Hydrogen (H₂) presents a promising clean and renewable energy carrier as it produces only water upon combustion, thereby contributing to CO₂ emission reduction. However, hydrogen is predominantly produced via steam-methane reforming, which is an environmentally unsustainable method^[1,2]. In contrast, “green” hydrogen can be generated through eco-friendly techniques, such as photocatalytic and photoelectrochemical reactions^[3,4].

Solar-driven photocatalytic water splitting using semiconducting materials such as titanium dioxide (TiO₂) and graphitic carbon nitride (g-C₃N₄) is well-established, while metal-organic frameworks (MOFs) have recently emerged as promising photocatalysts due to their high surface area and tunable structure^[5-7]. Over recent decades, various materials have been developed to enhance the efficiency of the hydrogen evolution reaction (HER). However, commercialisation remains challenging due to issues such as low efficiency, shorter lifetimes, and the high cost of stable plasmonic metal co-catalyst nanoparticles, including platinum (Pt) and gold (Au)^[8]. Photocatalysts such as g-C₃N₄^[9], ZnCdS^[10,11], and TiO₂^[12,13] have demonstrated potential for efficient photocatalytic H₂ generation. However, they suffer from various limitations, including a low surface area, rapid charge recombination, and reduced efficiency. Strategies to address these challenges include surface sensitisation^[14], bandgap engineering^[15], metal/non-metal doping^[16], heterojunction development^[17], spectral conversion^[18], co-catalyst coupling^[19], and modification with carbon-based materials^[20], leading to the development of novel semiconductor-based materials^[21].

Among various photocatalysts such as WO₃, g-C₃N₄, and TiO₂, the unique features of copper tungstate (CuWO₄) are favourable to the application in photocatalytic hydrogen production^[22]. CuWO₄ is a semiconductor photocatalyst that exhibits favourable bandgap properties due to the hybridised O-2p and Cu-3d orbitals^[23], which contribute to its valence band position on the normal hydrogen electrode (NHE) scale^[24]. Although TiO₂ is stable under photocatalytic reaction conditions, its activity is mainly limited to the UV region due to its wide band gap (~3.2 eV)^[25]. Meanwhile, g-C₃N₄, a visible light photocatalyst, has been extensively explored^[26]; however, it possesses limited photocatalytic performance due to rapid charge recombination. Similarly, WO₃ also possesses visible light absorption (with a band gap of ~2.4 eV)^[27], but the conduction band potential is insufficient for the HER. CuWO₄, on the other hand, has a narrow band gap (~2.2 eV) with improved solar light utilisation, a favourable band-edge position for reduction^[28], and has the potential for constructing a heterojunction with enhanced charge separation. However, pristine CuWO₄ suffers from high carrier recombination rates, limiting its HER performance. To enhance its efficiency, CuWO₄ has been combined with other materials to suppress electron-hole recombination^[29]. For example, Dinh et al. found that coating CuWO₄ nanoparticles with the conductive polypyrrole polymer enhanced the H₂ evolution rate to 365 µmol g^-1 h^-1[30]. Additionally, vanadium doping was found to reduce electron-hole recombination and improve the reduction potential, as reported by Le Minh Tri et al.^[31] Basyach et al.^[32] demonstrated that doping CuWO₄ with nickel and subsequently combining it with g-C₃N₄ (0.2Ni-CuWO₄/g-C₃N₄) significantly enhanced H₂ production through improved light absorption^[32], a reduced bandgap, and an increased surface area.

Various MOFs, including ZIF-67^[33], MIL-101^[34], and MIL-125^[35], have demonstrated potential as auxiliary component materials or as photocatalysts by themselves in solar-driven water-splitting, benefiting from their highly porous metal-coordinated architectures with organic linkers. The photocatalytic H₂ production efficiency of MOFs can be systematically tuned by modifying properties such as optical absorption, bandgap energy, and charge transfer dynamics^[36,37]. Functionalisation of UiO-66 and MIL-101 with amino groups (-NH₂), achieved by substituting terephthalic acid with amino terephthalic acid, significantly improves water splitting efficiency compared to their pristine forms^[38]. An effective approach to further enhance the H₂ production performance is to incorporate guest species into MOFs, to form core-shell structures or composite materials^[39]. Heterojunctions between metal oxide semiconductors and MOFs, such as TiO₂/MIL-101^[40] and ZnO/MIL-53^[41], exhibit superior photocatalytic activity due to improved charge separation and light absorption.

Similarly, the Zeolitic Imidazolate Framework-8 (ZIF-8) has been combined with other materials to improve the efficiency of photocatalytic water splitting, benefiting from its high surface area and the three-dimensional porous framework comprising zinc ions and organic imidazolates^[42]. In addition to structural benefits, ZIF-8 can influence photocatalytic efficiency by various physicochemical mechanisms^[43]. Abundant adsorption sites are provided by the porous structure and large surface area of ZIF-8, which facilitate the diffusion of reactants and products during the photocatalytic reaction^[44]. Moreover, coupling ZIF-8 with a semiconductor photocatalyst can increase the heterojunction interface contact area and promote interfacial charge transfer, thus enhancing charge separation^[45]. Therefore, ZIF-8 represents a promising platform for designing a composite with increased surface reactivity and photocatalytic efficiency. For instance, Qureshi et al. reported that the Ru nanoparticles achieved an exceptional H₂ evolution rate of 5,245 μmol g^-1 h^-1 in a Z-scheme heterojunction of g-C₃N₄ and ZIF-8^[46]. Cd- and Mn-doped ZIF-8 photocatalysts have demonstrated enhanced charge transfer and H₂ production^[47]; however, their poor stability restricts their practical application. Furthermore, reliance on costly noble metals (e.g., ruthenium) and hazardous elements (e.g., cadmium) undermines the sustainability and economic viability of green hydrogen production.

To address these challenges, researchers have modified ZIF-8 with wide bandgap semiconductors, such as Co-ZIF-8/TiO₂^[48], and ZIF-8/ZnO^[49], which exhibit improved photocatalytic performance compared to pristine ZIF-8, doped-ZIF-8, TiO₂, and ZnO. Despite some advancements, such heterojunctions continue to face issues of low photoconversion efficiency. Therefore, identifying new, more active semiconductors, such as CuWO₄, is crucial for enhancing the efficiency of photocatalysts in practical applications^[50].

However, the incorporation of ZIF-8 into CuWO₄ has been largely unexplored, despite significant progress in MOF-based composites for photocatalytic hydrogen production ^[51]. Although the photocatalytic activity of ZIF-8-based heterostructures has shown significant improvement, the relationship among the porous structure, interfacial charge transfer, and photocatalytic activity remains mostly unknown. Furthermore, the effect of ZIF-8 on the accessibility of active sites and the charge-carrier dynamics in CuWO₄-based heterostructures has not been investigated systematically. To address this gap, a ZIF-8@CuWO₄ heterostructure was prepared via an in situ growth strategy and evaluated for photocatalytic hydrogen evolution under UV and visible light irradiation. We investigated the synergistic interaction between ZIF-8 and CuWO₄ using detailed structural, textural, optical, and electrochemical characterisation. Our work provides new insight into the role of the MOF-semiconductor interface in enhancing photocatalytic hydrogen production and provides a promising strategy for designing efficient photocatalytic systems.

Zinc nitrate hexahydrate (Zn(NO₃)₂·6H₂O, ≥ 99%, Sigma Aldrich), copper nitrate hemi pentahydrate (Cu(NO₃)₂·2.5H₂O, ≥ 98%, Thermo Scientific), sodium tungstate dihydrate (Na₂WO₄·2H₂O, ≥ 99%, Sigma Aldrich), polyvinylpyrrolidone (PVP) ((C₆H₉NO)_x, M_W ~ 10,000, Thermo Scientific), 2-methyl imidazole (2-MIM, C₄H₆N₂, 99%, Sigma Aldrich), methanol (CH₃OH, 99.8%, Central chem), ethanol (C₂H₅OH, 96%, Central chem) were used as delivered by the suppliers, without further purification.

Figure 2 illustrates the XRD analysis, which was performed to elucidate the crystalline phases present in CuWO₄, ZIF-8, and their heterostructure (ZIF-8@CuWO₄). The XRD pattern of pure ZIF-8 exhibits multiple diffraction maxima within the 5-70° 2θ range, with prominent reflections at 2θ ≈ 7.49°, 10.48°, and 12.81°, corresponding to the [110], [200], and [211] crystallographic planes, respectively. By contrast, the diffraction pattern of pure CuWO₄ exhibits distinct maxima at 2θ ≈ 18.98°, 26.06°, 28.74°, 31.71°, and 36.94°, assigned to the [100], [$$\overline{1}$$01], [111], [$$\overline{1}$$$$\overline{1}$$1], and [020] planes of the CuWO₄ crystal structure. These diffraction signatures are consistent with the reference patterns from the ICDD PDF 01-070-1732 and COD PDF 7111973 databases for CuWO₄ and ZIF-8^[55,56], respectively. Pawley fittings were performed over the experimental data using the corresponding CIF files, further confirming the phase purity of the prepared CuWO₄ and ZIF-8 (see Supplementary Figure 1 for PXRD plots with Pawley refinements) and the coexistence of these phases within the ZIF-8@CuWO₄ heterostructure.

The size of CuWO₄ crystals in ZIF-8@CuWO₄ was estimated using the Scherrer equation [Supplementary Table 1], which indicates comparable sizes for CuWO₄, in both its pristine and composite forms, likely attributable to the post-synthetic ultrasonication and the in-situ growth process of ZIF-8.

The morphology and microstructures of ZIF-8, CuWO₄, and the ZIF-8@CuWO₄ composite are shown in Figure 3. ZIF-8 particles exhibit a nanoscale, pseudohexagonal morphology [Figure 3A], which is consistent with that of previously reported structures^[57]. The particle size of ZIF-8 could be considerably reduced by replacing the water solvent with methanol (see Supplementary Figure 2). Figure 3C confirms the successful synthesis of spherical CuWO₄ nanoparticles with an average diameter of ~64 ± 15 nm, with some degree of agglomeration. The incorporation of PVP and the application of ultrasonication during synthesis proved effective in achieving homogeneous particle size. The ZIF-8@CuWO₄ composite has a similar morphology and particle size distribution to pure CuWO₄ [Figure 3E and F]. In the heterostructure, the naturally pseudohexagonal ZIF-8 particulates are more rounded, which may be attributed to a greater density of nucleation sites, promoting preferential ZIF-8 growth with minimised surface energy in the presence of spherical CuWO₄ nanoparticles in the colloidal solution^[58]. Consequently, the coexistence of CuWO₄ and ZIF-8 phases at the nanoscale in the composite is expected to enhance photocatalytic efficiency due to the abundance of surface-active sites.

The XPS spectra for CuWO₄ and the ZIF-8@CuWO₄ composite are presented in Figure 4. In the XPS spectrum of pure CuWO₄ nanoparticles (see Figure 4A), characteristic peaks corresponding to Cu 2p, W 4f, and O 1s were identified. Consistent with energy-dispersive X-ray spectroscopy (EDS) results (see Supplementary Figure 3), the XPS spectrum of the ZIF-8@CuWO₄ composite (see Figure 4B) reveals the presence of Cu, W, O, Zn, N, and C.

XPS peak deconvolution was performed to elucidate the oxidation states and chemical environments of the elements present in pure CuWO₄ and the ZIF-8@CuWO₄ composite. Figure 5A shows that the C 1s spectrum of ZIF-8@CuWO₄ is deconvoluted into four components: a dominant C-C/C=C peak (pink) at 284.53 eV (42.13%)^[59], two more peaks at 285.36 (blue) and 286.29 (greenish) eV corresponding to C-N (34.23%) and N-C=N (13.17%) bonds of the imidazole ring^[60], and a peak at 291.78 eV (green) attributed to chemisorbed C=O species. The N 1s spectrum [Figure 5B] exhibits peaks at 398.48 (pink), 399.15 (blue), and 400.24 (greenish) eV, which respectively correspond to pyridinic, pyrrolic, and graphitic nitrogen in the composite^[61]. These are associated with enhanced photo-catalytically active sites in ZIF-8@CuWO₄^[61-63].

The Zn 2p spectrum [Figure 5C] displays spin-orbital splitting with Zn 2p_1/2 and Zn 2p_3/2 peaks at 1,044.8 and 1,021.75 eV, respectively. This fact is indicative of Zn²⁺ in the ZIF-8 lattice^[64]. The O 1s spectrum of the ZIF-8@CuWO₄ composite [Figure 5D] shows a peak at 530.2 (green) eV, which is assigned to lattice oxygen, as well as a peak at 531.42 (blue) eV, corresponding to adsorbed (or chemisorbed) oxygen^[64]. Additionally, Cu 2p spectra [Supplementary Figure 4A] reveal satellite features that are indicative of Cu²⁺ in the CuWO₄ lattice^[65]. In the ZIF-8@CuWO₄ composite, the binding energies of Cu 2p [Supplementary Figure 4B] and W 4f (Figure 4B, bottom) shift to lower values compared to pure CuWO₄, suggesting an increased surface electron density^[66]. This shift aligns with observations by Liao et al., who reported similar binding energy reductions for Cu 2p in nickel oxide-modified CuWO₄, indicating enhanced electron density and effective carrier transfer between NiO and CuWO₄^[66]. Comparable trends were observed by Vasilopoulou for WO₃ films synthesised under a hydrogen atmosphere, where reduced carrier recombination was noted relative to air-synthesised WO₃^[67]. Previous studies have also shown that WO₃·H₂O incorporation with ZIF-8 reduces charge carrier recombination and promotes photocatalytic activity^[68]. More importantly, referring to a similar reported ZIF-8@Cu-W₁₈O₄₉ system, where ZIF-8 was grown in the presence of Cu doped W₁₈O₄₉ nanowires, a similar negative shift in binding energy for W 4f XPS spectra indicated the robust electronic interaction between the heterointerface of ZIF-8 and Cu-W₁₈O₄₉^[69]. The observed binding energy shifts in ZIF-8@CuWO₄ imply the presence of facilitated electron transport pathways, suggesting the formation of a heterojunction within the composite.

The textural properties of the synthesised ZIF-8, CuWO₄, and ZIF-8@CuWO₄ composite were examined using the BET isotherms and BJH pore size distribution, which are shown in Supplementary Figures 5 and 6, respectively. The specific surface area and porous structure influence the kinetics of reactions by controlling the accessible surface-active sites, reactant adsorption, and mass transport in photocatalysis^[70,71]. As shown in Table 1, the compact structure of pure CuWO₄ with limited porosity was indicated by its low specific surface area of 3.55 m² g^-1. Compared with CuWO₄, ZIF-8 displayed a remarkably high BET surface area of 1,150.20 m² g^-1 due to its highly porous MOF^[72]. In ZIF-8@CuWO₄, the obtained surface area was 858.86 m² g^-1 after integrating CuWO₄ with ZIF-8, which is lower than that of ZIF-8 due to the partial occupancy or blocking of a fraction of micropores by CuWO₄ nanoparticles. However, this surface area remains considerably higher than that of CuWO₄. These results demonstrate that the ZIF-8 framework in ZIF-8@CuWO₄ is largely preserved while maintaining a high density of accessible active sites.

The BJH pore size distribution for ZIF-8@CuWO₄ is shown and compared with that of pristine CuWO₄ and ZIF-8 in Supplementary Figure 6. An almost featureless pore-size distribution was obtained for CuWO₄ with a weak pore volume centred around 180 nm, suggesting the presence of a limited number of interparticle voids. However, ZIF-8 showed a pore size distribution in the 1-4 nm pore size range along with a strong pore volume near 180 nm relative to CuWO₄, supporting its porous network and higher number of interparticle voids^[73]. In ZIF-8@CuWO₄, the pore size distribution from 1-4 nm with a much more pronounced peak around 180 nm was observed, which suggests the retention of intrinsic porosity of ZIF-8 and formation of additional voids and more accessible diffusion pathways in the heterostructure of ZIF-8@CuWO₄.

In line with these observations, the total pore volume for CuWO₄ and ZIF-8 increased from 0.026 and 0.238 cm³ g^-1 to 0.645 cm³ g^-1 for ZIF-8@CuWO₄, respectively. This increased pore volume can be attributed to the retained porosity and the formation of interparticle voids due to the incorporation of ZIF-8 with CuWO₄. The high surface area with this hierarchical porosity is expected to enhance reactant adsorption, mass transport, and the accessible active sites, thereby contributing to the enhanced photocatalytic hydrogen evolution of the ZIF-8@CuWO₄ heterostructure^[74]. In addition to these textural advantages, the improved photocatalytic activity is also a result of interfacial charge separation and electronic interactions, as analysed by the PL/TRPL, XPS, and Mott-Schottky measurements.

The UV-vis diffuse reflectance spectra (DRS) of pure ZIF-8, CuWO₄, and the ZIF-8@CuWO₄ composite were recorded and transformed using the Kubelka-Munk function to approximate the absorbance behaviour of these materials and are shown in Figure 6A. Pure CuWO₄ exhibited a broad absorbance band in both the UV and visible regions, with an absorption edge at 400-500 nm. In contrast, ZIF-8 displayed a sharp absorbance band peak at 220 nm, indicative of a wide band gap corresponding to the π-π* transitions in imidazole rings of the ZIF-8^[75]. For the ZIF-8@CuWO₄ composite, the absorption edge exhibited a redshift towards longer wavelengths, suggesting modified electronic interactions between the two components. Band gap energies were then estimated from Tauc plots assuming indirect allowed electronic transitions (n = 2), as commonly reported for CuWO₄-based systems [Figure 6B], yielding values of 2.22 eV for pure CuWO₄ (2.1-2.5 eV)^[76,77], and 5.13 eV for ZIF-8 (4.8-5.2 eV)^[77,78], consistent with previously reported data. The ZIF-8@CuWO₄ composite exhibited two apparent optical transitions at 2.13 and 4.99 eV, which are slightly shifted compared with individual components. Similar optical transitions have been observed by Anu et al. in CuO@ZnO nanocomposites with enhanced carrier density^[79]. Furthermore, Makula et al.^[80] demonstrated that a multiphase composite may possess multiple optical transitions, with each component contributing to the overall optical absorption while largely preserving its intrinsic electronic structure. Consequently, these transitions were associated with CuWO₄ (2.13 eV) and ZIF-8 (4.99 eV), whereas their slight shifts are attributed to interfacial electronic interactions within ZIF-8@CuWO₄ heterostructure, which was further supported by shifts in the PL quenching [Figure 6C], Mott-Schottky analysis [Figure 7A-C], estimated band positions [Figure 7D] and XPS valence band maxima [Figure 7E], potentially improving interfacial charge transfer and photocatalytic performance.

The recombination behaviour of photogenerated charge carriers in the synthesised photocatalysts was evaluated using photoluminescence (PL) spectroscopy, whereby changes in the PL intensity provide qualitative insight into electron-hole recombination^[81]. The steady-state PL emission spectra (400-700 nm) of pure ZIF-8, pure CuWO₄, and the ZIF-8@CuWO₄ composite, excited at 364 nm, are presented in Figure 6C. A pronounced PL quenching effect was observed in CuWO₄ and the ZIF-8@CuWO₄, in unambiguous contrast to the intense emission peaks of pure ZIF-8 at 435 and 454 nm. Among these, CuWO₄ exhibited the lowest PL intensity, establishing the relative recombination trend as ZIF-8 >> ZIF-8@CuWO₄ > CuWO₄. The suppressed PL intensity in the ZIF-8@CuWO₄ composite compared to pristine ZIF-8 signifies a substantial reduction in the electron-hole recombination process, which is attributed to enhanced charge transfer at the heterojunction interface. This is consistent with prior studies on hybrid WO₃/ZIF-8 composites^[82]. Despite lower radiative recombination of CuWO₄, the ZIF-8@CuWO₄ composite displays the highest photocatalytic hydrogen evolution (discussed in the latter section), suggesting that PL intensity is not an exclusive determining factor of photocatalytic performance. The activity of the ZIF-8@CuWO₄ composite is attributed to the enhanced interfacial charge transfer and favourable band alignment (discussed in Mott-Schottky analysis and Valence band XPS results), resulting from heterojunction formation.

Further investigation of charge-carrier dynamics was conducted by performing time-resolved luminescence (lifetime) measurements on pure CuWO₄ nanoparticles and the ZIF-8@CuWO₄ composite. The samples were excited at 360 nm, and emission was detected at 440 nm [Figure 6D]. The decay profiles were fitted using two exponential functions, which indicates the presence of distinct migration pathways and enables the determination of the average carrier lifetime (τ_avg). The extracted lifetimes for CuWO₄ and the ZIF-8@CuWO₄ were approximately 1.1 and 1.2 ns, respectively. These values are consistent with the previously reported lifetimes (0.4-2.5 ns) for the CuWO₄-based system^[83,84]. Although this difference is small (0.1 ns), it may indicate a slight improvement in the charge-carrier dynamics, on one hand, improved charge-carrier separation, and on the other hand, enhanced electron-hole transport, resulting in a migration of photogenerated charge carriers to the surface. These findings emphasise the importance of interfacial interactions within the composite and demonstrate the presence of a greater number of photogenerated carriers, further optimising the photocatalytic efficiency of CuWO₄ through ZIF-8 encapsulation^[85].

Mott-Schottky measurements were performed on ZIF-8, CuWO₄, and ZIF-8@CuWO₄ to confirm their redox potential by determining their conduction and valence band positions. This confirmed the characteristics of electron/hole transfer in the ZIF-8@CuWO₄ heterojunction. The flat-band potential of a semiconductor can be obtained through Mott-Schottky analysis. The positive slopes for ZIF-8@CuWO₄ [Figure 7A], ZIF-8 [Figure 7B] and CuWO₄ [Figure 7C] at three different frequencies, 1,000 (blue), 1,500 (orange) and 2,000 (green) Hz, respectively indicated that both are n-type semiconductors, and their flat-band potentials are estimated to be in proximity to their conduction bands, respectively^[86,87].

The flat-band potentials vs. RHE, for ZIF-8 and CuWO₄, were -0.844 and -0.542 V, respectively. In the previous report on CuWO₄/Cu_1-xZn_xWO₄/ZnWO₄ hybrid sandwiched heterojunction, Mott-Schottky analysis was consistent with their n-type behaviour, having flat-band potentials (varied from -0.2 to -0.3 V) for various Zn-doped CuWO₄^[88]. Similarly, Huang et al. obtained consistent n-type Mott-Schottky curves with a flat-band potential of -0.862 V for ZIF-8 in ZIF-8@ZIF-67 core-shell structures^[89]. Using the flat band potential data and the band gap estimated through the Tauc plot (see Supplementary Material Equations 1-3, where flat band potential values are converted to Energy vs. vacuum and approximated as conduction band potential), the conduction (CB) and valence bands (VB) positions for ZIF-8 and CuWO₄ were also calculated, and the results are shown in Table 2.

A comparison of both ZIF-8 and CuWO₄ revealed that ZIF-8 had a more negative reduction potential for H⁺/H₂. Contrarily, CuWO₄ requires less energy to excite the electrons for the photocatalytic reaction. In the ZIF-8@CuWO₄ heterostructure, a slight negative shift in the flat band potential to -0.878 V was observed [Figure 7A], which indicates the interfacial electronic interaction and equilibration of the Fermi level between ZIF-8 and CuWO₄^[90]. Together, the ZIF-8@CuWO₄ heterojunction offers better redox ability and can result in higher photocatalytic hydrogen production as compared with the pristine materials.

The interfacial interaction between CuWO₄ and ZIF-8 in the ZIF-8@CuWO₄ composite was further investigated and supported by analysing their valence band XPS spectra. Figure 7E represents the comparison of valence band XPS spectra, where the valence band maximum (VBM) was estimated by linear extrapolation of the leading edge of the valence band spectrum to the background level. The intercept was assigned as the VBM binding energy relative to the Fermi level (E_f = 0 eV). As shown in Figure 7E, the estimated VBM values for ZIF-8 and CuWO₄ were 2.31 and 1.38 eV, respectively. However, the VBM value for ZIF-8@CuWO₄ was 1.97 eV, which lies between those of pristine materials. Compared with pure ZIF-8, the VBM of the composite was shifted ~0.34 eV closer to the Fermi level, which suggests the redistribution of electrons within the ZIF-8@CuWO₄ heterojunction and the equilibration of the Fermi level between the contacting components^[91]. This change provides direct evidence of interfacial electronic coupling within ZIF-8@CuWO₄, which is favourable for the separation and migration of charge carriers. These VBM results are consistent with the Mott-Schottky measurements, where a similar shift for ZIF-8@CuWO₄ in flat band potential (-0.844 to -0.878 V) was observed after heterojunction formation, confirming that the interaction between ZIF-8 and CuWO₄ in the composite modified the electronic structure and promoted the interfacial charge transfer.

The photocatalytic performance of the synthesised materials was evaluated through HER under UV and visible light irradiation, using aqueous Na₂S and Na₂SO₃ as sacrificial agents [Supplementary Figure 7A]. Figure 8 shows the photoinduced real-time hydrogen evolution for pure ZIF-8, CuWO₄, and the ZIF-8@CuWO₄ composite monitored continuously by an online mass spectrometer under sequential radiation (visible and UV radiation, respectively). In the control experiment, negligible H₂ or O₂ production was detected in the absence of light irradiation or a photocatalyst. Under irradiation, both ZIF-8 and CuWO₄ exhibited measurable H₂ evolution, with respective rates of 1.5 and 2.9 μmol·g^-1·h^-1 for ZIF-8 and 2.5 and 26 μmol·g^-1·h^-1 for CuWO₄ under visible [Figure 8A] and UV light [Figure 8B], respectively. In the case of decorating CuWO₄ with ZIF-8 (ZIF-8@CuWO₄), the H₂ evolution rates of 44 and 1,112 μmol·g^-1·h^-1 were observed under visible and UV excitation, respectively, using 0.1 M Na₂S and 0.25 M Na₂SO₃ as sacrificial agents. This marked improvement in the photocatalytic activity for ZIF-8@CuWO₄ relative to individual components suggests enhanced interfacial charge transfer and suppressed electron-hole recombination within the heterojunction^[91]. Under visible light irradiation, the ZIF-8@CuWO₄ heterostructure achieved a slightly enhanced photocatalytic hydrogen evolution rate compared with CuWO₄, which is still very low in comparison with the hydrogen produced under UV-light irradiation by the composite. This limited performance can be attributed to the limited utilisation of the visible region by the composite and the low efficiency of photogenerated electron/hole pairs. However, the heterostructure formation promotes the interfacial charge separation within the ZIF-8@CuWO₄ composite, while cocatalyst loading could be used to further improve its visible-light photocatalytic activity.

The H₂ evolution of ZIF-8@CuWO₄ is lower than that of some of the state-of-the-art photocatalysts summarised in Table 3, yet remains within the performance range reported for several previously studied heterostructure photocatalyst systems. For instance, the heterostructure of TiO₂ with NH₂-MIL-125 MOF (TiO₂/NH₂-MIL-125) exhibited a comparable H₂ evolution rate of 44 μmol·g^-1·h^-1 under visible irradiation^[92]. Additionally, the evolution rates of H₂ obtained by noble-metal Pt islands in CuBi₂O₄^[93], CDs@BiVO₄ quantum dots^[94], and ZnO@rGO derived from ZIF-8@rGO over carbon sponge^[95] fall within a similar or lower activity range under similar experimental conditions. These comparisons indicate that the ZIF-8@CuWO₄ heterojunction demonstrates competitive photocatalytic performance among composite photocatalysts.

This enhancement in the photocatalytic activity of ZIF-8@CuWO₄ heterostructure, compared with pristine ZIF-8 and CuWO₄, is attributed to the increased porosity, more accessible active sites, and optimised optical and interfacial electronic properties as suggested by BET/BJH textural results, PL/TRPL studies, Mott-Schottky analysis, and valence band XPS estimations. Notably, the interaction between the pyrrolic and pyridinic nitrogen functional groups in ZIF-8 and the CuWO₄ phase enhances charge separation and electron-hole transfer rates^[78]. The high surface area and porosity of ZIF-8 provide a greater number of accessible catalytically active sites compared to pristine CuWO₄^[105]. In pure ZIF-8, the presence of Zn²⁺ ions and the 2-methylimidazole ligands in the framework results in intense PL emissions [Figure 6C] due to π-π* transitions^[106]. However, within the ZIF-8@CuWO₄ composite, the reduced band gap favours n-π* transitions, thereby promoting charge transfer^[106,107]. This synergistic effect, coupled with the electronic interactions at the interface, results in an efficient photocatalytic system as compared with as-synthesised pristine materials, where CuWO₄ enhances electron density while ZIF-8 provides extensive active surface sites^[42,82,85].

The impact of the sacrificial agent concentration on the photocatalytic reaction was also examined [Supplementary Figure 7B and Figure 8C]. During this reaction, the molarity of Na₂S increased from 0.1 to 0.35 M while the molarity of Na₂SO₃ remained constant at 0.25 M. This resulted in a significant improvement in the hydrogen evolution rate under UV irradiation, increasing from 1,112 to 2,531 μmol·g^-1·h^-1. However, under visible light, the efficiency was negligible (44 to 79 μmol·g^-1·h^-1) due to limited electron/hole pair production, which becomes the limiting factor as the concentration of Na₂S increases. Conversely, UV irradiation increases electron/hole pair generation, enabling Na₂S to effectively scavenge holes, and accelerates the hydrogen evolution performance^[108].

The proposed electron-hole transfer mechanism is illustrated in Scheme 2. Upon irradiation of light on ZIF-8@CuWO₄, both ZIF-8 and CuWO₄ generated electrons and holes in their CB and VB, respectively. Subsequently, the photogenerated electrons migrate from the CB of ZIF-8 to the CB of CuWO₄, facilitating the reduction of protons to generate H₂ at the CB of CuWO₄. Simultaneously, the photoinduced holes transfer from the VB of CuWO₄ to the VB of ZIF-8, thereby facilitating charge separation. The presence of Na₂S/Na₂SO₃ as sacrificial agents further enhances photocatalytic activity by scavenging the holes on the ZIF-8 surface and generating HS^- and S₂O₃^2- species. These species suppress charge carrier recombination and improve efficiency^[109]. This effective electron/hole transfer resulted in the formation of a Type-II heterojunction between ZIF-8 and CuWO₄ in the ZIF-8@CuWO₄ composite, as suggested by Mott-Schottky, Valence band XPS, and PL/TRPL results.

An important feature of the photocatalyst and its performance is the recyclability and reusability. Recyclability of ZIF-8@CuWO₄ composite as a photocatalyst is shown in Supplementary Figure 8. The photocatalytic H₂ evolution rate decreases from 2,531 to 1,336 μmol·g^-1·h^-1 after reusing the photocatalyst for a second time in the third cycle [Figure 9], during which the photocatalyst water-splitting experiment was run over ~15 h in the presence of Vis and UV light. The decrease in photocatalytic activities in each cycle may be due to the intrinsic instability of ZIF-8, which undergoes gradual hydrolysis under water and UV irradiation, due to intensified ligand optical absorption and defect formation^[110,111]. The morphological changes of the as-synthesised ZIF-8@CuWO₄ and recycled ZIF-8@CuWO₄ (after 3rd cycle) were compared in the respective SEM images (acquired at the same magnification) and are shown in Supplementary Figure 9A and B. The recycled composite displayed a transformation from dispersed nanoparticles to hollow microspherical aggregates, which further suggested the structural evolution of the ZIF-8@CuWO₄ heterostructure. These morphological changes could be associated with the accumulation of organic residue originating from the partial degradation of the ZIF-8 or oxidation products of the sacrificial agent on the catalyst surface. For further understanding, the XRD analysis of recycled ZIF-8@CuWO₄ composite was compared with the XRD patterns of standard reference and as-synthesised samples, as shown in Supplementary Figure 9C. The characteristic diffraction maxima for ZIF-8 at 2θ ≈ 7.5° and 12.8° became very weak, further suggesting the partial degradation of the ZIF-8 phase in ZIF-8@CuWO₄ during photocatalytic reaction. While the CuWO₄ phase was still dominantly observed, with several diffraction peaks shifted to higher 2θ values, suggesting structural modification of recycled ZIF-8@CuWO₄. Based on the proposed type-II heterojunction charge transfer mechanism, the partial degradation of ZIF-8 may be correlated with the hole-induced oxidation of the MOF framework during prolonged photoreaction. Although most holes are scavenged by the Na₂S/Na₂SO₃ sacrificial agents, a small fraction of holes may oxidise the ZIF-8 framework under prolonged irradiation, consistent with the observed structural changes. These findings demonstrate the necessity of carefully designing the MOF-based photocatalysts for practical use under visible irradiation. The ZIF-8@CuWO₄ composite remains stable under visible light and still achieves H₂ evolution rates of 44/79 μmol·g^-1·h^-1 (visible) and 1,112/2,531 μmol·g^-1·h^-1 (UV) from water without the need for a noble-metal co-catalyst.

A novel ZIF-8@CuWO₄ heterojunction composite was successfully synthesised by growing ZIF-8 around spherical CuWO₄ nanoparticles using an easy in situ sonication-assisted approach. This hybrid photocatalyst exhibits significantly improved charge separation and efficient interfacial charge transfer between its constituent phases, directly evidenced by the shifts observed in the Mott-Schottky and valence band XPS analysis. While retaining a high degree of light absorption from CuWO₄, the composite also benefits from increased structural porosity and density of surface-active sites primarily contributed by ZIF-8. Consequently, the ZIF-8@CuWO₄ composite achieved hydrogen production rates of 44 and 1,112 μmol·g^-1·h^-1 under visible and UV irradiation, respectively, exceeding the performance of the pristine ZIF-8 and CuWO₄. Reducing the overall electron/hole recombination through increasing the molarity of the sacrificial agent further improved this photocatalytic activity to 79 and 2,531 μmol·g^-1·h^-1 under visible and UV light, respectively. Despite ZIF-8@CuWO₄ photocatalytic activity decreasing over time under UV irradiation due to photo-accelerated hydrolysis, the composite remains stable under visible light. It achieves this comparable photocatalytic activity without the need for toxic or expensive noble-metal co-catalysts, highlighting its potential as a cost-effective, scalable platform for sustainable green hydrogen production.