From single-site tantalum complexes to nanoparticles of Ta_xN_y and TaO_xN_y supported on silica: elucidation of synthesis chemistry by dynamic nuclear polarization surface enhanced NMR spectroscopy and X-ray absorption spectroscopy

Abstract

Air-stable catalysts consisting of tantalum nitride nanoparticles represented as a mixture of Ta_xN_y and TaO_xN_y with diameters in the range of 0.5 to 3 nm supported on highly dehydroxylated silica were synthesized from TaMe₅ (Me = methyl) and dimeric Ta₂(OMe)₁₀ with guidance by the principles of surface organometallic chemistry (SOMC). Characterization of the supported precursors and the supported nanoparticles formed from them was carried out by IR, NMR, UV-Vis, extended X-ray absorption fine structure, and X-ray photoelectron spectroscopies complemented with XRD and high-resolution TEM, with dynamic nuclear polarization surface enhanced NMR spectroscopy being especially helpful by providing enhanced intensities of the signals of ¹H, ¹³C, ²⁹Si, and ¹⁵N at their natural abundances. The characterization data provide details of the synthesis chemistry, including evidence of (a) O₂ insertion into Ta–CH₃ species on the support and (b) a binuclear to mononuclear transformation of species formed from Ta₂(OMe)₁₀ on the support. A catalytic test reaction, cyclooctene epoxidation, was used to probe the supported nanoparticles, with 30% H₂O₂ serving as the oxidant. The catalysts gave selectivities up to 98% for the epoxide at conversions as high as 99% with a 3.4 wt% loading of Ta present as Ta_xN_y/TaO_xN_y.

---

Introduction

Dispersed species ranging from single-metal-atom complexes to clusters and nanoparticles (NPs) of metals and metal oxides are materials that find many applications, especially in catalysis.^1–3 Numerous methods have been used to synthesize such NPs, often involving colloidal, solution-based, or chemical or physical vapour-deposition techniques.^4,5 Solution-based methods often suffer from the need for capping agents or surfactants to stabilize the clusters or NPs, and deposition methods lack fine control and typically lead to highly nonhomogeneous materials.^6,7 In typical catalytic applications, NPs are dispersed on high-area porous supports such as silica, alumina, and zeolites.^8,9 The supported NPs are often unstable, undergoing agglomeration or sintering during operation.¹⁰

A foundation for the synthesis of NPs from molecular building blocks on supports is provided by surface organometallic chemistry. Thus, the supported species can be made without the need for templates, capping agents, or surfactants, and the syntheses thereby offer good prospects for control of the compositions and sizes of the dispersed species. Essential reactant species in such syntheses are the support surface functional groups, exemplified by the –OH groups on silica. Control of the sizes of NPs synthesized from single-metal-atom precursors on supports may be facilitated by the initial bonding of the precursor species to the support to minimize agglomeration. Thus, for instance, surface organometallic chemistry has guided the synthesis of monometallic (Pt) and bimetallic (Pt–Sn) clusters on silica, giving catalysts with high activities and selectivities for catalytic hydrogenolysis, isomerization, and dehydrogenation reactions.^11–13 However, only little work has been done to extend the concepts to non-metal NPs including semiconductors, and herein we demonstrate their use to guide the synthesis of silica-supported metal nitride semiconductor NPs.¹⁴ Such materials offer prospects as new catalysts, and they could also offer unique optoelectronic properties of potential value in displays, quantum computing, or photovoltaic devices.^8,15–22

Unsupported nanocrystalline samples consisting of TaON and Ta₃N₅ were used by Gao et al.¹⁷ to catalyze cyclooctene epoxidation, an industrially important reaction. Epoxidation reactions are sensitive to catalyst surface acidity or basicity, and by incorporating nitrogen in place of oxygen in the framework of Ta₂O₅, these authors tuned the catalyst basicity and improved its properties.

Syntheses of metal nitrides often lead to mixed phases or non-stoichiometric compositions. Strategies to control metal nitride synthesis have been developed, some involving the application of various nitriding agents such as gaseous or liquid NH₃, urea, cyanamide, etc.^20,22 A traditional method involving gaseous NH₃ is convenient for the synthesis of impurity-free metal nitrides, because the high treatment temperatures lead to the release of N₂ and H₂, which react with oxides and form water,^21,23–25 which can be removed as a gas. Hence, we reasoned that it would be of interest to attempt analogous syntheses to make dispersed metal nitrides on a support, being guided by surface organometallic chemistry and starting with molecular metal complexes anchored to the support. We chose silica as the support because its surface chemistry is well understood and the principles of surface of organometallic chemistry are a powerful guide to manipulating syntheses on it. We chose tantalum as a precursor metal to allow a comparison of our results with those of Gao et al. for the performance of tantalum nitride catalysts for cyclooctene epoxidation.

Understanding the chemistry of species dispersed on solid surfaces emerges best when a battery of complementary characterization techniques is applied to allow elucidation of elementary reactions. Dynamic nuclear polarization surface enhanced NMR spectroscopy (DNP SENS) is especially powerful for elucidation of dispersed surface species and their transformations. The technique requires a simple impregnation of the sample with a stable free radical such as bi-radical nitroxide (TEKPol).²⁶ The effective polarization transfer of free electrons of the radical to the surrounding nuclei (usually ¹H) induced by microwave radiation and subsequent cross polarization to the nuclei of interest (¹³C, ¹⁵N, ²⁹Si, ¹⁷O, etc.) are essential for the success of this recently developed technique. It allows NMR measurements with signal enhancements, leading to well-resolved spectra of any nuclei of interest.^26–30

Taking advantage of this technique, we now report two strategies for the synthesis of air-stable, supported NPs of Ta_xN_y and TaO_xN_y with sizes in the range of 0.5–3 nm, from well-defined molecular precursors without using any capping agents or surfactants. Such syntheses of finely sized NPs of tantalum nitride have not yet been known. The precursors, TaMe₅, 1, and the dimeric Ta(OMe)₅, 2, were first grafted onto silica (Aerosil SiO₂-700-dehydroxylated at 973 K (700 °C)). Subsequent treatments led to the formation of Ta_xN_y/TaO_xN_y NPs, which were tested as catalysts for epoxidation of cyclooctene. Conventional characterization techniques were sufficient to demonstrate the formation of the NPs, but understanding of the elementary steps of the surface chemistry required the application of DNP SENS.^26–30 Further details of the surface tantalum species, including interatomic distances and coordination numbers, were obtained from extended X-ray absorption fine structure (EXAFS) spectra.^31–34

Results and discussion

Summary of synthesis chemistry of silica-supported tantalum nitride nanoparticles from silica-supported tantalum complexes

To begin, we provide a summary roadmap, Scheme 1, illustrating the two synthetic strategies used to prepare the NPs referred to as Ta_xN_y/TaO_xN_y. These were prepared from TaMe₅, 1, and Ta(OMe)₅, 2 (which exists as a dimer). The evidence for this chemistry is presented in the paragraphs that follow. The spectroscopic data show that 1 and 2, upon grafting onto SiO₂-700, resulted in the formation of the supported species 1A and 2A, respectively, as shown in Scheme 1. 1A is formed as predominantly a monopodal species^35–37 that underwent treatment with O₂ to form the corresponding tantalum alkoxide 1B (Scheme 1). A tentative pathway corresponding to the O₂ insertion is given in Scheme S1 (in the ESI†).^38,391B and 2A are mixtures of monopodal and bipodal species, predominantly the former. A schematic representation of the associated surface reactions that occur during the grafting of 2 is provided in Scheme S2 (ESI†). The precursor 2 exists as a dimer in solution as well as in the solid-state.^40,41 A binuclear-to-mononuclear transition was observed on the silica surface during the grafting, as shown in Scheme S2.† Nitriding of 1B and 2A at 1173 K for 15 h resulted in the formation of the corresponding tantalum nitrides/oxynitrides that are depicted as 1C and 2B, respectively.

Scheme 1 Two different strategies used for the synthesis of Ta_xN_y/TaO_xN_y NPs (1C and 2B were handled in open air for all the characterizations except for the elemental analysis).

Characterization of silica-supported tantalum complexes: evidence from IR spectroscopy

IR spectra of SiO₂-700, 1A, 1B, 2A, and 2 are given in Fig. 1(A). The spectrum of the support (SiO₂-700) alone includes two prominent, well-known vibrations representing the O–H stretching of isolated silanols, at 3746 cm⁻¹, and lattice vibrations corresponding to Si–O–Si bridges, at 1985, 1864, and 1635 cm⁻¹. The IR spectrum of 1A includes bands corresponding to the stretching vibrations of C–H bonds (ν_–CH) of grafted Ta(CH₃)₅ at 2985 and 2932 cm⁻¹ (ν_asy) and at 2900 and 2855 cm⁻¹ (ν_sy). Bending vibrations of C–H groups (δ_–CH) were observed at approximately 1397 cm⁻¹.

Fig. 1 (A) IR spectra of SiO₂-700, 1A, 1B, 2A, and 2; (B) of 1C and 2B.

Treatment with 1 bar of O₂ resulted in the insertion of oxygen atoms into Ta–Me bonds (Scheme S1, ESI†). Corresponding changes in the C–H stretching vibrations are evident in the spectrum of 1B. Bands corresponding to the C–H stretching vibrations of Ta–O–CH₃ at 2960 and 2936 cm⁻¹ (ν_asy) and 2854 and 2840 cm⁻¹ (ν_sy) are evident in the spectrum; ν_–CH stretching vibrations corresponding to [(≡Si–O–CH₃)] (this notation represents methoxide groups bonded to the silica surface) were observed at 2992 (ν_asy) and 2856 cm⁻¹ (ν_sy).^42,43 Corresponding bending vibrations of C–H in Ta–O–CH₃ and ≡Si–O–CH₃ were observed at 1454 and 1436 cm⁻¹, respectively.

The spectrum of 2A includes vibrational features similar to those of 1B, which are assigned to the C–H stretching of Ta–O–CH₃ and ≡Si–O–CH₃ species. This comparison suggests the presence of similar types of surface species in the samples made from the two precursors. A magnified view of the spectra of 1B, 2A, and 2, Fig. S1 (ESI†), is in line with the suggestion, with the spectrum of 2 supporting the assignments as it shows the stretching and bending vibrations of the precursor complex Ta(OMe)₅ at approximately 2919, 2890, and 2823 cm⁻¹. However, grafting resulted in ∼10–30 cm⁻¹ shifts in the vibrational frequencies of Ta–O–CH₃ for the complexes 1B and 2A.

Because the grafting of 2 was accompanied by the release of methanol, we suggest that it might have opened up a nearby Si–O–Si bridge, resulting in the formation of ≡Si–OH and ≡Si–O–CH₃ moieties. Some of the released methanol might also have become adsorbed on the silica. Such adsorbed methanol would likely have been hydrogen bonded with remaining silanols—a possible explanation for the appearance of the broad band in the region of approximately 3400 cm⁻¹ in the spectrum of 2A.

Silica-supported tantalum complexes: further evidence from elemental analyses

Elemental analysis data presented in Table S2 (ESI†) (samples for elemental analysis were handled under controlled atmospheres) show that the C/Ta value obtained for 1B is 3.6 ± 0.3, which suggests that the major species formed was monopodal (theoretical C/Ta = 4). 2A has a C/Ta value of 4.6 ± 0.3, which accounts for the major fraction of monopodal species, with Ta–(O–CH₃)₄ formed, along with neighbouring ≡Si–OCH₃ groups formed as a result of the reaction of methanol with the strained siloxane bridges (theoretical C/Ta = 5). The elemental analysis data match the expected values for the structures of 1B and 2A given in Scheme 1.

Characterization of supported tantalum nitride NPs

X-ray diffraction patterns of supported tantalum nitride NPs. The XRD patterns of 1C and 2B include a broad peak corresponding to amorphous silica (Fig. S2, (ESI†)). No peaks representing Ta_xN_y/TaO_xN_y are evident, probably a consequence of the too low loading, amorphous nature, and/or the very small size of the particles.⁴⁴

Transmission electron microscopic analyses of supported tantalum nitride NPs. In order to understand the size, morphology, and crystalline nature of supported tantalum nitride (1C and 2B), TEM analyses were performed. The images provide evidence that very fine NPs were formed on the silica surface as a result of the nitridation treatment (Fig. 2(A)–(F)). Fig. 2(B) of 1C shows NPs with diameters in the range between 0.5 and 3 nm. The relatively large particle in the image of Fig. 2(B) is inferred to have formed by coalescence of smaller NPs under the influence of the high-energy electron beam. Fig. 2(B) and (C) show that the NPs are well dispersed on the support surface. Fig. 2(D) is a high-magnification image of 2B showing the NPs with diameters up to 3.5 nm (Fig. S3, (ESI†)). The slightly greater sizes of the NPs in sample 2B as compared with 1C suggest aggregation associated with the greater density of tantalum species in the former sample. Particle size distributions and additional images, determined by HRTEM and STEM, of 1C and 2B, are given in Fig. S3–S5 (ESI†). The average particle sizes of 1C and 2B samples are respectively 1.6 and 1.9 nm with a standard deviation of ±0.2 nm. These confirm the rather narrow size distribution of the NPs mentioned above, which we suggest may be associated with the preparation strategy. Thus, TEM analysis complements the observations from XRD analysis. Note that some micrographs acquired by HRTEM as shown for 1C in Fig. 2(A) were sufficient to resolve atomic planes with a d-spacing of 0.23 nm, which would possibly match with the {024} plane family of crystalline Ta₃N₅ (Cmcm (63)).²⁴

Fig. 2 TEM images of 1C and 2B. (A) Lattice-resolved image of one of the particles of 1C; (B) distribution of Ta_xN_y NPs on silica in 1C; (C) dark field STEM image showing the dispersion of the NPs in 1C; (D) high-magnification dark field STEM image of 2B showing the NPs; (E) NPs in 2B; and (F) low magnification dark-field STEM image of 2B.

IR spectroscopy and elemental analyses of supported tantalum nitride NPs. IR spectra of 1C and 2B (Fig. 1(B)) include a broad band centred at about 3300 cm⁻¹ corresponding to adsorbed water and a strong band at 1630 cm⁻¹ corresponding to the bending vibrations of adsorbed water. A very weak band assigned to Si–NH₂ is also present, at 1550 cm⁻¹, in the spectrum of 2B, and this decreased in intensity upon prolonged exposure of the sample to air.⁴⁵ Traces of isolated silanols were also present in both 1C and 2B as indicated by the weak band at 3746 cm⁻¹. Elemental analysis data presented in Table S2 (ESI†) show that 1C and 2B contained 4.2 ± 0.2 and 4.8 ± 0.2 wt% of Ta, respectively.

Textural properties of support and supported tantalum nitride NPs. BET surface area measurements were carried out to provide evidence of the textural properties of 1C, 2B, and SiO₂-700 (Table S3 (ESI†)). Dehydroxylation, grafting of the metal complexes, and subsequent nitriding at 1173 K to form NPs resulted in a reduction of the surface area by 15–20% (S_BET: 200 m² g⁻¹). Nitrogen adsorption–desorption isotherms of the samples 1C, 2B, and SiO₂-700 are presented in Fig. S6 (ESI†).

Evidence of tantalum oxidation states of supported NPs. X-ray photoelectron spectroscopic (XPS) analyses were performed to probe the nature of the surface composition and the oxidation state of Ta. XPS data corresponding to Ta 4f and N 1s of 1C and 2B are given in Fig. 3(A)–(D). The Ta 4f spectrum of 1C shown in Fig. 3(A) is composed of two sets of Ta 4f doublets. As for the doublet in the low energy region, the binding energy values from the Ta 4f spectrum (Ta 4f7/2 = 24.9 eV and Ta 4f5/2 = 26.8 eV) are in good agreement with the Ta 4f of the Ta₃N₅ phase (∼83%).⁴⁶ Another set of doublets observed in the high-energy region (Ta 4f7/2 = 27.7 eV and Ta 4f5/2 = 29.6 eV) most probably should be attributed to the TaO_xN_y phase (∼17%). This phase might have arisen from oxidation of the Ta₃N₅ that formed. Hence, it is concluded that 1C (Fig. 3(A)) is predominantly composed of tantalum nitride (Ta₃N₅) NPs.⁴⁷ In the case of 2B, the Ta 4f spectrum consists of three separate doublets (Fig. 3(B)) consisting of a weak low energy doublet, similar to the one observed for the TaN phase (Ta 4f7/2 = 23.3 eV and Ta 4f5/2 = 25.2 eV).⁴⁸ Another doublet at Ta 4f7/2 = 24.5 eV and Ta 4f5/2 = 26.35 eV and a high-energy doublet at Ta 4f7/2 = 26.5 eV and Ta 4f5/2 = 28.3 eV match well with the Ta₃N₅- and TaON-like phases, respectively.⁴⁹ The spin orbit splitting constants in all the cases are ∼1.9 eV, consistent with the literature.⁵⁰

Fig. 3 X-ray photoelectron spectra corresponding to Ta 4f of (A) 1C (B) 2B and N 1s of (C) 1C and (D) 2B.

XPS data corresponding to the N 1s of 1C are given in Fig. 3(C). N 1s spectrum shows contributions from three different types of nitrogen environments with their corresponding binding energies at 396.3, 398.5, and 400 eV, similar to the reported values for TaN, Ta_xN_y/Ta₃N₅, and TaO_xN_y phases, respectively. Slightly different contributions at 396, 397.7, and 398.5 eV were obtained for 2B as shown in Fig. 3(D).^51,52 For 1C the main core-level is located at 398.5 eV (∼83%) whereas for 2B the nitrogen core-levels are more evenly distributed among the three binding energies (396, 397.7, and 398.5 eV). The data suggest that the preparation method has a strong impact on the homogeneity of the final synthesized tantalum nitride NPs: starting from the Ta(Me)₅ precursor allows one to form the most homogenous tantalum nitride NPs. More spectra (C 1s, O 1s, and Si 2p) are provided for samples 1C and 2B in Fig. S7–S9, (ESI†). On the basis of the XPS analyses, we designate samples 1C and 2B as a mixture of Ta_xN_y and TaO_xN_y phases.

UV-visible spectroscopy of supported tantalum nitride NPs. Fig. 4(A) shows the UV-visible diffuse reflectance spectra of the supported tantalum nitrides 1C and 2B. Spectrum (a) shows a red shift of the onset of the absorption at about 636 nm, and the corresponding Tauc plot (Fig. 4(B)b) shows a direct band gap of 2.15 eV. On the other hand, the onset of absorption of the tantalum nitride 2B is at about 596 nm (Fig. 4(A)b) with a direct band gap of 2.29 eV (Fig. 4(B)b).⁵³ Bulk Ta₃N₅ is characterized by a band gap of 2.1 eV, whereas our as-synthesized Ta_xN_y/TaO_xN_y is characterized by band gaps of 2.15 and 2.29 eV for 1C and 2B, respectively, indicating a blue shift probably due to the presence of TaO_xN_y.⁵⁴

Fig. 4 (A) UV-visible diffuse reflectance spectra of (a) 1C and (b) 2B. (B) Tauc plot showing direct band gap of (a) 1C and (b) 2B.

Details of the molecular surface organotantalum chemistry

A more detailed characterization of the surface species depicted in Scheme 1 required spectroscopic evidence beyond that accessible from the results presented heretofore. NMR and EXAFS spectroscopies provide the further insights summarized in the following paragraphs.

Evidence from MAS DNP SENS. Although alkoxides of transition metals have been used extensively as precursors for the preparation of a wide variety of materials, the available crystallographic information characterizing such alkoxides is limited. Correspondingly, there is little such information about precursor 2. A comparison of the liquid and solid-state NMR spectra of 2 is also missing from the literature. Consequently, we recorded variable-temperature ¹H and ¹³C NMR spectra of Ta(OMe)₅, 2, in deuterated dichloromethane, with the results summarized in Fig. S10 and S11 (ESI†), respectively.

The ¹H NMR spectrum of dimeric tantalum pentamethoxide 2 in CD₂Cl₂ solution at 300 K includes two major resonances, at 4.31 ppm (corresponding to terminal OCH₃(a)) and at 4.05 ppm (corresponding to bridging O–CH₃(c)), and these indicate a rapid intermolecular exchange of terminal and bridging methoxy groups. The low-temperature ¹H NMR spectra include three signals with intensity ratios of 2 : 2 : 1, consistent with a dimeric bi-octahedral structure with two bridging methoxy groups. Significantly, the temperature dependence of the spectra indicates that the exchange of terminal methoxide groups occurs faster than the exchange of bridging and terminal methoxide groups.

When the temperature was decreased, the signal at approximately 4.31 ppm split into two distinct singlets, at δ 4.35 and 4.22 ppm, showing that two types of terminal OCH₃ groups exist, which are of type (a) and (b), respectively. A weak signal at about 3.33 ppm (CH₃–OH), corresponding to traces of methanol in the sample, is also evident in the spectra. We surmise that some of the methanol used in the preparation of 2 remained in the samples. A very weak signal at about 5.3 ppm corresponding to trace amounts of CH₂Cl₂ (from solvent CD₂Cl₂) is also evident.

Fig. S11 (ESI†) shows the corresponding ¹³C NMR spectra of 2 at 300 and 193 K. The intensity ratio of terminal carbons to bridging carbons was found to be 2 : 2 : 1, as expected for a dimer. The ¹H–¹³C HSQC spectra of Ta(OMe)₅ at 193 K (Fig. S12 (ESI†)) show that the signals corresponding to (a) and (b), representing the terminal methoxides, correlate with the carbon signals at 61 and 58 ppm, whereas the bridging methoxide correlates with the carbon signal at 60 ppm. Furthermore, as expected, the ¹H signal at 3.33 ppm, attributed to methanol, correlates with the carbon at 50 ppm. The weak ¹H signal at 5.3 ppm correlates with the ¹³C signal of CH₂Cl₂-D₂ (solvent), which appears at 54 ppm.

Fig. 5(A) and (B) is a representation of the solid-state ¹³C CP MAS NMR spectrum of 2 at room temperature and its corresponding two-dimensional ¹H–¹³C correlation (HETCOR) spectrum. The ¹³C signals include resonances at 60.4, 58.2, 57.1, and 55.1 ppm, corresponding to a-terminal (a), c-bridging (c), b-terminal (b) methoxides and traces of adsorbed methanol, respectively. Fig. 5(B) confirms these assignments and also indicates the direct correlation with the corresponding protons of various types of methoxides and methanol.

Fig. 5 (A) ¹³C CP MAS NMR spectrum of 2 (acquired at 400 MHz, with a 10 kHz MAS frequency, 5000 scans, a 4 s repetition delay, and a 2 ms contact time at room temperature). (B) 2D ¹H–¹³C CP MAS HETCOR spectrum of 2 (acquired at 9.4 T with 10 kHz MAS frequency, 1000 scans per t₁ increment, a 4 s repetition delay, 64 individual t₁ increments and a 0.2 ms contact time).

The ¹H MAS and ¹³C CP MAS NMR spectra of 1A show major peaks at 0.8 and 73 ppm, respectively, corresponding to methyl groups of Ta–CH₃ (Fig. S13(a) and (b) (ESI†)). The ¹³C CP MAS NMR spectra of 1B and 2A (Fig. S14(a) and (b) (ESI†)) show two prominent signals at about 60 and 49 ppm corresponding to Ta–OCH₃ and ≡Si–OCH₃, respectively (we emphasize that the acquisition time for each of the corresponding spectra was 4–5 days with poor signal-to-noise ratios).

It is thus clear that conventional NMR spectroscopy is limited for characterization of the surface chemistry involved in the preparation of Ta_xN_y NPs. Hence, we turned to DNP surface-enhanced NMR spectroscopy (SENS) for improved sensitivity. SENS has been used by means of dynamic nuclear polarization (DNP), a technique that allows short acquisition times and is of particular value for samples incorporating species at low concentrations on surfaces.²⁸ DNP SENS measurements were acquired at about 100 K, with TEKPol (stable radical) as the polarizing agent and 1,1,2,2-tetrachloroethane (TCE) and 1,2-dichlorobenzene (DCB) as the solvents.⁵⁵

The ¹H MAS DNP SENS spectrum of 1B in TCE (Fig. S15(a) (ESI†)) gave a very good proton enhancement (ε_H) of 112. The ¹³C CP MAS DNP SENS spectrum of 1B shows carbon signals corresponding to the solvent TCE, which overlaps the expected ¹³C chemical shifts of Ta–OCH₃ and ≡Si–OCH₃ (Fig. S15(b) (ESI†)). Hence, another solvent, DCB, was used to disperse the TEKPol, showing a proton enhancement of ε_H = 59 (Fig. S15(c) (ESI†)). The ¹³C CP MAS DNP SENS spectrum of 1B in DCB (Fig. 6(A)) shows the carbon signals corresponding to Ta–OCH₃ and ≡Si–OCH₃ at 60 and 49 ppm, respectively, without any interference from the solvent.⁵⁶ The spectra were recorded for the sample including ¹³C in natural abundance. We were also able to acquire a DNP SENS 2D ¹H–¹³C HETCOR NMR spectrum of 1B (Fig. 6(B)), which shows correlations of the ¹³C signal at 60 ppm with the ¹H signal at 4.1 ppm of Ta–OCH₃ and the ¹³C signal at 49 ppm with the ¹H signal at 3.4 ppm of ≡Si–O–CH₃.

Fig. 6 Schematic representation of 1B followed by (A) ¹³C CP MAS DNP SENS spectra (100 K, 400 MHz/263 GHz gyrotron) of 1B impregnated with 16 mM of TEKPol in DCB solution. The spectra were recorded using 128 scans, recycle delay of 3 s, contact time of 3 ms, and MAS frequency at 8 kHz. (* indicates the sidebands) (B). Two-dimensional ¹H–¹³C heteronuclear correlation (HETCOR) DNP SENS spectrum of 1B acquired with 2048 scans per t₁ increment, 96 individual increments, and 0.2 ms contact time. During t₁, e-DUMBO-1 homonuclear ¹H decoupling was applied and proton chemical shifts were corrected by applying a scaling factor of 0.57 (Table S1 (ESI†)).

The MAS DNP SENS spectrum of 2A in TCE shows a proton and carbon signal enhancement ε_H = 147 and ε_C = 62, respectively (Fig. S16(a) & (b) (ESI†)). Here, for the suppression of TCE solvent signals, we used a CP MAS NMR with echo. The corresponding ¹³C CP MAS echo DNP spectrum (Fig. 7(A)) clearly shows two peaks for the sample, at 60 and 49 ppm, corresponding to Ta–O–CH₃ and ≡Si–O–CH₃, respectively. As in the case of 1B, the 2D ¹H–¹³C HETCOR with echo spectrum (Fig. 7(B)) of 2A shows that the ¹³C of TaOCH₃ and ≡Si–O–CH₃ correlated with the corresponding ¹H signals. A weak tail at 60 ppm correlating with the ¹H at about 5.5 ppm might be evidence of a contribution from the solvent TCE.

Fig. 7 Schematic representation of 2A followed by (A) ¹³C CP MAS with echo DNP SENS spectra (100 K, 400 MHz/263 GHz gyrotron) of 2A impregnated with 16 mM TEKPOL in TCE solution. The spectra were recorded using 128 scans under microwave irradiation, recycle delay of 3 s, contact time of 3 ms, and MAS frequency at 8 kHz (B) 2D ¹H–¹³C HETCOR DNP SENS spectrum with an echo of 2A acquired with 2048 scans per t₁ increment, 96 individual increments and 0.2 ms contact time. During t₁, e-DUMBO-1 homonuclear ¹H decoupling was applied, and proton chemical shifts were corrected by applying a scaling factor of 0.57 (Table S1 (ESI†)).

The ²⁹Si MAS DNP SENS spectrum of 2A in TCE was acquired with only 1024 scans and showed the presence of Q and D type silicon at −100 and −16 ppm, respectively, attributed to the support Aerosil silica. The nature of the different Q-type Si and the D-type Si signals is explained in Fig. 8(A). The D-type signals are attributed mainly to Si atoms in (–O–Si(R₂)–O–)_x structures (This might be due to the contamination from grease used in the storage tubes). The two-dimensional ¹H–²⁹Si HETCOR DNP-SENS spectrum of 2A, Fig. 8(B), gives evidence of various Q types correlated with various surroundings of protons, namely, (a) that resulting from the interaction of Q3 with a corresponding OH proton (1.8 ppm); (b) that resulting from the direct interaction of Si of ≡Si–OCH₃ with the corresponding OCH₃ proton (3.2 ppm); and (c) that resulting from the Si of Si–O–Ta–OCH₃ with the respective proton (4.4 ppm).

Fig. 8 (A) ²⁹Si CP MAS DNP SENS spectra (100 K, 400 MHz/263 GHz gyrotron) of 2A impregnated with 16 mM TEKPOL in TCE solution. The spectra were recorded with 1024 scans, recycle delay of 3 s, contact time of 5 ms, and MAS frequency at 8 kHz. Separate Q and D type Si species are represented for clarity. (B). 2D ¹H–²⁹Si HETCOR DNP SENS spectrum of 2A acquired with 2048 scans per t₁ increment, 96 individual increments, and 0.2 ms contact time. During t₁, e-DUMBO-1 homonuclear ¹H decoupling was applied, and proton chemical shifts were corrected by applying a scaling factor of 0.57 (Table S1 (ESI†)).

Ta_xN_y/TaO_xN_y NPs obtained from both precursors 1B and 2A were subjected to DNP SENS analysis. The MAS DNP SENS spectra of 1C and 2B in TCE show proton enhancements of 119 and 101, respectively (Fig. S17 (ESI†)). The ¹⁵N CP MAS DNP SENS spectra of 1C and 2B were recorded with only 3072 scans (Fig. 9). In the case of 1C, the spectrum shows an intense signal at about 24 ppm that is attributed to Ta_xN_y (Fig. 9(A)). For 2B, in addition to the peak at 24 ppm, a signal at 10 ppm is clearly visible that can be assigned to TaO_xN_y (Fig. 9(C)). There is only one report that describes a similar type of NMR observation for such small clusters of tantalum nitride;⁵⁷ Wolczansky et al. demonstrated that similar types of alkylidene precursors generated clusters/oligomers of amorphous tantalum nitride at temperatures below 673 K and crystalline cubic tantalum nitride at temperatures above 1023 K. They also observed that the ¹⁵N signal corresponding to Ta_xN_y shifted upfield in the case of small partially amorphous clusters.

Fig. 9 ¹⁵N CP MAS DNP SENS spectrum (100 K, 400 MHz/263 GHz gyrotron) of (A) 1C and (C) 2B impregnated with 16 mM TEKPol in TCE solution. The spectra were recorded with 3072 scans under microwave irradiation, recycle delay of 3 s, contact time of 4 ms, and MAS frequency at 8 kHz. (B) and (D) Two-dimensional ¹H–¹⁵N HETCOR DNP SENS spectrum of 1C and 2B acquired with 2048 scans per t₁ increment, 96 individual increments, and 4 ms contact time. During t₁, e-DUMBO-1 homonuclear ¹H decoupling was applied, and proton chemical shifts were corrected by applying a scaling factor of 0.57. See Table S1† for the details.

In general, the ¹⁵N signal for a bulk transition metal nitride is expected as a broad band at ∼400 ppm.⁵⁸ For NPs of metal nitrides, the corresponding signal is shifted up field to 24 ppm (Fig. 9(A)). Moreover, as the coordinating nitrogen is replaced by oxygen, the ¹⁵N signal corresponding to TaO_xN_y is further shifted to 10 ppm (Fig. 9(C)). The ¹H–¹⁵N HETCOR DNP SENS spectra of 1C and 2B (Fig. 9(B) & (D)) recorded with the long contact time of 5 ms show a distant correlation with the ¹H of the solvent, which appears at about 5 ppm. Thus, the ¹⁵N DNP SENS results complement the XPS results.

Evidence from EXAFS spectroscopy. X-ray absorption spectroscopy (XAS) provides detailed information about the local electronic and geometric structure of an absorber atom and its near surroundings. XAS experiments characterizing SiO₂-700- supported tantalum complexes, namely, 1A, 1B, 2, and 2A were carried out at the Ta L_III edge (9881 eV), in a specially designed cell with the samples at liquid nitrogen temperature, under high vacuum to prevent exposure to air or moisture.⁵⁹

EXAFS data analysis was carried out with the software packages ATHENA and XDAP.^60,61 Reference files used for phase shift and backscattering amplitude corrections were calculated with the FEFF 7.0 code.⁶² A summary of the results is presented in Table 1, and the details of the fitting and analyses are presented in Fig. S18–S30 and Tables S6–S9 (ESI†).

Table 1 Summary of results of XAS characterizing SiO₂-700-supported Ta complexes^a

Sample	Shell	N	R (Å)	Δσ^2 × 10^3 (Å^2)	ΔE0 (eV)
a Notation: N, coordination number; R, absorber-backscatterer distance; Δσ^2, disorder term (Debye–Waller factor); ΔE0, inner potential correction; m, methyl; s, support; mo, methoxy; l, long; t-mo, terminal methoxy; b-mo, bridging methoxy. Estimated error bounds for each parameter are as follows: N ± 20%; R ± 0.02 Å; Δσ^2 ± 20%; ΔE0 ± 20%.
1A	Ta–Cm	4.0	2.20	3.0	−10.0
	Ta–Os	1.5	1.90	1.0	−9.7
1B Model I	Ta–Omo/s	2.0	1.86	3.7	3.2
	Ta–C	2.3	2.57	1.1	14.4
	Ta–Cmo	3.5	3.18	6.2	−14.7
	Ta–Ol	5.2	3.76	10.4	−6.6
1B Model II	Ta–Omo/s	2.0	1.85	3.6	4.6
	Ta–O	1.6	2.55	1.4	7.5
	Ta–Cmo	4.2	3.20	8.2	−14.7
	Ta–Ol	3.7	3.75	6.4	−5.8
2	Ta–Ot-mo	2.0	1.88	4.7	3.8
	Ta–Ob-mo	2.0	2.20	3.4	−2.7
	Ta–Cmo	4.0	3.15	9.9	−13.6
	Ta–Ta	1.2	3.40	4.5	−11.2
2A	Ta–Omo/s	3.3	1.88	6.3	1.5
	Ta–Cmo	3.3	3.11	0.1	−16.6
	Ta–Sis	2.8	3.33	5.2	3.7
	Ta–Ol	5.8	3.78	10.9	−8.3

---

The best-fit model for 1A is consistent with the expectation of methyl groups and support oxygen atoms bonded to the tantalum. The model thus gives evidence of a Ta–C contribution and a Ta–O contribution. The non-integer Ta–O coordination number (Table 1) could be an indication of the presence of some bipodal species, possibly formed under the influence of the X-ray beam during the measurement, as was observed for a silica-supported tungsten complex.⁶³ The results are nonetheless consistent with the expectation that the sample consisted of predominantly of monopodal species, [(≡SiO–)Ta(CH₃)₄].

In fitting the data characterizing sample 1B, two different models were considered to resolve O in the support and O in the methoxy groups. They are Models I and II, consisting of four shells, the maximum number justified by the data: (1) a Ta–O contribution (where O represents unresolved methoxy and support O atoms), (2) a Ta–C contribution, (3) another Ta–O contribution to represent the interaction of Ta with O atoms of the support at a distance greater than a bonding distance, and (4) another Ta-light atom contribution. In Model I, the light atom is C; in Model II, it is O. These models provided satisfactory fits of the data, with Model II giving a slightly better fit than the other. The important point is the confirmation of the presence of methoxy groups on the Ta, with Ta–O and a Ta–C contributions at appropriate distances for tantalum methoxy species (Table 1). But the results are less than quantitative, failing to resolve all of the Ta–C and Ta–O contributions, and thus reflecting the limitations of EXAFS spectroscopy for structures such as ours.

EXAFS spectroscopy was also used to characterize the dimeric precursor 2, for which there is no known crystal structure. The results (Table 1) demonstrate a Ta–Ta contribution, with a coordination number of 1 (within error) at a distance of 3.34 Å, consistent with a dimer. The EXAFS data also give evidence of methoxy groups (Table 1), but the C/Ta and O/Ta ratio suggested from the Ta–O and Ta–C coordination numbers is about 3, rather than the value of 5 expected for Ta(OMe)₅. Thus, the data are less than sufficient for a structure determination, and we postulate that the sample was degraded under the influence of the X-ray beam in the experiment, consistent with our observation that the second and third scans were different from the first (the third scan was used in the data analysis, because the data quality was higher than that of the earlier scans).

An approach similar to that described above for sample 1B was used in fitting the data characterizing the supported sample prepared from precursor 2, that is, 2A. Again, it was found that resolution of the various Ta–O contributions was not justified by the data, and so a simplified model was used (Table 1). The best-fit model leads to the important result that there is no EXAFS evidence of a Ta–Ta contribution, and thus there is support for the above-stated inference that the dimers broke and gave single-site species.

Thus, although EXAFS data were not sufficient to provide conclusive evidence of the exact structures of 1B and 2, the best-fit models support the following hypothesis: (a) transformation of 1 to 1B takes place by oxygen insertion (Ta–C to Ta–O–C), (b) 2 is a dimer with two bridging methoxide ligands with a Ta–Ta contribution, (c) 2A is a single-site species with no Ta–Ta contribution, and (d) 1B and 2A are similar types of surface species.

Catalytic properties of silica-supported tantalum nitride NPs

Ta_xN_y and TaO_xN_y NPs (1C and 2B) were tested as catalysts for cyclooctene epoxidation with liquid-phase reactants including 30 wt% H₂O₂ as the oxidant. Dilute hydrogen peroxide is one of the most convenient oxidants for epoxide formation because of its ease of handling and high oxygen content, and the relative lack of by-products in the reactions.^64,65 The reaction conditions were as follows: a volume of 0.92 mL of cyclooctene in 4 mL of methanol with 4–70 mg of catalyst was used in a batch reactor; 3 mL of H₂O₂ was used as the oxidant. The epoxidation reaction was carried out at 333 K for 18 h under refluxing conditions. Blank experiments were conducted in the absence of either a catalyst or H₂O₂ to assess the activity. No conversion was observed in the absence of the peroxide. In contrast, 8% conversion and approximately 98% selectivity were observed in the presence of H₂O₂ but in the absence of a catalyst. The catalyst performance data are presented in Fig. 10 and Table S4(a) (ESI†).

Fig. 10 Catalytic activities of 2B and 1C for cyclooctene epoxidation. (Black-conversion; blue-TON; red-selectivity) [reaction conditions: 0.92 mL of cyclooctene; 4–70 mg of catalyst; 4 mL of methanol and 3 mL of H₂O₂ (30 wt%); @ 333 K for 18 h].

Turnover numbers as high as 1395 were observed with 0.2 wt% of Ta in the catalyst. With an increase in Ta content, a gradual decrease of TON values was observed.⁶⁶ In contrast, conversion increased with increasing catalyst Ta content. The product formed was mainly cyclooctene epoxide with traces of cyclooctene diol. The observed selectivities were between 98 and 99% for the lower catalyst Ta loadings. A reaction with Aerosil silica but without NPs was conducted to provide evidence of the effect of the support. Results were similar to that of the blank reaction (performed with H₂O₂ but no catalyst) showing a conversion of ∼10% and a relatively low selectivity of ∼87%. Thus, silica is not just an innocent spectator in this partial oxidation reaction, but its effect under our reaction conditions was minimal, corresponding to its weak Brønsted acidity.⁶⁷ The catalyst performance data indicate that 1C (Ta content ∼0.7 wt%) shows only slightly better performance than 2B, which might be attributed to the smaller particles of 1C. For Ta contents below 3.4 wt%, the observed selectivities were ∼98%. Yet, similar selectivities were observed for various Ta contents. However, a further increase in Ta content resulted in a decline in the selectivity. A significant improvement in the conversion, from 23 to 99% or more, was achievable with increasing Ta content.

On the basis of reported¹⁷ mechanisms, we suggest that only surface Ta sites, which are either oxo or bis-hydroxide, could be the active sites (Scheme S3 (ESI†)). Because the surface to volume ratio of the NPs is high, they have more Ta sites than the corresponding masses of bulk tantalum nitrides and are evidently more efficient catalysts. Gao et al.¹⁷ reported that pure Ta₃N₅ has a higher basicity than TaON. High basicity might induce the decomposition of H₂O₂. As a result, the effective amount of H₂O₂ available in the reaction medium may not be sufficient for the formation of tantalum hydroperoxide active sites. Their study has shown that TaON with controlled basicity influences the epoxidation activity positively. The surface of tantalum nitride NPs undergoes oxidation to form a similar surface layer of TaO_xN_y which probably generates the active sites during the reaction, and this might be the reason for the observed similarities in activities of 1C and 2B. XPS results show the presence of TaO_xN_y phase in 1C.

Gao et al.¹⁷ used unsupported Ta₃N₅/TaON NPs (∼20 nm in average diameter) for the same reaction. They found approximately 57% conversion with Ta₂O₅, whereas Ta₃N₅ or TaON showed higher conversions, including, for example, 97% conversion with 96% selectivity with TaON nanoclusters being the best. By making use of supported NPs (0.5–3 nm in diameter) for the same reaction under similar reaction conditions, we observed conversions approaching 100% without loss of selectivity with a catalyst Ta content as low as ∼3.4 wt%. Notably, we did not observe any leaching of the active species from the surface. The substrate scope of the Ta_xN_y supported catalyst was probed with other substrates such as styrene, cyclohexene, and 1-octene and the results are presented in Table S4(b) (ESI†).

We compared the observed activities with literature values (shown for catalysts incorporating transition metals and supported on silica) for the catalytic cyclooctene epoxidation (Table S4(a) and S5 (ESI†)). Although a direct comparison of our data with literature data is not possible, we have sufficient data to infer that the activities reported in the literature are lower than that of the supported nitride/oxynitride catalyst reported here. At all of the Ta loadings tested, high selectivities were obtained for cyclooctene epoxide, both with samples 1C and 2B. These observations and the comparisons with the literature suggest that our metal nitride or metal oxynitride/silica samples could be promising alternatives to metal/silica catalysts for the cyclooctene epoxidation reaction. On the basis of the literature, we suggest a tentative mechanism for the cyclooctene epoxidation catalyzed by Ta_xN_y/TaO_xN_y NPs on silica, as summarized in Scheme S3.†^17,68–73

Conclusions

Highly dispersed Ta_xN_y/TaO_xN_y NPs with diameters in the range of 0.5 to 3 nm supported on silica were synthesized from supported tantalum complexes according to the strategy of surface organometallic chemistry. Characterization of the surface species at various stages of the synthesis by a range of spectroscopic and microscopic methods gave evidence of the chemistry. Especially helpful was solid-state NMR spectroscopy, specifically, DNP SENS, for understanding this chemistry—this technique enabled the sensitive detection of hetero nuclei including ¹³C, ¹⁵N, and ²⁹Si with the aid of the polarizing agent TEKPol. The results indicate oxygen insertion into supported tantalum alkyl complexes to form tantalum alkoxides at room temperature. NMR and EXAFS data provide evidence of a binuclear to mononuclear transformation of Ta₂(OMe)₁₀ upon grafting to the support. Irrespective of whether the precursor was mono- or binuclear, nitriding resulted in the formation of a mixture of Ta_xN_y and TaO_xN_y NPs from the molecular precursors, with the sizes depending on the precursor. The NPs catalyse cyclooctene epoxidation with selectivities of ∼98% at conversions of ∼99%.

Acknowledgements

We gratefully acknowledge the financial support provided by the King Abdullah University of Science and Technology (KAUST). We thank Dr Kun Li, Dr Nejib Hedili, and Dr Abdul-Hamid Emwas of the KAUST core labs for HRTEM, XPS, and NMR measurements, respectively, and Prof. Kazuhiro Takanabe for help with the ammonia treatments. We acknowledge Prof. Lyndon Emsley, EPFL, Switzerland, for helpful suggestions. We acknowledge beam time at beamline 4-1 of the Stanford Synchrotron Radiation Lightsource supported by the U.S. DOE Division of Materials Science under Contract No. DE-AC02-76SF00515.

Notes and references

1. A. Borodziński and M. Bonarowska, Langmuir, 1997, 13, 5613–5620.
2. W. Lu and C. M. Lieber, Nat. Mater., 2007, 6, 841–850.
3. J. Yang, T. Ling, W.-T. Wu, H. Liu, M.-R. Gao, C. Ling, L. Li and X.-W. Du, Nat. Commun., 2013, 4, 1695.
4. (a) X. Cheng, S. B. Lowe, P. J. Reece and J. J. Gooding, Chem. Soc. Rev., 2014, 43, 2680–2700; (b) C.-T. Ho, K.-B. Low, R. F. Klie, K. Maeda, K. Domen, R. J. Meyer and P. T. Snee, J. Phys. Chem. C, 2011, 115, 647–652.
5. L. Fan, M. Zhu, X. Lee, R. Zhang, K. Wang, J. Wei, M. Zhong, D. Wu and H. Zhu, Part. Part. Syst. Charact., 2013, 30, 764–769.
6. C. Zhen, L. Wang, G. Liu, G. Q. Lu and H.-M. Cheng, Chem. Commun., 2013, 49, 3019–3021.
7. T. Hisatomi, M. Otani, K. Nakajima, K. Teramura, Y. Kako, D. Lu, T. Takata, J. N. Kondo and K. Domen, Chem. Mater., 2010, 22, 3854–3861.
8. G. H. Carey, A. L. Abdelhady, Z. Ning, S. M. Thon, O. M. Bakr and E. H. Sargent, Chem. Rev., 2015, 115, 12732–12763.
9. J. P. Wilcoxon, T. Martino, E. Klavetter and A. P. Sylwester, in Nanophase Materials: Synthesis—Properties—Applications, ed. G. C. Hadjipanayis and R. W. Siegel, Springer Netherlands, Dordrecht, 1994, pp. 771–780,  DOI:10.1007/978-94-011-1076-1_80.
10. T. Bligaard, R. M. Bullock, C. T. Campbell, J. G. Chen, B. C. Gates, R. J. Gorte, C. W. Jones, W. D. Jones, J. R. Kitchin and S. L. Scott, ACS Catal., 2016, 6, 2590–2602.
11. F. Humblot, J. P. Candy, F. Le Peltier, B. Didillon and J.-M. Basset, J. Catal., 1998, 179, 459–468.
12. H. Zhu, D. H. Anjum, Q. Wang, E. Abou-Hamad, L. Emsley, H. Dong, P. Laveille, L. Li, A. K. Samal and J.-M. Basset, J. Catal., 2014, 320, 52–62.
13. C. R. Kagan and C. B. Murray, Nat. Nanotechnol., 2015, 10, 1013–1026.
14. A. P. Alivisatos, Science, 1996, 271, 933–937.
15. W. Jiang, C. Siming, S. Alwyn and L. Huiyun, J. Phys. D: Appl. Phys., 2015, 48, 363001.
16. X. Wang, G. Sun, N. Li and P. Chen, Chem. Soc. Rev., 2016, 45, 2239–2262.
17. Q. S. Gao, S. N. Wang, Y. C. Ma, Y. Tang, C. Giordano and M. Antonietti, Angew. Chem., Int. Ed., 2012, 51, 961–965.
18. Q. S. Gao, C. Giordano and M. Antonietti, Small, 2011, 7, 3334–3340.
19. Q. Gao, C. Giordano and M. Antonietti, Angew. Chem., 2012, 124, 11910–11914.
20. Q. Gao, C. Giordano and M. Antonietti, Small, 2011, 7, 3334–3340.
21. Z. Wang, J. Hou, S. Jiao, K. Huang and H. Zhu, J. Mater. Chem., 2012, 22, 21972–21978.
22. K. Maeda, ACS Catal., 2013, 3, 1486–1503.
23. Y. Sasaki, Z. Tokuyasu, Y. Ono, M. Iwasaki and S. Ito, Adv. Mater. Sci. Eng., 2009, 2009, 4.
24. Y.-C. Wang, C.-Y. Chang, T.-F. Yeh, Y.-L. Lee and H. Teng, J. Mater. Chem. A, 2014, 2, 20570–20577.
25. M. Jansen and H. P. Letschert, Nature, 2000, 404, 980–982.
26. (a) A. J. Rossini, A. Zagdoun, M. Lelli, A. Lesage, C. Copéret and L. Emsley, Acc. Chem. Res., 2013, 46, 1942–1951; (b) E. Pump, J. Viger-Gravel, E. Abou-Hamad, M. K. Samantaray, B. Hamzaoui, A. Gurinov, D. H. Anjum, D. Gajan, A. Lesage, A. Bendjeriou-Sedjerari, L. Emsley and J.-M. Basset, Chem. Sci., 2017, 8, 284–290.
27. L. Piveteau, T.-C. Ong, A. J. Rossini, L. Emsley, C. Copéret and M. V. Kovalenko, J. Am. Chem. Soc., 2015, 137, 13964–13971.
28. E. Bouleau, P. Saint-Bonnet, F. Mentink-Vigier, H. Takahashi, J. F. Jacquot, M. Bardet, F. Aussenac, A. Purea, F. Engelke, S. Hediger, D. Lee and G. De Paepe, Chem. Sci., 2015, 6, 6806–6812.
29. A. Lesage, M. Lelli, D. Gajan, M. A. Caporini, V. Vitzthum, P. Miéville, J. Alauzun, A. Roussey, C. Thieuleux, A. Mehdi, G. Bodenhausen, C. Coperet and L. Emsley, J. Am. Chem. Soc., 2010, 132, 15459–15461.
30. M. Valla, A. J. Rossini, M. Caillot, C. Chizallet, P. Raybaud, M. Digne, A. Chaumonnot, A. Lesage, L. Emsley, J. A. van Bokhoven and C. Copéret, J. Am. Chem. Soc., 2015, 137, 10710–10719.
31. O. A. Nikonova, G. A. Seisenbaeva, V. G. Kessler, P. A. Shcheglov and D. V. Drobot, Russ. J. Inorg. Chem., 2007, 52, 1687–1693.
32. C. Copéret, A. Comas-Vives, M. P. Conley, D. P. Estes, A. Fedorov, V. Mougel, H. Nagae, F. Núñez-Zarur and P. A. Zhizhko, Chem. Rev., 2016, 116, 323–421.
33. A. Johansson, M. Roman, G. A. Seisenbaeva, L. Kloo, Z. Szabo and V. G. Kessler, J. Chem. Soc., Dalton Trans., 2000, 387–394,  10.1039/A907720K.
34. M. T. Vandenborre, B. Poumellec, C. Alquier and J. Livage, J. Non-Cryst. Solids, 1989, 108, 333–337.
35. Y. Chen, E. Abou-hamad, A. Hamieh, B. Hamzaoui, L. Emsley and J.-M. Basset, J. Am. Chem. Soc., 2015, 137, 588–591.
36. Y. Chen, S. Ould-Chikh, E. Abou-Hamad, E. Callens, J. C. Mohandas, S. Khalid and J.-M. Basset, Organometallics, 2014, 33, 1205–1211.
37. R. R. Schrock and P. Meakin, J. Am. Chem. Soc., 1974, 96, 5288–5290.
38. E. C. Garrett, T. M. Figg and T. R. Cundari, Inorg. Chem., 2014, 53, 7789–7798.
39. S.-J. Chen and Z.-L. Xue, Organometallics, 2010, 29, 5579–5584.
40. L. G. Hubert-Pfalzgraf and J. G. Riess, Inorg. Chem., 1975, 14, 2854–2856.
41. D. C. Bradley and C. E. Holloway, J. Chem. Soc. A, 1968, 219–223.
42. R. P. Saint-Arroman, J.-M. Basset, F. Lefebvre and B. Didillon, Appl. Catal., A, 2005, 290, 181–190.
43. L. J. Burcham, G. Deo, X. Gao and I. E. Wachs, Top. Catal., 2000, 11, 85–100.
44. (a) Y. Luo, J. Yi, D. Tong and C. Hu, Green Chem., 2016, 18, 848–857; (b) R. K. Singha, S. Ghosh, S. S. Acharyya, A. Yadav, A. Shukla, T. Sasaki, A. M. Venezia, C. Pendem and R. Bal, Catal. Sci. Technol., 2016, 6, 4601–4615.
45. A. Noshay, M. Matzner and T. C. Williams, Ind. Eng. Chem. Prod. Res. Dev., 1973, 12, 268–277.
46. E. Nurlaela, S. Ould-Chikh, M. Harb, S. del Gobbo, M. Aouine, E. Puzenat, P. Sautet, K. Domen, J.-M. Basset and K. Takanabe, Chem. Mater., 2014, 26, 4812–4825.
47. L. Li, E.-W. Niu, G.-H. Lv, W.-R. Feng, W.-C. Gu, G.-L. Chen, G.-L. Zhang, S.-H. Fan, C.-Z. Liu and S.-Z. Yang, Chin. Phys. Lett., 2006, 23, 3018.
48. N. Arshi, J. Lu, Y. K. Joo, J. H. Yoon and B. H. Koo, Surf. Interface Anal., 2015, 47, 154–160.
49. H. X. Dang, N. T. Hahn, H. S. Park, A. J. Bard and C. B. Mullins, J. Phys. Chem. C, 2012, 116, 19225–19232.
50. A. Arranz and C. Palacio, Appl. Phys. A, 2005, 81, 1405–1410.
51. M. Chisaka, A. Ishihara, N. Uehara, M. Matsumoto, H. Imai and K. Ota, J. Mater. Chem. A, 2015, 3, 16414–16418.
52. M. Liao, J. Feng, W. Luo, Z. Wang, J. Zhang, Z. Li, T. Yu and Z. Zou, Adv. Funct. Mater., 2012, 22, 3066–3074.
53. Y. Kado, C.-Y. Lee, K. Lee, J. Muller, M. Moll, E. Spiecker and P. Schmuki, Chem. Commun., 2012, 48, 8685–8687.
54. E. Nurlaela, M. Harb, S. del Gobbo, M. Vashishta and K. Takanabe, J. Solid State Chem., 2015, 229, 219–227.
55. A. Zagdoun, G. Casano, O. Ouari, M. Schwarzwälder, A. J. Rossini, F. Aussenac, M. Yulikov, G. Jeschke, C. Copéret, A. Lesage, P. Tordo and L. Emsley, J. Am. Chem. Soc., 2013, 135, 12790–12797.
56. M. W. McKittrick and C. W. Jones, Chem. Mater., 2003, 15, 1132–1139.
57. M. M. Banaszak Holl, P. T. Wolczanski, D. Proserpio, A. Bielecki and D. B. Zax, Chem. Mater., 1996, 8, 2468–2480.
58. K. J. D. MacKenzie, R. H. Meinhold, D. G. McGavin, J. A. Ripmeester and I. Moudrakovski, Solid State Nucl. Magn. Reson., 1995, 4, 193–201.
59. R. E. Jentoft, S. E. Deutsch and B. C. Gates, Rev. Sci. Instrum., 1996, 67, 2111–2112.
60. E. A. Stern, M. Newville, B. Ravel, Y. Yacoby and D. Haskel, Phys. B, 1995, 208, 117–120.
61. M. Newville, J. Synchrotron Radiat., 2001, 8, 96–100.
62. S. I. Zabinsky, J. J. Rehr, A. Ankudinov, R. C. Albers and M. J. Eller, Phys. Rev. B: Condens. Matter Mater. Phys., 1995, 52, 2995–3009.
63. N. Maity, S. Barman, E. Callens, M. K. Samantaray, E. Abou-Hamad, Y. Minenkov, V. D'Elia, A. S. Hoffman, C. M. Widdifield, L. Cavallo, B. C. Gates and J.-M. Basset, Chem. Sci., 2016, 7, 1558–1568.
64. W. R. Sanderson, Pure Appl. Chem., 2000, 72, 1289–1304.
65. R. Noyori, M. Aoki and K. Sato, Chem. Commun., 2003, 1977–1986,  10.1039/B303160H.
66. S. Xue, G. Chen, Z. Long, Y. Zhou and J. Wang, RSC Adv., 2015, 5, 19306–19314.
67. P. Chandra, D. S. Doke, S. B. Umbarkar, K. Vanka and A. V. Biradar, RSC Adv., 2015, 5, 21125–21131.
68. P. J. Cordeiro and T. D. Tilley, ACS Catal., 2011, 1, 455–467.
69. D. A. Ruddy and T. D. Tilley, Chem. Commun., 2007, 3350–3352.
70. N. Morlanes and J. M. Notestein, Appl. Catal., A, 2010, 387, 45–54.
71. U. Arnold, R. Serpa da Cruz, D. Mandelli and U. Schuchardt, J. Mol. Catal. A: Chem., 2001, 165, 149–158.
72. N. A. Stephenson and A. T. Bell, J. Am. Chem. Soc., 2005, 127, 8635–8643.
73. M.-L. Lin, K. Hara, Y. Okubo, M. Yanagi, H. Nambu and A. Fukuoka, Catal. Commun., 2011, 12, 1228–1230.

---

Footnotes

† Electronic supplementary information (ESI) available: Complete experimental procedures, supporting characterization techniques, data and the details for the prepared compounds are provided. See DOI: 10.1039/c7sc01365e

‡ These authors contributed equally.

---