0 Introduction

Hydrogen as a sustainable, secure, clean and alternative energy source can ease the energy crisis and environment pollution faced in the present world^[1-2]. Electrochemical water splitting is considered as a highly effective method to produce hydrogen^[3]. However, it needs electrocatalysts to reduce overpotential due to low kinetics of water splitting. Pt is at present the most active catalyst for the hydrogen evolution reaction (HER)^[4-5]. However, its scarcity on the earth and high cost greatly restrict the industrial production of hydrogen. Therefore, it is necessary and urgent to find cost-effective and earth-abundant catalysts with high HER catalytic activity and excellent stability to facilitate translation of H₂O to hydrogen.

Transition metal dichalcogenides with generalized formula of MX₂ (M refers to transition metal; X represents a chalcogen such as S and Se) have received increasing interests due to their low cost and high abundance^[6-10]. Among them, MoS₂ has been considered as a promising candidate because of its high activity and stability^[11-14]. However, the experimental and computational studies have concluded that the catalytic activity mainly arises from the active sites located along the edges of 2-D MoS₂ layers which are under-coordinated and thermodynamically unfavorable, and the basal planes are catalytically inert^[15-17]. Although various methods have been adapted to expose more edge sites or enhance the intrinsic activity of the edge sites, the enhancement of HER electrocatalytic performance of MoS₂ still faces enormous challenges^[18-20]. Another transition metal dichalcogenides CoS₂, which was often utilized as electrode materials for supercapactiors, Li-ion batteries or electrocatalysts for oxygen reduction reaction (ORR)^[21-26], also has shown excellent HER electrocatalytic activity, even higher than MoS₂. In addition, CoS₂, which is an intriscally conductive metal in contrast to the other MX₂ such as FeS₂ ^[27] and NiS₂^[28], can improve the electron transportation from catalyst surface to the electrode, thus will reduce the overpotential needed to overcome the energy barriers and decrease the energy consumption. What's more, CoS₂ can exhibit high HER activity after conversion from thermodynamically favored semiconducting phase to a metastable metallic polymorph compared wih other transition metal dichalcogenideis, such as MoS₂ and WS₂^[29-32]. However, the key limitation for CoS₂ as HER electrocatalyst is poor stability in acid media, which may be due to the structural breakdown of CoS₂ nanostructure or the evolution of S during H₂ evolution^{[24, 33-37]}. The poor stability of CoS₂ materials has seriously restricted their practical applications as high-performance HER electrocatalysts. Therefore, it is significant to develop CoS₂-based electrocatalysts with excellent stability as well as high catalytic activity.

To improve the electrocatalytic activity and stability of CoS₂ for HER, carbon thin layer coating will be a promising method because of following advantages: (1) the carbon thin layer can well prevent CoS₂ from structural breakdown and the evolution of S; (2) the electronic interaction between CoS₂ and thin carbon layer will change the electronic structure of CoS₂, which will be beneficial for the improvement of electrocatalytic activity; (3) The carbon layer will provide "super-highways" for electron transfer to promote HER due to its high electrical conductivity. Based on the above considerations, we devote our attention to designing and synthesizing the novel CoS₂ yolk-shell (YS) spheres coated with carbon thin layers (CoS₂ YS@C spheres) as highly efficient HER electrocatalysts by a simple hydrothermal method. The CoS₂ YS@C spheres own hollow structure and high specific surface area, and they exhibit excellent catalytic activity with low onset potential of only about 20 mV, small Tafel slope of about 55 mV/dec, and small overpotential of about 90 mV at 10 mA/cm² in acidic solution. Especially, CoS₂ YS@C spheres also exhibit excellent stability at 10 mA/cm² for 10 h. This work provides a new revenue for the development of CoS₂-based electrocatalysts with high catalytic activity and excellent stability for HER.

1 Results and Discussion

Fig. 1 shows the schematic illustration of the fabrication of CoS₂ YS@C spheres. SEM images of CoS₂ YS spheres are shown in Figs. 2(a) and (b), which clearly shows that the diameters of CoS₂ YS spheres are about 2 μm and the surfaces of CoS₂ YS spheres are rough and made up of many small spheres. From the broken CoS₂ YS spheres, the YS structures are clearly seen as shown in Fig. 2(b), and the shell thickness is about 200 nm. This unique hollow and YS structures of CoS₂ will provide large surface areas, and they will be beneficial for the transportations of reactant and resultant and the enhancement of active sites. Then CoS₂ YS spheres coated with carbon thin layers (CoS₂ YS@C spheres) were achieved via hydrothermal treatment of CoS₂ YS spheres in glucose solution for 2 h. SEM image of CoS₂ YS@C spheres in Fig. 2(c) shows carbon layers are unifromly coated on the surfaces of CoS₂ YS spheres. Compared with those of CoS₂ YS spheres, the surfaces of CoS₂ YS@C spheres become smoother because of the coating of carbon layer. To investigate the thickness of carbon layer, HRTEM image of the edge layer of CoS₂ YS@C is measured (Fig. 2(d)), which shows that the carbon layer is uniform with thickness of about 10 nm. Fig. 2(e) shows that the inner CoS₂ owns clear lattice fringes of about 0.226 nm, which corresponds to (200) plane of CoS₂, and fast Fourier transform (FFT) parttern in Fig. 2(f) indicates the single crystal structure of CoS₂. The carbon layer is amorphous structure without lattice fringe, and FFT parttern in Fig. 2(g) confirms the amorphous structure of carbon thin layer. XRD pattern of CoS₂ YS@C is shown in Fig. 3(a), which shows that all the diffraction peaks are attributed to the standard cubic phase of CoS₂ (PDF 65-3322) and no diffraction peak of carbon is seen, furtherly confirming the crystalline structure of CoS₂ and amorphous structure of carbon layer. Here the thickness of carbon layer of CoS₂ YS@C sphere can be well controlled. When the hydrothermal treatment time of CoS₂ YS spheres in glucose solution is 3 h, the surface of CoS₂ YS@C sphere is also uniform and the carbon layer is about 15 nm. When the hydrothermal treatment time of CoS₂ YS spheres in glucose solution is 1 h, the unifrom carbon layer is about 6 nm. In order to investigate the effect of carbon layer on the electronic structure of CoS₂, XPS measurements of CoS₂ YS spheres and CoS₂ YS@C spheres were performed. In Co 2p region (Fig. 3(b)), the peaks of Co 2p_1/2 and 2p_3/2 of CoS₂ YS@C spheres at 793.9 and 778.8 eV both shift to lower binding energy compared with those of Co 2p_1/2 and Co 2p_3/2 of CoS₂ YS spheres at 795.0 and 779.9 eV, respectively, and the negetive shifts are about 1.1 eV. In S 2p region (Fig. 3(c)), the peaks of S 2p_1/2 and 2p_3/2 of CoS₂ YS@C spheres at 163.14 and 164.24 eV both shift to higher binding energy compared with those of S 2p_1/2 and S 2p_3/2 of CoS₂ YS spheres at 162.70 and 163.80 eV, respectively, and the positive shifts are about 0.44 eV. The negative shifts of Co 2p peaks and positive shifts of S 2p peaks well confirm the change of electronic structure of CoS₂ because of the strong electronic interaction between CoS₂ and carbon layers. XPS spectrum of C 1s of carbon layers is shown in Fig. 3(d), and it can be deconvoluted into three peaks at 284.8, 285.9 and 288.9 eV, which correspond to the bonds of C—C, C—O—C and O—C＝C, respectively. The existences of C—O and O—C＝C bonds in carbon layers will make CoS₂ YS@C spheres more hydrophilic, which is beneficial for the absorption of H₂O for HER^[38]. In addition, Raman measurements were used to investigate the effect of carbon layer on the electronic structure of CoS₂, and Raman spectra of CoS₂ and CoS₂ YS@C spheres are shown in Fig. 3(c). The peaks at 293.5 and 398.6 cm^-1 are observed for CoS₂, which correspond to the pure librational mode of dumb-bells (E_g) and in-phase stretching vibrations of S atom in the dumb-bells (A_g), respectively^[39], and they are in agreement with the data of CoS₂ single crystal^{[21, 40]}. However, for CoS₂ YS@C, there are about 6 cm^-1 negative shifts of E_g and A_g peaks (287.5 and 392.6 cm^-1) compared with those of CoS₂ YS spheres, as shown in Fig. 3(e), further confirming the change of electronic structure of CoS₂because of the strong electronic interaction between CoS₂ and carbon layers. To determine the content of C in CoS₂ YS@C sphere, TGA measurements of CoS₂ and CoS₂ YS@C spheres were studied in the air and the results are shown in Fig. 3(f). Compared with CoS₂, there is slow decrease among the temperature of 100—400 ℃ for CoS₂ YS@C, which corresponds to the lose of carbon in the sample. The compositions of C and CoS₂ in CoS₂ YS@C spheres are determined to be about 6% and 94% in weight, respectively.

The electrocatalytic activities of CoS₂ YS@C spheres with different carbon layer thickness are studied by linear sweep voltammetry (LSV) in 0.5 M H₂SO₄ solution at 2 mV/s. When the carbon layer thickness is 10 nm, the electrocatalytic activity of CoS₂ YS@C spheres reaches the highest level (the carbon layer thickness of CoS₂ YS@C spheres was kept to be 10 nm in all the following experiments). The electrocatalytic activities of CoS₂ YS@C spheres, CoS₂ YS spheres and carbon with the same loadings (1.02 mg/cm²) were compared and their polarization curves are shown in Fig. 4(a). Obviously, the HER catalytic activity of CoS₂ YS@C is much better than that of CoS₂ YS spheres and the carbon almost has no catalytic activity. The onset potential of CoS₂ YS@C is about 20 mV, which is much lower than those of CoS₂ YS spheres (85 mV), as shown in Fig. 4(b). The overpotential of CoS₂ YS@C at 10 mA/cm² is 89.3 mV, which is much smaller than 184.2 mV of CoS₂ YS spheres. In addition, the current density of CoS₂ YS@C at a given potential is much higher than that of CoS₂ YS spheres, as shown in Fig. 4(b). For instance, when the overpotential is 200 mV, the current density of CoS₂ YS@C spheres is about 81.23 mA/cm², which is about 6.5 times higher than that of CoS₂ YS speheres, as shown in Fig. 4(c), suggesting the important role of carbon thin layer for the enhancement of electrocatalytic activity of CoS₂ YS@C speheres.

The linear portions of Tafel plots were fit to Tafel equation (η=a+blogj, where j is current density, b is Tafel slope), yielding Tafel slope of about 55 mV/dec for CoS₂ YS@C spheres (Fig. 4(d)), which is much lower than that of CoS₂ (77 mV/dec). The Tafel slope of 55 mV/dec for CoS₂ YS@C indicates Volmer reaction has been taken place^[41-42], and the process to convert the protons into absorbed hydrogen atoms on CoS₂ YS@C surfaces becomes rate-determining step during HER. The exchange current density (j₀) of the catalyst can be calculated by extrapolating the Tafel plot. As expected, j₀ of CoS₂ YS@C spheres is 0.265 mA/cm² (Fig. 5(a)), which is much larger than that of CoS₂ (0.062 mA/cm²). Therefore, the CoS₂ YS@C spheres exhibit outstanding HER activity with low onset potential, high current density, low Tafel slope and high exchange current density, which are superior to most of the CoS₂-based electrocatalysts that have been ever reported in the acidic electrolyte.

In order to further provide the insight to CoS₂ YS@C sphere electrocatalysts, the electrochemical active surface area (ECSA) and electrochemical impedance spectroscopy (EIS) measurements were performed. Though it is difficult to obtain the accurate value of ECSA owing to the unclear capacitive behavior, it can be visualized by double layer capacitance (C_dl), which is proportional to the electrochemical surface area. The calculation of C_dl by cycle voltammograms (CVs) in 0.5 M H₂SO₄ is used to make comparison of ECSA. The C_dl of CoS₂ YS@C spheres is caculated to be 13.13 mF/cm² (Figs. 5(b) and (c)), which is much larger than that of CoS₂ YS spheres (6.56 mF/cm²), indicating that CoS₂ YS@C spheres own much larger ECSA than CoS₂ YS spheres. Nyquist plots of CoS₂ and CoS₂ YS@C spheres are shown in Fig. 5(d). The semicircle in the high frequency region is attributed to the charge transfer resistance (R_ct), which is related to the electrocatalytic kinetics, and a low value of semicircle is consistent with a fast reaction rate^[43]. From Fig. 5(d), it is clearly seen that CoS₂ YS@C spheres have much lower R_ct than CoS₂ YS spheres, indicating much faster reaction rate for CoS₂ YS@C spheres.

HER electrocatalytic activity of CoS₂ YS@C spheres is significantly better than those of CoS₂ and other CoS₂-based electrocatalysts reported in the litertatures. The enhancement of the catalytic activity of CoS₂ YS@C spheres can be ascribed to the rapid charge transfer based on analyses of EIS results and the large ECSA with more exposed active sites. Actually the better catalytic activity of CoS₂ YS@C spheres as electrocatalysts for HER is also due to the natural properties of CoS₂ YS@C spheres. As we all know, HER activity is strongly correlated with the chemisorption energy of atomic hydrogen to the electrocatalyst surface, and the hydrogen binding energy for an excellent HER electrocatalyst should be neither too high nor too low^[44-45]. The positive hydrogen binding energy on CoS₂ indicates a weak adsorption of H on CoS₂ surface^[45], which will be unfavourable to the reduction of H⁺ (i.e., Volmer step). Thus, an optimization of the electronic features is desired. It is notable that the surrounding elements have an important effect on the electron density around Co active sites. As we all know, carbon has a lower electronegativity compared with S, so the electron density around Co will increase by embedding carbon onto the surface of CoS₂. This phenomenon has been well demonstrated by XPS and Raman results. XPS binding energy of Co 2p of CoS₂ YS@C spheres obviously decreases compared with that of CoS₂, as shown in Fig. 3(b), indicating that the valence of Co in CoS₂ YS@C shperes is below +2 and the electronic density of Co will increase. In addition, E_g and A_g peaks of CoS₂ YS@C spheres shifting to low Raman shift compared with that of CoS₂ YS spheres also confirms the strong electronic interactions between CoS₂ and carbon thin layer, as shown in Fig. 3(e). Therefore, the increase of electronic density of Co will consequently enhance the strength of hydrogen binding energy to promote H_ads adsorption and thus will improve HER catalytic activity of CoS₂ YS@C spheres.

Besides the HER electrocatalytic activity, the stability is also one important criterion in evaluating the performance of electrocatalyst. The long-term HER electrocatalytic stabilities of CoS₂ YS spheres and CoS₂ YS@C spheres were tested through continuous electrolysis at 10 mA/cm² in 0.5 M H₂SO₄ for 10 h. As shown in Fig. 6(a), it is clearly seen that the CoS₂ YS@C spheres exhibit high durability with slight overpotential increase of about 32 mV at 10 mA/cm² after 10 h, whereas CoS₂ YS spheres exhibit very poor durability with obvious overpotential increase of about 517 mV after 8 h. To further gain insight into the stability of electrocatalysts, CoS₂ YS spheres and CoS₂ YS@C spheres after stability tests were further studied by SEM and XPS. It is observed that the surface morphology of CoS₂ YS@C still remains very well after stability tests, as shown in Fig. 6(b). However, for CoS₂ YS spheres, their surface morphology is seriously damaged after stability tests. So here the carbon thin layer plays a very important role for the protection from structural breakdown of CoS₂ YS@C spheres, as illustrated in Fig. 6(e). XPS characterization was also performed on CoS₂ and CoS₂ YS@C spheres after stability tests. It is clearly seen that the XPS peaks of Co 2p of CoS₂ YS@C spheres almost remain unchangeable compared with those of CoS₂ YS@C spheres before stablity tests, suggesting high chemical stability of CoS₂ YS@C spheres. However, for CoS₂ YS spheres, besides the peaks at 778.8 and 794.0 eV of Co 2p, two large new peaks appear at 783.1 and 799.9 eV, as shown in Fig. 6(c), which can be attributed to other forms of Co because of the oxidation of CoS₂. In addition, for CoS₂ YS spheres, a large peak appears at 169.2 eV in S 2p region, as shown in Fig. 6(d), indicating the evolution of S from CoS₂. However, for CoS₂ YS@C spheres, the oxidation of CoS₂ or the evolution of S are not observed as shown in Figs. 6(c) and (d). There fore, after coating carbon thin layers, CoS₂ YS@C spheres can efficiently prevent from the structural breakdown, CoS₂ oxidation or S evolution, which all are beneficial for the improvements of catalytic activity and stability.

2 Conclusions

CoS₂ YS spheres coated with carbon thin layers (CoS₂ YS@C spheres) were designed and fabricated as high-performance electrocatalysts for HER in acid media. The unique YS structure provides large space for the reactant and resultant, which is beneficial for the improvement of utilization of active sites. The carbon thin layer coating can efficently change the electronic structure of CoS₂ for the appropriate hydrogen adsorption energy to promote the Volmer step of HER, and it also can protect CoS₂ from the structural breakdown and can prevent the evolution of S from CoS₂ for the improvement of elelctrocatalytic activity and stability. Because of the above advantages, CoS₂ YS@C spheres exhibit superior catalytic activity with low onset potential of about 20 mV, small Tafel slope of about 55 mV/dec, and small overpotential of about 90 mV at 10 mA/cm². Especially, CoS₂ YS@C spheres also exhibit high durability with overpotential increase of only about 8% at 10 mA/cm² for 10 h. This work provides a new avenue for the design of high-performance electrocatalysts that are unstable during the catalysis process.

Supporting Information

Experimental Sections

The synthesis of CoS₂ spheres

CoS₂ spheres were synthesized according to the previous study reported by Lifang Jiao and her collabora-tors.¹ The details are listed as following: 1.65 mmol of CoCl₂ 6H₂O was dissolved in absolute ethanol and then was transferred into a 40 mL Teflon-lined stainless steel autoclave, then 4.1 mmol of sulfur powder was added into above solution. The Teflon-lined stainless steel autoclave was subsequently stirred for 30 min. The sealed tank was maintained at 240 oC for 24 h. After reaction was over, the autoclave cooled to the room temperature naturally. The precipitations were washed by ethanol three times and by distilled water one time and were collected by centrifugation. Finally, CoS₂ yolk-shell (YS) spheres were obtained after drying at 45 oC for 12 h.

The synthesis of CoS₂@C spheres

The synthesized CoS₂ YS spheres (60 mg) were added into 30 ml 0.05 M glucose solution and then were transferred into a 40 mL Teflon-lined stainless steel autoclave and the sealed tank was maintained at 180 ℃ for 2 h. After reaction, the autoclave cooled to the room temperature naturally. The precipitations were collected by centrifugation and washed by ethanol and distilled water, respectively. Finally, CoS₂ YS@C spheres were obtained after drying at 45℃ for 12 h. For comparisons, the different hydrothermal time, such as 1 h and 3 h, was also used for the fabrication of CoS₂@C with different thickness of carbon layer.

Material characterizations

The scanning electron microscope (SEM) and transition electron microscope (TEM) were undertaken on FEI Quanta 400 and FEI Tecnai G2 F30. Energy dispersive spectrum (EDS) mapping was investigated by INCA 300. X-ray powder diffraction (XRD) analysis was performed on Bruker D8 diffractometer using Cu K_α radiation. X-ray photoelectron spectroscopy (XPS) was processed using an ESCALAB X-ray photo-electron spectrometer. All XPS spectra were corrected using the C 1s line at 284.6 eV, and curve fitting and background subtraction were accomplished. The Fourier transform Raman (FT-Raman) spectrum was achieved on the Nicolet NXR 9650.

Electrochemical tests

CoS₂@C electrocatalysts were loaded on the glassy carbon electrode (GCE, diameter: 5mm) for testing in 0.5 M H₂SO₄ using three-electrode system. Saturated calomel electrode (SCE) and graphite rod were used as reference and counter electrodes, respectively. Typically, 10 mg catalysts and 20 μl 5% Nafion solution were dispersed in 0.48 ml ethanol through 30 min ultrasound to form homogeneous ink. For the tests, 10 μl catalyst ink was loaded onto the surface of GCE (loading 1.02 mg·cm^-2). All the potentials in this paper were referenced to a reversible hydrogen electrode (RHE) by following equation:

$ {\rm{E}}\left( {{\rm{RHE}}} \right) = {\rm{E}}\left( {{\rm{SCE}}} \right) + 0.241 + 0.059\;{\rm{pH}} $

(1)

Before measurements, the solution was purged with N₂ for 10 min to pip out O₂ dissolved in the solution. HER electrocatalytic activity of CoS₂ YS@C spheres was studied by linear sweep voltammetry (LSV) at the scan rate of 2 mV·s^-1. Double layer capacitance (C_dl) was measured by CVs using the same working electrode at the potential window of 0.19—0.27 V vs SCE. CVs were obtained at different scan rates of 2, 4, 6, 8, 10, 20 mV·s^-1. After plotting charging current density difference (Δj=j_a-j_c at the current density of 0.23 V) vs the scan rates, the slope, which is twice of C_dl, is used to represent ESCA. The chronopotentio-metry at 10 mA·cm^-2 was mesured to test the stability of CoS₂ YS@C spheres. The EIS measurements were conducted at overpotential of 0.24 V with the frequency ranging from 100 kHz to 0.1 Hz. For comparisons, bare C, CoS₂ and the physical mixture of CoS₂+C (with the same ratio of CoS₂ and C as that of CoS₂ YS@C spheres) were also mixed with Nafion solution and ethanol to form homogenous ink, respectively, and they were loaded onto the surface of GCE with the same loadings of 1.02 mg·cm^-2. The same tests of C, CoS₂ and CoS₂+C as those of CoS₂@C spheres were also measured.

N.A. represents the unknown data

References

[1] WANG Q, JIAO L, HAN Y, et al. CoS₂ hollow spheres: fabrication and their application in lithium-ion batteries. The Journal of Physical Chemistry C, 2011, 115, 8300.

[2] FABER M S, DZIEDZIC R, LUKOWSKI M A, et al. High-performance electrocatalysis using metallic cobalt pyrite (CoS2) micro-and nanostructures. Journal of the American Chemical Society, 2014, 136, 10053.

[3] HUANG J, HOU D, ZHOU Y, et al. MoS₂ nanosheet-coated CoS₂ nanowire arrays on carbon cloth as three-dimensional electrodes for efficient electrocatalytic hydrogen evolution. Journal of Materials Chemistry A, 2015, 3, 22886.

[4] WANG K, ZHOU C, XI D, et al. Component-controllable synthesis of Co(S_xSe_{1- x})₂ nanowires supported by carbon fiber paper as high-performance electrode for hydrogen evolution reaction. Nano Energy, 2015, 18, 1.

[5] PENG S, LI L, HAN X, SUN W, et al. Cobalt sulfide nanosheet/graphene/carbon nanotube nanocomposites as flexible electrodes for hydrogen evolution. Angewandte Chemie, 2014, 126(46):12802-12807

[6] FABER M S, LUKOWSKI M A, DING Q, et al. Earth-abundant metal pyrites (FeS₂, CoS₂, NiS₂, and their alloys) for highly efficient hydrogen evolution and polysulfide reduction electrocatalysis. The Journal of Physical Chemistry C, 2014, 118, 21347.

[7] XING W, ZHANG Y, XUE Q, et al. Highly active catalyst of two-dimensional CoS₂/graphene nanocomposites for hydrogen evolution reaction. Nanoscale research letters, 2015, 10, 1.

[8] LIU K, WANG F, XU K, et al. CoS_2xSe_2(1-x) nanowire array: an efficient ternary electrocatalyst for the hydrogen evolution reaction. Nanoscale, 2016, 8, 4699.

[9] OUYANG C, WANG X, WANG S. Phosphorus-doped CoS₂ nanosheet arrays as ultra-efficient electrocatalysts for the hydrogen evolution reaction. Chemical Communications, 2015, 51, 14160.  Dr. Wang Long received the Ph.D. degree in aerodynamic from Nanjing University of Aeronautics and Astronautics (NUAA) in 2014. He is currently a postdoctor at State Key Laboratory of Mechanics and Control of Mechanical Structures. He is actively pursuing research in wind turbine multi-objective design, aeroelastics and structural vibration analysis.

Acknowledgements

This work was supported by the National Basic Research Program of China (Nos.2015CB932304, 2016YFA-0202603), the Natural Science Foundation of China (No.91645104), the Natural Science Foundation of Guangdong Province (Nos.S2013020012833, 2016A010104004), and the Fundamental Research Fund for the Central Universities (No.16lgjc67).