines and ketenes, which are easily prepared in one-pot reactions from a plethora of commercially available compounds. 26 different (hetero) aromatic 3,3-oxindoles as well as spirooxindoles could be prepared following this method. Functional group tolerance is broad and the herein reported cascades nicely document the potential of polar-radical cross-over chemistry by benefting from both ionic as well as from radical bond forming reactions.

chemsum

{"title": "Oxindole synthesis <i>via</i> polar\u2013radical crossover of ketene-derived amide enolates in a formal [3 + 2] cycloaddition", "journal": "Royal Society of Chemistry (RSC)"}