The synthesis process of Zn_V-ZCS is shown in Fig. 1a. We use zinc acetate dihydrate, cadmium acetate dihydrate and L-cysteine as raw materials, Zn_V-ZCS samples were synthesized in hydrothermal environment with the strong coordination of Zn and L-cysteine mercapto (-SH). Scanning electron microscope (SEM) images (Fig. S1) show that both pure ZCS and Zn_V-ZCS-10 are granular morphology; Transmission electron microscope (TEM) (Fig. 1b) further observed that, Zn_V-ZCS-10 consists of numerous nanoparticles exhibiting amorphous-shaped morphology and dimensions of several tens of nanometers. Compared with pure ZCS (Fig. S2), the overall morphology has no obvious change, this shows that the introduction of vacancy does not cause large-scale structural damage to particles. High-Resolution Transmission Electron Microscopy (HRTEM) (Fig. 1c) clearly exhibits well-resolved lattice fringes showing an interplanar distance of approximately 0.33 nm, this matches the (002) crystallographic plane of Zn_0.4Cd_0.6S, indicating that the overall crystallinity is largely preserved. High-Angle Annular Dark-Field Scanning Transmission Electron Microscopy (HAADF-STEM) was used to characterize the morphology, while elemental mapping was applied to reveal the spatial distribution of elements (Fig. 1d). They illustrate that S, Zn, and Cd are homogeneously distributed throughout the particles, implying that the detected defects are predominantly atomic-scale point defects. X-ray Diffraction (XRD) (Fig. 1e) demonstrates that ZCS and Zn_V-ZCS-10 possess the same hexagonal phase (PDF#40-0836). The diffraction peaks of Zn_V-ZCS-10 become slightly wider and weaker. This suggests a small decrease in crystallinity. This change is due to the presence of Zn vacancies in Zn_V-ZCS-10. In the EPR spectrum, Zn_V-ZCS-10 exhibits a distinct signal at g = 2.004, whereas pristine ZCS shows a negligible response at this position. This signal is reasonably attributed to unpaired electrons associated with defect centers related to Zn vacancies [34,35,36]. ICP analysis indicates that the concentration of Zn vacancies in Zn_V-ZCS-10 is approximately 4.1226 wt%.

XPS measurements were conducted to elucidate the surface chemical environments of the samples (Fig. 2a–d). As shown in Fig. 2a, the survey spectrum reveals the existence of S, Zn, Cd, C and O. For the high-resolution S 2p spectrum (Fig. 2b), the peaks at 161.27 eV and 162.50 eV can be assigned to S 2p_3/2 and S 2p_1/2, respectively, indicating the typical sulfide S²⁻ state. In Fig. 2c, the Zn 2p peaks centered at 1021.73 eV and 1044.71 eV are attributed to Zn 2p_3/2 and Zn 2p_1/2, confirming the Zn²⁺ state. Meanwhile, the Cd 3d spectrum in Fig. 2d shows two peaks at 404.83 eV and 411.58 eV, corresponding to Cd 3d_5/2 and Cd 3d_3/2, which suggests that Cd mainly exists as Cd²⁺. Overall, the XPS results demonstrate that the chemical states of the elements in the sample remain essentially unchanged. The C 1s peak mainly arises from adventitious carbon and was used for binding-energy calibration, whereas the O 1s signal can be attributed to surface-adsorbed oxygen-containing species, such as hydroxyl groups or adsorbed water [37,38,39]. From the XPS results, it can be seen that the Zn 2p and Cd 3d peaks in Zn_V-ZCS-10 shift toward higher binding energy compared to those in ZCS, while the S 2p peak shifts toward lower binding energy. This indicates that Zn vacancies modulate the surface electronic structure of Zn_V-ZCS-10, leading to electron enrichment on the S atoms. The surface element content table (Table S3) obtained from XPS analysis shows that the Zn content in Zn_V-ZCS-10 is lower than that in ZCS, confirming the introduction of Zn vacancies in Zn_V-ZCS-10.

Quantitative evaluation of photocatalytic H₂O₂ production was carried out according to the calibration curve provided in Fig. S5. Under the identical experimental conditions, in one hour, the production amount of H₂O₂ in pure ZCS sample was 19.25 mmol/g, while the production amount of Zn_V-ZCS-10 reached 44.39 mmol/g, which is about 2.31 times that of pure ZCS (Fig. 3a). Compared with previously reported sulfide-based photocatalysts for H₂O₂ production, our as-prepared Zn_V-ZCS-10 exhibits excellent photocatalytic performance for H₂O₂ generation (Table S2). The enhanced activity suggests that introducing Zn vacancies effectively promotes H₂O₂ photosynthesis over ZCS. Fig. 3b shows the average H₂O₂ yields of the Zn_V-ZCS-X (where X represents the amount of L-cysteine used) samples after 1 h of irradiation, with error bars representing the standard deviation of three independent experiments. The results indicate that the optimal performance, reaching 44.39 mmol/g, is achieved when the amount of L-cysteine is 10 mmol. Fig. 3c shows the H₂O₂ production performance of Zn_V-ZCS-10 in different atmospheres. The H₂O₂ production can hardly be detected under an argon atmosphere, however, obvious catalytic activity can be observed in air and pure oxygen atmosphere, and the production amount in pure oxygen atmosphere is higher, which further demonstrates that O₂ is the dominant feedstock involved in photocatalytic H₂O₂ production. The AQY experiment (The specific methods can be found in the Supplementary Files) results (Fig. 3d) further reveal the wavelength-dependent photocatalytic performance of Zn_V-ZCS-10. The AQY increases from 6.31% at 365 nm to 10.33% at 425 nm, indicating that Zn_V-ZCS-10 exhibits the highest quantum efficiency in the 365–425 nm wavelength range. When the wavelength further increases to 520 and 920 nm, the AQY sharply decreases to 0.62% and 0.006%, suggesting that long-wavelength light contributes negligibly to the photocatalytic reaction. After five cycles, Zn_V-ZCS-10 still retains more than 78% of its initial activity, indicating acceptable photocatalytic stability (Fig. 3e). Additionally, we measured the XRD patterns and SEM images (Fig. S1) of the Zn_V-ZCS-10 samples before and after cycling. No significant changes can be observed from the results, indicating that the as-prepared catalyst possesses good structural stability.

To study how Zn vacancies affect the optical properties of ZCS, solid-state UV-Vis diffuse reflectance spectroscopy (UV-Vis DRS) was used to test ZCS and Zn_V-ZCS-10. The UV–vis DRS spectra (Fig. 4a) indicate that both samples possess pronounced light-harvesting capability in the visible region, while Zn_V-ZCS-10 shows overall stronger absorption in the 400–800 nm range, suggesting that the engineering of Zn vacancies enhances the ZCS light-harvesting capability. According to the Kubelka-Munk transformation and the Tauc plot fitting (inset of Fig. 4a), the ZCS optical band gap is about 2.17 eV. The band gap value for Zn_V-ZCS-10 is about 2.05 eV. The conduction band potential of Zn_V-ZCS-10 is about −0.60 V (vs. NHE), which is slightly more positive than that of ZCS (−0.73 V). Nevertheless, it remains more negative than the redox potential of O₂/·O²⁻ (−0.33 V vs. NHE), indicating that the photogenerated electrons still have sufficient thermodynamic driving force to reduce O₂ to ·O²⁻. Therefore, the enhanced photocatalytic H₂O₂ production of Zn_V-ZCS-10 should not be simply attributed to a more negative CB position, but rather to the combined effects of improved charge separation, increased carrier concentration, reduced interfacial charge-transfer resistance, and facilitated O₂ activation induced by Zn vacancies. The Mott-Schottky plots (Fig. 4c) show positive slopes. This shows both samples are n-type semiconductors. Under the same conditions, Zn_V-ZCS-10 has a much smaller slope than ZCS. From the Mott-Schottky equation, the slope of the 1/C²-V curve is inversely related to the donor density (N_D) [40]. A smaller slope means a higher carrier concentration in Zn_V-ZCS-10. The electrical properties were tested with an electrochemical workstation. The transient photocurrent curves (Fig. 4d) show Zn_V-ZCS-10 gives a higher and more stable photocurrent under on-off 425 nm light. Electrochemical impedance spectroscopy (EIS) results (Fig. 4e) show that Zn_V-ZCS-10 presents a markedly reduced Nyquist semicircle compared with ZCS. This indicates that the interfacial charge transfer resistance is lower. Linear sweep voltammetry (LSV, Fig. S3c) under light shows Zn_V-ZCS-10 has higher photocurrent densities in the tested potential range. In the dark, both samples show very low current. This indicates that the process is mainly driven by light. The photocurrent trend is similar to the photocatalytic activity. Photoluminescence (PL) spectra (Fig. 4f) show Zn_V-ZCS-10 has much lower emission intensity than ZCS. The presence of Zn vacancies is beneficial for inhibiting charge-carrier recombination while enhancing charge separation. The optical and electrochemical results show that Zn_V-ZCS-10 has a smaller band gap, a higher carrier concentration, and lower interfacial charge transfer resistance. This contributes to enhancing the photocatalytic H₂O₂ production.

To elucidate the photocatalytic H₂O₂ generation mechanism over Zn_V-ZCS-10, radical trapping and EPR experiments were performed. As shown in Fig. 5a, the H₂O₂ yield decreased sharply from 44.39 to 3.05 mmol/g after adding p-benzoquinone (p-BQ), demonstrating that ·O₂⁻ is the dominant intermediate. By contrast, tert-butanol and β-carotene only caused slight inhibition, suggesting that ·OH and ¹O₂ are not the major reactive species. In addition, the decreased H₂O₂ yield in the presence of KPS, an electron scavenger, confirms that photogenerated electrons are indispensable for O₂ reduction and subsequent ·O₂⁻ formation [41]. The time-dependent EPR spectra further show gradually enhanced DMPO-·O₂⁻ signals under irradiation, directly verifying the continuous generation of ·O₂⁻. Based on these results, the H₂O₂ generation over Zn_V-ZCS-10 primarily proceeds through a ·O₂⁻-mediated two-step single-electron oxygen reduction pathway, as illustrated in Fig. 5c. After introducing Zn vacancies into ZCS, although the conduction band position is lowered, it still satisfies the thermodynamic requirement for O₂ to obtain electrons and be reduced to ·O₂⁻. Upon light irradiation, electrons in the valence band of Zn_V-ZCS-10 are excited to the conduction band. Photocurrent and PL results indicate that the presence of Zn vacancies endows Zn_V-ZCS-10 with higher charge carrier separation efficiency than pristine ZCS, thereby generating more photogenerated electrons. Furthermore, XPS analysis reveals that the Zn vacancies on the surface of Zn_V-ZCS-10 can modulate the electronic structure of the active S sites, thereby accelerating the rate of H₂O₂ production via the two-step single-electron O₂ reduction pathway. Meanwhile, the photogenerated holes left in the valence band are effectively consumed by the sacrificial agent.