Introduction

The limited supply of energy and global climate change caused by burning fossil fuels are two serious challenges faced by humans in the future. Photoelectrochemical (PEC) and photocatalytic (PC) water splitting could be two potential approaches to counter these challenges because solar energy is a clean and inexhaustible energy source1,2. The PEC water splitting for hydrogen production has attracted extensive attention3,4, since the first report by Fujishima and Honda in 1972 using TiO2 semiconductor material as photoanode5. Although many semiconductor materials, such as TiO26,7, WO38,9, ZnO10,11, Fe2O312,13,14, CdS15,16, BiVO417,18 and so on, can be used as photoelectrode and show PEC activity, most of them have limited utility because of the high charge carrier recombination and large band gap, leading to the low efficiency for PEC water splitting. When a large band gap semiconductor with great surface area is coupled with a small band gap semiconductor with a more negative conduction band (CB) level, the photogenerated electrons in the CB can migrate from the small band gap semiconductor into the large band gap semiconductor and the photogenerated holes in the valence band (VB) move in the opposite direction19. Such the heterojunction structure can not only combine with the advantage of two semiconductors but also improve the separation and transport of photogenerated charges19,20,21. The heterojunction films, such as CdS/TiO222,23,24,25,26, Bi2S3/TiO227,28, CdS/ZnO29,30,31 and CdS(CdSe)/ZnO15,32,33 show highly efficient PEC hydrogen generation in the electrolyte containing sulphur ion, due to the expanded the spectral response and efficient carrier separation and transport in the heterojunction.

As a well-known photocatalyst, tungsten trioxide (WO3) with a large band gap (2.6 ~ 3.0 eV) has been extensively investigated, because of its excellent photocatalytic activity, high electron mobility and nontoxic nature8,34,35. More recently, the two-dimensional (2D) WO3 platelike films show an excellent PEC performance and have attracted intensive attention36,37,38,39. Compared to the film comprised of nanoparticles, the 2D platelike films can offer a direct electrical pathway for charge transport, resulting in an enhanced conductivity. It can effectively suppress the recombination of photogenerated electrons and holes in the transfer process40,41,42,43.

Bismuth trisulfide (Bi2S3) is a small band gap semiconductor (~1.3 eV) and has attracted great interest as a sensitizer for PEC and photovoltaic cells recently due to its large absorption coefficient and reasonable IPCE27,44,45,46. Moreover, the CB of Bi2S3 is more negative than that of WO347,48. It is reasonable to construct the Bi2S3/WO3 heterojunction photoelectrode. Recently, we have reported on the synthesis of Bi2S3/WO3 photoelectrode by a sequential ionic layer adsorption reaction (SILAR) process and the photoelectrodes exhibit an excellent PEC activity49. However, it may be weak to the interfacial contact of two components in the heterojunction film prepared by using the SILAR method. Because the SILAR method does not contain calcination process but only the wet-chemical deposition process. The weak interfacial contact will lead to the high resistance in the heterojunction interface. The in-situ growth method is considered as a good way to prepare ideal heterojunction with perfect interfacial contact50,51,52. In our study, we found that Bi2WO6 synthetized from WO3 can be used as interim product and Bi2S3 can be obtained from Bi2WO6 by a hydrothermal process. Thus, it is feasible to the formation of Bi2S3/WO3 heterojunction by in-situ growth method.

Herein, we have rationally designed and developed an in-situ growth method using Bi2WO6 as the interim product to synthesize Bi2S3 sensitized WO3 nanoplate arrays film. The as-prepared films as photoelectrodes show higher PEC activity than that of the Bi2S3/WO3 prepared by SILAR method. This may be because the heterojunction interface with less interfacial defects can be form by in-situ growth method, leading to a lower resistance and higher electron transit rate in the heterojunction interface.

Experiment Section

Preparation of WO3 platelike films

All chemicals were analytical grade. WO3 platelike films were prepared by hydrothermal method according to our previous work39. In a typical experiment, 0.231 g of sodium tungsten dehydrate (Na2WO4·2H2O) dissolved in 30 mL of deionized water at room temperature. Then, 10 mL of 3 M HCl was added to the solution under constant stirring, followed by the addition of 0.2 g of ammonium oxalate ((NH4)2C2O4). After several minutes of stirring, 30 mL of deionized water was added into it with continual stirring for 0.5 h. The as-prepared precursor was transferred into a 100 mL of Teflon-lined stainless autoclave. The FTO substrates with the conducting side facing down were immersed and leaned against the wall of the Teflon-vessel. The hydrothermal synthesis was carried out at 140 °C for 3 h. The as-prepared films were calcined at 500 °C for 1 h.

Preparation of Bi2WO6/WO3 film

The Bi2WO6/WO3 films were prepared through a simple soaking process. The prepared WO3 films were soaked in 0.2 M glacial acetic acid solution of Bi(NO3)3·5H2O for 12 h. To ensure the same thickness Bi2WO6 layer for all samples, the WO3 film was pulled from the solution with a pulling rate of 3 mm/s, not rinsing. Next, the films were dried in 25 °C for 1 h and then calcined at 520 °C in air for 4 h.

Preparation of Bi2S3/WO3 film

The Bi2S3/WO3 films could be obtained from hydrothermal Bi2WO6/WO3 film. 0.1 g thiourea as the sulphur source was dissolved in 50 mL of DI water at room temperature. Then, 100 μL of 2 M HCl was added to the solution under constant stirring. Next, the solution was transferred into an 80 mL of Teflon-lined stainless autoclave. The Bi2WO6/WO3 film was placed into the autoclave. The hydrothermal synthesis was carried out at 140 °C for 4 h. The as-prepared films were dried at 200 °C for 2 h. For comparison, we also prepared another kind of Bi2S3/WO3 and Bi2S3 films by the SILAR method according to our previous work49. Bi2S3/WO3 films prepared by the SILAR method have been annealed at low-temperature (160 °C, 2 h) after the Bi2S3 nanoparticles deposition.

Characterization and photoelectrochemical measurements

The crystalline phase of the sample was characterized by X-ray powder diffraction (XRD, Rigaku D/Max2500, Japan). The UV-Vis diffuse reflection spectra of WO3 and Bi2WO6/WO3 film and the Vis-NIR diffuse reflection spectra of Bi2S3/WO3 film were obtained using the UV-Vis 2450 (Shimadzu) and Vis-NIR U-4100 (Hitachi) spectrophotometer, respectively. The diffuse reflection spectra could convert into the absorption spectra by Kubelka-Munk function. Scanning electron microscope (SEM, Nova NanoSEM 230) was used to observe the surface morphology of samples. Transmission electron microscopy (TEM, TECNAI G2 F20, FEI) was operated at an accelerating voltage 200 kV to investigate the microstructure and the crystallinity of samples. On account of the nonuniformity of Bi2S3 thickness in s-Bi2S3/WO3 film (Supporting Information Fig. S1), the amount of Bi2S3 in the films was evaluated by the inductively coupled plasma massspectrometry (ICP-MS). The samples with 2 cm2 were soaked in 5 ml of 1 M HNO3 solution for 12 h and the Bi2S3 would be dissolved completely. The amount of Bi2S3 in the film was obtained to characterize the concentration of Bi3+ using ICP-MS. The concentration for Bi2S3/WO3 films prepared via an in situ approach and the SILAR method is 12.26 and 10.78 mg L−1, respectively. The PEC properties of samples were investigated in a typical three-electrode electrochemical cell using an electrochemical analyzer (Zennium, Zahner, Germany). The synthesized films were employed as the working electrode and a platinum foil and an Ag/AgCl/satd. KCl electrode were employed as the counter and reference electrodes, respectively. All electrochemical tests were conducted in an aqueous solution containing 0.1 M Na2S and 0.1 M Na2SO3 (pH ≈ 9). The illumination source was a 150 W xenon lamp (CHF-XM35, Beijing Trusttech Co. Ltd) with a 400 nm cutoff filter to remove UV irradiation. IPCE measurements were carried out using a xenon lamp (150 W, Oriel) with an AM 1.5 filter and a monochromator with a bandwidth of 5 nm. Intensity modulated photocurrent spectroscopy (IMPS) were recorded by using a Zahner CIMPS-2 system. A white light lamp emitting diode (λ = 540 nm) driven by a PP210 was used as lamp and the light intensity is adjusted to 50 mW cm−2. Light intensity modulation was driven by current modulation with a depth of 10%. The bias potential is at −0.4 V vs. Ag/AgCl during the IMPS measurement.

Results and Discussion

The depiction of the fabrication process for Bi2S3/WO3 films by in-situ growth method and surface morphology change of sample for each step are shown in Figs 1 and 2, respectively. First, the platelike WO3 arrays films were obtained via a hydrothermal process. The SEM images of platelike WO3 film in Fig. 2a,b reveal that the highly dense and uniform vertical nanoplates with edge length of 0.5–1.5 μm and thickness of 50–200 nm grown on a FTO substrate. Second, the Bi2WO6 was formed on WO3 nanoplate by a simple soaking and calcined process. The SEM images of Bi2WO6/WO3 film shown in Fig. 2c,d. As shown in Fig. 2a–d, the surface morphology of Bi2WO6/WO3 is different from that of the pristine WO3. Compared with the pristine WO3, the surface of Bi2WO6/WO3 nanoplates became rough and the surface textures reduce obviously. It may because the formation of the bigger molecule Bi2WO6 lead to the surface swell. Third, Bi2S3/WO3 films can be obtained from hydrothermal Bi2WO6/WO3 film. During the hydrothermal process, the S2− can be generated by the decomposition of thiourea and then the Bi2WO6 will react with the S2− to form the Bi2S3 on the surface of WO3 nanoplates53. It can be seen from Fig. 2e,f that the surface of Bi2S3/WO3 nanoplates is rougher than that of Bi2WO6/WO3.

Figure 1
figure 1

Depiction of the synthesis process of Bi2S3/WO3 films.

Full size image
Figure 2
figure 2

SEM images of (a,b) the WO3, (c,d) Bi2WO6/WO3 and (e,f) Bi2S3/WO3 film.

Full size image

Further, the transmission electron microscopy (TEM) was used to identify the elaborate structure of the composite nanoplates. Typical WO3 nanoplate is clearly shown in the low-resolution TEM image (Fig. S2). The uniform lattice fringe can be observed over an entire primary nanoplate from the high resolution TEM (HR-TEM) image of WO3 (Fig. S2b), revealing a single crystalline characteristic. The distance between each fringe is about 0.308 nm, matching well with the (112) of monocline phase WO3 (JCPDS 83-0950). The HR-TEM image recorded on the rim of a Bi2WO6/WO3 nanoplate was shown in Fig. S2c. The lattice spacing of 0.375 nm is consistent with the interplanar spacings of (111) planes of orthorhombic Bi2WO6. In addition, the lattice fringes with the spacing of 0.384 and 0.335 nm are in good agreement with the (002) and (120) planes of WO3, respectively. The TEM results for the Bi2WO6/WO3 nanoplate can confirm that Bi2WO6 layer can be formed on the surface of WO3 plate via an in situ approach. The HR-TEM image of the interface between Bi2S3 and WO3 is shown in Fig. 3b. The exact interface between the two phases is clearly shown in the image. The lattice spacing observed to be 0.384 corresponds to (002) plane of monocline phase WO3 (JCPDS 83–0950) and the lattice spacing of 0.312 nm is assigned to the (121) plane of Bi2S3 (JCPDS No. 84–0279). The two corresponding fast Fourier transform patterns (inset in Fig. 3b) confirm the single-crystal structure of the WO3 nanoplate and Bi2S3 layer. Especially, misfit dislocations or defect region are not seen near the physical interface, indicating the formation of an interface with less interfacial defects between Bi2S3 and WO3. Spatial elemental mapping was also performed on a single Bi2S3/WO3 plate to reveal the distribution of the two phases in the heterostructure. The mapping results confirm that the Bi2S3 is uniformly coated on WO3 plate to form core/shell structure (Fig. 4a–d).

Figure 3
figure 3

(a) TEM image of a Bi2S3/WO3 plate, (b) HRTEM image of the interface between Bi2S3 and WO3.

Full size image
Figure 4
figure 4

(a) Scanning transmission electron microscopy (STEM) image of Bi2S3/WO3 plate and (b–d) the corresponding elemental mapping images.

Full size image

The XRD was employed to examine the crystal structures and phase purity of the sample films. The XRD patterns of pristine WO3, Bi2WO6/WO3 and Bi2S3/WO3 films are shown in Fig. 5a. The peaks of pristine WO3 can be indexed to the diffractions from the monocline phase WO3 (JCPDS 83–0950). The three small diffraction peaks of the Bi2WO6/WO3 sample can be observed at 28.38, 33.05 and 47.16o, attributed to the characteristic diffraction peak of Bi2WO6 (JCPDS No. 73–2020). In addition, many weak diffraction peaks observed from the Bi2S3/WO3 sample, are identical to Bi2S3 (JCPDS No. 84–0279). The result indicates the formation of Bi2S3 on the surface of WO3 plate. The optical behavior of the prepared films was evaluated by using UV-vis and Vis-NIR absorption spectroscopy. Figure 6 shows the absorption spectra and Tauc plots of the WO3, Bi2WO6/WO3 and Bi2S3/WO3 film. For the pristine WO3, a clear absorption edge about 460 nm can be observed, corresponding to its indirect band gap energy (Fig. 6a). There is a weak red shift in the absorption spectra after the in situ growth of Bi2WO6 on the surface of WO3 plates. Differently, the Bi2S3/WO3 film shows strong absorption intensity at wavelength of ~950 nm (Fig. 6b), which is consistent with the reported absorption spectra of the Bi2S3/WO3 film49. The photograph of sample also supports the above results. As shown in Fig. 6 (inset), the color of the Bi2WO6/WO3 film almost is the same with the pristine WO3 film (Fig. 6a inset), while the Bi2S3/WO3 film shows black (Fig. 6b inset). In addition, the optical bandgap of samples have been calculated by the Tauc equation. As shown in the Fig. 6c,d, the bandgap energy of WO3, Bi2WO6/WO3 and Bi2S3/WO3 film is 2.60, 2.65 and 1.35 eV, respectively.

Figure 5
figure 5

XRD patterns the WO3, Bi2WO6/WO3 and Bi2S3/WO3 film.

Full size image
Figure 6
figure 6

(a,b) UV−vis and Vis-NIR absorption spectra of the WO3, Bi2WO6/WO3 and Bi2S3/WO3 films. (c,d) Tauc plots of the WO3, Bi2WO6/WO3 and Bi2S3/WO3 films.

Full size image

The PEC measurements were implemented in a three-electrode PEC cell using the prepared films as photoanodes under illumination of visible light. Linear sweep voltammetry (LSV) was employed to evaluate the PEC performance of films. For comparison, the Bi2S3/WO3 films prepared by the SILAR method also were used as photoanodes and denoted as s-Bi2S3/WO3 photoelectrodes. Likewise, the Bi2S3/WO3 prepared films by in-situ growth method were denoted as i-Bi2S3/WO3 photoelectrodes. Their PEC performances were recorded in the same condition. The LSV of photoelectrodes obtained in the dark and under illumination is shown in Fig. 7a. Under light illumination, the photocurrent density of the pristine WO3 photoelectrodes is negligible, while the photoelectrodes sensitized by Bi2S3 show significant photocurrent generation. This is because of the poor visible-light response for WO3. Compared to the s-Bi2S3/WO3 photoelectrodes (3.9 mA cm−2 at −0.1 V vs Ag/AgCl), the i-Bi2S3/WO3 photoelectrodes show a higher photocurrent density (8.0 mA cm−2 at −0.1 V vs Ag/AgCl). In addition, the stability of i-Bi2S3/WO3 photoelectrode is tested by performing long-duration PEC experiments. The testing experiment is carried out at the −0.1 V vs Ag/AgCl under continuous illumination and lots of H2 bubbles can be found on the surface of Pt electrode throughout the entire test. The result is shown in Fig. 7b. The photocurrent decreased by 23% after a 3600 s operation and could remain stable in the latter time period.

Figure 7
figure 7

(a) LSV scans of s-Bi2S3/WO3 and i-Bi2S3/WO3 films and (b) the photocurrent−time plot of i-Bi2S3/WO3 photoelectrodes.

Full size image

To investigate the quantitative correlation between the wavelength of the incident light and the PEC activity, incident photon-to-current conversion efficiency (IPCE) measurement was performed at a bias of −0.5 V vs. Ag/AgCl. The IPCE were estimated by the following relation15,17:

Where I is the photocurrent density, λ is the incident light wavelength and Jlight is the incident light power density. The IPCE of the WO3, s-Bi2S3/WO3 and i-Bi2S3/WO3 photoelectrodes are shown in Fig. 8a. The pristine WO3 photoelectrode exhibits the photoresponse only at the wavelength range of ~460 nm, because of the large band gap of WO3 (≥2.6 eV). Relative to the pristine WO3 photoelectrode, the photoelectrode decorated by Bi2S3 shows enhanced IPCE in the entire testing wavelength region due to the increased absorption by the Bi2S3. Importantly, the i-Bi2S3/WO3 photoelectrode shows the highest IPCE value, which is in accordance with the LSV result.

Figure 8
figure 8

(a) IPCE spectra and (b) complex plane plot of the IMPS response for s-Bi2S3/WO3 and i-Bi2S3/WO3 photoelectrodes.

Full size image

In order to understand the reason that the i-Bi2S3/WO3 photoelectrode has an enhanced PEC property, the electron transport properties of the photoelectrodes were investigated by using intensity modulated photocurrent spectroscopy (IMPS) analysis. Figure 8b shows the complex plane plot of the IMPS response for s-Bi2S3/WO3 and i-Bi2S3/WO3 photoelectrodes at −0.4 V vs. Ag/AgCl. The IMPS is popularly used to characterize the electron transport of PEC cells54,55. The electron transport time (τd) is the average time photogenerated electrons need to reach the back contact and can be obtained from the IMPS result by the following formula19,28,56,57:

where fmin is the frequency at the imaginary minimum. τd of the s-Bi2S3/WO3 and i-Bi2S3/WO3 photoelectrode are 15.87 and 3.80 ms, respectively. The results suggest the electron transit rate in i-Bi2S3/WO3 photoelectrode is faster than in the s-Bi2S3/WO3 photoelectrode. It may be attributed to the stronger contact and lower resistance in the interface between Bi2S3 and WO3 for the i-Bi2S3/WO3 photoelectrode. To further verify the IMPS results, we also measured IMPS at −0.6 and −0.2 V vs. Ag/AgCl and the results are shown in Fig. S3. Similarly, the τd of i-Bi2S3/WO3 photoelectrodes are smaller than that of s-Bi2S3/WO3 photoelectrodes at the same potential due to the higher fmin.

To further support the above ideas, the transient photocurrent plots were measured at −0.7 V and −0.5 V vs. Ag/AgCl and were shown in Fig. 8. The average transient time constant (τt) of photoelectrodes can be calculated and obtained from the transient photocurrent plots by using the kinetic equations as follows58,59,60,61:

Where D is defined as

Where I is the current, i and f are related to the initial and final steady states and t denotes time, respectively. The transient time constant, τt, can be defined as the time at ln D = −1. Generally, a longer transient time constant implies a smaller extent of recombination and the transient time may be treated as a lifetime of the photogenerated carriers. As shown in Fig. 9, the photocurrents were approximately zero in the dark. Once illumination, the photocurrents rapidly increase and then begin to decay. The decay of the photocurrents implies that the recombination of the photogenerated carriers occurs. The τt values of the s-Bi2S3/WO3 and i-Bi2S3/WO3 photoelectrode are shown in Table 1. The i-Bi2S3/WO3 photoelectrode has longer the transient time than that of the s-Bi2S3/WO3 photoelectrode, suggesting a smaller extent of recombination for the i-Bi2S3/WO3 photoelectrode. This may be because of the faster carries transit rate for the i-Bi2S3/WO3 photoelectrode, resulting in the reducing of the recombination of the photogenerated carriers.

Table 1 The average transient time constant of s-Bi2S3/WO3 (Ï„ s) and i-Bi2S3/WO3 (Ï„ i) photoelectrodes.
Full size table
Figure 9
figure 9

The transient photocurrent plots of (a,b) s-Bi2S3/WO3 and (c,d) i-Bi2S3/WO3 photoelectrodes at −0.7 and −0.5 V vs. Ag/AgCl, respectively.

Full size image

To better understand the charge separation and transfer process in the PEC cells, the schematic illustration of the charge transfer of the photoelectrode is illustrated in Fig. 10. Under visible light irradiation, the electrons are excited from the valence band (VB) of WO3 and Bi2S3 to their conduction band (CB). The accumulated electrons at the CB of Bi2S3 easily move to the CB of WO3 due to the more negative CB of Bi2S3 and then are collected by the FTO conductor49. It can also be supported from the analysis of Mott−Schottky results (Supporting Information Fig. S4). At the same time, the holes formed in the valence band (VB) of WO3 and Bi2S3 will be transferred to the semiconductor/electrolyte interface in the opposite direction and react with S2− to avoid the photocorrosion of Bi2S3. The electrons in the FTO are migrated to the Pt electrode/electrolyte interface by the external bias voltage and will reduce the H2O to H2. It should be noted that the PEC cells discussed here need the presence of Na2SO3 and Na2S as the sacrificial reductants to steadily generate H2, which has been widely reported15,24,31,33,62. The addition of the reductant will not affect the comparison of the PEC performance of photoelectrodes, but the development of efficient PEC cell with no need of sacrificial reductants is important and urgent for PEC water splitting and yet still very challenging.

Figure 10
figure 10

The schematic illustration of the charge transfer of the photoelectrode.

Full size image

Conclusions

In conclusion, we have synthesized Bi2S3 sensitized WO3 nanoplate arrays films via an in situ approach. The film characterization results suggest that the Bi2S3 layer was uniformly formed on the surface of WO3 nanoplates. The prepared films were used as photoanodes and the PEC performance was studied. The Bi2S3/WO3 films prepared via an in situ approach have a higher photocurrent density (8.0 mA cm−2 at −0.1 V vs Ag/AgCl) than that of the Bi2S3/WO3 films prepared by SILAR method. This may be due to the stronger contact interface for the Bi2S3/WO3 films prepared via an in situ approach, resulting in a higher electron transit rate and the reduced photogenerated carrier recombination. This versatile preparation method has potential to be applied in the synthesis of other hybrid films with less interfacial defects.

Additional Information

How to cite this article: Liu, C. et al. In situ synthesis of Bi2S3 sensitized WO3 nanoplate arrays with less interfacial defects and enhanced photoelectrochemical performance. Sci. Rep. 6, 23451; doi: 10.1038/srep23451 (2016).