Coordination-induced bond weakening of water at the surface of an oxygen-deficient polyoxovanadate cluster

Hydrogen-atom (H-atom) transfer at the surface of heterogeneous metal oxides has received significant attention owing to its relevance in energy conversion and storage processes. Here, we present the synthesis and characterization of an organofunctionalized polyoxovanadate cluster, (calix)V6O5(OH2)(OMe)8 (calix = 4-tert-butylcalix[4]arene). Through a series of equilibrium studies, we establish the BDFE(O–H)avg of the aquo ligand as 62.4 ± 0.2 kcal mol−1, indicating substantial bond weaking of water upon coordination to the cluster surface. Subsequent kinetic isotope effect studies and Eyring analysis indicate the mechanism by which the hydrogenation of organic substrates occurs proceeds through a concerted proton–electron transfer from the aquo ligand. Atomistic resolution of surface reactivity presents a novel route of hydrogenation reactivity from metal oxide surfaces through H-atom transfer from surface-bound water molecules.


Introduction
Manipulation of the reactivity of protons and electrons at electrochemical interfaces is integral for the development of improved energy storage and conversion processes. This fact renders an understanding of the basic science underlying proton-coupled electron transfer (PCET) at redox active surfaces important for the development of efficient and sustainable strategies for the production of commodity chemicals. One class of compounds that has risen to prominence in the area of PCET reactions are heterogeneous metal oxides (MO x ). These materials are ubiquitous in the activation of small molecules (e.g. O 2 , CO 2 , N 2 ), and also have been reported to facilitate the direct utilization of H 2 as a chemical fuel (e.g. water electrolyzers, fuel cells). [1][2][3][4][5] Despite the prevalence of MO x in energy storage and conversion schematics involving the manipulation of hydrogen atom (H-atom) equivalents (e − /H + ), there remain signicant challenges in obtaining an atomistic understanding of reaction pathways by which H-atoms are transferred from the surface of MO x to small molecule substrates. Mechanistic analysis is rendered particularly challenging due to difficultly in distinguishing between reactivity mediated by surface and inserted H-atom equivalents of extended MO x . [6][7][8] Furthermore, the lack of uniformity by which H-atom equivalents interact with the surface of MO x results in a wide variety of interpretations as to the mechanism of transfer of protons and electrons. Indeed, the inhomogeneity of MO x surfaces provokes questions related to the role of bridging vs. terminal oxido moieties in H-atom uptake and transfer reactions, as well as multi-site PCET invoking surface ligand centred proton-transfer processes (Fig. 1). 9,10 To experimentally model mechanisms of PCET at MO x surfaces, researchers have turned to the investigation of their molecular analogues. [11][12][13][14][15][16] One class of compounds that is particularly well-suited for these studies is polyoxometalates (POMs). POMs are three-dimensional MO x assemblies composed of early transition metal oxyanions linked together by edge-or face-sharing oxide ligands. Broadly speaking, POMs possess similar surface morphologies to their extended state congeners, however, these clusters limit their reactivity with Hatom equivalents to surface bound oxide ligands (i.e. no intercalation). While proton-responsive redox properties have been reported for POMs, 17 there remain limited examples in which thermodynamics or kinetics of PCET has been explored. The few studies that have investigated PCET at POM surfaces have revealed that the uptake of H-atom equivalents occurs primarily at bridging oxide positions. 12,18 An alternative mechanism for the generation of reactive Hatom equivalents at MO x surfaces invokes coordinationinduced bond weakening of O-H moieties upon association of water to a surface oxygen-atom (O-atom) vacancy ( Fig. 1). In this scenario, a coordinatively unsaturated, reduced metal centre at the surface of the material binds to water, resulting in a weakening of the O-H bond of the substrate. This phenomenon has been well-documented in the reactivity of water with reduced, mononuclear transition metal complexes, [19][20][21][22][23][24][25][26][27] however, to date, has not been invoked in extended solids. This is despite the fact that surfaces of reducible metal oxide materials are riddled with these defects sites. [28][29][30][31] As a result, our understanding as to how coordination-induced bond weakening impacts the design criteria for the formation of reactive H-atom equivalents at the surface of MO x systems from "green" substrates (e.g. H 2 O, NH 3 ) remains limited.
While touted as excellent models for studies investigating the surface reactivity of extended MO x materials, the isolation of POMs possessing single O-atom vacancy is exceedingly rare. Early examples of the formation of O-atom defect sites in "heteropoly brown" tungsten oxide assemblies emerged from the work of Launay and Pope and Piepgrass. In these studies, reduction of the polyoxotungstate in the presence of acid results in the formation of the 6e − /6H + reduced assembly [W VI 9 (-W IV OH 2 ) 3 ]. [32][33][34][35] Despite the observation of aquo ligands bound to the reduced tungsten(IV) centres, H-atom transfer reactivity is not observed. Instead, the aquo ligands are displaced in organic solvents by oxygenated substrates, resulting in reoxidation of the cluster via reductive cleavage of E]O bonds (e.g., O]AsR 3 , O]NPh). 36 Our group has also reported O-atom vacancy formation in a series of reduced polyoxovanadate-alkoxide clusters (POV-alkoxide); 13,37-43 most relevant to the studies reported here, we have described defect formation at the surface of the Lindqvist ion following the transfer of two H-atom equivalents to a terminal vanadyl site. 13 However, it was found that the affinity of the reduced V III ion toward acetonitrile results in rapid displacement of the transiently generated surface aquo ligand, preventing analysis of the surface bound water adduct.
Herein, we report the isolation of the rst aquo complex of a POV-alkoxide cluster. This unique compound is accessed via addition of two H-atom equivalents to a terminal oxido ligand of a surface vanadyl in a calix-functionalized hexavanadate cluster, (calix)V 6 O 6 (MeCN)(OMe) 8 , (calix)V 6 O 6 . Quantication of the strength of the resultant O-H bonds of the aquo ligand is determined through a series of equilibrium studies. Results reveal that coordination of the aquo ligand to the surface of the cluster at an oxygen atom point defect results in signicant weakening of the O-H bonds of water (∼50 kcal mol −1 ), rendering them reactive toward H-atom transfer. Further insight into the mechanism of delivery of H-atom equivalents to organic substrates is obtained through kinetic isotope effect studies and Eyring analysis. Our results indicate that the hydrogenation of the organic substrate relies on a rate determining concerted proton-electron transfer (CPET) step, as opposed to a stepwise transfer of protons and electrons. Our results indicate that the coordination of water to reactive surface defect sites may play a critical role as a source of environmentally benign H-atoms in hydrogenation schematics.

Results and discussion
Previously, our research group has reported changes in the electrochemical properties of the polyoxovanadate-alkoxide cluster following surface-functionalization of the assembly with a calix ligand. 44 The cyclic voltammogram of (calix)V 6 O 6 possesses three quasi-reversible redox events, shied by approximately +0.30 V in comparison to its non-substituted congener, [V 6 O 6 (OMe) 12 (MeCN)] 0 (MeCN = acetonitrile). Curious as to whether the changes in the redox potentials would inuence the reactivity of terminal vanadyl ligands with Hatoms, we set out to investigate the reactivity of (calix)V 6 O 6 with dihydrophenazine (DHP). DHP has been established to have weak N-H bonds (e.g. BDFE(N-H) avg of DHP in MeCN = 58.9 kcal mol −1 ), rendering this substrate a convenient entry point for our work.
To date, HAT reactions with POV-alkoxide clusters have been conducted in MeCN. The historical selection of this solvent is due to the breadth of reported thermochemical data (e.g. BDFE(E-H) values, pK a values, reduction potentials) for organic substrates. [45][46][47] This allows for facile benchmarking of the reactivity of surface oxido moieties with E-H bonds. Unfortunately, (calix)V 6 O 6 has limited solubility in MeCN, necessitating studies be performed in an alternative solvent. For this work, we elected to use tetrahydrofuran (THF) as solvent.
Addition of 1 equiv. of DHP to (calix)V 6 O 6 at room temperature in THF results in a colour change from brown to red over 1 h (Scheme 1). Analysis of the reaction mixture by 1 H NMR spectroscopy reveals formation of a new product, with four paramagnetically shied and broadened resonances (d = 28.71, 26.45, −5.10, and −11.61 ppm). Additional signals corresponding to the calix ligand are located in the diamagnetic region of the spectrum (Fig. S1 †). The 1 H NMR spectrum of the product is substantially different from that of the starting material, (calix)V 6 O 6 ( Fig. S2 †). In particular, two new, up-eld resonances are observed. Previous reports from our laboratory have concluded that similar, negatively-shied signals correspond to protons of methoxide ligands adjacent to a reduced vanadium centre embedded within the cluster core upon Oatom vacancy formation. 40,41,43,48 Indeed, we have recently reported the reduction of an oxidized form of the calix-substituted POV-alkoxide cluster, [(calix)V 6 O 7 (OMe) 8 ] 0 , by oxygen atom transfer to a tertiary phosphine; the product possesses a similar pattern of signals to that described here, albeit with resonances that possess different chemical shis due to the differing oxidation state distribution of vanadium ions of the starting . 49 Further support for the reduction of the cluster core was obtained through analysis of the product by electronic absorption spectroscopy (Fig. S3 †). This characterization technique has been well established by our group and others to report on the degree of reduction of the cluster core. [50][51][52][53][54][55] Comparison of the electronic absorption spectra of the product of H-atom transfer to (calix)V 6 O 6 reveals loss of an intervalence charge transfer band (V IV / V V ). This observation is consistent with formation of the desired oxygen-decient product, as vacancy formation is associated with the reduction of the single V V centre in (calix)V 6  Crystals suitable for structural analysis by single crystal X-ray diffraction were grown via slow diffusion of pentane into a concentrated solution of the product in THF ( Fig. 2 and Table  1). Renement of the data revealed formation of the expected reduced assembly, [(calix)V 6 O 5 (OH 2 )(OMe) 8 ] ((calix)V 6 O 5 (OH 2 )), with the new, "oxygen-decient" vanadium centre positioned cis to the calix-bound vanadium ion. The asymmetric unit contains one-half of a vanadium cluster on a crystallographic mirror plane. Bond valence sum calculations obtained from the V-O bond lengths of structurally unique vanadium centres conrm the spectroscopically proposed oxidation state assignments of metals contained within the cluster core (V III 2 V IV 4 , Table S2 †). The aquo ligand bound to the reduced vanadium centre is stabilized by hydrogen bonding interactions with two THF solvent molecules. Notably, the O4/O(THF) distances of 2.665(9) and 2.601(15)Å are quite short, suggesting the existence of strong H-bonding interactions between the coordinated aquo ligand and solvent. 56 Substantial elongation of the V4-O4 bond (2.052(4)Å) from that of the starting material (average V]O t bond length cis to calix ligand = 1.594Å) is observed, consistent with reduction of the vanadyl ion to a vanadium(III) aquo. Indeed, the V4-O4 bond length resembles values reported previously for molecular vanadium(III) aquo complexes (1.967(3)-2.086(2)Å). 57,58 The independent renement of the H-atoms of the aquo ligand (one unique) identied in the difference Fourier map further supports the assignment of an aquo ligand bound to the surface of the assembly.
While by no means is the formation of a vanadium(III) aquo moiety rare, 57,59-64 this result is signicant within the context of our work, given its relevance as the direct product of Scheme 1 Synthesis of (calix)V 6 O 5 (OH 2 ) by H-atom transfer. Previous work from our group has shown that the BDFE(O-H) of reduced POV-alkoxide clusters can be approximated by monitoring the extent to which H-atom transfer occurs between various organic reagents and a cluster. 12,13 The observation of quantitative reduction of the POV suggests that the BDFE(O-H) of the cluster is sufficiently larger than that of DHP, establishing an effective minimum value of the BDFE(O-H) avg for the reduced cluster in THF. However, despite the fact that there exists a substantial library of known thermochemical parameters of PCET reagents, the BDFE(N-H) avg for DHP is not reported in THF. In order to circumvent this obstacle, we turned to alternative methods for obtaining the free energy of the N-H bonds of DHP in THF. While THF has traditionally been underutilized in PCET chemistry, the thermodynamics of PCET reagents in MeCN are well established, including the BDFE(N-H) avg for DHP: 58.7 kcal mol −1 . 47 We are able to calculate the BDFE(N-H) avg for DHP in THF using eqn (1) 65 is the free energy of solvation for a hydrogen radical in the solvent of interest, and DG solv ðX Þ and DG solv ðXHÞ (solv = MeCN, THF) are the free energy of solvation for the radical and the reduced organic substrate, respectively.
The terms corresponding to ½DG solv ðX Þ À DG solv ðXHÞ are primarily dictated by the differences in hydrogen bonding between the radical and reduced version of the organic substrate. Due to the fact that both radical and reduced versions are approximately the same size, other factors that impact solvation can be ignored. 46 With this in mind, an empirical formula for calculating the difference in hydrogen bonding in aprotic solvents has been developed, 66,67 shown in eqn (2) Using these equations, we are able to obtain a calculated BDFE(N-H) avg for DHP in THF (59.2 kcal mol −1 ). This value only differs from the experimentally determined BDFE(N-H) avg for DHP in MeCN by 0.5 kcal mol −1 . 47 Accordingly, we are able to condently conclude that the BDFE(O-H) avg for (calix) V 6 O 5 (OH) 2 is greater than 59.2 kcal mol −1 in THF.
To further benchmark the reactivity of (calix)V 6 O 6 , we explored its behaviour with alternative H-atom transfer reagents that possess stronger E-H bonds. Hydrazobenzene is commonly employed as a reducing reagent (BDFE(N-H) avg,MeCN = 60.9 kcal mol −1 ). 47 Following the aforementioned strategy, we are able to calculate the average bond strength of the N-H bonds in THF (60.4 kcal mol −1 ). Upon the addition of one equivalent of hydrazobenzene to the oxidized cluster, quantitative formation of the reduced complex (calix)V 6 O 5 (OH 2 ) is observed (Fig. S4 †). The observed equilibrium between oxidized and reduced versions of the POV-alkoxide cluster following addition of H 2 NQ to (calix)V 6 O 6 suggests that H-atom transfer is active in both directions (i.e., the cluster is capable of donating and accepting H-atom equivalents). To further probe this nding, we evaluated the ability for H-atom transfer to occur from (calix) V 6 O 5 (OH 2 ) to an H-atom accepting reagent, naphthoquinone (NQ; Scheme 2). Upon addition of the organic substrate to the cluster, formation of (calix)V 6 O 6 is be observed via 1 H NMR spectroscopy (Fig. S6 †), along with the hydrogenated substrate, H 2 NQ. This observation is signicant, as this is the rst example in which H-atom transfer is found to occur at the surface of a polyoxometalate cluster from a coordinated aquo ligand. Additionally, this result implies that the O-H bond strengths of the aquo ligand are substantially weakened upon coordination to the reduced vanadium centre in comparison to that of free water (BDFE(O-H) avg,H2O = 108.9 kcal mol −1 ). 47 Although it is well established that water molecules are capable of associating to surface localized O-atom vacancies, Table 1 Selected bond distances and angles in (calix)V 6 O 6 and (calix) (5) To quantify the free energy of the O-H bonds of the aquo ligand, a series of equilibrium studies were performed (Fig. 3). To begin, we studied the reaction between the reduced cluster, (calix)V 6 O 5 (OH 2 ) and NQ (Scheme 2). Aer 12 hours, no signicant change in the product distribution was observed, indicating that the reaction mixture had reached equilibrium.   (Fig. 4). For example, the titanocene(III) complex reported by Cuevra binds to water, resulting in the formation of a highly reactive O-H bond due to the thermodynamic driving force for oxidation of the Ti(III) centre. 70,71 The change in BDFE(O-H) of water reported here, following coordination to a V(III) center embedded within a multimetallic MO x assembly, is smaller than that described for   The BDFE(O-H) avg obtained for the reduced cluster, (calix) V 6 O 5 (OH 2 ), closely resembles values previously reported by our group for reduced polyoxovanadate cluster containing a similar Lindqvist structure. 12 However, in our previous work, it was determined that reactivity occurs exclusively at bridging oxide ligands, whereas in this work, reactivity is limited to the terminal vanadyl site. The similarity between these two values is surprising, given the differences in the coordination geometry of the active O-atom. Computational analysis into H-atom uptake at ceria oxide nanoparticles reveals substantial differ-

Mechanistic analysis of PCET from (calix)V 6 O 5 (OH 2 )
Establishing of the BDFE(O-H) avg of (calix)V 6 O 5 (OH 2 ) allows for the prediction of the thermochemical driving force for PCET from the cluster surface to a given substrate. However, this information alone provides an incomplete description of charge transfer reactions. While thermodynamics are a critical factor to consider when designing systems for the mediation of PCET reactions, it is important to account for kinetic inuences on the reaction. For instance, PCET reactions that proceed via a stepwise mechanism, with either the electron or proton transferring initially, result in the formation of charged intermediates. These pathways oen require overcoming large activation barriers, and as a consequence, may experience sluggish reaction kinetics. However, the transfer of both electron and proton together in a single kinetic step, known as concerted proton-electron transfer (CPET), can, in some circumstances, avoid the formation of these high energy intermediates.
To probe the mechanism of H-atom transfer from the coordinated water molecule at the surface of the reduced vanadium oxide cluster, we performed a series of pseudo-rst order reaction analyses. We opted to investigate H-atom transfer between the reduced cluster, (calix)V 6 O 5 (OH 2 ), and an excess of the substrate, TEMPO (2,2,6,6-tetramethylpiperidin-1-oxyl; 20-120 equivalents; Scheme 3). TEMPO is an ideal substrate, as it allows for the interrogation of single vs. multi-HAT in the rate determining step of the reaction. This substrate is also exceptionally difficult to reduce in the absence of protons in THF (E c = −2.56 V vs. Fc +/0 , Fig. S8 †), eliminating the possibility of a stepwise ET-PT pathway.
Rate constants were determined by monitoring the change in the electronic absorption spectrum for (calix)V 6 O 5 (OH 2 ) over time. The growth of the low intensity absorption feature at 900 nm is indicative of the formation of the oxidized cluster, (calix)V 6 O 6 (vide supra). Measuring the change in absorbance as a function of time results in a kinetic trace as seen in Fig. S9. † Extraction of the pseudo rst-order rate constant (k obs ) is performed by determining the slope of the plot ln((A t − A inf )/(A 0 − A inf )) vs. time, where A t is absorbance at time = t, A inf is the absorbance at the end of the reaction, and A 0 is the initial absorbance ( Fig. S10; † see Experimental section for more details). Plotting k obs against the concentration of TEMPO reveals a linear correlation (Fig. 5), suggesting the reaction is rst order with respect to TEMPO. The linear relationship between time and the natural log of the concentration of the oxidized cluster noted in Fig. S10 † indicates that the reaction is also rst order with respect to cluster, translating to an overall second order rate law for the reaction. This suggests that the rate limiting step of the reaction involves the transfer of only one H-atom from the aquo ligand to a TEMPO molecule. The second order rate constant for PCET at (calix)V 6 O 5 (OH 2 ) (k PCET ) can be found from the slope in Fig. 5, where k PCET is equal to 1 2 × slope of the best t line to account for the two chemically identical O-H bonds of the aquo ligand, resulting in a k PCET value of 2.3 × 10 −2 M −1 s −1 .
Due to the fact that kinetic analysis of PCET between reduced polyoxometalates and organic substrates is a relatively new area, there are few examples available for comparison of the observed rate constant. One such study from our group determined the rate constant for the reduction of TEMPO by the reduced POValkoxide cluster, [V 6 O 11 (OH) 2 (TRIOL NO2 ) 2 ] −2 . Interestingly, the overall rate constant for this previous example is two orders of magnitude larger than the value reported in this work, despite the fact that H-atom transfer from (calix)V 6 O 5 (OH 2 ) is thermodynamically preferred. The observed discrepancy in the expected rates of reaction can be attributed to a kinetic solvent effect (KSE). The previous study from our group was performed in acetonitrile and observed no signicant interaction between the reduced cluster and solvent. Conversely, structural analysis of (calix)V 6 O 5 (OH 2 ) suggests strong interactions between the aquo ligand and THF. This H-bonding network must be disrupted prior to H-atom transfer. KSE has been shown to impart signicant kinetic barriers for PCET as a result of increases in the activation barrier required to reach the transition state of the reaction and is likely the cause of the relatively sluggish reaction kinetics observed here. 73 Next, kinetic isotope effect (KIE) studies were performed; formation of the deuterated derivative of the reduced cluster, (calix)V 6 O 5 (OD 2 ), was achieved via reduction of (calix)V 6 O 6 using the deuterated compound, d 2 -DHP (see Experimental section for details, Fig. S11-S13 †). Repeating the pseudo-rst order reaction experiments using the deuterated cluster, (calix)V 6 O 5 (OD 2 ), reveals a decrease in the overall rate of reaction (Fig. S14 †). Plotting the k obs values against the concentration of excess TEMPO results in a k PCET value of 1.4 × 10 −2 M −1 s −1 . Comparison rate constants of H-atom transfer from the protonated and deuterated clusters results in a KIE value of 1.61. While a KIE value > 1 suggests a rate determining step that includes the cleavage of the O-H bond at the cluster, this value is not substantial enough to conclusively assign a reaction mechanism. [74][75][76][77] Eyring analysis of the H-atom transfer process was performed to unambiguously distinguish between CPET and PT/ET pathways. We measured k obs under pseudo-rst order reaction conditions across a range of temperatures (25-65°C). From this information, determination of activation parameters for Hatom transfer from (calix)V 6 O 5 (OH 2 ) to TEMPO is possible (DH ‡ = 6.1 AE 0.9 kcal mol −1 ; DS ‡ = −41.7 AE 2.9 cal mol −1 K −1 ; DG ‡ = 18.8 AE 1.8 kcal mol −1 ; Fig. 6). The DH ‡ value is comparable to values obtained previously for a concerted PCET reaction at MO x compounds. 12,14,78,79 The relatively small magnitude of DH ‡ can be explained by the fact that (calix)V 6 O 5 (OH 2 ) is likely able to form a stable hydrogen bond to TEMPO, suggesting the reaction requires relatively little enthalpic energy to reach the activated transition state. 80 The large negative value for DS ‡ suggests a highly ordered intermediate is formed in the transition state of the rate determining step of H-atom transfer. This nding is consistent Fig. 5 Plot of k obs against the concentration of TEMPO initially in solution for the reaction between (calix)V 6 O 5 (OH 2 ) and TEMPO. Each sample contains 0.5 mM of (calix)V 6 O 5 (OH 2 ) in THF and run at 25°C. The linear trend of these results indicates the reaction is first order with respect to TEMPO. The second order rate constant, k PCET can be determined from the slope of the best fit line with a set y-intercept of 0, where 1 2 of the slope is equal to k PCET due to the presence of two chemically equivalent H-atoms at the surface of the reduced cluster. with a TEMPO molecule forming a H-bonding interaction with the surface aquo ligand of (calix)V 6 O 5 (OH 2 ) prior to the formal transfer of the H-atom. Previous reports of MO x compounds performing H-atom transfer reactions via CPET mechanisms found similarly negative DS ‡ values. This can be explained by the fact that CPET is inherently an inner sphere process; formation hydrogen bonds between the H-atom acceptor and donor are required for the reaction to occur. 78,79 Conclusion Here, we have described the synthesis and isolation of a reduced hexavanadate cluster, (calix)V 6 O 5 (OH 2 ). While previous studies from our laboratory have demonstrated the predilection of reduced polyoxovanadate clusters to form bridging hydroxide ligands, this work reports the rst instance of the formation of a stable aquo ligand at the surface of the cluster. Notably, H-atom transfer reactivity is observed from (calix)V 6 O 5 (OH 2 ), indicating that the cluster is capable of performing proton coupled electron transfer reactions using the Hatoms of water. Experimental analysis results in the direct measurement of the BDFE(O-H) avg of the aquo ligand (62.4 AE 0.2 kcal mol −1 ). Mechanistic analysis reveals that H-atom transfer reaction is likely occurring through a CPET pathway.
The experimentally determined BDFE(O-H) avg of the reduced cluster, (calix)V 6 O 5 (OH 2 ) (62.4 AE 0.2 kcal mol −1 ), demonstrates that coordination of water to an O-atom defect site results in a substantial increase in the thermodynamic driving force of H-atom transfer from water. Indeed, the BDFE(O-H) of water decreases ∼50 kcal mol −1 upon coordination to the cluster surface (BDFE(H 2 O) avg = 108.9 kcal mol −1 ). 47 This phenomenon, commonly referred to as coordination induced bond weakening, has been observed previously in mononuclear transition metal complexes. Signicantly, the BDFE(O-H) avg (calix)V 6 O 5 (OH 2 ) exists in a "sweet spot", weak enough to enable the cluster to act as a potent H-atom donor to a variety of small molecule substrates, while avoiding H 2 formation. This is in contrast to the BDFE(O-H) values reported for early transition metal aquo species, which have been reported at values below the point at which the formation of H 2 is thermodynamically favourable (BDFE(H 2 ) z 52 kcal mol −1 ). 81,82 Indeed, this nding indicates that reduced polyoxovanadatealkoxide clusters are promising candidates for use as catalysts for hydrogenation reactions.
The structural and electronic similarities of the POValkoxide cluster to bulk MO x allow for valuable insight into surface reactivity of these materials. Although many studies have demonstrated the ability for MO x to facilitate PCET, 10,83-86 identication of individual active sites and elucidation of precise reaction mechanisms is challenging, in part due to constraints of spectroscopic techniques available for heterogeneous systems. Furthermore, theoretical analysis of PCET at MO x surfaces oen localizes reactivity to bridging hydroxide ligands, disregarding the possibility of the participation of aquo moieties in H-atom transfer reactions. In light of this, the results reported here provide insight into a novel pathway of Hatom transfer at the surface of redox-active metal oxides, whereupon coordination of water to an O-atom defect site, the O-H bonds of the substrate are activated. Coordination-induced bond weakening enables hydrogenation of the desired substrate. The ability to derive H-atom equivalents from water is a signicant nding, as commonly H 2 used in these molecular transformations is obtained through the combustion of fossil fuels. When taken together, this work provides valuable insight into novel mechanisms of H-atom transfer at the surface MO x compounds, revealing design criteria for MO x -derived hydrogenation catalysts.

General considerations
All manipulations were carried out in the absence of water and oxygen using standard Schlenk techniques or in a UniLab MBraun inert atmosphere dry-box under a dinitrogen atmosphere. All glassware was oven-dried for a minimum of 4 h and cooled in an evacuated antechamber prior to use in the dry-box. Solvents were dried and deoxygenated on a glass contour system (Pure Process Technology, LLC) and stored over 3Å molecular sieves purchased from Fisher Scientic and activated prior to use. TEMPO was purchased from Sigma-Aldrich and used as received. Hydrazobenzene was purchased from TCI and used as received. The POV-alkoxide cluster, (calix)V 6 O 6 , was prepared according to previously reported procedure. 44 5,10-dihydrophenazine 87 and 5,10-dideuterophenazine, 13 were generated following literature precedent. 1,4-Naphthalenediol was formed following similar procedures for the formation of 1,4-hydroquinone, where substitution of the respective organic substrate resulted in the formation of the desired reduced organic compound. 45 1 H NMR spectra were recorded at 400 MHz or 500 MHz on a Bruker DPX-400 or Bruker DPX-500 spectrometer, locked on the signal of deuterated solvents. All chemical shis were reported relative to the peak of the residual H signal in deuterated solvents. CDCl 3 and THF-d 8 was purchased from Cambridge Isotope Laboratories, degassed by three freeze-pump-thaw cycles, and stored over fully activated 3Å molecular sieves. Infrared (FT-IR, ATR) spectra were recorded on a Shimadzu IRAffinity-1 Fourier transform infrared spectrophotometer and are reported in wavenumbers (cm −1 ). Electronic absorption spectra were recorded at room temperature in anhydrous THF in a sealed 1 cm quartz cuvette with an Agilent Cary 60 UV-vis spectrophotometer.
A single crystal of (calix)V 6 O 5 (MeCN)(OH 2 )(OMe) 8 ((calix) V 6 O 5 (OH 2 )) was placed onto a thin glass optical ber or a nylon loop and mounted on a Rigaku XtaLAB Synergy-S Dualex diffractometer equipped with a HyPix-6000HE HPC area detector for data collection at 173.00(10) K. A preliminary set of cell constants and an orientation matrix were calculated from a small sampling of reections. 88 A short pre-experiment was run, from which an optimal data collection strategy was determined. The full data collection was carried out using a Photo-nJet (Cu) X-ray source with frame times of 3.37 and 13.49 seconds and a detector distance of 34.0 mm. Aer the intensity data were corrected for absorption, the nal cell constants were calculated from the xyz centroids of 17 304 strong reections from the actual data collection aer integration. 88 The structure was solved using SHELXT 89 and rened using SHELXL. 90 The space group Pbcm was determined based on systematic absences and intensity statistics. Most or all non H-atoms were assigned from the solution. Full-matrix least squares/difference Fourier cycles were performed which located any remaining non H-atoms. All non H-atoms were rened with anisotropic displacement parameters. The O-H H-atoms (one unique) on atom O4 were found from the difference Fourier map and rened freely. All other H-atoms were placed in ideal positions and rened as riding atoms with relative isotropic displacement parameters.
Synthesis of (calix)V 6 A 20 mL scintillation vial was charged with (calix)V 6 O 6 (0.026 g, 0.019 mmol) and 6 mL THF. Dihydrophenazine (0.005 g, 0.025 mmol) was added as a solid with stirring. The reaction was stirred at room temperature for 1 h, whereupon a colour change from brown to red was observed. The crude solid dried under reduced pressure and was washed with pentane (3 mL × 3) to remove the by-product, phenazine. The red solid was extracted in THF and was dried in vacuo. Recrystallization of the product occurred through slow diffusion of pentane into THF resulting in the isolation of red, starburst-shaped crystals of the product, (calix)V 6 O 5 (OH 2 ), suitable for single crystal X-ray diffraction (0.011 g, 0.008 mmol, 57%). 1  A 20 mL scintillation vial was charged with (calix)V 6 O 6 (0.035 g, 0.026 mmol) and 6 mL THF. Deuterated dihydrophenazine (0.005 g, 0.028 mmol) was added as a solid with stirring. The reaction was stirred at 25°C for 2 hours, whereupon a color change from brown to red was observed. The crude solid dried under reduced pressure and was washed with pentane (3 mL × 3) to remove phenazine, ltering over Celite. The red solid was extracted in THF and was dried in vacuo which resulted in the formation of (calix)V 6 O 5 (OD 2 ) (0.016 g, 0.012 mmol, 45%). To conrm deuterium transfer to the cluster, 2 eq of TEMPO was added to the cluster in THF and 2 H NMR was taken, and TEMPO-D was observed (1.93 ppm, s). 1  In an N 2 -lled glove box, 1 mL of a 0.598 mM stock solution of (calix)V 6 O 5 (OH 2 ) in THF and 2 mL of THF are added to a longnecked UV-vis cuvette. The cuvette is capped with a rubber septum and removed from the glove box, where a control electronic absorption spectrum is collected. In the glovebox, in a separate cuvette, 1 mL of the (calix)V 6 O 5 (OH 2 ) stock solution is added, along with 1.9 mL of THF, along with 0.1 mL of a stock solution containing 0.62 mM naphthoquinone in THF. The sample was shaken several times to ensure homogeneity and allowed to sit at 25°C for 12 hours, whereupon an electronic absorption spectrum was collected. This procedure was repeated for other equivalents of naphthoquinone. The extent of the reaction is determined through the absorbance measured at 900 nm, where the intervalence charge transfer band of (calix) V 6 O 6 allows for the determination of the concentration of each cluster. The concentration of 1,4-naphthalenediol and naphthoquinone are established by relative amounts of reduced and oxidized cluster present in solution, under the assumption that H-atom transfer occurs solely from the cluster to substrate (e.g. for each oxidized cluster formed, we assume the reduction of one molecule of naphthoquinone). Calculating the BDFE(O-H) of the reduced cluster in solution can then be performed through methods adapted from the Mayer group, 10 using eqn (4 General procedure for measuring kinetics of PCET at POV clusters through electronic absorption Each sample was monitored using electronic absorption spectroscopy, where the absorbance at 900 nm is used to determine the concentration of the oxidized cluster, (calix)V 6 O 6 , present in solution. All experiments are performed in 3 mL of THF at the desired temperature. An initial sample of 0.5 mM (calix) V 6 O 5 (OH 2 ) was prepared in THF in an air free quartz cuvette. A sample of stock solution containing TEMPO was collected in an air-free syringe and both cuvette and syringe are removed from the glove box. The cuvette is inserted into the UV-vis spectrophotometer and allowed to equilibrate to the desired temperature. Data collection begins as soon as the organic substrate is rapidly injected into the cuvette. Once the sample reaches equilibrium, and the absorbance of the sample is no longer changing, the data collection is stopped. The sample is then disposed of in the waste, and the cuvette is washed and dried in an oven at 125°C for 2 hours before being brought back into the glove box. The absorbance of the sample was plotted versus time, and k obs extrapolated from a linear trend line of time vs. ln((A − A inf )/A 0 − A inf ), where A t is absorbance at time = t, A inf is the absorbance of the sample once equilibrium is reached, A 0 is the initial absorbance of the sample, k obs is the observed rate constant in pseudo-rst order reaction conditions, and t is time in seconds. The slope of the line of the plot is equal to k obs which can then be found for each of the different concentrations of TEMPO. k PCET can then be found plotting a line of k obs vs.
[TEMPO] from 1 2 of the slope to account for the two chemically equivalent hydrogens on the aquo moiety.
General procedure for determining activation energy for the oxidation of (calix)V 6 O 5 (OH 2 ) by HAT Activation parameters were determined using kinetic data obtained by measuring the initial rate of formation of the product cluster, (calix)V 6 O 6 , over a range of temperatures (25 / 65°C). Pseudo rst-order reaction conditions were used to simplify nding the observed rate constants for each temperature, where the concentration of TEMPO was held in excess over the cluster. Rate constants were found by dividing the observed rate constant by the concentration of TEMPO to obtain the rate constant, k. From the results collected in the variable temperature experiments, the activation parameters are able to be established using the linear form of the Eyring-Polanyi equation shown in eqn (5).
where T is the temperature, DH ‡ is the enthalpy of activation, DS ‡ is the entropy of activation, R is the universal gas constant (R = 1.987 × 10 −3 kcal K −1 mol −1 ), k B is the Boltzmann constant, and h is Planck's constant. Plotting ln(k/T) vs. 1/T gives a plot with a linear best-t line, from which the enthalpy of activation can be found by slope = −DH ‡ /R. In addition, the entropy of activation can be found from the y-intercept, where yintercept = ln(k B /h) + DS ‡ /R. From these parameters, the activation free energy can be determined at the desired temperature using eqn (6). All values are reported at 95% condence interval by linear regression using excel.

Conflicts of interest
There are no conicts to declare.