Enhanced Photocatalytic Activity of BiVO4/Bi2S3/SnS2 Heterojunction under Visible Light

Heterojunction photocatalysts have attracted a significant amount of attention due to their advantages over a single photocatalyst and, particularly, their superior spatial charge separation. Herein, the BiVO4/Bi2S3/SnS2 heterojunction was synthesized via solvothermal synthesis with different ratios of BiVO4 to SnS2. The photodegradation rate of the 0.03 BiVO4/SnS2 sample for rhodamine B removal is 2.3 times or 2.9 times greater than that of a single SnS2 or BiVO4, respectively. The chemical bond between photocatalysts is confirmed by X-ray photoelectron spectroscopy (XPS), and the synchronized shift observed in binding energies strongly indicates the electron screening effect at the heterojunction. A Z-scheme model is proposed to explain charge transfer pathway in the system, in which the formation of Bi2S3 plays a crucial role in the enhanced photocatalytic performance of the heterojunction.


Introduction
In recent years, a tremendous effort has been made to develop efficient photocatalysts due to their potential applications in water splitting [1][2][3], CO 2 reduction [4,5], and pollutant removal [6][7][8]. A good photocatalyst needs a suitable bandgap to work in the visible light region, in addition to satisfying other important criteria, e.g., abundance, low cost, nontoxicity, low electron-hole recombination rate, and stability. Photocatalysts such as SnS 2 and BiVO 4 fit most of these conditions; in particular, they can absorb visible light because their bandgaps are 2.18-2.4 eV [1,9]. Although both materials have many advantages as a single photocatalyst, incorporating them with different photocatalysts to form a heterojunction has proved very effective to attain superior photocatalytic activity [1,[10][11][12]. The interface can act as a bridge allowing electrons to migrate between photocatalysts, which, in turn, suppress the photoinduced electron-hole recombination and improve the performance of photocatalysts [11,13]. In addition, the heterojunction may also induce a Z-scheme system in which the redox ability is enhanced via coupling two or more narrow-bandgap semiconductors [1,14].
In this study, the BiVO4/Bi2S3/SnS2 heterojunction was thus synthesized by combining SnS2 and BiVO4 via ultrasonic mixing and solvothermal synthesis. The formation of Bi2S3 during the synthesis would then constitute a ternary heterojunction with SnS2 and BiVO4. Various molar ratios of BiVO4 to SnS2 were prepared to investigate the properties of three-phase photocatalysts (BiVO4, Bi2S3, and SnS2) and to optimize the amount of BiVO4 in the composite. Photocatalytic performance of the composite was compared with pure BiVO4 and SnS2, and the charge transfer pathway in the heterojunction system was examined. A Z-scheme model was then proposed to consistently explain the enhanced photocatalytic activity of BiVO4/Bi2S3/SnS2 heterojunctions.

Crystal Structure of the Samples
The crystalline phases of as-prepared samples with different ratios of BiVO4 to SnS2 were investigated through XRD (X-ray diffraction) patterns as shown in Figure 1. The characteristic peaks at 14.9°, 28.3°, 32.3°, 49.5°, and 52.4° correspond to (001), (100), (101), (110), and (111) planes of hexagonal SnS2, respectively (ICDD PDF No. 00-023-0677). Other visible peaks can be observed at 18.7°, 19.0°, 28.8°, and 30.5°, assigned to (110), (011), (1 21), and (040) planes of monoclinic BiVO4, respectively (ICDD PDF No. 00-014-0688). In addition to these obvious peaks, there are four small peaks at 22.3°, 23.6°, 24.9°, and 25.2°, which can be assigned to (202), (101), (130), and (310) planes of orthorhombic Bi2S3, respectively (ICDD PDF No. 01-089-8965). The XRD patterns of individual synthesized materials (BiVO4, SnS2, and Bi2S3) are shown in Figure S1. These characteristic diffraction peaks confirm three crystal phases of SnS2, B2S3, and BiVO4 in the composites. As the content of BiVO4 during the synthesis is reduced, the diffraction intensities of BiVO4 and Bi2S3 decrease and almost disappear for the 0.01 BiVO4/SnS2 sample due to the very low content of BiVO4 and Bi2S3 in the composite. According to the result, the content of Bi2S3 depends on the initial concentration of BiVO4 during the hydrothermal synthesis.  Figure 2a,b reveals agglomerates of SnS2 that were composed of ultrafine nanoparticles with diameter of approximately 50 nm. On the other hand, BiVO4 exhibits micron-sized polyhedral particles with a smooth surface, i.e., particular features of microcrystals with many facets, as observed in Figure 2c. The ultrafine nanoparticles of SnS2 and microparticles of BiVO4 explain a large difference  Figure 2a,b reveals agglomerates of SnS 2 that were composed of ultrafine nanoparticles with diameter of approximately 50 nm. On the other hand, BiVO 4 exhibits micron-sized polyhedral particles with a smooth surface, i.e., particular features of microcrystals with many facets, as observed in Figure 2c. The ultrafine nanoparticles of SnS 2 and microparticles of BiVO 4 explain a large difference (more than 100 times) in the specific surface area of SnS 2 and BiVO 4 in Table 1. Furthermore, Figure 2d shows a micrograph image of the 0.03 BiVO 4 /SnS 2 composite, in which the nanoparticles of SnS 2 are attached to, and cover, the particles of BiVO 4 . Based on the EDS elemental mapping of 0.03 BiVO 4 /SnS 2 in Figure 3, the elements of Sn, S, Bi, V, and O are uniformly dispersed in the composite, indicating good distribution of BiVO 4 and SnS 2 after the incorporation.

Morphology and Microstructure of the Samples
Catalysts 2020, 10, x FOR PEER REVIEW 3 of 13 (more than 100 times) in the specific surface area of SnS2 and BiVO4 in Table 1. Furthermore, Figure  2d shows a micrograph image of the 0.03 BiVO4/SnS2 composite, in which the nanoparticles of SnS2 are attached to, and cover, the particles of BiVO4. Based on the EDS elemental mapping of 0.03 BiVO4/SnS2 in Figure 3, the elements of Sn, S, Bi, V, and O are uniformly dispersed in the composite, indicating good distribution of BiVO4 and SnS2 after the incorporation.     (more than 100 times) in the specific surface area of SnS2 and BiVO4 in Table 1. Furthermore, Figure  2d shows a micrograph image of the 0.03 BiVO4/SnS2 composite, in which the nanoparticles of SnS2 are attached to, and cover, the particles of BiVO4. Based on the EDS elemental mapping of 0.03 BiVO4/SnS2 in Figure 3, the elements of Sn, S, Bi, V, and O are uniformly dispersed in the composite, indicating good distribution of BiVO4 and SnS2 after the incorporation.

Optical Absorption Properties
UV-Vis diffuse reflectance spectra (DRS) were used to study optical properties of as-prepared samples. The reflectance spectra were converted to absorption spectra via the Kubelka-Munk function as shown below, in Equation (1) [19,22]: where F(R), α, S, and R are the Kubelka-Munk function, absorption coefficient, scattering coefficient, and reflectance, respectively. As illustrated in Figure 4a, all composites exhibit absorption edges around the wavelength of 550 nm, indicating that they are photoactive in the visible light region. The curves of 0.06 BiVO 4 /SnS 2 and 0.12 BiVO 4 /SnS 2 are noticeably different from the remainder as their absorption regions extend beyond 550 nm. This could be attributed to an increase in the content of Bi 2 S 3 in the composite.

Optical Absorption Properties
UV-Vis diffuse reflectance spectra (DRS) were used to study optical properties of as-prepared samples. The reflectance spectra were converted to absorption spectra via the Kubelka-Munk function as shown below, in Equation (1) [19,22]: where F(R), α, S, and R are the Kubelka-Munk function, absorption coefficient, scattering coefficient, and reflectance, respectively. As illustrated in Figure   The bandgaps (Eg) can be estimated using the following formula: where hυ and B are incident photon energy and a constant associated with the material, respectively. The coefficient α can be obtained via the Kubelka-Munk function from Equation (1). The F(R) is commonly used to replace α in the Tauc plot ((αhυ) 2/n vs. hυ) [11,19,23]. In addition, SnS2, Bi2S3, and BiVO4 have been reported to be direct transition semiconductors, thus n = 1 is used in the equation [19,24]. Thus, the (F(R)hυ) 2 vs. hυ graph can be plotted as in Figure 4b, and the Eg values are evaluated by extrapolating the linear part of the curve to intercept the F(R) = 0 line. The Eg values of BiVO4, SnS2, and Bi2S3 are accordingly estimated to be 2.38, 2.18, and 1.46 eV, respectively, which are similar to those in previous reports [22,25]. In addition, the effective Eg values of 0.01 BiVO4/SnS2, 0.03 BiVO4/SnS2, 0.06 BiVO4/SnS2, and 0.12 BiVO4/SnS2 are estimated to be 2.17, 2.17, 2.11, and 2.04 eV, respectively. It appears that the apparent Eg value of the composite is decreased with an increase in BiVO4. This bandgap modification may be ascribed to small composition variation of SnS2, BiVO4, and Bi2S3, especially near the interface. This phenomenon is also reported in previous publications [3,16,19].

X-Ray Photoelectron Spectroscopy (XPS)
The XPS survey spectrum of 0.03 BiVO4/SnS2 composite is shown in Figure 5a, demonstrating that it consists of Sn, S, Bi, V, and O elements in agreement with the EDS result. The chemical state of each element, investigated via the high-resolution XPS analyses, is shown in Figure 5b,c. As observed in Figure 5b, two separate spectra with peaks at 494.89 and 486.49 eV with an energy difference of 8.4 eV correspond to Sn 3d3/2 and Sn 3d5/2 of Sn 4+ , respectively, for a common SnS2 [4,10,19]. On the other hand, for the 0.03 BiVO4/SnS2 composite, it is noted that both Sn 3d peaks shift to the higher binding The bandgaps (E g ) can be estimated using the following formula: where hυ and B are incident photon energy and a constant associated with the material, respectively. The coefficient α can be obtained via the Kubelka-Munk function from Equation (1). The F(R) is commonly used to replace α in the Tauc plot ((αhυ) 2/n vs. hυ) [11,19,23]. In addition, SnS 2 , Bi 2 S 3 , and BiVO 4 have been reported to be direct transition semiconductors, thus n = 1 is used in the equation [19,24]. Thus, the (F(R)hυ) 2 vs. hυ graph can be plotted as in Figure 4b, and the E g values are evaluated by extrapolating the linear part of the curve to intercept the F(R) = 0 line. The E g values of BiVO 4 , SnS 2 , and Bi 2 S 3 are accordingly estimated to be 2.38, 2.18, and 1.46 eV, respectively, which are similar to those in previous reports [22,25]. In addition, the effective E g values of 0.01 BiVO 4 /SnS 2 , 0.03 BiVO 4 /SnS 2 , 0.06 BiVO 4 /SnS 2 , and 0.12 BiVO 4 /SnS 2 are estimated to be 2.17, 2.17, 2.11, and 2.04 eV, respectively. It appears that the apparent E g value of the composite is decreased with an increase in BiVO 4 . This bandgap modification may be ascribed to small composition variation of SnS 2 , BiVO 4 , and Bi 2 S 3 , especially near the interface. This phenomenon is also reported in previous publications [3,16,19].

X-Ray Photoelectron Spectroscopy (XPS)
The XPS survey spectrum of 0.03 BiVO 4 /SnS 2 composite is shown in Figure 5a, demonstrating that it consists of Sn, S, Bi, V, and O elements in agreement with the EDS result. The chemical state of each element, investigated via the high-resolution XPS analyses, is shown in Figure 5b,c. As observed in Figure 5b, two separate spectra with peaks at 494.89 and 486.49 eV with an energy difference of 8.4 eV correspond to Sn 3d 3/2 and Sn 3d 5/2 of Sn 4+ , respectively, for a common SnS 2 [4,10,19]. On the other hand, for the 0.03 BiVO 4 /SnS 2 composite, it is noted that both Sn 3d peaks shift to the higher binding energy side by about 0.30-0.35 eV, compared with those of pure SnS 2 . The blueshift observed in the Sn 3d spectra is also found in, and synchronous with, the S 2p spectra.
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 13 energy side by about 0.30-0.35 eV, compared with those of pure SnS2. The blueshift observed in the Sn 3d spectra is also found in, and synchronous with, the S 2p spectra. As shown in Figure 5c, the S 2p3/2 (161.34 eV) and S 2p1/2 (162.52 eV) peaks, corresponding to typical Sn-S bonds of pure SnS2 [4,9,10], also exhibit a blueshift in the binding energy by 0.1-0.3 eV for the 0.03 BiVO4/SnS2 composite. Furthermore, the synchronized binding energy shift is also observed, with an opposite direction, for the Bi 4f spectra. The spectral peaks for the 0.03 BiVO4/SnS2 composite at 158.75 and 164.34 eV, which belong to Bi 4f7/2 and Bi 4f5/2 of BiVO4, respectively, exhibit the redshift, by 0.2-0.5 eV, compared with Bi 4f peaks (159.23 and 164.53 eV) of pure BiVO4 [15]. Importantly, the synchronous phenomena are reproducible.
The shift of binding energy in the core-level of XPS spectra should be largely attributed to the change in electron concentration of semiconductors due to the interaction between SnS2 and BiVO4 [26]. As the electron concentration decreases, the binding energy of the semiconductor increases, and vice versa, due to the so-called electron screening effect [27]. When a heterojunction is formed through chemical interaction, the Fermi energy levels of both materials are adjusted, allowing the electron transfer between materials to achieve equilibrium [10,26,27]. In the case of the 0.03 BiVO4/SnS2 composite, it implies that SnS2 has a higher Fermi energy level than that of BiVO4, thus the electron migrates from SnS2 to BiVO4. This consequently reduces the electron concentration of SnS2 and increases that of BiVO4. As a result, the Sn 3d and S 2p spectra of the composite exhibit the blueshift, while Bi 4f spectra exhibit the redshift.
Furthermore, the broad peaks of the Bi 4f exhibited in the 0.03 BiVO4/SnS2 composite may be caused by the presence of Bi2S3. The Bi 4f7/2 and Bi 4f5/2 spectral peaks of a typical Bi2S3 are reported to be at 158.4 and 163.8 eV [28], respectively, which are slightly lower than those of BiVO4. Moreover, the S 2p peaks at 161.45 and 162.81 eV in the 0.03 BiVO4/SnS2 composite could also correspond to S 2− of Bi2S3 [29]. The results from XPS spectra and XRD patterns (Section 2.1) strongly suggest the coexistence of SnS2, Bi2S3, and BiVO4 in the composite. Additionally, the shifts in binding energy As shown in Figure 5c, the S 2p 3/2 (161.34 eV) and S 2p 1/2 (162.52 eV) peaks, corresponding to typical Sn-S bonds of pure SnS 2 [4,9,10], also exhibit a blueshift in the binding energy by 0.1-0.3 eV for the 0.03 BiVO 4 /SnS 2 composite. Furthermore, the synchronized binding energy shift is also observed, with an opposite direction, for the Bi 4f spectra. The spectral peaks for the 0.03 BiVO 4 /SnS 2 composite at 158.75 and 164.34 eV, which belong to Bi 4f 7/2 and Bi 4f 5/2 of BiVO 4 , respectively, exhibit the redshift, by 0.2-0.5 eV, compared with Bi 4f peaks (159.23 and 164.53 eV) of pure BiVO 4 [15]. Importantly, the synchronous phenomena are reproducible.
The shift of binding energy in the core-level of XPS spectra should be largely attributed to the change in electron concentration of semiconductors due to the interaction between SnS 2 and BiVO 4 [26]. As the electron concentration decreases, the binding energy of the semiconductor increases, and vice versa, due to the so-called electron screening effect [27]. When a heterojunction is formed through chemical interaction, the Fermi energy levels of both materials are adjusted, allowing the electron transfer between materials to achieve equilibrium [10,26,27]. In the case of the 0.03 BiVO 4 /SnS 2 composite, it implies that SnS 2 has a higher Fermi energy level than that of BiVO 4 , thus the electron migrates from SnS 2 to BiVO 4 . This consequently reduces the electron concentration of SnS 2 and increases that of BiVO 4 . As a result, the Sn 3d and S 2p spectra of the composite exhibit the blueshift, while Bi 4f spectra exhibit the redshift.
Furthermore, the broad peaks of the Bi 4f exhibited in the 0.03 BiVO 4 /SnS 2 composite may be caused by the presence of Bi 2 S 3 . The Bi 4f 7/2 and Bi 4f 5/2 spectral peaks of a typical Bi 2 S 3 are reported to be at 158.4 and 163.8 eV [28], respectively, which are slightly lower than those of BiVO 4 . Moreover, the S 2p peaks at 161.45 and 162.81 eV in the 0.03 BiVO 4 /SnS 2 composite could also correspond to S 2− of Bi 2 S 3 [29]. The results from XPS spectra and XRD patterns (Section 2.1) strongly suggest the coexistence of SnS 2 , Bi 2 S 3 , and BiVO 4 in the composite. Additionally, the shifts in binding energy indicate that the interaction among SnS 2 , Bi 2 S 3 , and BiVO 4 in the 0.03 BiVO 4 /SnS 2 composite is a chemical bonding rather than a physical contact only.

Photocatalytic Activity
Photocatalytic activity was evaluated via concentration reduction in RhB (rhodamine B) dye solution over time, as shown in Figure 6a, where C 0 is the initial RhB concentration at the time of light irradiation (t = 0), and C t is the RhB concentration at any sampling time during irradiation. Blank RhB (a control test) in Figure 6a demonstrates that the RhB dye solution was stable under this test condition without any photocatalyst in the solution and that the photolysis of RhB dye was negligible over 240 min of visible light irradiation. Conversely, the concentration of RhB decreased over time for all other RhB solutions containing any photocatalyst, evidencing that photocatalysts used in this study respond to visible light (λ > 420 nm). Importantly, all composited powders exhibit higher photocatalytic degradation of RhB than that of bare SnS 2 , Bi 2 S 3 , or BiVO 4 under visible light irradiation for over 240 min. The temporal evolution of RhB absorption spectra for each photocatalyst can be seen in Figure S2. indicate that the interaction among SnS2, Bi2S3, and BiVO4 in the 0.03 BiVO4/SnS2 composite is a chemical bonding rather than a physical contact only.

Photocatalytic Activity
Photocatalytic activity was evaluated via concentration reduction in RhB (rhodamine B) dye solution over time, as shown in Figure 6a, where C0 is the initial RhB concentration at the time of light irradiation (t = 0), and Ct is the RhB concentration at any sampling time during irradiation. Blank RhB (a control test) in Figure 6a demonstrates that the RhB dye solution was stable under this test condition without any photocatalyst in the solution and that the photolysis of RhB dye was negligible over 240 min of visible light irradiation. Conversely, the concentration of RhB decreased over time for all other RhB solutions containing any photocatalyst, evidencing that photocatalysts used in this study respond to visible light (λ > 420 nm). Importantly, all composited powders exhibit higher photocatalytic degradation of RhB than that of bare SnS2, Bi2S3, or BiVO4 under visible light irradiation for over 240 min. The temporal evolution of RhB absorption spectra for each photocatalyst can be seen in Figure S2. The degradation rates were compared using a photodegradation rate constant (k), assuming a pseudo-first-order reaction model, ln(C0/Ct) = kt, as shown in Figure 6b. The kinetic model fits the data well, and the k value of the 0.03 BiVO4/SnS2 sample is about 2.3 times or 2.9 times greater than that of single SnS2 or BiVO4, respectively ( Table 1). The photocatalytic activity of the composite is enhanced as the content of BiVO4 is increased from 0.01 to 0.03, whereas further addition of BiVO4 leads to reduction in photocatalytic activity of the composites (0.06 BiVO4/SnS2 and 0.12 BiVO4/SnS2). It is thus reasonable to consider that an optimal molar ratio exists at a relatively low concentration of BiVO4. One possible reason could be the variation in the "effective" specific surface area. As shown in Table 1, the specific surface area of the composites mostly decreases with an addition of BiVO4, ascribed to the significantly small value of bare BiVO4 (0.6 m 2 /g). Furthermore, the interfacial reaction and agglomeration of powders may affect the effective active area for photocatalytic activity. As The degradation rates were compared using a photodegradation rate constant (k), assuming a pseudo-first-order reaction model, ln(C 0 /C t ) = kt, as shown in Figure 6b. The kinetic model fits the data well, and the k value of the 0.03 BiVO 4 /SnS 2 sample is about 2.3 times or 2.9 times greater than that of single SnS 2 or BiVO 4 , respectively ( Table 1). The photocatalytic activity of the composite is enhanced as the content of BiVO 4 is increased from 0.01 to 0.03, whereas further addition of BiVO 4 leads to reduction in photocatalytic activity of the composites (0.06 BiVO 4 /SnS 2 and 0.12 BiVO 4 /SnS 2 ). It is thus reasonable to consider that an optimal molar ratio exists at a relatively low concentration of BiVO 4 . One possible reason could be the variation in the "effective" specific surface area. As shown in Table 1, the specific surface area of the composites mostly decreases with an addition of BiVO 4 , ascribed to the significantly small value of bare BiVO 4 (0.6 m 2 /g). Furthermore, the interfacial reaction and agglomeration of powders may affect the effective active area for photocatalytic activity. As previously mentioned, the formation of Bi 2 S 3 is intensive as the amount of BiVO 4 in the composite is increased. Although a small amount of Bi 2 S 3 would be favorable for the composite, excess content may lead to deterioration of the photocatalytic performance because Bi 2 S 3 has low photocatalytic activity of RhB degradation compared to SnS 2 and BiVO 4 , possibly due to its rapid recombination rate of the photogenerated electron-hole [17]. Thus, excessive Bi 2 S 3 could act as a recombination center for electron-hole pairs in the composite, which is similar to the findings of previous studies [13,19].
In addition to the photocatalytic activity of the material, stability and reusability are also important to determine its value in practical applications. Thus, the reusability test was also conducted by collecting the used sample (0.03 BiVO 4 /SnS 2 ) after irradiation via centrifuge and putting it back into the fresh RhB solution. The same procedure for the photocatalytic activity measurement was repeated in this reusability test. Figure 6c shows the degradation efficiencies of 0.03 BiVO 4 /SnS 2 composite over four cycles, illustrating stability and reusability of the 0.03 BiVO 4 /SnS 2 composite. A slight loss (around 6%) of the degradation efficiency in the 4th run could be due to a loss of photocatalyst during the recovery process. Furthermore, the stability of the composite was also confirmed by XRD pattern before and after irradiation ( Figure S3), in which the crystal structure of 0.03 BiVO 4 /SnS 2 was not significantly altered after RhB degradation process.
To determine main reactive species responsible for RhB photodegradation by 0.03 BiVO 4 /SnS 2 during irradiation, scavenger tests were conducted using various types of scavengers. In this study, methanol (10 mM) [6], ascorbic acid (AA, 10 mM) [30], and isopropyl alcohol (IPA, 10 mM) [16] were used as the scavengers for superoxide radicals (•O 2 − ), holes (h + ), and hydroxyl radicals (•OH), respectively. As shown in Figure 6d, an addition of methanol into the system has only a slight effect on RhB degradation, indicating that •O 2 − radicals play a minor role in the degradation process. Conversely, AA (h + scavenger) and IPA (•OH scavenger) greatly suppress photocatalytic RhB degradation. The result strongly suggests that h + and •OH radicals are the main reactive species for the RhB photodegradation process with the 0.03 BiVO 4 /SnS 2 photocatalyst under visible light (λ > 420 nm) irradiation.

Possible Mechanism for Photocatalytic Activity Enhancement of BiVO 4 /Bi 2 S 3 /SnS 2 Heterojunction
A plausible photocatalytic mechanism of the BiVO 4 /Bi 2 S 3 /SnS 2 heterojunction could be explained by understanding the band energy structures of the photocatalysts as illustrated in Figure 7. The empirical Equations (3) and (4) are often used to estimate the band edge of the semiconductor such as conduction band edge potential (E CB ) and valence band edge potential (E VB ) [15,19]: interface, and their Fermi levels are aligned after contact; as a result, their band energy structures are accordingly adjusted, creating a heterojunction as depicted in Figure 7 [12,16]. To understand the charge-transfer pathway in the BiVO4/Bi2S3/SnS2 heterojunction system, we explore and examine two possible models for the charge transfer process as proposed in Figure 8. All of the photocatalysts in both models are capable of generating electron-hole pairs upon visible light illumination due to their narrow bandgaps. In model A, the photogenerated electrons (e -) would migrate from the CB (conduction band) of Bi2S3 to CBs of SnS2 and BiVO4. Thus, •O2 − radicals would not be produced by the heterojunction since the CBs of both SnS2 and BiVO4 are incapable of reducing dissolved O2 to produce •O2 − because their CBs are less negative than −0.33 eV (O2/ •O2 − ) [16]. On the other hand, the generated holes (h + ) would move from the VBs of both SnS2 and BiVO4 to the VB of Bi2S3. In this scenario, the heterojunction also could not oxidize H2O to form •OH because VB of Bi2S3 is less positive than 2.72 eV (for •OH/H2O) [8,16]. It has been reported that •O2 − , •OH, and h + play major roles in the RhB photodegradation process [16,32,33]. Although model A might improve charge separation of photoinduced electrons and holes, it would be less efficient as the composite would be unable to produce •O2 − and •OH to degrade the RhB dye. Moreover, the model is inconsistent with our trapping experiment (our scavenger test in Section 2.5) in which •O2 − and •OH were observed in the RhB degradation process.  Fermi level positions of SnS 2 , Bi 2 S 3 , and BiVO 4 with respect to their VBs (valence bands) can be determined from the potential difference between VBM (VB maximum) and Fermi level from the valence band XPS spectra in Figure S4 [12]. The difference between the Fermi level and VBMs of SnS 2 , Bi 2 S 3 , and BiVO 4 is found to be 1.3, 0.3, and 2 eV, respectively. The result indicates that SnS 2 and BiVO 4 are n-type semiconductors, whereas Bi 2 S 3 is a p-type semiconductor, which is in agreement with prior studies [2,6,16]. Our XPS results suggest that the photocatalysts are chemically bonded at the interface, and their Fermi levels are aligned after contact; as a result, their band energy structures are accordingly adjusted, creating a heterojunction as depicted in Figure 7 [12,16].
To understand the charge-transfer pathway in the BiVO 4 /Bi 2 S 3 /SnS 2 heterojunction system, we explore and examine two possible models for the charge transfer process as proposed in Figure 8.  To understand the charge-transfer pathway in the BiVO4/Bi2S3/SnS2 heterojunction system, we explore and examine two possible models for the charge transfer process as proposed in Figure 8. All of the photocatalysts in both models are capable of generating electron-hole pairs upon visible light illumination due to their narrow bandgaps. In model A, the photogenerated electrons (e -) would migrate from the CB (conduction band) of Bi2S3 to CBs of SnS2 and BiVO4. Thus, •O2 − radicals would not be produced by the heterojunction since the CBs of both SnS2 and BiVO4 are incapable of reducing dissolved O2 to produce •O2 − because their CBs are less negative than −0.33 eV (O2/ •O2 − ) [16]. On the other hand, the generated holes (h + ) would move from the VBs of both SnS2 and BiVO4 to the VB of Bi2S3. In this scenario, the heterojunction also could not oxidize H2O to form •OH because VB of Bi2S3 is less positive than 2.72 eV (for •OH/H2O) [8,16]. It has been reported that •O2 − , •OH, and h + play major roles in the RhB photodegradation process [16,32,33]. Although model A might improve charge separation of photoinduced electrons and holes, it would be less efficient as the composite would be unable to produce •O2 − and •OH to degrade the RhB dye. Moreover, the model is inconsistent with our trapping experiment (our scavenger test in Section 2.5) in which •O2 − and •OH were observed in the RhB degradation process.  A more plausible scenario might be that of model B in Figure 8, [16,32]. Thus, the Z-scheme model is consistent with the experimental result, and may be used to explain the charge-transfer pathway of the BiVO 4 /Bi 2 S 3 /SnS 2 heterojunction system. , > 95%) was purchased from Tokyo Chemical Industry (Tokyo, Japan). All chemicals were analytical grade and used without further purification. Deionized water (DI water) was obtained from the Direct-Q water purification system (Millipore).

Preparation of SnS 2
Preparation of SnS 2 was conducted via a typical synthesis method. The powder of SnCl 4 ·5H 2 O (5 mmol) was dissolved in 40 mL of 5% (v/v%) acetic acids under magnetic stirring. Then, 10 mmol amount of thioacetamide was added to the solution. After 30 min of vigorous stirring to achieve a homogeneous solution, the solution was transferred to a stainless-steel autoclave attached with a Teflon liner to fill 80% of its maximum capacity (50 mL). It was then put in a preheat electric oven at 150 • C for 12 h under autogenous pressure. After letting the autoclave cool to room temperature, SnS 2 precipitate was collected via centrifuge at 4000 rpm for 5 min. The SnS 2 was then washed several times with DI water and ethanol and dried at 90 • C overnight.

Preparation of BiVO 4
Typical hydrothermal synthesis was used to prepare BiVO 4 . The first solution was prepared by dissolving 2.43 g of Bi(NO 3 ) 3 ·5H 2 O in 20 mL of 2 M HNO 3 acid, and the second solution was made by dissolving an equimolar amount of NH 4 VO 3 in 2 M NaOH solution. Then, the second solution was poured drop by drop into the first solution. The clear solution turned yellow as BiVO 4 precipitate was formed. The mixture solution was continually stirred for another 10 min before adding 1 mL of acetic acid into the solution. After 1 h of stirring, the solution was transferred to the stainless-steel autoclave with a Teflon liner. The autoclave was heated at 180 • C for 24 h. Finally, BiVO 4 precipitate was collected, washed, and dried, following the same procedure described in Section 3.2.

Preparation of BiVO 4 /Bi 2 S 3 /SnS 2 Composites
The composite was prepared using an ultrasonic mixing and a solvothermal synthesis. A specific amount of BiVO 4 (0.01, 0.03, 0.06, and 0.12 mmol) was mixed with 1 mmol of SnS 2 in 40 mL of ethylene glycol. The mixture was thoroughly mixed via ultrasonication at 45 kHz for 1 h before transferring to the autoclave. It was then sealed and heated at 150 • C for 8 h. The final product of the composite was obtained through washing and drying, following the same procedure as stated in Sections 3.2 and 3.3. The composites with different ratios of BiVO 4 to SnS 2 were denoted, respectively, as 0.01 BiVO 4 /SnS 2 , 0.03 BiVO 4 /SnS 2 , 0.06 BiVO 4 /SnS 2 , and 0.12 BiVO 4 /SnS 2 . For reference, a bare Bi 2 S 3 was synthesized using solvothermal synthesis (detailed in the Supplementary Information).

Characterization
X-ray diffraction (XRD) measurement was conducted using a Rigaku RINT2100 at 40 kV and 30 mA with Cu Kα radiation (λ = 1.5418 Å), where the diffractogram was obtained via scanning the sample in a 2θ angle range from 10 • to 80 • . Microstructures and morphologies of the samples were investigated using an FE-SEM (Hitachi SU6600 Scanning Electron Microscope) equipped with Bruker EDX operated at 15 and 20 kV for SEM and EDX, respectively, where all samples were coated with Au via sputtering prior to the analyses. UV-Vis diffuse reflectance spectra (DRS) were obtained using a Lambda 750S UV/Vis/NIR Spectrophotometer with BaSO 4 as reference. X-ray photoelectron spectroscopy (XPS, JPS-9030 X-ray photoelectron spectrometer) was conducted with Mg Kα radiation using C 1s = 284.8 eV as reference. The Brunauer-Emmett-Teller (BET) specific surface area was measured and evaluated using nitrogen gas adsorption with a FlowSorb III 2305 Micromeritics Instrument (Shimadzu, Japan).

Photocatalytic Activity Measurement
Evaluation of photocatalytic activity was conducted via the photodegradation of RhB dye under visible light irradiation using a 500 W Xe lamp (Ushio, UXL-500D-O) equipped with a cutoff filter (λ > 420 nm). The intensity of light from the Xe lamp was 100 mW/cm 2 calibrated using a Spectroradiometer (S-2440 model II). After passing through the cutoff and water filters, the sample solution received around 40 mW/cm 2 of light intensity. In each measurement process, 30 mg of the photocatalyst was put into 80 mL of RhB solution (5 mg/L, 0.01 mmol/L) in a beaker with 100 mL capacity. The solution was magnetically stirred and maintained at 25 • C during irradiation. About 3 mL of sample solution was taken every 60 min, and the photocatalyst powder was filtered out by syringe filter (0.22 µm, PTFE). The absorbance of each RhB solution was measured using a UV-Vis spectrophotometer (Lambda 750S UV/Vis/NIR). Then, the concentration of RhB was determined from the absorbance intensity at λ max = 554 nm. Before irradiation, the establishment of adsorption-desorption equilibrium between photocatalyst and RhB solution was achieved to ensure an accurate result of photocatalytic activity.
Since SnS 2 exhibits exceptionally strong adsorptivity for RhB dye [23], an initial dye concentration (C 0 ) for each photocatalytic test would be different, depending on the ratio of BiVO 4 /SnS 2 composites, after achieving an adsorption-desorption equilibrium. To ensure that all samples had relatively similar initial concentrations at the start of illumination, all samples were subjected to the adsorption-desorption process twice. Each sample was added to a RhB solution, which was then agitated by sonication for 10 min and magnetically stirred for 60 min in the dark. Then, the sample was collected via centrifuge and put into a fresh RhB solution, where the adsorption-desorption procedure was performed again prior to the photocatalytic test. This devised method allowed an initial RhB concentration (C 0 ) to be similar for each photocatalytic test, using various composite ratios.

Conclusions
In summary, three-phase photocatalysts consisting of BiVO 4 , Bi 2 S 3 , and SnS 2 were prepared by chemical reaction between BiVO 4 and SnS 2 via solvothermal synthesis. The composite photocatalyst produced via a molar ratio of 1:0.03 (SnS 2 :BiVO 4 ) demonstrates the highest photocatalytic performance for RhB degradation among our prepared samples. The composite is proved to be stable after several cycles under visible light irradiation. The main reactive species for the photocatalytic degradation of RhB for 0.03 BiVO 4 /SnS 2 are h + and •OH, whereas •O 2 − also plays a minor role in the degradation process. The enhanced photocatalytic activity is attributed to the formation of Bi 2 S 3 , allowing a suitable condition for the electron pathway (Z-scheme) in the BiVO 4 /Bi 2 S 3 /SnS 2 heterojunction. We believe that our discovery of the beneficial formation of Bi 2 S 3 could encourage more research that focuses on materials prone to reaction with each other at elevated temperature and pressure. The finding may provide a different approach for preparation of ternary heterojunctions by taking advantage of the chemical reaction between combined photocatalysts.