Octahedral Tantalum Bromide Clusters as Catalysts for Light-Driven Hydrogen Evolution

The development of an efficient hydrogen generation strategy from aqueous protons using sunlight is a current challenge aimed at the production of low-cost, easily accessible, renewable molecular hydrogen. For achieving this goal, non-noble metal containing and highly active catalysts for the hydrogen evolution reaction (HER) are desirable. Octahedral tantalum halide clusters {Ta6(μ-X)12}2+ (X = halogen) represent an emerging class of such HER photocatalysts. In this work, the photocatalytic properties of octahedral aqua tantalum bromide clusters toward HER and in acid and homogeneous aqueous conditions were investigated. The [{Ta6Bri12}Bra2(H2O)a4]·4H2O (i = inner ligand; a = apical ligand) compound is revealed to be an efficient precatalyst in acid (HBr) conditions and with methanol as the sacrificial agent. A response surface methodology (RSM) study was applied for the optimization of the HER conditions, considering the concentrations of both additives (methanol and HBr) as independent variables. An optimal H2 production of 11 mmol·g–1 (TON = 25) was achieved, which displays exceptional catalytic properties compared to regular Ta-based materials. The aqua tantalum bromide clusters assist in the photocatalytic hydrogen generation in agreement with energy-conversion schemes, and plausible active catalytic species and a reaction mechanism were proposed from computational and experimental perspectives.


■ INTRODUCTION
In the past few years, a renewed interest for group 5 and 6 octahedral halide-bridged clusters has revealed many outstanding properties of this family of compounds with prospects of applications in energy conversion, catalysis, radiology, and materials science. 1 These entities belong to the large family of metal atom clusters, 2 with {M 6 (μ 3 -X) 8 } (M = Mo, W) and {M 6 (μ 2 -X) 12 } (M = Nb, Ta; X = halogens) cluster units as robust central entities, coordinated by six labile terminal (apical) ligands.−6 Studies on the catalytic properties the octahedral halide-bridged clusters and their cluster hybrids derivatives have mostly been focused on the molybdenum and niobium clusters as heterogeneous catalysts at elevated temperatures. 7In the past few years, these clusters have attracted interest in the field of the conversion of solar energy into clean fuels thanks to their optical and redox properties. 8,9n fact, octahedral metal-cluster-based hybrid systems combine the advantages of both molecular catalysts and semiconductors, viz., high catalytic activity, broad visible-light absorption, and long-term stability of light harvesters.Molecular clusters based on {Mo 6 (μ 3 -X 8 )} 4+ (X = Br, I) cluster cores have been shown to be efficient catalysts in photoassisted water reduction. 3,10,11Their facile ligand and counterion exchange facilitates cluster immobilization onto graphene surfaces to build suitable nanohybrid nanostructured materials for H 2 production.Recently, the coordinative anchoring of octahedral clusters onto graphene oxide (GO) has been extended to {Ta 6 Br i 12 } clusters, and the GO-cluster nanohybrids have demonstrated to be efficient photocatalysts in HER. 12 Even though heterogeneous systems are expected to exhibit higher H 2 production rates and longer stability than homogeneous ones, the activity of the nonsupported {Ta 6 Br i 12 } 2+ clusters as photocatalytically reactive sites in HER and in homogeneous conditions remains unexplored, whereas Vogler and Kunkely reported noncatalytic photoredox reactivity toward water reduction. 13The study of these measurements were performed with a conventional three-electrode configuration consisting of a glassy carbon working and a platinum auxiliary electrode and an Ag/AgCl/KCl reference electrode.The solvent was H 2 O. KNO 3 (1 M) was used as a supporting electrolyte.The concentration of the cluster was less than 1mΜ.Redox potential values (E 1/2 ) were determined as (E a + E c )/2, where E a and E c are anodic and cathodic peak potentials, respectively.
Photochemical H 2 Evolution Experiments.All the manipulations were performed under an argon atmosphere and using Schlenk techniques.For the photoreactivity and photocatalytic reactions, the aqueous mixtures were previously deoxygenated by bubbling dry argon for at least half an hour.
The photoreactivity experiments performed for in situ/in operando UV−vis monitoring were carried out in an UV−vis quartz cuvette and under an argon atmosphere.A Hamamatsu Xe lamp was used as the radiation source with a spot light positioned perpendicularly at 1.5 cm from the cuvette wall.Measurements were monitored every 5 min during the first hour of reaction and then every 30 min.
The HER experiments directed to molecular hydrogen quantification were carried out under a rigorous inert atmosphere and according to the following general procedure: the chosen reactor was a cylindrical quartz reactor of 55 mL and 144 mm in diameter, equipped with a valve coupled to a manometer to determine the pressure into the reactor.Figure S1   O was dissolved in 15 mL of pure water or the chosen aqueous mixture to obtain solutions with a concentration of 5 × 10 −4 M. For optimizing the catalytic properties, 10, 5, and 1 mg of the cluster compound were employed.The system was then purged with argon bubbling, and it was pressurized to a pressure between 0.25−0.3bar.In order to ensure that the cluster fully dissolved, the reactor was subjected to ultrasound for 3 min.The reactor temperature was set at 25 °C by means of a cooling system, and the homogeneity of the solution remained stable with constant magnetic stirring.Next, the vessel was irradiated with a Hamamatsu Xe lamp with a spot light placed at a distance of 5 cm above the reactor surface.The gas phase samples (500 μL) were collected with a Hamilton syringe and injected to the GC-TCD spectrometer.The hydrogen peak area was calculated to the corresponding concentration using the standard calibration curve as reference.The micromoles of hydrogen produced were calculated taking into account the ideal gas law (n = PV/RT) and these quantifications were determined considering the [{Ta 6 Br i 12 }Br a 2 (H 2 O) a 4 ]•4H 2 O as a photocatalyst.Control experiments showed the detection of atmospheric gases, exclusively.The absorption spectra and pH of the catalytic solution were analyzed before and after reaction.MS-ESI was employed for the identification of plausible cluster species during the HER reaction in methanol/water mixtures.
In order to evaluate the materials' stability, reuse tests were carried out for four cycles under the same conditions as the initial experiments.The cluster material was recovered by precipitation with HBr and, after careful decantation, the resulting green solid was isolated and dried at 50 °C overnight.The H 2 produced is reported as percentage with respect to the value of hydrogen obtained in the first use.
Experimental Design and Data Analysis.The Design RStudio Software was used for the statistical design of experiments and data analysis. 18In this study, the Central Composite Design (CCD) and Response Surface Methodologies (RSM) 19−22 were applied in order to optimize the two variables, methanol and bromidic acid concentrations (mol•L −1 ), to obtain the highest hydrogen production (μmol•g −1 ) of the system reaction.The CCD consists in an experimental design, used in RSM in order to obtain a second order (quadratic) model for one response variable, namely, the hydrogen produced in a photochemical reactions under specific conditions, without the need to use a full three level factorial design. 21,22The design proposed in this research corresponds to a 3 2 full factorial design that involves four experiments (experiment numbers from 10 to 13) as replicates of the central point.With the aim to determine the range of HBr concentration, the acid from 0.1 mol•L −1 was added until precipitation at 2.0 mol•L −1 .As a result of this study, we chose 1.0 mol•L −1 as the central concentration point for HBr.In the case of methanol, the central point was set to 4.94 mol• L −1 .To prevent the effect of the selection bias, all the experiments were doing randomly.
In order to obtain the optimum concentration of HBr and MeOH (independent parameters), the hydrogen produced in the process was analyzed as a response variable.The quadratic equation model for the estimation of the optimal conditions was obtained according to the eq (1): where β 0 indicates the offset (intercept), β i is the linear coefficients, β ii is the pure quadratic coefficients, β ij the spurious quadratic coefficients, k the number of factors studied, and e the random error.
In the study, the analysis of variance (ANOVA) was used for the graphical analyses of the data in order to obtain the interactions between the response and the independents variables.The dimensional plot and its respective contour plot were obtained using the same program for the data treatment (R and Rstudio) and based on the effects of the two independent variables. 18omputational Details.The molecular geometries of the electronic ground state of tantalum bromide clusters [{Ta 6 Br 12 }- 3, 4) were fully optimized both in gas and solvated phase at the density functional theory (DFT) level using the hybrid exchange correlation functional B3LYP, 23−25 coupled with the triple−ζ Ahlrichs' Def2−TZVPP basis. 26,27The geometries and energies included in this work refer to calculations in the solvated phase.No empirical dispersion was included.The choice of functional and basis set was based on a previous investigation on similar compounds. 28The optimized geometries were subsequently submitted to frequency calculation (in harmonic approximation).No negative frequencies were found.Thermochemical quantities (at T = 298.15K and p = 1 atm) were computed at the same level of theory.The atomic charges were computed by means of two different approaches, viz., by Mulliken's population analysis and the atomic polar tensor (APT) derived charges. 29he excitation energy and oscillator strength first ten singlet and ten triplet vertically excited electronic states were computed at TD-DFT CAM-B3LYP Def2−TZVPP level; 30 the states' nature was contextually characterized.
The solvent (water) was treated implicitly, within the framework of the polarizable continuum model using the integral equation formalism variant (IEF-PCM). 31The values of static dielectric constant (ε) and refraction index (n 2 ) were taken from the literature. 32The cavitation radii were the standard UFF radii, scaled by a factor α = 1.1; the scale factor for the metal atoms was modified to check the consistency of the results.Both equilibrium (eq) and nonequilibrium (neq) regimes as well as for the excited electronic states linear response (LR) and state specific (SS) solvation approaches were employed.
Ground state geometry optimizations and frequency calculations were repeated at the DFT M06-2X/Def2-TZVPP level, including solvents' effects by means of the SMD model as this level is acknowledged as more accurate for energy predictions. 33or all the calculations, the integration grid for the electronic density was set to 250 radial shells and 974 angular points.Accuracy for the two−electron integrals and their derivatives was set at 10 −14 a.u.The self-consistent field (SCF) algorithm used was the quadratically convergent procedure designed by Bacskay, 34 a method which is acknowledged as slower but more reliable than regular SCF with DIIS extrapolation.The convergence criteria for SCF were set at 10 −12 for root-mean-square (RMS) change in the density matrix and at 10 −10 for maximum change in the density matrix.Convergence criteria for geometry optimizations were set at 2 × 10 −6 a.u.for maximum force, 1 × 10 −6 au for RMS force, 6 × 10 −6 au for maximum displacement, and 4 × 10 −6 au for RMS displacement.All calculations were performed using the GAUSSIAN G16.C01 package.S2).Vogler and Kunkely established that the proton reduction reaction is achieved by simultaneous oxidation of the complex, promoted upon light irradiation and in the presence of acid as a proton source (eq 2). 13,17Hydrobromic acid was chosen in order to prevent eventual halogen exchange within the {Ta 6 Br i 12 } cluster core under the experimental conditions.
Two irradiation experiments were conducted, with and without HBr of 1.0 mol•L −1 in aqueous solution, respectively.The progress of the UV−vis spectra was monitored in situ/in operando in a spectrometric cuvette, and the H 2 generated was quantified when the reaction took place in a photochemical reactor.The UV−vis monitoring showed evolution of the characteristic spectra of [{Ta 6 Br i 12 }(H 2 O) a 6 ] 2+ in acidic conditions during a 5 h irradiation, with significant changes of the bands between 400 and 900 nm (Figure 1a), similar to that reported in HCl solution. 13The new bands are associated with the in situ generation of 3+ oxidized cluster species.In the absence of the acid, the characteristic bands of [{Ta 6 Br i 12 } - (H 2 O) a 6 ] 2+ persisted upon a 4 h irradiation (Figure 1b).This highlights the fact that no photoredox reaction takes place in the absence of the acid.However, a progressive and slight decrease of the intensity of these bands with reaction time was noticed, which was associated with a low-yield decomposition or hydrolysis of the complex toward Ta 2 O 5 under experimental conditions.
GC results (Figure S3) confirmed the production of H 2 according to 2. A small amount of H 2 (3 μmol•gr −1 of [{Ta 6 Br i 12 }Br a 2 (H 2 O) a 4 ]•4H 2 O cluster) was obtained after 1 h of irradiation in HBr, and this production remained stable for at least 6 h (Figure S3), confirming that no hydrolysis of the cluster is achieved in acidic conditions.This low hydrogen amount is comparable to that described in the presence of HCl, even though in that case, the features of the lamp used and the reaction time were not specified, 13 and corresponds to a 1% cluster conversion to the oxidized {Ta 6 Br i 12 } 3+ cluster species, according to 2. The progressive increase of the new cluster absorption bands (Figure 1(a)), associated with the 3+ oxidized cluster species, is also related to the light-promoted generation of {Ta 6 Br i 12 } 4+ cluster core species.The associated absorption bands are not detected, whereas the generation of 3+ species is promoted by the comproportionation reaction involving 2+ and 4+ species, as depicted in 3. 13 We performed a cyclic voltammetry experiment in order to verify the reversibility of the oxidized species, and two consecutive and quasi-reversible one-electron transfer processes are detected (Figure S4), associated with a consecutive two-step oxidation of the aqueous {Ta 6 Br i 12 } 2+ to lead {Ta 6 Br i 12 } 3+ and {Ta 6 Br i 12 } 4+ species.The first oxidation potential appears at 0.378 V (ΔE = 66 mV), and the second at 0.669 V (ΔE = 65 mV), which are similar to the previously published data recorded in acidic conditions. 36The positioning of the frontier orbitals from the redox and NIR absorption properties of the 2+ and 3+ cluster species suggests that the proton to H 2 photoreduction is thermodynamically favorable for both species.In fact, the two-electron transfer from {Ta 6 Br i 12 } 2+ species to aqueous protons is energetically more favorable to produce H 2 and {Ta 6 Br i 12 } 4+ in a 1:1 stoichiometric ratio, in agreement to studies found in the literature. 13No HER was recorded in pure water (Figure S3), which confirms the need of protons for promoting the photoredox reaction, nor from an aqueous solution of the cluster (without acid), suggesting that the presence of small amounts of Ta 2 O 5 does not promote the H 2 production under light.
In a second stage, the hydrogen production promoted by [{Ta 6 Br i 12 }Br a 2 (H 2 O) a 4 ]•4H 2 O was optimized in the presence of sacrificial electron donors and was monitored by GC and absorption spectroscopy.There is a wide scope of additional sacrificial reagents that are also commonly used in H 2 generation from protons. 37,38Among them, we chose cheap and abundant alcohol, such as methanol, and acid representatives, as acetic, lactic, and ascorbic acids, which also play the role as proton donors, thus avoiding the use of HBr.Longer irradiation times (24 h) were achieved, and sacrificial additives were used in 20% v/v.Whereas the photochemical reaction in the presence of ascorbic acid did not produce H 2 , upon 3 h of irradiation, the amount of H 2 produced by the acetic and lactic acids (22 μmol•g −1 ) was independent of the acid used (Figure 2) and above the amount registered in the absence of any sacrificial electron donor agent.The yield of H 2 obtained rose considerably on going from acetic acid (45 μmol•g −1 ) to lactic acid (215 μmol•g −1 ) at the end of the reaction (Figure 2), which indicates that the most efficient reaction corresponds to the lowest pK a of lactic acid.At this point, the absorption spectra showed that the cluster remained practically stable throughout the runs and revealed the presence of cluster species only in its reduced {Ta 6 Br i 12 } 2+ form, as expected for an ideal catalytic cycle (Figure S5a,b).Cluster hydrolysis was prevented by the use of lactic acid, and the pH remained unaltered during the reaction (pH = 0.68), whereas a slight cluster decomposition detected in the presence of acetic acid may be ascribed to the less acidic conditions (pH 1.27 and 1.29 before and after the reaction, respectively).
Methanol was proposed as an alternative sacrificial.Its use is attractive because of the well-known electron donating ability and its easy removal (thanks to its high volatility).−41 Due to the low pK a of methanol, it was also employed in combination with strong acids, such as HBr and H 3 PO 4 .Using methanol/H 3 PO 4 and MeOH/HBr mixtures, higher H 2 amounts (420 and 426 μmol•g −1 , respectively) were obtained after 24 h, and the fastest production rate was recorded in the presence of HBr during the first 3 h of irradiation (Figure 2).These H 2 yields are 2-fold higher than that obtained with lactic acid.As a control test, the activity recorded with methanol as the sole additive dropped to 61 μmol•g −1 (Figure 2).The UV− vis spectra (Figure S5c,d) established that the cluster species decomposed slightly in pure methanol, whereas they remained intact in the presence of the methanol/HBr mixture.The need for the acid conditions to preserve the cluster stability and promote hydrogen generation and is again confirmed.Surprisingly, the evolution of the reaction mixture with phosphoric acid toward oxidation and cluster decomposition was detected by UV−vis (Figure S5e) after exposure to air.The use of MeOH/HBr mixtures was proposed as indoneous in order to achieve the largest hydrogen production and to preserve tantalum cluster species.Control experiments confirmed the need for light, the cluster, and the additives to achieve these results (Table S1).In order to optimize the performance of the Ta-cluster/MeOH/HBr system, an appropriate experimental study was designed following response surface methodology (RSM) via Central Compose Design (CCD).HBr and MeOH concentrations were chosen as independent variables.The CCD matrix was obtained with 13 randomized experiments, and the results and predicted values are presented in Table 1.The maximum hydrogen production achieved was about 428 μmol•g −1 for the conditions defined in the experimental design, with a percentage error less than 5% in most cases.It is worth noting that the percentage error in the calculated values changes for HBr concentrations higher than 1.5 M due to the precipitation of the cluster and the modification of the reaction system since it is no longer a homogeneous reaction, as confirmed by the absorption spectra recorded under these conditions (Figure S6a).The increase in error was also attributed to the highest MeOH concentrations (≥7.41 mol•L −1 ).Absorption spectra showed that the cluster evolved toward oxidation and cluster decomposition in the presence of excess of alcohol in alcohol/ acid mixtures (Figure S6b,c), which thus prevents its role as the active species.
Based on the data analysis (Table S1), an empirical quadratic equation was proposed for the H 2 production using the tantalum bromide cluster as the active species (eq.4).where Y indicates the H 2 amount (μmol•g −1 ), A is the methanol concentration (v/v), and B is the HBr concentration (mol•L −1 ).The ANOVA analysis of the data (Table S1), calculated p value (p = 0.00005 < 0.005) and the lack of a fit value (0.003 < 0.05) led us to confirm that the proposed equation matches with the experimental data.The effect of variables A B on the hydrogen production (Y) is illustrated in Figure 3, which shows that the maximum value for Y is achieved when the MeOH and HBr concentrations are 4.83 and 0.7 mol•L −1 , respectively (optimal conditions).In addition, the effect of the acid concentration on variable Y is more marked due to cluster precipitation caused by an excess of HBr.
The optimal conditions were applied for an additional experiment, and the hydrogen quantity obtained was 442 μmol•g −1 , which is within the experimental error (3.6%).The tantalum cluster-based material was recovered as a green solid, and four recycling experiments were carried out following the optimal conditions described above.The cluster was stable after two reuses, but its performance in terms of activity halved after the third reuse (Figure 4a), probably due to the cluster decomposition associated with an aging of the recovered cluster after long exposure to irradiation.In an additional experiment, we monitored the HER activity during 3 days of illumination and observed that the activity decreases progressively under reaction conditions (Figure 4b).Then, the color of the reaction mixture progressively bleached, and a white precipitate appeared at the end of the reaction, associated with the formation of Ta 2 O 5 by cluster hydrolysis.The coded values (reported in brackets) for HBr (A) and methanol (B) concentration were set at five levels: − 1 (minimum), − 0.5 (minimal star point), 0 (central), + 0.5 (maximal star point), and +1 (maximum).These observations support the loss of stability of the in situ generated tantalum cluster active species in the reaction media.Whereas Ta 2 O 5 is a recognized as a photocatalyst, generally combined with noble metal cocatalysts, 42 we can infer from recycling and long-irradiation experiments that the Ta 2 O 5 produced does not contribute in the increase of the H 2 production.
The efficiency of the activity of [{Ta 6 Br i 12 }Br a 2 (H 2 O) a 4 ]• 4H 2 O (18 μmol•g −1 •h −1 ) remained constant even when less quantity of cluster compound is present (10 vs 18 mg), and in both cases, TON = 1 agrees with an stoichiometric transformation.The catalytic performance of the Ta 6 material was assessed by decreasing the amount of the cluster inverted during the reaction.Thus, we repeated experiments with 5 and 1 mg of cluster loading and obtained productions of 1.1 (TON = 3) and 11.0 mmol•g −1 (TON = 25), respectively.The catalytic performance of the HER achieved in the presence of the minimum cluster amount corresponds to 442 μmol•g −1 •h −1 (TOF = 3 × 10 −4 s −1 per cluster molecule).This value was 126-fold higher than that reported in the photoreduction of water vapor using the derived hybrid material {Ta 6 Br i 12 }@ GO 12 and 1 order of magnitude higher than the activities reported for tantalum oxides, oxynitrides, and nitrides in heterogeneous conditions, such as MTaO 3 (M = Li, Na, K, Mg), BaTa 2 O 6 , 43 Ta 2 O 5 , 44 TaON, 45 and Ta 3 N 5 nanoparticles. 46It is worth noting that the optimal catalytic performance of the Ta cluster improves the activities reported for tantalum solids in the presence of Pt as a cocatalyst and with more powered irradiation lamps.This is an additional advantageous feature from the point of view of sustainable chemistry, since the use of noble cocatalysts is avoided.S7) conditions and the recyclability experiments.
The kinetics of the catalytic reaction during the first 2−4 h was studied, when using 1, 5, and 18 mg of cluster compound (Figure 5a).With the minimum cluster loading, the hydrogen  intermediates involved in the photoredox reactions studied in this work.
The reaction mechanism proposed for the photoredox transformation was revisited via computational and experimental approaches and involving the aqua-hydroxo Ta 6 Br i 12 cluster core complexes, as the most plausible species involved in the catalytic transformation. 13In a first step, upon the photoexcitation of the {Ta 6 Br i 12 } 2+ cluster species (5), a deactivation (6) may follow competing with a two-electron oxidation of {Ta 6 Br i 12 } 2+ into {Ta 6 Br i 12 } 4+ (eq 7).Photoluminescence measurements of the catalytic reaction mixture were registered and showed no cluster emission within the visible and near-infrared (NIR) window (Figure S11), which confirms a fast photoinduced electron transfer process.

Figure 3 .
Figure 3. Surface (a) and contour (b) graphs representing the effect of MeOH and HBr concentrations on the hydrogen production.

Figure 4 .
Figure 4. (a) Recycling of [{Ta 6 Br i 12 }Br a 2 (H 2 O) a 4 ]•4H 2 O under optimal reaction conditions; (b) representation of the H 2 evolution from the reaction mixture exposed to long irradiation time.
illustrates the experimental layout for the experiments.For the optimization tests, 7.7 μmol of cluster (18 mg of [{Ta 6 Br i 2 35

Table 1 .
CCD, Predictive Values, and Experimental Results.a The catalytic activity of the [{Ta 6 Br i 12 }Br a 2 (H 2 O) a 4 ]•4H 2 O cluster is in the same order than that achieved in homogeneous conditions in the presence of octahedral {Mo 6 Br i 8 } 4+ cluster core catalysts (641 μmol•g −1 •h −1 for (Et 4 N) 2 [{Mo 6 Br i 8 } - F a 6 ] 10 ), with the advantage that the [{Ta 6 Br i 12 }Br a 2 (H 2 O) a 4 ]• 4H 2 O material stands out for its robustness in solution, as shown by the UV−vis spectra registered after reaction (Figure