The population of the earth was 7.7 billion at the end of 2018, which is expected to reach 9.5 billion by 2050 [1]. With such rapid growth in the population and the advancement of society, the need for energy will become larger and larger. Fossil energy as a non-renewable energy source is currently widely used. However, it cannot meet the continuous demand for energy over the next one or two hundred years. At the same time, the massive use of fossil energy has brought about a series of environmental pollution problems. Therefore, it is urgent to find new clean energy sources.

The radiant power of the sun at the earth's surface is as high as 10⁵ TW, which is far higher than the human demand for energy [2]. Therefore, solar energy can meet the energy needs if the conversion efficiency can be increased to 10% [3]. Converting solar energy into storable and transportable chemical energy is a viable method of developing and utilizing solar energy on a large scale, owing to the discontinuity and instability of the solar energy resource. Photoelectrochemical (PEC) water splitting can convert solar energy to clean hydrogen energy, which is a promising way of utilizing solar energy [4]. PEC water splitting was first reported by Fujishima and Honda in 1972 [5], subsequently, it has been studied extensively. A photoanode and a photocathode are necessary to construct a PEC water splitting system. The water oxidation reaction taking place at the photoanode is usually considered as the kinetic-controlling step because it involves four-electron transfer. Therefore, developing a high-performance photoanode is crucial to constructing an effective PEC water splitting system.

In order to realize large-scale high-efficiency PEC water splitting, the photoelectrodes need to satisfy the following conditions: (1) substantial amount of visible light must be absorbed; (2) it must display appropriate band alignments for water oxidation and reduction; and (3) it should be chemically stable and low cost. As an n-type semiconductor for the photoanode, WO₃ can meet the above requirements. Its valence band position is about 3 V vs. NHE, which can provide enough overpotential to oxidize H₂O to form O₂ [6]. Furthermore, WO₃, with the band gap of 2.6 eV, can absorb part of the visible light spectrum. Therefore, WO₃ has been identified as one of the most promising photoanode materials (as shown in Fig. 1).

WO₃ is a kind of yellow solid at room temperature that is stable in acidic solutions. As shown in Fig. 2(a), WO₃ exhibits the ABO₃ perovskite structure, in which the element A is missing and the element B is W [7, 8]. W is in the middle of the octahedron, whereas O is at the apex of the octahedron; each octahedron is connected by O points. However, actually prepared WO₃ reveals the structure of a distorted octahedron owing to antiferroelectric displacements of the W atoms and mutual rotations of the O octahedra. Pure WO₃ exhibits at least five phases (α, β, γ, δ, and ε) in the range 900 to 180 ℃, depending on the temperature [9-13]. In addition, orthorhombic [11] and hexagonal [14] phases were also reported, as shown in Fig. 2(b). The energy band structures of these phases were calculated by using a simplified model, shown in Fig. 2(c) [15].

In 1976, WO₃ was first reported as a photoanode for PEC water splitting [16]. However, the efficiencies of unmodified WO₃ photoanodes for water oxidation widely differ from their theoretical values because of excessive electron-hole recombination and poor water oxidation kinetics. In addition, the surface states of nano-WO₃ that originate from W-O dangling bonds or defects can also limit the water oxidation performance of WO₃ photoelectrode [17, 18]. Furthermore, side reactions take place on the WO₃ photoanode owing to its high valence band position, leading to low Faradic efficiencies (normally lower than 80%) of water oxidation on WO₃ photoanode [19, 20]. The anions (e.g., Cl⁻, SO₄²⁻) in the solution, which are present as supporting electrolytes, can interfere with the photooxidation of water [21]. Therefore, various strategies have been reported to optimize the WO₃ photoanode. Herein, an overview of the recently reported strategies, such as controlling the morphology, introducing defects, constructing a heterojunction, loading a cocatalyst, and using the plasmonic effect, will be provided.

Commonly, semiconductors with the morphologies of nanowire, nanoflake, and mesoporous are beneficial for the transfer of photogenerated holes from the bulk to the surface. Solvothermal technique was used by Grimes et al. [22] to grow WO₃ nanowire arrays on F-doped tin oxide (FTO) substrates, as shown in Fig. 3(a). This nanowire array morphology can be changed to two types of nanoflake arrays (Fig. 3(b) and (c)) by adjusting the solvothermal conditions. The study showed that the second type of nanoflake arrays (Fig. 3(c)) exhibited the best performance for PEC water oxidation, which resulted in the highest saturation photocurrent value of 1.43 mA/cm². The high performance was ascribed to the greater thickness of the second-type nanoflake arrays, because WO₃ is an indirect band gap semiconductor with a small light absorption coefficient and, therefore, greater thickness benefits light absorption. Although the first type of nanoflake arrays displayed almost the same thickness as that of the nanowires, these arrays exhibited a greater photocurrent density than that of the nanowires owing to the smaller band gaps and lower light scattering. The nanosheet structure was also synthesized by Zheng et al. [23] and other research groups [24-27] by using a similar method to confirm its good performance. Dense WO₃ nanosheets can also vertically grow on W substrate for PEC water splitting [28-30]. The WO₃ nanosheet prepared by this method displayed a low PEC water oxidation overpotential.

Recently, Kafizas et al. [31] fabricated a nanoneedle-structured WO₃ photoanode by a chemical vapor deposition method. This WO₃ photoanode was composed of a seed layer and a top layer of WO₃ nanoneedles. The WO₃ photoanode with a ∼300 nm seed layer and ∼5 μm thick nanoneedles showed the highest water splitting activity. WO₃ nanotubes and nanowires could also be synthesized by flame vapor deposition [32, 33] and other methods [34, 35].

Thin layer semitransparent WO₃ electrodes were prepared on conductive FTO substrates by using a one-step sol-gel method [36]. By employing a higher annealing temperature of 700 ℃, the photocurrent could reach 4.2 mA/cm² under simulated solar AM 1.5G irradiation. A thinner transparent WO₃ thin film can be prepared with the assistance of polyvinylpyrrolidone and ammonium metatungstate by using spin coating and screen printing methods [37]. This kind of transparent WO₃ thin film could be a potential candidate for use in tandem configurations for water splitting application. Surfactant-thermal-carbonization method was used to synthesize a tunable mesoporous WO₃ electrode by Yagi and coworkers [38]. Another morphology such as a mesoporous thin film WO₃ photoanode can be fabricated by a sol-gel dip coating approach [39] or foaming-assisted electrospinning strategy [40]. These structures showed improved solar energy conversion efficiencies. Pore-rich WO₃ is beneficial for the utilization of the photogenerated holes in water oxidation [41]. Unusual sandwich-structured WO₃ nanoplate arrays [42] were also fabricated for PEC water splitting. All these well-designed photoanodes display higher PEC water oxidation performances owing to a high surface area or small thickness, which facilitates easy transport of charge carriers to the surfaces.

Defects in materials can act as active sites in catalytic applications [43-47]. Hütter et al. [48] reported the effects of physical defects, such as microsized holes or cracks, and chemical defects, such as oxygen vacancies, on the PEC water splitting performance of WO₃. The results showed that physical defects inside the film increased the resistance to charge transfer and also resulted in a higher recombination rate, which inhibited the photocurrent generation. Chemical defects yielded an increased adsorption of OH groups on the film surface and enhanced the PEC efficiency. Excess of the chemical defects also inhibited the electron transfer, thus decreasing the photocurrent generation. Dual O and W vacancies on the surface of WO₃ were generated by dissolving ethylenediamine solution with Li, as shown in Fig. 4 [49]. The WO₃ surrounded by a defective overlayer exhibited a photocurrent density that was approximately 2.4 times higher and a marked cathodic shift in the onset potential for water oxidation. Similar work was carried out by Lee et al. [50], who employed acid-mediated hydrothermal treatment. The transfer ability of the charge carriers can be enhanced by the O vacancies present in WO₃ [20, 51, 52].

Self-doped WO_3−x nanoflake arrays could be synthesized by the serial process of dealloying of amorphous Fe-W alloy, thermal treatment, and cathodic polarization [53]. It was found that the optimized WO_3−x showed a photocurrent density of 8.7 mA/cm² in the presence of methanol as a hole scavenger, which was five times larger than that of pristine WO₃ nanoflakes.

Construction of a heterojunction is considered to be an effective method of improving the separation efficiency of the photogenerated charges or sometimes increasing the light absorption efficiency [54-62]. Herein, the combination of WO₃ with another semiconductor as the photoanode for PEC water oxidation is reviewed.

TiO₂ is a widely used photocatalyst [63-68]. It has the advantage of being stable and low cost. Coupling TiO₂ with WO₃ to form a WO₃/TiO₂ heterojunction photoanode is reported to improve the PEC water splitting performance. Ludwig et al. [69] reported the preparation of layered WO₃/TiO₂ films by magnetron sputtering. Various structures of WO₃/TiO₂ films could be prepared by adjusting the sputtering pressures and depositing the films in the form of wedges. The porous WO₃/TiO₂ photoanode showed significantly enhanced photocurrent densities, compared to those of individual TiO₂ and WO₃. The authors thought that the enhancement in the photoconversion efficiency was attributed to an increased surface area and efficient charge transfer within the layered nanostructures. Subsequently, more research on WO₃/TiO₂ heterostructure for water splitting [70] or photodegradation [71, 72] has been carried out and reported. Recently, the charge transfer mechanism in the WO₃/TiO₂ heterostructure for PEC water splitting was deeply studied [73-75]. A hole scavenger was used to realize effective charge separation at the heterojunction during the PEC tests. With Ni^Ⅱ meso-tetra(4-carboxyphenyl)porphyrin as a water oxidation catalyst that was loaded on WO₃/TiO₂ heterostructure, the water oxidation efficiency could reach 81% [76].

BiVO₄ with the band gap of 2.4-2.6 eV and a suitable band structure has gained considerable attention in the field of photocatalysis [77, 78]. Nosaka et al. [79] reported the deposition of a layer of BiVO₄ on WO₃ by spin coating to form WO₃/BiVO₄ heterostructure. The incident photon-to-current conversion efficiency (IPCE) of the WO₃/BiVO₄ electrode increased by ten times, compared to that of BiVO₄ electrode. The enhanced performance of the WO₃/BiVO₄ composite film electrode was mainly ascribed to the effective electron-hole separation at the semiconductor heterojunction. Subsequently, Grimes' team [80] reported the deposition of a BiVO₄ layer on WO₃ nanorods by spin coating. They found that the heterojunction structure offered an enhanced photoconversion efficiency and increased photocorrosion stability. Furthermore, the nanorod array films showed significantly improved PEC properties, compared to those of planar WO₃/BiVO₄ heterojunction films. The authors believed that the improved PEC properties could be ascribed to the high surface area and improved separation of the photogenerated charge carriers at the WO₃/BiVO₄ interface. These results were further confirmed by the subsequent work of Zheng and coworkers. They synthesized WO₃/BiVO₄ core/shell nanowire photoanode, in which BiVO₄ was the primary light-absorber and WO₃ acted as an electron conductor (Fig. 5) [81]. At a potential of 1.23 V vs. RHE, the photocurrent of 3.1 mA/cm² under simulated sunlight and an IPCE of ∼60% at 300−450 nm were obtained. In addition, Pihosh and coworkers fabricated well-separated and vertically oriented WO₃ nanorods capped with extremely thin BiVO₄ absorber layers for photocatalytic water oxidation [82]. The optimized WO₃/BiVO₄ photoanode modified with Co-Pi oxygen evolution cocatalyst showed remarkably stable photocurrents of 3.2 and 5.1 mA/cm² at 1.23 V vs. RHE in stable Na₂SO₄ electrolyte under simulated solar light at the standard 1 Sun and concentrated 2 Suns illuminations, respectively.

Recently, Yu and coworkers [83] reported WO₃/BiVO₄ spherical arrays for highly efficient PEC water splitting. The nanojunction, the W doping, and the O vacancies in the WO₃/BiVO₄ spherical array structure were believed to enhance the light harvesting efficiency through multiple light scattering and enhanced charge separation and transfer efficiencies. With the loading of cobalt phosphate, the 3D WO₃/BiVO₄/cobalt phosphate composites yielded a significantly improved PEC performance [84]. Further research demonstrated that the formation of a heterojunction between WO₃ and BiVO₄ could improve the PEC water splitting performance of WO₃ [85-90]

The dynamics of the photogenerated charge carriers in the WO₃/BiVO₄ heterojunction photoanode have been widely studied [91]. Research on charge separation showed that WO₃/BiVO₄ interfacial engineering was the preferred choice for improving the overall PEC activity [92] and that the separation in WO₃/BiVO₄ was wavelength-dependent [93]. Selli and coworkers [94] probed the photoinduced charge transfer dynamics of WO₃/BiVO₄ photoanodes through midinfrared transient absorption spectroscopy. Nanosecond midinfrared transient absorption experiments confirmed that charge carrier separation occurs in WO₃/BiVO₄ electrodes under visible-light excitation that persists up to the microsecond time scale.

Fe₂O₃ has a narrow band gap of about 2.1 eV, which suggests that it can absorb a large proportion of visible light. In addition, it offers the merits of excellent stability, abundance, and low cost [95, 96]. Zou's group [97, 98] prepared WO₃/Fe₂O₃ heterostructure for PEC water oxidation by the sol-gel method. The results showed that the photocurrent of the WO₃/Fe₂O₃ film was higher than that of WO₃ or Fe₂O₃ alone. The authors believed that a reasonable explanation for this could be that the photogenerated electrons transferred more easily in WO₃/Fe₂O₃ than in WO₃ or Fe₂O₃ owing to their unique conduction band structures. Therefore, the interface between WO₃ and Fe₂O₃ played an important role in improving the conversion efficiency. Subsequently, researchers deposited a thin layer of Fe₂O₃ nanoparticles on WO₃ by CVD to fabricate host-guest electrodes [99], which showed 20% increase in the photocurrent. The improvement was attributed to an increase in the absorbed photon conversion efficiency, especially for longer wavelengths, for which the photon penetration depth was large in hematite. Later, Diao and coworkers [100] reported the fabrication of WO₃ nanoneedle/α-Fe₂O₃ heterojunction photoanodes for efficient PEC water splitting. In this heterojunction, the high-aspect-ratio single crystalline WO₃ nanoneedles acted as both the photoanode framework and the light harvester in the UV and blue visible regions, while the porous layer of small Fe₂O₃ nanocrystals acted as a cooperative light harvester in the visible region. The WO₃/Fe₂O₃ heterojunction photoanode exhibited an extended light absorption band in the visible region and a significantly enhanced photocurrent density, which was ca. 1.6 times higher than that obtained for pure WO₃ NNs. Recently, NiFe-layered double hydroxide was decorated on the WO₃/Fe₂O₃ heterostructure to further improve its photocurrent density [101]. As shown in Fig. 6, the Ni²⁺ in the NiFe-layered double hydroxide was oxidized by the photogenerated holes to Ni⁴⁺, which in turn oxidized water and reduced back to Ni²⁺.

Recently, CQDs have attracted broad attention for photocatalysis application owing to the characteristics of stable photoluminescence and absorption in the UV and near-visible regions [102]. WO₃ nanoplate-decorated CQDs were successfully synthesized by Zheng and coworkers [103]. A WO₃ photoanode loaded with CQDs exhibited an enhanced photocurrent density under simulated solar light illumination. Simultaneously, the WO₃/CQDs photoanode showed strong visible light absorption. Meanwhile, the WO₃/CQDs photoanode revealed a low onset potential owing to changes in the band energy positions. The schematic of the process is shown in Fig. 7; WO₃ and CQDs are excited by the incident light to generate electrons and holes. These carriers are easily separated because of the staggered energy band positions, which facilitate the transfer of electrons and holes and thereby improve the PEC performance.

Further research revealed that the enhancement in the efficiency in the case of WO₃/CQDs was ascribed to the increased light harvesting ability of the CQDs and the accelerated charge transfer across the interface between the WO₃/CQDs electrode and the electrolyte [104, 105].

In addition to these above mentioned heterojunction structures, WO₃/Sb₂S₃ heterojunction [106], BiVO₄/WO₃/SnO₂ double-heterojunction [107], WO₃/Cu₂O heterojunction [108], WO₃/CuO heterojunction [109], WO₃/ZnWO₄ heterojunction [110], WO₃/CuWO₄ heterojunction [34, 111, 112], WO₃/Cr₂O₃ heterojunction [113], WO₃/conducting polymer heterojunction [114], and rGO-WO₃ composite [115] were fabricated for enhanced PEC water oxidation.

Both heterogeneous and homogeneous cocatalysts have been extensively studied for the photocatalytic [116-124] and photoelectrocatalytic [125-132] water splitting process.

To improve the stability of WO₃ in aqueous solutions with pH > 4, Wang and coworkers coated WO₃ with a Mn-based catalyst, which was prepared by thermally decomposing the Brudvig-Crabtree catalyst [(H₂O)-(terpy)Mn(O)₂Mn(H₂O)(terpy)] (NO₃)₃ (terpy = 2, 2':6', 2''-terpyridine) [133]. As shown in Fig. 8, the photogenerated holes transferred to the Mn catalyst, assisted by the formation of an electric field within the semiconductor. Then, the Mn catalyst was oxidized to a higher valence state, which oxidized water.

The Mn catalyst could act as a protective layer that maintained the stability of WO₃. WO₃ coated with Mn catalyst could produce more O₂ than WO₃ without the Mn catalyst. After 2 h, the performance of the WO₃ electrode with the Mn catalyst decreased by approximately 4%, whereas that of the WO₃ electrode without the catalyst dropped by 60% within 1 h. This is the first study that reports that WO₃ is stable in neutral solutions.

Choi's group [134] loaded Co-Pi onto WO₃ and studied its PEC water oxidation performance. The loading of Co-Pi increased the oxygen evolution Faradaic efficiency from 61% to nearly 100% at 0.8 V vs Ag/AgCl. Co-Pi could suppress the formation of peroxides, resulting in long-term stability of the WO₃/Co-Pi photoanode. In addition, Zhou's research group [135] loaded Co-Pi on WO₃ by photo-assisted electrodeposition. In this research, Co-Pi facilitated charge transfer through the formation of a junction between WO₃ and Co-Pi.

Spurgeon et al. [136] supported IrO₂ on WO₃ by sintering, sputtering, drop coating, and electrodeposition. The authors found that the coverage of the cocatalysts had a great influence on the performance of the WO₃/IrO₂ photoanode. Sputtering was the most effective method for loading IrO₂; therefore, it showed the largest photocurrent density. The WO₃/IrO₂ photoanode prepared by sputtering displayed the greatly increased Faradaic oxygen production efficiency of 100%.

He and coworkers [137] reported the electrochemical deposition of NiFe-LDH decorated WO₃ nanorod arrays as PEC oxygen evolution electrodes. Fig. 9 exhibits the typical morphology of the obtained WO₃@NiFe-LDH nanorod arrays. When the deposition time is 200 s, full coverage of the ultrathin NiFe-LDH on the WO₃ nanorod arrays can be observed. A shorter deposition time leads to a smaller amount of NiFe-LDH nanoflakes that do not cover the surface of the WO₃ nanorod arrays (Fig. 9a), whereas a longer deposition time results in thicker NiFe-LDH nanoflakes (Fig. 9(c)). The WO₃/NiFe-LDH electrode displays a lower onset potential for water oxidation than WO₃ and a higher photocurrent density. After optimization, the WO₃/NiFe-LDH electrode reveals an enhanced higher photocurrent density of 1.10 mA/cm² at the potential of 1.20 V vs. SCE. The authors thought that the promising performance was attributed to WO₃ providing a superior framework of photoanodes as well as to the prospective light harvesting property.

Ding's group [138] coated WO₃ with a thin FeOOH layer by photodeposition for PEC water oxidation. The WO₃/FeOOH anode displayed longer stability than the WO₃ anode owing to the protective effect of the FeOOH layer. Furthermore, as an excellent water oxidation catalyst, FeOOH improved the photocurrent of the WO₃ anode, which was 1.3 mA/cm² at 1.23 V RHE under simulated solar illumination. The authors found that a p-n junction was formed between WO₃ and FeOOH due to solvation effect that facilitated the transfer of the photogenerated carriers and thereby improved the PEC performance. FeOOH was also used to improve the performance of the WO₃ photoanode in a tandem water splitting device [139]. The similar CoOOH-loaded WO₃ photoanode also showed enhanced photoelectrocatalytic activity [140].

Recently, Li and coworkers [141] reported FeOOH and NiOOH co-deposited nanotube array-like WO₃ (WA) photoanode (WA-FeNi) for PEC water splitting. As shown in Fig. 10(a), FeOOH could reduce the recombination rate of the photogenerated carriers at the interface between WO₃ and NiOOH, whereas NiOOH decreased the recombination between NiOOH and the electrolyte and increased the water oxidation activity significantly.

CoO_x nanoparticles are known to be an excellent water oxidation catalyst [143-146]. Li and coworkers [142] reported the deposition of CoO_x nanoparticles on the surface of WO₃ plate arrays by chemical deposition method. The CoO_x/WO₃ film showed enhanced PEC performance, compared to that of a bare WO₃ film. The highest IPCE increased from 24.9% to 49.1% when CoO_x was introduced. The proposed process is presented in Fig. 10b. Upon illumination, WO₃ is excited to produce electrons and holes. Then, the Co in CoO_x is oxidized by the holes to Co⁴⁺, which in turn oxidizes water and is reduced back to Co²⁺. The electrons subsequently transfer to the cathode under an applied bias to reduce the protons.

Ding's group [147] also reported on the loading of CoO_x nanoparticles on WO₃ nanosheets for PEC water oxidation. CoO_x nanoparticles with the diameter of about 5 nm were loaded on WO₃ nanosheets by using hydrothermal method. Upon loading the CoO_x nanoparticles, the WO₃ photoanode showed improved charge separation efficiency and charge injection efficiency owing to the formation of a p-n junction between WO₃ and CoO_x and the excellent catalytic performance of CoO_x nanoparticles. Rotating ring-disk electrode system test results indicated that the selectivity towards the water oxidation reaction was enhanced. With the loading of the CoO_x nanoparticles, water was oxidized to O₂, rather than to H₂O₂, which improved the stability of the WO₃ photoanode.

Homogeneous cocatalysts can also accelerate the water oxidation reaction [148-156]. Bartlett's group [157] reported the loading of Fe(tebppmcn)Cl₂ (tebppmcn = tetraethyl N, N′-bis(2-methylpyridyl-4-phosphonate)-N, N′-dimethylcyclohexyldiamine) (1) with a phosphonate linkage to the WO₃ photoelectrode. The loading of 1 dramatically increased the rate of PEC water oxidation. At the same time, the selectivity of WO₃ for water oxidation increased from 56% to 79%.

Ma et al. [158] reported that bpy-Co²⁺ complex (bpy = 4, 4′-bipyridine) can be used as an efficient molecular catalyst to modify WO₃ nanoplate array photoanode. An enhanced photocurrent density of 2.88 mA/cm² at 1.5 V vs. RHE could be realized with the modified photoanode, which was 1.7 times larger than that of the WO₃ photoanode. The Ir complex [(H₄dphbpy)Ir^Ⅲ(Cp*)Cl]Cl (Ir-PO₃H₂; H₄dphbpy = 2, 2′-bipyridine-4, 4′-bisphosphonic acid, Cp* = pentamethylcyclopentadiene) was also reported to enhance the interfacial charge transfer in the WO₃ photoanode through the formation of the composite of WO₃ + Ir-PO₃H₂ [159]. This WO₃ composite photoanode showed significant improvements in both the photocurrent and Faradaic efficiency, compared to those of the bare WO₃ photoanode.

Plasmonic effect has been extensively introduced in the solar water splitting field in recent years [160-168]. Photoelectrodes with plasmonic effect usually combine a semiconductor with noble metal nanoparticles, which exhibit a strong surface plasmon resonance, for improving the photo-to-chemical conversion efficiency. Augustynski's group [169] reported on the loading of Ag nanoparticles to enhance the photocurrent of WO₃ for PEC water oxidation. This early research confirmed that the photo-to-chemical conversion efficiency and photocurrent of WO₃ photoanode could be improved by loading plasmonic Ag nanoparticles. Thereafter, Diao's group [165] reported the loading of Au nanoparticles on WO₃ photoanode. The Au/WO₃ photoanode exhibited a significantly improved photoactivity toward PEC water oxidation. Not only the saturation photocurrent density increased, but also a negatively shifted onset potential could be observed. Other similar works yielded the same conclusion [170-172].

A comparison of the performances, including the photocurrent density and stability time, is presented in Table 1.

In this review, we summarized a list of strategies aimed at improving the performance of WO₃-based photoanodes in PEC water splitting. As mentioned in the text, the inherent flaws of the pristine WO₃ photoanode suggest that it requires appropriate modifications for improved water oxidation performances, which include controlling the scale of WO₃ to the nanoscale range, which can shorten the transmission route of the photogenerated holes, introducing chemical defects into WO₃, which can supply ample active sites and thereby promote charge transfer, and forming a heterojunction with another semiconductor to efficaciously suppress the recombination of the photogenerated charge carriers. In some cases, the employed semiconductor may serve as a water oxidation catalyst as well; loading cocatalysts can increase the photocurrent and simultaneously reduce the onset potential by promoting the charge transfer and consumption in surface reactions. Cooperation with metal particles exhibiting plasmonic effect can also enhance the photo-to-chemical conversion efficiency.

Generally, the PEC water splitting process involves three steps: light absorption, photogenerated charge carrier separation, and reactions on the electrodes involving these carriers. Only when the efficiencies of all the three steps are improved concurrently can a PEC cell display optimal performance. Although several attempts have been made to improve the PEC water splitting performance of WO₃ photoanode, these studies have mainly focused on one step of the PEC water splitting process. Fully taking into account two or three steps of the process can improve the PEC water splitting to a higher level in future studies.

This work was financially supported by the National Natural Science Foundation of China (21808189, 21663027), the Science and Technology Support Project of Gansu Province (1504GKCA027), the Program for Innovative Research Team (NWNULKQN-15-2), the Opening Project of Guangzhou Key Laboratory of Environmental Catalysis and Pollution Control (GKLECPC-12), and the Opening Project of Key Laboratory of Green Catalysis of Sichuan Institutes of High Education (LYJ18205).