Electrochemically Assisted Photocatalysis of Hybrid WO₃/TiO₂ Films: Effect of the WO₃ Structures on Charge Separation Behavior

Abstract

In this study, the photocatalysis of hybrid WO₃/TiO₂ films with different loadings of WO₃ were investigated with and without potential bias. It was clearly indicated that hybrid WO₃/TiO₂ films show less photo-reactivity under only UV-irradiation, while more effective photocatalysis under potential bias than either TiO₂ or WO₃ by themselves, their photocatalytic performance depending on the loadings of WO₃. In particular, a hybrid WO₃/TiO₂ film involving an amorphous-like WO₃ phase plays a significant role in an enhancement of the electrochemically assisted photocatalysis.

Introduction

Since the effect of Honda–Fujishima on the photocatalysis of TiO₂ was discovered [1], different kind of semiconductor photocatalysts have been extensively studied by many researchers [2–7]. In particular, TiO₂ is interesting as photocatalyst, photoinduced super-hydrophilic materials, while tungsten trioxide (WO₃) as photo-anode, electro- or photo- chromic materials and photocatalyst [8–14]. In recent years, a considerable attention has been paid to photo-electrochemical properties of the WO₃/TiO₂ double layers, such as photocatalysis, photochromism, and photo-rechargeable battery [15–20]. Under the band gap excitation of WO₃ and/or TiO₂ layers by themselves, photoinduced electrons transfer to the conduction band of WO₃ layers, while holes transfer to the valence band of TiO₂ layers. So that, they play a significant role in maximizing the photocurrent generation efficiency to promote charge separation by accumulating electrons and holes in two different semiconductor layers with suppression of charge recombination.

Our present study deals with electrochemically assisted photocatalysis of hybrid WO₃/TiO₂ film electrodes with different loadings of WO₃. Special attention has been focused on effect of the band structures of an amorphous-like WO₃ or a polycrystalline WO₃ coupled with TiO₂, on the charge separation behavior under UV-irradiation.

Experimental Details

Film Preparation

TiO₂ (anatase structure) and tungstic acid (H₂WO₄) were purchased from Kanto Chemicals, and ammonium tungstate, (NH₄)₁₀W₁₂O₄₁·5H₂O was purchased from Kishida Chemicals. Four different types of WO₃-based solid films were prepared by the following procedures: (1) WO₃/TiO₂ samples with different WO₃ loadings were prepared by an impregnation of an aqueous solution of ammonium tungstate into TiO₂. The obtained powder was mixed with triethyleneglycol to form slurry, and it was spread on a conductive indium tin oxide glass (ITO, 10 Ω/square) by squidgie method, followed by thermal treatment at 773 K for 15 min in air, and labeled as x-WO₃/TiO₂ (x represents weight percent of WO₃); (2) A polycrystalline WO₃ material was prepared by thermal decomposition of ammonium tungstate at 873 K for 6 h in air. The polycrystalline WO₃ powder was treated in the same manner on an ITO glass, followed by thermal treatment in air at 773 K for 15 min, and labeled as c-WO₃; 3) A physically mixed powder (40 wt.% as WO₃) containing both 0.6 g of TiO₂ and 0.4 g of c-WO₃ was treated in the same manner on an ITO glass, followed by thermal treatment in air at 773 K for 15 min, and labeled as 40c-WO₃/TiO₂; 4) The tungstic acid was also treated in the same manner on an ITO glass, followed by thermal treatment at 573 K for 15 min in air, and labeled as a-WO₃. The mass of each film was adjusted to be about 1.5 mg cm⁻² on an ITO glass unless the mass of film was especially mentioned.

Film Characterization

Scanning Electron Microscopy (SEM) images were taken at 20 kV accelerating voltage in a JEOL JSM-840A. Atomic Force Microscopy (AFM) images were obtained in a Shimadzu SPM-9500J by tapping mode using an etched silicon tip. The X-ray diffraction (XRD) patterns were obtained with a RIGAKU RINT2000 using Cu K_α radiation (λ = 1.5417 Å) in a scan range of 20–40° and a scan speed of 2° min⁻¹. The UV–Vis spectroscopic measurements in diffuse reflectance mode were carried out using a Shimadzu UV–Vis recording spectrophotometer, UV-2200A. The collected data were transformed using the Kubelka–Munk function.

Photoelectrochemical Measurements

Photoelectrochemical measurements were carried out by a Potentiostat/Galvanostat (HABF 5001, HOKUTO DENKO) in a standard two-compartment cell in which three electrodes were connected by the bridge as shown in Fig. 1. The I−V characteristics of film electrodes were obtained at the potential sweep rate of 10 mV s⁻¹. Black light of the fluorescent lamp (0.53 mW·cm⁻² at 365 nm) was used for photoirradiation of the films. The electrolyte solution was used for 0.5 M Na₂SO₄ aqueous solution involving 0.5 M HCOOH. Also, the pH of aqueous solutions was adjusted by aqueous solutions of H₂SO₄ or NaOH. They were bubbled with N₂ gas for 30 min under vigorous stirring before use. The products after photoreaction were analyzed by gas chromatography equipped with a thermal conductivity detector (TCD) for the analysis of CO₂.

Fig. 1

Schematic diagram of two-compartment cell employed in electrochemically assisted photocatalysis. (CE: Pt wire, RE: Ag/AgCl electrode, WE: film electrode)

Results and Discussion

SEM and AFM images from top views of a 40-WO₃/TiO₂ film was observed as shown in Fig. 2. The film was seen to form the secondary particle having porous surface by an aggregation of the primary particles from the SEM image. Electrolyte solutions involving formic acid can be easily penetrated into the porous film, which is expected to exhibit high photocatalytic activity. The AFM image of 40-WO₃/TiO₂ shows that this film consists of secondary particles (about 300–1,000 nm in diameter) that are in contact with each other.

Fig. 2

SEM and AFM micrographs from top views of 40-WO₃/TiO₂ (a and b)

Figure 3 shows X-ray diffraction patterns of WO₃/TiO₂ films with different loadings of WO₃. Structural change of WO₃ was observed to transform from an amorphous-like structure to a polycrystalline structure as an increase of WO₃ on TiO₂, that is, 20- and 40-WO₃/TiO₂ involve an amorphous-like WO₃ phase, whose diffraction pattern is similar with that of a-WO₃, while 60-WO₃/TiO₂ and 40c-WO₃/TiO₂ involve a polycrystalline WO₃ as a major phase [20]. Here, an amorphous-like WO₃ structure can be defined to be as the intermediate between an amorphous WO₃ and a polycrystalline WO₃.

Fig. 3

XRD patterns of TiO₂ (a), 20-, 40-, 60-WO₃/TiO₂ (b–d), 40c-WO₃/TiO₂ (e), c-WO₃ (f) and a-WO₃ (g) (*: WO₃; #: anatase TiO₂)

The photocurrent responses as a function of applied potentials (I−V characteristics) for film electrodes under photoirradiation and in dark are shown in Fig. 4. Firstly, the saturated photocurrent was obtained at +1.0 V vs. Ag/AgCl. Secondly, a flatband potential of the film was defined to be as a zero photocurrent potential, i.e., the photocurrent can be seen at more anodic potential, while no more effect of electrochemical intercalations of protons are observed at more cathodic than the flatband potential.

Fig. 4

Dependence of the photocurrent as a function of applied voltage (I−V characteristics) under UV-irradiation (up) and in dark (down): TiO₂ (a), 40-WO₃/TiO₂ (b), c-WO₃ (c) and a-WO₃ (d). The pH of an aqueous solution was adjusted at 2.1

Influence of film thickness on the saturated photocurrent at +1.0 V versus Ag/AgCl is shown in Fig. 5. The photocurrent for each film electrodes was saturated at more than 1.5 mg cm⁻² of film mass. Given the condition of the saturated photocurrent of the film electrodes, it was found that (1) the photocurrent efficiency depend on loadings of WO₃ on TiO₂, and was maximized at 40 wt.% WO₃ and then decreased at loadings higher than 60 wt.%; (2) 40c-WO₃/TiO₂ exhibits lower photocurrent efficiency than either WO₃ or TiO₂ by themselves as shown in Fig. 6.

Fig. 5

Dependence of the saturated photocurrent density as a function of mass of films for TiO₂ (a), 40-WO₃/TiO₂ (b), and c-WO₃ (c). The pH of an aqueous solution was adjusted at 2.1

Fig. 6

Yields of CO₂ photo-formed with (gray bar) and without (white bar) potential bias at +1.0 V versus Ag/AgCl, and saturated photocurrent density (black circle). The pH of an aqueous solution was adjusted at 2.1

Figure 6 also shows the yields of CO₂ for the photocatalytic decomposition of formic acid under UV-irradiation with or without potential bias at +1.0 V versus Ag/AgCl. With no bias, all of the hybrid WO₃/TiO₂ film electrodes exhibit lower photocatalytic activity than either TiO₂ or WO₃ by themselves. On the other hand, under potential at +1.0 V versus Ag/AgCl, UV-irradiation of hybrid WO₃/TiO₂ film electrodes show higher photocatalytic activity than either TiO₂ or WO₃ by themselves, while it was found to be maximized at 60 wt.% WO₃ and then decreased at loadings higher than 40 wt.%, which corresponds with the efficiency of the saturated photocurrent. In fact, this profile can be explained by the transformation from an amorphous-like WO₃ (at ∼40 wt.%) to a polycrystalline WO₃ structure (60∼ wt.%) as evidenced by the XRD data as shown in Fig. 3. Furthermore, 40c-WO₃/TiO₂ was seen to exhibit lower efficiency for the electrochemically assisted photocatalysis than 40-WO₃/TiO₂. These results indicate that the hybrid WO₃/TiO₂ film having an amorphous-like WO₃ can play an important role in more effective charge separation than the film having a polycrystalline WO₃ on photo-electrochemical cells.

Figure 7 shows the potential shifts of photocharge under UV-irradiation, and discharge under dark condition. Photocharged potentials were saturated at around −0.49 V for TiO₂, −0.29 V for 40-WO₃/TiO₂, +0.03 V for a-WO₃ and +0.03 V for c-WO₃ versus Ag/AgCl, respectively, under UV-irradiation for 60 min, while 40-WO₃/TiO₂ and a-WO₃ exhibit blue coloration. When light is turned off, both of TiO₂ and c-WO₃ exhibit rapid self-discharge and/or discharge, while photocharged 40-WO₃/TiO₂ and a-WO₃ exhibit a typical Faradic discharge behavior (bleaching color) by galvanostatic discharge at the rate of 10 μA cm⁻². Photocharge and discharge processes are associated with intercalation of protons into WO₃ to form tungsten bronze (H _x WO₃) and deintercalation of protons, respectively, accordingly to the following process of (1) [17, 19, 21].

$$ {\text{WO}}_{3} + xe^{ - } + x{\text{H}}^{ + } \underset{{{\text{discharge}}}}{\overset{{{\text{photocharge}}}}{\rightleftarrows}}{\text{H}}_{x} {\text{WO}}_{3} $$

(1)

Fig. 7

Changes in potential by photocharge for 60 min, and subsequently galvanostatic discharge at 10 μA·cm⁻² on TiO₂ (a), 40-WO₃/TiO₂ (b), c-WO₃ (c) and a-WO₃ (d). The pH of an aqueous solution was adjusted at 2.1

Moreover, XRD data clearly indicate that tungsten bronze (H _x WO₃) (x: 0.23 or 0.33) is formed through the intercalation of protons depending on the photocharged potentials. The reason for the suppression of the photo-reactivity of the hybrid WO₃/TiO₂ films under no bias can be explained by the formation of tungsten bronze.

Figure 8 shows the linear correlation between the pH of aqueous solutions and the flatband potentials (U _fb) on TiO₂, 40-WO₃/TiO₂ and WO₃ film electrodes. The flatband potentials of TiO₂ and WO₃ film electrodes follow the theoretical equation of (2) through (de)protonation on the surface. The E ⁰ and pH in the equation of (2) represent the flatband potential at pH = 0 and the pH of aqueous solutions, respectively. Such phenomena were also observed in a hybrid WO₃/TiO₂ film electrode, suggesting that it also gives a property of metal-oxide semiconductor such as WO₃ and TiO₂.

$$ U_{{{\text{fb}}}} \, = \,E^{0} - 0.059{\text{ pH}} $$

(2)

Fig. 8

Dependence of the pH of aqueous solutions on the flatband potentials for TiO₂ (a), 40-WO₃/TiO₂ (b), c-WO₃ (c) and a-WO₃ (d)

The flatband potentials of WO₃ coupled with TiO₂ depend on the loadings of WO₃ on TiO₂, and they shift towards the positive potential from −0.54 V to +0.07 V as an increase of loadings of WO₃ as shown in Fig. 9.

Fig. 9

Dependence of the flat band potentials as a function of loadings of WO₃ on TiO₂. White circle shows the flatband potential of a-WO₃. The pH of an aqueous solution was adjusted at 2.1

As shown in Fig. 10, it was observed that the photocurrent efficiency for the film electrodes is not influenced by the pH of aqueous solutions.

Fig. 10

Dependence of the pH of aqueous solutions on the saturated photocurrent density for TiO₂ (a), 40-WO₃/TiO₂ (b), c-WO₃ (c) and a-WO₃ (d)

The bandgap energy of TiO₂ was determined to be E _g = 3.40 eV by threshold against energy (hv) as shown in Fig. 11 [I]. Likewise, the band gaps of 40-WO₃/TiO₂, a-WO₃ and c-WO₃ were estimated to be E _g = 3.35 eV for 40-WO₃/TiO₂, E _g = 3.35 eV for a-WO₃ and E _g = 2.75 eV for c-WO₃, respectively (cf. Fig. 11 [II]) [20]. It is reported that amorphous WO₃ shows larger band-gap energy (3.35 eV) than polycrystalline WO₃ (2.5–2.8 eV) as shown in the reference [13]. The similar phenomena can be also observed in the case of amorphous silicon (1.8 eV) compared with polycrystalline silicon (1.1 eV) [22]. It can be assumed that band-gap widening may be possibly explained by structural fluctuation of amorphous WO₃ or Si. On the other hand, amorphous-like WO₃ can be observed on hybrid WO₃/TiO₂ in the low loadings of WO₃, which cause the band-gap widening of WO₃. Thus, the band-gap widening of amorphous-like WO₃ may be possibly explained by (i) structural fluctuation or (ii) high dispersibility of WO₃ particles on TiO₂ surface.

Fig. 11

An estimation of the bandgap of TiO₂ [I], and absorption spectra of TiO₂ (a), 40-WO₃/TiO₂ (b), c-WO₃ (c) and a-WO₃ (d) [II]

Taking the results of flatband potentials and band gaps into consideration, the energy band structure for an amorphous-like WO₃ coupled with TiO₂ (a-WO₃/TiO₂) is proposed in Fig. 12. As a whole, the photo-induced holes and electrons on the films participate in the oxidation and the reduction of substrates, respectively. In our system, the electrons go through outer electric circuit towards counter electrode under potential bias, probably causing the reduction of oxygen. The photo-induced electrons and holes effectively transfer to the conduction band of WO₃, and to the valence band of TiO₂, respectively, by accumulating electrons and holes in two different semiconductor mixture layers. In general, the formate anion (HCOO⁻) can be oxidized by a hole to form HCOO·, followed by automatic oxidation to CO₂, H⁺ and e^- from the simple occurrence of the photocurrent doubling [10]. So that, it can be supposed that one photon gives the photoreaction of HCOOH → CO₂ + 2H⁺ + 2e⁻. It can be assumed that photoirradiation of a-WO₃/TiO₂ gives more effective charge separation (h ⁺···e ^-) than a polycrystalline WO₃ coupled with TiO₂ (c-WO₃/TiO₂), since the potential of valence band of a-WO₃ is more anodic than that of c-WO₃.

Fig. 12

Schematic energy diagram for the photo-induced charge transfer on the a-WO₃/TiO₂ film electrode. The pH of an aqueous solution was adjusted at 2.1

Conclusions

A hybrid WO₃/TiO₂ film electrode was shown to perform more effective electrochemically assisted photocatalysis (photocurrent efficiency) than either TiO₂ or WO₃ by themselves due to the effective charge separation. In particular, the hybrid WO₃/TiO₂ film having an amorphous-like WO₃ can play an important role in more effective charge separation than the film having a polycrystalline WO₃.