Dipole synergy enables fast directional charge transport for solar hydrogen and benzaldehyde coproduction

Abstract

Artificial photosynthesis for hydrogen evolution coupled with benzyl alcohol photoreforming faces efficiency challenges due to insufficient charge directional transfer. Here, we report a synergistic strategy integrating the polarization electric field from asymmetric Zn₃In₄S₉ and the interface dipole field induced by MoS₂ to drive fast charge dynamics. The optimized 6%-MoS₂/Zn₃In₄S₉ exhibits notable photocatalytic performance, generating 41.19 mmol g^-1 h^-1 of hydrogen and 43.33 mmol g^-1 h^-1 of benzaldehyde, which is 11.8 to 12.2 times higher than that of Zn₃In₄S₉. Notably, apparent quantum yields reach 36.6% ± 0.7% for hydrogen and 40.0% ± 0.3% for benzaldehyde (3 times), while retaining 93.8% and 87% of initial activity after 30 hours, demonstrating high stability. In this work, we reveal that the intrinsic dipole of Zn₃In₄S₉ generates a polarization electric field, suppressing bulk charge recombination. Concurrently, the MoS₂-induced interface dipole field creates a fast electron transport pathway from Zn₃In₄S₉ to MoS₂.

This work demonstrates how engineered dipole synergy between two semiconductor materials creates efficient charge transport pathways, enabling simultaneous production of clean hydrogenand high-value chemicals from solar energy.

Introduction

Hydrogen energy, recognized as the most promising clean energy carrier of the 21st century, combines the advantages of zero carbon emissions and high energy density, positioning it as a crucial option for future energy solutions¹. Photocatalytic technology has the potential to convert abundant solar energy into hydrogen energy². However, traditional water-splitting methods face limitations due to the bottleneck of the oxidation half-reaction. Although sacrificial agents such as triethanolamine, lactic acid, and Na₂S/Na₂SO₃ can enhance hydrogen production efficiency, their use results in energy waste of holes (h⁺) in the valence band (VB) of the catalyst, increases costs, emits CO₂, and exhibits slow oxidation kinetics^3,4. These factors significantly constrain the scalability of the technology. To address this issue, we achieved the dual objectives of electron-driven hydrogen production and hole-mediated organic conversion. Studies utilizing benzyl alcohol (BA) as a representative medium demonstrate that its hydroxyl group can effectively capture h⁺ and facilitate proton-driven hydrogen production through a dehydrogenation reaction. Concurrently, BA is selectively oxidized to benzaldehyde (BAD), which serves as a fundamental raw material in both the pharmaceutical and spice industries, thereby establishing an economic cycle model⁵.

The core of photocatalytic technology lies in the development of high-performance photocatalytic materials. Currently, sulfides⁶, metal and nonmetal oxides^7,8, and nitrogen-carbon compounds represent the mainstream materials in photocatalysis⁹. The two-dimensional hexagonal crystal system ZnIn₂S₄ has demonstrated significant potential for photocatalytic hydrogen production¹⁰. This structure possesses -SH surface active sites, demonstrating excellent physicochemical stability and an appropriate band structure. These characteristics enable the selective activation of C-H bonds and the dehydrogenation of organic substrates, which hold significant application value¹¹. However, the photocatalytic performance of ZnIn₂S₄ is constrained by the rapid recombination of electron-hole (e⁻-h⁺) pairs within the bulk phase. To mitigate this challenge, researchers have explored various strategies, including the introduction of cocatalysts¹², heterojunction engineering¹³, morphology control¹⁴, and elemental doping¹⁵, all aimed at enhancing the photocatalytic activity of ZnIn₂S₄. While these efforts have effectively alleviated the e⁻-h⁺ complexation on the material surface, the e⁻-h⁺ complexation in the bulk phase remains a significant challenge that requires urgent attention¹⁶. When the crystal structure symmetry of the photocatalyst is disturbed, the positive and negative charge centers within it are displaced, resulting in the formation of a dipole moment. This phenomenon subsequently induces a polarization electric field (PEF) within the bulk phase¹⁷. This finding introduces a novel strategy to enhance charge separation efficiency in the bulk phase of photocatalysts, which is anticipated to overcome existing technical bottlenecks and facilitate the further advancement of photocatalytic technology.

In this work, we address the challenge of insufficient bulk phase charge separation efficiency by leveraging the PEF induced by the intrinsic dipole moment in asymmetric Zn₃In₄S₉. Furthermore, we introduce an economically efficient co-catalyst, MoS₂, at the Zn₃In₄S₉ interface, which induces the formation of an interface dipole field (IDF) to establish a pathway for fast electron directional transfer. Experimental results indicate that the synergistic effect of PEF and IDF, induced by dipole-field engineering, significantly enhances the fast directional transfer of electrons from the conduction band (CB) of Zn₃In₄S₉ to MoS₂. The photocatalytic generation rates of H₂ and BAD for the 6%-MoS₂/Zn₃In₄S₉ composite reached notable values of 41.19 mmol g⁻¹ h⁻¹ and 43.33 mmol g⁻¹ h⁻¹, respectively. These rates are 11.8 and 12.2 times higher than those observed for Zn₃In₄S₉. The apparent quantum yield (AQY) of H₂ and BAD reached 36.6% ± 0.7% (n = 3 independent experiments) and 40.0% ± 0.3% (n = 3), respectively, when subjected to monochromatic light irradiation at a wavelength of 420 nm. The 6%-MoS₂/Zn₃In₄S₉ material demonstrates high long-term stability in photocatalytic H₂ and BAD production, retaining 93.8% and 87% of its initial productivity rates after 30 h. These results are competitive with those reported in previous studies. The synergistic effect of PEF and IDF facilitates the formation of fast electron transport channels. This insight provides a novel perspective on the directional movement of photo-generated charge carriers, clarifying the mechanism behind efficient synergistic photocatalytic H₂ production and the selective oxidation of BA to BAD.

Results

Photocatalyst characterization

The MoS₂ nanospheres formed a distinctive “balls-on-ball” structure on the surface of Zn₃In₄S₉ microspheres through the solvothermal method. (Fig. 1a). Initially, Zn²⁺ and In³⁺ ions, characterized by their high charge density and small ionic radius, preferentially adsorb onto the surface of MoS₂ nanospheres, demonstrating strong adsorption properties. Subsequently, S^2- ions participate by forming ligand bonds with the adsorbed ions, which initiates the nucleation of Zn₃In₄S₉. As the reaction progresses, additional ions are attracted to the nuclei, leading to the expansion of the clusters. This process culminates in the formation of Zn₃In₄S₉ nanoparticles, representing crystal growth. As these particles mature, Zn₃In₄S₉ microspheres coalesce into larger structures, ultimately arranging into a compact “balls-on-ball” configuration. To thoroughly investigate the surface morphology and structural characteristics of the catalysts, we employed a series of microscopy techniques, including scanning electron microscopy (SEM), transmission electron microscopy (TEM), high-resolution transmission electron microscopy (HRTEM), and atomic force microscopy (AFM). The results indicate that Zn₃In₄S₉ and MoS₂ display distinct nanosphere morphologies, with a notable size difference: ~3.5 μm for Zn₃In₄S₉ and around 500 nm for MoS₂ (Supplementary Fig. 1a–d, and Supplementary Fig. 2a, c). The SEM images of the 6%-MoS₂/Zn₃In₄S₉ reveal that the MoS₂ nanospheres are firmly adhered to the surface of the Zn₃In₄S₉ microspheres (Fig. 1b, and Supplementary Fig. 1e, f). The lattice spacings of Zn₃In₄S₉ (0.32 nm) and MoS₂ (0.62 nm), corresponding to the (102) and (002) crystal planes, respectively, were further elucidated in the HRTEM images of the 6%-MoS₂/Zn₃In₄S₉ (Fig. 1c, and Supplementary Fig. 2b, d)^18,19. High-angle annular dark field (HAADF) and energy dispersive X-ray spectroscopy (EDS) analyses revealed a uniform distribution of Zn, In, Mo, and S elements within the 6%-MoS₂/Zn₃In₄S₉ (Fig. 1d–g, and Supplementary Fig. 3). The AFM measurements indicated that the height fluctuation of Zn₃In₄S₉ microspheres ranged from 10 nm to −11.4 nm, while the height variation of the composite material introducing with 6% MoS₂ expanded to ±20 nm (Fig. 1h, i). Within a specific observation area, the maximum thickness of Zn₃In₄S₉ was measured at 7.4 nm, whereas the maximum thickness of the 6%-MoS₂/Zn₃In₄S₉ increased to 19.2 nm (Fig. 1j). These changes demonstrate that MoS₂ nanospheres are firmly adhered to the surface of Zn₃In₄S₉ microspheres, resulting in an overall increase in thickness, which aligns with the observations from SEM. Overall, these findings confirm the robust construction of the 6%-MoS₂/Zn₃In₄S₉ heterojunction interface, suggesting that the interactions at the interface may contribute to the establishment of fast electron transport channels²⁰.

Fig. 1. Synthesis and morphological characterization of photocatalysts.

a Schematic preparation of 6%-MoS₂/Zn₃In₄S₉ by the solvothermal method. (b) SEM image, (c) HRTEM image, and EDS elemental mapping of (d) Zn, (e) In, (f) Mo, and (g) S in 6%-MoS₂/Zn₃In₄S₉. AFM images of (h) ZCS and (i) 6%-MoS₂/Zn₃In₄S₉. j Corresponding height profiles of Zn₃In₄S₉ and 6%-MoS₂/Zn₃In₄S₉.

To systematically investigate the crystal structure, chemical bonding characteristics, and surface electronic states of the materials, this study employed a multi-scale characterization approach that integrates X-ray diffraction (XRD), Raman spectroscopy, in situ X-ray photoelectron spectroscopy (in situ XPS), X-ray absorption near-edge structure (XANES), and extended X-ray absorption fine structure (EXAFS). XRD analysis (Fig. 2a, and Supplementary Fig. 4a) revealed that MoS₂ exhibited distinct diffraction peaks at 13.7° (002), 33.3° (100), 39.4° (103), and 58.8° (110) (PDF #00-037-1492)¹⁹. The Zn₃In₄S₉ and y-MoS₂/Zn₃In₄S₉ composite exhibited characteristic peaks at 21.3° (006), 27.6° (102), 30.0° (104), 39.4° (108), 47.1° (110), 52.2° (116), and 55.6° (202) (PDF # 04-009-4784)²¹. Notably, only the diffraction signals of Zn₃In₄S₉ were observed in the y-MoS₂/Zn₃In₄S₉ composite, with no detectable peaks corresponding to MoS₂. This phenomenon is likely attributed to the low loading of MoS₂ and the physical shielding effect caused by the larger Zn₃In₄S₉ microspheres enveloping the smaller MoS₂ nanospheres. Raman spectra further verified the lattice vibrational properties of the composites (Supplementary Fig. 4b). The Raman peaks observed at 244, 300, and 354 cm⁻¹ correspond to the LO₁, LO₂ longitudinal optical vibrations, and A_1g vibrational modes of the crystal structure of Zn₃In₄S₉²². The FT-IR analysis revealed that MoS₂ exhibits three prominent absorption peaks at 1608 cm⁻¹, 1400 cm⁻¹, and 1135 cm^−123,24. Meanwhile, Zn₃In₄S₉ displays four distinct absorption peaks at 1605 cm⁻¹, 1400 cm⁻¹, 1043 cm⁻¹, and 637 cm^−125–27. Additionally, the absorption peak at 3555 cm⁻¹ is attributed to the elongation oscillation of O-H bonds in water (Supplementary Fig. 5)²⁸. In situ XPS analysis revealed the presence of Zn, In, Mo, and S on the surface of 6%-MoS₂/Zn₃In₄S₉, with calibration set at C 1 s 284.8 eV (Supplementary Fig. 6). The binding energy of 6%-MoS₂/Zn₃In₄S₉-dark indicated that the Zn 2p_3/2 level was observed at 1022.0 eV, In 3d_5/2 at 445.1 eV, Mo 3d_5/2 at 229.3 eV, and S 2p_2/3 at 161.9 eV, corresponding to the +2, +3, +4, and −2 valence states, respectively (Supplementary Fig. 7)^29,30. It is noteworthy that the binding energies of Zn 2p and In 3 d are elevated in the 6%-MoS₂/Zn₃In₄S₉-dark composite compared to Zn₃In₄S₉ (Supplementary Fig. 9). In contrast, the binding energies of Mo 3 d and S 2p are reduced, indicating the transfer of interfacial electrons from Zn₃In₄S₉ to MoS₂³¹. To verify the reproducibility of the binding energy shift trend, we conducted two independent in situ XPS measurements on 6%-MoS₂/Zn₃In₄S₉, ensuring strict calibration of the C 1 s reference peak at 284.8 eV. The results indicated a significant change in binding energy under light conditions. The Zn 2p_3/2 binding energy increased from 1022.0 eV in the dark to 1022.2 eV under light, while the In 3d_5/2 binding energy rose from 445.1 eV to 445.3 eV. In contrast, the Mo 3d_5/2 binding energy decreased from 229.3 eV to 229.1 eV, and the S 2p_3/2 binding energy also decreased from 161.9 eV to 161.8 eV (Supplementary Fig. 7). This reproducible pattern of binding energy shifts confirms the stable electron transfer from Zn₃In₄S₉ to MoS₂ (Supplementary Fig. 8)³². XANES and EXAFS techniques were employed to investigate the local structure of the materials. The variation in the photon energy of the MoS₂ absorption edge, relative to the Mo foil, indicates changes in its local coordination environment (Fig. 2b)³³. Additionally, the raw oscillatory signals in k-space from Mo foil, MoS₂, and 6%-MoS₂/Zn₃In₄S₉ are processed and Fourier transformed into R-space (Fig. 2c). In the R space EXAFS analysis, the following bond length changes were observed: the Mo-Mo bond length of Mo foil was 2.45 Å, while the Mo-S bond length of MoS₂ was 1.88 Å. In the 6%-MoS₂/Zn₃In₄S₉ composite material, the Mo-Mo bond length increases to 2.61 Å, and the Mo-S bond length decreases to 1.86 Å (Fig. 2d). Bond length variations arise mainly from electron redistribution driven by IDF³⁴. Fig. 2e compares the Zn K-edge XANES spectra for Zn foil, Zn₃In₄S₉, and 6%-MoS₂/Zn₃In₄S₉. Compared to Zn foil, the absorption edge of Zn₃In₄S₉ slightly shifts towards the high-energy region. This shift indicates that Zn predominantly exists in the +2 valence state. This observation is consistent with the conclusions drawn from XPS regarding its valence state^17,18. It is noteworthy that, compared to Zn₃In₄S₉, the white line peak intensity of the 6%-MoS₂/Zn₃In₄S₉ composite material has decreased. This decrease can be attributed to the introduction of MoS₂, which results in a reduction of the coordination number of Zn sites and weakens the density of empty states³⁵. The EXAFS fitting analysis in both K-space and R-space reveals that the 6%-MoS₂/Zn₃In₄S₉ exhibits a distinct Zn-S coordination peak at 1.85 Å, which is significantly more pronounced than the Zn-Zn metallic bond peak observed at 2.26 Å in the Zn foils (Fig. 2f, g). This observation suggests the formation of stronger covalent interactions between the Zn and S atoms. (Supplementary Table 1). Compared to Zn₃In₄S₉, the reduced Zn-S bond length in 6%-MoS₂/Zn₃In₄S₉ suggests a higher Zn-S bond energy and enhanced chemical stability of the material³⁵. Further analysis of the Zn and Mo K-edge EXAFS signals through wavelet transform (WT) reveals that the WT contour plots of 6%-MoS₂/Zn₃In₄S₉ exhibit significant differences compared to those of MoS₂ and Zn₃In₄S₉ (Fig. 2h, i, and Supplementary Fig. 10). This significant change indicates the formation process of the IDF, characterized by the directional migration of electrons from Zn₃In₄S₉ to MoS₂³⁵. The redistribution of interface charges not only optimizes the local coordination environment of the material but also establishes a channel for the transport of fast charge carriers during photocatalytic reactions³⁵. To further investigate the band structure of photocatalysts, we conducted analyses using ultraviolet-visible (UV-vis) diffuse reflectance spectroscopy, Mott Schottky (MS) testing, and XPS-VB spectroscopy measurement. In comparison to Zn₃In₄S₉, the 6%-MoS₂/Zn₃In₄S₉ demonstrates improved performance in visible light absorption. This enhancement is characterized by a slight redshift of the absorption edge and an increased capacity to absorb a broader spectrum of visible wavelengths (Supplementary Fig. 11). The flat-band potential of Zn₃In₄S₉, measured at −0.92 V, was evaluated using MS spectroscopy (Supplementary Fig. 12). This value was then converted to a reversible hydrogen electrode (RHE) potential through the application of the Nernst equation ($E_{RHE}=E_{Ag/Cl}+0.05916pH+E_{Ag/Cl}^0$)²³. The CB potential of Zn₃In₄S₉ RHE was calculated to be −0.3 eV. The VB potential was determined using the energy band formula ($E_{VB}=E_{CB}+E_g$), yielding a value of 2.27 eV, which is consistent with the findings from XPS-VB spectroscopy (Supplementary Fig. 13a)³⁶. The work function of 6%-MoS₂/Zn₃In₄S₉ (Φ = 2.09 eV) is lower than that of Zn₃In₄S₉ (Φ = 2.31 eV), indicating that the interfacial modification with 6% MoS₂ effectively reduces the energy barrier and facilitates the formation of fast electron transport channels (Supplementary Fig. 13b, d)³⁷. This observation confirms the establishment of an IDF. The same methodology was employed to calculate the energy band structures of the other photocatalysts, culminating in the generation of the energy band diagrams for each photocatalyst. (Supplementary Fig. 14, Supplementary Table 2). To evaluate the physical properties of the materials, we examined the specific surface area and pore size distribution of the samples (Supplementary Fig. 15, Supplementary Table 3). All the photocatalysts demonstrated type IV isotherms, while Zn₃In₄S₉ and 6%-MoS₂/Zn₃In₄S₉ showed H4 hysteresis loops as well as mesoporous structures. Significantly, 6%-MoS₂/Zn₃In₄S₉ revealed a greater specific surface area, offering additional active sites for the photocatalytic processes.

Fig. 2. Atomic-scale structural and electronic properties characterization of photocatalysts.

a XRD of MoS₂, Zn₃In₄S₉, and y-MoS₂/Zn₃In₄S₉. b Compare the Mo K-edge XANES spectra among Mo foil, MoS₂, and 6%-MoS₂/Zn₃In₄S₉ composites. Present (c) K-space and (d) R-space EXAFS fitting spectra for Mo foil, MoS₂, and 6%-MoS₂/Zn₃In₄S₉. e Contrast the Zn K-edge XANES spectra of Zn₃In₄S₉ and 6%-MoS₂/Zn₃In₄S₉ with the reference spectrum of Zn foil. (f) K-space and (g) R-space EXAFS fitting spectra for Zn foil, Zn₃In₄S₉, and 6%-MoS₂/Zn₃In₄S₉. Illustrate the WT-EXAFS analysis for the (h) Mo and (i) Zn K-edges in 6%-MoS₂/Zn₃In₄S₉.

Dipole-synergistic driving of charge dynamics

We employed density functional theory (DFT) on idealized structural models to investigate the possible formation mechanism of the PEF in the bulk phase of Zn₃In₄S₉. Additionally, we explored the IDF generated at the Zn₃In₄S₉ interface modified with MoS₂, with calculations suggesting a potential fast electron transport pathway established by this synergistic effect. The electrostatic potential (V) of Zn₃In₄S₉ along the [001] polarity axis exhibits a potential difference (ΔV₁) of 2.51 eV and a dipole moment (μ₁) of 2.95 eÅ (Fig. 3a, and Supplementary Table 4). This represents a significant enhancement compared to ZnIn₂S₄, which shows a ΔV₄ of 2.21 eV and a μ₃ of 2.61 eÅ. This indicates that the bulk-phase asymmetry of the Zn₃In₄S₉ crystal exceeds that of ZnIn₂S₄ (Supplementary Figs. 16–17). Consequently, the bulk-phase environment of Zn₃In₄S₉ is more conducive to suppressing the ineffective complexation of photogenerated e⁻-h⁺ pairs. The significant abrupt change in electrostatic potential at the interface between the [In-S] and [Zn-S] layers indicates active charge transfer between the two layers²². This asymmetric structure leads to spontaneous polarization within the bulk of Zn₃In₄S₉, forming a PEF. The PEF drives electrons across the interface from the [Zn-S] layer into the [In-S] layer, effectively suppressing the ineffective recombination of e⁻-h⁺ pairs within the Zn₃In₄S₉ bulk phase. It is noteworthy that the [Mo-S] and [In-S] interfaces of MoS₂/Zn₃In₄S₉ exhibit a significant potential difference (ΔV₃ = 9.63 eV). This observation suggests the presence of substantial charge transfer between the [Mo-S] and [In-S] interfaces (Fig. 3b). Furthermore, it indicates that the MoS₂-modified Zn₃In₄S₉ interface generates an IDF, which provides a foundation for the establishment of fast electron directional transport channels. To verify the direction of electron transfer at the MoS₂ and Zn₃In₄S₉ interface, we conducted charge density difference calculations. The color distribution visually illustrates the regions of electron enrichment (yellow) and depletion (green) at the MoS₂/Zn₃In₄S₉ interface (Fig. 3c). Electrons are enriched on the MoS₂ side, while electron depletion is evident on the Zn₃In₄S₉ side, thereby confirming the transfer of electrons from Zn₃In₄S₉ to MoS₂. To accurately quantify the electron gain and loss between each atom, we employed the Bader charge analysis method (Fig. 3d, e). The results indicate that the Bader charges of Zn, In, and S atoms in Zn₃In₄S₉ are −0.017 eV, −0.174 eV, and −0.095 eV, respectively, all demonstrating characteristics of electron loss. In contrast, the Bader charges of Mo and S atoms in MoS₂ are 0.073 eV and 0.373 eV, respectively, which signifies a strong electron capture capability. Based on these findings, we have designed a schematic diagram that illustrates the mechanism of fast electron directional transport (Fig. 3f). The asymmetric structure of Zn₃In₄S₉ initially establishes a PEF, which facilitates the transfer of electrons from the [Zn-S] layer to the [In-S] layer. Subsequently, the IDF generated at the interface between MoS₂ and Zn₃In₄S₉ further accelerates the migration of electrons towards MoS₂. In summary, these observational results provide robust support for our proposed synergistic strategy. integrating the PEF of Zn₃In₄S₉ with the IDF induced by MoS₂ to achieve fast charge dynamics. Piezoresponse Force Microscopy (PFM) has emerged as a significant experimental tool for probing the existence of PEF, owing to its reversible linear relationship between electric field and mechanical deformation³⁸. The PFM results demonstrate that ZnIn₂S₄, Zn₃In₄S₉, and the 6%-MoS₂/Zn₃In₄S₉ composite exhibit typical piezoelectric behavior, characterized by distinct butterfly-shaped amplitude loops and well-defined positively and negatively polarized regions (Fig. 3g, h, Supplementary Fig. 18). This observation confirms the presence of internal PEF in these materials. Additionally, the slope of the amplitude loop serves as an indicator of the strength of piezoelectric polarization¹⁶. Zn₃In₄S₉ (1.61 mV/V) exhibits stronger spontaneous polarization compared to ZnIn₂S₄ (1.21 mV/V). Compared to Zn₃In₄S₉, the 6%-MoS₂/Zn₃In₄S₉ demonstrates a polarization strength that is 3.52 times greater than that of Zn₃In₄S₉ (Fig. 3i).

Fig. 3. Revealing the synergistic effect of PEF and IDF in driving charge separation.

The distribution of electrostatic potential (V) along the [001] polarity axis is presented for (a) Zn₃In₄S₉ and (b) MoS₂/Zn₃In₄S₉. c Differential Charge Density and (d, e) Bader Charge Analysis of MoS₂/Zn₃In₄S₉. f Schematic diagram that elucidates the electron transfer mechanism involved in the synergistic effects of PEF and IDF in MoS₂/Zn₃In₄S₉. Butterfly amplitude loops and phase curves of (g) Zn₃In₄S₉ and (h) 6%-MoS₂/Zn₃In₄S₉. i Comparison of the slopes of the hysteresis loop of ZnIn₂S₄, Zn₃In₄S₉, and 6%-MoS₂/Zn₃In₄S₉.

To fundamentally investigate the mechanism of constructing electron highways through IDF engineering, we systematically characterized the materials using ultraviolet-assisted Kelvin probe force microscopy (KPFM) under 365 nm illumination. The analysis of the contact potential difference (CPD) for Zn₃In₄S₉ under both dark and illuminated conditions revealed a substantial increase in CPD, rising from 4.5 mV in the dark to 414.6 mV in the light (Fig. 4a, b). The significant enhancement of CPD indicates that, under light excitation, the intrinsic dipole moment in the asymmetric structure of the Zn₃In₄S₉ bulk phase spontaneously generates a PEF. This phenomenon greatly accelerates the effective migration of photo-generated h⁺ from the bulk phase of Zn₃In₄S₉ to its surface³⁹. The efficiency of charge carrier separation can be quantitatively assessed by calculating the CPD difference between light and dark conditions (ΔCPD = CPD_light - CPD_dark)^22,40. Comparative analysis revealed that the 6%-MoS₂/Zn₃In₄S₉ exhibited a ΔCPD value of 451.6 mV, which is significantly higher than that of Zn₃In₄S₉, recorded at 410.1 mV (Fig. 4c–f). The optimal loading of MoS₂ at 6% on Zn₃In₄S₉ leads to the formation of an IDF that effectively suppresses interfacial e⁻-h⁺ recombination, while simultaneously establishing an fast unidirectional charge transmission channel. The surface potentials of Zn₃In₄S₉ and 6%-MoS₂/Zn₃In₄S₉ under dark conditions are 3 mV and 12 mV, respectively (Supplementary Fig. 19). Based on these surface potential values, we can utilize specific calculation formulas (detailed formulas are provided in the Supplementary Information) to compute their surface charge densities (ρ) and PEF intensities⁴¹. Furthermore, when calculations are performed using the Zeta potential in conjunction with the Gouy-Chapman equation, we find that the surface charge densities of Zn₃In₄S₉ and 6%-MoS₂/Zn₃In₄S₉ are 28.5 mC m^-2 and 41.1 mC m^-2, respectively (Supplementary Fig. 20a). Therefore, the PEF intensity of 6%-MoS₂/Zn₃In₄S₉ is 2.41 times that of Zn₃In₄S₉ (Supplementary Fig. 20b). KPFM analysis reveals that charge dynamics are driven by the synergistic action of the PEF from the intrinsic dipole moment in asymmetric Zn₃In₄S₉ and the IDF induced by MoS₂. The carrier kinetic behaviors of Zn₃In₄S₉ and 6%-MoS₂/Zn₃In₄S₉ were systematically investigated using femtosecond transient absorption spectroscopy (fs-TAS) under pump light excitation at 340 nm. As shown in Fig. 4g, h, Zn₃In₄S₉ and 6%-MoS₂/Zn₃In₄S₉ exhibit significant negative ground state bleaching (GSB) signals in the 450-510 nm range. This phenomenon originates from the energy dissipation associated with the compounding process of photoexcited e^--h⁺ pairs⁴². After laser excitation, the ΔA signal peaks rapidly within 1 ps and subsequently decays gradually, reflecting the relaxation process of the excited state⁴³. This dynamic evolution encompasses the migration of localized carriers to the excited state, which ultimately transforms into free carriers. The pseudo-color mapping further validated the time-domain evolution characteristics of the fs-TAS (Fig. 4i, j). The carrier lifetimes of Zn₃In₄S₉ and 6%-MoS₂/Zn₃In₄S₉ were determined to exhibit both fast and slow components, respectively, by fitting the decay curves at 475 nm using a double-exponential function. The fast decay lifetime (τ₂) is associated with the rapid recombination of excited state electrons and h⁺ within the bulk phase of Zn₃In₄S₉, while the slow decay lifetime (τ₁) is attributed to the trans-interfacial transfer of electrons from the CB of Zn₃In₄S₉ to MoS₂⁴⁴. The τ₁ (526.9 ps and 677.02 ps) and the τ₂ (11940.2 ps and 18191.69 ps) for Zn₃In₄S₉ and 6%-MoS₂/Zn₃In₄S₉ (Fig. 4k, l, and Supplementary Table 5), respectively. It is worth noting that the average decay lifetime (τ_ave) of 6%-MoS₂/Zn₃In₄S₉ (τ_ave = 10674.63 ps) is significantly prolonged compared to Zn₃In₄S₉ (τ_ave = 7374.87 ps). The fluorescence lifetime range of 6%-MoS₂/Zn₃In₄S₉, spanning from 694 to 1021 ps, is the broadest when compared to that of Zn₃In₄S₉ (686 to 739 ps) and 18%-MoS₂/Zn₃In₄S₉ (539 to 927 ps). Notably, the maximum value exceeds 1000 ps (Supplementary Fig. 21). fs-TAS and fluorescence lifetime testing demonstrated that an optimal loading of MoS₂ (6%) induces an IDF at the Zn₃In₄S₉ interface, effectively suppressing e⁻-h⁺ recombination at this interface and establishing an fast directional transport channel for charges to MoS₂. Consequently, this significantly prolongs the lifetime of photogenerated charge carriers. To investigate the migration efficiency of interfacial photogenerated carriers in gradient-MoS₂ decorated asymmetric Zn₃In₄S₉ structures, we conducted a series of tests including electrochemical impedance spectroscopy (EIS), surface photocurrent (SPC), linear scanning voltammetry (LSV), photoluminescence spectroscopy (PL), time-resolved photoluminescence spectroscopy (TRPL), and cyclic voltammetry (CV). Compared to ZnIn₂S₄, Zn₃In₄S₉ exhibits a smaller EIS semicircle radius, indicating a reduced interfacial charge transfer resistance. Furthermore, Zn₃In₄S₉ demonstrates a significantly enhanced photocurrent response, suggesting improved efficiency in photogenerated carrier generation and separation, which facilitates rapid charge migration (Supplementary Fig. 22a, b). The EIS results indicate that the semicircle radius of the 6%-MoS₂/Zn₃In₄S₉ in the low-frequency region is significantly reduced compared to that of Zn₃In₄S₉ and y-MoS₂/Zn₃In₄S₉ (Supplementary Fig. 22a, Supplementary Fig. 23a). This finding suggests that an appropriate amount of MoS₂ loading can greatly reduce the interface resistance of Zn₃In₄S₉, thereby accelerating the transfer of electrons from the Zn₃In₄S₉ interface to MoS₂⁴⁵. The 6%-MoS₂/Zn₃In₄S₉ composite demonstrates the highest photocurrent density, attributed to the formation of an IDF induced by the optimal amount of (6%) MoS₂ (Supplementary Fig. 22b, and Supplementary Fig. 23b). This IDF significantly enhances the separation and transport of photogenerated carriers⁴⁶. The LSV results demonstrated that at a photocurrent density of 20 mA/cm², the hydrogen precipitation overpotential of 6%-MoS₂/Zn₃In₄S₉ was −1.63 V. This value represents the lowest hydrogen precipitation overpotential among Zn₃In₄S₉ and other composite photocatalysts, indicating a more rapid hydrogen evolution reaction (HER) rate. (Supplementary Fig. 22c, and Supplementary Fig. 23c)⁴⁷. To investigate the e^--h⁺ coupling and lifetime mechanism in greater depth, we conducted PL and TRPL tests using an excitation wavelength of 525 nm. The 6%-MoS₂/Zn₃In₄S₉ composite exhibited the lowest fluorescence signal intensity (Supplementary Fig. 22d, and Supplementary Fig. 24a) and the longest e^--h⁺ pairs lifetime (5.47 ns) compared to the other samples: Zn₃In₄S₉ (1.51 ns), 9%-MoS₂/Zn₃In₄S₉ (3.30 ns), and 18%-MoS₂/Zn₃In₄S₉ (2.82 ns) (Supplementary Fig. 22e, and Supplementary Fig. 24b). The IDF induced by the optimal amount of MoS₂ (6%) in the Zn₃In₄S₉ interface establishes a highway for electron transport, significantly reducing the recombination of e^--h⁺ pairs at the Zn₃In₄S₉ interface and consequently extending their lifetime (Supplementary Table 6)^13,48. The CV evaluation revealed that the 6%-MoS₂/Zn₃In₄S₉ exhibited the highest electrochemically active surface area, measuring 0.31 μF (Supplementary Fig. 22f, Supplementary Fig. 23d), in comparison to other samples: Zn₃In₄S₉ (0.11 μF), 3%-MoS₂/Zn₃In₄S₉ (0.254 μF), 9%-MoS₂/Zn₃In₄S₉ (0.250 μF), and 18%-MoS₂/Zn₃In₄S₉ (0.22 μF). This observation was recorded at scanning rates ranging from 20 to 90 mV/s (Supplementary Fig. 25), indicating that the 6%-MoS₂/Zn₃In₄S₉ composite possesses a greater number of active sites and enhanced photocatalytic activity.

Fig. 4. Dipole-engineered fast charge transport pathways.

Three-dimensional CPD of (a, b) Zn₃In₄S₉ and (c, d) 6%-MoS₂/Zn₃In₄S₉ under dark and light conditions. The line scan surface potential distribution of (e) Zn₃In₄S₉ and (f) 6%-MoS₂/Zn₃In₄S₉ under dark and light conditions. fs-TAS monitors the photogenerated charge dynamics of (g) Zn₃In₄S₉ and (h) 6%-MoS₂/Zn₃In₄S₉. 3D pseudo-color representations of fs-TAS over (i) Zn₃In₄S₉ and (j) 6%-MoS₂/Zn₃In₄S₉ at 340 nm excitation. Femtosecond resolved transient absorption kinetics and the corresponding fitting results for (k) Zn₃In₄S₉ and (l) 6%-MoS₂/Zn₃In₄S₉.

Solar hydrogen coupled with alcohol photoreforming

To thoroughly investigate the photocatalytic activity, we assessed the performance of various photocatalysts in the dehydrogenation reaction and the synergistic conversion of BA to BAD. The H₂ and BAD production rates of Zn₃In₄S₉ were measured at 3.49 mmol g⁻¹ h⁻¹ and 3.55 mmol g⁻¹ h⁻¹, which are 2.3 and 1.5 times higher than those of ZnIn₂S₄, respectively. (Fig. 5a, d). In contrast, the 6%-MoS₂/Zn₃In₄S₉ composite, modified with 6% MoS₂, exhibited significantly enhanced photocatalytic activity, with H₂ and BAD production rates elevated to 41.19 mmol g⁻¹ h⁻¹ and 43.33 mmol g⁻¹ h⁻¹, representing increases of 11.8 and 12.2 times compared to those of Zn₃In₄S₉, respectively. (Supplementary Table 7). This is attributed to the introduction of a moderate amount (6%) of MoS₂, which induces an IDF at the Zn₃In₄S₉ interface. This IDF establishes an fast electron transmission channel from the Zn₃In₄S₉ interface to MoS₂, significantly suppressing the ineffective recombination of e^--h⁺ pairs at the Zn₃In₄S₉ interface. The performance is competitive with that of many reported photocatalysts in photocatalytic BA dehydrogenation systems (Supplementary Fig. 26, and Supplementary Table 8). To elucidate the synergistic mechanism of PEF-IDF, we constructed a 6%-MoS₂/ZnIn₂S₄ control system, with structural integrity confirmed through XRD and XPS characterization (Supplementary Fig. 27). The catalytic activity of this system, measured in terms of H₂ production (5.24 mmol g⁻¹ h⁻¹) and BAD yield (6.71 mmol g⁻¹ h⁻¹), was found to be only 12.6%–15.5% of that observed for 6%-MoS₂/Zn₃In₄S₉ (Supplementary Fig. 28). The complete inactivity of pure MoS₂ conclusively demonstrates the dominant role of the bulk photoelectrochemical functionality of Zn₃In₄S₉. Notably, the physically mixed 6%-MoS₂/ZnIn₂S₄ exhibited a 96% reduction in hydrogen production activity (1.67 mmol g⁻¹ h⁻¹) compared to the sample synthesized via solvothermal methods. This significant order-of-magnitude difference provides compelling evidence for the necessity of IDF formation, ultimately elucidating the fundamental mechanism underlying the synergy between PEF and IDF in catalytic enhancement. To validate the advantages of MoS₂, we prepared a control sample of 6%-Co₉S₈/Zn₃In₄S₉. XRD and XPS characterizations confirmed its successful preparation (Supplementary Figs. 29, 30). However, its photocatalytic performance was significantly inferior to that of the 6%-MoS₂/Zn₃In₄S₉. Specifically, the H₂ yield (5.12 mmol g⁻¹ h⁻¹), BAD yield (3.28 mmol g⁻¹ h⁻¹), and BAD selectivity (53.7%) were all much lower (Supplementary Fig. 31). This difference is primarily attributed to weaker interfacial dipole effects and an insufficient capability for proton reduction, which underscores the irreplaceable role of MoS₂ in interface engineering and catalytic activity. We extracted 100 μL from the 6%-MoS₂/Zn₃In₄S₉ solution at 30-min intervals during the reaction process for liquid chromatography analysis. The results indicated a significant decrease in the BA concentration over time, while the BAD concentration continuously increased (Supplementary Fig. 32). The infrared thermal imaging of 6%-MoS₂/Zn₃In₄S₉ over time under 300 W xenon lamp irradiation (Supplementary Fig. 33). The cooling system of the photoreactor ensures a uniform heat distribution throughout the solution, resulting in minimal temperature variation. Consequently, the 6%-MoS₂/Zn₃In₄S₉ does not exhibit significant heat accumulation, and the reaction process is devoid of any photothermal effects. The experimental results indicated that the hydrogen production rate remained relatively constant across varying temperatures, suggesting that temperature is not a critical factor influencing catalytic performance. To investigate the key active species involved in the selective oxidation of BA to BAD, we conducted control experiments under an argon atmosphere and utilized free radical scavengers to eliminate specific active species (Fig. 5b, Supplementary Table 9). The experimental results indicated that the incorporation of triethanolamine (TEOA) effectively eliminated h⁺ and significantly decreased the rates of H₂ production and BAD synthesis⁴⁹. This suggests that h⁺ plays a critical role in the process. In contrast, the introduction of K₂S₂O₈ (KPS) effectively removed electrons and significantly inhibited H₂ production⁵⁰. This observation confirms the critical role of photogenerated electrons in the reduction of H⁺ protons to H₂. The introduction of benzoquinone (BQ) and tert-butanol (t-BuOH) effectively eliminates •O²⁻ and •OH, respectively, while having minimal impact on hydrogen production and BAD synthesis^50,51. This observation suggests that h⁺ and electrons serve as the primary active species within the reaction system. To evaluate the photocatalytic activity of the 6%-MoS₂/Zn₃In₄S₉ photocatalysts with greater accuracy and comprehensiveness, we designed a series of controlled-variable experiments utilizing band-pass filters at various wavelengths (λ = 350, 380, 420, 450, 475, and 500 nm). The results indicated that the production of both H₂ and BAD reached their peaks at 1126.76 μmol and 1240.10 μmol, respectively, under 420 nm irradiation. Meanwhile, the apparent quantum yields (AQY) were determined to be 36.6 ± 0.7% for H₂ and 40.0 ± 0.3% for BAD (n = 3). (Fig. 5c, Supplementary Table 10). It is noteworthy that in three independently repeated AQY experiments, the AQY values for H₂ and BAD production exhibited notable stability, further corroborating the high activity of 6%-MoS₂/Zn₃In₄S₉ (Supplementary Fig. 34). We fully recognize the potential impact of temperature on the kinetics of the photocatalytic reaction under strictly controlled conditions (λ > 420 nm illumination, 6%-MoS₂/Zn₃In₄S₉). We assessed the H₂ production rate at temperature intervals of 4 °C, 8 °C, 12 °C, 16 °C, 20 °C, and 24 °C (Fig. 5e). To highlight the advantages of BA, we evaluated the H₂ production rate of the 6%-MoS₂/Zn₃In₄S₉ photocatalyst under identical conditions in sacrificial systems (Fig. 5f), including pure water cracking, Na₂S/Na₂SO₃, TEOA, and lactic acid⁵². The results indicated that the H₂ production rate of the reaction system utilizing BA was optimal, with its H₂ production performance improved by factors of 27.66, 6.88, 5.92, and 4.7, respectively, when compared to the other systems. This finding suggests that BA, as a specialized h⁺ transport medium, can significantly enhance the utilization efficiency of photogenerated h⁺. The 6%-MoS₂/Zn₃In₄S₉ demonstrates broad applicability for the oxidative dehydrogenation of various aromatic alcohols, facilitating the efficient production of corresponding aromatic aldehydes alongside H₂ evolution (Supplementary Fig. 35, and Supplementary Table 11). Additionally, we assessed the stability of the 6%-MoS₂/Zn₃In₄S₉. The photocatalytic system demonstrated excellent stability, maintaining 93.8% and 87% of its initial H₂ and BAD production rates, respectively, after 10 cycles (Fig. 5g). We analyzed the 6%-MoS₂/Zn₃In₄S₉ after the 10th cycle using XRD, Raman, and XPS (Supplementary Figs. 36–38). The results indicated that the position and intensity of the peaks remained consistent before and after the reaction. This study further confirms the high efficiency and stability of the 6%-MoS₂/Zn₃In₄S₉ photocatalyst.

Fig. 5. Solar hydrogen production synergized with photocatalytic aromatic alcohol conversion.

a Photocatalytic BA dehydrogenation rates of various photocatalysts (300 W Xe lamp; λ > 420 nm; light intensity 100 mW/cm²; catalyst 20 mg; solvent 50 mL water with 3 mL BA; irradiation time 3 h; sampling was performed every 30 min). b In the presence of different free radical scavengers, the photocatalytic production rates of H₂ and BAD by 6%-MoS₂/Zn₃In₄S₉ were tested after 3 h of light irradiation. c H₂ and BAD yields of 6%-MoS₂/Zn₃In₄S₉ at distinct wavelengths (λ = 350, 380, 420, 450, 475, and 500 nm) and the corresponding AQY of H₂ and BAD. d The photocatalytic production rates of H₂ and BAD by various photocatalysts and the selective efficiency of BAD. e H₂ production rates for temperature intervals of 4 °C, 8 °C, 12 °C, 16 °C, 20 °C and 24 °C. f Comparison of H₂ production rates of 6%-MoS₂/Zn₃In₄S₉ photocatalysts under different sacrificial agent systems (pure water, Na₂S/Na₂SO₃, TEOA, lactic acid) and BA photosynthesis systems. g Cycling test of H₂ and BAD production rates over 6%-MoS₂/Zn₃In₄S₉ photocatalysts (3 h per cycle, 10 cycles of 30 h).

Cascaded C-H activation pathway

To elucidate the mechanistic insights into synergistic photocatalytic H₂ evolution coupled with cascaded C-H activation during BA to BAD conversion, we conducted an in situ electron paramagnetic resonance (in situ EPR) test to monitor the fluctuations in the concentrations of reactive electron (e^-), h⁺, and α-hydroxybenzoyl (C_α) radical intermediates throughout the reaction course. We selected TEMPO as an efficient spin-trapping agent to capture and measure the photogenerated e^- and h⁺ under varying conditions. In the absence of light, TEMPO displays a distinct EPR signal characterized by a typical triplet state spectral ratio of 1:1:1^53,54. Upon introducing light into an aqueous system containing the catalyst with BA, we observed that the decay of the EPR signal was most pronounced in the experimental group utilizing 6%-MoS₂/Zn₃In₄S₉ (Fig. 6a). This observation suggests that this 6%-MoS₂/Zn₃In₄S₉ is capable of generating the highest concentration of reactive e^- under light excitation, which subsequently reduce TEMPO to TEMPOH. Furthermore, the photogenerated h⁺ exhibited the capacity to oxidize TEMPO to TEMPO⁺⁵⁵. Notably, the 6%-MoS₂/Zn₃In₄S₉ demonstrated the most pronounced reduction in EPR signals, thereby further confirming its high efficiency in generating reactive h⁺ under light conditions (Fig. 6b). To verify the key intermediate in the selective oxidation of BA to generate BAD, we introduced DMPO as an additional spin-trapping agent. This approach successfully captured the EPR signals of the α-hydroxyphenyl radical (•CH(OH)Ph). Six characteristic peaks were observed in the EPR spectra, coinciding with those of the DMPO-C_α radical adduct⁵⁶. This not only confirms the generation of the -CH(OH)Ph radical but also highlights the central role of direct oxidation of BA by the h⁺ in this reaction pathway. Under light conditions, the 6%-MoS₂/Zn₃In₄S₉ exhibited a significantly stronger DMPO-C_α radical signal (Fig. 6c), thereby demonstrating its high efficacy in facilitating the selective oxidation of BA to BAD. We conducted an in-depth analysis of the free energy changes of intermediate species and transition states between Zn₃In₄S₉ and MoS₂/Zn₃In₄S₉ using DFT calculations (Fig. 6d). The adsorption energy of Zn₃In₄S₉ modified with MoS₂ for BA molecules decreased from −1.55 eV to −0.94 eV (Supplementary Table 12), which significantly diminishes the adsorption effect. This alteration facilitates the reduction of the energy barrier (ΔG₂) in the rate-determining step of the BA oxidation reaction, thereby accelerating the overall catalytic reaction rate. A deuterium labeling experiment has been meticulously designed to investigate the source of hydrogen during the photocatalytic dehydrogenation reaction of BA. The primary innovation of this experiment lies in the utilization of D₂O (deuterated water) as the reaction medium, replacing conventional H₂O (pure water), to effectively trace the migration trajectories of hydrogen and deuterium atoms. The gas generated under these conditions served as a hydrogen source and participated in the hydrogenation reduction reaction of 1-vinylnaphthalene (C₁₀H₇CH = CH₂). Upon completion of the experiment, we conducted a gas chromatography-mass spectrometry (GC-MS) analysis of the solution in reactor 2 (Fig. 6e, Supplementary Fig. 39). The analytical results indicated the formation of four distinct products during the hydrogenation of 1-vinylnaphthalene (C₁₀H₇CHCH₂) to 1-ethylnaphthalene: C₁₀H₇CH₂CH₃ (I, m/z = 156), C₁₀H₇CH₂CH₂D (II, m/z = 157), C₁₀H₇CHDCH₃ (III, m/z = 157), and C₁₀H₇CHDCH₂D (IV, m/z = 158)²². Products I and II yield C₁₀H₇CH₂⁺ (m/z = 141) following the removal of alkyl groups (-CH₃ and -CH₂D), whereas products III and IV generate C₁₀H₇CHD⁺ (m/z = 142) after the corresponding alkyl groups are removed. These products provide direct verification of the effective integration of hydrogen atoms released during the oxidation of BA and deuterium atoms supplied by D₂O in the hydrogenation reaction, offering conclusive evidence for the multiple sources of hydrogen. Finally, we conducted liquid chromatography analysis on the solution in reactor 1, as well as on the BA and BAD standard samples, to confirm the peak times of each component (Supplementary Fig. 40). The results indicated that the peak times of BA and BAD in the solution from reactor 1 were highly consistent with those of the standard samples, and a significant quantity of BAD was detected in the solution from reactor 1. Through an in-depth analysis of theoretical calculations and experimental results, this study effectively elucidates the mechanism of fast charge transfer driven by the synergistic effects of PEF and IDF in photocatalytic hydrogen production and the photoreforming of aromatic alcohols (Fig. 6f). The intrinsic dipole of asymmetric Zn₃In₄S₉ generates a PEF that effectively suppresses bulk-phase charge recombination. Following photoexcitation, the PEF facilitates the rapid migration of electrons to the Zn₃In₄S₉ interface. Pure Zn₃In₄S₉ exhibits limited photocatalytic activity due to insufficient charge separation solely driven by its bulk PEF, coupled with sluggish surface reaction kinetics for both water and BA oxidation. Notably, MoS₂ induces an IDF at the Zn₃In₄S₉ interface, creating an fast directional electron transport channel that allows for the efficient injection of electrons into MoS₂ while retaining h⁺ in the VB of Zn₃In₄S₉. The spatially separated h⁺ exhibit strong oxidizing properties, which enable them to cleave the α-C-H bond in BA, generating the •CH(OH)Ph radical and a proton (H). Following this, the O-H bond experiences hole-induced oxidation, resulting in its cleavage and the release of another proton (H), ultimately yielding BAD through a double-hole oxidation process. The enriched electrons on MoS₂ can synergistically reduce protons released from BA oxidation and those generated from water dissociation, thereby achieving efficient simultaneous hydrogen production. This bifunctional charge utilization mechanism, which integrates directed fast charge transport and cascading redox reactions, establishes a paradigm for the development of high-performance artificial photosynthetic systems.

Fig. 6. Unveiling the cascaded C–H activation mechanism.

In situ EPR analysis of (a) e⁻, (b) h⁺, and (c) C_α radical concentrations in different catalysts. d DFT theoretical calculation model for Gibbs free energy changes of adsorbed species *C₆H₅CH₂OH, *C₆H₅CHOH, and *C₆H₅CHO on Zn₃In₄S₉ and 6%-MoS₂/Zn₃In₄S₉. e Deuterium labeling experiments investigating the hydrogen production mechanism and BA oxidation in 6%-MoS₂/Zn₃In₄S₉. f Dipole-field catalytic mechanism for hydrogen production and high-value chemical synthesis in 6%-MoS₂/Zn₃In₄S₉.

Discussion

In summary, we elucidate the mechanism of carrier dynamics regulation facilitated by the synergistic action of the PEF generated by the intrinsic dipole moment within the bulk phase of asymmetric Zn₃In₄S₉ and the IDF induced by MoS₂. The systematic study that combines PFM analysis and fs-TAS demonstrates that the spontaneous polarization of dipole moments in the asymmetric structure of Zn₃In₄S₉ generates a PEF, which significantly suppresses the ineffective recombination of e^- and h⁺ in the bulk phase of Zn₃In₄S₉. Notably, the Zn₃In₄S₉ interface, modified with 6% MoS₂, induces the formation of IDF, which facilitates the construction of fast charge transfer channels directed towards MoS₂. The synergistic effect of PEF and IDF enables the 6%-MoS₂/Zn₃In₄S₉ to achieve notable photocatalytic performance, with hydrogen evolution and BAD production rates of 41.19 mmol g⁻¹ h⁻¹ and 43.33 mmol g⁻¹ h⁻¹, respectively. These rates are 11.8 and 12.2 times higher than Zn₃In₄S₉. Notably, the 6%-MoS₂/Zn₃In₄S₉ achieves 36.6 ± 0.7% AQY for H₂ and 40.0 ± 0.3% for BAD (n = 3) at 420 nm, with high stability ( > 90% activity retention after 30 h). Our research findings provide significant insights into the fundamental mechanisms by which dipole synergy facilitates fast directional charge transport for the solar-driven co-production of hydrogen and value-added chemicals.

Methods

Synthesis of photocatalysts

Synthesis of y-MoS₂/Zn₃In₄S₉. In a 60 mL water-ethanol solution, 3 mmol of ZnCl₂ (Sinopharm Chemical Reagent Co., Ltd., ≥98.0%) and 4 mmol of InCl₃·4H₂O (Beijing InnoChem Science & Technology Co., Ltd., 99.9%) were dissolved with vigorous stirring for 20 min. Subsequently, 9 mmol of thioacetamide (C₂H₅NS, Beijing InnoChem Science & Technology Co., Ltd., 98.0%) was added, and stirring was continued for an additional 20 min. Subsequently, varying amounts of flower-like MoS₂ (4.8, 9.6, 14.4, 19.2, 24, and 28.8 mg) were added to the mixture and stirred for an additional 30 min. The mixture was reacted at 160 °C for 12 h. The resulting product was then successively washed with ethanol and distilled water, followed by drying for subsequent use. The resulting product was designated as y-MoS₂/Zn₃In₄S₉, where y represents the molar ratio of MoS₂ to Zn₃In₄S₉ at 3%, 6%, 9%, 12%, 15%, and 18%, respectively. For detailed preparation procedures of additional catalysts, please refer to the supplementary information.

Photocatalytic activity measurements

The production of solar hydrogen synergizes effectively with organic synthesis. In this experiment, 20 mg of the photocatalyst was uniformly dispersed in a three-necked glass reaction flask containing 50 ml of H₂O and 3 ml of BA (Sinopharm Chemical Reagent Co., Ltd., ≥99.0%) solution. Subsequently, the mixed system was subjected to sonication for 20 min. Prior to sealing the reactor, argon gas was employed to effectively eliminate air and dissolved oxygen from both the reactor and the solution. Following this, the circulating water system was activated to maintain the reactor temperature at 4 ^oC. Subsequently, a xenon lamp (CEL-HXF300-T3) emitting wavelengths (λ > 420 nm) was activated to irradiate the solution for a duration of up to three hours. During this interval, hydrogen (H₂) was collected from the reactor every 30 min, and its concentration was analyzed using a gas chromatography system (GC-7920, Beijing, China). After 3 h, 5 mL of the solution was extracted, filtered by centrifugation, and passed through a 0.22 μm pore size filter. The concentrations of BA and BAD were then determined using liquid chromatography (Agilent 1260 Infinity II, inert Sustain C18). The yields of H₂, BA, and BAD were quantified using the external standard method, with the relevant standard curves presented in Supplementary Fig. 41. For further details regarding characterization methods, parameter settings, and catalyst activity testing, please refer to the supplementary information. The reaction selectivity is calculated using the following formula (1):

$$\mathrm{Selectivity}\left(\%\right)=\left(\frac{{\mathrm{C}}_2}{{\mathrm{C}}_0{-\mathrm{C}}_1}\right)\times 100\%$$ 1

In the photocatalytic system, the concentrations of BA and its derivative, BAD, after the reaction are denoted as C₁ and C₂, respectively. C₀ represents the initial concentration of BA.

Author contributions

Z.W. was responsible for sample preparation, executing all experimental procedures, collecting and analyzing data, and drafting the preliminary manuscript. D.Z. specifically conducted the theoretical calculations using DFT. K.T., G.C., Y.L., Y.J., S.(F).L., and S.-T. L. meticulously reviewed and provided insightful comments on the experimental data, contributing significantly to the feedback process. J. Y. was responsible for designing the experimental plan and conducting the final review and confirmation of the manuscript after incorporating diverse opinions. All authors engaged in comprehensive discussions regarding the manuscript results, thereby ensuring the accuracy and thoroughness of the content.

Peer review

Peer review information

Nature Communications thanks the anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data produced in this study are presented within the article and Supplementary Information, while the raw data are accessible in the accompanying Source Data file. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-66003-4.