Preparation of CdS-CoSx photocatalysts and their photocatalytic and photoelectrochemical characteristics for hydrogen production

Lizi Chu a,1 , Yuan Lin a,1 , Yunpeng Liu a , Hongjuan Wang a , Qiao Zhang b , Yuhang Li b , Yonghai Cao a , Hao Yu a , Feng Peng b,*

a School of Chemistry and Chemical Engineering, Key Laboratory of Fuel Cell Technology of Guangdong Province, South China University of Technology, Guangzhou, 510640, China b Guangzhou Key Laboratory for New Energy and Green Catalysis, School of Chemistry and Chemical Engineering, Guangzhou University, Guangzhou, 510006, China

Highlights

 CoSx as co-catalyst is loaded on CdS nanorod by a facile method.
CdS-CoSx has good activity and stability for photoelectrochemical (PEC) reaction.
PEC activity of CdS-CoSx as photoanode for H2 production is the same as that of CdSePt.
An abnormal relationship between photocurrent and hydrogen production is found.
The reaction network in Na2S eNa2SO3 system for PEC H2 production is proposed.

graphical abstract

ABSTRACT

A facile method of loading CoSx nanosheet onto CdS nanorod has been designed, and the prepared CdS-CoSx composite catalyst exhibited significantly improved performance for photocatalytic hydrogen evolution compared with CdS catalyst. This composite catalyst was also used as a photoanode for photoelectrochemical (PEC) hydrogen production. The hydrogen production rate reached 168.6 mmol cm2 h1 (37.77 L m2 h1 ) under the simulated solar light, which is 2.7 times that of CdS and the same as that of CdSePt. In addition, in the Na2SeNa2SO3 system for PEC hydrogen production, an abnormal relationship between photocurrent and the hydrogen production yield was found. By designing a series of experiments, the photocatalytic and photoelectrochemical characteristics for hydrogen production were reasonably revealed for the first time. In this work, the prepared structured catalyst is easy to be recycled, and CoSx can replace precious metal Pt, showing a promising application.

© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Introduction

Due to the exhaustion of non-renewable resources such as fossil fuels and increasingly serious environmental pollution, the search for clean and non-polluting sustainable energy has attracted extensive interest of researchers around the world. Hydrogen is an excellent energy carrier due to its advantages of high energy density and environmental friendliness. Photocatalytic water-splitting is a promising way to produce hydrogen. However, there is a long way to go for its large-scale application for hydrogen production because of the relatively low energy conversion efficiency of photocatalysis at present. Therefore, it is still a challenge to find a photocatalytic material with high utilization efficiency of sunlight and good stability [1e11].

Among many photocatalytic materials, metallic sulfides have been widely studied due to their narrow band gaps and high utilization of solar light. CdS with the band gap of 2.4eV, as a typical semiconductor photocatalytic material, has good response to visible light and high conduction band position, which is favorable to reduce Hþ to hydrogen. Therefore, it has become a research hotspot [12e18]. However, CdS is easily oxidized by photogenerated holes, resulting in self-corrosion. Moreover, CdS has high photogenerated carrier recombination rate, which leads to its poor photocatalytic performance for hydrogen production [19e21]. Therefore, loading co-catalysts or combining other semiconductors to form heterostructures are effective ways to improve the photocatalytic activity and stability for hydrogen production [22e27]. It is well known that Pt and other precious metals are excellent co-catalysts for hydrogen evolution, but the high price and scarce resource make them unsuitable to be applied in a large scale [28e30]. Therefore, it is highly necessary to look for cheap and abundant non-precious metal catalysts to replace Pt and other precious metals as co-catalysts of CdS for photocatalytic hydrogen production [31e38].

Recently, non-noble metal sulfide has become a research hotspot of co-catalyst, such as Co9S8, which is a very promising catalytic material for photocatalytic hydrogen production because of narrow band gap and high flat band potential [39e42]. CoSx is widely used in electrocatalytic water decomposition and oxygen reduction with excellent electrochemical properties [43,44]. However, as far as we know, CoSx has not been widely used in photocatalytic water-splitting. Qiu et al. [40] designed and synthesized a kind of direct Z-type semiconductor heterostructure photocatalytic material Co9S8/CdS with hollow Co9S8 nano-cube supported CdS quantum dots. In this study, hollow cube Co(OH)2 was used as a template, and dimethyl sulfoxide (DMSO) was used as a solvent and S source to form Co9S8 by ion exchange with OH. In visible light, compared with the pure hollow Co9S8 cube and CdS quantum dot, the hydrogen production yield of the Co9S8/CdS catalyst was improved by 134 times and 9.1 times, respectively. In addition, the hydrogen production activity of the Co9S8/CdS catalyst remained relatively stable for 25 h. However, this method required the use of toxic organic reagents and complex preparation processes. Reddy [45] designed and synthesized a Z-type photocatalyst composed of CoeC@Co9S8 doubleshell nanocage supported mesoporous reticular CdS, which had a hydrogen production rate of 26.69 mmol g1 h1 and an apparent quantum yield of 7.82% at 425 nm for 5 h. Wang et al. synthesized a photocatalyst with layered Co9S8@ZnIn2S4 cage heterostructure [46], which effectively promoted the separation of photogenerated charges and exposed the redox active sites. The hydrogen production efficiency reached 6.25 mmol g1 h1 with good stability, but the disadvantage was that the powder sample was difficult to be recycled. Zheng et al. [41] designed and synthesized a CoS bifunctional catalyst with the hydrogen production rate of 1.197 mmol g1 h1 , and the oxygen yield of up to 65%. Lin's research group [47] used Co9S8 for the first time to prepare a ternary TiO2/Co9S8/POM composite material as photoanode for photoelectrochemical (PEC) water oxidation, and the current density was as high as 1.12 mA cm2 vs RHE at AM1.5. However, the preparation process of this material was too complicated.

Here, a simple method was designed to synthesize a composite catalyst by using CdS nanorod array (NRs) as skeleton, and then loading CoSx nanosheet onto the CdS nanorod. According to the previous work of our research group [48], firstly, one-dimensional CdS NRs were grown on FTO, and then the intermediate containing Co was loaded onto the CdS NRs through simple chemical bath deposition (CBD) [49] in cobalt chloride and urea solution. Finally, CdS-CoSx was obtained through simple ion exchange. The preparation process is simple and the prepared structured catalyst is easy to be recycled. The photocatalytic hydrogen evolution performance of CdS-CoSx composite catalyst was significantly improved compared with that of single component catalyst CdS. The prepared composite catalyst was used as a photoanode for PEC hydrogen production, and its performance was comparable to that of CdSePt sample. So CoSx can replace precious metal Pt, showing a promising application. In addition, in Na2SeNa2SO3 system for PEC hydrogen production, we found that the measured photocurrent showed an opposite change rule with the hydrogen production yield. By designing a series of experiments, the reaction network for this photocatalytic system was analyzed for the first time, and the photocatalytic and photoelectrochemical hydrogen production process was explained reasonably.

Experimental

Catalysts preparation

One-dimensional CdS NRs were synthesized according to the hydrothermal methods reported previously by our group [48]. In a typical experiment, a clean 3  5 cm2 FTO with the conductive side facing down was immersed in Teflon-lined stainless steel autoclave (100 mL) containing 0.05 mol/L cadmium nitrate (99% Aladdin), 0.05 mol/L thiourea (99% Aladdin) and 0.01 mol/L glutathione (99% Aladdin), and then the autoclave was transferred to an oven and maintained at 210 C for 12 h. After that, the CdS NRs film on FTO was alternately rinsed with distilled water and alcohol for several times, dried naturally, then annealed at 450 C for 2 h under pure Ar.

The obtained CdS NRs film on FTO (conductive side facing down) was immersed in mixed solution containing 0.01 mol/L urea (99% Aladdin) and 0.015 mol/L cobalt chloride (99% Aladdin) under 90 C for 1 h, 2 h, 3 h, 4 h, respectively, to load the intermediate containing cobalt (Co(CO3)0.35Cl0.20(OH)1.10, LCCH) [49] onto the CdS nanorods. After cooling down, the asprepared samples were transferred to Teflon-lined stainlesssteel autoclaves (100 mL) containing 5 mmol/L Na2S solution. Then the autoclave was maintained at 90 C for 21 h. Finally, the sample was rinsed, dried, and the resultant sample was labeled as CdS-CoSx-t, where “t” represented the chemical deposition time (1e4 h). The catalyst preparation process is illustrated in Scheme 1.

For comparison, CdSePt catalyst was prepared using the currently accepted photochemical reduction method. 300 mL 0.01 mol/L of H2PtCl6 and 300 mL of absolute ethanol were uniformly mixed, dropped on the CdS nanorods prepared above. The sample was then in situ photo-deposited under AM 1.5. The obtained sample was labeled as CdSePt.

Material structure and photoelectric chemical characterizations

The crystal phase composition of the photocatalysts were characterized by X-ray powder diffractometry (XRD, Bruker D8 Advance). The surface microstructure and elemental mappings of the photocatalysts were measured by field emission scanning electron microscope (SEM, ZEISS Merlin, In-lens, Germany), transmission electron microscopy (TEM, JEM2100F, Japan) and energy dispersive X-ray detector (EDX, Bruker, XFlash 5030T, Germany). The surface chemistry state was analyzed by X-ray photoelectron spectroscopy (XPS, Krato, Britain). The optical properties of the samples were examined by ultravioletevisible diffuse reflectance spectroscopy (DRS, U3010, Hitachi, Japan). The photoluminescence (PL) spectra of samples were tested by an F-7000 fluorescence spectrophotometer (Hitachi, Japan). The slit width of excitation and emission was 10 nm, the PMT voltage was 700 V, the excitation wavelength was 400 nm, and the scanning speed was 1200 nm min1 . The Surface photovoltaic (SPV) of samples were tested by surface photovoltage test system (CEL-SPS1000). The element content samples were analyzed by inductively coupled plasma atomic emission spectrometer (ICP-OES, Agilent 720 ES).

Photoelectrochemical performances of samples were performed using CHI660D electrochemical workstation with a three-electrode setup. Platinum net was used as the counter electrode, Ag/AgCl electrode was used as the reference electrode, and the as-prepared samples were served as the working electrode. The tests were conducted in electrolyte of 0.5 M Na2S-0.5 M Na2SO3 mixed solution and light source was 300 W Xenon lamp with AM1.5 filter. Mott-Schottky plots acquired at 1000 Hz in 0.5 M Na2S-0.5 M Na2SO3 mixed solution under dark condition.

Measurement of photocatalytic hydrogen production

The photocatalytic hydrogen evolution of the samples was carried out in a quartz closed photocatalytic hydrogen production system (CEL-SPH2N-D, Beijing). A piece of the structured catalyst (active area 3  3 cm2 ) was immersed in 150 ml of a mixed solution containing 0.5 M Na2S-0.5 M Na2SO3. The photocatalytic reaction was conducted under visible light irradiation (300 W Xe lamp with 420 nm cut-off filter) and vacuum environment of less than 0.09 Mpa and temperature of 5 C. Hydrogen was detected using GC-9790 gas chromatograph with molecular sieve 5 A packed column and thermal conductivity detector.

Measurement of photoelectrochemical hydrogen production

The photoelectrochemical reaction was implemented in CHI660D electrochemical workstation with three-electrode system. Platinum net was used as the counter electrode, Ag/ AgCl electrode was used as the reference electrode, the asprepared sample was served as the working electrode (effective area 1  1.5 cm2 ). The hydrogen is collected by drainage as shown in Fig. S1. In the reaction system, the sacrificial agent is the mixed solution of 0.5M Na2S-0.5M Na2SO3. The simulated solar light source is 300 W Xenon lamp with light intensity of 82 mW cm2 , and the reaction temperature is room temperature (25 C).

Results and discussion

Material structure characterizations

Fig. 1A shows that pure CdS is hexahedral prism structure, growing uprightly on FTO. Fig. 1(BeF) are SEM images of CdS NRs coated with CoSx. It can be seen clearly that CoSx nanosheets are uniformly distributed on the outer surface of CdS NRs, and the loading of CoSx nanosheets gradually increases with the increase of deposition time. When the deposition time reaches 4 h, excess CoSx completely covers the entire CdS NRs (Fig. 1F).

Fig. 2 shows TEM images of CdS-CoSx-3 and CdS. As shown in Fig. 2A, CdS exhibits short rod shape, the average length and average width of CdS have been measured to be 775 nm and 340 nm respectively. Fig. 2B is the HRTEM image of CdS, where the lattice diffraction fringes of the CdS (002) crystal plane can be observed. From Fig. 2C, it can be seen clearly that CoSx nanosheets are layered on the surface of CdS. But it is hard to observe any lattice fringes of CoSx in Fig. 2D, indicating that CoSx exists mainly in amorphous form. The surface element distribution mappings of CdS-CoSx-3 are shown in Fig. 2(EeG), corresponding to Cd, S, and Co, respectively, indicating that CoSx nanosheets are uniformly distributed on the surface of the CdS NRs. The TEM images of other CdS-CoSx samples are shown in Fig. S2 (H-J).

The XPS spectra of CdS-CoSx-3 are shown in Fig. 3. In Fig. 3A, the characteristic peaks at 411.73 eV and 404.98 eV correspond to 3d3/2 and 3d5/2 of Cd2þ, respectively. In Fig. 3B, the characteristic peaks at 162.50 eV and 161.30 eV are attributed to 2p1/2 and 2p3/2 of S2, respectively. The weak peak for elemental S can be found at around168.3 eV. Fig. 3C can be fitted to three pairs of peaks. One pair of the satellite peaks of Co are located on 803.38 eV and 787.04 eV. The other pair of peaks at 798.75 eV and 783.29 eV are assigned to 2p1/2 and 2p3/2 of Co3þ, and the third pair of peaks at 797.05 eV and 781.13eV are ascribed to 2p1/2 and 2p3/2 of Co2þ, respectively. According to the calculated peak areas, the content ratio of Co2þ to Co3þ is 1.76. Therefore, the prepared CoSx is composed of Co2þ and Co3þ, mainly Co2þ. Compared with the XPS spectra of CdS as shown in Fig. S3, it can be found that the peak positions of Cd and S in CdS-CoSx-3 have a little shift after loading CoSx, which is probably caused by the strong interaction between CdS and CoSx. The XPS spectra for CdS-CoSx-3 sample before and after cycling stability test are shown in Fig. S4. It is obvious that the chemical state of CdS-CoSx-3 after stability test has hardly changed.

The XRD patterns of CdS-CoSx, CdS and CoSx powders are shown in Fig. 4A and Fig. S5. There is no significant difference between CdS and CdS-CoSx, and all samples have diffraction peaks corresponding to CdS and FTO, but no diffraction peaks of CoSx are found in CdS-CoSx samples. The possible reason is that CoSx mainly exists in amorphous form, or the load of CoSx is too low. The XRD patterns of used CdS-CoSx-3 and CoSx powder are shown in Fig. S5. It obvious that the physical structure of CdS-CoSx-3 has hardly changed after stability test. The actual contents of Co, Cd, S elements in the CdS-CoSx-3 and CdS measured by inductively coupled plasma optical emission spectrometer (ICP-OES) are shown in Table S2. It is known that the real Co content of the prepared samples with optimum performance is 2.1%. Fig. 4B shows that the absorption intensity of the samples in the visible and ultraviolet light region increases with the extension of deposition time. When the deposition time extends to 4 h, the absorbance of CdS-CoSx4 in the 300e500 nm region decreases, indicating that the excessive loading of CoSx is not conducive to the absorption of visible light. The curves of (Ahn) 2 vs hn for all samples are shown in Fig. S6. It can be obtained that the band gap of CdS and CdS-CoSx-3 is 2.25 eV and 2.32 eV, respectively. Fig. 4C shows that the fluorescence intensity of CdS-CoSx is higher than that of CdS. This is because the carrier concentration of CdS-CoSx increases significantly after loading CoSx, leading to an increase in the amount of photogenerated carrier recombination. However, the fluorescence intensity of CdS-CoSx-3 is the lowest among all CdS-CoSx samples, indicating that proper CoSx loading can reduce the recombination of photogenerated carriers.

Fig. 5A and B show the time-resolved photoluminescence (TRPL) patterns of CdS and CdS-CoSx-3, respectively. TRPL spectra were fitted in polynomial R(t) ¼ A1e(t/t1) þ A2e(t/t2). The fluorescent lifetime t1 and t2 were calculated to be 1.24 ns, 3.94 ns for CdS and 1.16 ns, 2.88 ns for CdS-CoSx-3, respectively. It is obvious that the fluorescence lifetime of CdS-CoSx3 is slightly lower than that of CdS, which is possibly due to the increase of charge transfer rate of CdS after loading CoSx. Fig. 5C exhibits the surface photovoltage (SPV) spectra of CdS and CdS-CoSx-3. It can be seen that the SPV value of CdS is higher than that of CdS-CoSx-3, indicating that less photogenerated electrons or more holes in CdS migrated to the surface after loading CoSx, and it is possibly because of the lower fermi level of CoSx than that of CdS. As discussed above, we deem that an appropriate amount of CoSx can effectively promote the transfer capability of photogenerated charge carrier in CdS.

Photocatalytic and photoelectrochemical hydrogen production

In order to study the carrier concentration and photocatalytic properties of all the samples, the Mott-Schottky curves and photocatalytic H2 production were tested. Fig. 6A shows the slopes of normalized Mott-Schottky plots. As exhibited in Table S1, the calculated carrier concentration increases along with the increase of deposition time, but when the deposition time increases to 4 h, the excessive loading of CoSx reduces the carrier concentration. The calculated ECB values of CdS and CdS-CoSx-3 are shown in Fig. S7. Combined with Fig. S6, it can be deduced that EVB values of CdS and CdS-CoSx-3 are 3.33 eV and 3.35 eV, respectively. Under visible light irradiation, the hydrogen evolution rates of all samples are displayed in Fig. 6B. CdS without co-catalyst has almost no hydrogen produced by photocatalysis. After loading CoSx, the hydrogen production rate of CdS-CoSx increases rapidly, among which the hydrogen production rate of CdS-CoSx-3 is the highest (3.5 mmol cm2 h1 ). The cyclic stability of CdS-CoSx-3 for photocatalytic hydrogen production is displayed in Fig. 6C, demonstrating that the photocatalytic stability of CdS-CoSx-3 is very good after at least 5 cycles. However, the photocatalytic hydrogen production activity of CdS-CoSx is about 1/4 of that of CdSePt, indicating that the activity of CoSx as a co-catalyst is still lower than that of precious metal Pt (Fig. 6B-d and f). The PEC hydrogen production was carried out with the schematic device shown in Fig. S1 and the results shown in Fig. 7A. It can be seen clearly from Fig. 7A that the PEC hydrogen production rate of CdS-CoSx samples is greatly improved compared with that of photocatalytic hydrogen production (Fig. 5B).

The PEC hydrogen production rate of CdSCoSx-3 is 2.7 times that of CdS, reaching 168.6 mmol cm2 h1 (37.77 L m2 h1 ), which is the same as that of CdSePt. The recycling stability of CdS-CoSx-3 for PEC hydrogen production is shown in Fig. 7B. After 5 cycles, the hydrogen production rate of the sample was reduced to 74%, indicating that the sample was subjected to some degree of photo-corrosion during PEC hydrogen production.

The linear sweep voltammetric (LSV) curves of all samples under dark and AM1.5 illumination are shown in Fig. 8A. At the positive bias voltage, all samples have much higher current densities under light irradiation than those in the dark, owing to the transfer of photogenerated electrons from the sample to counter electrode (Pt net). After deposited Pt on CdS, the photocurrent of CdSePt was obviously lower than that of pure CdS, which is possibly because of the excellent electron capture ability of the loaded Pt [35]. Fig. 8B shows the timepassed photocurrent curves (i-t curve) of the as-prepared samples at the bias voltage of 1.0 V (vs. RHE). It is obvious that the photocurrent density of CdSePt exhibits the smallest among all the samples. As the loading time of CoSx increases from 1 to 3 h, the photocurrent of CdS-CoSx decreases gradually. We deem that it is also because of the electron capture ability of the loaded CoSx as a reductive cocatalyst. While the loading time of CoSx reaches to 4 h, the photocurrent density of CdS-CoSx obviously increases. As shown in the SEM images (Fig. 1), the loaded CoSx obviously agglomerates in the CoSxCdS-4 sample. It is reported that the bulk CoSx shows the properties of semiconductor [40]. According to the band structure of CoSx and CdS, the electrons on the CB of CoSx might transfer to the CB of CdS, hence increases the photocurrent density of CdS-CoSx.

The reaction network analysis for photocatalytic hydrogen production

In the general PEC hydrogen production process, the photoelectrons of the working electrode are transferred to the counter electrode Pt by the external circuit to generate hydrogen. Therefore, the larger photocurrent corresponds to the higher photoelectrochemical hydrogen production rate. However, in this study, it is very interesting that the smaller photocurrent resulted in the higher photoelectrochemical hydrogen production rate. In fact, this phenomenon has also been found in TiO2/MoS2 system [50].

In order to explain this abnormal result, a series of experiments (Fig. S8) were designed to explore the reaction process and the results were shown in Fig. 9. From Fig. 9A, it can be seen that in 0.5 M Na2S-0.5 M Na2SO3 solution (Fig. S8A), Pt alone produces little hydrogen. While CdS-CoSx-3 can produce hydrogen with the average hydrogen production rate of 44 mmol cm2 h1 for 5 h. It should be noted that the photocatalytic hydrogen production rate of CdS-CoSx-3 in Fig. 9 is much higher than that in Fig. 5 because of different reaction temperatures and light sources. When Pt and CdS-CoSx-3 were simultaneously inserted into 0.5 M Na2S-0.5 M Na2SO3 solution (Pt and CdS-CoSx-3 were not connected, Fig. S8B), the hydrogen production rate reaches 70 mmol cm2 h1 , with 60% increase based on CdS-CoSx-3. For CdS alone, the average hydrogen production rate is 8 mmol cm2 h1 for 5 h. However, the hydrogen production rate does not increase significantly when the Pt was inserted into solution. The hydrogen production performance of all samples was tested in the same case of Pt (Fig. S8B), and their hydrogen production rates are shown in Fig. S9. From these figures, all the photocatalysts exhibit enhanced hydrogen production rates when Pt is inserted simultaneously into the solution even if the Pt is not connected with the photocatalysts.

To this interesting phenomenon, we deduced the reason is that Na2S2O3 would be formed during the photocatalytic reaction in Na2SeNa2SO3 solution, then Na2S2O3 reacts with water to release H2 on the Pt electrode, thus increasing the hydrogen production rate. In order to verify our conjecture, Pt alone was inserted in 0.5M Na2S-0.5M Na2S2O3 solution under irradiation for 3 h, as shown in Fig. S8A, large amounts of hydrogen were observed in this system. However, H2 was not produced in the 0.5M Na2S-0.5M Na2S2O3 solution under illumination without Pt, as shown in Fig. 9B. Therefore, it is preliminarily confirmed that Na2S2O3 reacts with water to release H2 on Pt electrode.

In order to further verify the formation of Na2S2O3, we designed another experiment, as shown in Fig. S10. Firstly, a piece of CdS-CoSx-3 alone was placed in 0.5M Na2S-0.5M Na2SO3 solution under irradiation for 3 h, and then CdS-CoSx3 was taken out. Finally, Pt was put into the above solution and irradiated for 1 h. As expected, 1.2 mL hydrogen was produced, further indicating that the intermediate Na2S2O3 is formed in 0.5 M Na2S-0.5 M Na2SO3 system. Na2S2O3 cannot be completely consumed in time on the CdS-CoSx catalyst to generate hydrogen, while, Pt can enhance the redox reaction of Na2S2O3 with water to release H2. Therefore, when Pt and CdS-CoSx are both present in Na2SeNa2SO3 solution, the hydrogen production rate can be greatly improved.

Base on the above discussion, the photocatalytic reaction network in this catalytic system is proposed as follows (Eqs. (1)e(6))