Introduction

Inspired by the natural process of photosynthesis, the hydrogen produced by an artificial system using energy from the sun is emerging as one of the most fascinating approaches suitable for the conversion and storage of solar energy1,2,3. This scheme requires the use of sensitizers in a multicomponent solution that are essential for capturing, delivering and converting solar energy and later storing it in the form of chemical bonds in a high-energy density fuel, such as hydrogen, with the aid of catalytically active sites4,5,6,7,8,9,10,11. The development of such an artificially intermolecular system requires a fundamental understanding of electron-transfer reactions between the individually constructed components for interesting catalytic reactions. In recent years, considerable progress has been achieved in developing highly absorbing photosensitizers (PSs) and noble-metal-free water reduction catalysts (WRCs)12,13,14,15,16,17. The use of iridium PSs with transition metal-based WRCs, such as cobalt or iron complexes, has been shown to be preferable because such systems typically possess remarkable activity with a limited operational lifetime for the formation of hydrogen from water18,19,20,21. The existence of metal-carbon sigma bonds in cyclometalated Ir(III) emissive species improves their photostability compared to Ru(II) diimines under illumination22. However, for these artificial systems, the photodecomposition of both the sensitizer and catalyst remains a problem that terminates the practical water-splitting reaction in the production of clean fuel23. For solar fuel conversion, the stability under catalytic functional conditions is a critical issue that has led to the search for long-term sustainable artificial systems in which PSs and WRCs are capable of withstanding degradation, which continues to be a central challenge in the development of solar fuels.

To address this issue, many approaches have been employed to improve the durability of iridium PSs, such as placing bulky pendant groups on the backbone of the cationic iridium complexes due to their steric protection24 or using neutral tris-cyclometalated iridium complexes due to their intrinsic high stability to photolysis25,26,27. In addition to the rhodium, palladium or platinum WRCs, molecular catalysts based on cobalt18,19, iron20,21 and nickel28 have been suggested as cheaper alternatives, but there is still abundant room for improvement of the catalysts' stability. In the search for stable photocatalytic systems, we are interested in highly dispersed inorganic nanoparticles (NPs) as working catalysts that can reduce water to hydrogen in the solution phase. Such nanomaterials with superior photostability are particularly attractive for performing this reaction because redox catalysis can be independently optimized by varying the size, shape or surface ligands of the NPs. Molybdenum disulfide appeared to be a promising candidate because it is active for the evolution of H2 at low electrochemical overpotentials for water reduction, which has been identified as an electrocatalyst and a semiconducting photocatalyst for the heterogeneous hydrogen evolution reactions29,30,31,32,33,34. Few investigations have investigated the reduction of water in multicomponent photochemical molecular systems involving the utilization of inorganic catalysts composed of earth-abundant elements because NPs in a stable colloidal form are needed for such applications in aqueous media. Early studies have shown that colloidal MoS2 is active as a WRC replacement for traditional Pt group metal catalysts for hydrogen photoproduction. However, colloidal MoS2 exhibited rather undesirable photocatalytic lifetimes and approached a turnover number (TON) of only 75 with cationic [Ru(bpy)3]2+ (bpy = 2,2′-bipyridine) chromophores35. For further progress with this interesting catalyst, higher conversion efficiencies are required for solar fuel conversion. In addition, a better understanding of the interaction between the new sensitizers and the dispersed MoS2 NPs in the photocatalytic fuel-forming reaction is essential.

To this end, we developed multicomponent hydrogen production systems based on a series of Ir(III) PSs (as shown in Fig. 1) in combination with the colloidal MoS2 WRC. This series allows for the evaluation of the photohydrogen-evolution activity as a function of structure. Due to the functional groups on the PSs, different potential interactions may exist between the PSs and the MoS2 NPs, which may directly interfere with the overall efficiency of the solar energy conversion. In particular, iridium PSs that were functionalized with pendant carboxylic acids as anchoring groups might allow for binding of the molecule chemisorbed onto the surface of the colloidal catalyst to promote electron transfer, which appears to be important for catalytic activity. By controlling the size of the colloidal MoS2 NPs, the selected systems have the ability to produce remarkably high amounts of H2 with an apparent quantum yield (AQY) of up to 12.4%. Further investigation reveals that colloidal MoS2 exhibits desirable activity and long-term stability, which holds great promise for continuing progress in the clean and sustainable production of hydrogen with novel photocatalytically active materials.

Figure 1
figure 1

Formulas of the Ir(III) photosensitizers investigated in this work (3: R = H; 4: R = CF3; 5: R = t-Bu).

Full size image

Results

Characterization of colloidal MoS2 nanoparticles and Ir(III) complexes

The transparent colloidal MoS2 suspensions were obtained according to a previously published protocol35. The poly(vinylpyrrolidone) (PVP) was used as a surface ligand to stabilize the NPs in a fine colloidal state in the aqueous solution. The colloidal suspension of MoS2 clearly displayed a typical Tyndall effect, suggesting the formation of stable MoS2 NPs (Supplementary Fig. S1) and the colloidal suspension was notably stable (no deposition) for several weeks under an air atmosphere. X-ray photoelectron spectroscopy (XPS) was utilized to characterize the chemical states of Mo and S in the colloidal NPs. The binding energies of Mo 3d3/2, Mo 3d5/2, S 2p1/2 and S 2p3/2 were determined to be 232.1, 228.8, 162.8 and 161.7 eV, respectively, indicating the presence of a tetravalent molybdenum ion (Mo4+) and a divalent sulfur ion (S2−) in the resulting colloidal solution. Quantification by XPS indicated a Mo to S ratio of approximately 1:2, which showed that the NPs were present in the form of MoS2. Transmission electron microscopy (TEM) images showed that MoS2 NPs were well-dispersed and appeared to be spherical (Fig. 2a–e). As revealed by the TEM, the dispersion and size distribution of the colloidal MoS2 NPs was controlled by adjusting the ratio of the stabilizer of PVP to MoS2. When a ratio of 40 or 80 was used, the majority of the colloids were less than 10 nm in diameter with better dispersion. When less PVP was added, a small fraction of particles unavoidably agglomerated into large ones. When the colloid preparation was conducted without PVP or in an aqueous solution with PVP, a MoS2 colloid failed to form and the particle size of the prepared MoS2 NPs was greater than 20 nm, as shown in Figs. 2e and 2f, respectively. Due to the relatively low contrast, the TEM micrographs do not provide an unambiguous NP diameter. However, the difference is clear and is expected to lead to a significant impact on their catalytic performance. The diffuse absorption spectrum of colloidal MoS2 in methanol is rather featureless (Fig. S3), which is consistent with that of the MoS2 colloids reported in the literature35.

Figure 2
figure 2

TEM images of MoS2 prepared in methanol with ratio of PVP (monomeric units) to MoS2 of 2.5 (a), 10 (b), 40 (c), 80 (d), 0 (e) and SEM image of MoS2 particles prepared in aqueous solution with ratio of PVP to MoS2 at 40 (f).

The diameter and dispersion of the NPs improves as the ratio of PVP to MoS2 increases, as shown by the curved arrow.

Full size image

All of the iridium complexes were successfully synthesized according to standard procedures and fully characterized. Various cyclometalating and ancillary ligands helped modulate the photophysical and electrochemical properties of the target light-harvesting complexes and provided further insight into the potential relationship between the substituents and the PS performance. In particular, the use of the COOH group in the ancillary ligand of 2,2′-bipyridine-4,4′-dicarboxylate (H2bpdc) was expected to control the funneling of excitation energy toward the photochemical reaction center. Among these iridium PSs, complexes 1 and 2 appear to be electrically neutral, with one of the carboxyl group losing its proton to become a negatively charged carboxylate group (–COO−)36, while the other Ir(III) compounds were isolated as their hexafluorophosphate (PF6−) salts.

Photophysical and electrochemical properties

The absorption and emission spectra of all of the Ir(III) complexes measured in a methanol solution at 278 K are shown in Fig. 3 and the data are summarized in Table 1. All of the complexes exhibit intense absorption bands in the UV region, which corresponds to the spin-allowed singlet ligand-centered (1LC) π → π* transitions of both the C∧N and N∧N ligands. The less intense absorption features are observed in the visible region and are primarily attributed to metal-to-ligand charge transfer (MLCT), which may also include some ligand-to-ligand charge transfer (1LLCT). Upon excitation at 350 nm, all of the complexes exhibit intense orange to green luminescence with decay lifetimes between 0.20 and 0.68 μs (Fig. S4). To further elucidate the nature of the excited states, density functional theory (DFT) calculations were performed on all of the Ir(III) complexes. The contour plot of the frontier molecular orbitals is shown in Table S1. For all of the complexes, the highest occupied molecular orbitals (HOMOs) are distributed between the Ir(III) center and the π orbitals of the cyclometalated ligands. In contrast, the lowest unoccupied molecular orbitals (LUMOs) are primarily localized on the ancillary bipyridine ligand with small contributions from the Ir orbitals, which is in agreement with the results from previous studies23,36. For 1 and 2, the presence of the electronic densities of the LUMO on the anchoring units is believed to facilitate electron transfer from the excited complex to the catalytic reaction center (e.g., MoS2). These results also suggest that the excited states of all of the complexes are primarily due to the contribution of a mixture of 3MLCT [dπ(Ir) → π*N∧N</emph>] and 3LLCT (πC∧N</emph> → π*N∧N</emph>) transitions.

Table 1 Photophysical and electrochemical data of Ir(III) complexes 1–6
Full size table
Figure 3
figure 3

(a) Absorption spectra and (b) normalized emission spectra of complexes 1–6 in methanol at room temperature.

Full size image

The electrochemical behaviors of all of the Ir(III) complexes were measured using cyclic voltammetry in an acetonitrile solution and the electrochemical data are presented in Table 1. All of the compounds exhibit an oxidation potential in the range of +1.02 to +1.52 V (vs. Ag/AgCl) and a reduction couple at ca. −1.65 to −1.21 V (vs. Ag/AgCl). Based on the compositions of the HOMOs and LUMOs, the oxidation process is related to a metal-aryl centered process for each complex, while the reduction occurs on the ancillary bipyridine ligand with little contribution from the iridium metal center, which was determined in the study on the related cyclometalated Ir(III) complexes.

The excited-state reduction potentials, E(PS*/PS−), are obtained from E(PS*/PS−) = Ered + E0-0, in which Ered is the first reduction potential of Ir(III) compounds and E0-0 is the zero-zero excitation energy obtained from the optical spectra. The calculated E(PS*/PS−) values of the complexes, which range from +1.22 to +1.56 V versus NHE, are more positive than the oxidation potential of both ascorbic acid (H2A) and triethanolamine (TEOA) suggesting that the electron transfer from the electron donor to the Ir(III) species via a reductive quenching pathway is thermodynamically feasible37,38. To confirm these analyses, quenching experiments were conducted in methanol/water (1:1, v/v) solutions of these Ir(III) complexes with the addition of quenchers. The excited state of all of the Ir(III) complexes were quenched by both H2A and TEOA following Stern-Volmer plots (Fig. S5) and the quenching rate constants are summarized in Table S4.

Hydrogen production

Photocatalytic hydrogen production using a freshly prepared colloidal suspension of MoS2 NPs was initially operated in a methanol/water solution with iridium complex 1 as a PS and H2A as a sacrificial reducing (SR) agent under visible light irradiation of the system. No H2 formation was detected after 8 h of irradiation in the absence of one of the components (i.e., Ir(III) PS, MoS2 NPs WRC or H2A) or when the reaction was performed without irradiation. Optimization of the reaction medium for hydrogen production was performed in a reaction medium with various methanol–water volume ratios. The TON based on MoS2 increased from 60 to 211 as the ratio of methanol to water changed from 1:4 to 1:1 (v/v) (Entries 1 and 2 in Table 2) while the other reaction conditions remained unchanged. A further increase in the ratio of methanol/water to 4:1 resulted in a decrease in the amount of hydrogen production to 114 H2 TON based on MoS2 (Entry 3 in Table 2). The photochemical process involving intermolecular electron transfer is sensitive to changes in the lower dielectric constant of the reaction media. The highest catalytic efficiency was achieved in a 1:1 methanol-water solution, which could provide a balance between the benefit of the lower dielectric constant at higher methanol concentration and the need for water as a proton source for extracting hydrogen39. Under the condition of taking methanol as a reaction medium without water, the declined activity was observed with a TON of approximately 40; in this case, ascorbic acid can act as a source of protons.

Table 2 Photoinduced hydrogen evolution with Ir(III) complexes 1–6a
Full size table

As a hydrogen evolution catalyst, the activity of MoS2 NPs is correlated with the number of unsaturated sulfur atoms on the edges of the NPs40, which enable them to absorb and release hydrogen. Therefore, the particle size and dispersion of the MoS2 NPs play a key role in its catalytic activity. In the present study, this factor was adjusted by varying the initial ratio of PVP to MoS2 toward the target. Fig. 4a shows the effect of the colloidal MoS2 NPs with various ratios of PVP to MoS2 on the hydrogen evolution efficiency using complex 1 (20 μM) as the PS and H2A as the electron donor under visible light irradiation (λ > 420 nm). When the ratio of PVP/MoS2 changes from 2.5 to 40, the hydrogen evolution efficiency in the homogeneous-like system increased as the particle size of MoS2 decreased, which may be due to the presence of much more active sites due to the smaller size and better dispersion of the MoS2 NPs. In addition, the existence of PVP allows for the enhancement of energy transfer in the solar-fuel converison41. The system with a PVP/MoS2 ratio of 40 exhibits the highest activity for hydrogen production with a TON of 211 based on MoS2 NPs after 20 h of irradiation. Although a smaller diameter of MoS2 particles was observed for a PVP/MoS2 ratio of 80, a lower photocatalytic activity was observed, which may be due to the disruption of electron transfer between the PS and WRC affecting the photocatalytic reactions when a large number of PVP surround MoS2 in the reaction solution. When MoS2 NPs in the bulk powder form were used as a WRC in an heterogeneous system of Ir(III)-MoS2-H2A, no H2 was detected suggesting that it is quite different from the NPs. Due to the size-dependent energy shift, the conduction band potential is not sufficiently large enough for bulk MoS2 to function as a catalyst to reduce protons42. In addition, when PVP was employed in place of the colloidal MoS2 NPs under identical conditions, hydrogen was not produced, which indicated that participation of PVP in the reaction can be ignored.

Figure 4
figure 4

Hydrogen production using (a) varying ratios of PVP to MoS2 with 20 μM 1 and 20 μM MoS2; (b) varying the concentration of colloidal MoS2 NPs with 100 μM 1; (c) varying the concentration of 1 with 20 μM MoS2 NPs; (d) 20 μM 1 and a 20 μM concentration of various WRC. WRC: [Rh] = [Rh(dtb-bpy)3](PF6)3, [Co]′ = [Co(bpy)3]Cl2, [Co]″ = [Co(dmgH)2(H2O)2] and [Pt] = K2PtCl4. All of the reactions were studied in 1:1 (v/v) methanol-water media (100 mL) in the presence of 100 mM H2A at pH 6.5 (λ > 420 nm). Here, the pH value was not optimized.

Full size image

To gauge the effect of the concentrations of the PS and WRC, hydrogen production was conducted with various concentrations of MoS2 colloid or 1 when all of the other conditions remained constant. As shown in Fig. 4b, a substantial increase in hydrogen evolution with complex 1 was observed as the initial MoS2 concentration increased. For MoS2 NPs concentration of 5, 10, 20, 50 and 100 μM in the presence of 100 μM PS and 100 mM H2A, 282, 489, 731, 1290 and 1923 μmol of H2 was obtained with the TON varying from 1128 to 384 based on MoS2 after 20 h of irradiation. Similarly, an increase in the total hydrogen yield with a fixed concentration of MoS2 NPs was also observed when the concentration of the added PS increased from 5 μM to 100 μM. In addition, the TON increased from 82 to 731 based on MoS2 (Fig. 4c). Therefore, the concentration of both the WRC and PS are critical parameters that affect the catalytic production of hydrogen from water.

The performance of colloidal MoS2 as a WRC for hydrogen photoproduction was evaluated in comparison to other WRCs, such as colloidal platinum generated in situ from K2PtCl4, [Co(bpy)3]Cl2, [Co(dmgH)2(H2O)2] (dmgH2 = dimethylglyoxime) or [Rh(dtb-bpy)3](PF6)3 (dtb-bpy = 4,4′-di-tert-butyl-2,2′-bipyridine), in association with 1 under irradiation for 20 h and the same experimental conditions. As shown in Fig. 4d, the TONs based on these species are quite low. The Rh WRC is more active with 30 TON obtained after a 20 h of irradiation. With colloidal MoS2 NPs, a more than ten-fold enhancement compared to the Pt catalyst was observed and hydrogen production slowed down but still retained some activity after the cessation of the reaction (after 20 h), which observed in the kinetics plots in Figs. 4a–c. With the other WRCs due to their poor stability, the reactions lasted approximately 8–12 h. In particular, colloidal platinum easily agglomerates in the presence of methanol as demonstrated by the observation of the formation of Pt particles as a black particulate matter during irradiation. Therefore, the dispersed MoS2 NPs exhibited considerable durability and significantly greater catalytic activity than commonly utilized WRCs for hydrogen production, which may be due to more efficient charge transfer occurring within the cooperative components of the photochemical reactions.

After approximately 20 h of irradiation, the hydrogen-evolution activity decreased sharply due to the decomposition of at least one component of the Ir(III) PS and MoS2 NPs in the presence of a large excess of H2A. To confirm this decomposition, the PS or catalyst were re-added when the H2 evolution ceased. When the MoS2 NPs were re-added to the reaction solution, only 16.5 μmol H2 was regenerated after another 20 h of irradiation, whereas the parallel experiment showed that the re-addition of Ir(III) PS resulted in substantial recovery of catalytic activity, as shown in Fig. 5. These results clearly indicated that the MoS2 NPs were stable during the reaction. After the H2 production leveled off in this system, the decomposition products were extracted from the reaction mixture and characterized by gas chromatography-mass spectrometry (GC-MS). There was a strong ion appearing at m/z = 223, which was characteristic of the 2-(4-trifluoromeththylphenyl)pyridine that dissociated from complex 1 after the photocatalytic reaction (Fig. S6). The result further confirmed that the decomposition of the PS may account for the cessation of the activity of the reaction system.

Figure 5
figure 5

H2 production as a function of time showing the stability of the system with re-addition of 1 after 20 h of irradiation (λ > 420 nm).

The experiment was performed in 1:1 (v/v) methanol-water media (100 mL) in the presence of 20 μM 1, 20 μM MoS2 and 100 mM H2A at the original pH of 6.5.

Full size image

Based on the aforementioned results, the use of and electron-donating agent, such as triethanolamine (TEOA), instead of H2A was investigated in the multicomponent photocatalytic water reduction systems (at a favorable pH of 7.0) (Entry 4 in Table 2). A much higher TON of 1421 based on the catalyst (20 μM MoS2 NPs) was obtained after 12 h of irradiation compared to that of H2A, which had a H2 TON of 731 after 20 h of irradiation under the same conditions. In addition, the turnover number frequency (TOF) with TEOA was as high as 118 h−1, which is higher than that of H2A (i.e., TOF of H2A was 37 h−1). It should be noted that the lifetime of the H2 evolution system with TEOA was relatively short where 87% of the total hydrogen was evolved during the first four hours. Under the same conditions using TEOA during the course of a 12 h experiment, the TON reached a maximum value of approximately 3124 at concentrations of MoS2 NPs down to 1 μM. While H2A quenched the excited state of 1 more efficiently than TEOA, H2A was more readily oxidized than TEOA and less effective than TEOA in performing the reaction. This result is most likely due to the pH of the reaction with TEOA under optimized reaction conditions. The photocatalytic performance is closely related to the pH value of the reaction system. The effect of pH on the production of H2 was investigated in the 1-MoS2-TEOA systems, as shown in Fig. 6a. Maximum production of H2 occurred at a pH of 7, which is consistent with the observations in related photocatalytic systems. In addition, the apparent quantum yield (AQY) in the 1-MoS2-TEOA system was determined to evaluate the photon-to-H2 efficiency under irradiation of monochromatized light at 350, 380, 400, 420, 440 and 480 nm under the given reaction conditions. A maximum AQY of 12.4% was obtained under irradiation at 400 nm (light intensity: 0.40 mW/cm2). This value compares favorably with other iridium-based multicomponent systems for the photogeneration of hydrogen using colloidal Pt as the catalyst19,20. As shown in Fig. 6b, the AQY as a function of the increasing wavelength is in agreement with the characteristic features of the excitation spectra of compound 1 (Fig. S7), which indicates that the occurrence of hydrogen evolution reaction was plausibly governed by the excited state of 1.

Figure 6
figure 6

(a) Hydrogen production as a function of pH in the 1-MoS2-TEOA systems. HCl (approximately 38%) was added to each solution to adjust the pH. (b) Hydrogen production quantum yield plotted as a function of the wavelength of the incident light in systems with TEOA at pH 7 after initial 4 h of irradiation.

Full size image

Discussion

The contribution from the system based on colloidal MoS2 NPs as the catalyst with an iridium sensitizer for photoinduced H2 production was impressive based on the aforementioned results. The size and dispersion of the MoS2 NPs has an obvious effect on the improvement of the catalytic performance in photocatalytic H2 production. The amount of unsaturated sulfur atoms, which reside on the MoS2 NPs edges and are responsible for the high catalytic activity, increases as the particle size decreases, resulting in better dispersion and increased catalytic activity. In addition, the ability to perform as an active site is controlled by the value of the band gap in the nanoscale MoS2 due to quantum size effects42, as shown in supplementary Fig. S8. The reduction potential of the MoS2 NPs shifts to more negative values as the size of the MoS2 particle decreases, which offers the potential for proton reduction at an appreciable rate. However, the conduction band value of bulk MoS2 is more positive than the proton reduction potential E(H+/H2) making bulk MoS2 inactive for the formation of hydrogen.

The photocatalytic reaction in the multicomponent system involves electron transfer, which proceeds either by oxidative or reductive quenching. To determine which quenching is operative in the reaction described here, fluorescence quenching within the PS, the WRC and SR agent must be investigated. In this study, the excited state (PS*) of complex 1 was substantially quenched by both H2A and TEOA. However, the contribution of the WRC of MoS2 to the quenching processes was difficult to estimate due to the optical properties of MoS2. In these MoS2-based systems, the amount of the SR agent is more than 1000 times that of the catalyst. The predominant pathway for electron transfer was proposed to proceed via reductive quenching derived from the rapid oxidation of H2A or TEOA combined with the strong oxidizing nature of the excited state of the PS. Previous studies have shown that the excited state of a neutral iridium complex27 (i.e., Ir(ppy)3 (ppy = 2-phenylpyridine)) cannot be quenched reductively by TEOA but can be oxidatively quenched by [Co(bpy)3]2+. Under conditions identical to those used for hydrogen production by 1, hydrogen was not produced by the Ir(ppy)3-MoS2-H2A system under 12 h of visible light irradiation, which indicated that the electron was not transferred from the PS* to the catalyst, resulting in suppression in the production of hydrogen. Therefore, the reductive quenching pathway appears to be consistent with the results reported here for the photochemical system catalyzed by MoS2 NPs. In the hydrogen formation process, sensitizer 1 harvests visible light and the excited state of the sensitizer is reductively quenched by the SR agent (H2A) to generate the activated reduced iridium species that directly delivers reducing equivalents to the MoS2 NPs WRC where hydrogen is evolved.

Using adsorbing moieties, the sensitizer might be localized on the semiconductor nanoparticles to improve the electron transfer efficiency, which has been extensively employed in dye-sensitized solar cells43,44. The Fourier transform IR (FT-IR) spectroscopy was used for extracting information of the binding of PS 1 on the MoS2 NPs surface (Fig. S9). Two strong bands in the FT-IR spectrum of 1 appeared at approximately 1608 cm−1 and 1373 cm−1 corresponding to the asymmetric and symmetric stretch modes of the carboxylate group, respectively. For comparison of the characteristic peaks associated with the carboxylate group, the 1-sensitized MoS2 NPs as a powder with diameters up to 20 nm was investigated; in the case of the colloidal MoS2 NPs, the IR bands involving the PVP were superimposed over the relevant peaks. Due to the interaction between PS 1 and the MoS2 NPs, the corresponding peaks in the adsorbed state occurred at 1602 cm−1 and 1402 cm−1. Based on spectral changes and the difference between the symmetric and asymmetric bands of the carboxylate group, the binding of the carboxylate groups of PS 1 on the MoS2 NPs surface was probably via a bridging bidentate mode45. The emission spectrum of PS 1 adsorbed on MoS2 NPs was similar with that of 1 in CH2Cl2 solution (Fig. S10); however, the Ir-based emissions were not detected when using other Ir(III) complexes with other substituents such as 3 or 6 as a PS for sensitizing the MoS2 NPs. The photoluminescence studies provide a supplementary evidence of chemical interaction between 1 and the MoS2 NPs. Inspired by this possibility, the use of carboxylate groups on iridium sensitizer 1 was expected to promote charge transfer to the catalyst resulting in improved performance of the hydrogen evolution systems. After the successful application of complex 1, complex 2 bearing a H2bpdc ligand with structure similar to 1 was evaluated. As expected, when H2A was used as the SR agent, complex 2 also exhibited efficient activity with a 426 H2 TON based on MoS2 for hydrogen production, which was lower than that observed for complex 1 (Entry 5 in Table 2). This result is due to a smaller quenching rate constant for H2A and a shorter excited-state lifetime of 2 relative to 1. To further address the relationship between the performance of the photocatalytic reaction and the structure of the sensitizer, other Ir(III) complexes (3–6) with other substituents, such as –CH2OH or –NH2 (see Fig. 1), which allow for structural analogues of complexes 1 and 2, were examined as PSs for hydrogen evolution under the same conditions. Only 4, 3, 3 and 8 TON based on MoS2 were observed after 20 h of irradiation for systems based on complexes 3, 4, 5 and 6 (Entries 6, 7, 8 and 9 in Table 2), respectively. In addition, the parent chromophore [Ir(ppy)2(bpy)]+ was also used as a reference PS. The unbound chromophore [Ir(ppy)2(bpy)]+ does not possess a moiety in the pendent groups eliminating the possibility of direct interaction between the PS and the WRC similar to complexes 1 and 2 and 48 TON based on MoS2 was observed (Entries 10 in Table 2). A “standard” PS of [Ru(dmphen)3](PF6)2 (dmphen = 4,7-dimethy-1,10- phenanthroline) was used under the same conditions (Entry 11 in Table 2) and only 92 μmol of H2 with respect to MoS2 was observed, which is comparable to the reported data using [Ru(bpy)3]2+. The performance of these iridium PSs were also investigated using TEOA as a SR agent and TONs of 12, 15, 9 and 10 were observed after 12 h of irradiation for system based on complexes 2, 3, 4, 5 and 6, respectively, which are much less than those of complexes 1 and 2.

Hydrogen is formed by photoinduced electron transfer. However, the contribution of this process to the reaction is not completely understood. In this study, the photophysical properties and the electrochemical behaviors of PSs are similar but result in completely different catalytic properties. It has been proposed that the electron transfer rates between the molecular PS and the WRC play an important role in determining the overall hydrogen evolution efficiency. In previous reports, the weak adsorption effects between iridium PSs and platinum from the vinyl or pendant pyridyl functionalities can not only facilitate fast25, directional electron transfer from the PS to the WRC but also effectively suppress the probability of bipyridyl photodissociation resulting in highly efficient activities for light-driven water reduction systems10. Studies on dye-sensitized MoS2 NPs with improved photoelectrochemical property are rare. However, a previous study has suggested that electron injection can occur when Ru(4,4′-dicarboxy-2,2′-bipyridine)2cis(NCS)2 is adsorbed on the MoS2 nanoclusters via the carboxylate groups46. In this study, the photocatalytic results confirmed the existence of carboxylate groups by which electron can easily transfer from the adsorbing PS to the MoS2 NPs, as shown in Fig. 7. Ease of electron transfer is extremely important in determining the enhancement of the PS performance for hydrogen production. For complexes 1 and 2, the DFT results indicated that the electron resides on the side of the COOH moiety upon excitation. The electron might be easier to transfer to the MoS2 NPs via COOH, which may provide the active electron transfer channel. In contrast, the non-adsorbing PS may exhibit little or no attachment to the same catalyst in the colloidal MoS2 NPs leading to a lower photocatalytic activity.

Figure 7
figure 7

Schematic representation of the electron transfer from PS 1 to the MoS2 NPs.

Full size image

In summary, molybdenum disulfide is an interesting hydrogen evolution catalyst because it is composed of more abundant and economical materials than are Pt-based catalysts. The utility of colloidal MoS2 as a catalyst coupled with iridium sensitizers for efficient photoinduced hydrogen production was developed using a three-component systems. The combined system exhibited much better catalytic activity than that observed for the K2PtCl4, [Co(bpy)3]Cl2, [Rh(dtb-bpy)3](PF6)3 and [Co(dmgH)2(H2O)2] catalysts under identical conditions. Due to cooperative effects, the conversion efficiency reached 12.4% at 400 nm in the Ir(III)-MoS2-TEOA system. For Ir(III) sensitizers, the substituent groups in the ancillary ligand were determined to influence their catalytic performance. The use of carboxylate groups leads to the formation of electron transfer channels promoting an increase in the electron transfer from the sensitizers to the colloidal MoS2 NPs, which positively affects the activity of the catalytic system. We anticipate that this study will provide new insights into producing a robust photochemical system with an improved conversion efficiency and we believe that the promising MoS2 NPs-supported catalysts has the potential for use in a hybrid energy conversion system without the use of noble-metal complexes.

Methods

Syntheses

All of the iridium complexes (1–6) were prepared via the known two-step bridge-splitting procedure. The colloidal MoS2 NPs were prepared via a one-step reaction of (NH4)2MoS4 and hydrazine hydrate in a methanol solution according to a previously reported method.

Photogeneration of hydrogen experiments

The reactions were performed in a Pyrex vessel attached to a closed gas-circulation glass system and an evacuation system. The samples contained different concentrations of the PS, WRC and SR agent in a 100 mL mixed methanol–water solution with gentle magnetic stirring during the H2 production experiments. The system was well-evacuated and subsequently backfilled with argon prior to side visible-light irradiation using a 300 W Xenon lamp equipped with a cut-off filter (radiation wavelength > 420 nm). The evolved gases were periodically monitored in situ using an online gas chromatograph with a thermal conductivity detector (Shimadzu GC-8A, argon as a carrier gas and MS-5A column).