1. Introduction

Green hydrogen, as a multifunctional and sustainable energy carrier, is increasingly recognized as a key contributor in the global transition toward a low-carbon economy^[1,2]. Photocatalytic overall water splitting (OWS) has garnered significant attention for its potential to produce green hydrogen from water utilizing solar energy^[3-8]. This process represents an artificial photosynthesis reaction without any emission of CO₂. Moreover, the produced green hydrogen can be further utilized for CO₂ hydrogenation, thus contributing to CO₂ reduction^[9-12]. Typically, photocatalytic OWS can be achieved via two distinct ways: one-step and two-step photoexcitation routes^[13,14]. For the one-step photoexcitation system, both the H₂ evolution reaction (HER) and O₂ evolution reaction (OER) occur on a single photocatalyst. It means the conduction band minimum (CBM) and valence band maximum (VBM) of the semiconductor photocatalyst should straddle the potentials of H⁺/H₂ and H₂O/O₂, respectively. In addition, to maximize the theoretical solar-to-hydrogen (STH) conversion efficiency, the semiconductor’s absorption edge should be narrow enough to harvest more photons from sunlight. However, as the absorption edge increases, the corresponding redox driving force for water splitting decreases significantly, limiting the number of reported photocatalysts capable of achieving visible-light driven one-step photoexcitation OWS^[15]. For the two-step photoexcitation system (commonly referred to as Z-scheme system), the HER and OER are performed over two different photocatalysts, leaving the residual electrons in the O₂ evolution photocatalyst (OEP) and holes in the H₂ evolution photocatalyst (HEP) “recombine” to form a closed cycle. In such case, the requirement of band structure of the applied photocatalysts is alleviated. Specifically, the HEP possesses a CBM more negative than the H⁺/H₂ potential, while the OEP possesses a VBM more positive than the H₂O/O₂ potential^[16]. Therefore, several narrow band gap semiconductors unsuitable for the one-step photoexcited approach can be employed in Z-scheme OWS systems^[17-20]. Both ways expand the possibilities for solar-driven photocatalytic OWS process, contributing to rapid advances in this field and enabling the development of diverse photocatalyst systems with varying materials, routes, and efficiencies.

Besides the aforementioned materials and efficiencies, the existing form of the photocatalyst during the reaction is a critical factor for the future practical applications. Generally, the photocatalytic OWS is evaluated using powder photocatalyst suspended in an aqueous solution (commonly referred to as the powder suspension system)^[21]. However, such system faces significant challenges for large-scale implementation, because it requires continuous stirring to maintain the uniform dispersion and is also difficult to recover/replace the photocatalyst^[22]. To tackle those problems, an immobilized photocatalyst system is developed, in which the powder photocatalyst is fixed as a thin layer on a substrate, and the formed photocatalyst sheet structure is stable under various reaction conditions, such as stirring and the pH value^[23]. Based on it, an integrated system featuring scalability, facile recovery/replacement of photocatalysts, and no extra operation for photocatalyst dispersion can be built (Figure 1), showing great potential for large-scale application of photocatalytic OWS^[24].

Display Full Size

Figure 1. Schematic diagram of the immobilized photocatalyst system for scalable OWS. OWS, overall water splitting.

Although immobilized photocatalyst systems for scalable OWS have only been explored for about a decade, significant progress has been achieved. To date, various immobilized photocatalyst systems have been developed, ranging from one-step photoexcitation route to Z-scheme approach mediated by metals, metal oxides, nonmetals, or biofilms^[25-29]. Correspondingly, multiple large-scale fabrication techniques have been established, such as programmed spraying, screen printing, particle transfer method, and suction filtration method^[30-32] (Figure 2). For example, Wang et al. reported an immobilized photocatalyst device with a size of 3 × 3 cm² prepared via the particle transfer method that could deliver an STH conversion efficiency of 1.2% at 331 K and 10 kPa without external bias. In this device, a Z-scheme OWS system was constructed by employing Ru/Cr₂O₃-loaded La and Rh codoped SrTiO₃ (SrTiO₃:La,Rh), Mo doped BiVO₄ (BiVO₄:Mo), and C as an HEP, an OEP, and an electron mediator, respectively^[28]. This efficiency is the highest among the reported particulate photocatalyst sheet systems for OWS. What’s more, a 100 m² immobilized Al-doped SrTiO₃ (SrTiO₃:Al) device fabricated via the programmed controlled spraying achieved a peak STH conversion efficiency of 0.76% under natural sunlight and operated safely for more than one year, providing a practical reference for the large-scale applications of such devices^[30]. These progresses underscore the critical role of immobilized photocatalyst devices in advancing the OWS research from laboratory prototype to large-scale demonstration, demonstrating the great potential for future practical applications. However, to the best of our knowledge, no specific review has yet systematically addressed immobilized photocatalyst devices for scalable OWS for green hydrogen production.

Display Full Size

Figure 2. Schematic diagram of the classified illustration content for the immobilized photocatalyst devices towards OWS. OWS: overall water splitting.

To fill this gap and provide a systematic overview of immobilized photocatalyst devices for efficient OWS, this review is therefore prepared. Firstly, the basics of the photocatalytic OWS process are introduced. Then, why the immobilized photocatalyst devices need to be developed is answered. Subsequently, the key advances in such devices for scalable OWS are classified and illustrated. Finally, the conclusion and future prospects in this evolving field are proposed. It is hoped that this review can motivate more researchers across interdisciplinary fields to advance the development of efficient and scalable photocatalytic OWS systems and offer valuable guidance for their practical applications.

2. Basics of Photocatalytic OWS

2.1 Photocatalytic OWS reaction process and evaluation

As a thermodynamically unfavorable process, the OWS reaction demands an external energy input to overcome the intrinsic energy barrier (Equation 1)^[33].

(1)$H_{2}O\to H_2+\frac{1}{2}{O}_2\Delta G^{\circ }=237.13\mathrm{kJ}/\mathrm{mol}$

When the CBM is more positive than the H⁺/H₂ reduction potential (-0.41 V vs. NHE at pH = 7) and the VBM is more negative than the H₂O/O₂ oxidation potential (0.82 V vs. NHE at pH = 7), the photogenerated electrons can reduce H⁺ to produce H₂, while holes can oxidize H₂O to produce O₂. Therefore, semiconductors with bandgaps larger than 1.23 eV that can straddle the H⁺/H₂ and H₂O/O₂ redox potentials are thermodynamically capable of driving the one-step photoexcited OWS process (Figure 3a). The light absorption wavelength range of the photocatalyst determines the maximum theoretical limit of STH conversion efficiency. It means the semiconductor bandgap should be as narrow as possible while still satisfying the thermodynamic requirements for water splitting. However, narrowing the bandgap generally compromises the redox ability and escalates the chance of electron-hole recombination^[34].

Display Full Size

Figure 3. Schematic energy diagrams corresponding to various categories of photocatalytic OWS systems. (a) One-step photoexcitation; (b) Two-step photoexcitation employing an aqueous redox mediator; (c) Two-step photoexcitation employing a solid-state electron mediator. OWS: overall water splitting; CBM: conduction band minimum; VBM: valence band maximum; E_g: bandgap energy; HEP: H₂ evolution photocatalyst; OEP: O₂ evolution photocatalyst; NHE: normal hydrogen electrode; Ox: oxidant; Red: reductant.

Inspired by the natural photosynthesis process in the plant, two-step photoexcited OWS aims to mimic natural photosynthesis, primarily comprising an HEP and an OEP^[35]. The Z-scheme system alleviates the stringent band structure requirements for photocatalysts, as it only requires that the HEP possess a CBM more negative than the H⁺/H₂ potential, and the OEP own a VBM more positive than the H₂O/O₂ potential. In this system, H⁺ is reduced to H₂ at the HEP, while H₂O is oxidized to O₂ at the OEP, leaving the photoexcited holes in the HEP and electrons in the OEP to recombine via an aqueous redox mediator or a solid-state electron mediator to complete the whole reaction cycle^[36] (Figure 3b,c).

In a Z-scheme OWS system, aqueous redox mediators are widely employed to facilitate electron transfer between the OEP and the HEP^[37]. Generally speaking, to facilitate efficient reduction and oxidation processes, the redox potential of the oxidant/reductant must lie between the H⁺/H₂ and H₂O/O₂ potentials. Numerous aqueous redox couples have been successfully implemented in Z-scheme OWS systems, including IO₃^-/I^-, I₃^-/I^-, Fe³⁺/Fe²⁺, [Co(bpy)₃]³⁺/[Co(bpy)₃]²⁺, [Co(phen)₃]³⁺/[Co(phen)₃]²⁺, and [Fe(CN)₆]^3-/[Fe(CN)₆]^4-[38-42]. Alternatively, electrons can be transferred between the OEP and the HEP through solid-state electron mediators. As early as 2006, Tada et al. reported a Z-scheme system using a three-component CdS-Au-TiO₂, where Au nanoparticles served as the recombination channel of electrons from TiO₂ and holes from CdS^[43]. Since then, several electron conductors, such as Ag, Au, Rh, Ni, C, indium-doped tin oxide (ITO), reduced graphene oxide (rGO), and carbon nanotube (CNT), have been identified as effective solid-state electron mediators for Z-scheme OWS systems^[26-28].

To enable a fair comparison of photocatalytic OWS performance reported by different groups, the STH conversion efficiency and apparent quantum efficiency (AQE) have been introduced and widely adopted as activity indexes^[44-46]. Specifically, the STH conversion efficiency is defined as the ratio of the chemical energy stored in hydrogen to the energy supplied by the incident solar light, as described in Equation 2:

(2)$\mathrm{STH}\left(\%\right)=\frac{(r_{H_2}\times \Delta G_r)}{(P\times S)}\times 100$

where $r_{H_2}$, ΔG_r, P, and S represent the H₂ evolution rate during the OWS reaction, the Gibbs energy of the reaction, the energy intensity of the AM1.5G (simulated sunlight), and the irradiated area, respectively.

The AQE, on the other hand, is generally defined as the ratio of the number of photons utilized in the OWS reaction to the number of incident photons at a specific wavelength. Its calculation follows Equation 3:

(3)$\mathrm{AQE}\left(\%\right)=\frac{(A\times R\times N_a)}{I}\times 100$

where A, R, N_a, and I denote constant, H₂ or O₂ evolution rate during the photocatalytic OWS reaction, the Avogadro constant and the incident photons, respectively. Herein, the values of A for H₂ and O₂ evolution are 2 and 4 in a one-step photoexcited OWS process, while 4 and 8 in a Z-scheme OWS approach, respectively.

2.2 Comparison of the powder suspension system and the immobilized photocatalyst system

Generally, the powder suspension form is applied for the particulate photocatalysts to evaluate photocatalytic OWS performance^[47]. In this case, the introduced powder photocatalyst is well-dispersed in the reaction solution, ensuring maximal light absorption and sufficient contact with the aqueous solution, which facilitates the screening and optimization of new photocatalysts. Consequently, this form is widely adopted at a laboratory scale for fundamental research^[48]. However, maintaining a stable dispersion of photocatalyst particles requires continuous mechanical stirring or circulation, which consumes external energy and becomes increasingly impractical as the reactor size increases^[49]. Moreover, the recovery and reuse of particulate photocatalysts from suspensions are challenging, as particles tend to aggregate or settle, resulting in photocatalyst loss and increased operational complexity and nondeterminacy^[50-52].

Immobilizing photocatalysts on designated substrates can effectively overcome the aforementioned issues. In such immobilized systems, stirring or circulation is no longer required to maintain the powder suspension, thereby reducing operational energy consumption and simplifying the reactor design for large-scale deployment^[53]. Furthermore, the immobilized system ensures the efficient reuse of photocatalysts without the risk of aggregation or loss, addressing the recyclability challenge inherent to powder suspension system^[54]. The robust physical contact among photocatalysts, the solid-state electron mediator and the substrate can also enhance the Z-scheme OWS process and the mechanical stability of the integrated system, enabling sustained long-term outdoor operation, which is crucial for practical applications^[55]. Compared with the powder suspension system, the immobilized photocatalyst system is more desirable for future industrial-scale implementation. Consequently, it is primarily focused in this review, and its key progress in photocatalytic OWS is summarized and elaborated in the following section. A detailed comparison between the powder suspension system and the immobilized photocatalyst system is provided in Table 1.

3. Progress of Photocatalytic OWS over the Immobilized Photocatalyst Systems

For the construction of immobilized photocatalyst devices, one-step photoexcitation systems are typically fabricated using a drop-casting process. In contrast, two-step photoexcitation systems generally require the solid-state electron mediator to effectively “link” the HEP and the OEP, consequently deriving various markedly different manufacturing processes^[56]. Therefore, this section will systematically review the recent advances in immobilized photocatalyst systems based on their different reaction approaches, photocatalysts, and preparation methods.

3.1 Immobilized photocatalyst systems in a one-step photoexcited approach

In the one-step photoexcitation approach, the OWS reaction takes place on a single photocatalyst. When it is immobilized into a device, challenges associated with photocatalyst dispersion, cocatalyst detachment during the mechanical stirring, photocatalyst recovery/replacement and the scalability can be effectively mitigated compared with powder suspension system^[24]. This section will introduce and analyze the progress and functions of such immobilized devices, providing insights for the design of efficient devices, with the aim of further deepening the understanding of the relationship between device structure and large-scale photocatalytic performance.

The primary challenge in fabricating devices by immobilizing powder photocatalysts on substrates lies in the dense accumulation of particulate photocatalysts, which hinders efficient mass and gas transfer processes. Xiong et al. addressed this issue by incorporating inert SiO₂ particles to disperse the applied Rh_2-xCr_xO₃-modified (Ga_1–xZn_x)(N_1–xO_x) photocatalyst^[57]. The resulting composite was deposited onto a 5 × 5 cm² glass plate via a drop-casting method, forming an immobilized photocatalyst device for one-step photoexcited OWS. Notably, the inclusion of hydrophilic SiO₂ particles with different sizes can significantly improve the porosity and hydrophilicity of the photocatalyst layer, thereby facilitating the mass transfer of water and gases and yielding photocatalytic activity comparable to that of the powder suspension system (Figure 4a). Meanwhile, the fabrication process of the device is simple, scalable, and flexible to diverse substrates, showing considerable potential for large-scale applications.

Display Full Size

Figure 4. (a) Schematics of photocatalyst devices prepared without SiO₂, with nanometer-sized SiO₂, and with micrometer-sized SiO₂, respectively^[57]; (b) Illustrations of the 1 × 1 m² water-splitting device; (c) A photograph of a 1 × 1 m² SrTiO₃:Al device. Republished with permission from^[25].

Building on the above result of SiO₂-mediated mass and gas transfer, Goto et al. have further developed a scalable OWS device^[25]. A demo with a size of 1 × 1 m² was fabricated using RhCrO_x-modified SrTiO₃:Al photocatalyst, comprising an array of nine photocatalyst sheets, each with dimensions of 33 × 33 cm² (Figure 4b,c). A suspension containing particulate photocatalyst, SiO₂ nanoparticles, and binder was deposited onto glass substrates via a drop-casting process, forming a particle layer with a thickness of 10-20 μm. The device employs transparent acrylic plates as the window material and is entirely tilted at 10° to facilitate the release of gas bubbles through buoyancy. A 4 mm thin water layer is designed to minimize water pressure. Under natural sunlight irradiation (65-75 mW cm^-2), the system achieved an STH conversion efficiency of approximately 0.4%, representing the highest efficiency for photocatalytic OWS at the square-meter scale at that time. The system eliminates the need for forced convection and utilizes hydrophilic window materials to prevent bubble accumulation, providing a feasible device design scheme for large-scale OWS and introducing a comprehensive system concept including gas-liquid separation.

Based on the above-mentioned experience in 1 m² of system, a 100 m² of OWS system was further designed and ultimately assembled in 2021, which is the largest in area ever reported^[30]. This achievement demonstrated its large-scale feasibility, with the design emphasizing operability, safety, and efficiency. The system comprises 1,600 panel reactor units (each with an area of 625 cm²), tilted at 30° to maximize light reception (Figure 5a). Each reactor features an ultraviolet-transparent glass window and a polycarbonate housing, with a 0.1 mm water layer between glass and photocatalyst sheet to reduce the load and prevent oxyhydrogen accumulation (Figure 5b). Supported by aluminum brackets, units use polyurethane tubes with different inner diameters for the corresponding transport of gas and water. The photocatalyst sheet is composed of SiO₂ nanoparticles, calcium chloride, and the photocatalyst of SrTiO₃:Al loaded with Rh/Cr₂O₃ and CoO_x cocatalysts, uniformly prepared via programmed spraying. Frosted glass serves as the substrate of the photocatalyst sheet to ensure wettability and mechanical strength. However, long-term outdoor operation, particularly in winter, can cause the photocatalytic particle layer to peel off due to repeated freezing and thawing of water within the panel reactors. Efficient separation of H₂ from O₂ is achieved using a commercial polyimide hollow fiber membrane, with a transmembrane pressure difference generated by a vacuum pump serving as the driving force. The high selectivity, stemming from the permeation rate of H₂ being ten times that of O₂, enables H₂ production with a purity greater than 94% and a recovery rate of 73%. This membrane technique surpasses alternative separation methods in energy efficiency and operational simplicity, making it suitable for outdoor, large-scale applications. But the current membrane is not specifically designed for H₂-O₂ separation. It is necessary to develop membranes with higher H₂ permeability and lower O₂ permeability, and optimize the gas-handling processes to reduce the energy consumption and equipment cost. Over more than a year of operation, the system achieves a maximum STH conversion efficiency of 0.76%. This value represents a peak under optimized conditions, and the system continues to face severe challenges during long-term operation. Comprising 1,600 reactor units, the system involves regular replacement of photocatalyst sheets, a maintenance process that remains cumbersome. Despite the current low efficiency and energy negative balance, the accumulation in module design, system assembly, and operational experience can benefit the development of practical, large-scale OWS systems (Figure 5c).

Display Full Size

Figure 5. (a) The image of a panel reactor unit with 625 cm²; (b) Side-view of the structure of the panel reactor unit; (c) An overhead view of the 100 m² solar H₂ production system consisting of 1,600 panel reactor units and a hut housing a gas separation facility (indicated by the yellow box). Republished with permission from^[30]; (d) Synergetic effect mechanism of promoting the forward H₂ and O₂ evolution reactions while inhibiting the reverse H₂-O₂ recombination in the photocatalytic OWS; (e) A 4 × 4 cm² Rh/Cr₂O₃/Co₃O₄-loaded InGaN/GaN nanowire wafer under the illumination of the concentrated natural solar light. Republished with permission from^[58]. OWS: overall water splitting.

In addition to applications under natural sunlight, concentrated sunlight is also considered, in which the utilization of infrared light and its thermal effect can be further modulated to enhance the STH conversion efficiency. For instance, Zhou et al. synthesized an InGaN/GaN nanowire (NW) photocatalyst thin film with high crystallinity and a well-aligned structure on commercial silicon wafers via molecular-beam epitaxy^[58]. Due to the variation of In content along the growth direction, InGaN/GaN NWs exhibit a multi-band structure, enabling a broad visible-light response range (400-700 nm)^[59,60]. Furthermore, a temperature-dependent strategy was implemented in this study. The device incorporates a heat-insulating layer to heat the reaction system to an optimal temperature of approximately 70 °C by harvesting the previously wasted infrared light (Figure 5d). This temperature enhances mass transfer and activates chemical bonds, promoting the forward HER and OER while inhibiting reverse H₂-O₂ recombination. Under the optimized conditions, the system achieves an STH conversion efficiency of 9.2% in pure water under concentrated simulated solar light. A scaled-up 4 × 4 cm² test achieves an STH value of 6.2% under natural sunlight, confirming the operational feasibility (Figure 5e). This study has successfully addressed the efficiency bottlenecks of photocatalytic OWS. Nevertheless, the validation of this strategy is currently restricted to small-sized photocatalyst devices, and the performance under large-scale application scenarios still requires further exploration. To provide a brief summary of these studies, the aforementioned representative one-step photoexcited OWS devices are summarized in Table 2.

3.2 Immobilized photocatalyst systems in a Z-scheme approach

Unlike the one-step photoexcitation approach that only needs a single photocatalyst, the Z-scheme route typically involves an HEP, an OEP and a solid-state electron mediator. In an immobilized photocatalyst system, the electron mediator should maintain intimate contact with the HEP and the OEP while avoiding the light absorption of photocatalysts^[61]. To date, various Z-scheme OWS systems have been developed, along with the corresponding immobilization methods. In this section, the research progress of the immobilized photocatalyst devices in a Z-scheme approach is outlined, including device design and performance characteristics using different types of solid-state electron mediators, such as metals, metal oxides, non-metals, and biofilms.

3.2.1 Metal as a solid-state electron mediator

Metals are firstly used as the solid-state electron mediators in Z-scheme OWS systems, playing a critical role in facilitating electron transfer from an OEP to an HEP. In 2015, Wang et al. first fabricated an immobilized photocatalyst device with an area of 9 cm² via the particle transfer method, where SrTiO₃:La,Rh and BiVO₄ functioned as the HEP and OEP, respectively, and an Au layer acted as the solid-state electron mediator^[26]. In this system, the Au layer facilitates the charge transfer between the HEP and the OEP. The photocatalytic water splitting activity of this device was 6 and 20 times higher than those of the HEP/OEP systems in powder suspension form and without a metal layer, respectively.

Subsequently, building upon this research, Wang et al. further optimized several aspects, including the photocatalyst composition and the design and preparation of the Au layer, to achieve superior performance (Figure 6a)^[31]. The optimized device consists of SrTiO₃:La,Rh, BiVO₄:Mo, and an Au layer (Figure 6b). Meanwhile, a Cr₂O₃ layer photodeposited on Ru cocatalyst-modified SrTiO₃:La,Rh can effectively suppress the reverse reaction of H₂-O₂ recombination. Additionally, annealing at 573 K for 20 min can reduce the contact resistance between the semiconductors and the Au layer, thereby enhancing electron transfer efficiency. As a result, the device achieves an STH conversion efficiency of 1.1% and an AQE of 33% at 419 nm. It represents the highest performance to date for Z-scheme OWS devices using SrTiO₃. However, its activity was significantly affected by the operating pressure. When the background pressure increased from 5 kPa to 10 kPa, the photocatalytic activity decreased to 23%. This decline is attributed to the fact that at higher pressure, H₂ and O₂ bubbles require a larger number of gas molecules to desorb from the surface of the photocatalyst sheet, and simultaneously, uninhibited backward reactions further contribute to the activity loss (Figure 6c). Although the particle transfer method faces challenges for large-scale manufacturing, this study provides a valuable technical reference for optimizing process parameters in large-scale devices.

Display Full Size

Figure 6. (a) Schematic diagram of the preparation steps for the SrTiO₃:La,Rh/Au/BiVO₄:Mo device via the particle transfer method; (b) Schematic diagram of OWS over the Ru-SrTiO₃:La,Rh/Au/BiVO₄:Mo device; (c) Time courses of the OWS reaction over a Cr₂O₃/Ru-SrTiO₃:La,Rh/Au/BiVO₄:Mo device under simulated sunlight (AM 1.5G) at 288 K and 5 kPa (open markers) and 331 K and 10 kPa (closed markers). Republished with permission from^[31]; (d) Schematic representation of fabricating alternate BiVO₄ and Au:Cu₂O strips with fixed interspace on FTO(BiVO₄||Au:Cu₂O@FTO); (e) Sketch map of photogenerated charge carriers (electrons and holes) between strips in the integrated system; (f) Photocatalytic activity of OWS over the patterned sheet (3 cm × 3 cm) deposited with Pt cocatalyst under visible light. Republished with permission from^[69]. HEP: H₂ evolution photocatalyst; OEP: O₂ evolution photocatalyst; CB: conduction band; VB: valence band; OWS: overall water splitting; FTO: fluorine-doped tin oxide.

Based on the aforesaid prototype of SrTiO₃:La,Rh/Au/BiVO₄:Mo device, this study has extended beyond metal oxide photocatalysts to explore other types of narrow bandgap photocatalysts, thereby broadening the utilized wavelength of the incident light. For example, a series of metal oxynitrides (e.g., LaMg_1/3Ta_2/3O₂N), metal oxysulfides (e.g., La₅Ti₂Cu_0.9Ag_0.1O₇S₅) and metal selenides (e.g., (ZnSe)_0.5(CuGa_2.5Se_4.25)_0.5) have been successfully applied in the immobilized photocatalyst devices, some of which even have absorption edges as long as 700 nm, demonstrating the great potential of such device for efficient Z-scheme OWS^[62-67].

In traditional Z-scheme systems, the disordered distribution between an OEP and an HEP tends to form a Type-II charge transfer pathway, reducing the utilization efficiency of charge carriers^[68]. To address this issue, Zhen et al. developed a patterned “artificial leaf”^[69]. By employing the photolithography technique to construct alternating strips of the OEP (BiVO₄) and the HEP (Au-supported Cu₂O, denoted as Au:Cu₂O) (Figure 6d), coupled with an ultrathin TiO₂ layer on the top surface, the unidirectional Z-scheme charge transfer is achieved (Figure 6e). This design eliminates competing charge transfer pathways by spatially separating active units, enabling stable production of H₂ and O₂ in pure water under visible light irradiation, with a near stoichiometric ratio of 2:1. The respective evolution rates for H₂ and O₂ are 156 and 74 μmol·m^-2, (Figure 6f). This work validates the efficacy of the patterned design, which can be scaled up via mature micro/nanofabrication techniques such as photolithography, providing a new paradigm for the development of next-generation, highly efficient Z-scheme OWS devices.

In addition to Au, some other metals with varying work functions, such as Ni and Rh, have also been identified as solid-state electron mediators in the SrTiO₃:La,Rh/Au/BiVO₄:Mo prototype device^[26]. Results indicate that the case of Rh layer exhibits an activity even comparable to that of the Au layer, while the case of Ni layer shows lower activity, partly due to the corrosion during the reaction. At present, the effect of metal layer’s work function cannot be determined unambiguously, as its chemical stability under the reaction condition varies. Further study is still needed to elucidate the mechanism by which electrons can efficiently migrate across Schottky barriers.

3.2.2 Metal oxide as a solid-state electron mediator

When metals are applied as electron mediators in Z-scheme OWS, their possible shortcomings, such as high cost, relatively high oxygen reduction reaction (ORR) activity, and poor chemical stability during the photocatalytic reaction, restrict their broader application. Metal oxide offers a promising alternative as solid-state electron mediators. Compared with metal, it is cost-effective, and shows low ORR activity and high chemical stability, rendering it more suitable for practical OWS devices. For instance, ITO has a transmittance of 56%-81% in the 420-540 nm wavelength range and much lower ORR activity than that of Au, which can reduce optical absorption loss and prohibit the side reaction. Based on this, Wang et al. developed an immobilized photocatalyst device prepared in the screen printing route. The device consists of Cr₂O₃/Ru-modified Rh-doped SrTiO₃ (SrTiO₃:Rh) as an HEP, RuO₂-BiVO₄:Mo as an OEP, and ITO as a solid-state electron mediator^[27]. Such preparation route, which does not require vacuum conditions, allows the direct fabrication of a device with an area of 30 × 30 cm² (Figure 7a,b). This system achieves an STH conversion efficiency of 0.4% and can realize stable OWS process at 5-95 kPa. Furthermore, this device can be integrated with the acetogenic bacterium Sporomusa ovata to form a bio-abiotic hybrid system. In this system, the evolved H₂ and the photogenerated electrons can be directly utilized to drive selective CO₂ reduction to acetate, achieving a solar-to-acetate efficiency of 0.7% without the need for organic additives^[70]. The transparency of the ITO layer enables it to simultaneously harvest light for both water splitting and bacterial metabolism, while its high chemical stability prevents interference with microbial activity. This integration demonstrates a scalable pathway for closed-loop carbon conversion, where solar-generated acetate is directly used for electricity generation in bioelectrochemical systems, marking a transition from isolated photocatalysis to integrated, sustainable energy cycles (Figure 7c).

Display Full Size

Figure 7. (a) Illustration of the preparation steps for a SrTiO₃:Rh/np-ITO/BiVO₄:Mo device via the screen printing method; (b) Photograph of a printed SrTiO₃:Rh/np-ITO/BiVO₄:Mo device with a size of 30 × 30 cm². Republished with permission from^[27]. (c) Schematic diagram illustrating that the sunlight-driven S.ovata|Cr₂O₃/Ru-SrTiO₃:La,Rh|ITO|RuO₂-BiVO₄:Mo hybrid supplies CH₃COO^– ions to a biohybrid electrochemical system to generate current and close the carbon cycle. Republished with permission from^[70]. IO: inverse opal; ITO: indium-doped tin oxide.

Similar to ITO, In@InO_x has also been successfully applied as a solid-state electron mediator to link SrTiO₃:Rh and BiVO₄:Mo photocatalysts^[71]. It is demonstrated that besides functioning as a charge conductor, In@InO_x can act as a particle binder to maintain the structural stability of the device. The presence of InO_x can suppress the ORR, thus enhancing photocatalytic performance.

3.2.3 Non-metal as a solid-state electron mediator

Besides metals and metal oxides, non-metals can also serve as effective solid-state electron mediators due to their good electron conductivity and relatively low ORR activity. Among non-metals, carbon has been extensively investigated because of its earth-abundant nature and low cost. For example, Wang et al. developed an immobilized photocatalyst device based on a carbon conductor layer (SrTiO₃:La,Rh/C/BiVO₄:Mo) for efficient Z-scheme OWS^[28]. This device achieves an STH conversion efficiency of 1.2% at 331 K under a background pressure of 10 kPa. At near atmospheric pressure (91 kPa), the STH conversion efficiency is maintained at around 1.0%. This demonstrates that the carbon-based photocatalytic device can efficiently operate under the ambient condition, offering an economically viable strategy for the design of practical photocatalytic OWS systems.

In addition, carbon-derived materials such as rGO and CNT have also been proven effective as solid-state electron mediators. In 2025, Gu et al. developed an immobilized photocatalyst device, where Sm₂Ti₂O₅S₂ and BiVO₄ particles were interconnected by rGO or CNT to mediate the interfacial charge transfer^[32]. However, unlike two-dimensional rGO, CNT does not excessively wrap photocatalysts or block their active sites; so the device using CNT exhibits superior performance. Notably, this device is fabricated via a simple filtration process (Figure 8a). This approach eliminates the need for vacuum/high-temperature processes or organic media, offering low-cost and straightforward operation. Moreover, large-scale production can be readily achieved through repeated filtration or redispersion, enabling the device to be scaled up to large dimensions (verified device with a diameter of 7 cm) (Figure 8b). The system achieves an STH conversion efficiency of 0.4% at 323 K and 10 kPa, operates stably at near atmospheric pressure (1 bar), and continuously produces H₂ and O₂ for 270 h without noticeable attenuation (Figure 8c).

Display Full Size

Figure 8. (a) Schematic diagram of the fabrication procedures for a Sm₂Ti₂O₅S₂/CNT/BiVO₄ device via a facile filtration method; (b) A photo of the device with a diameter of 7 cm; (c) Time courses of the evolved H₂ and O₂ gases and their evolution rates over the Sm₂Ti₂O₅S₂/CNT/BiVO₄ device irradiated with visible light (λ > 420 nm) at 288 K under an initial pressure of 5 kPa. Republished with permission from^[32]. STOS: Sm₂Ti₂O₅S₂; CNT: carbon nanotube; BVO: BiVO₄; IPA: isopropanol.

3.2.4 Biofilm as a solid-state electron mediator

Biofilms have recently been verified to be feasible for solid-state electron mediators. More importantly, their inherent capabilities for self-regeneration and environmental responsiveness can avoid the frequent maintenance of the constructed device and endow it with long-term operational stability. For instance, Wang et al. developed a sustainable hybrid Z-scheme device for visible-light-driven OWS by integrating inorganic photocatalysts with conductive bacterial biofilms^[29]. The device employs Cr₂O₃/Ru/SrTiO₃:Rh as an HEP and CoO_x/BiVO₄:Mo as an OEP. It is fabricated by drop casting a mixture of photocatalysts onto a glass plate, followed by biofilm growth that forms a conformal conductive layer through oxidative polymerization of pyrrole molecules (Figure 9a). It enables the release of H₂ and O₂ at a stoichiometric ratio of 2:1. Owing to the self-regenerative property of biofilms and the chemical stability of polypyrrole, the device exhibits less than 10% efficiency loss after 100 h of the reaction. Additionally, it shows excellent resistance to pressure variations, retaining 91% of its activity when the background pressure is enhanced from 4.5 to 100 kPa (Figure 9b). Importantly, biofilms can be prepared on a large scale through simple cultivation, and devices with different areas (2.25-12.25 cm²) show a standard deviation of only 2.4% in performance (Figure 9c). The process does not require sophisticated equipment such as vacuum evaporation, making it suitable for scalable production.

Display Full Size

Figure 9. (a) Illustration of the fabrication procedures for the [Bi]-biofilm/PPy-[Sr] device on a glass substrate preceding its transfer to another substrate; (b) Effect of background pressure on H₂ and O₂ evolution rates for the [Bi]-biofilm/PPy-[Sr] photocatalyst device; (c) Gas evolution rates for [Bi]-biofilm/PPy-[Sr] devices with different areas^[29]. [Bi]: CoO_x-BiVO₄:Mo; [Sr]: Ru@Cr₂O₃-SrTiO₃:Rh; PPy: polypyrrole.

Among the discussed electron mediators, noble metal mediators offer high charge transfer efficiency in Z-scheme OWS systems but are hampered by distinct drawbacks, such as high cost, pressure sensitivity, and susceptibility to side reactions like the ORR. Metal oxides present a more balanced alternative, with moderate cost, superior stability under ambient conditions, and relatively low ORR activity. However, their higher interfacial resistance limits charge transfer efficiency. Non-metals possess attractive features like low cost and low ORR activity, yet their performance strongly depends on the degree of integration between the photocatalyst particles and the non-metal mediator. Biofilms, although attractive for several advantages such as self-regeneration and simple preparation, demonstrate lower charge transfer efficiency and are prone to deactivation under high temperatures or extreme pH conditions. Therefore, the selection of a proper electron mediator for effective photocatalytic Z-scheme OWS will necessarily be a compromise.

3.2.5 Free of solid-state electron mediator

Beyond the systems described above that utilize metallic or non-metallic media, a special case exists in which the interfacial charge transfer between HEP and OEP occurs directly through physical contact, without the use of a solid-state electron mediator. Sasaki et al. reported a mediator-free Z-scheme OWS system comprising SrTiO₃:Rh and BiVO₄ photocatalysts^[72]. In this system, direct contact is achieved through powder aggregation under acidic conditions, allowing interparticle electrons to transfer from the CBM of BiVO₄ to the Rh³⁺/Rh⁴⁺ redox level in SrTiO₃:Rh, thereby ensuring efficient charge transfer. Based on this result, they further developed BiVO₄–SrTiO₃:Rh composite photocatalysts, where SrTiO₃:Rh particles attached to BiVO₄ particles^[73]. Such composite photocatalysts can realize effective interparticle electron transfer without adjusting the pH value of the reaction solution, enabling pure water splitting with an AQE of 1.6% at 420 nm. These series works collectively provide a new strategy for constructing the immobilized photocatalyst device in a Z-scheme approach for OWS, avoiding potential drawbacks associated with electron mediators during the reaction, such as light-blocking effects and competitive occupation of active sites.

In this section, recent advances in immobilized photocatalyst devices for efficient and scalable OWS using one-step or two-step photoexcitation approaches are classified and summarized. For the case of the one-step photoexcitation approach, the whole OWS reaction can occur over a single photocatalyst, featuring a relatively simple structure, but it faces stringent restrictions on the band structure and active sites. In contrast, the Z-scheme route generally requires the HEP, the OEP and an electron mediator, all of which must be carefully designed and coupled for the effective interfacial charge transfer. Specifically, various types of solid-state electron mediators for immobilized Z-scheme OWS devices, such as metals, metal oxides, non-metals and biofilms, are concluded and illustrated. A concise overview of aforementioned representative Z-scheme OWS devices is provided in Table 3. Based on the current research progress, it is expected that further breakthroughs in the aspects, including but not limited to material screening, device construction, and system integration, will be made to advance the development of the immobilized photocatalyst systems toward practical applications.

4. Conclusion and Outlook

This review systematically summarizes the main research progress of the immobilized photocatalyst devices for scalable OWS in terms of materials screening, device construction, and system integration, aiming to provide valuable guidance for translating efficient photocatalytic OWS systems from laboratory prototypes to engineering implementation. Despite rapid progress in this field, the STH conversion efficiency and the scalability level remain below expectations. To further address these challenges, the following issues are highlighted, and corresponding recommendations for future development are proposed.

(1) Material screening and innovative synthesis. To achieve a higher STH conversion efficiency, it is crucial to develop efficient photocatalysts with narrow bandgaps^[74-78]. Furthermore, photocatalyst environmental safety must be fully considered, such as the heavy metal contamination caused by Cd²⁺ leaching from Cd-based photocatalysts. The traditional materials discovery relies on empirical iterative optimization or sequential testing, which is time-consuming, labor-intensive and constrained by human experience, making it difficult to identify globally optimal combinations. Artificial intelligence (AI), by learning structure-activity relationships, can precisely guide high-throughput screening and enable rapid evaluation of hundreds of combinations^[79-83]. For instance, Li et al. employed machine learning to screen ternary organic heterojunction photocatalysts from a vast combinatorial space^[82]. By training a model on electronic and structural descriptors, they identified effective candidates with high H₂ evolution rates ever reported while reducing experimental screening by over 80%. Such integration of high-throughput screening with AI/machine learning can realize the transformation of future materials from design to targeted synthesis, offering an effective solution toward efficient and environmentally friendly photocatalysts.

(2) Design and construction of the innovative devices. Current immobilization techniques for powder photocatalysts often suffer from weak interfacial adhesion and low mass transfer efficiency^[84-88]. To address these limitations, it is essential to modify the interfacial contact between the substrate and the particulate photocatalysts, such as roughening, chemical treatment, and plasma activation^[89-92]. But for certain non-oxide photocatalysts, substrate roughening requires careful pre-treatment to preserve chemical stability, as excessive roughness can induce cracking in brittle lattices. In addition, alternative approaches, including binder-free assembly, low-temperature sintering, and self-assembly driven by charge or polarity differences between components, may also offer promising solutions. However, low-temperature sintering is limited to conditions below the decomposition point of materials to prevent degradation. These strategies will not only improve the mechanical stability of the photocatalyst layer, but also enhance the electronic coupling between photocatalysts and mediators, ultimately promoting immobilized photocatalyst devices^[93].

(3) Understanding of the reaction mechanism. To provide a solid theoretical basis for the design of immobilized photocatalyst devices, it is crucial to deepen the understanding of the entire reaction process. Herein, advanced in situ time- and space-resolved analytical techniques, combined with theoretical calculations, are highly recommended for probing the real reaction process at both temporal and spatial dimensions. These techniques can help elucidate the details of the rate-determining step, and facilitate the design and construction of efficient photocatalyst systems^[94-96].

(4) Development of application-oriented and scalable devices and systems. The scaling of OWS systems presents multifaceted challenges, such as safety risks, net energy output, and integration with hydrogen infrastructure. Overcoming these interconnected barriers requires the development of efficient, practical, and easily scalable novel technologies. Thin film preparation techniques, such as roll-to-roll processing^[97,98], 3D printing^[99-101], and interfacial polymerization^[102,103], show substantial application potential. Such methods not only eliminate energy-intensive vacuum requirements and enable continuous operation, but also facilitate the design of integrated systems that are inherently safer and more resource-efficient. Furthermore, the development of efficient H₂-O₂ separation membranes (e.g., metal-organic frameworks^[104]) and advanced H₂ liquefaction technologies can mitigate the risk of explosive H₂-O₂ mixture while reducing the operational cost. By integrating these key technologies, it is expected that the net energy output can become positive.

To summarize, the immobilized photocatalyst devices for OWS have emerged as a promising approach for practical solar hydrogen production, exhibiting significant advantages over the powder suspension systems, such as scalability, facile recovery/replacement of photocatalysts, and the elimination of additional operations to disperse photocatalysts. Over the past decade, remarkable progress has been achieved in this field, particularly in materials screening, device construction, and system integration for scalable applications. Nonetheless, current technologies still require further development to meet the demands of large-scale implementation. It is our sincere hope that this review will attract more scientists from diverse disciplines to advance this field and accelerate its transition to industrial-scale applications.

Authors contribution

Chen S, Ma B: Supervision, writing-review & editing.

Shao Y, Wang S: Conceptualization, writing-original draft.

All of the authors discussed and approved the final version of the manuscript.

Conflicts of interest

The authors declare that they have no conflicts of interest.

Ethical approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and material

Not applicable.

Funding

This work was financially supported by the National Natural Science Foundation of China (22272082, 22572089), the Funds for International Cooperation and Exchange of the National Natural Science Foundation of China (W2421036), Beijing-Tianjin-Hebei Natural Science Foundation Cooperation Project (25JJJJC0010), and the Fundamental Research Funds for the Central Universities, Nankai University (63213098).

Copyright

© The Author(s) 2025.