Carbon Dioxide Photoreduction on the Bi2S3/MoS2 Catalyst

The photocatalytic activity of a material is contingent on efficient light absorption, fast electron excitation, and control of the recombination rate by effective charge separation. Inorganic materials manufactured in unique shapes via controlled synthesis can exhibit significantly improved properties. Here, n-type Bi2S3 nanorods (with good optical activity) were wrapped with two-dimensional (2D) p-type MoS2 sheets, which have good light absorption properties. The designed p-n junction Bi2S3/MoS2 composite exhibited enhanced light absorption over the entire wavelength range, and higher carbon dioxide adsorption capacity and photocurrent density compared to the single catalysts. Consequently, the activity of the 1Bi2S3/1MoS2 composite catalyst for the photocatalytic reduction of carbon dioxide was more than 20 times higher than that of the single catalysts under visible-light irradiation at ≤400 nm, with partial selectivity for CO conversion. This is attributed to the p-n heterojunction Bi2S3/MoS2 composite designed in this study, the high light absorption of n-Bi2S3, accelerated electron excitation, and the electron affinity of the 2D sheet-p-MoS2, which quickly absorbed excited electrons, resulting in effective charge separation. This ultimately improved the catalytic performance by continuously supplying catalytically active sites to the heterojunction interfaces.


Introduction
The greatest challenge for an eco-friendly CO 2 -to-fuel system, which produces fuel by photocatalytically reducing carbon dioxide under visible light, is the development of stable and highly efficient photocatalysts. Chalcogenides are currently highly topical as candidate materials for this purpose and extensive studies of these materials have been conducted. Chalcogenides are an attractive research theme because their physicochemical properties vary with their unique shape and size. Furthermore, the shape and size of these materials can be easily controlled by the synthesis methods [1,2]. In particular, chalcogenides have small band-gaps and large extinction coefficients, and can be utilized in various applications such as batteries [3], gas sensors [4], and photodetectors [5]. In recent years, due to their wide light absorption ability, the application of chalcogenides as photocatalysts that are active under visible-light irradiation has been actively studied.
Chalcogen compounds, such as CdS [6], ZnS [7], SnS 2 [8], NiS [9], Bi 2 S 3 [10], CuS [11], and MoS 2 [12], can be used for water decomposition or CO 2 conversion, and these have produced quite reliable results. Specifically, CdS has been evaluated as one of the most promising catalyst candidates for water decomposition under visible light, which is thermodynamically advantageous, as the positions of the valence and conduction bands overlap with the oxidation and reduction potentials of water [13,14]. However, CdS suffers from the inherent disadvantage of low efficiency due to inefficient charge separation between the electrons and holes generated by the absorption of photons, and is self-decomposing due to its structural instability in water [15]. Therefore, the catalytic performance varies widely depending on the degree of overlap between the redox potential of the selected chemical reaction and the band-gap of the catalyst. Thus, designing a stable and highly active chalcogen photocatalyst for a given reaction is a very attractive challenge. Additionally, by controlling the morphology of chalcogenides, significantly improved properties can be derived from the unique shapes generated at various stages of synthesis.
Here, p-type MoS 2 with excellent light absorption ability is prepared as 2D-structured nanosheets, and n-type Bi 2 S 3 with excellent optical activity and a nanorod morphology is also prepared. A heterojunction comprising the two types of particles is designed to create a heterojunction composite system. Weng and coworkers suggested that the use of MoS 2 nanosheets to coat the surface of Bi 2 S 3 discoids by a facile anion-exchange strategy accelerates the light-harvesting efficiency and charge separation, and helps faster charge transport and collection, leading to higher activity in the photocatalytic reduction of Cr(VI) under visible-light irradiation (λ > 400 nm) [16]. As another example, Ke and coworkers reported that a prepared Bi 2 O 3 /Bi 2 S 3 /MoS 2 nanocomposite exhibits improved photocatalytic activity for water oxidation (529.1 µmol h −1 g −1 cat ), with a value that is 1.5 and 12.5 times higher than that of pure Bi 2 O 3 and MoS 2 , respectively, under simulated solar-light irradiation [17]. Ultimately, in this study, we propose that p-type MoS 2 particles, which lack electrons, will easily accept electrons excited by n-type Bi 2 S 3 particles that have a narrow band-gap (which facilitates excitation of the electrons by the absorbed visible light [18]). Consequently, it was expected that the cycle of electron flow would proceed smoothly and separation between the electrons and holes would progress effectively. Furthermore, we anticipated that sulfur deficient sites [19] would be formed at the new interfaces between the two particles, which would eventually act as reaction sites, leading to enhanced catalytic activity. Figure 1 shows the X-ray diffraction patterns of the as-synthesized Bi 2 S 3 , MoS 2 , 1Bi 2 S 3 /1MoS 2 , 2Bi 2 S 3 /1MoS 2 , and 1Bi 2 S 3 /2MoS 2 particles. The XRD (X-ray diffraction) pattern of as-synthesized Bi 2 S 3 indicated the formation of orthorhombic phase Bi 2 S 3 (JCPDS No. 002-0391) [20]. No other crystalline phases were detected in the product. The XRD patterns show that the sample was mainly composed of rhombohedral MoS 2 , and the diffraction peaks at 2θ = 14.5 • , 33 • , 40 • , and 58 • are ascribed to the (002), (100), (103), and (110) lattice planes of crystalline MoS 2 (JCPDS card No. 037-1492) [21]. Extremely broad reflection peaks were observed in the profile of the MoS 2 particles, which indicated that the obtained MoS 2 sample was nano-sized. The XRD profile of the heterojunction Bi 2 S 3 /MoS 2 catalysts clearly showed the characteristic peaks of Bi 2 S 3 , but the peak assigned to the (002) plane of MoS 2 was hardly visible. The expanded spacing of the MoS 2 (002) plane resulted from the synthesis conditions. The spacing usually increases in the alkali solution, and in addition, the environmental temperature also has a critical effect. In particular, the expanding becomes large in both the alkali environment and the low temperature. At this time, it is reported that the new shifted peak (2θ = 9.3 • ) associated with the expanded (002) d-spacing was observed [22]. However, at temperatures higher than 220 • C, the spacing of the (002) plane of MoS 2 did not be expanded [23]. In this study, perhaps both the alkali environment and the low temperature (180 • C) caused the expansion of the (002) plane of MoS 2 in Bi 2 S 3 /MoS 2 heterojunction materials. Furthermore, in the 1Bi 2 S 3 /2MoS 2 sample, a new peak associated with the expanded (002) plane was also seen at 2θ = 9.3 • . The intensity of the characteristic peaks of the Bi 2 S 3 particles increased in proportion to the concentration added during the synthesis step. However, the crystallite sizes obtained from the Scherer equation [24] based on the (111) plane, which is the characteristic peak of Bi 2 S 3 nanoparticles, decreased slightly with increasing addition of MoS 2 .

Physical Properties of Catalysts
The crystallite sizes were 10, 9, 8, and 7 nm, respectively, for Bi 2 S 3 , 1Bi 2 S 3 /1MoS 2 , 2Bi 2 S 3 /1MoS 2 , and 1Bi 2 S 3 /2MoS 2 . From these results, we predicted that the crystallite sizes of Bi 2 S 3 in the Bi 2 S 3 /MoS 2 heterojunction composites did not change and the shape was maintained, even when different amounts of Bi 2 S 3 and MoS 2 were incorporated into the heterocapsulated catalysts. The Bi 2 S 3 particles comprising 50-60 nm wide and 100-150 nm long rods are shown in the TEM (Transmission Electron Microscope) image in Figure 2. The MoS 2 sample had a wide and thin sheet morphology. In the heterojunction Bi 2 S 3 /MoS 2 samples, the rod-shaped Bi 2 S 3 was wrapped by a thin sheet of MoS 2 . Increasing the amount of MoS 2 led to complete wrapping of the Bi 2 S 3 particles by MoS 2 . From these results, it was deduced that the two particles were uniformly structured and aligned for wrapping by electrostatic attraction during heterojunction formation.  Figure 3C) were performed for 1Bi 2 S 3 /1MoS 2 as a prototype of the heterojunction catalysts, and the results are shown in Figure 3. The fringe spacing parallel to the nanorod was estimated to be 3.55 and 6.34 Å in Bi 2 S 3 and MoS 2 , which indicates that crystal growth occurred preferentially in the (111) and (002) directions, respectively. Additionally, some of the characteristic peaks identified in the XRD pattern could be seen in this figure. The SAED pattern of Bi 2 S 3 recorded for the regularly spaced orthorhombic phase exhibited diffraction rings, thereby indicating that the orthorhombic phase was polycrystalline and could be readily indexed as the Bi 2 S 3 phase. A circle attributed to the (002) plane was also observed, which also corresponds to the rhombohedral MoS 2 nanosheet. Based on these results, we deduced that each of the crystallized particles, 1Bi 2 S 3 and 1MoS 2 , existed stably in the 1Bi 2 S 3 /1MoS 2 heterojunction particle with no crystal change. However, element mapping by TEM confirmed that all components (1Bi 2 S 3 or 1MoS 2 ) of the catalyst were uniformly dispersed in the catalyst; particularly, it was confirmed that MoS 2 uniformly wrapped Bi 2 S 3 as an outer layer. Moreover, there were almost no impurities such as C and O. The EDS (Energy Dispersive X-ray Spectroscopy) spectrum in Figure 4 showed the presence of only Bi and S on the surface of Bi 2 S 3 , and the atomic ratio of Bi:S was close to the stoichiometric value of 2:3, as shown in the inset table. In the EDS spectrum of the MoS 2 particles, only Mo and S components were observed, and no other impurities were found. The ratio of Mo:S, as shown in the inset table (Figure 4), was about 1:2. The atomic ratio of Bi 2 S 3 and MoS 2 once again suggested reliability of the synthesis. These ratios were almost the same as the theoretical values used in the synthesis step. On the other hand, all the components of the Bi 2 S 3 /MoS 2 heterojunction particles were detected in the EDS spectrum, and the intensities of the peaks were proportional to the contents. The theoretical metal/S ratio of the 1Bi 2 S 3 /1MoS 2 particles is 1:1.68. However, the actual surface ratio of Mo + Bi/S was 1:1.73, where the sulfur content was relatively high. In the 2Bi 2 S 3 /1MoS 2 heterojunction particle, the theoretical ratio of Mo + Bi/S is 1:1.6, but the actual surface ratio was 1:1.9. For the 1Bi 2 S 3 /2MoS 2 heterojunction particles, the theoretical ratio of Mo + Bi/S is 1:1.75, but the surface composition ratio of Mo + Bi/S was 1:1.99. This result suggests that more elemental sulfur is exposed on the surface of the Bi 2 S 3 /MoS 2 heterojunction particles than on the surfaces of the pure particles. Ultimately, this result is also related to the disulfide peaks observed in the XPS (X-ray Photoelectron Spectroscopy) profile and may be responsible for defects in the edges of the heterojunction particles. Typical high-resolution quantitative Bi 4f, Mo 3d, and S 2p in Bi 2 S 3 , MoS 2 , and 1Bi 2 S 3 /1MoS 2 XPS spectra are presented in Figure 5. The characteristic orbital peaks occurred at different positions in the full scan for each sample (Figure 5a). The highest intensity peaks for the Mo 3d, Bi 4f, and S 2p states are presented in Figure 5b. The deconvoluted peaks of the Mo 3d 5/2 and Mo 3d 3/2 states, located at binding energies of 229.1 and 232.4 eV, respectively, are attributed to Mo 4+ in the MoS 2 sample [25]. A small peak due to Mo 6+ [26] was observed at 235.5 eV, but this peak was negligibly small. The Mo 3d 5/2 peak located at the lower binding energy of 228.4 eV in the profile of the 1Bi 2 S 3 /1MoS 2 sample corresponds to the reduced Mo atoms due to sulfur vacancies [27]. Moreover, an overlapping peak of Mo 6+ and a singlet S 1s peak was observed at 225.7 eV due to sulfur defects [27]. Additionally, Kwok and coworkers reported the presence of edge S-defect sites (S 2p) in MoS 2 nanosheets within a mesoporous silica shell [28], such as Mo V=O sites that are comparable to molybdenum oxysulfides (MoO x S y ), bridging disulfides (S-S) br 2− , shared disulfides (S-S) sh 2− , and terminal disulfides (S-S) t 2− .
Notably, the disulfide ligands at these defect sites can be easily removed to expose unsaturated Mo 4+ sites, and have been suggested as the active sites. In this study, the O 1s and S 2p peaks were also observed in the full range spectrum of MoS 2 (Figure 5a). This result means that Mo present in the 1Bi 2 S 3 /1MoS 2 heterojunction catalyst is not present as MoS 2 , but as MoO x S y . Eventually, this results in more defects in MoS 2 , which are believed to act as catalytically active sites. Two peaks respectively corresponding to the Bi 4f 5/2 and Bi 4f 7/2 states of Bi 3+ appeared at 163.8 and 158.4 eV in the profile of the Bi 2 S 3 sample (Figure 5b). Bi 2 S 3 remain unchanged after the formation of Bi 2 S 3 /MoS 2 heterojunction. The peaks at 161 and 162.3 eV could be assigned to the S 2p state, and the S 2s XPS peak at the higher binding energy of 225.7 eV is consistent with the Bi-S species [29]. From these results, we confirmed that after heterojunction formation, there was no significant change in the Bi 2 S 3 crystal, but the oxidation state of Mo in the MoS 2 crystal was lowered. Ultimately, this is believed to lead to the formation of S defects and to increase the capacity for carbon dioxide adsorption.

Carbon Dioxide Photoreduction on Catalysts
Before entering the full-scale CO 2 photoreduction experiment, an isotopic experiment using 13 CO 2 as the substrate was performed under the same photocatalytic reaction conditions to further verify the source of the generated CO, and the products were analyzed by gas chromatography and mass spectrometry. The peak at m/z = 29 could be assigned to 13 CO, indicating that the carbon source for the formation of CO was the CO 2 used. When the CO 2 was replaced with N 2 , no detectable products were formed. When CO 2 was subsequently added to the reaction mixture with the clean photocatalyst (Ar gas/H 2 O/catalyst/CO 2 gas/light on), the product formation yield increased, providing strong evidence of product formation from CO 2 . This means that the final carbon compounds produced were exclusively generated by the reduction of externally injected CO 2 , which was the only carbon source in this study. Additionally, isotopic study of 1Bi 2 S 3 /1MoS 2 in the presence of moist 13 CO 2 was performed, as shown in Figure 6. Various signals of isotopic carbon compounds, 13 C, 13 CH 2 , 13 CH 3 , 13 CH 4 , 13 CO, 13 CH 3 OH, and H 13 COOH were observed at m/z = 13, 15, 16, 17, 29, 33, and 47, respectively, and the 13 CO 2 carbon source was detected at m/z = 45. Generally, the overall photoreduction reaction of CO 2 to CH 4 is a very complicated process that includes the transfer of eight electrons and eight protons, breaking of C-O bonds, and the formation of C-H bonds. There are two pathways, and both pathways preferentially follow the two-electron mechanism. Here, the formaldehyde pathway (fast hydrogenation pathway) follows the path CO 2 → HCOOH → HCHO → CH 3 OH → CH 4 , which is thermodynamically feasible [30]. However, no HCHO was detected in the GC-MS analysis. The favorable production of HCOOH as an intermediate increases CO production, and the favorable production of HCHO increases CH 4 production. Mass analysis showed that the 1Bi 2 S 3 /1MoS 2 catalyst produced slightly more CH 3 OH than CH 4 . Overall, the production of CH 4 from the complete reduction of CO 2 in the presence of the 1Bi 2 S 3 /1MoS 2 heterojunction catalysts seems to be unfavorable.  Figure 7, the amount of CO generated increased almost linearly with the irradiation time for all of the catalysts. However, the amount of CO generated with the pure MoS 2 and Bi 2 S 3 catalysts was minimal, whereas the conversion to CO in the presence of the heterojunctioned Bi 2 S 3 /MoS 2 catalysts increased surprisingly. There was no selectivity for CH 4 when only Bi 2 S 3 was used, whereas the heterojunctioned catalysts exhibited selectivity. The total amount of CO and CH 4 generated from CO 2 was similar for the 2Bi 2 S 3 /1MoS 2 and 1Bi 2 S 3 /2MoS 2 catalysts after 10 h. However, a high CO evolution of 40 µmol g −1 (42.5 µmol·g −1 for total amounts of CO and CH 4 ) was achieved with the 1Bi 2 S 3 /1MoS 2 catalyst, having a junction comprising 1:1 Bi 2 S 3 and MoS 2 , under photo-illumination for 10 h. These results reveal that the activity of 1Bi 2 S 3 /1MoS 2 for CO 2 reduction was comparable to that of other catalysts. The total yield obtained with the 1Bi 2 S 3 /1MoS 2 composite did not change significantly after five cycles (1 cycle corresponds to 10 h of reaction), as shown in Figure 7B), indicating the excellent stability of the 1Bi 2 S 3 /1MoS 2 heterojunction catalyst.

Optical Properties of Bi 2 S 3 /MoS 2 Catalysts
The UV-vis diffuse reflectance spectra were used to analyze the light absorption and energy band features of the Bi 2 S 3 /MoS 2 catalysts, as illustrated in Figure 8a. The absorption edge of Bi 2 S 3 was close to 970 nm; thus, the compound exhibited a photo-response from the visible to near-infrared region. In comparison, the pure MoS 2 nanosheets also presented a broad absorption band around 670 nm, which indicates that they can absorb photons over almost all of the visible region. Furthermore, the Bi 2 S 3 /MoS 2 heterojunction catalysts absorbed light over the full spectral region due to the transition of MoS 2 and Bi 2 S 3 . Other factors affect the observed photocatalytic efficiency, such as the optical range, containment of the charge transfer, specific surface area, and the suppression of the recombination of electron-hole pairs in the MoS 2 /Bi 2 S 3 heterojunction. Nevertheless, the highest absorbance point was around 670 nm. To determine the relative band-gap and the energy levels of all the catalysts, the band-gap energy (E g ) was determined by applying the following formula based on the differential reflectance spectroscopy (DRS) data [31]: the hetero-junction particles are direct band-gap semiconductors. The reflectance data can be converted to absorption according to the Kubelka−Munk (K-M) theory: F(R ∞ ) = (1−R ∞ )2/2R ∞ , where, R ∞ is reflectance and F(R ∞ ) is the K-M function [32]. The band gap of the samples can be estimated by using Tauc plot: (αhν) 1/n = A (hν−E g ), where, α, hν, A, and E g are the absorption coefficient, incident light frequency, proportionality constant, and band gap, respectively. The value of exponent n determines the nature of electronic transition; n = 1/2 for direct transition and n = 2 for indirect transition. The linear extrapolation of (αhν) 1/n to zero give the band gap energies of the samples. According to Tauc's relation, the plotting of (αhυ) 1/2 versus the photon energy (hυ) gives a straight line in a certain region. The extrapolation of this straight line will intercept the (hυ)-axis to give the value of the indirect optical energy gap (E g ). From the Tauc plots presented in Figure 8B, the E g s of Bi 2 S 3 and MoS 2 (determined from the plot of (ahν) 2 versus (hν)) were found to be 1.32 and 1.58 eV, respectively. The band-gap of the heterogeneous catalysts was intermediate between the band-gap of these two particles.
To determine the valence band (VB) and conduction band (CB) edge potential of the semiconductors, the VB positions were determined from XPS, as shown in Figure 9a; the VBs of MoS 2 and Bi 2 S 3 were 0.4 and 1.05 eV, respectively. Thus, by combining the XPS spectral data and band-gaps in Figure 8, the CB position was determined by applying the following equation, E g = CB − VB [33]. In conclusion, the conductance bands of MoS 2 and Bi 2 S 3 were −0.62 and −0.88 eV, respectively, as illustrated in Figure 9b. Based on these findings, it can be seen that the Bi 2 S 3 rod and MoS 2 sheet can form a perfect heterojunction.
To understand the charge carrier transfer behavior in the catalysts, photocurrent experiments were carried out and the photocurrent responses are shown in Figure 10. For the photocurrent density cycle, the analytical samples were prepared as pastes and coated on the cell-type FTO (Fluorine doped tin oxide) glass. This cell was used as the working electrode, and the Pt-coated FTO glass was used as the counter-electrode. The two electrodes were connected to form a single system. An iodolyte electrolyte (AN-50, Solarnonix) was used and sufficiently immersed in the electrode, and a potential of 0 V was applied. The current was measured under one-sun illumination with a 2000 Solar Simulator (IVIUM STAT, ABET Technologies). After the initial 1 min stabilization period, the current was measured during periodic irradiation at intervals of 30 s. The cell area (0.4 cm 2 ) was divided by the measured current, and the current density was calculated. The photocurrent was 30 nA for Bi 2 S 3 and 1185 nA for MoS 2 . When the amount of MoS 2 in the heterojunction catalyst was increased, the photocurrent increased with the active photoirradiation (light on) time; when the amount of Bi 2 S 3 was increased, the current was small but stable during active photoirradiation. The 1Bi 2 S 3 /1MoS 2 heterojunction showed the best photocurrent response, which was 1.5 and 58 times that of pristine MoS 2 and Bi 2 S 3 , respectively. This result confirms slower recombination of the charge carriers in the 1Bi 2 S 3 /1MoS 2 heterojunction, indicating significant improvement of the photocatalytic performance. Moreover, the steady and reproducible photocurrent response after several cycles suggests that the 1Bi 2 S 3 /1MoS 2 heterojunction possessed good photostability.    Figure 11 shows the CO 2 gas adsorption capacity (CO 2 -TPD (Temperature programmed desorption) curve) of the Bi 2 S 3 , MoS 2 , and 1Bi 2 S 3 /1MoS 2 particles. In order to remove impurities, the particles were treated at 300 • C for 30 min, and CO 2 gas was adsorbed at 50 • C for 2 h. The temperature was increased to 550 • C at a rate of 5 • C min −1 to determine the desorption temperature and curves in order to determine the amount of CO 2 gas adsorbed on the catalysts. The results indicated that CO 2 gas was desorbed above 300 • C; the amount of CO 2 desorbed was small, but the desorption temperature was high in the case of Bi 2 S 3 . The CO 2 desorption curves were separated into peaks at 324.2 and 484.4 • C for the MoS 2 sample, and the amount of CO 2 gas desorbed by the MoS 2 sample was significantly greater than that desorbed by Bi 2 S 3 . As noted above, this was attributed to the amount of sulfur vacancies in MoS 2 , as shown in the XPS profile. CO 2 is expected to be adsorbed in the vacant sulfur sites, and therefore, more sulfur vacancies lead to stronger attraction between CO 2 gas and the particles. However, in the case of 1Bi 2 S 3 /1MoS 2 , desorption occurred at lower temperatures and the highest desorption was achieved at 392.6 • C. However, the desorption amount decreased relative to that achieved with the single MoS 2 catalyst. This is because the amount of MoS 2 present per unit weight was reduced. Nevertheless, the amount of CO 2 adsorbed was high enough to react with water. Moreover, the lower CO 2 desorption temperature suggests that the adsorbed CO 2 can easily be converted in the next step. This also means that even after conversion to the product by the reaction with water, the product was easily desorbed, resulting in less coking on the catalyst. This in turn extended the catalytic performance.

Catalytic Mechanism for 1Bi 2 S 3 /1MoS 2
To comprehend the method of transfer of the photo-induced charge, a possible mechanism for photocatalysis by the Bi 2 S 3 /MoS 2 heterojunction was proposed, as shown in Scheme 1. When the Bi 2 S 3 /MoS 2 heterojunction is excited by visible light, the electrons move from the CB of the Bi 2 S 3 rods to the CB of the MoS 2 nanosheets. The holes are then transferred from the VB of MoS 2 to the VB of Bi 2 S 3 . Excited electrons naturally flow through the CB surface of Bi 2 S 3 to the CB of MoS 2 at lower potential (−0.62 eV vs. NHE (Normal Hydrogen Electrode)); thus, the density of electrons at the interface of these two particles will plausibly increase. Finally, CO 2 gas is adsorbed on the electron-rich junction surface or the VB surface of MoS 2 based on the gas adsorption data presented in Figure 11, and these electrons can reduce CO 2 to produce CO. This leads to efficient separation between the electrons and holes in this Bi 2 S 3 /MoS 2 heterojunction catalytic system. In theory, the photo-induced electrons in the VB of Bi 2 S 3 (0.4 eV vs. NHE) are not effective for the decomposition of water to produce H + ions; thus, the CB of Bi 2 S 3 cannot participate in reduction to yield H radicals or H 2 gas. This reduces the possibility of interaction of the CO intermediate with hydrogen, which in turn leads to lower CH 4 yields. Overall, the yield of CO was increased and the yield of CH 4 was very low.

Preparation of Catalysts
As shown in Scheme 2, the catalysts were prepared by a hydrothermal method. The homogenous solution was stirred for 4 h, then transferred to a Teflon-lined autoclave, heated to 140 • C at a rate of 10 • C min −1 , and maintained for 10 h. The solution was then slowly cooled at room temperature to obtain a precipitate that was separated by centrifugation and washed several times with absolute ethanol. The washed sample was dried at 70 • C for 8 h to obtain the final compound. MoS 2 was synthesized in the following order: 0.41 g (5.4 mmol) of thiourea was uniformly dissolved in 35 mL of DI water with stirring. Hexaammonium heptamolybdate tetrahydrate ((NH 4 ) 6 Mo 7 O 24 4H 2 O, M W = 1235.9 g mol −1 , Aldrich) was slowly added to the above solution and stirred for 4 h. This solution was transferred to a Teflon-lined autoclave, heated to 220 • C at a rate of 10 • C min −1 , and then held for 18 h. At the end of the reaction, the autoclave was cooled to room temperature, and the precipitate obtained was centrifuged and washed several times with anhydrous ethanol. The compound was dried at 70 • C for 8 h to obtain the final product. Finally, heterojunction Bi 2 S 3 /MoS 2 catalysts were synthesized in the following manner: thiourea was dissolved in 35 mL of DI (Deionized) water. To this solution, the appropriate amounts of (NH 4 ) 6 Mo 7 O 24 4H 2 O and Bi(NO 3 ) 3 5H 2 O were slowly added and then stirred for 4 h for homogenization. The solution was transferred to a Teflon-lined autoclave, heated to 180 • C at a rate of 10 • C min −1 , and held for 14 h. After the reaction, the autoclave was cooled to room temperature, and the precipitate obtained by centrifugation was washed several times with anhydrous ethanol. Finally, the compound was dried at 70 • C for 8 h to obtain the final product. The molar ratios of Bi 2 S 3 :MoS 2 in the obtained samples, were 1:1, 2:1, and 1:2. In total, five types of catalysts, Bi 2 S 3 , MoS 2 , 1Bi 2 S 3 /1MoS 2 , 2Bi 2 S 3 /1MoS 2 , and 1Bi 2 S 3 /2MoS 2 , were synthesized and their optical properties and activities were investigated.

Characterization
Powder XRD patterns of the prepared catalysts were collected with an XPert Pro MPD PANalytical 2-circle diffractometer using Cu-Kα radiation (λ = 1.5406 Å), a working voltage of 30 kV, and a current of 15 mA in the angle of diffraction (2θ) range 10-80 • . The morphologies and compositions of the as-prepared samples were characterized by transmission electron microscopy (TEM, H-7600, Hitachi, operating at 200 kV, Japan), energy dispersive spectrometric element mapping (EDS, EX-250, Horiba), and high-resolution transmission electron microscopy (HRTEM, JEM-2100 F, Jeol). The optical properties were confirmed from the UV-visible diffuse reflectance spectra collected on a Neosys-2000 (UV-vis, Scinco Co. Korea) spectrometer using BaSO 4 as the reference sample, photoluminescence spectroscopy (PL, Perkin Elmer, He-Cd laser source, wavelength of 300 nm), and a photocurrent-time response system (IVIUM STAT, ABET Technologies). A 150 W Xe lamp was used as the light source. A saturated calomel electrode (SCE), and saturated 0.1 M Na 2 SO 4 solution were used as the counter-electrode and electrolyte, respectively. The working electrode was prepared via a dip-coating method. Time-dependent photocurrent curves were constructed by the i-t curve method.

Photocatalytic Carbon Dioxide Conversion Reaction
The photoreduction of CO 2 in the presence of the synthesized Bi 2 S 3 , MoS 2 , 1Bi 2 S 3 /1MoS 2 , 2Bi 2 S 3 /1MoS 2 , and 1Bi 2 S 3 /2MoS 2 particles with H 2 O was performed in a closed cylinder-type quartz vessel (length: 15.0 cm; diameter: 1.0 cm; total volume: 12.50 mL). A schematic diagram of the photoreactor is provided in a previous report [34]. Here, 0.2 g of the catalyst and 40.0 L of distilled water were placed in the photoreactor. High-purity CO 2 gas (99.99%) was used as the reactant. The chamber was purged with the CO 2 gas before irradiation. The water was injected and the reactor was sealed; the outside of the reactor was wrapped with a heat tape and kept at 80 • C for 30 min to vaporize the water in the reactor. The reaction was performed after cooling the reactor to 25-30 • C. At this time, the pressure in the reactor was close to atmospheric pressure. Here, in general, the activity of a photocatalyst that decomposes organic molecules in a solution shows a tendency to increase steadily up to a reaction temperature of 60 • C, but decreases above it [35]. When the temperature is increased, it causes the bubbles formation in the solution which results in the generation of free radicals. Additionally, the increase in temperature helps the degradation reaction to overcome electron-hole recombination. In addition to that, the increasing temperature may enhance the oxidation rate of organic molecules at the interface. Zhang et al. also reported [36] when temperature increased from 323 K to 343 K, the yield of the photocatalytic reaction increased by two times. Although raising temperature is an effective strategy to obtain higher yields, the photoreduction of carbon dioxide has been reported to have the best performance at the low reaction temperature of 25 • C in some researches [37]. Thus, there is always an optimum temperature range to get the best. Here, the photoreactor in this study was an open system in which a one sun lamp illuminates a quartz photoreactor from above, with a diameter of up to 10 cm × 10 cm radiated from one sun. The distance from the end of one sun lamp to the reactor was 1.0 cm and the inner diameter of the reactor was 1.0 cm, thus the distance between the lamp and the photocatalyst sample was 2.0 cm. Moreover, the laboratory was maintained at 18 • C by the air conditioner to suppress the photoreactor temperature from rising above 25 • C by the one sun lamp. The reactor chamber was then closed and the lamp was switched on. A natural solar simulator (a xenon lamp, 150 mW cm −2 , 620 nm, 2000 Solar Simulator, ABET Tech) was used as the irradiation source. The solar light intensity of the solar simulator was calculated with a digital lux meter and the average light intensity was determined as approximately 1,200,000 Lux. Photoreduction was carried out at room temperature and atmospheric pressure. The product gases were analyzed using a gas chromatography (GC; Master GC, Scinco, Korea) instrument. Hydrocarbons such as methane, ethane, and propane were quantified by flame ionization detection (FID), and quantitative analysis of CO, H 2 , O 2 , CH 3 OH, etc., was performed with thermal conductivity detectors (TCD). The minimum detection limit of the GC was 1.5 pg carbon/s for FID, 2.5 ng mL −1 (standard), and 400 pg mL −1 (µ-TCD) for TCD. The product selectivity was calculated using the following equation: Ci (%) = Ci moles of the product/total moles of C produced × 100%. Furthermore, trace amounts of gases that are difficult to analyze by GC were analyzed by using a mass spectrometer (BELMASS, BEL Japan Inc., Japan).
When the catalyst is repeatedly done to run-off rather than continuously running the reaction for a long time, the catalyst structure is more damaged and so the activity of the catalyst is lowered. That is, more important than the lifetime of the catalyst is to confirm the recycling of the catalyst. Generally, in the thermal catalyst system, a catalyst used in the reaction is calcined at a high temperature, and a regeneration method for removing carbon coking accumulated on the catalyst surface is performed. However, a general regeneration method is not standardized in the photocatalyst system. Therefore, some researchers, including us, choose their own special regeneration method. In this study, the recycling experiment was carried out in the following manner: After each run of photocatalysis reactions for 10 h, the light was switched off, the reactor was disassembled, and then the photocatalyst was separated from reactor. The used catalyst was washed with distilled water for three times, filtered and dried at 60 • C. It was then thermal-treated for 2 h under N 2 at 150 • C. The catalyst was then stored in the dark for 24 h to minimize contact with external light until it was repacked into the photoreactor. This is to ensure that all regenerated catalysts are in the same environment as much as possible. The reactor was again installed, and then the gas inside the reactor was ejected. The stored catalyst was again fixed in the reactor, and the reactor was purged with CO 2 to displace the air inside the reactor. At this time, the reactor valve is closed/opened and repeated three times while purging CO 2 into the reactor. After ejecting all the air gas inside the reactor, both valves of the reactor are closed, water was injected in, vaporized it, and the reaction was initiated by switching on the light source. This experiment was repeated five times.

Conclusions
Bi 2 S 3 /MoS 2 heterojunction catalysts comprising Bi 2 S 3 rods wrapped with MoS 2 nanosheets were prepared via a hydrothermal method. The optimal photocatalytic performance was achieved with the 1Bi 2 S 3 /1MoS 2 heterojunction, for which CO 2 reduction to CO after 10 h was almost 40 and 16 times higher than that achieved with pure MoS 2 and Bi 2 S 3 , respectively. The novelty of this study is that CO was generated with sufficiently high selectively to preclude the need for secondary gas separation. Additionally, the enhanced photocatalytic efficiency could be attributed to the combination of Bi 2 S 3 and MoS 2 , which extends the light absorption ability and formation of the Bi 2 S 3 /MoS 2 heterojunction with a tight face-to-face connection and a good energy band match between MoS 2 and Bi 2 S 3 . The interfacial interaction between MoS 2 and Bi 2 S 3 can accelerate the separation of the photo-induced carriers and increase the sulfur defects.