M.
Dave‡
a,
A.
Rajagopal‡
a,
M.
Damm-Ruttensperger
a,
B.
Schwarz
a,
F.
Nägele
b,
L.
Daccache
b,
D.
Fantauzzi
*cd,
T.
Jacob
*bcd and
C.
Streb
*acd
aInstitute of Inorganic Chemistry I, Ulm University, Albert-Einstein-Allee 11, 89081 Ulm, Germany. E-mail: carsten.streb@uni-ulm.de
bInstitute of Electrochemistry, Ulm University, Albert-Einstein-Allee 47, 89081 Ulm, Germany. E-mail: timo.jacob@uni-ulm.de
cHelmholtz Institute Ulm for Electrochemical Energy Storage, 89081 Ulm, Germany
dKarlsruhe Institute of Technology, P.O. Box 3640, 76021 Karlsruhe, Germany. E-mail: donato.fantauzzi@kit.edu
First published on 1st March 2018
Molybdenum sulfides are highly active hydrogen evolution reaction (HER) catalysts based on earth abundant elements. Here, the molybdenum sulfide anion [Mo3S13]2− is used as a molecular model to rationalize HER reactivity of Mo–S-catalysts. For the first time, homogeneous, visible light-driven HER activity of [Mo3S13]2− is reported and high reactivity is observed (turnover number TON ∼23000, maximum turnover frequency TOFmax ∼156 min−1). Experimental and theoretical studies shed light on the catalytic role of terminal disulfide ligands (S22−) and show that these ligands modulate catalyst redox-activity and electron transfer in solution. Partial substitution of the terminal disulfides with water ligands leads to the most active catalytic species, e.g. [Mo3S11(H2O)2]. In contrast, complete substitution of the terminal disulfides results in a significant loss of reactivity. These results could lay the foundations for the knowledge-based development of homogeneous and heterogeneous molybdenum sulfide catalysts.
Fig. 1 Molecular molybdenum sulfides as models for amorphous MoS2+x catalysts. Top: simplified structure of amorphous MoS2+x based on disulfide-bridged [Mo3S13]2− units, as proposed by Artero, Tran et al.12 Bottom: Ball-and-stick representation of the molecular molybdenum sulfide prototypes [Mo3S13]2− (={Mo3}), [Mo3S7Cl6]2− (={Mo3}–Cl) and [Mo3S7Br6]2− (={Mo3}–Br). Color scheme: MoIV: teal; S: yellow, Cl: green, Br: brown. |
In this study, we build on these ground-breaking results and report the high visible light-driven HER activity of a molecular molybdenum sulfide together with experimental and theoretical studies to rationalize the high reactivity observed. Terminal disulfide ligands are identified as key species, which tune energetics of the hydrogen formation, control the redox-properties of the catalyst and modulate catalyst–photosensitizer interactions in solution. Exchange of the terminal disulfides with water or halide ligands results in significant changes in HER reactivity. These changes are rationalized using experimental and theoretical methods, providing unprecedented molecular-level insights into catalyst reactivity and deactivation.
Notably, the initial turnover frequency (TOF) in water (∼37 min−1, determined after tirrad = 5 min) decreases sharply to ∼12 min−1 (tirrad = 30 min), after which TOF values remain constant (Fig. 2b). This behaviour was interpreted as a structural change of the catalyst, leading to the formation of a less HER-active species. Structural considerations led us to suggest that the most likely deactivation mode is an exchange of the terminal disulfide ligands with solvent (i.e. water) ligands. This mode of ligand exchange is documented in the literature22,26,27 and leads to species such as [Mo3(S2,bridging)3S(H2Oterminal)6]4+, see Fig. 2d.26 In contrast, exchange of the bridging disulfide ligands on {Mo3} (or related thiomolybdates) is not known and would most likely lead to cluster decomposition and formation of colloidal molybdenum sulfide particles.14 This was not observed.
To support that exchange of the terminal disulfides occurs under reaction conditions, we performed in situ Raman spectroscopy on {Mo3} in the water-containing reaction solution. Raman spectroscopy is an ideal tool to examine structural changes in {Mo3} as the bridging and terminal disulfide ligands give two well-separated signals at 552 cm−1 (bridging) and 521 cm−1 (terminal), respectively (Fig. 2c).7 A mixture of MeOH and H2O (1:1, v/v) was used as solvent to reduce the water content of the solution and slow down ligand exchange so that spectroscopic measurements became possible. Further, higher catalyst concentrations (0.3 mM) were used to obtain detectable Raman signals. HER catalytic experiments verified that the system is still active under these conditions and indeed shows increased activity (Table 1, entry 2) compared with the purely aqueous system. As shown in Fig. 2c, with increasing reaction time, a decrease of the signal intensity for terminal disulfides is noted, indicating exchange of these ligands. In contrast, the signal associated with the bridging disulfides remains constant. The ligand exchange processes occur over a period of ∼1 h, which qualitatively corresponds to the period of time during which TOF decrease is observed (Fig. 2b). Note that quantitative correlation between both measurements is not directly possible due to different experimental conditions required for catalysis and Raman spectroscopic analyses. Further evidence of the proposed ligand exchange was provided by high resolution ESI mass spectrometry on the {Mo3} reaction solution, where deprotonated aquo species such as [Mo3S7O3]2− (exp: 280.76 m/z; calcd: 280.76 m/z) were detected, see ESI.† In sum, these experimental findings suggest that exchange of the terminal disulfide ligands by aquo ligands occurs under the given homogeneous HER conditions in the presence of water, leading to significant reactivity changes (see Fig. 2d for proposed exchange mechanism). Note that we also examined whether exchange of the terminal disulfides with MeOH ligands could occur, however, UV-vis spectroscopy showed no ligand exchange when {Mo3} is dissolved in pure MeOH (see ESI†).
No | Catalyst | Solvent | TONb/— | TOFmax/min−1 |
---|---|---|---|---|
a catalytic conditions: [catalyst] = 0.3 μM, [Ru(bpy)3]2+ = 20 μM, [ascorbic acid] = 0.1 M, light source, LED, λmax = 470 nm, P ∼ 40 mW cm−2. b Determined after tirrad = 6 h, standard deviation ± 3–4%. | ||||
1 | {Mo3} | H2O | 2850 | 37 |
2 | {Mo3} | MeOH:H2O (1:1, v/v) | 3950 | 54 |
3 | {Mo3} | MeOH:H2O (10:1, v/v) | 23000 | 156 |
4 | {Mo3}, tirrad = 24 h | MeOH:H2O (10:1, v/v) | 41000 | 156 |
5 | {Mo3} | MeOH | 8850 | 47 |
6 | {Mo3}–Cl | MeOH:H2O (10:1, v/v) | 500 | 1.0 |
7 | {Mo3}–Br | MeOH:H2O (10:1, v/v) | 60 | 0.7 |
No | Catalysta | Photosensitizer | Electron donor | TON/—([catalyst]) | TOFmax/min−1 | Reference |
---|---|---|---|---|---|---|
a Ligand abbreviations: bdt = 1,2-benzene dithiolate; PentPy = mono-hydroxylated pentapyridyl ligand; DPA-py = N,N-bis(2-pyridinylmethyl)-2,2′-bipyridine-6-methanamine; pyS = pyridine-2-thiolate, tacn = 1-[(4-CO2Et-3,5-H-2-pyridyl)methyl]-4,7-dimethyl-1,4,7-triazacyclononane. | ||||||
1 | [Co(bdt)2]− | [Ru(bpy)3]2+ | Ascorbic acid | 2700 (5 μM) | ∼15 | 28 |
2 | [CoBr(PentPy)]Br | [Ru(bpy)3]2+ | Ascorbic acid/P(C2H4COOH)3 × HCl | 33300 (1 μM) | ∼100 | 29 |
3 | [Fe2S2]-dendrimer | [Ir(ppy)2(bpy)]+ | Triethyl amine | 22200 (1 μM) | ∼120 | 30 |
4 | [Co(DPA-py) (OH2)]3+ | [Ru(bpy)3]2+ | Ascorbic acid | 4400 (0.5 μM) | ∼25 | 31 |
5 | [Ni(pyS)3]− | Fluorescein | Triethyl amine | 5500 (4 μM) | ∼5 | 32 |
6 | [CoII(OTf)2(tacn)] | [Ir(ppy)2(bpy)]+ | Triethyl amine | 9000 (0.25 μM) | ∼870 | 33 |
7 | [Mo3S13]2− | [Ru(bpy)3]2+ | Ascorbic acid | 41000 (0.3 μM) | 156 | This work |
Based on these insights, we hypothesized that lower HER reactivity can be expected for {Mo3} in the absence of aquo ligands. Thus, we performed HER catalysis under our standard conditions using pure MeOH as a solvent (Table 1, entry 5). This led to a significantly lower performance (TON ∼ 8,850, TOFmax ∼ 47 min−1) compared with the MeOH:H2O mixtures described above, thereby substantiating our hypothesis that exchange of terminal disulfides with aquo ligands can be used to optimize HER activity of {Mo3}.
To investigate the reactivity of the [Mo3(S2,bridging)3S]4+-core in the absence of any terminal disulfide ligands, we assessed the catalytic performance of the halide-terminated species {Mo3}–Cl and {Mo3}–Br (see Fig. 1) under our standard HER-catalytic conditions in MeOH:H2O (10:1, v/v). Both halide-substituted species show significantly lower reactivity based on TON and TOF values compared with {Mo3}, see Fig. 3a and Table 1, entries 6/7. This demonstrates that simple replacement of the terminal disulfides with other common anionic ligands does not lead to a stabilization of the HER-catalytic activity and emphasizes that the terminal disulfide ligands introduce specific structural and electronic effects, which contribute to the increased HER activity of {Mo3}, see below.
Since the disulfide-aquo ligand exchange is dependent on the underlying chemical equilibria shown in Fig. 2d, we suggested that using Le Chatelier's principle, it should be possible to slow down the full exchange of terminal disulfides. As an initial proof of principle, we performed the standard {Mo3} HER catalysis in MeOH:H2O (1:1, v/v) in the presence of ammonium polysulfide (NH4)2Sx, which is used as disulfide source in the synthesis of {Mo3}.14,21,22 Under these conditions, we note that catalyst deactivation proceeds at a slower rate compared to the polysulfide-free reference reaction (Fig. 3b, TON increase ∼30%), suggesting that polysulfides could improve the HER activity of {Mo3}.
To this end, free energies (ΔGs) of formation of intermediate states involved in the catalytic process were calculated, which provide information on the thermodynamic contributions to activation energies.34 Of course, a more rigorous analysis of the overall reaction kinetics would additionally require detailed investigations on the energetics of transition states.
In our studies special care was taken to define the reference energies of the solvated proton and electron to be in agreement with the experimental conditions.15,35,36 Careful referencing is important, as even small changes of the reference energies might perturb the free energy values. Details of the theoretical approach used and a full list of the pathways studied are given in the ESI.†
Our calculated reaction path analysis is in line with previous experimental reports on heterogenized {Mo3} electrocatalysts15 and suggests a Volmer–Heyrovsky-type mechanism:6,15 in step one, a proton-coupled electron transfer (Volmer step, Fig. 4a) is energetically favored and leads to a hydrogen atom bound at a bridging disulfide. Note that this step leads to a cleavage of the disulfide bond, resulting in the formation of a hydrogen sulfide (HS−) ligand, see Fig. 4e.37 In step 2 (Heyrovsky step, Fig. 4a), a second proton-coupled electron transfer to the catalyst is energetically favored, resulting in the formation and release of a H2 molecule and re-generation of the native catalyst. In addition to proton-coupled electron transfers, we also studied individual protonation or electron-transfer steps (Fig. 4b). In comparison to the proton-coupled electron transfers, these individual steps were always energetically non-favoured, see details in the ESI.†15,34
Fig. 4 Theoretical investigation of the HER catalytic reactivity of proposed catalytic intermediates. For each species, free energy changes during a Volmer–Heyrovsky-type H2 evolution was calculated. (a) Illustration of the Volmer–Heyrovsky mechanism: step 1 corresponds to the Volmer step, i.e. a proton-coupled electron transfer to the catalyst. Step 2 corresponds to the Heyrovsky step, i.e. a proton-coupled reductive hydrogen evolution. (b) Illustration of the H2 formation processes calculated, exemplified for [Mo3S13]2−. Note that proton-coupled electron transfers (diagonal lines) were the energetically favoured processes. Other pathways such as stepwise electron-proton transfer were energetically non-favoured and therefore unlikely (see ESI†). (c) Proton reduction free-energies for {Mo3} and {Mo3}-derivatives where one, two or three terminal disulfides have been replaced by two, four or six water ligands. H atom binding occurs at the bridging disulfide ligands. (d) Free energies for hydrogen binding by [Mo3S11], where a vacant coordination site is generated by loss of one terminal disulfide ligand. Here, the energies for H atom binding on a bridging disulfide, on the Mo (i.e. formation of a MoV hydride, MoV–H) and on the terminal disulfide were calculated. (e) Calculated binding modes for H-atom binding at bridging disulfide ligands in [Mo3S13]2− and at a Mo center (forming a MoV-hydride) in [Mo3S11]. Note that weakly stabilizing S⋯H hydrogen bonds37 (d(S⋯H) ∼2.6–2.8 Å) to the apical sulfide ligand are observed in both binding modes (dashed magenta lines). |
We then examined, how the exchange of terminal disulfides by aquo ligands on {Mo3} affects the hydrogen evolution energetics. To this end, we studied the species [Mo3S13−x(H2O)x](2−x)− with x = 0, 2, 4, and 6. As shown in Fig. 4c, the free energy of the Volmer step for native {Mo3} (7.5 kcal mol−1) is relatively high. Significantly lower free energies are observed when one or two disulfides are exchanged with water ligands, i.e. x = 2 (0.0 kcal mol−1) and x = 4 (−3.4 kcal mol−1). Exchange of all three terminal disulfides (x = 6) leads to the highest free energies of the Volmer step (11 kcal mol−1).
These findings are in line with the experimental results discussed above: in MeOH:H2O mixtures (Fig. 2b), we note a reproducible initial TOF increase (tirrad = 0–20 min), which indicates the conversion of {Mo3} into a catalytically more active species. This would be in line with the substitution of aquo ligands on {Mo3}, leading to the species [Mo3S11(H2O)2] and/or [Mo3S9(H2O)4]2+. The TOF decrease after tirrad = 20 min indicates that catalyst deactivation becomes the dominant process. This would be in line with the formation of [Mo3S7(H2O)6]4+ as main species, where the highest formation energies for the Volmer step (Fig. 4c) suggest low HER reactivity.
The calculations also shed light on the significantly faster deactivation of {Mo3} under aqueous conditions: the high water concentration results in faster exchange of terminal disulfides with aquo ligands and thus, faster formation of the less reactive species [Mo3S7(H2O)6]4+.
In the next step, we compared how complete substitution of terminal disulfides with chloride ligands affects the HER energetics: for {Mo3}–Cl we calculated a free energy of the Volmer step (12.1 kcal mol−1, see ESI†). This is in line with the experimentally observed low HER activity for {Mo3}–Cl (Table 1, entry 6). As a final step, we assessed whether molybdenum hydride (MoV–H) species could be relevant for hydrogen evolution by {Mo3}. This is a relevant question as recent studies on solid-state Mo–S HER electrocatalysts suggest that MoV–H species might be involved in the catalytic cycle.12,13 Here, we investigated the proton reduction by [Mo3S11], i.e. a molecular model for a Mo–S catalyst featuring vacant terminal molybdenum coordination sites (Fig. 4e). The species can be formed from {Mo3} by loss of one terminal disulfide (Fig. 4d). For this species, we also obtained a Volmer–Heyrovsky type mechanism. In the Volmer step, H atom binding at a bridging disulfide (−6.5 kcal mol−1) and at the vacant Mo center (−8.7 kcal mol−1, leading to a MoV–H) are energetically comparable (Fig. 4d). In contrast, H atom binding at the terminal disulfide leads to an energetically stabilized species where the subsequent Heyrovsky step would be significantly uphill (Fig. 4d). This finding suggests that for {Mo3} and related Mo–S catalysts, two HER pathways could be energetically possible, and the exact route might depend on the chosen reaction conditions. This could indeed help to consolidate the divergent reports on Mo–S HER catalysts where either hydrogen binding by bridging disulfides6–9 or MoV hydride formation12,13 are described as the key mechanistic steps.
In summary, theoretical analysis supports our experimental findings: partial exchange of terminal disulfides with aquo ligands leads to species such as [Mo3S7(H2O)x](2−x)- (x = 2, 4) where hydrogen evolution should be energetically favoured. In contrast, complete exchange of the terminal disulfides with aquo or chloride ligands (giving [Mo3S7(H2O)6]4+ or [Mo3S7Cl6]2−, respectively) results in species where HER is expected to be energetically more demanding. These findings are in line with recent studies on structurally related homogeneous10,38 and heterogeneous12 Mo–S electrocatalysts, where partial replacement of terminal disulfides with oxo ligands is suggested to give catalytically more active species.
Electrochemical studies were performed to assess the redox-activity of {Mo3}, {Mo3}–Cl and {Mo3}–Br under reductive conditions. To this end, linear sweep voltammetry was performed between E = 0 and −1.5 V (referenced against Fc+/Fc in de-gassed DMF containing 0.5 vol% H2O and 0.1 M nBu4NPF6 as electrolyte, [catalyst] = 1 mM). Under identical experimental conditions, the onset of reduction for {Mo3} is observed at less negative potentials (Eonset ∼ −1.2 V) compared to the halide-substituted species {Mo3}–Cl and {Mo3}–Br (Eonset ∼ −1.4 V), see Fig. 5a. This suggests that electron transfer to {Mo3} features favorable redox-potentials for the reductive electron transfer required to initiate HER catalysis. The observed overpotentials are a result of the given experimental conditions (low proton donor concentration, use of glassy carbon electrode).
The solution interactions between [Ru(bpy)3]2+ and {Mo3}, {Mo3}–Cl or {Mo3}–Br were explored in MeOH:H2O (10:1, v/v) using emission spectroscopy to analyze the quenching of the [Ru(bpy)3]2+-3MLCT emission depending on catalyst concentration. The resulting Stern–Volmer plots (Fig. 5b) show significantly higher quenching efficiency for {Mo3} compared to {Mo3}–Cl and {Mo3}–Br in the catalytic concentration range. Quenching is assigned to energy transfer between photosensitizer and catalyst, so that these initial data suggest higher solution interactions between {Mo3} and [Ru(bpy)3]2+ compared with the halide-substituted species.
Footnotes |
† Electronic supplementary information (ESI) available. CCDC 1533536 and 1533537. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c7se00599g |
‡ These authors contributed equally. |
This journal is © The Royal Society of Chemistry 2018 |