Design of WO₃-Ov/In₂S₃ Photocatalysts

WO₃ nanoparticles with Ovs (WO₃-Ov) were prepared by annealing pristine WO₃ nanoparticles in 450 °C for 1 h at 5 % H₂/Ar atmosphere. In₂S₃ was subsequently grown on the surface of WO₃-Ov via a hydrothermal method. The synthesis process of WO₃-Ov/In₂S₃ heterojunction structure is illustrated in Figure 1a. The obtained samples with different contents of In₂S₃ are denoted as W_OvIn1, W_OvIn2, and W_OvIn3, while the heterojunction samples of In₂S₃ grown on pristine WO₃ nanoparticles are denoted as WIn1, WIn2, and WIn3. It can be seen from field emission scanning electron microscopy (FESEM) (Figure S1) that the size of pristine WO₃ nanoparticles is ~100 nm. After the introduction of Ovs, the morphology remains unchanged. FESEM images show that, as the amount of In₂S₃ increases, both the surface of WO₃-Ov/In₂S₃ (Figure S2) and WO₃/In₂S₃ samples (Figure S3) are covered with increasing amount of hierarchical structures. Furthermore, transmission electronic microscopy (TEM) confirms the hierarchical architecture of W_OvIn2 (inset of Figure 1b). Figure 1b illustrates that the hierarchical In₂S₃ is composed of plenty of nanoflakes assembled on the surface of the WO₃-Ov nanoparticle. It should be noted that this hierarchical morphology can provide a large number of active sites to participate in the photocatalytic reaction, which benefits the photocatalytic performances. In high solution TEM images (HRTEM, Figure 1c), In₂S₃ displays clear lattice fringes with an interlayer spacing of 0.269 and 0.301 nm, which can be assigned to the (022) and (211) crystal planes of β-In₂S₃, respectively. Inset of Figure 1c presents the profiles of corresponding lattice fringes. Furthermore, high angle annular dark-field (HAADF) image of W_OvIn2 (Figure 1d) and corresponding elemental mapping images of In, S, W, and O (Figure 1e) reveal that hierarchical In₂S₃ with uniform composition are anchored on WO₃-Ov nanoparticles.

Characterization of WO₃-Ov/In₂S₃ Photocatalysts

To detect the unpaired electrons in WO₃-Ov, electron paramagnetic resonance (EPR) measurement was carried out. As shown in Figure 2a, the g value at 2.003 corresponds to unpaired electrons in the Ov site.² It is apparent that after the hydrogen thermal reduction process, the signal area at g=2.003 exhibits significant enhancement, confirming that a considerable number of Ovs were introduced to WO₃ with this simple method. X-ray diffraction (XRD) was adopted to figure out the crystal structure of WO₃-Ov, In₂S_3, and WO₃-Ov/In₂S₃. The peaks of WO₃-Ov can be well indexed to the monoclinic phase (JCPDS card no. 71–2141), which is consistent with pristine WO₃ (Figure S4). It can be concluded that the hydrogen thermal annealing keeps WO₃ in its pristine crystal phase. As shown in Figure 2b, the diffraction peaks of In₂S₃ can be indexed to tetragonal β-In₂S₃ (JCPDS card no. 84–1385). XRD patterns of WO₃-Ov/In₂S₃ verify the successful synthesis of WO₃-Ov/In₂S₃ heterojunction. Moreover, with increasing content of In₂S₃ in WO₃-Ov/In₂S₃ heterojunction, the peak intensities of the In₂S₃ phase increase accordingly. The matching of the XRD peaks of WO₃-Ov/In₂S₃ to those of the combination of WO₃-Ov and In₂S₃ single component indicates that the crystal structure is not altered in the hybridization process. Similarly, XRD patterns of WIn1, WIn2, and WIn3 (Figure S4) also confirm the successful construction of WO₃/In₂S₃ heterojunctions. Raman spectra were conducted to investigate the structural information of the heterojunction. As shown in Figure 2c, Raman peaks of WO₃-Ov at 714 and 807 cm⁻¹ can be assigned to the stretching vibrations of longer and shorter W−O−W bonds in WO₆ octahedron units, respectively. The peaks at 269 and 331 cm⁻¹ are attributed to the bending vibrations of the bridging oxygen atoms in W−O−W.¹⁸ In comparison, the main Raman peak of WO₃-Ov demonstrates obvious peak shifting and broadening compared with pristine WO₃ (Figure S5a), which is related to the deficiency of oxygen atoms and non-stoichiometry on the WO₃ surface.¹⁹ The Raman spectra of In₂S₃ show a distinct peak at 306 cm⁻¹, which is associated with A_1g mode of β-In₂S₃.²⁰ The Raman spectra of WO₃-Ov/In₂S₃ are composed of oscillation peaks of WO₃-Ov and In₂S₃. It can be observed that all the peaks in WO₃-Ov/In₂S₃ shift to lower wavenumber compared with pristine WO₃-Ov, confirming the chemical bonding effect between heterointerface.^21-23 Raman spectra of WO₃/In₂S₃ are depicted in Figure S5b, where the main peak of WO₃/In₂S₃ shows a similar shift tendency compared with pristine WO₃. X-ray photoelectron spectroscopy (XPS) was performed to examine the valence state change of the elements in WO₃-Ov/In₂S₃ heterojunctions. XPS survey spectra (Figure S6) show the existence of W, O, In, and S in this heterojunction. Pristine In₂S₃ demonstrates two peaks located at 444.8 and 452.3 eV, which correspond to In 3d_5/2 and 3d_3/2, respectively (Figure S7a). After hybridization, these two peaks shift to higher binding energies. Moreover, W 4 f XPS spectra of WO₃-Ov (Figure S7b) demonstrate two peaks at 37.6 and 35.5 eV.^{18, 24} W 4f peaks of WO₃-Ov/In₂S₃ samples show a negative shift compared with WO₃-Ov. After conjunction, S 2p XPS (Figure S8a) shows that S 2p_3/2 and 2p_1/2 locate at 161.7 and 162.9 eV, respectively.^{25, 26} Whereas, in these heterojunctions, S 2p_3/2 and 2p_1/2 shift to higher binding energy. The positive shifts of In 3d and S 2p XPS peaks indicate the loss of electrons in In₂S₃ upon conjunction.

Meanwhile, O1s XPS (Figure S8b) demonstrates that the lattice O of WO₃-Ov/In₂S₃ moves to lower binding energy compared with WO₃-Ov, showing the same peak shifting tendency as W 4 f XPS after hybridization. It can be concluded from W 4 f and O 1s XPS that WO₃-Ov accepts electrons after being in contact with In₂S₃. Further, it can be presumed that upon the contact of WO₃-Ov and In₂S₃ with different E_f, electrons spontaneously transfer from In₂S₃ to WO₃-Ov until E_f is equilibrated at the heterointerface. Consequently, the flow of electrons results in the formation of an internal electric field between the heterointerface of the two semiconductors, pointing from In₂S₃ to WO₃-Ov. XPS of WO₃-Ov and WO₃ are exhibited in Figure S9 and S10 to compare their chemical states. XPS of WO₃/In₂S₃ are shown in Figure S11, the detailed analysis can be found in Supporting Information.

The electronic structure and coordination environment of In and W before and after conjunction were investigated by X-ray absorption near edge structure (XANES). It can be seen that In K pre-edge peak position of W_OvIn2 centers at a higher energy than that of In₂S₃, indicating that the In element in W_OvIn2 possesses higher valence state than that in In₂S₃ (Figure 2d),^{27, 28} which is consistent with the In 3d XPS results. Fourier-transformed extended X-ray absorption fine structure (EXAFS) (Figure 2e) was obtained to acquire quantitative structural parameters. It can be noticed that the near-edge absorption energy of W_OvIn2 shifts to higher R direction (2.02 Å) with increased intensity compared with In₂S₃ (1.99 Å), implying that the chemical bonding character of the In₂S₃ has been changed after the hybridization. Figure 2f presents the fitting result of the extended In K-edge. Table S1 summarizes the quantitative structural parameters fitting result of the extended In K-edge. W L₃-edge XANES spectra (Figure 2g) shows that the white line intensity of as prepared WO₃-Ov is slightly lower than pristine WO₃. The weaker white line absorption peak is related to the increase of W 5d band electron densities in WO₃-Ov.^{29, 30} After the growth of In₂S₃, W_OvIn2 shows a lower intensity peak, revealing the higher average electron density and the decrease of oxidation states of W compared with WO₃-Ov, strongly verifying that WO₃-Ov accept electrons in this W_OvIn2 heterojunction.²⁹ EXAFS spectra of W L₃-edge for WO₃, WO₃-Ov and W_OvIn2 are displayed in Figure 2h. EXAFS curve fitting of W-L₃ (Figure 2i) shows that compared with WO₃, the coordination number in WO₃-Ov decreases, which is related to the deficiency of oxygen atoms. Moreover, as for W_OvIn2, apart from W−O bond, W−S bond (2.39 Å) as chemical covalent bond forms in W_OvIn2. It can be verified that the growth of In₂S₃ changes the W chemical bonding environment in WO₃-Ov samples. The fitting parameters of W L₃-edge EXAFS are summarized in Table S2. The wavelet transform contour plots of In K-edge weighted EXAFS for In₂S₃ and W_OvIn2 (Figure S12), and W L₃-edge for WO₃, WO₃-Ov and W_OvIn2 (Figure S13) also highlight minor differences in W_OvIn2 and the two single components.³¹ The above results provide evidence that, after the combination, the electrons transfer from In₂S₃ to WO₃-Ov. And the W−S chemical bonding between WO₃-Ov and In₂S₃ is formed at the interface, which may provide an electron transport pathway during photocatalytic process. The microscopic environment and local electronic structure of WIn2 samples are depicted in Supporting Information (Figures S14–S17, Table S3 and S4). The neighboring structure around W atoms and In atoms of WIn2 shows differences compared with WO₃ and In₂S₃ components. k³-weighted EXAFS data and fit data are shown in Figure S18 and Figure S19. Figure S20 shows crystal structure models of In₂S₃ and WO₃ used in EXAFS fitting.

Band Structure Analysis

The light absorption property of the photocatalysts was tested by UV–Vis diffuse reflectance spectroscopy (DRS). As shown in Figure S21, compared with pristine WO₃, WO₃-Ov shows evidently enhanced photo absorption in the visible region. After hybridization, WO₃-Ov/In₂S₃ heterojunctions show absorption edge from 550 to 580 nm (Figure S22a), which is similar to In₂S₃. The absorption abilities of WO₃-Ov/In₂S₃ heterojunctions is stronger than that of WO₃/In₂S₃ heterojunctions (Figure S22b), which is beneficial to their photocatalytic properties. The optical photographs of WO₃-Ov/In₂S₃ heterojunctions and WO₃/In₂S₃ heterojunctions are displayed in Figure S23a and S23b, respectively, showing that the existence of Ovs deepens the color of the heterojunctions.

Tauc plot can be obtained based on UV–Vis DRS and Kubelka–Munk equation to analyze the band gap value (E_g) of the semiconductors. As shown in Figure S24, E_g of WO₃-Ov and WO₃ is estimated to be 2.71 and 2.89 eV, respectively, which verifies that the introduction of Ov can slightly lower E_g. In addition, E_g of In₂S₃ is calculated to be 2.17 eV. Ultraviolet photoelectron spectroscopy (UPS) was carried out to assess the band structure of the semiconductor photocatalyst. As presented in Figure S25a, the secondary cutoff binding energy ($$ ) is measured to be 17.01 eV for WO₃-Ov. By applying the equation: $$ , the work function ($$ ) of WO₃-Ov can be calculated to be 4.21 eV. Hence, the E_f of WO₃-Ov can be determined as −0.19 V (vs. NHE). As shown in Figure S25b, $$ of In₂S₃ was measured to be 17.36 eV, $$ and E_f of In₂S₃ can be estimated to be 3.86 eV and −0.54 V (vs. NHE), respectively. Mott–Schottky plots show positive slopes, indicating the n-type behavior of WO₃-Ov (Figure S26a) and In₂S₃ (Figure S26b).³² The conduction band potential (E_CB) can be evaluated by Mott–Schottky plots. The flat band potentials (E_fb) of WO₃-Ov and In₂S₃ are estimated to be −0.09 and −0.65 V (vs. NHE), respectively.³³ It is confirmed by previous studies that E_CB of n-type semiconductor is ~0.2 V negative than the E_fb.²² Therefore, the E_CB of WO₃-Ov and In₂S₃ is estimated to be −0.29 and −0.85 V (vs. NHE), respectively.

Potential of valence band (E_VB) can be obtained according to the equation of E_g=E_CB - E_VB,³⁴ which can be estimated to be 2.42 and 1.32 V (vs. NHE) for WO₃-Ov and In₂S₃, respectively. Figure 3a displays the detailed band structures of WO₃-Ov and In₂S₃ photocatalysts. EPR measurements were performed to further probe the reactive species in photocatalytic reactions, where 5,5-dimethyl-l-pyrroline N-oxide (DMPO) was adopted as a spin-trapping agent. Under light irradiation, hydroxyl ( ⋅ OH) and superoxide ( ⋅ O₂⁻) radicals can be detected. As presented in Figure 3b and 3c, there is no signal observed under dark condition. Once irradiated by light, WO₃-Ov is more favorable for generating ⋅ OH⁻ rather than ⋅ O₂⁻, while In₂S₃ promotes the production of ⋅ O₂⁻ instead of ⋅ OH⁻. These can be attributed to the fact that the VB position of WO₃-Ov is more positive than the potential of H₂O/ ⋅ OH and CB position of In₂S₃ is more negative than the potential of O₂/ ⋅ O₂⁻.^{26, 35-37} In WO₃-Ov/In₂S₃ heterojunction structure, the recombination of photogenerated holes and electrons is inhabited effectively, which enables higher amount of photogenerated holes and electrons to oxidize H₂O and reduce O₂, respectively, causing the enhanced signal of ⋅ O₂⁻ and ⋅ OH compared with the individual photocatalysts.^{38, 39} Moreover, in situ irradiation X-ray photoelectron spectroscopy (ISI-XPS) directly shows the transfer of photogenerated electrons from WO₃-Ov to In₂S₃ under light irradiation (Figure S27a and S27b), the detailed analysis can be found in Supporting Information. The results of EPR and ISI-XPS strongly verify the type of S-scheme heterojunction of WO₃-Ov/In₂S₃. The charge transfer route in this S-scheme heterojunction can be proposed based on the above experiments and calculation results. As shown in Figure 3d, when WO₃-Ov and In₂S₃ contact together, an intimate interface forms. Due to the difference of E_f between the two semiconductors, electrons will spontaneously transfer from In₂S₃ with higher E_f to WO₃-Ov with lower E_f until E_f at the interface reaching equilibrium. Due to the redistribution of electrons on the interface of WO₃-Ov and In₂S₃, the band alignment near the interface changes correspondingly, and an internal electric field is built. In details, WO₃-Ov with electrons accumulation shows downward band bending at the interface, while In₂S₃ with electron loss shows upward band bending. When irradiated by the light, electrons in VB absorb photon energy and jump into CB, causing the photogenerated electrons and holes in CB and VB, respectively. With the joint effect of internal electric field, band bending at the interface and Coulomb interaction, the photogenerated electrons in CB of WO₃-Ov tend to transfer to VB of In₂S₃ and combine with the holes. Through this pathway, the photogenerated carriers with low redox potentials merge, leaving the carriers with higher redox abilities separated apart, which significantly improves the efficiency of photocatalysis. Band structure of WO₃/In₂S₃ is analyzed in details in the Supporting Information (Figure S28 and S29).

Photocatalytic and Photoelectrochemical Performance

To evaluate the photocatalytic performance of the heterojunction, Rhodamine B (RhB) degradation property was tested. As shown in Figure 4a, under the irradiation of visible light, WO₃-Ov/In₂S₃ composites show significantly enhanced RhB photocatalytic degradation rate compared with pristine In₂S₃. Furthermore, the pseudo-first-order equation was utilized to quantitatively evaluate the degradation kinetics of RhB and the fitting curves are illustrated in Figure 4b. It can be seen that W_OvIn2 exhibits the highest K values (0.071) toward RhB degradation, followed by W_OvIn3 (0.058) and W_OvIn1 (0.044). Notably, W_OvIn2 shows the optimal performance, which is 8-folds higher than pristine In₂S₃. The photocatalytic RhB degradation performance of WO₃/In₂S₃ hybridizations is also provided in Figure S30. Pseudo first-order kinetics fitting of RhB degradation of WO₃/In₂S₃ composites is depicted in Figure S31. In order to compare the rate constants (K, min⁻¹) for WO₃-Ov/In₂S₃ and WO₃/In₂S₃, a histogram is depicted in Figure 4c. The photocatalytic RhB degradation abilities of the WO₃-Ov/In₂S₃ heterojunctions are higher than those of WO₃/In₂S₃ composites with the same In₂S₃ content. It can be presumed that the composites with specific electron transfer channels can remarkably improve the photocatalytic performance. Afterwards, the photocatalytic performance of W_OvIn2 was tested for 4 cycles to study its stability. Figure 4d demonstrates that there is no decay during the 4 cycles, revealing the high stability of W_OvIn2 S-scheme photocatalyst. The Brunauer–Emmett–Teller (BET) surface areas of all the photocatalysts are summarized in Table S5. K values after normalizing the specific surface are demonstrated in Figure S32 to eliminate the effect of adsorption ability. It can be seen that the creation of Ovs boosts the intrinsic photocatalytic degradation performance and W_OvIn2 shows the optimal performance. Furthermore, the photocatalytic phenol degradation performance of the as prepared samples was investigated (Figure S33), the detailed discussions are illustrated in Supporting Information. Transient photocurrent response and electrochemical impedance spectroscopy (EIS) were carried out to study charge transfer and separation efficiency of the photocatalysts. Charge transfer resistance can be evaluated by the semicircle diameter of EIS (Figure 4e). W_OvIn2 shows the lowest interfacial charge transfer resistance, verifying the efficient separation of photogenerated carriers. As shown in Figure 4f, the photocurrent density of WO₃-Ov/In₂S₃ hybridizations is higher than that of In₂S₃ under light irradiation, indicating a higher light harvesting ability and lower carrier recombination rate. Transient photocurrent response of WO₃/In₂S₃ (Figure S34) shows that WIn2 possesses the highest photo response current among WO₃/In₂S₃ composites. Photoluminescence (PL) spectra were collected to investigate the charge transfer and separation kinetics of the photocatalysts. As shown in Figure S35, compared with pristine In₂S₃, the PL emission intensity peaks of WO₃-Ov/In₂S₃ quenches, confirming an enhanced photogenerated carrier separation in WO₃-Ov/In₂S₃.²² W_OvIn2 shows the weakest peak intensity, which can be attributed to the fast photogenerated electrons transfer from WO₃-Ov to In₂S₃ at the interface of the W_OvIn2. It can be explained that the fast electron transfer channel at the heterointerface of WO₃-Ov and In₂S₃ builds a more efficient nonradiative decay pathway.

Density Functional Theory (DFT) Calculations

In order to unveil the effect of Ovs in the photocatalytic reaction, the related DFT calculation was carried out. As shown in Figure 5a, the partial density of states (PDOS) of WO₃-Ov at E_f demonstrates increment compared with WO₃ without Ovs, and the metallicity of WO₃-Ov is enhanced. It can be manifested that the W-5d and S-3p orbital overlap and hybridize to form covalent bond. W−S chemical bonds at the heterointerface can act as specific “bridge” to accelerate charge transfer between In₂S₃ and WO₃-Ov. Moreover, to visualize the charge transfer at the WO₃/In₂S₃ interface, the calculated charge density difference was employed to trace the electron transfer behavior. The charge accumulation and depletion are denoted in yellow and blue regions, as shown in Figure 5b and Figure S36a, where substantial charge transfer at the WO₃-Ov /In₂S₃ interface can be observed.²² From the Bader charge analysis,⁴⁰ In₂S₃ transfers 1.342 e⁻ to WO₃-Ov, verifying that the charge transfer at the interfaces makes the heterostructure structures tighter, and is consistent with the experimental results. The planar-averaged differential charge density as a function of z for WO₃-Ov/In₂S₃ is depicted in Figure S36b. The charge transfer at the heterointerface would subsequently induce the internal electric field in WO₃-Ov/In₂S₃ heterostructure and further efficiently steer the separation of photogenerated charge carriers. In order to further investigate the spatial distribution of photogenerated electrons and holes and visualize the impact of surface Ovs, additional first-principles calculations were conducted. Our findings indicate that WO₃-Ov generates photogenerated electrons from CB, which can subsequently be transferred to VB of In₂S₃ to combine with the photogenerated holes. The system of WO₃-Ov/In₂S₃ exhibits spatial separation between the valence band maximum (VBM, Figure 5c) and conduction band minimum (CBM, Figure 5d), effectively suppressing electron-hole recombination. Compared with band-decomposed charge density distribution of WO₃/In₂S₃ (Figure S37), the presence of Ovs leads to a greater electron localization of both VBM and CBM in the interface layer of the heterojunction, thereby forming a rapid charge transport channel between In₂S₃ and WO₃-Ov.

Fs-TAS Monitor Photogenerated Charge Dynamics

Furthermore, fs-TAS was performed to study the carrier dynamics (excited at 340 nm). As shown in Figure 6a, WO₃ exhibits a distinct positive photoinduced absorption band (400–500 nm), which can be attributed to the signal from photogenerated holes.^{41, 42} Compared with pristine WO₃, fs-TAS spectra of WO₃-Ov show significant difference (Figure 6b) with the appearance of negative absorption bands above 500 nm. The negative absorption bands result from the re-trapping of the photoexcited electrons from CB to vacancy centers,⁴² indicating the kinetic behavior of photogenerated carriers can be effectively steered by Ovs defects. The fs-TAS spectra of W_OvIn2 are shown in Figure 6c. The intensity of negative photobleaching signal gradually increases in 0.5 ps, then decays within 5 ps, which is related to the kinetic behaviors of photogenerated electrons. The fs-TAS spectra of WIn2 demonstrate difference with W_OvIn2 (Figure 6d). Meanwhile, the pseudo color TA plots of WO₃, WO₃-Ov, W_OvIn2 and WIn2 are shown in Figure 6e, respectively. Furthermore, the decay curves of WO₃-Ov and W_OvIn2 at 598 nm were recorded and fitted by a two-exponential function to investigate the kinetics of photogenerated electrons (Figure S38), where the short time τ₁ and the long time τ₂ correspond to photogenerated electrons trapped by the shallow and deep trapping states, respectively.^{43, 44} Compared with WO₃-Ov (τ₁=0.65 ps, τ₂=6.03 ps), the lifetime of W_OvIn2 decreases (τ₁=0.39 ps, τ₂=3.16 ps), which can be inferred that an additional channel for electron transfer exists at the interface between WO₃-Ov and In₂S₃, accelerating the relaxation process of electrons. To estimate the decay kinetics of photogenerated holes, time profiles of fs-TAS at 428 nm were probed. Compared with WO₃-Ov (τ_ave=194.4 ps) (Figure 6f), the average decay lifetime (τ_ave) of W_OvIn2 is prolonged to 528.1 ps (Figure 6g), suggesting that photogenerated holes in the composite possess much longer lifetime and higher priority to participate in photo oxidation reactions.⁴⁵ As comparison, the τ_ave of WIn2 is 265.9 ps (Figure 6h), shorter than that of W_OvIn2, further suggesting the creation of Ovs prolongs the lifetime of the photo-induced holes, thus favoring the improvement in photocatalytic activity. The fitting results are summarized in Table S6 and S7, which confirms the formation of S-scheme charge transfer between WO₃-Ov and In₂S₃. Based on the above analysis, the mechanism of Ovs in the S-scheme heterojunction can be proposed, as depicted in Figure 6i. Ovs can generate impurity trap states in WO₃-Ov which partially trap photogenerated electrons, thus inhibit direct recombination of the photogenerated electrons and holes. After hybridizing with In₂S₃, a new interfacial electron transfer pathway may be generated, the electron transfer and migration can be promoted by the internal electric field at the interface.⁴⁶ The rapid electron trapping as well as the fast charge separation at the interface of WO₃-Ov/In₂S₃ suppress electron-hole recombination. In addition, the existence of Ovs narrows the band gap of WO₃-Ov, resulting in the enhancement of light absorption and the increased number of photogenerated carriers to participate the reaction.