Photocatalytic H₂ evolution over semiconductors is a promising approach to convert solar energy to clean hydrogen energy [1-4]. A large number of photocatalysts have been developed and investigated for photocatalytic H₂ evolution, CO₂ reduction, and environmental contaminant removal, including metal oxides, sulfides, oxynitrides, and graphitic carbon nitride (g-C₃N₄) [5-34]. Among these photocatalysts, sulfides are excellent candidates for photocatalytic H₂ evolution because of their suitable band gap energies and sites for facilitating the photocatalytic H₂ reaction. Ternary chalcogenide ZnIn₂S₄, a layered structure with a band gap energy of 2.0–2.4 eV, has attracted attention in recent years because of its favorable visible-light harvesting capability, photostability, and low toxicity [35-41]. However, pure ZnIn₂S₄ usually suffers from rapid recombination of photogenerated electron–hole pairs and a large kinetic barrier for H₂ evolution reaction at the surface sites. As a result, the photocatalytic H₂ evolution performance over pure ZnIn₂S₄ is not satisfactory. To improve the photocatalytic H₂ evolution activity of ZnIn₂S₄, much effort has been made to suppress recombination of charge carriers and promote charge transfer and separation, such as depositing noble metal cocatalysts and combining ZnIn₂S₄ with graphene, carbon nanotubes, carbon quantum dots, and other semiconductors [42-49]. Although loading the precious metal Pt on the surface of ZnIn₂S₄ can enhance the photocatalytic H₂ evolution activity, this strategy has high cost for the photocatalytic reaction and is thus not suitable for practical application. Therefore, it is imperative to discover low-cost, earth-abundant, low-toxicity, and high- efficiency materials to replace noble metals as cocatalysts for the photocatalytic H₂ evolution reaction for practical application [50-52].

Recently, the layered MoS₂ structure has received extensive attention as a cocatalyst in the photocatalytic H₂ evolution reaction [53-71]. For example, Ye's group [53] coupled exfoliated MoS₂ monolayers with commercialized CdS to form composites that exhibit significantly enhanced photocatalytic H₂ evolution activities. Zhao et al. [54] synthesized MoS₂/Cd_0.5Zn_0.5S composites by a two-step hydrothermal route. They achieved an extremely high H₂ evolution rate of 12.30 mmol/h/g at the optimal loading amount of MoS₂. Yu et al. [55] used the adsorption in situ transformation method to prepare amorphous MoS_x/g-C₃N₄ hybrids. Photocatalytic experimental results showed that the H₂ evolution activity of g-C₃N₄ is markedly improved compared with pure g-C₃N₄. Meng's group [56] anchored MoS₂ nanosheets on the surface of ZnIn₂S₄ by the in situ photoassisted deposition method. They achieved a H₂ evolution rate of 8.047 mmol/h/g, which is 28 times higher than that of pure ZnIn₂S₄ under identical conditions. It has been suggested that facilitation of the photocatalytic H₂ evolution using MoS₂ as a cocatalyst originates from the unsaturated edge sulfur atoms acting as active sites to rapidly capture protons from aqueous solution and then promote direct reduction of H⁺ to H₂ [57]. Based on the aforementioned reports, designing and loading MoS₂ on the surface of semiconductors could be an effective method to improve the photocatalytic H₂ evolution activity. Herein, a two-step hydrothermal route was used to fabricate MoS₂/ZnIn₂S₄ composites. A significantly enhanced photocatalytic H₂ evolution activity was achieved in the presence of a small amount of MoS₂ as a cocatalyst compared with pure ZnIn₂S₄.

The ZnIn₂S₄ samples were fabricated according to our previously reported method [36]. Typically, Zn(NO₃)₂·6H₂O (1.0 mmol) and In(NO₃)₃·4.5H₂O (2 mmol) were dissolved in deionized water (70 mL) and the pH value of the solution was then adjusted to 1.0 by adding hydrochloric acid (1 mol/L). Next, thioacetamide (TAA, 10 mmol) was added to the above solution with intense magnetic stirring. After stirring for 30 min at room temperature, the mixture was transferred to a 100-mL Teflon-lined stainless steel autoclave and heated at 160 ℃ for 12 h in an electric oven. The mixture was then naturally cooled to room temperature. The obtained light yellow product was collected by centrifugation, washed with deionized water and ethanol three times, and then dried at 60 ℃ for 12 h.

The MoS₂/ZnIn₂S₄ composites were synthesized by hydrothermal treatment. The typical procedure using the 1 wt% MoS₂/ZnIn₂S₄ composite as an example was as follows. As-fabricated ZnIn₂S₄ (0.2 g) was dispersed in deionized water (60 mL) with ultrasonic treatment. (NH₄)₆Mo₇O₂₄·4H₂O (0.0022 g) and thiourea (0.0041 g) were then added to the above suspension. After stirring for 30 min, the suspension was transferred to a 100-mL Teflon-lined stainless steel autoclave and kept at 210 ℃ for 24 h. The resultant product with a theoretical mass ratio of 1 wt% MoS₂ in the MoS₂/ZnIn₂S₄ composite was washed with deionized water and ethanol several times, and then dried at 60 ℃ for 12 h. Samples with different theoretical MoS₂ mass ratios (0.5–20 wt%) were fabricated by changing the added amounts of (NH₄)₆Mo₇O₂₄·4H₂O and thiourea. The pure MoS₂ sample was prepared by the same procedure except that the ZnIn₂S₄ precursor was not added. Pure ZnIn₂S₄ treated by the same second hydrothermal procedure without adding MoS₂ and 1 wt% Pt-loaded ZnIn₂S₄ obtained by the light-assisted reduction method were used as controls.

The crystal phases were analyzed by a Shimadzu XRD-7000 diffractometer operating at 40 kV and 30 mA with Cu K_α irradiation. Morphology analysis and energy dispersive spectroscopy (EDS) were performed with a JSM-7100F field emission scanning electron microscope (FESEM). High-resolution transmission electron microscopy (HRTEM) observation was performed with a JEOL JEM 2100F electron microscope operating at 200 kV accelerating voltage. The X-ray photoelectron spectroscopy (XPS) measurements were performed with a VG Multilab 2000 spectrometer operating at 300 W with an Al Kα source. Raman spectroscopy was performed with a Jobin Yvon LabRAM HR800 spectrometer. The ultraviolet-visible (UV-vis) diffuse reflectance absorption spectra were recorded with a Purkinje General TU-1901 spectrophotometer using BaSO₄ as the reference standard. The Brunauer-Emmett-Teller (BET) specific surface areas of the samples were determined with a Micromeritics ASAP 2020 nitrogen adsorption apparatus. The photoluminescence (PL) spectra were recorded with a PerkinElmer LS55 fluorescence spectrometer at an excitation wavelength of 350 nm. The elemental contents of the samples were determined by a Shimadzu EDX-7000 X-ray fluorescence (XRF) spectrometer with a Cu target in standardless mode.

The photocatalytic H₂ evolution experiments were performed in a 300-mL Pyrex glass reactor. A 300 W Xe lamp (PLS-SXE300, Beijing Trusttech Co. Ltd., China) with a 420-nm cut-off filter was used as the light source. In a typical photocatalytic H₂ evolution experiment, the photocatalyst (80 mg) was dispersed in an aqueous solution (80 mL) containing lactic acid (10 mL) by ultrasonic treatment. Before irradiation, the reactor was thoroughly evacuated to remove air. During irradiation, continuous magnetic stirring was performed to maintain the photocatalyst in suspension. After irradiation for 1 h, 100 μL of the gas was intermittently sampled and the amount of hydrogen was detected by a gas chromatograph (SP7820, thermal conductivity detector, X13 molecular sieve column, N₂ carrier gas).

The apparent quantum efficiency (AQE) for H₂ evolution was determined in a 75-mL Pyrex glass reactor. Differing from the above experiments, 20 mg of the photocatalyst was dispersed in 20 mL aqueous solution containing 2 mL lactic acid. The other photoreaction conditions were to the same as the above experiments except that a band-pass filter (λ = 420 nm) was used instead of the cut-off filter. The average intensity and area of irradiation were determined to be 22.8 mW/cm² and 9.62 cm², respectively. The AQE for H₂ evolution was determined by

Electrochemical impedance spectroscopy (EIS) and the photocurrent tests were performed using an electrochemical system (CHI 760D, China) with a three-electrode cell. The working electrodes were prepared as follows. A certain amount of the photocatalyst was mixed with ethanol (1 mL) and Nafion aqueous solution (1 mL, 5 wt%) by ultrasonic treatment. The mixture (0.1 mL) was then dropped on indium tin oxide (ITO) glass (1 cm × 1 cm). After evaporation of ethanol in air, the photocatalyst was attached to the ITO glass surface. A platinum plate and Ag/AgCl electrode were used as the counter and reference electrodes, respectively. Na₂SO₄ solution (0.1 mol/L) was used as the electrolyte. A 300-W Xe lamp with a 420 nm cut-off filter acted as the light source.

The crystal phases of the samples were identified by XRD. As shown in Fig. 1, pure ZnIn₂S₄ shows characteristic diffraction peaks of the (006), (102), (110), (116), and (202) crystal planes, which can be indexed to the hexagonal phase of ZnIn₂S₄ (JCPDS No. 65-2023) [36]. For MoS₂, the diffraction peaks located at 2θ = 14°, 33°, 39°, and 59° are assigned to the (002), (100), (103), and (110) crystal planes of MoS₂ (JCPDS No. 37-1492) [58]. The XRD patterns of the MoS₂/ZnIn₂S₄ composites with MoS₂ mass ratios of 0.5–10 wt% are similar to pure ZnIn₂S₄, which may be because of the low contents and high dispersion of MoS₂ in the composites. In contrast, the diffraction peaks of the (002) and (100) crystal planes ascribed to MoS₂ are observed in the XRD pattern of the 20 wt% MoS₂/ZnIn₂S₄ composite. The above results suggest that loading MoS₂ on the surface of ZnIn₂S₄ by hydrothermal treatment does not influence the crystal phase of ZnIn₂S₄. To accurately determine the MoS₂ contents in the MoS₂/ZnIn₂S₄ composites, XRF spectroscopy was performed. The mass ratios of MoS₂ in the 0.5, 1, 3, 5, 10, and 20 wt% MoS₂/ZnIn₂S₄ composites are estimated to be 0.47, 0.67, 2.46, 4.23, 8.89, and 17.52 wt%, respectively, which are close to the theoretical contents.

FESEM images of pure ZnIn₂S₄, pure MoS₂, and the 1 wt% MoS₂/ZnIn₂S₄ composite are shown in Fig. 2(a)–(c). Pure ZnIn₂S₄ forms relatively uniform spheres (average diameter 3 μm) composed of a large number of interleaving flakes (Fig. 2(a)). Pure MoS₂ forms micrometer-sized sheets. MoS₂ aggregation is greatly inhibited for the 1 wt% MoS₂/ZnIn₂S₄ composite. There are some agglomerated sheets with sizes of 0.1–0.5 μm distributed on the surface of ZnIn₂S₄. The size and morphology of ZnIn₂S₄ are not obviously affected by the second hydrothermal process. HRTEM was performed to reveal the structure of the MoS₂/ZnIn₂S₄ composite (Fig. 2(d)). The MoS₂/ZnIn₂S₄ composite is composed of microsized ZnIn₂S₄ spheres and nanosized MoS₂ agglomerates, which agrees with the SEM observations. HRTEM was also performed to clearly identify the structure of the 1 wt% MoS₂/ZnIn₂S₄ composite (Fig. 2(e)). A distinct heterostructure interface is observed, indicating that MoS₂ is in close contact with ZnIn₂S₄. The lattice spacing of 0.63 nm corresponds to the (002) plane of MoS₂, whereas the lattice spacing of 0.32 nm corresponds to the (102) plane of ZnIn₂S₄. The elemental composition of the composite was determined by EDS measurements (Fig. 2(f)). The composite is composed of Zn, In, Mo, and S. The above results confirm coexistence of MoS₂ and ZnIn₂S₄ in the composite.

To determine the chemical composition and the valence states of the constituent species, XPS was performed. Fig. 3(a) shows the XPS survey spectra of pure ZnIn₂S₄ and the 1 wt% MoS₂/ZnIn₂S₄ composite. In the survey spectrum of the 1 wt% MoS₂/ZnIn₂S₄ composite, there is a weak Mo 3p peak, indicating that the composite contains little Mo. The Zn 2p spectra are shown in Fig. 3(b). In the high-resolution Zn 2p XPS spectrum of ZnIn₂S₄, the peaks at binding energies of 1045 and 1022 eV can be attributed to Zn 2p_1/2 and Zn 2p_3/2, respectively. The Zn 2p peaks are slightly shifted in the spectrum of the 1 wt% MoS₂/ZnIn₂S₄ composite. Fig. 3(c) shows the high-resolution In 3d XPS spectra. The peaks at about 445.0 and 452.5 eV are assigned to In 3d_5/2 and In 3d_3/2. Compared with pure ZnIn₂S₄, the In 3d XPS peaks for the 1 wt% MoS₂/ZnIn₂S₄ composite are shifted toward higher binding energy. The high-resolution Mo 3d XPS spectra are shown in Fig. 3(d). The peaks at 229.6 and 233.1 eV are attributed to Mo 3d_5/2 and Mo 3d_3/2, in accordance with a previous report [56]. The high-resolution S 2p XPS spectrum of ZnIn₂S₄ can be fitted as two peaks (Fig. 3(e)). The peaks at binding energies of 161.8 and 162.9 eV are associated with S 2p_3/2 and S 2p_1/2, respectively. Similar to the Zn 2p and In 3d peaks, the S 2p peaks in the MoS₂/ZnIn₂S₄ composite are also shifted to higher values. These shifts are indicative of electronic interaction and transfer between ZnIn₂S₄ and MoS₂, which is in agreement with a previous report [59]. The XPS results confirm close contact between ZnIn₂S₄ and MoS₂.

Raman spectroscopy is a useful technique to investigate the microscopic structures of samples. Fig. 4 shows the Raman spectra of ZnIn₂S₄ and the 1 wt% MoS₂/ZnIn₂S₄ composite. For ZnIn₂S₄, there are five characteristic peaks at 121, 243, 294, 343, and 367 cm^-1, which is consistent with a previous report. The strong peak at 121 cm^-1 confirms the layered structure of ZnIn₂S₄, and the other four peaks are related to the LO₁, TO₂, LO₂, and A_1g modes of ZnIn₂S₄ [37]. For the 1 wt% MoS₂/ZnIn₂S₄ composite, the Raman peaks corresponding to ZnIn₂S₄ are present in the spectrum. The Raman peaks of MoS₂ are also present at about 382 and 408 cm^-1, which can be attributed to the E_2g and A_1g modes of MoS₂ [53]. In the spectrum of the 1 wt% MoS₂/ZnIn₂S₄ composite, the representative peaks of ZnIn₂S₄ have shifted from 121 to 128 cm^-1 and from 294 to 304 cm^-1 compared with the spectrum for ZnIn₂S₄. These red shifts indicate the strong interaction between ZnIn₂S₄ and MoS₂, which is in agreement with the XPS investigation.

UV–vis diffuse reflectance absorption spectroscopy was performed to investigate the optical absorption properties of the samples (Fig. 5). Pure ZnIn₂S₄ exhibits a strong absorption band edge at about 525 nm, which is related to the intrinsic band gap absorption. Using the equation E_g = 1240/λ, where E_g and λ are the band gap energy and absorption band edge of the semiconductor, the band gap energy of ZnIn₂S₄ is about 2.36 eV. Notably, the MoS₂/ZnIn₂S₄ composite shows enhanced absorption intensity in the visible region compared with pure ZnIn₂S₄. Furthermore, the absorption intensity increases with increasing mass ratio of MoS₂ in the composite, which is consistent with the color of the samples changing from light yellow to black.

The BET surface areas and pore size distribution curves of pure ZnIn₂S₄, MoS₂, and the 1 wt% MoS₂/ZnIn₂S₄ composite were obtained by nitrogen adsorption–desorption measurements (Fig. 6). The three isotherms are very similar and type Ⅳ according to the Brunauer–Deming–Deming–Teller classification. Furthermore, the shapes of the hysteresis loops in the range 0.5–1.0 P/P₀ are type H3, which indicates formation of porous structures in the samples. The BET specific surface areas of pure ZnIn₂S₄, MoS₂, and the 1 wt% MoS₂/ZnIn₂S₄ composite are estimated to be 56.0, 48.9, and 49.9 m²/g, respectively. The pore size distribution curves indicate that the samples exhibit wide pore size distributions (20 to 300 nm), verifying the presence of mesopores and macropores in the samples.

The photocatalytic H₂ evolution activities of the samples with different mass ratios of MoS₂ were evaluated under visible light irradiation using lactic acid as a sacrificial reagent. For comparison, photocatalytic H₂ evolution experiments of pure ZnIn₂S₄ treated by the second hydrothermal procedure (denoted as the reference sample) and 1 wt% Pt-loaded ZnIn₂S₄ were also performed under the same reaction conditions. The results are shown in Fig. 7. Pure ZnIn₂S₄ and ZnIn₂S₄ treated by the second hydrothermal procedure show poor photocatalytic H₂ evolution activity with H₂ evolution rates of 23 and 45 μmol/h, respectively, which can be ascribed to fast recombination of the photogenerated electron–hole pairs and the deficiency of reactive active sites. The slight enhancement of the photocatalytic H₂ evolution activity for ZnIn₂S₄ treated by the second hydrothermal procedure may be because of the increase of the crystallinity of ZnIn₂S₄. When a small amount of MoS₂ is added to ZnIn₂S₄ to form a composite, the photocatalytic H₂ evolution activity markedly improves. With increasing MoS₂ content, the photocatalytic activity initially increases and then decreases. When the MoS₂ amount in the composites is about 1 wt%, the H₂ evolution rate reaches the highest value of 343 μmol/h, corresponding to an AQE of about 3.85% at 420 nm monochromatic light. The 1 wt% MoS₂/ZnIn₂S₄ composite has a 14.9 times faster H₂ evolution rate than pure ZnIn₂S₄ under visible light illumination. The H₂ evolution rate decreases with increasing amount of MoS₂ above 1 wt%. This may be because MoS₂ covers the active sites on the surface of ZnIn₂S₄ and shields light absorption of ZnIn₂S₄, which might influence excitation of ZnIn₂S₄. As a control, the photocatalytic H₂ evolution rate of 1 wt% Pt-loaded ZnIn₂S₄ is only 120 μmol/h under the same conditions, which is lower than that of the 1 wt% MoS₂/ZnIn₂S₄ composite. This indicates that MoS₂ is a suitable substitute for Pt in cocatalysts for the photocatalytic H₂ evolution reaction.

The photocatalytic stability of a semiconductor is crucial for its practical application. The photostability of ZnIn₂S₄ and the 1 wt% MoS₂/ZnIn₂S₄ composite were investigated in three consecutive runs with a total time of 15 h with the sacrificial reagent solution periodically replaced in each run (Fig. 8). The photocatalytic H₂ evolution rate decreases because of consumption of the sacrificial reagent in each run. When the sacrificial reagent solution is changed, the photocatalytic H₂ evolution activity recovers. In the third run, the photocatalytic H₂ evolution activity of the MoS₂/ZnIn₂S₄ composite exhibits a slight decrease, indicating that photocorrosion occurs. The crystal phase of the recycled 1 wt% MoS₂/ZnIn₂S₄ composite after 15 h photocatalytic reaction was determined by XRD (Fig. 9). The intensities of the diffraction peaks in the XRD pattern of the used sample decrease compared with those before the reaction, suggesting that photocorrosion occurred during the photocatalytic reaction.

EIS tests were performed to investigate interfacial charge transfer and the separation efficiency over pure ZnIn₂S₄ and the 1 wt% MoS₂/ZnIn₂S₄ composite. As shown in Fig. 10(a), the diameter of the Nyquist circle for the 1 wt% MoS₂/ZnIn₂S₄ composite is less than that for pure ZnIn₂S₄, indicating that the composite has more rapid interfacial charge transfer and effective charge separation, which is beneficial for improvement of the photocatalytic H₂ evolution activity. Transient photocurrent response measurements were also performed to verify effective charge transfer and separation in the MoS₂/ZnIn₂S₄ composite. Generally, the photocurrent is caused by diffusion of photogenerated electrons to the back contact and photogenerated hole capture by electron donors in the electrolyte [60]. In Fig. 10(b), the photocurrent density of the 1 wt% MoS₂/ZnIn₂S₄ composite is higher than that of pure ZnIn₂S₄, which suggests that the MoS₂/ZnIn₂S₄ composite undergoes more effective charge transfer and separation than ZnIn₂S₄. PL spectroscopy was also performed to investigate the separation efficiency and recombination of photogenerated charge carriers (Fig. 11). The 1 wt% MoS₂/ZnIn₂S₄ composite exhibits a lower PL peak intensity than pure ZnIn₂S₄, indicating that the charge carrier recombination rate of the MoS₂/ZnIn₂S₄ composite is suppressed under visible light irradiation. These results confirm that coupling MoS₂ with ZnIn₂S₄ to form a composite can facilitate separation of photogenerated electron–hole pairs.

Based on the above results, a possible mechanism for the improvement in the photocatalytic H₂ evolution activity is shown in Fig. 12. The conduction band (CB) and valence band (VB) potentials of ZnIn₂S₄ were estimated by the empirical equation E_CB = χ -E^c -0.5E_g, where E_CB is the CB potential, χ is the electronegativity of the semiconductor, which is the geometric mean of the electronegativities of the constituent atoms, and E^c is the energy of free electrons on the hydrogen scale (about 4.5 eV). For ZnIn₂S₄, the calculated χ and E_g values are 4.82 and 2.36 eV, respectively [45]. Thus, the E_CB value of ZnIn₂S₄ is estimated to be -0.86 eV. Using the equation E_VB = E_CB + E_g, the VB potential E_VB = 1.50 eV. For MoS₂, E_CB = -0.12 eV and E_g = 1.8 eV versus the normal hydrogen electrode potential [54, 58]. Under visible light irradiation, the photogenerated electrons are easily excited from the VB to the CB of ZnIn₂S₄. Because of the matched energy band positions of ZnIn₂S₄ and MoS₂, the photoexcited electrons can easily transfer from ZnIn₂S₄ to MoS₂, which promotes separation and suppresses recombination of the photogenerated electron–hole pairs. MoS₂ then acts as active sites to reduce H⁺ to H₂. The holes in the VB of ZnIn₂S₄ would be consumed by the sacrificial reagent to generate the oxidation products. Therefore, this coupled system is not only favorable for separation of photogenerated electron–hole pairs, but it also improves the photocatalytic H₂ evolution activity.

MoS₂/ZnIn₂S₄ composites have been fabricated by a two-step hydrothermal process and applied to photocatalytic H₂ evolution with lactic acid as a sacrificial reagent. A remarkable improvement in photocatalytic H₂ evolution activity is achieved by adding a small amount of MoS₂ to ZnIn₂S₄, which can be attributed to efficient transfer and separation of photogenerated electron–hole pairs. Among the composites, the 1 wt% MoS₂/ZnIn₂S₄ composite shows the highest photocatalytic H₂ evolution rate of 343 μmol/h under visible light irradiation, which is even higher than that of 1 wt% Pt-loaded ZnIn₂S₄. The corresponding AQE is about 3.85% at 420 nm monochromatic light. This work provides an approach to prepare high-performance MoS₂-based photocatalysts for many potential applications.