High-efficiency water oxidation and energy storage utilizing various reversible redox mediators under visible light over surface-modified WO₃

Abstract

Tungsten trioxide (WO₃) powder, treated with various metal salt solutions, was used for the photocatalytic oxidation of water into O₂; the reaction was accompanied by the reduction of various redox oxidants. The photocatalytic activity of WO₃ was remarkably improved by thermal treatment with alkali metal and silver salt aqueous solutions. Cs-treated WO₃ showed the highest activity. The WO₃ particles were covered with a very thin layer of a cesium tungstate species, and the ion-exchangeable sites on the WO₃ surface were formed by Cs-treatment. The activity of Cs-treated WO₃ was further improved by ion exchange of H⁺ and Fe²⁺ ions and was 24 times higher than the activity of WO₃ without treatment. All methods of O₂ evolution using IO₃⁻ and VO₂⁺ on WO₃ were also significantly improved with Cs-treatment, which suppressed the reverse reaction of the oxidation of the various redox reductants. The optimized WO₃ in Fe(ClO₄)₃ aqueous solution had a high quantum yield (31% at 420 nm) and solar-to-chemical energy conversion efficiency (0.38%).

---

Introduction

Water splitting by photocatalysis has been studied as an ideal candidate for solar energy conversion and storage.^1–7 Water reduction and oxidation must be separately induced on a semiconductor particle in conventional one-electron excitation systems. Therefore, a photocatalyst material should have a band structure that can thermodynamically induce water-oxidation and water-reduction reactions (conduction band potential < 0 V vs. RHE, valence band potential > +1.23 V vs. RHE).⁴ Moreover, it should be stable against photo-generated holes because some semiconductor materials can oxidize themselves by such holes in the absence of a reducing agent. Therefore, there are very few reports on direct water splitting by visible light, even though visible-light photocatalysts are necessary for developing highly efficient solar-light-driven water splitting.^8,9 We first reported the stoichiometric decomposition of water into H₂ and O₂ using only visible light (λ > 420 nm) via a two-photon excitation system (Z-scheme system) that mimicked natural photosynthesis, which is a similar two-photon energy conversion system.¹⁰

nOx + 2H₂O → nRed + 4H⁺ + O₂

ΔG (eV) = +1.23 − E_ox/red (−0.059 pH) (1)

nRed + 4H⁺ → nOx + 2H₂

ΔG (eV) = + E_ox/red (+0.059 pH) (2)

2H₂O → 2H₂ + O₂

ΔG (eV) = +1.23 (3)

In these equations, Ox is the redox oxidant, red is the redox reductant, E_ox/red is the redox potential of the redox reagent, ΔG is the Gibbs free energy change, and 0.059 pH is the term added when the redox potential is pH independent. In this system, water is decomposed via the combination of two photocatalytic reactions (eqn (1) and (2)), which proceed more easily thermodynamically compared to direct water splitting. Therefore, this system can potentially use a semiconductor for water-reduction or water-oxidation potential of one side of the reaction, suggesting that the potentials of the conduction and valence bands are not restricted compared to the conventional one-photon excitation system. Moreover, separation of the evolved H₂ and O₂ is possible. Visible-light-responsive systems combined with various photocatalysts and redox reagents for water splitting have been widely reported recently (Fig. 1).^11–20 However, the efficiency of these systems remains low. Therefore, techniques that can improve the quantum efficiency (QE) of the photocatalytic reactions for eqn (1) and (2) are important for high-efficiency water splitting. I₃⁻/I⁻ (+0.55 V vs. RHE),^17,19 Fe³⁺/Fe²⁺ (+0.77 V vs. RHE),^21,22 VO₂⁺/VO²⁺ (+1.00 V vs. RHE),²³ and IO₃⁻/I⁻ (+1.09 V vs. RHE)^{10,12–15,18,24} have been reported as reversible redox ions that utilize eqn (1) and (2). With respect to stability, cost, and redox potential, Fe³⁺/Fe²⁺ is an excellent redox mediator. It was reported that the Fe²⁺ ion was produced in relatively high QE over a WO₃ photocatalyst under visible light using organic compounds as sacrificial electron donors.²⁵ However, the photocatalytic activity of Fe²⁺ production without sacrificial electron donors for eqn (2) was very low (QE < 0.4% at 405 nm).^20,21,26,27 We reported in a short communication that the oxidation of water into O₂ over WO₃ in a FeCl₃ aqueous solution was improved by Cs-treatment;²⁰ however, the efficiency was low (quantum yield (QE) = 19% at 420 nm) and the mechanism was unclear. One of the reasons for low efficiency is that the reverse reactions easily proceed in semiconductor photocatalysts, as shown in Fig. 2.⁷ Ideally, only the redox oxidant (Ox) reduction (shown in Fig. 2 as a solid line) and water oxidation selectively proceed in an O₂-evolution photocatalyst. However, a redox reductant (Red) is accumulated in the reactant solution as the reaction proceeds. As a result, Red oxidation (shown in Fig. 2 as a broken line) preferentially proceeds instead of water oxidation because of the thermodynamic advantage. Consequently, the apparent reaction efficiency decreases significantly. Therefore, it is important to develop techniques that suppress this undesirable back reaction. In this study, we investigated the oxidation and reduction properties of a WO₃ photocatalyst systematically treated with various ions using various redox mediators. We discovered that the undesirable back reactions of the Reds and holes were significantly suppressed by the presence of a thin cesium tungstate layer on the WO₃ surface. Optimized WO₃ in an aqueous Fe(ClO₄)₃ solution showed a very high QE (31% at 420 nm) and solar-to-chemical energy conversion efficiency (0.38%).

Fig. 1 Two-photon excitation (Z-scheme) system combined with various types of photocatalysts and redox reagents.

Fig. 2 Forward and back reactions that proceed in water oxidation and redox oxidant reduction.

Experimental

Preparation

WO₃ (Kojundo Chemical Laboratory Co., Ltd.) powder was used as the starting material. The WO₃ powder was first thermally treated in air at 973 K to improve its crystallinity. Some metal salts such as alkali, alkaline earth, and transition-metal salts were used for surface modification of the WO₃ photocatalyst surface modifications were conducted by an impregnation method. 0.5 g of WO₃ powder was dispersed in 0.5 mL of 4.3–215 mM metal salt solution, and then the slurry was evaporated to dryness followed by thermal treatment at 373–973 K for 10–600 min. Then, these catalysts were stirred in H₂SO₄ or FeSO₄ aqueous solutions for 30 min for ion exchange, if necessary. TiO₂ powder (rutile, Toho Titanium Co., Ltd.) was used as the photocatalyst based on literature.²⁸ BiVO₄ was prepared from Bi₂O₃ and V₂O₅ in a 0.75 M HNO₃ aqueous solution at room temperature according to literature.²⁹

Characterization

Diffuse reflection spectra were obtained using a UV-vis-NIR spectrometer (JASCO Corporation, UbestV-570). Phase purity of the obtained modified WO₃ powder was confirmed by X-ray diffraction (PANalytical, EMPYREAN) and Raman spectroscopy (JASCO Corporation, NRS-1000, excitation at 532 nm). X-ray photoelectron spectra were obtained using an X-ray photoelectron spectrometer (Ulvac-Phi, XPS-1800). WO₃ particles were observed using scanning electron microscopy (SEM, Hitachi, Ltd., S-4800) and transmission electron microscopy (TEM, Hitachi High-Technologies Corporation, H-9000NAR). The particles for TEM samples were sliced into thin sections by focused ion beam method. The adsorption amounts of Fe²⁺ and Fe³⁺ on various WO₃ catalysts were determined by color reactions using phenanthroline and chloride ions, respectively.

Photocatalytic reactions

Photocatalytic reactions were performed in a side-window cell made of Pyrex connected to a gas-closed circulation system. A 300 W Xe illuminator (ILC Technology, Inc., CERMAX-LX300) attached to a cut-off filter (HOYA Corporation, L42) was employed for visible-light irradiation. The photocatalyst powder (0.4 g) was dispersed in the Fe³⁺ aqueous solution (300 mL) using a magnetic stirrer. The initial pH in the Fe³⁺ solution was adjusted to 2.3 for all photocatalytic reactions. An aqueous VO₂⁺ solution (pH 0.7) was prepared by dissolving V₂O₅ in a 1 M H₂SO₄ aqueous solution. An aqueous IO₃⁻ solution (pH 6.5) was prepared by dissolving NaIO₃ in pure water. For the reaction where the IO₃⁻ ion was used as the redox mediator, the Pt cocatalyst was loaded on the WO₃ photocatalyst according to a literature method.¹⁰ NaI, VOSO₄, and FeSO₄ were dissolved in the reactant solutions as Red sources, if necessary. The amount of the evolved O₂ was determined using on-line gas chromatography (Shimadzu Corporation, MS-5A column, TCD, Ar carrier). The apparent QE was measured using monochromatic light through a bandpass filter. The number of incident photons was determined using a Si photodiode proofread by NMIJ (National Metrology Institute of Japan). The solar-to-chemical energy-conversion efficiency (η_sun) was measured using a solar simulator (Sanei Denki Co., adjusted to AM 1.5 and 1 SUN by a spectroradiometer). Solar-energy-conversion efficiency was defined by the following eqn (4).

Solar-energy conversion (%) = [output energy as Fe(II) ions]/[energy of incident solar light] × 100 = [Gibbs free energy change, ΔG₂₉₈⁰] × [rate of O₂ evolution × 2]/[energy of incident solar light] × 100 (4)

Preparation and photoelectrochemical measurement of WO₃ photoelectrodes

A WO₃ electrode (sputtering film on conducting glass purchased from NSG Techno-Research Co. Ltd., thickness: ∼1 μm, aggregation of the particles: ∼50 nm) was calcined at 923 K for 30 min. A Cs₂CO₃ aqueous solution was dropped onto the WO₃ electrode, which was then calcined at 773 K for 10 min. In addition, the WO₃ electrode was soaked in H₂SO₄ or FeSO₄ aqueous solution for 30 min. The photoelectrochemical measurements were performed using a potentiostat (BAS Co.) and a Pyrex glass cell. A Pt wire and Ag/AgCl electrode were used as the counter and reference electrodes, respectively.

Results and discussion

Effect of thermal treatment using various metal aqueous solutions for WO₃ photocatalyst on photocatalytic water oxidation accompanied by Fe³⁺ reduction

Fig. 3 shows the reaction rates for water oxidation into O₂ using WO₃ photocatalysts treated with various metal aqueous solutions. Native WO₃ showed little activity for water oxidation, as reported previously.^20,21,26,27 The activity of WO₃ was remarkably improved by thermal treatment at 773 K using MCl (M = Na, K, Rb, and Cs) and AgNO₃ aqueous solution. WO₃ treated with cesium aqueous solution (Cs–WO₃) showed the best activity. The X-ray diffraction patterns, Raman spectra, and diffuse reflection spectra (DRS) of the WO₃ photocatalysts barely changed with the Cs–WO₃ (Fig. S1(I)–(III)),† suggesting that the bulk of WO₃ was unchanged. On the other hand, Cs⁺ was detected on Cs–WO₃ by XPS measurement even after washing thoroughly with pure water (Table 1). Cs₂CO₃ (or CsOH) is usually readily soluble in water. Therefore, it was determined that a water-insoluble Cs⁺ species formed on the WO₃ surface. The elemental ratio of Cs/W estimated by X-ray photoelectron spectroscopy (XPS) on the Cs–WO₃ surface after calcination at 773 K (0.22–0.25) had almost the same value as that found on Cs–WO₃ without calcination (0.19–0.21). The radius of Cs⁺ is too large for it to penetrate the bulk. Thus, most of the Cs⁺ is likely present on the WO₃ surface, which is within the detection depth of XPS (ca. 2 nm). Fig. 4 shows the SEM images of WO₃, with and without Cs-modification. A nanostep structure with plane terraces formed on the WO₃ particles with Cs-modification; however, the native WO₃ particles without any treatment had characterless and roundish surfaces, suggesting that the surface of WO₃ was reconstructed by the Cs-modification. It has been reported that a similar surface nanostep structure forms on NaTaO₃ particles by doping with La or alkaline earth metals, and this surface structure promotes the charge separation of photo-generated electrons and holes, resulting in increased activity for water splitting compared to the activity on nondoped NaTaO₃.^30,31 Therefore, there is a possibility that the high activity of Cs–WO₃ might result because of this surface morphology change. Fig. 5 shows the TEM images of WO₃ and Cs–WO₃. The entire Cs–WO₃ particle was covered by thin compound layers (ca. 2 nm) with different lattice spacing compared to the WO₃ bulk; however such a thin film was not observed on the WO₃ surface. The lattice spacing of the thin surface layers is smaller than that of the WO₃ bulk, despite the fact that the ion radius of Cs⁺ (1.67 Å) is much large than that of W⁶⁺ (0.60 Å). Therefore, it was determined that the crystal structure of the surface layers became different from the ReO₃ structure (WO₃ bulk) by incorporating Cs⁺. From the above results, it was surmised that the high activation of the WO₃ photocatalyst treated with Cs⁺ was derived from the reconstructed surface structure and/or the presence of a water-insoluble thin film of cesium tungstate on the WO₃ surface.

Fig. 3 Water-oxidation reaction rates of WO₃ photocatalysts treated with various metal aqueous solutions. Impregnation reagent amount: 1 mol% for WO₃. Impregnation condition: 773 K for 30 min. Catalyst: 0.4 g, reactant solution: 300 mL of 2 mM Fe₂(SO₄)₃, light source: 300 W Xe-arc lamp (λ > 420 nm), reaction cell: side-irradiation cell. Initial pH of reactant solution was always adjusted to be 2.3 using sulfuric acid. WO₃ was calcined at 973 K for 2 h in air as a pretreatment.

Table 1 Effect of ion-exchange treatment of Cs–WO₃ and WO₃ on photocatalytic activities for water oxidation accompanied with Fe³⁺ reduction^d

Catalyst	Ion-exchange treatment solution^b	Surface atom ratio (W : Cs: Fe)	O2-evolution rate/μmol h^−1
a Cs amount: 2.2 mol% for WO3.b Ion exchange was conducted for 30 min.c Activity of calcined catalyst at 673 K for 30 min after H^+ exchange.d Catalyst: 0.4 g, reactant solution: 300 mL of 2 mM Fe2(SO4)3, light source: 300 W Xe-lamp (L42).
Cs–WO3^a	Pure water	1 : 0.25 : 0	118
Cs–WO3^a	H2SO4	1 : 0.16 : 0	156 (27)^c
Cs–WO3^a	Fe2(SO4)3	1 : 0.12 : 0.14	116
Cs–WO3^a	FeSO4	1 : 0.14 : 0.02	200
WO3	Pure water	1 : 0 : 0	8.1
WO3	H2SO4	1 : 0 : 0	7.5
WO3	Fe2(SO4)3	1 : 0 : 0.10	8.7
WO3	FeSO4	1 : 0 : 0.02	8.4

---

Fig. 4 Scanning electron microscopy images of (a) WO₃ and (b) Cs–WO₃ photocatalysts. Cs amount: 1 mol% for WO₃.

Fig. 5 Transmission electron microscopy images of (a and b) WO₃ and (c–e) Cs–WO₃ photocatalysts. Cs amount: 1 mol% for WO₃.

Fig. 6 shows the time course of O₂ evolution over WO₃ photocatalysts, with and without Cs-treatment. Cs–WO₃ showed the highest activity when an aqueous CsCl solution (2.2 mol% for WO₃) was impregnated at 773 K for 30 min. The total amount of O₂ gas reached ca. 300 μmol over Cs–WO₃, thus agreeing with the stoichiometric amount expected from Fe³⁺ (1200 μmol) in the solution. The stoichiometric amount of Fe²⁺ was detected in the solution after the photoreaction; Fe³⁺ was not detected after the photoreaction. The activity of the second run over Cs–WO₃ (c) was much higher than that of the first run (b), though an improvement in the second run was not observed for WO₃ with no Cs-treatment. Therefore, it was determined that the further improvement was derived from the changing surface condition of the Cs–WO₃ photocatalyst during the photoreaction.

Fig. 6 Photocatalytic O₂ evolution over (a) WO₃ without Cs-treatment, (b) Cs–WO₃ (1^st run), and (c) Cs–WO₃ (2^nd run) under visible-light irradiation. Cs amount: 2.2 mol% for WO₃. Broken line shows upper limit of O₂ evolution expected from the amount of Fe³⁺ (1200 μmol) added to solutions. Second run reaction was performed by exchanging reactant solution of 1^st run with fresh Fe₂(SO₄)₃ aqueous solution. Catalyst: 0.4 g, reactant solution: 300 mL of 2 mM Fe₂(SO₄)₃, light source: 300 W Xe-arc lamp (λ > 420 nm), reaction cell: side-irradiation cell. Initial pH of reactant solution was always adjusted to 2.3 using sulfuric acid.

Analyses of detail surface conditions of Cs–WO₃ using XPS and DRS measurements

Some alkali complex oxides possessing layered or tunnel structures have ion-exchange abilities. The ratio of Cs/W based on the XPS decreased when the stirring treatment with 1 M H₂SO₄ solution was performed for Cs–WO₃. Moreover, the Na 1s signal was detected on the Cs–WO₃ surface after performing stirring treatment with 1 M NaCl solution; however, there was no signal derived from Na⁺ for native WO₃ (Fig. S2†). These results suggest that the cesium tungstate species of the thin WO₃ surface layers have an ion-exchange ability and can be exchanged for H⁺ and Na⁺. As for the ion-exchange candidates, H⁺, Fe²⁺, and Fe³⁺ ions are present in the reactant solution. Table 1 shows the effects of ion-exchange treatments for Cs–WO₃ and WO₃ on photocatalytic activities. The photocatalytic activities of native WO₃ had no practical impact on the ion-exchange treatments using various aqueous solutions because it has no ion-exchange ability. On the other hand, ion-exchanged Cs–WO₃ with 1 M H₂SO₄ solution (H–Cs–WO₃) has a higher activity compared to Cs–WO₃ washed with pure water. Here, the activity of H–Cs–WO₃ drastically decreased (from 156 to 27 μmol h⁻¹) when H–Cs–WO₃ underwent thermal treatment at 673 K for 30 min. It is likely that this inactivation was related to the extinction of H⁺-exchanged sites via a dehydration reaction. It was noted that the surface structure with nanostep and plane terrace on H–Cs–WO₃, as shown in Fig. 4, remained after this thermal treatment (Fig. S3†). Therefore, the high activation of WO₃ by Cs-modification is mainly caused by the introduction of ion-exchange sites on the WO₃ surface, and the contribution of the surface structure with nanostep and plane terrace is not very large. We also found that the photocatalytic activity of H–Cs–WO₃ was further improved from 156 to 200 μmol h⁻¹ by ion exchange for the Fe²⁺ ion (Fe–H–Cs–WO₃); however, the activity of H–Cs–WO₃ was unchanged after stirring in a Fe³⁺ solution, as shown in Table 1. The color of the Cs–WO₃ powder also changed only after stirring in the Fe²⁺ solution. Fig. 7 shows the DRS spectra of ion-exchanged Cs–WO₃ with Fe²⁺ and Fe³⁺ aqueous solutions. The specific absorption band developed in the wavelength range 450–600 nm on the Cs–WO₃ photocatalyst after the stirring treatment with the Fe²⁺ solution, but not with the Fe³⁺ solution. Therefore, it was determined that this specific absorption was derived from the Fe²⁺ ion incorporated on the WO₃ surface by an ion-exchange reaction. Cs⁺- and H⁺-ion-exchanged sites on the WO₃ surface may not directly replace the Fe³⁺ ion because of the difference in the atomic value. Consequently, experiments indicated that the ion-exchange sites formed on the cesium tungstate species on the thin WO₃ surface layers, and Cs⁺ ions were exchanged for H⁺ and Fe²⁺, improving the photocatalytic activity. It was noted that the Fe/W ratios on WO₃ and Cs–WO₃ treated with Fe³⁺ solution were higher than those treated with Fe²⁺ solution (Table 1). This result also indicates that the characteristic changes in the powder color and photocatalytic ability were caused by a small amount of Fe²⁺ incorporating in the ion-exchange sites, and not by Fe³⁺ adsorption on the WO₃ surface. Fe–H–Cs–WO₃ showed no activity when the irradiation wavelength was longer than 500 nm. Therefore, the absorption band derived from the Fe²⁺ ion is not available for the water-oxidation reaction in the presence of the Fe³⁺ ion. Here, we examined the ion-exchange effects for H⁺ and Fe²⁺ over M–WO₃ (M = Na, K, Rb, and Ag) on photocatalytic activities. The activities of all WO₃ photocatalysts improved with ion-exchange treatments (Fig. S4†). Therefore, it was concluded that the high activation of M–WO₃ (M = Na, K, Rb, and Ag) photocatalysts is derived from a mechanism similar to Cs–WO₃ and is based on the ion-exchange ability.

Fig. 7 (i) Reflection spectra of Cs–WO₃. (a) Powder dispersion in 10 mM H₂SO₄ aq. sol., (b) 4 mM Fe³⁺ sol., and (c) 4 mM Fe²⁺ aq. sol., (ii) magnification of (i). Cs amount: 1 mol% for WO₃.

Effects of condition of reactant solution over Cs–WO₃ photocatalyst on photocatalytic water oxidation in the presence of Fe³⁺ ions

In the case of Fe³⁺ reduction, Fe³⁺ ions must preferentially arrive at the photocatalyst surface and easily receive the photo-generated electrons from the photocatalyst.²⁸ Fe³⁺ ions form various types of complex ions, depending on the coexisting anion species.^32–34 The adsorption and reaction behaviors of the Fe³⁺ ion are changed by the state of the Fe³⁺ complex ion. Therefore, it is likely that the state of a Fe³⁺ complex ion affects the photocatalytic activity for Fe³⁺ reduction. Table 2 shows the photocatalytic O₂-evolution activity over Fe–H–Cs–WO₃ under various reactant solution conditions. The activities strongly depended on the type of iron source. Fe–H–Cs–WO₃ showed the highest activity in an aqueous Fe(ClO₄)₃ solution. There was a poor correlation between the reaction rate and the adsorption amount of Fe³⁺ ions. Therefore, it is likely that there are other factors affecting the photocatalytic activities. This tendency of the coexisting anion species was also observed in the case of other photocatalysts, native WO₃, BiVO₄, and TiO₂ (Table 3). The photocatalytic activities of WO₃, BiVO₄, and TiO₂ using the Fe(ClO₄)₃ solution were 4.4, 4.7, and 12 times higher than those using the Fe₂(SO₄)₃ solution. The cyclic voltammetry (CV) profiles of various Fe³⁺ solutions changed depending on the type of iron solution (Fig. 8). The order of redox potentials estimated by the CVs was: SO₄²⁻ (+0.72 vs. RHE) < Cl⁻ (+0.77 vs. RHE) < NO₃⁻ = ClO₄⁻ (+0.79 vs. RHE). There was a good correlation between the redox potential and reaction rate. The negative shift of the redox potential shows that it is thermodynamically difficult to reduce the Fe³⁺ ion. On the other hand, the difference between the conduction band potentials of these photocatalysts (WO₃,³⁵ BiVO₄,³⁵ and TiO₂³⁶ and the Fe^3+/2+ redox potential are estimated to be ca. 0.3, 0.7, and 0.8 V, respectively. There was no correlation between the improvement in the photocatalytic activity and the value of the potential difference. Therefore, we could not explain the photocatalytic behaviors using only the potential differences. Fe³⁺ ions in aqueous Fe₂(SO₄)₃ or FeCl₃ solutions mainly form Fe(SO₄²⁻)_x(H₂O)_6−x or Fe(Cl⁻)_x(H₂O)_6−x and Fe(H₂O)₆ in aqueous Fe(NO₃)₃ and Fe(ClO₄)₃.^32–34 The difference of coordinate species for Fe³⁺ ions may affect the reducibility of Fe complex ions; it may be difficult to reduce the Fe³⁺ ion if SO₄²⁻ or Cl⁻ ions are introduced in the Fe³⁺ complex ions compared to only H₂O coordination. Therefore, it was concluded that the photocatalytic reaction rates were strongly affected by the coordination of the Fe³⁺ complex ion and thermodynamic reducibility of the Fe complex ions.

Table 2 Photocatalytic O₂ evolution over Fe–H–Cs–WO₃ in various reactant solution conditions^b

Catalyst	Fe^3+ source	Adsorption amount/μmol g^−1		O2-evolution rate/μmol h^−1
		Fe^3+	Fe^2+
a Cs amount: 2.2 mol% for WO3.b Catalyst: 0.4 g, reactant solution: 300 mL of 2 mM Fe(III), light source: 300 W Xe-lamp (L42). Nd, not detected.
Cs–WO3^a	Fe2(SO4)3	16	nd.	200
Cs–WO3^a	FeCl3	14	nd.	241
Cs–WO3^a	Fe(NO3)3	—	—	290
Cs–WO3^a	Fe(ClO4)3	18	nd.	297

---

Table 3 Effect of types of Fe³⁺ solution over TiO₂, WO₃, and BiVO₄ photocatalyst on reaction rates for water oxidation^b

Catalyst	Fe^3+ source	pH adjuster^a	Irradiation wavelength	O2-evolution rate/μmol h^−1
a Initial pH was always adjusted to 2.3.b Catalyst: 0.4 g, reactant solution: 300 mL of 2 mM Fe(III), light source: 300 W Xe-lamp.
TiO2	Fe2(SO4)3	H2SO4	λ > 300 nm	30
WO3	Fe2(SO4)3	H2SO4	λ > 420 nm	8.1
BiVO4	Fe2(SO4)3	H2SO4	λ > 420 nm	47
TiO2	Fe(ClO4)3	HClO4	λ > 300 nm	370
WO3	Fe(ClO4)3	HClO4	λ > 420 nm	35
BiVO4	Fe(ClO4)3	HClO4	λ > 420 nm	221

---

Fig. 8 Cyclic voltammetry at Pt-electrode (SA: 0.01 cm²) in (a) 2.0 mM Fe₂(SO₄)₃ in 0.1 M Na₂SO₄, (b) 4.0 mM FeCl₃ in 0.1 M NaCl, (c) 4.0 mM Fe(NO₃)₃ in 0.1 M NaNO₃, and (d) 4.0 mM Fe(ClO₄)₃ in 0.1 M NaClO₄. pH value of iron solutions was adjusted to 2.3 using (a) H₂SO₄, (b) HCl, (c) HNO₃, and (d) HClO₄. Scan rate: 5 mV s⁻¹.

Fig. 9 shows the action spectrum of water oxidation over a Fe–H–Cs–WO₃ photocatalyst in Fe(ClO₄)₃. The apparent QE at 420 nm was 31%. This is the highest value among all the photocatalysts using reversible redox mediators under visible light. It was confirmed that the QE estimated by the Fe²⁺-production rate had the same value as that of the efficiency estimated by the O₂-production rate. The QE at 365 nm was 21%. The Fe³⁺ solution shows a strong absorption in the UV light region (Fig. 9(c)). Therefore, it is likely that the QE was low because of the light-shield effects of the Fe³⁺ reactant solution. Fig. 10 shows the time course of O₂ evolution over a Fe–H–Cs–WO₃ photocatalyst in Fe(ClO₄)₃ when a solar simulator (AM 1.5) is used as the light source. O₂-evolution rate at Fe³⁺/Fe²⁺ = 1 was 94 μmol h⁻¹ (irradiation area: 9 cm²). The water oxidation reaction accompanied by Fe³⁺ reduction is the energy-storage reaction shown in eqn (5). The solar-to-chemical energy-conversion efficiency (η_sun) was estimated to be 0.38%. This is also the best value among all the photocatalysts using reversible redox mediators.

2Fe³⁺ + H₂O → 2Fe²⁺ + 2H⁺ + 1/2O₂ (pH = 2), ΔG = 66.1 kJ (5)

Fig. 9 (a) Action spectrum of water oxidation in the presence of Fe³⁺ ions. Absorbance spectra of (b) Cs–WO₃ photocatalyst and (c) Fe(ClO₄)₃ solution.

Fig. 10 Time course of photocatalytic water oxidation accompanied with Fe³⁺ reduction using solar simulator (AM 1.5). Broken line shows upper limit of O₂ evolution expected from the amount of Fe³⁺ (600 μmol) added to solution. Cs–WO₃: 0.4 g, reactant solution: 300 mL of 2 mM Fe(ClO₄)₃, reaction cell: side-irradiation cell. Initial pH of reactant solution was always adjusted to 2.3 using perchloric acid.

Mechanism of activation by Cs treatment

We investigated the photoelectrochemical performance of WO₃ electrodes to elucidate the role of surface treatment on WO₃ photocatalysts because reductive and oxidative reactions can be individually evaluated by photoelectrochemical methods. Table 4 shows the O₂-production photocurrent at 0.8 V vs. RHE and Fe-reduction current at 0.45 V vs. RHE for WO₃ electrodes with and without surface treatment, respectively. The O₂-production photocurrent of the H–Cs–WO₃ electrode was much higher than that of the WO₃ electrode without treatment, indicating that water oxidation is promoted on the H–Cs–WO₃ surface; further improvement by the partial ion exchange for Fe²⁺ was very small. On the other hand, the Fe³⁺-reduction current of the Fe–H–Cs–WO₃ electrode was higher than that of the other electrodes, suggesting that Fe³⁺ reduction is promoted on the Fe–H–Cs–WO₃ surface. The H⁺-exchanged site would be strongly adsorbed with water as H₃O⁺, which is suitable for the H₂O oxidation. On the other hands, the Fe²⁺-exchange site may act as the effective reduction site of Fe³⁺ adsorbed on the outer surface of the photocatalyst, because the electron transfer between the same element ions (Fe²⁺ and Fe³⁺) is ideal. Consequently, both reduction and oxidation abilities of WO₃ improved at different sites formed by Cs-modification and followed by ion exchange for Fe²⁺ and H⁺ ions, resulting in the surface-modified WO₃ photocatalyst showing remarkably high activity. The suppression of the reverse reaction between the photo-generated holes and Fe²⁺ ions as shown in Fig. 2 is important for realizing highly efficient water oxidation. Fig. 11(a and b) shows the influence of the concentration of coexisting Fe²⁺ ions with 2 mM of Fe³⁺ ions over the WO₃ photocatalyst, with and without surface modification, on the photocatalytic activity for water oxidation into O₂. The activity of native WO₃ decreased when 0.5 mM of Fe²⁺ ions coexisted in the reactant solution. This indicates that the Fe²⁺ ion is easily oxidized compared to water on a normal WO₃ surface. On the other hand, the activity of the Fe–H–Cs–WO₃ photocatalyst was unaffected by the concentration of the Fe²⁺ ion. This suggested that surface modification improves the oxidation and reduction abilities of WO₃; it also improves the suppression effect on the undesirable reaction between the photo-generated holes and Fe²⁺ ions. Fig. 11(c–f) shows the suppression effect for WO₃ photocatalysts with and without surface modification on the undesirable reaction when other reversible redox mediators are used (VO₂⁺/VO²⁺: +1.00 vs. RHE and IO₃⁻/I⁻: +1.09 vs. RHE). The photocatalytic activities of Pt/WO₃ and WO₃ in the presence of IO₃⁻ and VO₂⁺ ions for O₂ evolution remarkably improved (3.5 and 7.3 times, respectively) with surface modification. Among the reactions using different redox mediators, water oxidation to O₂ gas was the common reaction, suggesting that an excellent water-oxidation site is probably formed on the surface-modified WO₃ photocatalyst. Interestingly, the undesirable reactions were also effectively suppressed over the Fe–H–Cs–WO₃ photocatalyst, while the activity of native WO₃ drastically decreased when 0.5 mM of I⁻ or VO²⁺ ions were also present in the reactant solution. Therefore, an excellent suppression effect for the undesirable reaction between the photo-generated holes and Red was obtained, regardless of the type of redox reagent. Thus, this ideal reaction selectivity arose because of the suppression of the undesirable reaction between the photo-generated holes and various Reds and the construction of an effective site that can induce only water oxidation. According to the results mentioned above, a few layers of cesium tungsten species initially formed on the WO₃ surface by thermal treatment using a cesium aqueous solution. Cs⁺ sites in cesium tungsten species have ion-exchange ability, and Cs⁺ easily exchanged for H⁺ and Fe²⁺ in acidic Fe²⁺ solutions. Electrochemical technique showed that the H⁺-exchange site promotes water oxidation, and the Fe²⁺-exchange site promotes Fe³⁺ reduction. Moreover, the photo-generated hole can selectively oxidize water to O₂ even when various Reds were present. Consequently, the surface-modified WO₃ photocatalyst showed remarkably high activity for water oxidation (Fig. 12). A redox potential is usually more negative than the water oxidation potential (+1.23 vs. RHE) in the case of water oxidation with a positive Gibbs free energy change. Therefore, the red is easily oxidized compared to water, even though this reaction is undesirable. In this study, the undesirable reaction was effectively suppressed on a Fe–H–Cs–WO₃ photocatalyst. These results provide a promising route for improving various photocatalytic water-splitting techniques such as the Z-scheme system, photocatalysis-electrolysis system,²⁰ and Fe²⁺ ion fuel cells.^37,38

Table 4 O₂-production photocurrent and Fe³⁺-reduction current over various surface-treated WO₃ photoelectrodes

Photoelectrodes	O2-production photocurrent at 1.0 V mA^−1	Fe^3+-reduction current at 0.4 V mA^−1
WO3	0.11	0.45
Fe–WO3	0.12	0.47
H–Cs–WO3	0.20	0.49
Fe–H–Cs–WO3	0.20	0.55

---

Fig. 11 Effect of concentration of (I) Fe²⁺, (II) VO²⁺, and (III) I⁻ ions over WO₃ photocatalyst (a), (c), and (e) with and (b), (d), and (f) without surface ion-exchange sites on photocatalytic reaction rates for water oxidation in the presence of (I) Fe³⁺, (II) VO₂⁺, and (III) IO₃⁻ ions. In case of reaction using IO₃⁻ ion as redox mediator, Pt cocatalyst was loaded on WO₃ photocatalyst.

Fig. 12 Description of highly activated WO₃ surface.

Conclusions

WO₃ powders modified with various metal salt solutions were studied as photocatalysts for the oxidation of water to O₂ gas in an aqueous Fe³⁺ solution under visible light (λ > 420 nm). The activity of WO₃ was remarkably improved by thermal treatment at 773 K using MCl (M = Na, K, Rb, and Cs) and a AgNO₃ aqueous solution. WO₃ treated using a cesium aqueous solution had the highest activity. WO₃ particles were covered with a very thin layer of cesium tungstate species, which had a different structure from the WO₃ bulk. It was observed that the Cs⁺ site of the cesium tungstate species on WO₃ had an ion-exchange ability, and the photocatalytic performances of the WO₃ photocatalyst were further improved by the ion-exchange treatment with H⁺ and Fe²⁺ ions. The photocatalytic activity for water oxidation in the presence of Fe³⁺ ions was strongly dependent on the types of Fe³⁺ solutions. The best activities were obtained in a Fe(ClO₄)₃ aqueous solution over Cs-treated WO₃ and various other photocatalysts. The apparent QE at 420 nm and solar-to-chemical energy-conversion efficiency over Cs-treated WO₃ were 31% and 0.38%, respectively. These values are the best among the photocatalysts using various reversible redox mediators under visible light. Electrochemical analyses suggested that Fe²⁺- and H⁺-incorporated sites on the WO₃ surface promoted Fe³⁺ reduction and water oxidation, respectively. Moreover, the surface of the Fe–H–Cs–WO₃ photocatalyst had the ability to suppress undesirable reverse reactions between the photo-generated holes and Fe²⁺ ions. The improvement of the water oxidation and suppression of the reverse reaction were still obtained by Cs-treatment, even when other redox ions (IO₃⁻/I⁻ and VO₂⁺/VO²⁺) were used. Therefore, the Cs-treated WO₃ has an excellent activity and a broad utility for the photocatalytic oxidation reaction of water to O₂ using various reversible redox mediators.

Acknowledgements

This research was supported by The Funding Program for Next Generation World-Leading Researchers (NEXT Program).

Notes and references

1. J. S. Lee, Catal. Surv. Asia, 2005, 9, 217–227.
2. A. J. Esswein and D. G. Nocera, Chem. Rev., 2007, 107, 4022–4047.
3. F. E. Osterloh, Chem. Mater., 2008, 20, 35–54.
4. A. Kudo and Y. Miseki, Chem. Soc. Rev., 2009, 38, 253–278.
5. Y. Inoue, Energy Environ. Sci., 2009, 2, 364–386.
6. K. Maeda and K. Domen, J. Phys. Chem. Lett., 2010, 1, 2655–2661.
7. R. Abe, Bull. Chem. Soc. Jpn., 2011, 84, 1000–1030.
8. K. Maeda, K. Teramura, D. Lu, T. Takata, N. Saito, Y. Inoue and K. Domen, Nature, 2006, 440, 295; K. Maeda, K. Teramura and K. Domen, J. Catal., 2008, 254, 198–204.
9. Y. Lee, H. Terashita, Y. Shimodaira, K. Teramura, M. Hara, H. Kobayashi, K. Domen and M. Yashima, J. Phys. Chem. C, 2007, 111, 1042–1048.
10. K. Sayama, K. Mukasa, R. Abe, Y. Abe and H. Arakawa, Chem. Commun., 2001, 2416–2417; K. Sayama, K. Mukasa, R. Abe, Y. Abe and H. Arakawa, J. Photochem. Photobiol., A, 2002, 148, 71–77.
11. H. Kato, M. Hori, R. Konta, Y. Shimodaira and A. Kudo, Chem. Lett., 2004, 33, 1348–1349; Y. Sasaki, A. Iwase, H. Kato and A. Kudo, J. Catal., 2008, 259, 133–137.
12. R. Abe, T. Takata, H. Sugihara and K. Domen, Chem. Commun., 2005, 3829–3831; R. Abe, M. Higashi and K. Domen, ChemSusChem, 2011, 4, 228–237.
13. M. Higashi, R. Abe, K. Teramura, T. Takata, B. Ohtani and K. Domen, Chem. Phys. Lett., 2008, 452, 120–123; M. Higashi, R. Abe, T. Takata and K. Domen, Chem. Mater., 2009, 21, 1543–1549.
14. M. Higashi, R. Abe, A. Ishikawa, T. Takata, B. Ohtani and K. Domen, Chem. Lett., 2008, 37, 138–139.
15. K. Maeda, H. Terashima, K. Kase, M. Higashi, M. Tabata and K. Domen, Bull. Chem. Soc. Jpn., 2008, 81, 927–937; K. Maeda, M. Higashi, D. Lu, R. Abe and K. Domen, J. Am. Chem. Soc., 2010, 132, 5858–5868.
16. Y. Sasaki, H. Nemoto, K. Saito and A. Kudo, J. Phys. Chem. C, 2009, 113, 17536–17542.
17. R. Abe, K. Shinmei, K. Hara and B. Ohtani, Chem. Commun., 2009, 3577–3579.
18. M. Tabata, K. Maeda, M. Higashi, D. Lu, T. Takata, R. Abe and K. Domen, Langmuir, 2010, 26, 9161–9165.
19. Y. Miseki, S. Fujiyoshi, T. Gunji and K. Sayama, Catal. Sci. Technol., 2013, 3, 1750–1756.
20. Y. Miseki, H. Kusama, H. Sugihara and K. Sayama, J. Phys. Chem. Lett., 2010, 1, 1196–1200.
21. J. R. Darwent and A. Mills, J. Chem. Soc., Faraday Trans. 2, 1982, 78, 359–367.
22. W. Erbs, J. Desilvestro, E. Borgarello and M. Grätzel, J. Phys. Chem., 1984, 88, 4001–4006.
23. Y. Miseki, H. Kusama and K. Sayama, Chem. Lett., 2012, 41, 1489–1491.
24. S. S. K. Ma, K. Maeda, R. Abe and K. Domen, Energy Environ. Sci., 2012, 5, 8390–8397.
25. T. Arai, M. Yanagida, Y. Konishi, H. Sugihara and K. Sayama, Electrochemistry, 2008, 76, 128–131.
26. G. R. Bamwenda, K. Sayama and H. Arakawa, J. Photochem. Photobiol., A, 1999, 122, 175–183.
27. G. R. Bamwenda, T. Uesigi, Y. Abe, K. Sayama and H. Arakawa, Appl. Catal., A, 2001, 205, 117–128.
28. T. Ohno, D. Haga, K. Fujihara, K. Kaizaki and M. Matsumura, J. Phys. Chem. B, 1997, 101, 6415–6419.
29. A. Iwase, H. Kato and A. Kudo, J. Sol. Energy Eng., 2010, 132, 021106.
30. H. Kato, K. Asakura and A. Kudo, J. Am. Chem. Soc., 2003, 125, 3082–3089.
31. A. Iwase, H. Kato and A. Kudo, ChemSusChem, 2009, 2, 873–877.
32. H. Kato, Y. Sasaki, A. Iwase and A. Kudo, Bull. Chem. Soc. Jpn., 2007, 80, 2457–2464.
33. S. P. Moulik and K. K. S. Gupta, J. Indian Chem. Soc., 1972, 49, 447.
34. R. A. Whiteker and N. Davidson, J. Am. Chem. Soc., 1953, 75, 3081–3085.
35. S. J. Hong, S. Lee, J. S. Jang and J. S. Lee, Energy Environ. Sci., 2011, 4, 1781–1787.
36. L. Kavan, M. Gratzel, S. E. Gilbert, C. Klemenz and H. J. Scheel, J. Am. Chem. Soc., 1996, 118, 6716–6723.
37. S. Cheng, B. A. Dempsey and B. E. Logan, Environ. Sci. Technol., 2007, 41, 8149–8153.
38. H. Eom, K. Chung, I. Kimand and J. Han, Chemosphere, 2011, 85, 672–676.

---

Footnote

† Electronic supplementary information (ESI) available: XRD patterns, Raman spectra, and reflectance spectra of (a) WO₃ and (b) Cs–WO₃ (S1), XPS spectra of (a) WO₃ and (b) Cs–WO₃ after stirring treatment in 1 M NaCl solution (S2), SEM images of H–Cs–WO₃ (a) with and (b) without thermal treatment at 673 K for 30 min (S3), and reaction rates for water oxidation over WO₃ photocatalysts treated with various metal salt solutions. Stirring treatment was conducted with (a) water and (b) FeSO₄ solution (S4). See DOI: 10.1039/c3ra47772j

---