Composites Based on Electrodeposited WO₃ and TiO₂ Nanoparticles for Photoelectrochemical Water Splitting

Abstract

Photoelectrochemically active WO₃ films were fabricated by electrodeposition from an acidic (pH 2), hydrogen-peroxide-containing electrolyte at −0.5 V vs. SCE. WO₃-TiO₂ composites were then synthesized under the same conditions, but with 0.2 g/L of anatase TiO₂ nanoparticles (⌀ 36 nm), mechanically suspended in the solution by stirring. After synthesis, the films were annealed at 400 °C. Structural characterization by XRD showed that the WO₃ films exhibit the crystalline structure of a non-stoichiometric hydrate, whereas, in WO₃-TiO₂, the WO₃ phase was monoclinic. The oxidation of tungsten, as revealed by XPS, was W⁶⁺ for both materials. Ti was found to exist mainly as Ti⁴⁺ in the composite, with a weak Ti³⁺ signal. The efficiency of the WO₃ films and composites as an oxygen evolution reaction (OER) photo-electrocatalyst was examined. The composite would generate approximately three times larger steady-state photocurrents at 1.2 V vs. SCE in a neutral 0.5 M Na₂SO₄ electrolyte compared to WO₃ alone. The surface recombination of photogenerated electron–hole pairs was characterized by intensity-modulated photocurrent spectroscopy (IMPS). Photogenerated charge transfer efficiencies were calculated from the spectra, and at 1.2 V vs. SCE, were 86.6% for WO₃ and 62% for WO₃-TiO₂. Therefore, the composite films suffered from relatively more surface recombination but generated larger photocurrents, which resulted in overall improved photoactivity.

1. Introduction

The adverse environmental impact of using fossil fuels for energy needs and generating H₂ for chemical processing has led to significant research toward finding alternative and sustainable energy carriers, such as hydrogen (H₂), with H₂ produced by solar water splitting is more advantageous compared to other forms of green hydrogen [1,2,3]. The photoelectrochemical (PEC) water-splitting process includes light harvesting, photogenerated charge separation, and surface catalysis, and is an effective strategy as an alternative to fossil fuels. The reduction of water to produce hydrogen is coupled with the more kinetically hindered oxygen evolution reaction, which is typically carried out on metal oxide photoanodes [4,5]. Although TiO₂ is recognized as an effective inexpensive photocatalyst in the UV range, [6,7], the addition of tungsten trioxide (WO₃) is widely known to lower the band gap (~2.75 eV) and hence, allows absorption to occur within the visible range of the solar spectrum, [8,9,10,11]. However, WO₃ alone suffers from high recombination of photon-generated electron–hole pairs [12] and is susceptible to photocorrosion caused by peroxo-species formed on the surface during the photoelectrochemical reaction [13]. Tungsten trioxide doped by TiO₂ shows a red-shifted optical response in the visible light range due to a band gap decrease from a lowering of the conduction band edge, and can act as a photocatalyst. Momeni and Nazari [14] reported a change in band gap from 3.23 eV for bare TiO₂ to 2.78 eV for WO₃-doped TiO₂ nanotubes. Also, doping TiO₂ with cations like tungsten (W⁶⁺) introduces energy levels in TiO₂ just below the conduction band that acts as an electron trap increasing the lifetime of charge carriers, and thus retarding recombination [15].

Heterojunctions of WO₃/TiO₂ layers have been used to increase the efficiency of the photocatalysts [16,17,18,19,20] by aligning the band structures of the two coupled semiconductors to encourage migration of photogenerated electrons (e⁻) and holes (h⁺) in separate directions across the heterojunction boundary, which reduces the electron–hole recombination. The highly oxidative valence band of WO₃, arising from the O 2p orbitals, also helps to drive oxidation reactions, such as PEC oxygen evolution or degradation of dyes. WO₃/TiO₂ composites have been reported to perform as stable and efficient photoanodes in water splitting [21,22,23]. Numerous reports of dye degradation with tungsten doped TiO₂ or heterojunctions are reported through the reactions of a photogenerated superoxide [24,25,26,27,28]. One such heterojunction unique to electrodeposition is the work presented by Martin et al. [29]. They electrodeposited WO₃ onto a metallic Ti electrode then heat treated it to create a WO₃/TiO₂/Ti layered material that exhibited high photocurrent values toward the photoelectrocatalytic degradation of propyl paraben. An alternative strategy to reduce recombination rates is to apply a potential or bias and photogenerated electrons are drawn away from the electrode–electrolyte interface through the external surface. This electrically enhanced photocatalysis was observed for electrochemically prepared TiO₂/WO₃ layers on stainless steel for the oxidation of organics [30].

Nanostructuring is also important for generating quantum confinement effects that alter the electron and hole transport behaviors, shifting of the electronic band structure, and the large surface-to-volume ratio can significantly increase the surface reaction sites [31]. However, it can also introduce defects or additional grain boundaries that can promote recombination [32].

As mentioned, synergistic effects, as a result of combining different photoactive materials, provide a promising pathway to improve the photoelectrochemical performance in heterojunction devices for solar water splitting [33]. To this end, a nanocomposite WO₃ -TiO₂ film was electrodeposited with a WO₃ matrix incorporating nanoparticles of TiO₂, in contrast to layered materials or those with a matrix of TiO₂. The electrodeposition of WO₃ films followed the approach by Pauporté [34], where the electrochemical deposition mechanism is multistaged and occurs through the reduction of a peroxy-tungstate species as outlined in Equations (1) to (3) [35,36,37].

$${2\mathrm{W}\mathrm{O}}_4^{2-}+{4\mathrm{H}}_2{\mathrm{O}}_2\to {\mathrm{W}}_2{\mathrm{O}}_{11}^{2-}+{2\mathrm{O}\mathrm{H}}^{-}+{3\mathrm{H}}_2\mathrm{O}$$

(1)

$${\mathrm{W}}_2{\mathrm{O}}_{11}^{2-}+\left(2+3\mathrm{x}\right){\mathrm{H}}^{+}+3\mathrm{x}{\mathrm{e}}^{-}\to 2{\mathrm{H}}_{\mathrm{x}}\mathrm{W}{\mathrm{O}}_3+{\displaystyle \frac{(2+\mathrm{x})}{2}}{\mathrm{H}}_2\mathrm{O}+{\displaystyle \frac{(8-\mathrm{x})}{4}}{\mathrm{O}}_2$$

(2)

$${\mathrm{H}}_{\mathrm{x}}\mathrm{W}{\mathrm{O}}_3\to {\mathrm{W}\mathrm{O}}_3+\mathrm{x}{\mathrm{H}}^{+}+\mathrm{x}{\mathrm{e}}^{-}$$

(3)

In this study, the approach is modified by adding TiO₂ nanoparticles to the sodium tungstate–hydrogen peroxide electrolyte. The electrodeposition is then carried out to simultaneously electrodeposit a WO₃ film and capture TiO₂ nanoparticles, resulting in a one-step synthesis of a composite with improved photoactivity for water splitting. Moreover, the use of an advanced non-stationary photoelectrochemical technique (IMPS) is demonstrated to probe charge carrier photogeneration in the space-charge layer and to elucidate important kinetic constants for charge transfer and recombination.

2. Materials and Methods

2.1. Composite Electrodeposition

A schematic representation of the cell and process used to deposit WO₃ and WO₃-TiO₂ films is shown in Figure 1. All used chemicals were of ACS reagent grade. The electrolyte to electrodeposit WO₃ films was prepared from 0.025 M Na₂WO₄·2H₂O (Fisher Chemical, Hampton, NH, USA), 0.075% (v/v) H₂O₂ (Fisher Chemical, Hampton, NH, USA) which was 1.25 mL of 30% hydrogen peroxide, and the pH was adjusted to 1.4 with sulfuric acid (Fisher Cchemical, Hampton, NH, USA). To electrodeposit composite WO₃-TiO₂ films, the same electrolyte composition was used, but additionally, a particle loading of 0.2 g/L of anatase TiO₂ (Thermo Fisher Scientific, Waltham, MA, USA) of average ⌀ 36 nm was added. The nanoparticles were mechanically suspended in the electrolyte by using a magnetic stir bar. Electrodeposition was carried out in a three-electrode cell, with a stainless-steel (type 316) working electrode (plates of 2 cm × 2 cm dimensions were used as the substrate, with a circular working area of 1.226 cm²), a Pt counter electrode, and an SCE reference electrode. The substrate was positioned in a holder parallel to the bottom of the electrodeposition bath. Deposition occurred during an applied potential of −0.5 V vs. SCE for 5 min.

2.2. Structural and Surface Morphology Characterization

The as-deposited WO₃ and WO₃-TiO₂ films were thermally annealed at 400 °C for two hours, with a 2 °C min⁻¹ temperature ramp-up rate. The surface morphology was characterized by a Hitachi TM 3000 scanning electron microscope (Hitachi Ltd., Tokyo, Japan). XRD diffraction patterns were obtained with a Rigaku MiniFlex II x-ray diffractometer (Rigaku Corporation, Tokyo, Japan).

The XPS analyses were carried out with a Kratos Axis Supra spectrometer (Kratos Analytical Limited, Manchester, UK), using a monochromatic Al K(alpha) source (15 mA, 15 kV). Survey scan analyses were carried out on the area of 300 × 700 µm at a pass energy of 160 eV. High-resolution analyses were also carried out on the area of 300 × 700 µm, but at a pass energy of 20 eV. The XPS signal due to adventitious carbon located at 284.8 eV was used as a binding energy (BE) reference.

2.3. PEC Characterization

The photoelectrochemical response of the resulting deposits was examined with linear sweep voltammetry, chronoamperometry, and IMPS, using a ZAHNER Zennium CIMPS-QE/IPCE system (Zahner-Elektrik, Kronach, Germany). The photoelectrochemical cell included a quartz window for front-illumination of the films, a platinum counter electrode, and an SCE reference electrode. A neutral 0.5 M Na₂SO₄ electrolyte was used. The light source was a variable-intensity LED, emitting monochromatic 365 nm UV light.

A typical characterization procedure of an electrochemically deposited and annealed WO₃ or WO₃-TiO₂ film consisted of these steps:

Voltammetry experiments were conducted from 0.2 V to 1.6 V (vs. SCE) at 5 mV s⁻¹, with a 4-s on/off illumination pulse (2 s on/2 s off). Chronoamperometry at a constant potential of 1.2 V used an illumination pulse of 120 s (60 s on/60 s off). Light intensity was set to 30 mW cm⁻². IMPS was measured at steady state potentials: 0.6 V, 0.8 V, 1.0 V, 1.2 V (100 mW cm⁻² light intensity, 10% amplitude modulation, 10 kHz to 0.1 Hz).

3. Results and Discussion

3.1. Structure and Surface Morphology

Figure 2 shows the XRD patterns of electrodeposited WO₃ and WO₃-TiO₂ films that had been annealed at 400 °C for 2 h. The XRD pattern of the annealed stainless steel substrate is also included. It contains two peaks characteristic of austenite iron (JCPDS No. 00-023-0298) at 43.5° and 50.7°, and the substrate peaks also emerge in the diffractograms of the deposited films. The diffractograms of WO₃ and WO₃-TiO₂ reveal the radically different phase structure of the films. WO₃-TiO₂ displays all the characteristic peaks of monoclinic WO₃ [38,39]. However, there are a couple of relatively low-intensity peaks that cannot be attributed to this phase, in particular, at 27.6° and 49.5°. The analysis of WO₃ reveals the origin of these peaks. The diffractogram of WO₃ contains a very intense peak at ~ 24.1°, which may correspond to the (200) face of monoclinic WO₃. There is also a broad hump at a slightly lower 2Θ of 23.3°, and two small peaks at 27.8° and 49.5°. These reasonably correspond to the unidentified peaks from WO₃-TiO₂ and have been attributed to hydrated forms of WO₃ in the literature [40,41]. It is claimed that the formation of WO₃·xH₂O can occur through tuning of the intermediate peroxo-complex and thus is related to the H₂O₂ content in the solution [40]. The electrodeposition of a hydrated tungsten oxide has also recently been reported under conditions very similar to those used in this study [42]. Therefore, the formation of monoclinic WO₃ when electrodepositing the composite must have been related to an equilibrium shift in the intermediate peroxo-complex caused by the interaction of solution species with nanoparticulate TiO₂, possibly solvation. The crystallite sizes of WO₃, as approximated by the Scherer equation, are 41 ± 9 nm for WO₃ and 27.4 ± 3.3 nm for WO₃-TiO₂. Regarding the presence of TiO₂, in the diffractogram of the composite film, a very faint hump at ~25° can perhaps be distinguished, which could be related to the main (101) peak of anatase TiO₂.

The visual appearance and SEM surface morphology images of WO₃ and WO₃-TiO₂ films are shown in Figure 3. The WO₃ film (Figure 3a) is inhomogeneously polychromic, which is indicative of a compact thin film of slightly varying thicknesses. The surface morphology, as observed by SEM, is smooth and fine-grained, as was also observed by Pauporté when first reporting on WO₃ electrodeposition by this method [34]. For both films, signs of emerging cracks can be seen. It was noted earlier [43,44] that cracking and delamination occur if the layer is thicker (i.e., after longer deposition times). Accordingly, the WO₃-TiO₂ film is already cracked (Figure 3b), which may indicate a larger thickness. Scattered agglomerates of white particles, which are most likely TiO₂, can be observed on the surface. Visually, the film appears “cloudy” because TiO₂ particles had settled at the surface of the film during deposition. These particles were not always well-adhered to the surface. If the electrodeposition time was prolonged, they would detach from the surface when immersed in an electrolyte for further characterization. For this reason, the PEC characterization was carried out on films that had been deposited for 5 min.

The results of an EDX line scan (Figure 3c,d) show that a TiO₂ nanoparticle agglomeration was observed on the surface. Titanium was detected, albeit with a very weak signal. It is also interesting to note that the tungsten signal increases along with titanium. This may simply be a topographical effect (i.e., a stronger signal obtained from a point closer to the detector), but it could also indicate that WO₃ had electrodeposited onto TiO₂.

The near-surface electronic structure of the WO₃ and WO₃-TiO₂ films was investigated by XPS. It was found that for both films, the core-level structure of W 4f deconvolutes into a single doublet with W 4f_7/2 and W 4f_5/2 peaks at binding energies of 35.46 eV and 37.6 eV, respectively (Figure 4a,b). The position of these peaks, and the spin-orbit splitting energy of 2.14 eV, indicate the total dominance of the W⁶⁺ oxidation state [45,46]. Interestingly, the O 1s core-level spectra reveal that oxygen can be found in at least two oxidation states (Figure 4c,d). The O 1s peak seen here at 530.22 eV for WO₃ and 530.48 eV for WO₃-TiO₂ is commonly attributed to lattice oxygen (i.e., the W-O bond) [47].

The origin of the second much broader peak at ~531.4 eV is slightly more complicated to explain. First and foremost, this cannot be Ti-related O because (1) the lattice oxygen peak energy is expected to be around 529–530 eV [48] and (2) the same peak also exists for WO₃ without TiO₂. A couple of other possibilities that are often reported are oxygen from non-stoichiometric WO_3-x or -OH groups incorporated within the lattice, which can result in peaks in the O 1s spectra at approximately 532.5 eV [47,49]. However, these energies also do not match. A very recent work by T. J. Frankcombe and Y. Liu describes what is most likely the cause of these peaks [50]. They found that a peak at this energy (commonly ascribed in the literature to oxygen vacancies) is instead caused by O 1s electrons from water molecules that are chemisorbed to the surface or surface oxygen passivated by hydrogen. This agrees well with the W 4f spectra, which show 100% dominance of W⁶⁺ and no trace of reduced oxidation states that would be expected with a large number of oxygen vacancies. The existence of chemisorbed water may also be related to the observation of hydrates in XRD characterization (Figure 2). Lastly, no Ti is found on WO₃, and only a very weak Ti 2p signal is obtained from the WO₃-TiO₂ film (Figure 4e,f). The core level spectrum deconvolutes into one doublet with peaks at 459.25 eV and 465.1 eV, corresponding to Ti⁴⁺ from anatase TiO₂ [51]. An indistinct hump at 461.5 eV may also be observed. It is on the verge of being noise, but a peak at this position has been attributed to Ti³⁺ [52]. The reduction of titanium dioxide nanoparticles is feasible under cathodic conditions in an acidic electrolyte, so the signal is attributed to Ti³⁺ here as well.

3.2. Photoelectrochemical Characterization

Initially, the PEC properties of WO₃ and WO₃-TiO₂ films were characterized by carrying out a potential scan with intermittent on/off light pulses (Figure 5a). Two differences are immediately apparent: (1) the WO₃-TiO₂ film generates larger photocurrents throughout the entire potential range and (2) the photocurrent onset potential of the composite is more cathodic (~ 0.2 V vs. SCE) in comparison to just WO₃ (~ 0.35 V vs. SCE). As expected, the magnitude of the photocurrent increases with the applied potential as the space-charge layer, where the photoexcitation of charge carriers occurs and becomes wider due to the increasing strength of the applied electrical field. It is also worth noting that both WO₃ and WO₃-TiO₂ films show an appreciable amount of surface hole recombination, which is evident from the initial photocurrent overshoot that occurs at the instant light is turned on.

An additional difference between the activities of these films is noted at higher applied potentials of ≥~1 V vs. SCE, which roughly corresponds to 1.23 V vs. RHE in neutral pH (Figure 5b). In this case, particularly for WO₃-TiO₂, the electrochemical current rises sharply and is most probably attributable to the onset of electrochemical water splitting/oxygen evolution (OER). A similar effect, although with much lower current densities, is observed for the annealed stainless steel substrate. Thus, the composite films exhibit better electrochemical OER activity. This may simply be an effect of increased surface area [53] or the improved OER mechanistic kinetics of WO₃ [54].

Because the profile of the photocurrent pulse provides useful information about the material photoactivity and surface recombination [55], potentiostatic photocurrent pulses were obtained at 1.2 V vs. SCE and with a longer on/off cycle duration of 120 s (Figure 6). This made it possible to observe the initial photocurrent overshoot immediately upon turning on the light, and the subsequent decay over several pulses.

At the moment the light is turned on, the photocurrent density (j_t=0) is proportional to the maximum amount of photogenerated electron–hole pairs. For a photoanode, the electrons move to the back contact and are registered as photocurrent, and the holes move toward the electrode–electrolyte interface, where they are either lost to surface recombination or are transferred into the electrolyte to carry out an oxidative photoelectrochemical reaction. After this, the photocurrent decays until a steady state value (j_ss is assumed to be the photocurrent at the end of the 60 s pulse) is reached. In this case, it is observed that the WO₃-TiO₂ film displays slightly more relative surface recombination than WO₃. For example, for the WO₃ film, the j_ss/j_t=0 ratio is 0.8; whereas, for the comparable composite film, it is lower at 0.77. However, the WO₃-TiO₂ film also generates larger photocurrents overall (~40 µA cm⁻² compared to just under 12 µA cm⁻² for WO₃). A similar result was observed by Castro et al., who found that a hydrothermally synthesized WO₃ composite with 40 wt% of TiO₂ exhibited larger photocurrents than WO₃ [21]. Similarly, WO₃/TiO₂ heterostructures on reduced graphene oxide were shown to generate significantly larger photocurrents than the plain films alone [22].

Lastly, similar to the initial jump, there is an “overshoot” to the downside when illumination is turned off. This occurs because of the recombination of holes trapped on the surface with electrons—a phenomenon similar to the decay of the photocurrent.

3.3. IMPS Study

IMPS is a non-stationary photoelectrochemical analysis method, modeled by Ponomarev and Peter [56], to determine the features of the space charge region relative to the reaction rate constants. A sinusoidal modulation is applied to the intensity of illumination, and the working electrode under investigation is kept at potentiostatic conditions. In this case, the (photo)current response is proportional entirely to the incident photon flux, i.e., conversion efficiency. It must be noted that here, the light intensity was increased in comparison to previous experiments (30 mW cm⁻² to 100 mW cm⁻²). This is because the signal for IMPS is not obtained from the entire photocurrent, but from the light-perturbation-induced photocurrent, which corresponds to the photocurrent generated by the amplitude of light intensity modulation (10%). If this light intensity is too low (e.g., 3 mW cm⁻² as 10% of 30 mW cm⁻²), the spectra can be noisy. However, photogenerated charge carrier recombination is generally larger at higher light intensities [57], so care should be taken to find the optimal measurement conditions for a particular system.

IMPS Nyquist spectra obtained for WO₃ are shown in Figure 7, where H represents the complex transfer function of the resulting current measured by controlling the light intensity, normalized to the geometric electrode area. The first quadrant reflects the response from charge-transfer kinetic and surface recombination. The low-frequency intercept H’_LF (left side) when H’’→0 represents the differential steady-state photocurrent increase due to a differential increase in the light intensity and should be zero if no photoelectrochemical reactions occur. ω_LF, the radial frequency of the highest point of the recombination semicircle, is a product of first-order rate constants of photogenerated charge transfer and recombination and is inversely related to the recombination time constant.

Figure 7 for WO₃ shows a non-zero H’_LF, suggesting that an electrochemical reaction occurs at all applied potentials. In Figure 8 for WO₃-TiO₂ at 0.6 and 0.8 V, this intercept is nearly zero and thus, electron–hole generation and surface recombination primarily occur. A significant shift in this intercept occurs at 1 V where a significant amount of OER is expected to commence driven by the applied potential. The intercept when the imaginary part is zero at a higher frequency (right side) represents the amplitude of the flux of charge carriers to the surface. This intercept occurs at a value of H’ that can be expressed as Equation (4).

$${\mathrm{H}}^{\prime }=j_h{\displaystyle \frac{C_H}{C_H+C_{SC}}}$$

(4)

where j_h is the hole current amplitude, C_H and C_SC are the Helmholtz and space charge layer capacitances, respectively.

Since the Helmholtz layer size is typically smaller than the width of the depletion, the space charge layer, then it is expected that the Helmholtz capacitance is higher, C_H > C_SC [58]. Then, this intercept equals j_h, i.e., the hole current. At higher frequencies relative to this intercept, the influence of the capacitance of the semiconductor space charge dominates and the response moves to the lower quadrant, with a negative imaginary transfer function. The discussed parameters that had been obtained from the spectra are presented in

Table 1. Parameters, obtained from interpretation of IMPS Nyquist plots.

	WO3			WO3-TiO2
E, V	ωLF, Rad s−1	H’LF, µA cm−2 W−1 m2	jh, µA cm−2 W−1 m2	ωLF, Rad s−1	H’LF, µA cm−2 W−1 m2	jh, µA cm−2 W−1 m2
0.6	69.6	0.058	0.135	546.2	0.004	0.511
0.8	37.0	0.097	0.208	271.3	0.010	0.867
1.0	26.4	0.180	0.276	162.7	0.284	1.208
1.2	26.4	0.330	0.378	124.0	1.028	1.653

. The overall trend is that, as the applied potential is raised, H’_LF increases because more photogenerated holes are transferred as surface recombination is suppressed. Moreover, the IMPS spectra are not normalized to the amplitude of the photogenerated hole current, and therefore, larger IMPS spectra magnitudes of WO₃-TiO₂ suggest higher conversion efficiencies overall, corresponding well to LSV and potentiostatic pulse data.

In IMPS spectra, the high-frequency semicircle occurs because of the RC time constant of the electrochemical cell, i.e., the series resistance of the substrate and the total capacitance. Conversely, the low-frequency semicircle describes the recombination and charge transfer kinetics. Kinetically, these processes are described by first order rate constants: k_tr—the rate constant of charge transfer across this same interface and k_rec—the rate of electron–hole recombination at the semiconductor/electrolyte interface. The spectra shown in Figure 5 and Figure 6 exhibit some characteristic tendencies. Most notably—the radius of the low frequency semicircle decreases as the applied potential is increased. In IMPS, the value of the low frequency intercept H’_LF corresponds to the DC photocurrent, scaled by the 10% modulation, and is expressed as Equation (5).

$${\mathrm{H}}_{LF}^{\prime }=j_h{\displaystyle \frac{k_{tr}}{k_{tr}+k_{rec}}}$$

(5)

Here, k_tr/k_tr + k_rec corresponds to η_trans—the fraction of hole flux from the space charge region that is involved in the charge transfer reaction, or the transfer efficiency. The transfer efficiency increases with applied potential as photogenerated charge carriers can participate in an electrochemical reaction. Therefore, electron–hole recombination is significant at less positive potentials (the low frequency intercept value is closer to 0), because there is no electrochemical reaction, but at 1.2 V, it decreases considerably with the occurrence of OER.

Several important parameters can be calculated from IMPS spectra. Firstly, the transfer efficiency is an indicator of how many of the holes that had arrived at the semiconductor/electrolyte interface that were transferred into the electrolyte to provide the photoelectrochemical reaction. The data for WO₃ and WO₃-TiO₂ films are shown in Figure 9. As expected, the efficiencies are larger with increasing applied potential. Plain WO₃ films have larger photogenerated transfer efficiency over the entire measured potential range, while the TiO₂-containing composites have total recombination at lower potentials, but improve rapidly at higher potentials.

To fully discern reaction kinetics, the rate constants should be considered. Both k_tr and k_rec can be calculated from two characteristic points of the IMPS spectrum: the low frequency intercept H’_LF (Equation (5)) and ω_LF, the radial frequency of the highest point of the recombination semicircle, which is described by Equation (6).

$${\omega }_{LF}=k_{tr}+k_{rec}$$

(6)

The spectra were analyzed and the rate constants are presented in Figure 10 in relation to applied potential. Firstly, it can be seen that for both measured photoelectrodes, k_rec decreases with applied potential (Figure 10a), which is expected from the added potential bias as has been discussed earlier.

The k_rec values of WO₃-TiO₂ are larger, showing faster recombination kinetics. However, when considering the kinetic transfer constant, it is apparent that k_tr increases significantly for the composite films at higher potentials, compared to WO₃ (Figure 10b). Thus, there is a higher OER reaction rate on the WO₃-TiO₂ under a biased photoelectrochemical environment. At an applied potential of 1.2 V vs. SCE, the k_tr of the composite is three times larger than that of the respective WO₃ film, consistent with the voltammetry result in Figure 5 and the chronoamperometry in Figure 6.

While the addition of discrete TiO₂ nanoparticles to WO₃ generated a higher photocurrent compared to WO₃ alone, it had the negative effect of increasing the recombination rate constant, particularly near open circuit conditions at small applied potentials. The increased recombination rate may be due to the introduction of grain boundaries and local potential perturbation by the particles. The recombination rate constant decreased with applied potential, and in the applied potential range for conventional electrolysis, the WO₃-TiO₂ composite greatly exceeded the performance of the WO₃. At an even larger applied potential exceeding 1.2 V vs. SCE, evident in the polarization curve in Figure 5, the WO₃-TiO₂ composite OER current density greatly exceeded WO₃ even without light; thus, the TiO₂ addition to WO₃ improved its electrocatalytic performance. Georgieva et al. [30] also found an enhanced photocatalysis for TiO₂/WO₃ layers on stainless steel compared to WO₃ but attributed the enhancement to reduced recombination. The results presented here are in contrast to this reasoning, elucidated by the additional evidence from IMPS. The enhancement in our work is due to larger charge transfer kinetics and not suppressed recombination.

4. Conclusions

In this study, WO₃ and composite WO₃-TiO₂ films were prepared by electrodeposition from an acidic peroxo-tungstate precursor solution containing suspended anatase TiO₂ nanoparticles for the formation of the composite. The annealed films were found to be crystalline, with WO₃-TiO₂ having a distinct monoclinic WO₃ structure, but plain WO₃ displaying characteristics of hydrate formation. XPS showed the dominance of W⁶⁺ in both samples. In addition, the inclusion of Ti⁴⁺ and a small amount of Ti³⁺ was proven by deconvoluting the Ti 2p core level spectra. The films prepared in this study were found to be photoelectrochemically active, and WO₃-TiO₂ generated significantly larger photocurrent magnitudes compared to the WO₃ alone. However, the WO₃-TiO₂ composite photocurrents possessed considerable recombination at and near open circuit conditions. While the overall transfer efficiency is higher for WO₃, applying a bias potential resulted in a significant increase in the OER current density of the WO₃-TiO₂ composite: it is three times higher at an applied potential of 1.2 V vs. SCE, and an associated expected decrease in the electron–hole recombination rate. The implications of these results show that electrodeposition of a WO₃-TiO₂ composite from an electrolyte containing a peroxo-tungstate precursor and TiO₂ nanoparticles result in significant changes in the material’s structure, morphology, and consequently, PEC activity. The synthesis of composites is facile, they exhibit improved PEC performance as photoanodes when compared to WO₃, but also suffer from higher photogenerated electron–hole recombination at lower applied potentials.