Fabrication of an Efficient N, S Co-Doped WO3 Operated in Wide-Range of Visible-Light for Photoelectrochemical Water Oxidation

In this work, a highly efficient wide-visible-light-driven photoanode, namely, nitrogen and sulfur co-doped tungsten trioxide (S-N-WO3), was synthesized using tungstic acid (H2WO4) as W source and ammonium sulfide ((NH4)2S), which functioned simultaneously as a sulfur source and as a nitrogen source for the co-doping of nitrogen and sulfur. The EDS and XPS results indicated that the controllable formation of either N-doped WO3 (N-WO3) or S-N-WO3 by changing the nW:n(NH4)2S ratio below or above 1:5. Both N and S contents increased when increasing the nW:n(NH4)2S ratio from 1:0 to 1:15 and thereafter decreased up to 1:25. The UV-visible diffuse reflectance spectra (DRS) of S-N-WO3 exhibited a significant redshift of the absorption edge with new shoulders appearing at 470–650 nm, which became more intense as the nW:n(NH4)2S ratio increased from 1:5 and then decreased up to 1:25, with the maximum at 1:15. The values of nW:n(NH4)2S ratio dependence is consistent with the cases of the S and N contents. This suggests that S and N co-doped into the WO3 lattice are responsible for the considerable redshift in the absorption edge, with a new shoulder appearing at 470–650 nm owing to the intrabandgap formation above the valence band (VB) edge and a dopant energy level below the conduction band (CB) of WO3. Therefore, benefiting from the S and N co-doping, the S-N-WO3 photoanode generated a photoanodic current under visible light irradiation below 580 nm due to the photoelectrochemical (PEC) water oxidation, compared with pure WO3 doing so below 470 nm.


Introduction
The development and utilization of hydrogen energy is considered to be one of the significant ways to resolve the energy crisis and environmental pollution [1][2][3].
At present, there are many strategies to produce hydrogen by solar energy, including electrolytic and solar thermal water splitting, PEC water splitting, and so on [4]. Among them, PEC water splitting could directly convert abundant solar energy into clean hydrogen energy. Therefore, it is regarded as one of promising ways and has attracted considerable attention since the TiO 2 photoanode was first reported by Honda and Fujishima [5][6][7][8][9][10]. However, the half-reaction of PEC water oxidation on photoanode is considered to be a key process to affect the efficiency of fuel generation due to the difficult kinetic nature. Moreover, the bandgap of TiO 2 is too wide (3.0-3.2 eV) to respond to the visible light of sun spectrum, being consequently responsible for low efficiency in the utilization of solar light.
So, it is of great importance to develop a stable and robust semiconductor photoanode with narrow bandgaps to enhance the absorption of solar light.
Since WO 3 was reported as a PEC photoanode by Hodes in 1976 [27], it has attracted immense attention because of its visible light response (bandgap, E g = 2.6-2.8 eV), strong absorption within the solar spectrum and good photochemical stability under acidic conditions. However, as the WO 3 photoanode cannot respond to visible light above 460 nm, its solar spectrum utilization is still low. Taking this disadvantage into account, enhancing the light absorption at longer wavelengths is the key to improving the solar energy conversion efficiency of the WO 3 photoanode. Therefore, extension of light absorption to longer wavelengths by bandgap engineering of WO 3 is an important and interesting research subject in the related field.
In recent years, the research mostly focused on single doping WO 3 with selective nonmetallic elements (C, N, S) [37][38][39][40], as well as molecules (N 2 , Xe and CO) [41][42][43] to enhance the light absorption. However, attention has scarcely been focused on the multielement co-doped WO 3 yet so far. We noted that co-doped with two or more nonmetallic elements was widely reported in TiO 2 systems [44][45][46][47][48][49][50], where the photocatalytic activities of TiO 2 were further improved compared to single doping due to their excellent visible light photocatalysis caused by the narrowed bandgap. This indicated that nonmetallic element co-doped TiO 2 could enhance the visible light, but also reduce the recombination rate of photo-induced electron-hole pairs. WO 3 exhibits property similar to that of TiO 2 because the VB of WO 3 and TiO 2 are mainly composed of O 2p orbitals. It is confirmed that the effective nonmetallic doping induces hybridization of the outer orbitals of the doped elements and the VB of TiO 2 to form a new energy level at the top of the VB and reduce the bandgap of TiO 2 . This suggests that co-doping of WO 3 with two or more nonmetallic elements is a promising route to improve the absorption efficiency of WO 3 .
Herein, we reported the first simultaneous synthesis of S-N-WO 3 using (NH 4 ) 2 S as N and S atom source. In this strategy, S-N-WO 3 exhibited a narrower energy bandgap compared with the pure one. It is attributed to the delocalization of the N 2p orbit with the O 2p orbit after doping of N. Furthermore, S-N-WO 3 extended its optical response range to longer wavelength visible light because of the fact that 3s (S 6+ ) orbitals can be delocalized with W 5d and O 2p orbitals to form a new intermediate level above the VB top. Therefore, the absorption threshold of S-N-WO 3 can be lowered by co-doping with the S and N elements. Based on this transition, the performance of S-N-WO 3 for PEC water oxidation is superior to that of pure WO 3 .

Fabrication of Electrodes
In a typical procedure, an (NH 4 ) 2 S-derived precursor powder (800 mg), PEG (400 mg), and Marpolose (80 mg) were mixed in water (0.6 mL) under slow stirring for 4 h to form a smooth paste without bubbles. The resulting paste was squeezed on a clean FTO glass substrate by a doctor-blade coater and dried at 80 • C for 15 min. After repeating the procedure two times, the electrodes were calcined at 450 • C in O 2 flow for 1.5 h to give different WO 3 electrodes. The pure WO 3 electrode was fabricated by the same method using a precursor prepared without addition of (NH 4 ) 2 S.
All PEC measurements were examined in a two-compartment PEC cell separated by a Nafion membrane using an electrochemical analyzer (Shanghai Chenhua Instrument Co., Ltd., Shanghai, China, CHI660E). A three-electrode system was employed using different WO 3 electrodes and Ag/AgCl electrodes in one cell as the working and reference electrodes, respectively, and a Pt wire-in the other cell as the counter electrode. The linear sweep voltammograms (LSV) were measured at a scan rate of 5 mV s −1 . Light (λ > 450 nm, 100 mW cm −2 ) was irradiated from the backside of the working electrode using a 500 W xenon lamp with a UV-cut filter (λ > 450 nm). The output of light intensity was calibrated as 100 mW cm −2 using a spectroradiometer (USR-40, Ushio Shanghai Inc., Shanghai, China). Photoelectrocatalysis was conducted under potentiostatic conditions at 0.5 V at 25 • C with illumination of light (λ > 450 nm, 100 mW cm −2 ) for 1 h. All the PEC experiments were carried out under argon atmosphere in an aqueous 0.1 M phosphate buffer solution (pH 6.0). The amounts of H 2 and O 2 evolved were determined from the analysis of the gas phase of counter and working electrode compartments, respectively, using gas chromatography (Shimadzu GC-8A with a TCD detector and molecular sieve 5 A column and Ar carrier gas). A monochromic light with 10 nm bandwidth was provided by a 500 W xenon lamp using a monochromator for incident photon-to-current conversion efficiency (IPCE) measurements.
Raman spectra of the WO3 samples exhibited the characteristic peaks of the monoclinic WO3 at 135.   Figure 2II, where the uniform distribution of W and O (Figure 2IIc,d) are confirmed. While the signals of both S and N can be clearly detected on the same structural portion, no other impurity elements were observed in the samples. However, both N and S mappings exhibited higher distribution due to the presence of higher contents in the WO3−15 sample. The atom number ratios of W/N as well as W/S were calculated from EDS data to exhibit that it increases with an increase in the nW:n(NH4)2S ratio from 1:0 to 1:15 and thereafter decreased above 1:15 ( Figure S1 and Table 1).

Figure 1. (A) XRD patterns and (B) Raman spectra of (a) WO
Raman spectra of the WO 3 samples exhibited the characteristic peaks of the monoclinic WO 3 Figure 2II, where the uniform distribution of W and O (Figure 2IIc,d) are confirmed. While the signals of both S and N can be clearly detected on the same structural portion, no other impurity elements were observed in the samples. However, both N and S mappings exhibited higher distribution due to the presence of higher contents in the WO 3 −15 sample. The atom number ratios of W/N as well as W/S were calculated from EDS data to exhibit that it increases with an increase in the n W :n (NH4)2S ratio from 1:0 to 1:15 and thereafter decreased above 1:15 ( Figure S1 and Table 1).   The chemical composition and valence states of different WO3 samples were investigated through XPS. The spectra were calibrated with the C 1s peak as reference. As shown in Figure S2, the XPS survey spectrum of WO3-0 depicts that no other impurity signals, besides the C 1s line, were detected and only W and O. The high-resolution XPS spectrum of W 4f exhibited two peaks at 37.7 eV and 35.5 eV associated with the spin-orbit doublet of W 4f7/2 and W 4f5/2, respectively, for a W 6+ state in WO3 [11,51]. The apparent peaks at 531.0 eV and 530.2 eV in the XPS spectrum of O 1s can be assigned to the H2O and W-O species, respectively [52,53]. The XPS spectra of W 4f doublet for WO3−5, WO3−10, and WO3−15 samples are shown in Figure 3A. Three of the samples exhibited two characteristic peaks at 38.1 eV and 35.9 eV corresponding to 4f5/2 and W 4f7/2 components of the WO3 lattice similar to WO3−0. The components with binding energies 530.8 and 532.0 eV in the high-resolution O 1s spectra ( Figure 3B) are correspondent to the W-O and hydrocarbonate species, respectively. The XPS spectrum in an N 1S region of 399-404 eV ( Figure  3C) exhibited two peaks at 400.2 eV and 402.2 eV, as obtained by two-bands deconvolution. The former one is ascribed to the binding energies of W-O-N, and the latter one is attributed to surface adsorbed (NOx, NH3) and/or nitrogen trapped in the surface layers as γ-N2 [38,[54][55][56]. Considering that no peaks that correspond to W2N or WN were observed in the XRD patterns, we confirmed the substitution of O in WO3 by N element and  The chemical composition and valence states of different WO 3 samples were investigated through XPS. The spectra were calibrated with the C 1s peak as reference. As shown in Figure S2, the XPS survey spectrum of WO 3 -0 depicts that no other impurity signals, besides the C 1s line, were detected and only W and O. The high-resolution XPS spectrum of W 4f exhibited two peaks at 37.7 eV and 35.5 eV associated with the spin-orbit doublet of W 4f 7/2 and W 4f 5/2 , respectively, for a W 6+ state in WO 3 [11,51]. The apparent peaks at 531.0 eV and 530.2 eV in the XPS spectrum of O 1s can be assigned to the H 2 O and W-O species, respectively [52,53]. The XPS spectra of W 4f doublet for WO 3 −5, WO 3 −10, and WO 3 −15 samples are shown in Figure 3A. Three of the samples exhibited two characteristic peaks at 38.1 eV and 35.9 eV corresponding to 4f 5/2 and W 4f 7/2 components of the WO 3 lattice similar to WO 3 −0. The components with binding energies 530.8 and 532.0 eV in the high-resolution O 1s spectra ( Figure 3B) are correspondent to the W-O and hydrocarbonate species, respectively. The XPS spectrum in an N 1S region of 399-404 eV ( Figure 3C) exhibited two peaks at 400.2 eV and 402.2 eV, as obtained by two-bands deconvolution. The former one is ascribed to the binding energies of W-O-N, and the latter one is attributed to surface adsorbed (NO x , NH 3 ) and/or nitrogen trapped in the surface layers as γ-N 2 [38,[54][55][56]. Considering that no peaks that correspond to W 2 N or WN were observed in the XRD patterns, we confirmed the substitution of O in WO 3 by N element and the formation of W-O-N banding. In the high-resolution XPS spectra, the S 2p ( Figure 3D) peak at 168.7 eV was observed for WO 3 −15 (no signals for the two other samples), and it is Nanomaterials 2022, 12, 2079 6 of 13 assigned to the S 2p orbits in the +6 oxidation state [40,57]. The formation of W-S bonding instead of W-O bonding can be confirmed by the following two reasons: (1) the binding energy of 168.7 eV for W-S is different from that of 169.9 eV for the SO 4 2− , (2) S 2− doping may only occur with difficulty because the S 2− radius (1.70 Å) is significantly larger than O 2− (1.22 Å). Generally, the larger the ionic radius is, the doping would be more difficult to occur due to higher formation energy. Therefore, the replacement of W 6+ by S 6+ is more favorable than replacing O 2− with S 2− . Furthermore, the XPS results also demonstrate that the S-N-WO 3 could be formed when the n W :n (NH4)2S ratio was over 1:5. Compared to that of the WO 3 −0, the positive shifts of 0.4 eV and 0.8 eV for W 4f and O 1s can be seen, which is attributed to the electron transfer from the dopant energy level to the CB of WO 3 . It is considered that this transfer can be benefitial to improving the optical properties of WO 3 . Nanomaterials 2022, 12, x FOR PEER REVIEW 6 of 13 the formation of W-O-N banding. In the high-resolution XPS spectra, the S 2p (Figure 3d) peak at 168.7 eV was observed for WO3−15 (no signals for the two other samples), and it is assigned to the S 2p orbits in the +6 oxidation state [40,57]. The formation of W-S bonding instead of W-O bonding can be confirmed by the following two reasons: (1) the binding energy of 168.7 eV for W-S is different from that of 169.9 eV for the SO4 2− , (2) S 2− doping may only occur with difficulty because the S 2− radius (1.70 Å) is significantly larger than O 2− (1.22 Å). Generally, the larger the ionic radius is, the doping would be more difficult to occur due to higher formation energy. Therefore, the replacement of W 6+ by S 6+ is more favorable than replacing O 2− with S 2− . Furthermore, the XPS results also demonstrate that the S-N-WO3 could be formed when the nW:n(NH4)2S ratio was over 1:5. Compared to that of the WO3−0, the positive shifts of 0.4 eV and 0.8 eV for W 4f and O 1s can be seen, which is attributed to the electron transfer from the dopant energy level to the CB of WO3. It is considered that this transfer can be benefitial to improving the optical properties of WO3. To further reveal the mechanism of S and N co-doped WO3, it is necessary to discuss the influence of the nW:n(NH4)2S ratio on the content of each element into the WO3 lattice ( Figure 4). The contents of (a) O, (b) W, (c) N, and (d) S were calculated from XPS data ( Figure 3 and Table 1). For WO3−0, the atom percent of W and O were 14.99% ± 1.2 and 44.82% ± 0.3, respectively. For WO3−5, no S element was doped into the WO3 lattice, only N element (1.64% ± 0.15). Compared to WO3−0, almost no change was observed for the W content (14.97% ± 1.0), but a decreasing trend was seen for the O (43.89% ± 0.8) content. As increasing the ratios from 1:5 to 1:15, the N content increased from 1.64% ± 0.15 to 5.82% ± 0.12, but the O content decreased from 40.4% ± 0.7 to 37.37 ± 0.8. It suggests that the higher nW:n(NH4)2S ratios could lead to more oxygen defects due to N doping. Special attention should be paid to the change trend of W content, which decreased with the appearance of the S element from 1:10 (12.7% ± 0.7) due to the substitution of W 6+ by S 6+ . The significantly higher contents for both N (5.82% ± 0.12) and S (5.85% ± 0.18) were obtained at 1:15 than at other ratios. Such high N and S contents can improve the absorption of visible light to further narrow the bandgap of WO3. Thereafter, the increase of atom To further reveal the mechanism of S and N co-doped WO 3 , it is necessary to discuss the influence of the n W :n (NH4)2S ratio on the content of each element into the WO 3 lattice (Figure 4). The contents of (a) O, (b) W, (c) N, and (d) S were calculated from XPS data ( Figure 3 and Table 1). For WO 3 −0, the atom percent of W and O were 14.99% ± 1.2 and 44.82% ± 0.3, respectively. For WO 3 −5, no S element was doped into the WO 3 lattice, only N element (1.64% ± 0.15). Compared to WO 3 −0, almost no change was observed for the W content (14.97% ± 1.0), but a decreasing trend was seen for the O (43.89% ± 0.8) content. As increasing the ratios from 1:5 to 1:15, the N content increased from 1.64% ± 0.15 to 5.82% ± 0.12, but the O content decreased from 40.4% ± 0.7 to 37.37 ± 0.8. It suggests that the higher n W :n (NH4)2S ratios could lead to more oxygen defects due to N doping. Special attention should be paid to the change trend of W content, which decreased with the appearance of the S element from 1:10 (12.7% ± 0.7) due to the substitution of W 6+ by S 6+ . The significantly higher contents for both N (5.82% ± 0.12) and S (5.85% ± 0.18) were obtained at 1:15 than at other ratios. Such high N and S contents can improve the absorption of visible light to further narrow the bandgap of WO 3 . Thereafter, the increase of atom percent for W and O and decrease for S and N was observed at higher n W :n (NH4)2S ratios, and it may correspond to limitations in the substitution capacity of the WO 3 lattice. percent for W and O and decrease for S and N was observed at higher nW:n(NH4)2S ratios, and it may correspond to limitations in the substitution capacity of the WO3 lattice.

The Optical Properties of S-N-WO3
The DRS and the corresponding Tauc plots for the WO3 samples with changes in the ratio of nW:n(NH4)2S are exhibited in Figure 5. As shown in Figure 5A, the WO3-0 can only absorb light below 470 nm. However, a significant redshift in the absorption edge with new shoulders appearing at 470-650 nm can be seen in N-doped WO3 or the S-N co-doped one. It was found that the absorption properties increased when increasing the ratio of nW:n(NH4)2S below 1:15, and then they decreased when further increasing the addition of (NH4)2S. Absorption above 700 nm was observed for S-N-WO3 samples due to the formation of lattice defects caused by doping, in contrast to the negligible absorption for neat WO3. Furthermore, Tauc plots based on DRS data are shown in Figure 5B. The bandgap was determined by this technique in different materials [58][59][60]. It was reported that WO3 has an indirect optical bandgap. The Tauc plots for WO3−0 provided the absorption energy of 2.64 eV, which is in agreement with the bandgap energy of WO3 reported previously [11]. The Tauc plots for S-N-WO3 samples exhibited two different slopes due to the appearance of the new shoulders. Therefore, the estimated band energies for S-N-WO3 samples were obtained from the slopes, as displayed in Table 1. For WO3−5, the bandgap was reduced because a new intermediate N 2p orbital could be formed between the CB and the VB owing to N doping. It was observed that, in WO3 co-doped with S and N, the bandgap further decreased due to the formation of an intrabandgap above the VB edge and a dopant energy level below the CB of WO3.

The Optical Properties of S-N-WO 3
The DRS and the corresponding Tauc plots for the WO 3 samples with changes in the ratio of n W :n (NH4)2S are exhibited in Figure 5. As shown in Figure 5A, the WO 3 -0 can only absorb light below 470 nm. However, a significant redshift in the absorption edge with new shoulders appearing at 470-650 nm can be seen in N-doped WO 3 or the S-N co-doped one. It was found that the absorption properties increased when increasing the ratio of n W :n (NH4)2S below 1:15, and then they decreased when further increasing the addition of (NH 4 ) 2 S. Absorption above 700 nm was observed for S-N-WO 3 samples due to the formation of lattice defects caused by doping, in contrast to the negligible absorption for neat WO 3 . Furthermore, Tauc plots based on DRS data are shown in Figure 5B. The bandgap was determined by this technique in different materials [58][59][60]. It was reported that WO 3 has an indirect optical bandgap. The Tauc plots for WO 3 −0 provided the absorption energy of 2.64 eV, which is in agreement with the bandgap energy of WO 3 reported previously [11]. The Tauc plots for S-N-WO 3 samples exhibited two different slopes due to the appearance of the new shoulders. Therefore, the estimated band energies for S-N-WO 3 samples were obtained from the slopes, as displayed in Table 1. For WO 3 −5, the bandgap was reduced because a new intermediate N 2p orbital could be formed between the CB and the VB owing to N doping. It was observed that, in WO 3 co-doped with S and N, the bandgap further decreased due to the formation of an intrabandgap above the VB edge and a dopant energy level below the CB of WO 3 . Figure 6 is the relation between the absorbance value at 600 nm (Abs 600 ). The Abs 600 value is a measure of the increase/decrease of the shoulders at 470-650 nm. Compared with WO 3 −0, the Abs 600 increased from 0.02 to 0.11 with an increase in the ratio of n W :n (NH4)2S from 1:5 to 1:15, and, thereafter, decreased over 1:15 to 0.06 at 1:25. The dependency of Abs 600 on the n W :n (NH4)2S ratio agrees to the cases of the N and S content ( Figure S1), indicating that the longer wavelength absorption due to the shoulders can be attributed to doping of N and S into a WO 3 lattice.  Figure 6 is the relation between the absorbance value at 600 nm (Abs600). Th value is a measure of the increase/decrease of the shoulders at 470-650 nm. Co with WO3−0, the Abs600 increased from 0.02 to 0.11 with an increase in the nW:n(NH4)2S from 1:5 to 1:15, and, thereafter, decreased over 1:15 to 0.06 at 1:25. The ency of Abs600 on the nW:n(NH4)2S ratio agrees to the cases of the N and S content (Fi indicating that the longer wavelength absorption due to the shoulders can be at to doping of N and S into a WO3 lattice.

Photoelectrocatalytic Properties
The LSVs for these electrodes calcined at 450 °C were measured with choppe light irradiation to study their PEC water oxidation performance. The photoano rents of these electrodes were observed above 0.1 V vs. Ag/AgCl due to water ox   Figure 6 is the relation between the absorbance value at 600 nm (Abs600). The Abs600 value is a measure of the increase/decrease of the shoulders at 470-650 nm. Compared with WO3−0, the Abs600 increased from 0.02 to 0.11 with an increase in the ratio of nW:n(NH4)2S from 1:5 to 1:15, and, thereafter, decreased over 1:15 to 0.06 at 1:25. The dependency of Abs600 on the nW:n(NH4)2S ratio agrees to the cases of the N and S content ( Figure S1), indicating that the longer wavelength absorption due to the shoulders can be attributed to doping of N and S into a WO3 lattice.

Photoelectrocatalytic Properties
The LSVs for these electrodes calcined at 450 °C were measured with chopped visible light irradiation to study their PEC water oxidation performance. The photoanodic currents of these electrodes were observed above 0.1 V vs. Ag/AgCl due to water oxidation.

Photoelectrocatalytic Properties
The LSVs for these electrodes calcined at 450 • C were measured with chopped visible light irradiation to study their PEC water oxidation performance. The photoanodic currents of these electrodes were observed above 0.1 V vs. Ag/AgCl due to water oxidation. The photocurrent of 1.15 mA cm −2 at 1.0 V for WO 3 −15 was the highest in comparison to other samples. Moreover, as shown in Figure 7B, the dependency of the photocurrent at 1.0 V on the n W :n (NH4)2S ratio for each electrode is in agreement with the N and S contents. Figure 7C exhibits that the photocurrent at 0.68 V vs. Ag/AgCl (1.23 V vs. RHE) under visible-light irradiation chopped was stable during PEC water oxidation (5 min) for these electrodes. The photocurrent of the WO 3 −15 electrode (1.0 mA cm −2 ) was higher than those of the WO 3 −0, WO 3 −5, WO 3 −10, WO 3 −20, and WO 3 −25 by a factor of 83 (0.012 mA cm −2 ), 3.6 (0.28 mA cm −2 ), 1.4 (0.71 mA cm −2 ), 1.6 (0.62 mA cm −2 ), and 2.3 (0.44 mA cm −2 ), respectively.
1.0 V on the nW:n(NH4)2S ratio for each electrode is in agreement with the N and S contents. Figure 7C exhibits that the photocurrent at 0.68 V vs. Ag/AgCl (1.23 V vs. RHE) under visible-light irradiation chopped was stable during PEC water oxidation (5 min) for these electrodes. The photocurrent of the WO3−15 electrode (1.0 mA cm −2 ) was higher than those of the WO3−0, WO3−5, WO3−10, WO3−20, and WO3−25 by a factor of 83 (0.012 mA cm −2 ), 3.6 (0.28 mA cm −2 ), 1.4 (0.71 mA cm −2 ), 1.6 (0.62 mA cm −2 ), and 2.3 (0.44 mA cm −2 ), respectively.  Figure 8A). A higher photoanodic current due to water oxidation was observed for the WO3−15 electrode. Compared with the electrodes prepared at other nW:n(NH4)2S ratios, the highest charge amount passed and the amount (nO2) of O2 evolved during the 1 h photoelectrocatalysis for WO3−15 were 2.12 C and 5.36 mmol (98% Faradaic efficiency), respectively (Figure 8B and Table S2). These results clearly prove that the doping of S and N enhances the PEC performance of WO3−15 in application to water oxidation.   Figure 8A). A higher photoanodic current due to water oxidation was observed for the WO 3 −15 electrode. Compared with the electrodes prepared at other n W :n (NH4)2S ratios, the highest charge amount passed and the amount (n O2 ) of O 2 evolved during the 1 h photoelectrocatalysis for WO 3 −15 were 2.12 C and 5.36 mmol (98% Faradaic efficiency), respectively ( Figure 8B and Table S2). These results clearly prove that the doping of S and N enhances the PEC performance of WO 3 −15 in application to water oxidation. Figure 7C exhibits that the photocurrent at 0.68 V vs. Ag/AgCl (1.23 V vs. RHE) under visible-light irradiation chopped was stable during PEC water oxidation (5 min) for these electrodes. The photocurrent of the WO3−15 electrode (1.0 mA cm −2 ) was higher than those of the WO3−0, WO3−5, WO3−10, WO3−20, and WO3−25 by a factor of 83 (0.012 mA cm −2 ), 3.6 (0.28 mA cm −2 ), 1.4 (0.71 mA cm −2 ), 1.6 (0.62 mA cm −2 ), and 2.3 (0.44 mA cm −2 ), respectively.  Table S2). These results clearly prove that the doping of S and N enhances the PEC performance of WO3−15 in application to water oxidation.  The action spectra of IPCE for these electrodes are shown in Figure 9. In Figure 9A, for WO 3 −0, the photocurrent was not observed above 470 nm, which is consistent with the bandgap energy of WO 3 . For the WO 3 −5 electrode, the onset wavelength for photocurrent generation was at least 520 nm, which, due to N doping, is significantly longer than that of WO 3 −0. The energy of the onset wavelength for WO 3 −5 (520 nm, 2.38 eV) was lower than the main bandgap excitation for WO 3 −5 (2.43 eV). This suggests that the photocurrent was generated based on the bandgap excitation, and the bandgap excitation occurs through collateral excitation from intermediate N 2p orbital to CB for the WO 3 −5 electrode. The onset wavelengths for WO 3 −10, WO 3 −15, WO 3 −20, and WO 3 −25, due to the S and N co-doping, are considerably shifted to the wavelengths (580 nm) longer than that of single N-doped WO 3 −5. However, for all of S-N-WO 3 electrodes, the photocurrent at longer wavelengths longer than 580 nm could not be detected due to the limited current detection level of the employed apparatus. For the electrodes prepared at different n W :n (NH4)2S ratios, the IPCE values at 450 nm (IPCE 450 ) are shown in Figure 9B; the IPCE 450 for WO 3 −5 electrode (0.63%) was 4.2 times higher than that of WO 3 −0 (0.15%), basically due to the formation of the formation of N doping . It precipitously increased at the ratios of 1:5 to 1:15, indicating that S and N co-doping plays a positive role in not only the increase in the onset wavelength but also in the increase in the IPCE 450 . The maximum IPCE 450 of WO 3 −15 (5.81%) was obtained, which was 9.2 times higher compared to that of the WO 3 −5 electrode due to co-doping by S and N. It is suggested that the highest contents of S and N into WO 3 lattice can effectively increase the electron transport rate and further inhibit recombination of electron-hole pairs in the film. When increasing the n W :n (NH4)2S ratios, the IPCE 450 for WO 3 −20 and WO 3 −25 reduced to 1.99% and 1.46%, respectively. However, they were still higher than that of the WO 3 −5 electrode. The relationship between IPCE 450 and n W :n (NH4)2S ratio is consistent with the Abs 600 value in DRS data ( Figure 6), indicating that the S and N co-doping is responsible for the lengthening of the onset wavelength for PEC water oxidation.
for WO3−0, the photocurrent was not observed above 470 nm, which is consistent w bandgap energy of WO3. For the WO3−5 electrode, the onset wavelength for photoc generation was at least 520 nm, which, due to N doping, is significantly longer th of WO3−0. The energy of the onset wavelength for WO3−5 (520 nm, 2.38 eV) was than the main bandgap excitation for WO3−5 (2.43 eV). This suggests that the photoc was generated based on the bandgap excitation, and the bandgap excitation through collateral excitation from intermediate N 2p orbital to CB for the WO3−5 ele The onset wavelengths for WO3−10, WO3−15, WO3−20, and WO3−25, due to the S co-doping, are considerably shifted to the wavelengths (580 nm) longer than that of N-doped WO3−5. However, for all of S-N-WO3 electrodes, the photocurrent at wavelengths longer than 580 nm could not be detected due to the limited current de level of the employed apparatus. For the electrodes prepared at different nW:n(NH4)2S the IPCE values at 450 nm (IPCE450) are shown in Figure 9B; the IPCE450 for WO3trode (0.63%) was 4.2 times higher than that of WO3−0 (0.15%), basically due to t mation of the formation of N doping. It precipitously increased at the ratios of 1:5 t indicating that S and N co-doping plays a positive role in not only the increase in th wavelength but also in the increase in the IPCE450. The maximum IPCE450 of W (5.81%) was obtained, which was 9.2 times higher compared to that of the WO3−5 ele due to co-doping by S and N. It is suggested that the highest contents of S and N int lattice can effectively increase the electron transport rate and further inhibit recom tion of electron-hole pairs in the film. When increasing the nW:n(NH4)2S ratios, the I for WO3−20 and WO3−25 reduced to 1.99% and 1.46%, respectively. However, the still higher than that of the WO3−5 electrode. The relationship between IPCE nW:n(NH4)2S ratio is consistent with the Abs600 value in DRS data ( Figure 6), indicatin the S and N co-doping is responsible for the lengthening of the onset wavelength fo water oxidation.

Conclusions
Nitrogen and sulfur co-doped crystalline WO 3 was synthesized by thermal decomposition of (NH 4 ) 2 S-derived precursor, in which (NH 4 ) 2 S acted as a sulfur source, as well as the nitrogen source for doping. The addition of (NH 4 ) 2 S has an effect on the physiochemical properties, and the performance of PEC water oxidation of the WO 3 -0 and S-N-WO 3 electrodes was investigated to characterize the co-doping of S and N into the WO 3 lattice and reveal the mechanism of superior performance for PEC water oxidation using the S-N-WO 3 photoanode. S-N-WO 3 exhibited the optimum n W :n (NH4)2S ratio at 1:15 for the high concentration of both S and N elements. The S and N co-doping is responsible for the significant redshift in the absorption edge, with a new shoulder appearing at 470-650 nm compared to that of WO 3 −0. The S-N-WO 3 photoanode is able to utilize visible light at wavelengths below 580 nm for PEC water oxidation, in contrast to the WO 3 −0 photoanode being able to work below 470 nm. The IPCE (5.81%) at 450 nm for S-N-WO 3 photoanode calcined at 450 • C was higher than that (0.15%) for WO 3 −0 by 38.7 times due to the codoping of S and N. The S-N-WO 3 photoanode is expected to be applied for PEC water splitting cell as an artificial photocatalyst to improve the solar energy conversion efficiency.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/nano12122079/s1, Figure S1: (A) Relationship between the relative contents of N, S and n W :n (NH4)2S ratio; Figure S2: (A) the XPS survey spectrum and (B) XPS spectra in (A) W 4f, (B) O 2p regions for WO 3 −0; Table S1: Atomic percent of surface W, O, N, and S estimated by XPS; Table S2: Summary of PEC water oxidation in a 0.1 M phosphate buffer solution (pH 6.0) for 1 h using different WO 3 electrodes calcined at 450 • C.