CdS Nanoparticles Decorated 1D CeO 2 Nanorods for Enhanced Photocatalytic Desulfurization Performance

: CdS nanoparticles were constructed onto one-dimensional (1D) CeO 2 nanorods by a two-step hydrothermal method. The X-ray diffraction (XRD), transmission election microscopy (TEM), Raman spectra, X-ray photoelectron spectra (XPS) and UV-Vis diffuse reﬂection spectroscopy (DRS) techniques were used to characterize these CdS/CeO 2 nanocomposites. It was concluded that when the molar ratio of CdS and CeO 2 was 1:1, the nanocomposites exhibited the best photocatalytic desulfurization activity, reaching 92% in 3 h. Meanwhile, transient photocurrent (PT) measurement, photoluminescence (PL) spectra and electrochemical impedance spectroscopy (EIS) measurement indicated that the modiﬁcation of CeO 2 nanorods by CdS nanoparticles could signiﬁcantly inhibit the recombination of photogenerated electrons and holes. In addition, the possible mechanism of photocatalytic oxidation desulfurization of the nanocomposites was proposed. This study may provide an effective CeO 2 -based photocatalyst for


Introduction
Sulfur oxides (SO x ) emitted from vehicle fuel combustion have been shown to cause acid rain and haze, which has become a serious environmental pollution problem. Therefore, it is urgent to remove sulfur-containing organic compounds from fuel to reduce sulfur content [1,2]. Subsequently, a series of desulfurization technologies have been developed, such as hydrodesulfurization [3], oxidative desulfurization [4], adsorptive desulfurization [5] and biodesulfurization [6]. Among them, hydrodesulfurization is the predominant technology currently used in industry, which requires hydrogen consumption to achieve a high pressure and high temperature. In addition, it is difficult to remove dibenzothiophene (DBT) and its derivatives from fuel. Alternatively, photocatalytic desulfurization is a new desulfurization technology using a semiconductor photocatalyst for catalytic oxidation desulfurization, which has the advantages of mild reaction conditions, low energy consumption and environmental friendliness. In the process of photocatalytic desulfurization, DBT and its derivatives can be oxidized to corresponding sulfones, which have polarity and can be easily removed by solvent extraction [7,8]. In recent years, a series of photocatalytic desulfurization catalysts have been developed, such as a TiO 2 -based photocatalyst [9,10], bismuth series material [11,12], g-C 3 N 4 framework material [13,14], metal organic framework material [15], etc.
CeO 2 is an important rare earth oxide semiconductor material, which is commonly used as a photocatalyst for pollutant degradation and energy conversion. However, CeO 2 is a wide band gap semiconductor, which usually responds in the ultraviolet region, and ultraviolet light only accounts for 5% of the sunlight, which greatly hinders its application [16][17][18]. Therefore, the heterojunction of CeO 2 and another narrow band gap semiconductor is expected to improve the photocatalytic performance. Moreover, onedimensional (1D) nanomaterials have attracted much attention in recent years due to their

Synthesis of CeO 2 Nanorods
A total of 0.02 mol Ce (NO 3 ) 3 · 6H 2 O was dissolved in 5 mL deionized water, and then 35 mL 10 M NaOH solution was added. After being fully stirred, the mixture was transferred to a 50 mL Teflon hydrothermal reactor and heated to 110 • C for 10 h; the obtained products were centrifugally washed and dried, and then calcined at 400 • C for 2 h to obtain CeO 2 nanorods.

Synthesis of CdS/CeO 2 Composite
A certain amount of prepared CeO 2 nanorod powders were ultrasonically dispersed in 20 mL of deionized water, and then 20 mL of 0.1 M CdCl 2 ·2.5H 2 O solution was added. After full stirring, 20 mL of 0.1 M Na 2 S solution was slowly added, and then the mixture was transferred to a Teflon hydrothermal reactor and heated to 160 • C for 24 h; then the suspension was filtered, washed and dried to finally obtain the CdS/CeO 2 composite. The molar ratios of CdS to CeO 2 in the prepared samples were 1:2, 1:1 and 2:1, respectively.

Characterization
X-ray diffraction (XRD) measurement was performed on an X-ray diffractometer with Cu Ka radiation (Rigaku, D/max-RB, λ = 0.15406 nm); Raman spectra were measured by a Thermo Fisher Scientific DXR Raman spectrophotometer and the excitation laser wave length was 532 nm. The morphology was observed by a JEOL JEM-2100 transmission election microscope (TEM) equipped with Gatan 832 CCD operating at a voltage of 200 kV; X-ray photoelectron spectroscopy (XPS) analysis was carried out using a Thermo Fisher Scientific ESCALAB 250 spectrometer with the mono Al Kαradiation (1486.6 eV). Photoluminescence (PL) spectra were collected on a PerkinElmer LS45 fluorescence spectrometer with an excitation wavelength of 400 nm; ultraviolet-visible diffuse reflectance spectra (UV-Vis DRS) were measured by a Shimadzu UV-2450 spectrophotometer equipped with an integrating sphere.

Photoelectrochemical Measurements
The photoelectrochemical tests were carried out on an LK5600 photoelectrochemical workstation. For the photocurrent measurement, the prepared sample was deposited on the surface of FTO glass as the working electrode, the Ag/AgCl electrode as the reference electrode, the Pt plate as the counter electrode and the 0.1 M Na 2 SO 4 solution as the supporting electrolyte, and the 300 W xenon lamp as the light source. For electrochemical impedance spectroscopy (EIS) measurements, the three electrodes system was also used. The electrolyte solution was 0.5 M KCl solution containing 0.01 M K 3 Fe(CN) 6 /K 4 Fe(CN) 6 (molar ratio 1:1), performed at bias voltages 0.5 V, in the frequency range of 0.1 Hz-100 kHz with oscillation potential amplitudes of 0.01 V at room temperature.
The electrolyte solution was 0.5 M KCl solution containing 0.01 M K3Fe(CN)6/K4Fe(CN)6 (molar ratio 1:1), performed at bias voltages 0.5 V, in the frequency range of 0.1 Hz-100 kHz with oscillation potential amplitudes of 0.01 V at room temperature.

Photocatalytic Desulfurization Measurement
A total of 0.08 g DBT was dissolved in 200 mL n-octane as model oil; subsequently, a 0.1 g photocatalyst was added under constant stirring, and then appropriate H2O2 was added (moral ratio of O/S = 4:1). After dark adsorption for 30 min, the photocatalytic desulfurization process was carried out in a photochemical reaction instrument with a xenon lamp (300 W, with ultraviolet cut-off filter (λ > 420 nm)). The dispersion was collected every 30 min and extracted with acetonitrile, and the sulfur content was determined on a UV fluorescent sulfur analyzer. The desulfurization rate D (%) was obtained according to the following formula: D = (1 − Ct/C0) × 100%, where C0 is the initial sulfur content and Ct is the sulfur content of the solution at reaction time t.

XRD Patterns Analysis
The structures of the samples are confirmed by XRD analysis in Figure 1

Raman Spectroscopy Analysis
Raman spectroscopy with an excitation wavelength of 532 nm was used to measure the structure and electronic properties of the materials. As shown in Figure 2, for pure CeO2, the peak at 465 cm −1 can be attributed to the F2g Raman active interior phonon mode of the fluorite cubic structure [23]. As for pure CdS, the two dominant peaks at 300 cm −1 and 600 cm −1 are attributed to the first-order longitudinal optical (1LO) and the secondorder longitudinal optical (2LO) phonon modes of CdS, respectively [24]. Moreover, it can

Raman Spectroscopy Analysis
Raman spectroscopy with an excitation wavelength of 532 nm was used to measure the structure and electronic properties of the materials. As shown in Figure 2, for pure CeO 2 , the peak at 465 cm −1 can be attributed to the F2g Raman active interior phonon mode of the fluorite cubic structure [23]. As for pure CdS, the two dominant peaks at 300 cm −1 and 600 cm −1 are attributed to the first-order longitudinal optical (1LO) and the second-order longitudinal optical (2LO) phonon modes of CdS, respectively [24]. Moreover, it can be observed that all the composites have peaks corresponding to both CeO 2 and CdS with varying intensities. The Raman peak intensity of the CdS/CeO 2 composites is lower than that of pure CeO 2 or pure CdS. This may be attributed to the scattering loss caused by defects in the heterojunction [25,26]. With the increase in CeO 2 content, the characteristic peak intensity of CeO 2 in the CdS/CeO 2 composite gradually increases. Therefore, the Raman measurement results further demonstrate the existence of CdS and CeO 2 in the composite. be observed that all the composites have peaks corresponding to both CeO2 and CdS with varying intensities. The Raman peak intensity of the CdS/CeO2 composites is lower than that of pure CeO2 or pure CdS. This may be attributed to the scattering loss caused by defects in the heterojunction [25,26]. With the increase in CeO2 content, the characteristic peak intensity of CeO2 in the CdS/CeO2 composite gradually increases. Therefore, the Raman measurement results further demonstrate the existence of CdS and CeO2 in the composite.    The well-defined heterojunction structure may facilitate the separation of photogenerated electrons and holes, improving the photocatalytic efficiency.

XPS Spectra Analysis
The surface chemical states of the prepared samples are studied by XPS spectroscopy. The high-resolution XPS spectra of Ce 3d, O 1s, Cd 3d and S 2p are shown in Figure 4a-d, respectively. In Figure 4a, the characteristic peaks of Ce 3d are divided into eight fitting peaks, with the six peaks labeled as v, v 2 , v 3 , u, u 2 and u 3 representing Ce 4+ , and the other two peaks labeled as v 1 and u 1 corresponding to Ce 3+ , which indicates the presence of mixed Ce 4+ and Ce 3+ states in the samples [27]. According to the charge neutrality condition, the appearance of Ce 3+ is an indicator of oxygen vacancies in CeO 2 .The binding energy peaks labeled at 529.6 eV, 530.6 eV and 531.7 eV correspond to lattice oxygen (OL), oxygen vacancies (V O ) and chemisorbed oxygen (OA) in CeO 2 nanorods, respectively ( Figure 4b). Therefore, it indicates that there is a certain amount of V O in CeO 2 . When CdS is loaded, all Ce 3d and O L peaks shift slightly, indicating that there is charge transfer between CeO 2 and CdS [28,29]. As can be seen in Figure 4c, the peaks at 404.7 eV and 411.6 eV correspond to the Cd 3d 5/2 and Cd 3d 3/2 of Cd (II) states in the CdS sample. As shown in Figure 4d, the XPS spectrum of S 2p exhibits two peaks at 161.5 and 162.6 eV, which are attributed to S 2p 3/2 and S 2p 1/2 , respectively [30,31]. It is worth noting that the peaks of Ce 3d, O 1s, Cd 3d and S 2p are shifted slightly in CdS/CeO 2 (1:1) compared with pure CeO 2 nanorods and CdS, indicating that a heterojunction is formed between CdS nanoparticles and CeO 2 nanorods, which is conducive to the effective separation of photogenerated electrons and holes.

XPS Spectra Analysis
The surface chemical states of the prepared samples are studied by XPS spectroscopy. The high-resolution XPS spectra of Ce 3d, O 1s, Cd 3d and S 2p are shown in Figure 4a-d, respectively. In Figure 4a, the characteristic peaks of Ce 3d are divided into eight fitting peaks, with the six peaks labeled as v, v2, v3, u, u2 and u3 representing Ce 4+ , and the other two peaks labeled as v1 and u1 corresponding to Ce 3+ , which indicates the presence of mixed Ce 4+ and Ce 3+ states in the samples [27]. According to the charge neutrality condition, the appearance of Ce 3+ is an indicator of oxygen vacancies in CeO2.The binding energy peaks labeled at 529.6 eV, 530.6 eV and 531.7 eV correspond to lattice oxygen (OL), oxygen vacancies (VO) and chemisorbed oxygen (OA) in CeO2 nanorods, respectively (Figure 4b). Therefore, it indicates that there is a certain amount of VO in CeO2. When CdS is loaded, all Ce 3d and OL peaks shift slightly, indicating that there is charge transfer between CeO2 and CdS [28,29]. As can be seen in Figure 4c, the peaks at 404.7eV and 411.6 eV correspond to the Cd 3d5/2 and Cd 3d3/2 of Cd (II) states in the CdS sample. As shown in Figure 4d, the XPS spectrum of S 2p exhibits two peaks at 161.5 and 162.6 eV, which are attributed to S 2p3/2 and S 2p1/2, respectively [30,31]. It is worth noting that the peaks of Ce 3d, O 1s, Cd 3d and S 2p are shifted slightly in CdS/CeO2(1:1) compared with pure CeO2 nanorods and CdS, indicating that a heterojunction is formed between CdS nanoparticles and CeO2 nanorods, which is conducive to the effective separation of photogenerated electrons and holes.    (Figure 5b), respectively [32,33]. Therefore, CeO2 nanorods can only absorb UV light, and the increase in the CdS nanoparticles would result in the  Figure 5 shows the ultraviolet-visible diffuse reflectance spectra (UV-Vis DRS) of CeO 2 , CdS and the CdS/CeO 2 composites. As shown in Figure 5a, the absorption band edge of CeO 2 is about 400 nm, and that of CdS is about 550 nm. For the CdS/CeO 2 composites, the absorption edge is red shifted with the increase in the molar ratio of CdS. According to the Tauc plot of (αhυ) 2 vs. (hυ), the band gap energy (E g ) values of CeO 2 , CdS/CeO 2 (1:2), CdS/CeO 2 (1:1), CdS/CeO 2 (2:1) and CdS are estimated to be 3.1 eV, 2.45 eV, 2.4 eV, 2.36 eV and 2.3 eV (Figure 5b), respectively [32,33]. Therefore, CeO 2 nanorods can only absorb UV light, and the increase in the CdS nanoparticles would result in the increase in light adsorption in the visible region.  Figure 6a shows the photocatalytic removal of dibenzothiophene (DBT) from model oil with different photocatalysts. Under dark conditions, all photocatalysts have low desulfurization capacity for DBT. When only H2O2 exits without any catalyst, the desulfurization percentage is only about 7.5%, indicating that H2O2 has a certain oxidation capacity but the efficiency is relatively low. The photocatalytic desulfurization activities of different samples are as follows: CdS/CeO2 (1:1) > CdS/CeO2 (2:1) > CdS > CdS/CeO2 (1:2) > CeO2, indicating that the heterojunction formed by CdS and CeO2 can significantly improve the photocatalytic desulfurization activity. When the molar ratio of CeO2 and CdS is 1:1, the highest desulfurization efficiency can reach 92% in 3 h. Meanwhile, the corresponding photocatalytic desulfurization kinetic curves over the prepared photocatalysts are shown in Figure 6c.The reaction data are fitted by a first-order model as depicted by the formula [34]: ln(C0/C) = kt + b, where k is the pseudo-first-order rate constant, and the relationship between ln(C0/C) and catalytic reaction time t is considered to be linear. The corresponding desulfurization rate constants (k) are estimated to be 0.08109 h −1 , 0.27446 h −1 , 0.22957 h −1 , 0.57659 h −1 and 0.39763 h −1 for the CeO2, CdS, CdS/CeO2 (1:2), CdS/CeO2 (1:1) and CdS/CeO2 (2:1). It is clear that the photocatalyst CdS/CeO2 (1:1) displays the best photocatalytic activity among these prepared photocatalysts. The reusability of the CdS/CeO2 (1:1) composite is also investigated. As shown in Figure 6c, after repeated use for three times, the desulfurization percentage is about 75%, indicating that the prepared composite has relatively good stability. The desulfurization percentage decreases during the cycle may be attributed to the easy filling of the oxygen vacancy in the photocatalytic process [35].  Figure 6a shows the photocatalytic removal of dibenzothiophene (DBT) from model oil with different photocatalysts. Under dark conditions, all photocatalysts have low desulfurization capacity for DBT. When only H 2 O 2 exits without any catalyst, the desulfurization percentage is only about 7.5%, indicating that H 2 O 2 has a certain oxidation capacity but the efficiency is relatively low. The photocatalytic desulfurization activities of different samples are as follows: CdS/CeO 2 (1:1) > CdS/CeO 2 (2:1) > CdS > CdS/CeO 2 (1:2) > CeO 2 , indicating that the heterojunction formed by CdS and CeO 2 can significantly improve the photocatalytic desulfurization activity. When the molar ratio of CeO 2 and CdS is 1:1, the highest desulfurization efficiency can reach 92% in 3 h. Meanwhile, the corresponding photocatalytic desulfurization kinetic curves over the prepared photocatalysts are shown in Figure 6c.The reaction data are fitted by a first-order model as depicted by the formula [34]: ln(C 0 /C) = kt + b, where k is the pseudo-first-order rate constant, and the relationship between ln(C 0 /C) and catalytic reaction time t is considered to be linear. The corresponding desulfurization rate constants (k) are estimated to be 0.08109 h −1 , 0.27446 h −1 , 0.22957 h −1 , 0.57659 h −1 and 0.39763 h −1 for the CeO 2 , CdS, CdS/CeO 2 (1:2), CdS/CeO 2 (1:1) and CdS/CeO 2 (2:1). It is clear that the photocatalyst CdS/CeO 2 (1:1) displays the best photocatalytic activity among these prepared photocatalysts. The reusability of the CdS/CeO 2 (1:1) composite is also investigated. As shown in Figure 6c, after repeated use for three times, the desulfurization percentage is about 75%, indicating that the prepared composite has relatively good stability. The desulfurization percentage decreases during the cycle may be attributed to the easy filling of the oxygen vacancy in the photocatalytic process [35]. photocatalytic activity among these prepared photocatalysts. The reusability of the CdS/CeO2 (1:1) composite is also investigated. As shown in Figure 6c, after repeated use for three times, the desulfurization percentage is about 75%, indicating that the prepared composite has relatively good stability. The desulfurization percentage decreases during the cycle may be attributed to the easy filling of the oxygen vacancy in the photocatalytic process [35].

Photoelectrochemical and PL Analysis
The photocurrent measurement (PM), electrochemical impedance spectroscopy (EIS) and photoluminescence spectroscopy (PL) are used to analyze photogenerated electrons and holes separation efficiency. As shown in Figure 7a, when the light is turned on, all samples generate photocurrent, indicating that all the samples have light response. It is worth noting that the photocurrent of the CdS/CeO2 composite sample is significantly higher than that of the pure CeO2 and CdS, indicating a higher charge separation efficiency [34,36]. The EIS Nyquist plots are displayed in Figure 7b; the arc radius of the CdS/CeO2 composite sample is obviously smaller than those of the CeO2 and CdS samples, indicating that the charge transfer resistance of the CdS/CeO2 composite is lower [37,38]. The PL spectra of CeO2, CdS and CdS/CeO2 (1:1) are shown in Figure 7c; the CdS/CeO2 composite exhibits weaker PL intensity than that of pure CeO2 and CdS, which indicates that the combination of CeO2 and CdS can effectively inhibit the recombination of electrons and holes [39,40]. Therefore, the well-defined heterojunction structure is conducive to the transportation and separation of photogenerated electrons-holes, and enhances the photocatalytic activity.

Photoelectrochemical and PL Analysis
The photocurrent measurement (PM), electrochemical impedance spectroscopy (EIS) and photoluminescence spectroscopy (PL) are used to analyze photogenerated electrons and holes separation efficiency. As shown in Figure 7a, when the light is turned on, all samples generate photocurrent, indicating that all the samples have light response. It is worth noting that the photocurrent of the CdS/CeO 2 composite sample is significantly higher than that of the pure CeO 2 and CdS, indicating a higher charge separation efficiency [34,36]. The EIS Nyquist plots are displayed in Figure 7b; the arc radius of the CdS/CeO 2 composite sample is obviously smaller than those of the CeO 2 and CdS samples, indicating that the charge transfer resistance of the CdS/CeO 2 composite is lower [37,38]. The PL spectra of CeO 2 , CdS and CdS/CeO 2 (1:1) are shown in Figure 7c; the CdS/CeO 2 composite exhibits weaker PL intensity than that of pure CeO 2 and CdS, which indicates that the combination of CeO 2 and CdS can effectively inhibit the recombination of electrons and holes [39,40]. Therefore, the well-defined heterojunction structure is conducive to the transportation and separation of photogenerated electrons-holes, and enhances the photocatalytic activity. indicating that the charge transfer resistance of the CdS/CeO2 composite is lower [37,38]. The PL spectra of CeO2, CdS and CdS/CeO2 (1:1) are shown in Figure 7c; the CdS/CeO2 composite exhibits weaker PL intensity than that of pure CeO2 and CdS, which indicates that the combination of CeO2 and CdS can effectively inhibit the recombination of electrons and holes [39,40]. Therefore, the well-defined heterojunction structure is conducive to the transportation and separation of photogenerated electrons-holes, and enhances the photocatalytic activity.

Photocatalytic Mechanism
In order to analyze the mechanism of photocatalytic desulfurization, it is necessary to determine the conduction band (CB) and valence band (VB) positions of the semiconductors. The value of the valence band (VB) and conduction band (CB) of the semiconductor can be obtained according to the following formula [37,41]: ECB = χ − E0 − 0.5Eg, where χ is the electronegativity of a semiconductor, the geometric mean of the electronegativity of the constituent atoms, E0 is the energy of free electrons on the hydrogen scale  [42]. Hydroxyl radicals (•OH) and holes (h + ) with strong oxidation can be oxidized with DBT into DBTO2. Finally, due to the strong polarity of DBTO2, it can be removed by extraction separation to achieve the purpose of desulfurization [2,7,43]. In addition, the surface VO of CeO2 plays an important role in the process of photocatalytic desulfurization. It not only serves as an electron reservoir to store the photogenerated electrons and inhibit the recombination of photogener-

Photocatalytic Mechanism
In order to analyze the mechanism of photocatalytic desulfurization, it is necessary to determine the conduction band (CB) and valence band (VB) positions of the semiconductors. The value of the valence band (VB) and conduction band (CB) of the semiconductor can be obtained according to the following formula [37,41]: E CB = χ − E 0 − 0.5E g , where χ is the electronegativity of a semiconductor, the geometric mean of the electronegativity of the constituent atoms, E 0 is the energy of free electrons on the hydrogen scale (4.5 eV) and E g is the band gap of the semiconductor. DRS shows that the band gaps (E g ) of CeO 2 and CdS are 3.1 and 2.3 eV, respectively. The calculated CB positions of CeO 2 and CdS are −0.49 and −0.65 eV, respectively, and the calculated VB positions of CeO 2 and CdS are about 2.61 and 1.65 eV, respectively. When the composite is irradiated under visible light, CdS can be activated, and the electrons in the full VB of CdS can migrate to the empty CB, leaving a considerable number of positively charged holes at the corresponding position. Due to the CB potential of CdS (−0.65 eV) being lower than that of CeO 2 (−0.49 eV), the excited electrons in CdS are transferred to the CB of CeO 2 through the heterojunction interface, which leads to the photogenerated electrons (e − ) and holes (h + ) being effectively separated, inhibition of the recombination of photocarriers, and an improvement in the photocatalytic activity. At the same time, the photogenerated electrons react with H 2 O 2 to generate hydroxyl radicals (•OH) and hydroxyl (-OH). However, since the VB potential of CdS (1.65 eV) is lower than the redox potential of -OH/•OH (1.99 eV), holes cannot react with hydroxyl (-OH) to form hydroxyl radicals (•OH) [42]. Hydroxyl radicals (•OH) and holes (h + ) with strong oxidation can be oxidized with DBT into DBTO 2 . Finally, due to the strong polarity of DBTO 2 , it can be removed by extraction separation to achieve the purpose of desulfurization [2,7,43]. In addition, the surface V O of CeO 2 plays an important role in the process of photocatalytic desulfurization. It not only serves as an electron reservoir to store the photogenerated electrons and inhibit the recombination of photogenerated electrons and holes, but also acts as an active site to promote the adsorption and activation of H 2 O/-OH [29,44]. Figure 8 illustrates the possible photocatalytic desulfurization mechanism.

Conclusions
In conclusion, one-dimensional CeO2 nanorods were modified by CdS nanoparticles to form a heterojunction photocatalyst. The nanocomposites show great enhanced photocatalytic activity for photocatalytic desulfurization, and CdS/CeO2 (1:1) shows the highest desulfurization rate of 92% in 3 h under visible light. The nanocomposites demonstrate outstanding photocatalytic activity, and the desulfurization rate reaches 92% for DBT in 3 h. The excellent photocatalytic desulfurization performance is attributed to the transfer of visible light excited electrons on CdS nanoparticles to CeO2 nanorods, which effectively inhibits the photocorrosion of CdS with narrow band gap and improves the light trapping ability of CeO2 with a broad band gap. This work may provide a new idea for the design and construction of photocatalytic deep desulfurization materials.

Conclusions
In conclusion, one-dimensional CeO 2 nanorods were modified by CdS nanoparticles to form a heterojunction photocatalyst. The nanocomposites show great enhanced photocatalytic activity for photocatalytic desulfurization, and CdS/CeO 2 (1:1) shows the highest desulfurization rate of 92% in 3 h under visible light. The nanocomposites demonstrate outstanding photocatalytic activity, and the desulfurization rate reaches 92% for DBT in 3 h. The excellent photocatalytic desulfurization performance is attributed to the transfer of visible light excited electrons on CdS nanoparticles to CeO 2 nanorods, which effectively inhibits the photocorrosion of CdS with narrow band gap and improves the light trapping ability of CeO 2 with a broad band gap. This work may provide a new idea for the design and construction of photocatalytic deep desulfurization materials.