Iron–thiolate complexes of the type [Fe2(μ-bdt)(CO)6−xP(OMe3)x] (bdt=S2C6H4=benzenedithiolate, x≤2) are simplified models of iron–iron hydrogenase enzymes. Recently, we have shown that these water-insoluble organometallic complexes, when included into micelles formed by sodium dodecyl sulfate (SDS), are good catalysts for the electrochemical production of hydrogen in aqueous solutions at pH<6. We herein report that the all-CO derivative [Fe2(μ-bdt)(CO)6] (1), owing to its comparatively low reduction potential, is also a robust molecular catalyst for visible-light-driven production of H2 in aqueous SDS solutions at pH 10.5. Irradiation at λ=455 nm of a system consisting of complex 1, Eosin Y as a sensitizer, and triethylamine as an electron donor produced up to 0.86 mL of H2 in 4.5 h, corresponding to a turnover number of 117 mol of H2 per mol of catalyst. In the presence of a large excess of sensitizer, the production of H2 lasted for more than 30 h, stressing the relative stability of complex 1 under the photocatalytic conditions used herein. Thermodynamic considerations and UV/Vis spectroscopy experiments suggest that the catalytic cycle begins with the photo-driven reduction of complex 1. The reduced intermediate reacts with a proton source to yield iron hydride. Subsequent reduction and protonation steps produce H2, regenerating the starting complex. As a result, the iron–thiolate complex 1 is a versatile proton reduction catalyst that can utilize either solar or electrical energy inputs, providing a starting point for the construction of noble metal-free molecular systems for renewable H2 production.
Splitting water with sunlight to produce H2 is thought to provide an effective means of storing solar energy.1–4 Numerous photocatalytic systems, consisting of a proton-reduction catalyst, a sensitizer, a sacrificial electron donor, and a proton source, have been described, but most of them used catalysts and/or sensitizers based on platinum-group metals (PGMs),5, 6 which limit practical use. During the last decade, electro- and photocatalytic production of H2 has been revisited using molecular catalysts formed from earth-abundant materials, guided by the idea that they will in the future compete favorably with PGM-based catalysts.7–10 Catalysis of proton reduction by cobalt11 and nickel12, 13 complexes have been extensively studied. Examples of molybdenum- and iron-based catalysts have also been reported.14–16 Significant advances have been realized with iron–thiolate complexes of the type [Fe2(μ-SRS)(CO)6−xLx] (R=organic group, L=electron-donor ligand, x≤4),17–25 which are simplified models of the Fe2S2 subunit of iron–iron hydrogenase enzymes.26 The utilization of these iron–thiolate complexes for visible light-driven H2 production has been recently reviewed.27, 28 In most cases, yields in the range of ≤4 mol of H2 per mol of catalyst (hereafter referred to as turnover number or TON) have been reported in organic solvents or organic solvent/water mixtures. A significantly higher TON of 200 has been reported for an iron–thiolate catalyst bearing an electron-accepting bridging ligand used in combination with a tenfold excess of ruthenium-based sensitizer and ascorbic acid as a proton and electron donor in DMF/water (1:1 v/v), but H2 production stopped rapidly.29 Decomposition of most iron–thiolate complexes has been established to occur within hours upon irradiation in organic solvents.30 Recently, a system consisting of an iron–thiolate catalyst protected by a dendritic framework, an excess of an iridium-based sensitizer, and triethylamine in acetone/water (9:1 v/v) has been shown to produce H2 at an initial rate exceeding 2 s−1 during the first hour of irradiation, achieving a TON of 22 000 in 8 h.31 Although the observation of such a high rate of H2 production was encouraging with regard to the intrinsic activity of iron–thiolate catalysts, utilization of a large excess of expensive PGM-based sensitizers was required. To overcome this issue, Beller et al. have developed a PGM-free sensitizer based on a copper(I) complex with polypyridine ligands. Its utilization in combination with Fe3(CO)12 in a mixture of THF, water, and triethylamine (4:1:1 v/v/v) afforded about 480 mol of H2 per mol of Cu in 24 h of irradiation with visible light.32 An alternate approach has been proposed by Rauchfuss et al., demonstrating that the direct irradiation of an unsymmetrically substituted iron-hydride complex in the presence of triflic acid and ferrocene derivative in dichloromethane afforded H2; however, the TON was only 4.33
There are few examples of iron–thiolate-based photocatalytic systems for the production of H2 in pure water. A photocathode was assembled by immobilization of [Fe2(μ-S2)(CO)6] within an array of InP nanocrystals on a gold electrode, producing H2 on the nanomol scale.34 Homogeneous photocatalytic systems have been reported to exhibit a higher activity. A system achieving a TON of 500 at wavelengths >400 nm was built from a water-soluble iron–thiolate catalyst bearing hydrophilic ether chains, CdTe quantum dots as a sensitizer, and ascorbic acid as a proton source and electron donor.35 Using a supramolecular approach, a sulfonate-functionalized diiron–thiolate catalyst, and a xanthene dye [i.e., Eosin Y (EY2−) or Rose Bengal (RB2−)] were included into the hydrophobic cavity of cyclodextrin, increasing substantially the lifetime of the resulting photocatalytic system. TONs of up to 75 were achieved in an aqueous solution of triethylamine at pH 10 upon irradiation at >450 nm during a period of 24 h.36 The prevailing catalytic mechanism invokes the photodriven reduction of the iron–thiolate catalyst to form a putative Fe0FeI species, which reacts with a proton source to yield iron hydride. Subsequent reduction and protonation steps produce H2, regenerating the starting complex. Accordingly, it is anticipated that both a large driving force for electron transfer to the catalyst and a long lifetime of the Fe0FeI-reduced intermediate are essential factors to ensure an efficient production of H2.
In the search for a functional mimic of the active site of iron–iron hydrogenases, we have placed specific focus on rigid and unsaturated bridging ligands, which stabilize the Fe0FeI and Fe0Fe0 oxidation states of iron–thiolate complexes.37–39 This approach has been well illustrated by [Fe2(μ-bdt)(CO)6] (1; bdt=benzenedithiolate; Scheme 1), which is an easily synthesized proton-reduction catalyst combining reversible reductive electrochemistry and good activity at mild potentials in organic solvents: E1/2=−1.30 V versus ferrocenium/ferrocene (Fc+/0) in MeCN.40–42 Recently, we have demonstrated that complex 1 was still reduced at mild potentials in aqueous solutions [E1/2=−0.74 V versus the standard hydrogen electrode (SHE) at pH 7], when dispersed into micelles formed by sodium dodecyl sulfate (SDS).43, 44 At the same time, the catalytic efficiency for electrochemical proton reduction at pH<6 was significantly increased relative to that in organic solvents.
Encouraged by these results, we reasoned that complex 1 included in SDS micelles could potentially be an efficient proton-reduction catalyst for photocatalytic H2 production in water. Herein we report the performance of a PGM-free system consisting of complex 1, EY2− as a sensitizer, and triethylamine (Et3N) as a sacrificial electron donor in an aqueous SDS solution. To gain mechanistic understanding of the photocatalytic H2 production, we compared the TON values obtained upon replacing the all-CO complex 1 with substituted derivatives, that is, [Fe2(μ-bdt)(CO)5P(OMe)3] (2) and [Fe2(μ-bdt)(CO)4(P(OMe)3)2] (3) (Scheme 1), which are reduced at more negative potentials.
Results and Discussion
Photocatalytic H2 production experiments were carried out in aqueous solutions containing complex 1 (0.1 mm), EY2− (0.2 mm), Et3N (10 % v/v), and SDS (10 mm). The pH value was adjusted to 10.5 by adding HCl, and the solution was purged with argon. The amount of H2 formed upon irradiation at λ=455 nm with a light-emitting diode (LED, 0.3 W) was quantified by GC analysis using Ar as an internal reference (see the Supporting Information, Figure S1). Under these conditions, up to 0.86 mL of H2 was produced in 4.5 h, corresponding to a TON of 117 (Figure 1, Table 1). The average value of the initial H2 production rate was up to 1.96×10−7 mol min−1 (Figure 2). Additionally, a rate of photon absorption of 1.73×10−5 mol min−1 was calculated by performing chemical actinometry analysis (see the SI, Figure S3), yielding an apparent quantum yield of about 1.1 %. With =0.05 mM, the initial rate of H2 production was 0.2 mL h−1 (Figure 2), corresponding to a turnover frequency of 1.1 min−1 (see the SI, Figure S2). The PGM-free system 1/EY2−/Et3N in an aqueous SDS solution at pH 10.5 is, therefore, competitive with most of the systems working in water and using iron-,36, 45, 46 nickel-,47 or cobalt-based48, 49 catalysts. We anticipate that the comparatively high H2 production yield is the result of the mild reduction potential and the stability of the reduced form of complex 1 in aqueous SDS solutions (vide infra).
Table 1. Photocatalytic activity for H2 production in aqueous micellar solutions from systems consisting of complexes 1, 2, or 3 as a catalyst, EY2− or RB2<M-> as a photosensitizer (PS), and Et3N as an electron donor.
[a] All values are the average of at least two independent experiments. Experimental conditions: [catalyst]=0.1 mM; [Et3N]=10 % v/v; [SDS]=10 mM, Ar-purged aqueous solution (3.2 mL) at pH 10.5; irradiation at λ=455 nm using a LED (0.3 W).
No H2 was detected in control experiments performed in the absence of light or any of the three components (i.e., complex 1, EY2−, or Et3N), confirming the photocatalytic production of H2. Moreover, the performance of the system was not significantly altered upon addition of a large excess of Hg in solution, providing evidence that photocatalysis proceeded homogeneously.50, 51
To verify the influence of SDS micelles on the H2 production rate, we carried out additional experiments in a mixture of ethanol and water (1:1 v/v), in which complex 1 is slightly soluble. Under otherwise similar conditions, TONs achieved in the absence of SDS were about four times lower, stressing the beneficial effect of the inclusion of complex 1 into SDS micelles.
Experiments carried out in the Et3N concentration range of 1–10 % v/v and the pH value range of 9.5–11.0 showed that maximum TON was achieved at a concentration of 10 % v/v for Et3N and at pH 10.5 (see the SI, Figures S4 and S5). No detectable amount of H2 was produced at pH 7.7 with triethanolamine (10 % v/v) as electron donor.
The addition of excess amounts of sensitizer improved significantly the lifetime of the overall photocatalytic system. The production of H2 stopped after about 5 h in the presence of 2 equiv of EY2−, whereas it lasted for more than 30 h in the presence of 20 equiv of EY2− (Figure 3, Table 1). H2 production was, therefore, mainly limited by the decomposition of EY2−, a problem that has already been pointed out in previous reports.36 Unexpectedly, the maximum TON achieved in the presence of a large excess of EY2− was about half that with 2 equiv of EY2− under otherwise similar conditions (Table 1). We explain this result by the low power (0.3 W) of the LED used for irradiation, which was not able to illuminate the entire volume of the vial at high concentrations of dyes ([EY2−]=2 mM). On the other hand, decreasing the concentration of complex 1 by a factor of two had little effect on the initial rate of H2 production (Figure 2), confirming that the photocatalytic process was not limited by the intrinsic activity of the iron–thiolate catalyst. Due to the weak solubility of complex 1 in aqueous SDS solutions,43 its concentration could not be increased much more than 0.1 mM.
As stated above, the catalytic cycle for H2 production is thought to begin by the photodriven reduction of the iron–thiolate catalyst (Scheme 2). Previous work on xanthene dyes EY2− and RB2− has shown that reductive electron transfer occurs preferentially from the long-lived triplet-state species, which is formed by intersystem crossing (ISC).52 The reversible potentials for the couples EY−/3*EY2− and RB−/3*RB2− were estimated to be −0.87 and −0.78 V versus SHE at pH 7.53 These potentials are both more negative than the reduction potential of complex 1 in aqueous SDS solutions (E1/2=−0.74 V vs. SHE at pH 7),43 indicating that the electron transfer from the transient species 3*EY2− and 3*RB2− to complex 1 is thermodynamically favorable. The system 1/RB2−/Et3N produced H2, but at a slower initial rate and with a TON that was half that of the system 1/EY2−/Et3N (Table 1). The measured decrease of activity agrees well with our estimation of the driving force for electron transfer to complex 1 in the respective systems (0.04 V with 3*RB2− vs. 0.13 V with 3*EY2−). To confirm the results, we performed experiments with the P(OMe)3-substituted complexes 2 and 3, of which the reduction potentials are about 0.15 and 0.3 V more negative than that of complex 1. As expected, the H2 production activity decreased significantly for the system 2/EY2−/Et3N, achieving a TON of 4 in 4.5 h (Table 1). Moreover, only a small amount of H2 was produced by the system 3/EY2−/Et3N (TON<2 in 4.5 h) and no detectable amount of H2 was produced by 3/RB2−/Et3N. In addition, no H2 production was detected with Eosin B (EB2−) used in combination with complex 1 and Et3N at pH 10.5, most certainly because the potential of the couple EB−/3*EB2− is significantly less negative than that of the couple EY−/3*EY2−. Taken all together, the results show that the H2 production activity of the systems depends directly on the amplitude of the driving force for electron transfer from the triplet state of the xanthene dye to the iron–thiolate catalyst.
Recently, Sun et al. have described a system using [Fe2(μ-pdt)(CO)5P(CH2OH)3] (pdt=S2C3H6=propanedithiolate) as a proton-reduction catalyst in combination with EY2− and Et3N in ethanol/water at pH 10.46 The reduction potential of the propanedithiolate complex was about −1.2 V versus SHE, a potential much more negative than that of EY−/3*EY2−. Therefore, the catalytic mechanism was proposed to begin with the reduction of [Fe2(μ-pdt)(CO)5P(CH2OH)3] by a neutral alkyl radical formed from Et3N after reductive quenching of 3*EY2−. Herein, considering the effect of the driving force for electron transfer on the H2 production activity established above, we infer that the reduced form of the benzenedithiolate complex 1 is formed through direct electron transfer from 3*EY2− because it is a long-lived species and the reaction is thermodynamically favorable by 0.13 V.
To gain further understanding of the photodriven reduction of the iron–thiolate catalysts, we measured UV/Vis spectra of the system 1/EY2−/Et3N in Ar-purged aqueous SDS solutions at pH 12, a value for which we have shown that there is no H2 production (see the SI, Figure S5). Before irradiation, the absorption spectrum was simply the sum of the individual components: absorption at around λ=335 nm for complex 1 and around λ=520 nm for EY2− (Figure 4 A). After 1 min of irradiation at λ=455 nm, the band associated with complex 1 disappeared, whereas a new band appeared at about λ=570 nm. Moreover, bleaching of the absorption of EY2− occurred. According to literature,19, 54 the results strongly support the formation of a Fe0FeI species through a photoinduced reduction of complex 1. The decay of the absorption band at λ=570 nm was less than 20 % in 2 h (see the SI, Figure S6), confirming the stability of the reduced form of 1. In contrast, only weak signals were detected in the range 550–600 nm after irradiation of systems 2 and 3/EY2−/Et3N for 1 min at pH 12 (Figure 4 B), indicating a slow or hindered formation of Fe0FeI species. Consequently, the different rates of formation of the reduced form of iron–thiolate catalysts at pH 12 correlate with photocatalytic activity for H2 production at pH 10.5.
The second step in the proposed catalytic cycle is the reaction of the reduced Fe0FeI with a proton source to yield HFeIIFeI (Scheme 2). This transient species is a key intermediate in the electro- and photocatalytic H2 production by iron–thiolate complexes.41, 46 In a previous study, we have shown that proton-reduction electrocatalysis by complex 1 was almost suppressed at pH>6,43 suggesting that protonation of the reduced form of complex 1 was slow or hindered at these pH values. The observation of photocatalytic H2 production at pH 10.5 in aqueous Et3N solutions is, therefore, puzzling. We determined that the pH value of the solution remained unchanged in the course of experiments even after prolonged irradiation times. In addition, no photocatalytic activity was observed at pH 7.7 in the presence of triethanolamine, whereas UV/Vis spectra of the system 1/EY2−/triethanolamine confirmed the formation of reduced Fe0FeI species after irradiation (see the SI, Figure S7). Although proton release by the oxidized form of Et3N has been reported,55 we do not have at present any satisfactory assumptions to explain the mechanism by which irradiation of the system 1/EY2−/Et3N in aqueous SDS solutions at pH 10.5 induces protonation of the reduced form of complex 1.
Further protonation of the HFeIIFeI intermediate is an unfavorable reaction under the photocatalytic conditions used herein (pH 10.5).56 To close the photocatalytic cycle, H2 production has to occur either through reduction and protonation of the HFeIIFeI intermediate, which is more easily reduced than the starting FeIFeI complex,56 or through bimolecular combination of two iron hydride molecules. Although the bimolecular pathway cannot be excluded, we favor the mechanism involving the reduction and protonation of the iron hydride intermediate because the initial rate of H2 production is not very sensitive to the concentration of catalyst for ≤0.1 mM (Figure 2).
The iron–iron hydrogenase model [Fe2(μ-bdt)(CO)6] (1) is a robust molecular photocatalyst for H2 production in aqueous SDS solutions at pH 10.5, achieving TONs of up to 117 when used in combination with EY2− as a sensitizer and Et3N as an electron donor. The lifetime of this PGM-free system was clearly limited by the stability of EY2− under photocatalytic conditions. Moreover, the initial rate of H2 production (1.1 min−1 mol H2 per mol of catalyst) was not very sensitive to the concentration of complex 1, indicating that the intrinsic activity of this iron–thiolate catalyst is not limiting. More studies are needed to further reveal the details of the catalytic mechanism, in particular the nature of the proton source that reacts with the Fe0FeI reduced species at pH 10.5 and, therefore, the rate-determining step in the catalytic cycle. Importantly, complex 1 can photocatalyze proton reduction at basic pH values, which is also the most favorable condition to accelerate the catalysis of water oxidation. These results present a starting point for the construction of a PGM-free molecular device for renewable production of H2 from sunlight and water. Current lines of investigations include: (i) the utilization of a more powerful light source (i.e., an artificial sun) and a photoreactor with optimized geometry and (ii) the association of complex 1 with a semiconductor-based sensitizer to increase the lifetime and efficiency of the photocatalytic system.
Iron–thiolate complexes 1, 2, and 3 were synthesized as previously described.44 Et3N, EY2−, RB2−, and EB2− (Sigma–Aldrich), and K3[Fe(C2O4)3]⋅3 H2O (Strem) were used as received. All the solutions were prepared from deionized water (Milli-Q) and analytical grade chemicals.
The precise volume of the vial used as a photoreactor was determined from its weight when fully filled with water. A total volume of around 4.9 mL was usually measured. The pH value of the solutions to be irradiated was measured by using a pH meter (Mettler Toledo). Irradiation was carried out using a blue LED (Thorlabs M455L2, λ=455 nm). Its power under experimental conditions was measured with a power meter (Thorlabs PM100A) equipped with an integrating sphere (photodiode power sensor S142C). The gas in the headspace of the vial was analyzed by GC (Varian 3900 equipped with a thermal conductivity detector). A sample volume of 50 μL was injected under N2 gas flow into a molecular sieve column (Porapak Q80/100, 2 m×1/8 in; ca. 0.05 cm) heated at 50 °C. Absorption spectra were measured under an Ar atmosphere in quartz cells (optical path length of 1 cm) using an UV/Vis spectrophotometer (Jasco V-670).
In a typical experiment, a vial (4.9 mL) was filled with an aqueous solution of Et3N (3 mL, 10 % v/v) acidified to pH 10.5 with HCl (concentrated). Aqueous SDS (66 μL, 0.5 M), iron–thiolate complex (50 μL, 6 mM in MeOH), and EY2− (100 μL, 6 mM in MeOH) were added. The vial was capped with a silicone/ polytetrafluoroethylene (PTFE) septum, purged under a flow of Ar (30 min), and placed at a distance of 1 cm from a LED (0.3 W). Irradiation was performed at λ=455 nm under stirring. The amount of H2 in the headspace of the vial was measured by GC analysis using Ar as an internal standard (see the SI, Figure S1). The quantum yield was calculated using ferrioxalate as a chemical actinometer (see the SI, Figure S3).
This work was supported by Agence Nationale de la Recherche (ANR, BLANC SIMI9/0926-01, “TechBioPhyp”) and Centre National de la Recherche Scientifique (CNRS, défi Transition Energétique, PE “AZIN”).