Effect of Surface Passivation on Photoelectrochemical Water Splitting Performance of WO 3 Vertical Plate-Like Films

WO3 vertical plate-like arrays provide a direct pathway for charge transport, and thus hold great potential as working electrodes for photoelectrochemical (PEC) water splitting. However, surface recombination due to surface defects hinders the performance improvement. In this work, WO3 vertical plate-like arrays films with HfO2 passivation layer were fabricated via a simple dip-coating method. In the images of transmission electron microscope, a fluffy layer and some small sphere particles existed on the surface of WO3 plate. X-ray photoelectron spectroscopy (XPS) showed a higher concentration of Hf element than the result of energy-dispersive X-ray spectroscopy (EDX), which means that HfO2 is rich on the surface of WO3 plates. A higher photocurrent under visible light irradiation was gained with surface passivation. Meanwhile, the results of intensity modulated photocurrent spectrum (IMPS) and incident photon to current conversion efficiency (IPCE) indicate that HfO2 passivation layer, acting as a barrier for the interfacial recombination, is responsible for the improved photoelectrochemical performance of WO3 vertical plate-like arrays film.


Introduction
Photoelectrochemical (PEC) water splitting is a significant technology for relieving the energy crisis and environmental issues [1].In the past decades, numerous researchers have focused on discovering novel solid-state catalysts that can effectively split water using either unbiased or externally biased electrodes.Since wide-band gap semiconductors like TiO2 [2] and ZnO [3] were studied as photocatalysts for water-splitting, more recent efforts have turned to WO3 due to its bandgap of 2.6 eV [4], its photostability in acidic conditions [5] and its good hole mobility [6].
Recently, WO3 has been grown on the conductive substrate by hydrothermal technique, which has produced nanowire or nanoflake arrays by adjusting solution composition [7][8][9].The WO3 film with special oriented or one dominant facet was fabricated using different capping agents [10,11].In other works, WO3 vertical plate-like and rod-like array films have also been obtained by the hydrothermal method without seed layer [5,12].The above vertically aligned films show satisfactory photoelectrochemical properties because two-dimensional arrays can provide a direct pathway for efficient charge transport.At the same time, co-catalysts like ZnO, BiVO4 and TiO2 were loaded on the WO3 electrodes to improve photoelectrochemical performance [13][14][15].In order to enhance catalytic performance, some strategies (doped with metal or nonmetal elements [16][17][18] and modified with noble metals [19,20]) have also been reported.Furthermore, surface passivation processes were introduced to PEC, with the aim of decreasing surface charge recombination.Alumina overlayer as a passivation layer loading on the WO3 surface by atomic layer deposition, which reduced the number of electron trapping sites on the surface of WO3, eventually facilitated the photoelectron transfer to the external circuit in the presence of a positive bias [21].
HfO2 has been considered one of the most promising candidates of passivation layer in dye sensitized solar cells (DSSC) due to its wide band gap and chemical stability.Baroughi et al. found that the incorporation of HfO2 layer resulted in a 69% improvement in the efficiency of the cell [22].Kim et al. gained a similar result, and the lifetime of photoelectron increased after introducing hafnium oxide layer that could effectively retard the electron recombination process between nanocrystalline TiO2 and electrolyte [23].Ziegler et al. deposited HfO2 or TiO2 compact layer on the surfaces of ITO nanowires, and then coated porous TiO2 layer.After coating of the compact layer, the efficiency of DSSC had been improved.Even though the wide band gap of HfO2 was conducive to retarding the electron transfer from indium tin oxides (ITO) to electrolyte, the charge collection was also suppressed because the conduction band of HfO2 is more negative than that of TiO2.The photocurrent of DSSC with HfO2 compact layer was not as well as that with TiO2 compact layer [24].Hence, the high band gap materials (such as HfO2) are not preferentially recommended to deposit between conductive substrate and photoresponse material.
In our previous paper [25], HfO2 had been loaded on the surface of WO3 nanoparticles, and then prepared into film as photoanodes.Despite the fact that HfO2 has made progress in decreasing the surface defects of the WO3 nanoparticles and improving the photoelectrochemical performance, an unwilling phenomenon occurs.Because the WO3 particles were covered with HfO2 before the film was prepared, and the charge transport in WO3 nanoparticles film occurred via a random walk in a trap-limited diffusion process, a certain amount of photo-generated electrons transferred to fluorine-doped tin oxide (FTO) through HfO2 layer, and the HfO2 might have acted as a compact layer when it was inserted between the interfaces of FTO and WO3 (Scheme 1).The more negative conduction band (CB) edge position of HfO2 that forms an energy barrier would suppress electrons transport, similar to the conclusion of Ziegler et al. [24].Hence, a surfacial passivation layer without the compact layer should exhibit better photoelectrochemical performance in theory.WO3 vertical plate-like arrays provide a direct pathway for charge transport, and thus the electron transfer to FTO through the inner instead of surface of WO3 plate.However, the surface recombination due to surface defects also hinders the performance improvement, and the surface passivation is still needed (Scheme 2).In addition, the particle size of HfO2 and WO3 nanoparticles might be similar, and it is hard for us to distinguish them from too low to distinguish from each other in the previous work.Thus, the obviously different size of HfO2 and WO3 plate provided a method to confirm and demonstrate the existence of surface passivation layer.Scheme 2. Scheme of the structures of WO3 and HfO2/WO3 films.It is easy for the photogenerated electrons to recombine with photo-generated holes at the surface defect points of the WO3 plates.With the passivation layer of HfO2, the surface defect points are covered so that the recombination is inhibited at the surface of WO3 plates.
Herein, in this paper, we reported a dip-coating method to load HfO2 on the surface of WO3 vertical plate-like arrays.The HfO2 could be clearly seen in the images of scanning electron microscopy and transmission electron microscopy, which verified that HfO2 existed as overlayers on the surface of WO3 plates.Meanwhile, the influence of HfO2 passivation layer on the photoelectrochemical performance of WO3 is studied elaborately using several methods.

Characterization of the As-Prepared Films
SEM and TEM were utilized to investigate the morphology and microstructure of samples.Typical SEM images of HfO2/WO3 and WO3 are presented in Figure 1.In the top-views (Figure 1a,b), the low-magnification SEM images show that both samples exhibit plate-like structure with the plate thickness of 240-400 nm.The cross-sectional views (Figure 1c,d) show that the plate-like arrays grow perpendicular to the substrate, and the thickness of each film is about 1.16 μm.In other words, the structures are well retained after dip-coating and annealing.In the inset of Figure 1b, pure WO3 presents a smooth surface.However, some spherical particles are tightly attached to the surface of HfO2/WO3 plates (inset of Figure 1a).Correspondingly, EDX spectra of the two samples were also recorded and shown in Figure 1e,f, which show that Hf exists in the HfO2/WO3 film.More details of size, shape and structural features can be analyzed from the TEM images (Figure 2). Figure 2a displays the plate-like structure of WO3 film after the surface passivation process, and a fuzzy layer covers the surface of the plate.According to EDX analyses of the different parts, the nanoparticles at the edge of the plate (blue circle) is enriched with Hf relative to the bulk of the plate (red circle).It means that a layer of HfO2 was loaded on the surface of WO3 plate.In Figure 2b, high-resolution transmission electron microscopy (HRTEM) image exhibits that the interplanar distance of the nanoparticle is about 0.283 nm, which is in agreement with the (111) crystal plane of HfO2.Therefore, we can verify that Hf exists as HfO2, which covers the surfaces of the WO3 vertical plate-like arrays.X-ray photoelectron spectroscopy (XPS) was employed to further confirm the surface composition and chemical states of WO3 and HfO2/WO3, as shown in Figure 3. Figure 3a presents low-resolution full range XPS spectra of both samples.The total XPS spectra indicate the presence of tungsten and oxygen, and hafnium element can also be seen in the spectrum of HfO2/WO3.Meanwhile, the carbon peak (C1s) appearing at about 284 eV is due to adventitious carbon species from the XPS instrument.In Figure 3b, for pure WO3, W4f can be deconvoluted into a doublet with binding energy peaks at 35.70 eV and 37.85 eV, resulting from the emission of W4f7/2 and W4f5/2 core-levels that might belong to the W 6+ oxidation state of tungsten atoms [18].For HfO2/WO3, the characteristic peaks of W 6+ 4f7/2 and W 4f5/2 are shifted about 0.08 eV to a higher binding energy, indicating that some Hf atoms might have embeded into the WO3 lattice.In Figure 3c, for WO3, the peaks at 530.5 and 531.8 eV can be attributed to the lattice oxygen in WO3 and OH group, respectively [26].For HfO2/WO3, an additional peak at 530.7 eV is assigned to Hf-O band [27].Figure 3d shows the Hf 4f spectrum of HfO2/WO3, and two peaks at 17.39 and 18.99 eV can be ascribed to Hf 4f7/2 and Hf 4f5/2 [28].This indicates that the Hf element exists in the state of HfO2.At the same time, the ratio of [Hf]/[W] = 0.28 has also been estimated, which is higher than the result of EDX (Figure 4c).Considering the detecting depth is different between EDX and XPS, it can be concluded that most of Hf atoms were enriched on the surface of WO3 plates in the form of HfO2, which is in line with the results of TEM in the Figure 2. The corresponding XRD patterns in Figure 4 indicate that both of the samples match well with the monocline WO3 (PDF#72-0677).Moreover, the peaks at 37.77°, 51.76° and 65.74° belong to the (200), ( 211) and (301) facets of the tetragonal structure of SnO2 (PDF#46-1088), indicating that the samples are loaded on the FTO.The peaks corresponding to HfO2 cannot be found, which is originated from two reasons.One is that the concentration of HfO2 is less than the detection limit of XRD.The other is that the HfO2 in the sample are dispersed on the surface of WO3 nanoplates uniformly and sparsely, which has been confirmed by TEM.The light-absorbance properties of WO3 vertical plate-like film before and after the surface passivation were probed with UV-vis diffuse reflectance spectroscopy, as shown in Figure 5. Compared with WO3, a broad background absorption in the visible light region is observed after the surface passivation, which is similar to Kim's work [23].It can be ascribed to the fact that the surface of WO3 plate has been changed after the loading of HfO2.The absorption edge of HfO2/WO3 has no obvious shift compared with WO3, and it is 470 and 473 nm for WO3 and HfO2/WO3, respectively.Correspondingly, the band gap of WO3 and HfO2/WO3 is 2.64 and 2.62 eV, respectively, which is estimated using the following equations [18]: The slight difference of band gap between two samples is because that several Hf atoms are embedded in the WO3 lattice, which is consistent with the result of DFT calculation in the work of Valentin et al [29].

Photoelectrochemical Measurements of the Films
Figure 6a shows linear sweep voltammetry measurements to determine the photocurrent densities of WO3 and HfO2/WO3 photoanodes.The photocurrent density of WO3 increases gradually with the increase of applied potential.After loading of HfO2, the photocurrent is enhanced compared to the pristine WO3, which is similar to the alumina/Fe2O3 films [30].The photocurrent values of WO3 and HfO2/WO3 at 1.0 V (vs.Ag/AgCl) are about 0.75 and 0.90 mA/cm 2 , respectively.The stability of WO3 and HfO2/WO3 were also investigated and measured at 1.0 V (vs.Ag/AgCl).In Figure 6b, the photocurrent value rapidly decreases to about zero as soon as the light is chopped.It means that there is no other reaction occurred under dark condition.At the same time, the photocurrent values of WO3 and HfO2/WO3 decrease 29.0% (from 0.81 to 0.57 mA/cm 2 ) and 24.4% (from 0.94 to 0.71 mA/cm 2 ), respectively, after irradiation of 1180 s.This indicates that the HfO2/WO3 film shows better stability compared to the bare WO3 film.The photoelectrochemical performances of both photoanodes are also estimated by the light energy to chemical energy efficiency, which is calculated with the following equation [31,32].
In this equation, ji, Ere°, Eapp and Io are the photocurrent density, the standard reversible potential, the applied potential of the working electrode and power density of the incident light, respectively.As illustrated in Figure 6c, the maximum photoconversion efficiency of HfO2/WO3 (0.30%) is greater than that of WO3 (0.25%).Electrochemical impedance spectroscopy (EIS) analysis has been regarded as a useful tool for investigating the kinetics of photoelectrochemical processes.To better understand the dynamics of electron transport and reveal the difference in the interfacial characteristics of the WO3 vertical plate-like film with and without the surface passivation layer, the EIS of photoelectrodes were measured, as shown in Figure 7.In each Nyquist plot (Figure 7a), the semicircle denotes the charge transfer at the photoanode/electrolyte interface, which can be fitted by using an equivalent circuit containing a series resistance (Rs), a charge transfer resistance (Rct) and a constant phase element (CPE) (inset of Figure 7a).The results of the EIS analysis show that a smaller value of Rct is seen for HfO2/WO3 (433.2 ohm) compared with WO3 (584.2 ohm), which implies that HfO2 layer promotes more efficient charge transfer at the photoanode/electrolyte interface.The bode phase plots of EIS spectra (Figure 7b) display the frequency peaks of the charge transfer process at the interface of each photoanode.The maximum oscillation frequency (fmax) of the impedance semicircle of HfO2/WO3 is less than that of WO3.Considering electron lifetime (τe) is inversely proportional to fmax, the corresponding lifetime of photoelectrons in HfO2/WO3 increases compared to WO3.According to the equation [33]: The values of τe are 3.8 and 6.5 ms for WO3 and HfO2/WO3, respectively.The above discussion indicates that HfO2 coverage acts as a barrier for the recombination of photo-generated electrons and holes at the photoanode/electrolyte interface, decreasing the recombination rate of electron-hole pairs.Moreover, the CB of HfO2 is less than that of WO3, the HfO2 layer on the surface of WO3 plates as an electron-blocking layer may hinder the photo-generated electrons from transferring to the interface of photoanode/electrolyte.The effect might be non-significant because the plate-like arrays with good charge transport ability and positive bias for photoanode have good effect on transferring the photoelectrons to FTO and external circuit.Furthermore, the valence band of HfO2 (3.0-3.2 eV) [34,35] is slightly less than that of WO3 (3.0-3.35eV) [36,37].It is hard to say that the small difference of valence band has obvious effect on the transferring hole from WO3 to HfO2, but we believe that the negative effect would not occur in this system.Anyhow, these facilitate the charge transfer efficiently so as to enhance the photoelectrochemical performance.
Intensity modulated photocurrent spectrum (IMPS) data were recorded and used to investigate electron transport.Figure 8 shows the complex plane plots of the IMPS responses for both samples.The response appears in the fourth quadrant of the complex plane and displays one semicircle, where the frequency at the imaginary minimum of the semicircle can be used to calculate the time constant of the charge transfer process.The electron transport time (τtr) can be estimated from the equation [14]: Figure 8. Intensity-modulated photocurrent spectroscopies of WO3 and HfO2/WO3.
The electron transport times calculated for WO3 and HfO2/WO3 are 0.96 and 1.20 ms, respectively.The longer electron transport time might be caused by some Hf atoms occupying into the WO3 lattice, which affect the crystallinity of WO3 plates so that the charge transport ability of HfO2/WO3 is not as good as that of WO3.Even though a long transport time provides no benefit to the photoelectrochemical properties, the photocurrent of HfO2/WO3 is still higher than that of WO3.Hence, the passivation layer, which acts as a barrier for the interfacial recombination, is the main reason of the improved photoelectrochemical performance.
Incident photon-to current efficiency (IPCE), which reflects the number of electrons in the external circuit produced by an incident photon at a given wavelength divided by the number of incident photons, is an effective tool to investigate the quantitative correlation of light absorption on the WO3 vertical plate-like film with and without the surface passivation layer.Figure 9 shows the plots of both samples at a bias of 1.0 V.The HfO2/WO3 exhibits a higher IPCE value compared to WO3, in agreement with the results of linear sweep photovoltammetry measurements (Figure 6a).Both samples display a strong photon response in the region of 310-460 nm, and the maximum values are 30.92%and 36.54% for WO3 and HfO2/WO3, respectively.Based on the above results and the enhancement of IPCE existing in the whole photoresponse region, we can conclude that the HfO2 passivation layer, which inhibits the recombination of electrons and holes, is the major reason for enhancing photoelectrochemical performance of WO3 vertical plate-like arrays.

Sample Preparation
All chemicals were analytical grade and used without further purification.WO3 vertical array film was fabricated via a facile hydrothermal method [38].To synthesize the WO3 vertical array film with a HfO2 surface passivation layer (HfO2/WO3), as-prepared WO3 film was immersed in a 40 mL concentrated sulfuric acid solution that 30 mg HfO2 had been dissolved in.After dipping of 2 min, the substrate with WO3 film was taken out and immersed in deionized water immediately.Then, the film was dried naturally, and the two-step dipping procedure was repeated again.Finally, the prepared sample was obtained after being dried in an oven at 60 °C and annealed at 450 °C for 30 min.For comparison, the bare WO3 film was also treated in a similar method without the addition of HfO2.

Structure Characterization
The structure and morphology of prepared samples were characterized with a field emission scanning electron microscope (FESEM, JSM-7600F, JEOL Company, Tokyo, Japan) and a transmission electron microscope (TEM, TECNAI G2 F20, FEI, Japan) equipped with an energy-dispersive X-ray (EDX) spectrometer.The crystal phases were identified by X-ray diffraction (XRD, D/Max2250, Rigaku Corporation, Japan) with Cu Kα radiation.X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi, Thermo Fisher-VG Scientific) was used to determine the chemical composition and the concentration of the atoms.UV-visible (UV-vis) absorption spectra in the range of 400-600 nm were detected by a diffused reflectance ultraviolet and visible spectrophotometer (UV-vis, PGeneral TU-1901, Beijing, China).

Photoelectrochemical Measurements
A standard three-electrode system was used to evaluate the photoelectrochemical properties with an electrochemical analyzer (Zennium, Zahner, Germany).The sample served as the photoanode, a Pt foil as the counter electrode and an Ag/AgCl (saturated KCl) as the reference electrode.An aqueous solution of 0.2 M Na2SO4 was used as the electrolyte solution.Linear sweep voltammetry were measured in a potential range from 0 V to 1.4 V (vs.Ag/AgCl) under AM 1.5G illumination.The electrochemical impedance spectra (EIS) were measured at 0.7 V (vs.Ag/AgCl) with perturbation amplitude of 10 mV, frequency of 0.1 Hz~10 kHz.Mott-Schottky plots were monitored at the AC frequency of 1 kHz, and the potential ranged from 0 to 1.0 V. Intensity modulated photocurrent spectrum (IMPS) data were recorded using a Zahner CIMPS-2 system (Zahner CIMPS-2, Zahner, Germany).Meanwhile, the incident photon to current conversion efficiency (IPCE) was carried out as a function of wavelength from 310 to 700 nm with Si solar cell as standard.

Conclusions
In summary, we report a simple dip-coating and annealing method to synthesize HfO2 passivation layer loaded on the surface of WO3 vertical plate-like arrays.The differences in morphology between WO3 and HfO2/WO3 can be clearly seen in the SEM and TEM measurements, and the HfO2 nanoparticles can be distinguished from the WO3 plates due to different sizes and interplanar distances.The different concentrations of Hf detected by EDX and XPS indicate that HfO2 is rich on the surface of WO3 plates.HfO2/WO3 photoanode exhibits a higher photocurrent under visible light irradiation compared to WO3 photoanode.At the same time, the electron transport time has been slightly increased, which suggests charge transport ability has not been modified.Hence, the improvement of IPCE confirms that HfO2 passivation layer, acting as a barrier for the interfacial recombination, is in favor of the improvement the photoelectrochemical performance of WO3 vertical plate-like arrays film.

Scheme 1 .
Scheme 1. Scheme of electron transport processes in the different section of photoanode.(a) WO3 nanoparticle film coating of HfO2 passivation layer; (b) photo-generated electrons transfer from WO3 to FTO directly; (c) photo-generated electrons transfer from WO3 to FTO through HfO2 layer; (d) a structure of FTO/WO3/HfO2; and (e) a compact HfO2 layer inserted between FTO and WO3.In the diagram, the arrows represent electron transfer from WO3 to FTO, and the width of arrow represents the ability of electron transport.

Figure 2 .
Figure 2. (a) Transmission electron micrographs and (b) high resolution TEM (HRTEM) image of WO3 vertical plate-like film after the surface passivation; and EDX spectra corresponding to different sections in the inset of (a): (c) blue circle and (d) red circle.

Figure 5 .
Figure 5. UV-vis absorbance spectra of WO3 vertical plate-like film with and without the surface passivation layer.

Figure 6 .
Figure 6.(a) Linear sweep voltammetry curves of WO3 vertical plate-like film with and without the surface passivation layer under light illumination; (b) photocurrent-time plot of the samples under illumination; and (c) Photoconversion efficiency of the as-prepared samples.

Figure 7 .
Figure 7. EIS plots of the WO3 vertical plate-like film before and after the surface passivation: (a) Nyquist plots and (b) Bode plots; equivalent circuit in the inset of (a).

Figure 9 .
Figure 9. Incident photo to current conversion efficiency of WO3 vertical plate-like film with and without the surface passivation layer.