*3.2. Z-Schemes*

Z-scheme photocatalysts for overall water-splitting are a combined system involving two photon excitation processes (Figure 5) [50]. The two–photon excitation system was proposed by Bard et al. in 1979 which mimicked natural photosynthesis [51]. A Z-scheme is composed of one H2 evolution photocatalyst and another O2 evolution photocatalyst with electron mediator. Most of Z-scheme construction was demonstrated using Pt co-catalyst loaded SrTiO3, TaON, CaTa2O2 N and BaTa2O2N for hydrogen evolution and Pt/WO3 for oxygen evolution. Photocatalytic activity depends on pH level, concentration of electron mediator and type of co-catalysts. For example, Hideki Kato showed that pH level affects photocatalytic activity of Z-scheme consisting of Pt/SrTiO3Rh–WO3–FeCl3 system [52]. The pH value was adjusted using sulfuric and perchloric acids between 1.3 and 2.5. It was shown that the best photocatalytic activity was achieved at pH 2.4 and subsequent increasing of pH led to decrease of the activity. The sulfate ions-induced formation of [Fe(H2O)5(SO4)]+ species around pH 2.4. Under 48 h of solar illumination, the Z-scheme generated both 890 and 450 μmol H2 and O2, respectively.

Yugo Miseki reported a Z-scheme system with an oxygen evolution photocatalyst of PtOx/H–Cs–WO3 [53]. The Z-scheme water-splitting efficiency with PtOx/H–Cs–WO3 was 3-fold higher than that of using PtOx/WO3. Adding Cs+ ions to the PtOx/WO3 significantly improved the oxygen evolution rate. IO3− ion was used as an electron acceptor in this work. The apparent quantum yield at 420 nm was 20% which is the best index among photocatalysts using the IO3− redox.

Another Z-scheme, containing g-C3N4–WO3 photocatalysts, demonstrated enhanced H2 evolution [54]. The high photocatalytic activity is most likely due to direct electron transfer from WO3 to g-C3N4 in the Z-scheme.

One Z-scheme system consisting of graphitic nitrite g-C3N4 and WO3 nanocomposites modified with co-catalyst Ni(OH)x showed the highest hydrogen production rate of 576 μmol/(g.h). Photogenerated electrons and holes are efficiently separated by combination of g-C3N4 and Ni(OH)x. The electron spin resonance (ESR) technique used DMPO (5,5-dimethyl-1-pyrroline N-oxide) as a trapping agen<sup>t</sup> of •O2<sup>−</sup> and •OH radicals to register the change of the active oxidizing species in aqueous systems. As a result, the Z-scheme charge separation mechanism explained the high hydrogen production rate [55].

**Figure 5.** Schematic illustration of photocatalytic water-splitting by Z-system. C.B.—conduction band; V.B.—valence band; Eg—bandgap. Reproduced from [50], with permission from American Chemical Society, 2010.

Recently, construction of Z-scheme using ZnO–WO3−<sup>x</sup> nanorods was successfully synthesized by the solvothermal method [56]. A high photocurrent value of 3.38 mA/cm<sup>2</sup> at 1.23 V vs. RHE, which is 3.02-fold higher than pure ZnO, was obtained by an effective Z-scheme charges-transfer process. Red shift of optical absorption and better electrochemical performances were achieved by decoration of ZnO nanorods with WO3−<sup>x</sup> nanoparticles.

Sayama et al. reported [57] a stoichiometric production of H2 and O2 using a mixture of Pt-WO3 and PtSrTiO3 (Cr–Ta-doped) in NaI media. The Pt-loaded SrTiO3 (Cr–Ta-doped) produced H2 of 0.8 μmol h−<sup>1</sup> from an aqueous NaI solution while the Pt-loaded WO3 produced O2 at an initial rate of 84 μmol h−<sup>1</sup> in an aqueous NaIO3 solution under visible light (l > 420 nm) separately. The H2 evolution rate from the mixed photocatalyst system (1.8 μmol <sup>h</sup>−1) was higher than that from Pt–SrTiO3 (Cr–Ta-doped) alone (0.8 μmol <sup>h</sup>−1), indicating that addition of the Pt–WO3 effectively reduced the IO3− ion to I<sup>−</sup>.

Even though the band position of WO3 is suitable for O2 evolution, doping WO3 with a metal can shift the energy level. Wang [58] studied electronic properties of WO3 using density functional theory (DFT) calculations with a hybrid calculation. Replacing W by Mo and Cr in the lattice can modify the bandgap of WO3 and improve absorption of visible light. The effect of replacing O atoms by S anions was simulated by substitution along the Z direction in the unit cell. The DFT results predicted that there is a decrease in energy gap (2.21 eV) as well as a conduction band shift up, which is beneficial for HER. The authors also tested the effect of doping WO3 with Ti, Zr and Gf metals, resulting in a predicted upward shift of the conduction band like the case with S anions.

Mg-doped WO3 photocatalysts [59] have been studied experimentally. The conduction band edge for p-type Mg-doped WO3 was −2.7 eV vs. Saturated calomel electrode (SCE) at pH 12, which is more negative than the reduction potential of H2. Hydrogen generation of 3 μmol/gh was achieved by doping WO3 with Mg (5–10 wt%). Doping has also been done using other metals, including Mo [59].

#### **4. Heterostructured WO3 Nanocomposites for Photoelectrochemical Cell Systems**

Photocatalytic activity of WO3 depends on the crystal structure, morphology and surface areas. High surface areas of WO3 usually increase the photo activity via providing more reaction sites. The certain morphology increases electron mobility, thus demonstrates better photocatalytic activity. For example, one dimensional WO3 demonstrates relatively high photoactivity relative to nanoparticles. In the two dimensional WO3 nanomaterials, it is very important to have optimum grain size which lead to high photoactivity. The crystal structure is critical for the photoactivity of WO3. Furthermore, monoclinic structure of WO3 offers different photocatalytic activity relative to other crystalline structure including tetragonal, etc.

Anodization [60,61], solvothermal [56], hydrothermal [62,63], spin coating [64], electrodeposition [65,66] and sol–gel [67] methods were used to fabricate different morphologies and structures.

## *4.1. Crystalline Structure*

Many research efforts have been performed to investigate the effect of crystal structure on the tungsten photocatalytic activity. It was found that the monoclinic crystalline phase demonstrated stronger oxidation activity than other crystal phases such as hexagonal and orthorhombic. The monoclinic phase was found to be the most stable at room temperature [68–72]. Increase of the temperature gradually transformed WO3·0.33H2O from orthorhombic into anhydrous hexagonal and a final stable form monoclinic (Figure 6a). As the temperature transited from 400 to 500 ◦C, the color of the film turned into yellowish color obviously, which is corresponded to a red shift (Figure 6b). The photocurrent density increases until 500 ◦C, then it starts to decrease (Figure 6c). The monoclinic structure of WO3 at 500 ◦C showed the highest photoelectrochemical performance, on the contrary the orthorhombic WO3·0.33H2O exhibited the lowest photocurrent density [68–70,73–75].

**Figure 6.** (**a**) Crystal-unit cells for orthorhombic WO3·0.33H2O, hexagonal WO3 and monoclinic WO3; (**b**) absorption peaks of WO3 films with/without heat treatment; (**c**) linear-sweep voltammetry of WO3 photoanodes at different temperatures under chopped illumination. Reproduced from [68], with permission from American Chemical Society, 2016.

Nayak et al. [76] used combination of a facile precipitation and solvothermal methods to fabricate WO3 nanowires. The precipitation method produced (Figure 7a–d) WO3·H2O nanoplates with an orthorhombic phase, later the solvothermal approach was used to form WO3 nanowires with a monoclinic phase (Figure 7e–h). The photocurrent density obtained from WO3 monoclinic structure was 21-fold higher than that of WO3·H2O orthorhombic phase. This enhancement was ascribed to the presence of different phases between WO3·H2O nanoplates and WO3 nanowires or the high crystallinity of WO3 nanowires, which minimized the barrier of charge transfer at the interfacial charge and enhanced the PEC performance.

**Figure 7.** SEM images. (**a**) Stacked WO3·H2O square nanoplates synthesized at room temperature, WO3 nanowires evolving after (**b**) 3 h, (**c**) 6 h and (**d**) 12 h solvothermal treatment of WO3·H2O nanoplates at 200 ◦C; (**<sup>e</sup>**–**h**) corresponding XRD patterns. Reproduced from [76], with permission from American Chemical Society, 2017.

The effects of crystal phase on the photocatalytic performance has been broadly explored. Park et al. [77] found that the annealing treatment reduced the surface disorder induced by water via improvement of the crystallinity or oxygen deficiencies of WO3, led to enhancement of the PEC performance. Zeng's group explained the formation of peroxo species on the surface of WO3·H2O as it has low degree of crystallinity. As the annealing temperature of WO3 reached 500 ◦C, highly reactive (002) facets were formed to reduce defects, thus to minimize the recombination of electron–hole pairs [75]. The same conclusion was obtained by Su's group [72]. From the above studies, monoclinic WO3 demonstrated higher PEC performance than that of as-prepared hydrated WO3.

Recent investigations demonstrated that with surface engineering certain crystal planes possess preferences on the photoexcited electrons and holes, which lead to either preferential oxidation or preferential reduction reactions [78]. Furthermore, photo–electrochemical efficiency has been improved via exposing the high surface crystalline surface [79,80]. Among the three crystal planes or facets of WO3 which are (200) with 1.43 J/m2, (020) with 1.54 J/m<sup>2</sup> and (002) with (1.56 J/m2) facet of WO3, the crystal facet (002) showed preference for adsorbing the reaction species due to its highest surface energy [81]. Wang et al. confirmed this via DFT calculations [82]. The dangling O atoms of the weakest W–O bond on the (002) crystal plane of the monoclinic WO3, offer plentiful active sites for H2O and organic molecules through the hydrogen bond. Oxidization of water and degradation of organics on the (002) easily occur via consuming photo-excited holes and generating active oxygen species, which reduce the recombination of photogenerated carriers [83–85].

The morphology of the WO3 films can be controlled by synthesis parameters such as synthesis time, temperature and the amount of the capping agen<sup>t</sup> [75]. HRTEM study revealed that annealing WO3·H2O plates transformed along with the (020) crystal face into WO3 plates with preferentially (002) facet. At annealing 500 ◦C, the WO3 showed 1.42 mA cm<sup>−</sup><sup>2</sup> at 1.23 V vs. RHE, which is relatively high current density. This is explained due to reduction of peroxo species on the surface of WO3. The high energy crystal plane of WO3 nanoplate enhanced PEC water-splitting. Zhang et al. [83] compared monoclinic WO3 nanomultilayers which has preferable (002) facet with that of WO3 nanorods and found that WO3 nanomultilayers performed higher photocurrent densities than the WO3 nanorods. These results were explained not by the specific surface area of WO3 nanorods, but the presence of highly reactive (002) facets of WO3, which contributed to the improved PEC water-splitting performance.

In addition, increasing studies have been made investigating the e ffect of the (002) crystal plane of 2D monoclinic WO3 on PEC water-splitting.

To enhance the PEC water-splitting performance of WO3, most of studies have been focused on engineering morphology, crystallinity, heterojunction, oxygen vacancy, doping and co-catalysts for enhancement of photocatalytic hydrogen evolution.

According to the crystalline structure of WO3, it is confirmed that the monoclinic phase of WO3 demonstrated higher OER than the hexagonal or orthorhombic phases since it is the most stable phase at room temperature and presence of highly reactive (002) facets.

## *4.2. Morphologic E*ff*ect*

Various WO3 nanomaterials with di fferent morphologies including nanorods [86], nanoflake [87], nanotubes [88,89], nanoplates [90] and nanoparticles [91,92] were synthesized by various methods to provide active sites for catalysis. It was found that morphology change of WO3 can significantly influence photocatalytic activity.

Ma and other authors [93] obtained nanoplates of WO3 by topological method using Na2WO4 and HBF4 and mentioned that intrinsic crystal lattice of tungsten acid plays important role to obtain morphology of final products. The crystal lattice of H2WO4 has (WO6) octahedra layers with normal direction (010) and each layer is linked to each other via hydrogen bonds. That is why H2WO4 tends to form platelike nanocrystals with (010) direction. Another factor a ffecting morphology is addition of directing agents for nucleation and crystal growth. Interaction of H2WO4 crystal planes and HBF4 can be reason of formation plate morphology of WO3. Meng and others [94] synthesized hierarchical structure using citric acid C6H8O7 and found that (–COOH) functional groups a ffect growth of nanoplate. They concluded [93,94] that uniform platelike morphology is favorable for gas sensing because it has more active sides for absorption of gas molecules. In addition, much work was done on WO3 crystal growth using fluoroboric acid [95], polyethylene glycol (PEG) [96], polyvinyl alcohol (PVA) [87].

Strategy of increasing active sides for suitable absorption of light is a way to enhance photoelectrochemical performances of photocatalysts. For example, Jiao et al. demonstrated di fferent morphology of tungsten trioxide hydrate (3WO3·H2O) films which grown by hydrothermal method using Na2SO4, (NH4)2SO4 and CH3COONH4 as capping agents, respectively. Platelike, wedgelike and sheet like nanostructures can be obtained as shown in Figure 8 [97]. From Figure 8e it can be seen that sheet like nanostructures had the highest photocurrent density (1.15 mA/cm2) under illumination and high photocatalytic activity for photodegradation of methanol. This was in good agreemen<sup>t</sup> with UV-vis absorbance spectroscopy results (Figure 8d). The authors believe that the reason for high current density of sheetlike morphology can be explained by the existence of small pores among sheets. This may be beneficial for accelerating the interface electron kinetics between the sheet and electrolyte due to its large active surface area.

**Figure 8.** SEM images for WO3 hydrate film with sheetlike nanostructures with (**a**) (NH4)2SO4, (**b**) Na2SO4 and (**c**) CH3COONH4 as capping agent; (**d**) absorption spectra of WO3 films; (**e**) current-potential scans of WO3 films measured in darkness and under illumination in a 1-M H2SO4 electrolyte containing 0.1-M methanol. Reproduced from [97], with permission from American Chemical Society, 2011.

Davidne et al. [98] reviewed WO3 nanostructures and studied the effect of morphology on photocatalytic activity for decomposition organic dyes. Nanostructures such as nanoplate, nanoneedle, nanorods and nanowire were obtained by hydrothermal method. It was found that photocatalytic efficiency has good correlation with band gap, crystalline phase, morphology and oxidation state. Nanoneedles with hexagonal structure showed the best photocatalytic efficiency in contrast to others.

Monoclinic nanorods showed higher photocurrent density than (2.09 mA/cm2) nanoplates (1.61 mA/cm2) in the hydrogen evolution reaction (HER) [96]. Some results concluded that 1D structures have high optical, electrical, photoconductivity properties and fast charge transportation [99–101]. Vertically oriented nanorods and nanoneedles have remarkable PEC results [65,66,102] due to high interfacial contact area which improves redox contact area and efficient light scattering.

However, some authors believed that 2D nanostructures like nanoplates have higher specific surface area than one-dimensional (1D) materials such as nanorods and nanowires. For example, Su and others demonstrated better photoelectrochemical characteristics and optical properties of WO3 nanoflakes than WO3 nanorods [87]. Hammad et al. [103] fabricated WO3 nanorods (with a diameter 7 nm, length up to 700 nm) and WO3 nanoplates with width 700 nm on fluorine-doped tin oxide (FTO) substrate via hydrothermal treatment. Results of electrochemical spectroscopy showed that nanoplates have better contact with substrate than nanorods which led to high photocurrent density of 400 μA/cm<sup>2</sup> over 350 μA/cm2. Through changing concentration of HCl acid, Zhou et al. [104] synthesized perpendicularly oriented WO3 nanorods and nanoplates at different amount of acid. WO3 nanoplate arrays also showed a superior photocurrent density of 1 mA/cm<sup>2</sup> at 1.6 V vs. RHE than nanorods of 0.8 mA/cm2. A high photocurrent density may be related with a long carrier diffusion length through 2D nanostructures comparing with 1D and efficient charge transportation and separation. Contradictory results between 2D and 1D nanostructures allow them to conclude that comparing different morphologies under different conditions do not give us true information. This is because one morphology can show different results depending on its morphologic parameter such as length, thickness and diameter.

In the aspect of photocatalytic efficiency evaluation in combination with morphology, WO3–BiVO4 nanostructures have been mostly studied. Lee and others [65] fabricated WO3–BiVO4, TiO2–BiVO4, Fe2O3–BiVO4 and SnO2–BiVO4 nanostructures and showed that PEC characteristics of bare WO3 dramatically increased after modification with BiVO4. The SEM and cross-sectional images of WO3 nanorods coated with BiVO4 are presented in Figure 9. They concluded that pairing WO3 with BiVO4 creates very promising photoanodes for water oxidation than others.

**Figure 9.** (**<sup>a</sup>**,**b**) Top and cross-section SEM images of the optimum BiVO4/70◦-WO3 nanorods; (**c**) expanded image of BiVO4/WO3 nanorods, (**d**) enhancement of photocurrent density of BiVO4/WO3 nanorods (70◦) nanorods; (**e**) schematic illustration of BiVO4/WO3 nanorods. Reproduced from [65], with permission from Elsevier, 2016.

Improving PEC characteristics can be achieved by also using core–shell structures. Spatial separation of photogenerated charges between the core and shell is beneficial, however, excited charges stay inside and do not react with electrolyte. Nevertheless, fast transportation of charges to surface can diminish shell thickness [105]. Rao et al. [106] synthesized core–shell nanostructures of WO3–BiVO4 to improve light absorption and charge separation. A photocurrent and an incident photon-to-current conversion efficiency reached 3.1 mA/cm<sup>2</sup> and ~60% at 300−450 nm, respectively at 1.23 V vs. RHE under simulated sunlight.

Enhanced PEC performance was obtained by designing yolk-shell-shaped WO3/BiVO4 heterojunction which produced a photocurrent density of 2.3 mA/cm<sup>2</sup> with the highest value of ~5.0 mA/cm<sup>2</sup> after adding a Fe–Ni co-catalyst at a bias of 1.23 V vs. RHE under AM 1.5 illumination (100 mW/cm2) [107]. These noticeable photocurrent density results demonstrated that core–shell structures may be potentially viable for photocatalytic applications.

Nanostructures with nanoporosity have shown a better PEC activity due to their large specific surface areas, relatively higher light absorption rate and excellent charge collection efficiency [108–110]. A high surface area of porous nanostructures makes them promising electrode materials for electrochemical surface reactions [111–113]. Furthermore, the nanoporous structure creates the depletion layer and reduced diffusion distance to the photoelectrodes/electrolyte interface, which diminish recombination of electrons and holes [114–116]. Song et al. [117] used versatile foaming-assisted electrospinning method to produce mesoporous WO3 nanobelts which enhanced the PEC water-splitting performance compared with the as-prepared WO3 nanofiber and WO3 nanobelt samples. Shin et al. [114] used a laser ablation method to produce tree-like nanoporous WO3 photoanode

for a photoelectrochemical water–oxidation performance. Both SEM and TEM image in Figure 10, show 1D treelike morphology with a thickness ~3.2 μm and many clusters with nanoporous with average size of ~60 nm. The photocurrent density of treelike porous structures was 9-fold higher (1.8 mA/cm<sup>2</sup> at 1.23 V vs. RHE)) than dense WO3. A quantum efficiency (QE) or incident photon-to-electron conversion efficiency (IPCE) was 70% at 350–400 nm.

**Figure 10.** (**a**) SEM image of the WO3 photoanode; (**b**) TEM image of the nanoporous WO3 clusters, the inset: a high resolution TEM image; (**<sup>c</sup>**,**d**) schematic illustrations of charge transport/transfer processes. Reproduced from [114], with permission from Royal Society of Chemistry, 2015.

Fujimoto et al. [115] synthesized porous BiVO4 using the auto combustion method. Adding oxidizing agen<sup>t</sup> NH4NO3 and subsequent decomposition of organic additive after heating allowed to create porous film with small crystalline BiVO4 nanoparticles during the synthesis. The optimized WO3/BiVO4 film produced a maximum IPCE value of 64% at 440 nm with photocurrent density of 3.43 mA cm<sup>2</sup> at 1.23 V vs. RHE (under one sun illumination).

Finally, it is obvious that different morphologies as a factor of synthesis method produce different PEC results and play an important role in configuration of water-splitting devices. However, it is hard to conclude that one morphology is more beneficial than others. In fact, other factors such as substrate on which the structure is grown, electrolyte, capping agents and, etc. may be cause of change in photocatalytic activity.

#### *4.3. Binary Structures of Hierarchical Architectures Based on WO3 Semiconductors*

As mentioned above, single photocatalysts cannot satisfy all the requirements needed for water-splitting PEC systems. Therefore, scientists focus most of their attention on creating different kinds of heterostructure architectures from different various materials such as metal oxide/metal oxides [47,118–120], metal oxide/metals [92,121–123] and metal oxide/inorganic compounds [124] to create efficient systems for various spheres. Mixing several materials is a method commonly used

to improve separation of charge carriers, photoelectrode stability, absorption of visible light, suitable carrier diffusion length and effective surface charge transfer.

#### 4.3.1. Metal Oxide/Metal Oxide Binary Heterostructures

The sensitivity of TiO2 can also be obtained by modification the surface using semiconductor photocatalyst with a smaller bandgap as WO3 [125–127]. A WO3/TiO2 photocatalytic system was published in 1998 [128]. The photocatalytic activity of WO3 coated with TiO2 was tested for the oxidation of water using iron (III) acceptor. It was also found that iron (III) ion is preferred more than iron (II) ion as an electron acceptor for oxygen evolution.

Amorphous α-TiO2 can be used to passivate the surface of WO3 based nanostructured photoanodes. For example, Yang and et al. [129] demonstrated high photocurrent with 1.4 mA/cm<sup>2</sup> at 0.8 V in 0.1-M Na2SO4 electrolyte using WO3 nanoflakes coated amorphous α-TiO2 films. Passivation of WO3 by α-TiO2 was realized through the O2−W6<sup>+</sup> bonding at contact surface between WO3 and α-TiO2. Hence, passivation of surface allows to decrease recombination and improve PEC oxidation.

Hierarchical WO3/TiO2 composites for hydrogen evolution was fabricated by Momeni [130] using the anodization method. TiO2 nanotubes with a diameter ranged 80–110 nm were modified by WO3. Controlling the concentration of the Na2WO4 solution allowed them to achieve the highest amount of H2, with 2.14 mL/cm<sup>2</sup> under 120 min of solar illumination, which is approximately 3.02-fold higher than bare samples with TiO2 nanotubes (0.71 mL/cm2). It also showed increase of photocurrent value from 0.81 to 1.61 mA/cm<sup>2</sup> after modification proved the effectiveness of the coupled WO3/TiO2 system. The anodization method was also successfully used to prepare the hybrid WO3/TiO2 nanotube photoelectrodes [131] which showed better photo conversion efficiency, STH efficiency and H2 generation.

Many other studies highlighted that coupled WO3–TiO2 systems have better characteristics. For example, the highest photocatalytic activity of nanocomposites particles for degradation of Rhodamine B [60,61], decomposition of 1,4-dichlorobenzene (DCB) aqueous solution [132] and azo dyes [133], for effective catalytic oxidation cyclopentene to glutaraldehyde [133] were obtained.

The effectiveness of photocatalysts was also attained by engineering morphology and specific surface area of the material since electron-hole transfer occurs on the surface. According to some studies, although electron–hole pairs can be generated in volume, they can annihilate before they reach the surface due to the low diffusion length.

Most of authors have shown that dual heterostructures of WO3 and BiVO4 are effective for driving water oxidation reactions [64,134,135]. BiVO4 is an n-type semiconductor-like WO3 which has a bandgap around 2.4 eV [26,136]. The theoretical solar-to-hydrogen efficiency using this material (9.8%) is more than that of WO3 (4.8%)-based systems. Despite the fact that BiVO4 is a direct semiconductor unlike WO3, it has a poor charge transport and a short hole-diffusion length (~60 nm). Coupling WO3 with BiVO4 decreases recombination of photogenerated charge carriers and improves the efficiency of overall water-splitting systems. A dynamic of photogenerated carriers and effective charge separation of WO3–BiVO4 heterojunctions was explained by Grigioni using femtosecond transient absorption spectroscopy [137]. They determined the position of the WO3 conduction bands (+0.41 V vs. RHE) and BiVO4 (+0.02 V vs. RHE) by testing the photocatalytic reduction of thionine. The charge separation mechanism of BiVO4–WO3 system is shown in Figure 11a. When comparing the flat band position of BiVO4 in the WO3–BiVO4 composite, a shift of 170 mV is observed. This shift was explained by electron equilibrium between the two materials due to a Fermi level shift. Photoelectrons transfer from BiVO4 to WO3 while holes localize in BiVO4, so it is possible to separate photogenerated charge carriers spatially (Figure 11b).

**Figure 11.** (**a**) Diagram of the band edge positions of pure WO3 and BiVO4 and (**b**) for a WO3–BiVO4 composite under solar irradiation. Reproduced from [137], with permission from American Chemical Society, 2015.

The morphology of WO3 and the decoration method with BiVO4 are also very important issues. Chae et al. synthesized mesoporous WO3 films followed by a coating of BiVO4 to research particle sizes and shapes, as well as the effects of the photoanode thickness. Large nanoplates showed a high injection efficiency while nanospheres enhanced the charge-separation efficiency [138]. Pihosh et al. synthesized WO3–BiVO4 vertically oriented nanorods by combining the glancing-angle deposition and normal physical sputtering techniques [66]. The photocurrent density achieved 3.1 mA/cm<sup>2</sup> at 1.23 V RHE under illumination of one sun. A nanopillar morphology of WO3–BiVO4 photoanodes fabricated by electrostatic spraying method also produced a photocurrent up to 3.2 mA/cm<sup>2</sup> [139]. An enhanced photocurrent density of 4.55 mA/cm<sup>2</sup> was achieved by using a WO3–BiVO4 photoanode [65]. Deposition of BiVO4 nanodots on WO3 nanorods had an increased photon to hydrogen efficiency of 80% at 1.23 V vs. RHE, which is higher than the theoretical efficiency for bare BiVO4. Rao et al. fabricated WO3–BiVO4 core shell nanowires and showed that the photoanodes demonstrated a η*arb* × η*sep* up to 53%. A combination of BiVO4 with more conductive WO3 leads to effective charge carrier separation and the photocurrent achieved 3.1 mA/cm<sup>2</sup> at 1.23 V vs. RHE.

Iron is an abundant and important metal in the earth's crust, so its use is considered economically viable. Oxidation of iron can lead to formation of the known hematite phase α-Fe2O3 which has semiconductor properties. It has good stability in most electrolytes pH > 3 and has a narrow bandgap (~2.2 eV) which can absorb 40% of the solar spectrum. Although hematite electrodes are well studied for PEC system, photoconversion efficiency is still lower than the theoretical value due to low hole mobility (~2–4 nm). Moreover, poor electrical conductivity, high recombination rate of electron-hole pairs [140] and the slow kinetics of oxygen evolution [141] limits its use. Some studies focus on binary heterostructures with WO3–hematite α-Fe2O3 photoanodes [142]. A photocurrent of 1.66 mV/cm<sup>2</sup> was observed at 1.23 V RHE, while the photon to current efficiency was 73.7% at 390 nm. Schematic illustration of WO3-Fe2O3 composite nanosheets and bandgaps are shown in Figure 12a. The optical absorption measured by a UV-vis diffuse reflectance spectroscopy was found to be improved for the composite WO3 and Fe2O3 material (Figure 12b). Luo published enhanced electrochemical characteristics of a WO3@Fe2O3 photoelectrode compared to bare WO3 and Fe2O3 [143]. Effective photoelectrochemical splitting of seawater with Fe2O3/WO3 nanorods was achieved by Li et al. [144]. Although optical absorption is promising, the photocurrent of Fe2O3/WO3-based photoanodes is still low.

**Figure 12.** (**a**) WO3–α–Fe2O3 nanocomposites and its band diagram for PEC water-splitting (**b**) UV-vis absorption spectra of WO3@Fe2O3. Reproduced from ref [142], with permission from John Wiley and Sons, 2016.

#### 4.3.2. Metaloxide/Inorganic Compounds Heterostructures

The heterostructures formed from WO3 and sulfur components have narrow bandgaps. For example, antimony sulfide, Sb2S3, (1.7–1.9 eV), bismuth sulfite, Bi2S3, (~1.3 eV) and tungsten disulfide, WS2, (~1.3 eV) [145] are very effective. For example, Zhang [146] synthesized WO3/Sb2S3 heterostructures via a simple hydrothermal method to improve PEC performances. Tungsten trioxide nanorods and nanoplates were synthesized by controlling the concentration of acid and tungsten precursor along with subsequent growth of Sb2S3 nanoparticles. It was demonstrated, that WO3/Sb2S3 heterostructures have better electrochemical and optical characteristics than pristine WO3.

A high photocurrent of 5.95 mV/cm<sup>2</sup> was achieved using a three-dimensional WO3/Bi2S3 heterojunction [147]. Bi2S3 is also a n-type semiconductor with bandgap 1.3 eV and has more negative conduction band edge than WO3. A WO3/Bi2S3 heterojunction was fabricated by combining of hydrothermal method, SILAR (successive ionic layer absorption and reaction) process and CBD (chemical bath deposition). Relatively high light absorption, small electron transfer impedance and high charge carrier were proved by UV-vis, EIS (Electrochemical Impedance Spectroscopy) and Mott-Shottky methods.

Despite the fact that the WO3 bandgap energy is not suitable for hydrogen evolution, it is still useful for solving problems such as high electron-hole recombination rates and poor electrical conductivity of some photocathodes [145]. A study of WO3@WS2 core–shell nanostructures fabricated by plasma assisted sublimation was published by Kumar et al. [145]. The highest achieved cathodic photocurrent was 16.2 mA/cm<sup>2</sup> for WS2 at 0.3 V vs. RHE. Sulfurization of the WO3 surface forms a WS2 layer with a rich defect structure, resulting in a high catalytic activity.

#### 4.3.3. Metal Oxide/Plasmon Particle Systems

The plasmonic effect induced by noble metal particles plays an important role in decorating photoelectrodes. Photoactivity of photocatalysts can be increased by enhancing light scattering and SPR [148,149]. Moreover, noble metals play a role as a co-catalyst for OER due to good electrical contact between the metal and semiconductor [150,151]. Altering the surface properties of WO3 photoanodes with plasmonic nanoparticles Au and Ag has shown enhanced visible light absorption and high photocurrent density [152]. Hu showed a high faradic efficiency of 94% for WO3@Au composites [149]. Enhanced photocurrent density and morphology of heterostructure is presented in Figure 13a,b. Modified WO3 by plasmonic Ag and Pt nanoparticles showed enhanced photocurrent of 1.13 mA/cm<sup>2</sup> at 1.23 V vs. RHE under AM 1.5G illumination in a 0.2 M Na2SO4 solution, which is nearly 3.32 times that of bare WO3 [153]. The photocurrent density for binary systems are represented in Table 1.

**Figure 13.** (**a**) Current -potential curves of WO3@Au composite with HAuCl4 concentrations of 8, 24 and 240 umol in 0.1-M Na2SO4 electrolyte and (**b**) SEM images of WO3@Au composites. Reproduced from ref [149] with permission from Springer Nature, 2016.



**Table 1.** *Cont*.

#### *4.4. Ternary Systems for E*ffi*cient Water Decomposition*

A typical heterojunction between two dissimilar semiconductors comes to equilibrium without any external electric field. The result shown above is a potential difference that appears at the interface as an internal electric field. This field accelerates charge carrier drift and decreases the number of electron–hole recombination, improving the semiconductor's photocatalytic activity [155]. The exploitation of multilayer structures in photocatalysis is considered more beneficial. An illustration of the advantages of a cascade transition of charge carriers is well explained in numerous works [156] and [157]. The authors of [156] work investigated the photocatalytic properties of a composite of CdS, TiO2 and tungsten trioxide. Since, TiO2 has the conduction band edge which is between conduction band edges of CdS and WO3, in such ternary composites electronic transitions are cascading. After generation of excitons in CdS, electrons easily migrate to TiO2 and WO3 of potential difference as shown in Figure 14.

**Figure 14.** Schematic representation of a comparison of electronic transition processes in binary and ternary hybrid. Reproduced from [156], with permission from American Chemical Society, 2011.

In addition to ternary composites made only of semiconductor materials, there are also ternary structures based on semiconductors and metal complexes. An example of such ternary photocatalysts can be photonic heterostructure of CdS–Au–WO3. Cui et al. reported, that Au nanoparticles deposited between WO3 and CdS leads to form heterostructure which had photocatalytic properties superior to similar two-phase systems. It was thought that such amelioration caused by a synergistic integration of photon absorption with effective electron transfer in the heterostructure [62]. The use of doping elements—or modification by particles such as CdS quantum dots [158] in usual two-component heterojunctions—is also referred to as a ternary structure. Doping elements such as Yb [159] and Mo have been shown to suppress charge carrier recombination during photocatalysis [160], improving the efficiency of reactions. In ref. [161], high photocurrent density was achieved by doping active materials

to make bilayers WO3/MoBiVO4 (BiV0.95Mo0.05O4). The photocurrent of the Mo-doped content increased by a factor of 3 and 1.5 relative to pristine photoanodes based on WO3 and BiVO4/WO3, respectively. Another promising way to create performable photocatalysts is a combination of a catalyst with heterostructures. Some examples of such formations are materials which obtained by the deposition of catalytic coatings NiOOH [162] and FeOOH [163] on the surface of two-phase WO3 structures. The deposited materials suppress both the degradation of the photoactive material and the Faradic losses [164], accelerating the photoelectrochemical reaction processes. Shouli Bai et al. also combined catalyst layers with a heterojunction by depositing NiFe bimetallic complexes onto a WO3/Fe2O3 surface [165]. Their strategy increased the photocurrent density of the ternary photoanode system to 3.0 mA cm<sup>−</sup>2, which, according to Shouli Bai, is 5 and 7-fold higher relative to that of pristine WO3 and α-Fe2O3 structures, respectively. The role of the catalyst here is to improve the absorption of holes accumulated on the electrode surface. In another study, the deposition of CoPd bimetallic nanoparticles onto the surface of WO3/α-Fe2O3 photoanodes causes a cathode shift of the initial potential, increasing the photocurrent density from 0.15 to 0.5 mA/cm<sup>2</sup> during water oxidation at 1.23 V relative to RHE when illuminated with AM 1.5 G [166]. Substitution of iron oxide with cadmium sulfide in heterostructures based on WO3 also makes it possible to sensitively increase the photoresponse of electrodes during water decomposition. In [167] work preparation of such ternary compositions conducted via three simple hydrothermal, impregnation and photo-assisted deposition steps. Thus, authors obtained rodlike structures with a performance of photocurrent at the level of 5.85 mA/cm<sup>2</sup> at 1.23 V (vs. RHE). Sun with co-workers explain this phenomenon by creation a larger built-in potential at interface WO3/CdS formed via impregnating appropriate CdS onto surface of WO3. This drives transport of electrons from CdS to WO3 with improvement of exciton separation. In this case, not the entire charge is torn well enough. Part of the charge recombines due to the weak involvement of holes in the valence band of cadmium sulfide in the process of water oxidation. Decoration surface of WO3/CdS heterojunctions with Co-Pi co-catalyst advances the transfer kinetics of charge advanced which is positive to suppression of charge recombination. In this case, the mechanism of improving charge transfer to the sites of redox half-reactions is also achieved by adding phosphate anions to the electrolyte. In fact, the use of various electrolytes, such as glycerol-water mixture [168], in its effect on the parameters of the transferred charges between photocatalytic coatings and a split liquid. Varieties of compositions of working electrolytes and a list of structural heterojunctions, as well as geometric schemes that receive the influence and influence of all this on the photoresponse of PEC systems are shown in Table 2. It is known that tungsten trioxide is widely used as the primary semiconductor material in three transient systems in photocatalysis. Therefore, numerous of recent works dedicated to photoinduced dye degradation processes [169] and the direct splitting of water [170,171] were carried out with exploiting WO3. Liu formed an effective three-component photoanode based on tungsten trioxide nanosheets synthesized by hydrothermal method and decorated with ZnxBi2S3<sup>+</sup>x quantum dots via layer-by-layer adsorption [172]. In obtained core/shell structure of ZnxBi2S3<sup>+</sup>x/WO3, surface ZnxBi2S3<sup>+</sup>x served as a protective layer for tungsten trioxide. Comparable photocatalytic studies conducted under visible light irradiation in 0.1-M Na2S and 0.1-M Na2SO3 aqueous solution at pH ~9 showed that ZnxBi2S3<sup>+</sup>x/WO3 composite has higher photocatalytic activity then Bi2S3/WO3.structure. The photocurrent density was 7 mA/cm<sup>2</sup> at a bias voltage of −0.1 V. This attributed to the involvement of ZnS nanostructures with high photocatalytic properties [173]. Designing semiconducting heterocomposites via a surface–interface engineering approach showed high effectiveness for enhancing exciton separation/transportability and photoelectrochemical features. The photoactive layer of 2D BiVO4-layer/WO3 array modified with cobalt phosphate studied for oxygen evolution and showed 1.8 mA/cm<sup>2</sup> at 1.23 V vs. RHE in a phosphate buffer under an AM1.5G sun. The result is 5 and 12-fold higher than that of bare WO3 and BiVO4 photoanodes, respectively [174]. Morphology of WO3 nanostructures affects to charge separation ability in the active layer and to charge collection efficiency in the WO3/BiVO4 heterojunction. The low-dimensional nanosphere WO3 layer showed higher photocatalytic efficiency than the platelike or rodlike ones [71].

There are three different component systems with different morphology in which the hierarchy of the band structure observed for effective water-splitting. The main types of morphology include multi-heterojunction-based photocatalysts composed of WO3 nanorods, Pt nanoparticles and TiO2 nanoparticles [175], hierarchical heterostructures with core and double shells [176], rectangular AgIn(WO4)2 nanotubes which showed excellent photocatalytic properties for decomposing water to evolve H2 [177] and linked porous structures such as WO3/porous–BiVO4/FeOOH [178]. Special attention should be paid to the three-component system made on spiral WO3 nanostructures decorated with doped Mo and BiVO4 nanoparticles [179]. Based on the assertions that 1D structures charge carriers go straight to the terminals or half-reaction centers [180], Xinjian Shi et al. used a spring morphology with an increased surface area while retaining the properties of 1D structures. As a result of the successive deposition of these structures onto conductive glass, a spiral heterocomposite WO3/(W, Mo)–BiVO4 with a length of 5.5 μm was obtained and the photocurrent density exceeded 3.9 mA/cm2. The process of making triple transitions is possible using various techniques, such as electrochemical reduction-oxidation technology [181], a hybrid synthetic method, including electrodeposition and hydrothermal treatment [182], a solvothermal-calcination process [183], an electrostatic-driven self-assembly correlated with ion-exchange process [162] and a polymer complex method [184]. Jun Lv et al. obtained photoactive LiCr(WO4)2. After heat treatment at 700 °C for 5 h, crystal structures of tungstate were obtained, consisting of alternating layers of WO6 and CrO6 octahedra lying on the YZ plane. The WO6 octahedra are connected by separating edges, leading to the appearance of unrelated zigzag rows along the *Y*-axis. On the other hand, CrO6 octahedra not related to each other. Photocatalytic tests of LiCr(WO4)2 showed that the release of H2 proceeds upon irradiation with visible light up to 540 nm [184]. Reaching the rest of the visible spectrum remains the goal. Ji Hyun Baek et al. developed a photoanode based on BiVO4/WO3/SnO2 using a simple method of applying layers on conductive glass to obtain a thin double-heterojunction coating on the order of 320 nm. A characteristic feature of BiVO4/WO3/SnO2 is the large bandwidth of visible light, which allowed the researchers to combine the photoanode with a perovskite solar cell into a tandem PEC system. This allowed the reaction to split water without applying a bias voltage [26]. In general, the development of PEC cells using independent absorbents of incident light is a promising direction, so the next section will deal tandem hydrogen production systems.

#### *4.5. WO3-Based Tandem PEC Cells*

Two strategies can produce tandem cells for photoinduced water-splitting. One strategy used is to increase the capture of photons is a tandem system of a photoanode and photocathode with n- and p-conductivities of active layers, respectively. The splitting of water under light irradiation in this type of PEC cell occurs as a result water oxidation by photogenerated holes on the photoanode surface. Meanwhile, reduction to molecular hydrogen is initiated by electrons on the photocathode surface. At the same time, negative charge carriers generated in the photoanode are directed under the action of the field, to positively charged carriers in the photocathode material, where they recombine. Robert Coridan et al. investigated the photocatalytic properties of Si/WO3 heterojunctions and Si/ITO/WO3 arrays in a core-shell manner [185]. The operation of the tandem structure depends not only on the bandgap of the semiconductors used but also on the alignment of the strip edge and the state of the electrical connection between photo absorbents. When a mismatch of conductivity levels and valence levels of semiconductors included in tandem circuits occurs, low efficiencies of PEC hydrogen evolution cells [186]. A similar effect was observed in the work of Heli Wang et al. in which they combined n-type tungsten trioxide and hematite nanorod metal oxides with p-GaInP2. It was found that even when photo electrodes are illuminated with a source with a power of 1 <sup>W</sup>/cm2, a photocurrent appears but the density values of which are rather weak. This is due to low electron mobility of in the hematite layers, short hole-diffusion length, and insufficient potential difference between the levels of the conduction band and valence semiconductors, which help reduce charge recombination [187]. Geometrical optimization of the morphology of the active layers of photoelectrodes refers to an increase

the density of short circuit current [188]. Investigating the PEC properties of the tandem structure of WO3/Si, Zhuo Xing et al. concluded that it was necessary to add an intermediate layer between p and n semiconductors to reduce the number of recombinations of photoinduced charges. In [189], metal W served as an intermediate layer, resulting in a WO3/W/Si ternary structure demonstrated an increase in the photogenerated current density by a factor of 10 compared to the WO3/Si structure.

Another possible way for general water-splitting without assistance is to combine photoelectrodes with photovoltaic cells to form a tandem PEC/PV cell. In one study [190], organic molecules were used as sensitizers in a tandem PEC, which is a powerful strategy for designing hydrogen evolution systems since they allow large-scale modification of photoelectrodes by adjusting the dye redox potentials or redox mediators.

In the tandem devices shown in Figure 15, the BiVO4/WO3 photoelectrode absorbs short-wavelength photons and more extended wavelength absorbs by a dye/TiO2 electrode [185–190]. This method offers better concession between device performance, complexity and stability [3]. In addition to scientific methods, when choosing materials, morphology and hierarchy of architecture, engineering aspects related to the spatial and reciprocal arrangemen<sup>t</sup> of the physical elements of tandem structures are also important. For efficient use of incident photons, Pihosh Y. and colleagues produced a PEC-PV tandem system based on WO3 NRs/BiVO4+CoPi photoanode and an AlGaAsP solar cell, which were placed at 45◦ relative to each other using a V-shaped stand [102]. This design allows the passage of reflected light from the photoanode to the surface of the solar cell.

**Figure 15.** Schematic illustration of a PEC tandem system based on BiVO4/WO3 and DSSC. Reproduced from [191], with permission from Springer Nature, 2016.

Thus, the characteristic differences between ternary systems and binary systems are the improvement of photocatalytic properties and corrosion resistance. As described above, tandem structures provide an operating mode for photoelectrochemical processes in a wide range of the radiation spectrum. It increases the number of components of the hydrogen evolution cells, which leads to a complication of the assembly of heterostructures and to a high cost of the obtained layers. Therefore, when choosing which or used components of tertiary composites, pay attention to postprocessing, including thermal.


#### **Table 2.** Photocurrent densities of ternary heterostructures.


#### **Table 2.** *Cont*.

#### **5. Conclusions and Outlook**

The goal of research in the field of photoinduced decomposition of water is to develop high performance photocatalytic systems with high STH efficiency. The transmission of the photocatalytic systems from the field of laboratory research to the large scale production is a key point. The principle of using semiconductor coatings based on tungsten trioxide for PEC cells is justified by the economic aspects associated with the low cost of the material, as well as with its physicochemical properties. Using nanotechnology and nanomaterials is a suitable method for addressing several of the issues listed above. Metal oxide nanoparticles can be obtained by a wide range of physical and chemical methods. They can be classified as top-down and bottom-up methods. Top-down approaches rely on physical processes, such as abrasion or ball milling. Nano powders produced in this method usually exhibit wide distribution sizes and their size, shape and morphology are difficult to control. In addition, possible structural and surface impurities can have a significant effect on surface chemistry and the catalytic properties of nanomaterials. Low-dimensional structures are most advantageous from the

point of view of e ffective absorption of light with the generation of charge carriers, migration of charge carriers to the surfaces of the material, which fit over the exciton lifetime, as well as possessing a fairly significant semiconductor/electrolyte contact area. The mixing of semiconductors (i.e., the formation of composites) is also an accepted strategy for the development of photocatalysts that respond to radiation in the visible range. This strategy is based on a hierarchical architecture for connecting a wide-gap semiconductor with a narrow-gap semiconductor with a more negative level of the conduction band. Thus, the conduction of electrons can be introduced from a narrow-gap semiconductor into a wide band semiconductor, leading to better absorption for the mixed photocatalyst. An additional advantage of using composite semiconductor photocatalysts are that it reduces carrier recombination by facilitating electron transfer crossing interface of particles. In photocatalyst composites, semiconductor particles stay in electronic contact individually. For a successful combination of semiconductors, the certain requirements are needed to be met: the conductivity level of the narrow-gap semiconductor should be more negative than the level of the wide-gap semiconductor; the position of the conductivity level of the wide-gap semiconductor should be more negative than the recovery potential; electron injection should be quick. All of the steps can improve the characteristics of the material, as well as eliminate the influence of its shortcomings on the process of splitting water under the action of light. In any case, a review of the literature in this area indicates a special level of development in the field of photoelectrochemistry for hydrogen evolution using active materials from pure tungsten trioxide or in various compositions with it. However, so far, the complete and qualitative decomposition of water and the generation of hydrogen under the influence of sunlight has a low rate, which indicates insu fficient feasibility of industrial use of existing technologies. Based on the current trend towards the creative and experimental activity of researchers in this direction, the authors of this article express deep confidence in the imminent achievement of quantum e fficiency of PEC systems su fficient for universal use in human life in the near future.

**Author Contributions:** N.N. designed and supervised the whole manuscript. He wrote the whole manuscript together with Y.S., A.M. and N.B.; Y.S. contributed to write Sections 2 and 4.1; A.M. wrote Sections 3, 4.2 and 4.3. N.B. wrote Sections 4.4 and 4.5; all authors wrote introduction, conclusion and outlook together. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partially funded by Nazarbayev University FGRG gran<sup>t</sup> number SEDS2020 016 and the Young Scientist Program of the Committee of Science of the Ministry of Education and Science of the Republic of Kazakhstan, gran<sup>t</sup> number AP08052381.

**Acknowledgments:** The authors would like to express special thanks to Aygul Nuraje from BTHS for proof-reading and editing the manuscript. N.N. greatly acknowledge financial supported from Nazarbayev University. N.B. thanks the Young Scientist Program of the Ministry of Science and Education of the Republic of Kazakhstan for financial support.

**Conflicts of Interest:** The authors declare no potential conflict of interest.
