Electrochemically activated cobalt nickel sulfide for an efficient oxygen evolution reaction: partial amorphization and phase control

Yu-Rim Hong ab, Sungwook Mhin c, Kang-Min Kim a, Won-Sik Han b, Heechae Choi de, Ghulam Ali f, Kyung Yoon Chung f, Ho Jun Lee a, Seong-I. Moon g, Soumen Dutta g, Seho Sun g, Yeon-Gil Jung h, Taeseup Song *g and HyukSu Han *a
aKorea Institute of Industrial Technology, 137-41 Gwahakdanji-ro, Gangneung-si, Gangwon 25440, Republic of Korea. E-mail: hhan@kitech.re.kr
bDepartment of Chemistry, Seoul Women's University, Seoul, Republic of Korea
cKorea Institute of Industrial Technology, 156 Gaetbeol-ro, Yeonsu-gu, Incheon 406-840, Republic of Korea
dInstitute of Inorganic Chemistry, University of Cologne, 50939 Cologne, Germany
eInstitute of Inorganic Chemistry, University of Cologne, Greinstr. 6, 50939, Cologne, Germany
fCenter for Energy Convergence Research, Korea Institute of Science and Technology, Hwarang-ro 14-gil 5, Seongbuk-gu, Seoul 02792, Republic of Korea
gDepartment of Energy Engineering, Hanyang University, Seoul 04763, Republic of Korea. E-mail: tssong@hanyang.ac.kr
hSchool of Materials Science and Engineering, Changwon National University, Changwon, Gyeongnam 641-773, Republic of Korea

Received 22nd October 2018 , Accepted 18th November 2018

First published on 20th November 2018

It has recently been demonstrated that the OER activity of transition metal sulfides (TMSs) could be enhanced by the introduction of a thin amorphous layer on a pristine surface. We report here a novel strategy to enhance the OER by developing cobalt nickel sulfide (CoxNi1−xS2, CNS) with a high density of crystalline and amorphous phase boundaries. Electrochemical activation (ECA) can partially amorphize hollow CNS nanoparticles derived from surface-selective sulfidation. The ECA-treated CNS (ECA-CNS) electrocatalyst, which is comprised of CNS nanodots separated by thin amorphous layers, shows high densities of crystalline and amorphous phase boundaries. This catalyst shows superior OER catalytic performance with a current density of 10 mA cm−2 at a small overpotential of 290 mV, a low Tafel slope of 46 mV dec−1, a high mass activity of 217 A g−1, a high turnover frequency of 0.21 s−1 at an overpotential of 340 mV, and excellent stability in alkaline media.

1. Introduction

Electrochemical water splitting is the most advanced green method of producing hydrogen from water due to its simplicity and reasonably high energy conversion efficiency.1,2 However, the slow kinetics of the oxygen evolution reaction (OER) necessitates the use of expensive novel metals, including Pt, Ru, and Ir, as catalysts. Therefore, tremendous efforts are underway to develop low-cost, high-performance, and earth-abundant electrocatalysts for electrochemical water splitting.3–17 In particular, Co- and Ni-containing compounds with boron (B), nitrogen (N), phosphorus (P), and sulfide (S). These compounds exhibited improved OER activity and stability compared to IrO2 and RuO2 electrocatalysts in alkaline solutions.18–25 To enhance the electrocatalytic activity of Co- and Ni-containing compounds, their crystalline phases and/or crystalline planes were controlled. This strategy is referred to as phase- or face-engineering.26–32

Amorphous Co and Ni compounds have been explored as high-performance electrocatalysts.14,25,33,34 Relative to their crystalline counterparts, amorphous materials possess a large number of randomly oriented bonds; these result in relatively high quantities of unsaturated surface sites in the amorphous phase, which facilitate the adsorption of reactants.20,35 The local structural flexibility inherent to amorphous phases also enhances electrocatalytic performance by accelerating changes associated with the bonding of intermediates on the catalyst surface. However, interestingly, purely amorphous electrocatalysts often exhibit lower catalytic activity than the crystalline and amorphous mixed phase possibly due to lower conductivity.36,37 In addition, the amorphous surface layer of transition metal sulfides (TMSs), which is formed during electrochemical reactions, enables significant improvement of the OER activity.25,36,38,39 These findings imply that the interface between the crystalline and amorphous phases may be responsible for the observed enhancement in the OER activity. For TMSs, from a thermodynamic point of view, the amorphous layer on the surface of the electrocatalyst is spontaneously formed during electrochemical reactions under oxidative conditions. Although the phase- or face-engineered TMS electrocatalysts with nano-dimensions could significantly enhance the electrochemical performances, the synthesis of these compounds with a specific phase or plane is challenging in nanostructured materials due to the facile formation of secondary phases, surface oxidation, and the segregation of impurities. Therefore, a new materials design and synthesis strategy for TMS nanomaterials must be developed for their practical use as catalysts.

We report herein a rational approach to boost the OER activity of TMSs, cobalt nickel sulfide, by engineering the crystalline/amorphous phase boundaries. Notably, CoS2 has a similar crystal structure to hydrogenase, which typically has a Ni center that is critical for the water splitting reaction. Thus, the incorporation of Ni into CoS2 can promote of the catalytic properties. Considering the Co–Ni–S ternary system, although there has been some research on hydrogen evolution reaction (HER) catalysis, a deep understanding of the catalytic mechanism of the OER remains elusive. Furthermore, most of the research on the Co–Ni–S ternary system has been restricted to spinel compounds (i.e., CoNi2S4 and Co2NiS4), and there are few reports on CoxNi1−xS2 (CNS, space group Pa[3 with combining macron]) as catalysts for water oxidation. Hence, we designed hollow nanostructured CNS with well-defined cavities and chemical functionality in order to utilize intriguing structural features such as a kinetically favorable open structure, large surface area, and surface permeability of shell materials. Briefly, Ni–Co hydroxide nanoprisms were synthesized and a surface-selective sulfidation process using different diffusion kinetics for each ion was followed to obtain CNS with a uniquely hollow interior structure. The crystalline/amorphous phase boundaries of the hollow nanostructure of CNS were engineered by the electrochemical activation (ECA) process. The ECA-treated CNS (ECA-CNS) has a unique nanostructure consisting of highly crystalline CNS nanodots separated by a partially amorphized phase. The resulting high-density crystalline/amorphous phase boundaries could significantly boost the electrocatalytic OER performance of CNS, which is one of the highest among the recently reported Co and Ni sulfides and oxides including carbon-based materials. Based on the first principles calculation combined with X-ray absorption spectroscopy analysis, the excellent OER performance of ECA-CNS was attributed to the undercoordinated active centers in CNS originating from its intrinsic crystal structure of Pa[3 with combining macron] and the parallelized OER reaction pathway at the crystalline–amorphous interface.

2. Experimental section

2.1 Synthesis of Ni–Co precursors

All the chemicals were purchased from Sigma Aldrich and were directly used without further purification. Typically, a mixture of metal acetate tetrahydrate (1.28 g) with a 2[thin space (1/6-em)]:[thin space (1/6-em)]1 molar ratio of Ni[thin space (1/6-em)]:[thin space (1/6-em)]Co was dissolved in 200 mL of ethanol at room temperature. The solution was then refluxed at 85 °C for 4 h. The precipitate was centrifuged and rinsed with ethanol several times. The resultant powder is subsequently dried in air at 60 °C.

2.2 Synthesis of CNS-rGO

CNS nanoparticles were supported on reduced graphene oxide (rGO), a conductive porous support, to facilitate charge and mass transport during water oxidation. CNS-rGO was prepared via a facile one-step hydrothermal process. In a typical procedure, GO was prepared via the chemical exfoliation of graphite powder following a modified Hummers' method. A solution of GO (20 mL, concentration of 0.2 g L−1) in 50 mL of deionized (DI) water was mixed with 40 mg of the Ni–Co precursor and 0.0563 g of thioacetamide, which is used as the sulfur source, by vigorous stirring. Subsequently, the solution was further ultrasonicated for 20 min to homogeneously disperse the Ni–Co precursor in the GO solution. The resulting mixture was then transferred to a 200 mL Teflon®-lined stainless-steel autoclave and hydrothermally reacted at 200 °C for 24 h. The resultant product was washed several times with DI water and freeze-dried at −50 °C for 3 days. For the synthesis of CS-rGO and NS-rGO, Ni and Co precursors were prepared by using the same procedure as for the Ni–Co precursor and mixed with a GO solution containing thiourea for further hydrothermal reaction.

2.3 Synthesis of Ni2CoS4

The Ni–Co precursor (80 mg) was dispersed in 40 mL of ethanol containing 0.1125 g of thioacetamide. The mixed solution was transferred into a 200 mL Teflon®-lined stainless-steel autoclave and hydrothermally reacted at 120 °C for 6 h. The precipitate was centrifuged and washed with ethanol several times. The resultant powder was subsequently dried in air at 60 °C.

2.4 Characterization

Scanning electron microscopy (SEM; model S4800; Hitachi) was used to obtain microstructural images of each sample. Transmission electron microscopy (TEM) images and the corresponding energy dispersive X-ray (EDX) mapping images were acquired using a Talos F200X (Thermo Fisher Scientific) microscope equipped with an EDX analyzer at 200 kV. X-ray diffraction (XRD) patterns were measured using a D/MAX-2500/PC (Rigaku) diffractometer with Cu-Kα radiation (λ = 0.15418 nm) at 40 kV and 100 mA. Raman spectroscopy was performed using a dispersive laser spectrophotometer (model NRS-3100; JASCO) at room temperature with an excitation wavelength of 633 nm. X-ray photoelectron spectroscopy (XPS) spectra were recorded for the samples using a VG ESCALAB 200i instrument (Thermo Fisher Scientific). XPS survey and high-resolution scans were conducted with pass energies of 100 and 20 eV, respectively. The specific surface area was measured using nitrogen adsorption–desorption tests (model TriStar II 3020; Micromeritics) following the Brunauer–Emmett–Teller (BET) theory. The experimental methods for electrochemical characterization, X-ray absorption spectroscopy (XAS), and density functional theory (DFT) are summarized in the ESI.

3. Results and discussion

3.1 Materials design strategy: diffusion kinetics controlled sulfidation and electrochemical activation process

Scheme 1 shows the materials design strategy that was used for the synthesis of ECA-CNS. Nanoprisms of Ni–Co acetate hydroxide, synthesized in accordance with a previous report, were used as precursors for preparing CNS nanoparticles (Scheme 1a).39 CNS nanoparticles, supported on rGO sheets, were produced by hydrothermally reacting the Ni–Co precursor with thioacetamide as a sulfur source. The formation of hollow CNS structures occurs via the reaction of sulfide ions with the transition metal ions of the Ni–Co nanoprisms governed by the diffusion effect resulting in the formation of a thin sulfide layer (Scheme 1b).40,41 The newly formed thin layer of Ni–Co sulfides can act as a physical barrier preventing outside sulfide and inside metal ions from undergoing a direct chemical reaction. Further reaction relies on the relative diffusion kinetics of metal and sulfide ions through the thin sulfide barrier. Given the relatively small size of transition metal cations compared to that of sulfide anions, the outward diffusion of metal ions should be much faster than the inward diffusion of sulfide ions. Therefore, sulfidation and the consequent formation of CNS nanoparticles occur primarily on the outer surface of the sulfide barrier, while the interior of the nanoparticles becomes empty. In addition, high isostatic pressure incurred during hydrothermal processing changes the initial nanoprism morphology to a rounded nanoparticle (Scheme 1c). Subsequent ECA treatment collapses the shell and the CNS nanoparticles are deposited onto rGO supports (Scheme 1d). Finally, the ECA treatment partially amorphizes or oxidizes the CNS nanoparticles, resulting in a high density of crystalline/amorphous phase boundaries.
image file: c8ta10142f-s1.tif
Scheme 1 Schematic illustration of the preparation of (a–c) hollow CNS-rGO nanoparticles in a GO solution and (d) morphology change after the ECA process.

3.2 Structural and compositional characterization

The morphology of the Ni–Co precursors with a Ni[thin space (1/6-em)]:[thin space (1/6-em)]Co molar ratio of 2[thin space (1/6-em)]:[thin space (1/6-em)]1 is presented in Fig. S1a, which shows a highly homogeneous tetragonal nanoprism structure with an average particle size of ca. 200 nm. The XRD pattern of the Ni–Co precursor matches that of a tetragonal Ni–Co acetate hydroxide phase, (Ni,Co)5(OH)2-(CH3COO)8·2H2O (Fig. S1b). The prepared Ni–Co nanoprisms were ultrasonically mixed in an aqueous solution of GO, which was prepared by following a modified Hummers' method.42 The resultant mixture was then hydrothermally reacted to convert the Ni–Co hydroxide precursors to CNS nanoparticles supported on rGO sheets. Fig. S2 shows FE-SEM and low-resolution TEM images of CNS-rGO. Hollow CNS nanoparticles with a diameter of ca. 100 nm were dispersed uniformly onto ultrathin rGO sheets. No aggregation of CNS particles was observed due to the high surface area of the rGO. A nitrogen adsorption–desorption isotherm of CNS-rGO was measured, yielding a BET surface area of 71.4 m2 g−1 (Fig. 1a), which is comparable to those of other rGO-based materials reported elsewhere.42 High-resolution TEM micrographs provided further insight into the detailed structure of CNS-rGO. Fig. 1(b) shows the hollow structure of CNS where CNS nanoparticles mainly populated the outer shell surface. The average size of a CNS nanoparticle at the outer surface was ca. 30–50 nm. The HR-TEM image and fast Fourier transform (FFT) pattern (Fig. 1(c and d), respectively) indicate that CNS nanoparticles are crystallized in the cubic CoxNi1−xS2 phase (space group Pa[3 with combining macron]), which is consistent with XRD analyses; the XRD pattern of CNS-rGO (Fig. S3) indicates a cubic CoxNi1−xS2 phase (ICDD 98-602-4479).
image file: c8ta10142f-f1.tif
Fig. 1 (a) N2 adsorption–desorption isotherm of CNS-rGO and (b) TEM image of CNS-rGO with a typical hollow structure. (c) HR-TEM image of CNS-rGO and (d) the corresponding FFT pattern. (e) XPS spectra of CNS-rGO before and after the electrochemical activation process. (f) TEM image of ECA-CNS with the (g) corresponding elemental mapping images for Co, Ni, S, O, S, and N. (h) HR-TEM image of ECA-CNS (i) showing CNS nanodots separated by amorphous phases with (j) a magnified view of the crystalline/amorphous phase boundary.

3.3 Effect of electrochemical activation on nanostructure: high density crystalline–amorphous phase boundary

ECA treatment in an alkaline solution (1 M KOH) significantly changed the overall morphology and chemical bonding states of the CNS nanoparticles by partial oxidation. Fig. 1e shows the XPS spectra of O 1s for CNS-rGO before and after ECA. In particular, the formation of a surface hydroxide/oxyhydroxide layer on CNS-rGO due to partial oxidation can be clearly seen by the increased peak intensity at 535.2 eV attributed to a surface dominated by oxygen species. In addition, a positive peak shift for lattice oxygen species centered at 531.3 eV indicates the formation of surface oxygen defect/vacancy species, revealing that partial amorphization occurred on the surface of CNS during ECA treatment.25 The hollow structure of the CNS nanoparticles disappears after ECA treatment. However, CNS nanodots with a size of ca. 5–10 nm are formed in the ECA-CNS nanoparticles (Fig. 1f and h). Energy-dispersive X-ray spectroscopy (EDS) mapping revealed uniform distributions of Co, Ni, S, N, and O in the ECA-CNS nanoparticles (Fig. 1g). Oxygen is likely a by-product of the ECA treatment due to partial oxidation, while nitrogen may be incorporated from thioacetamide, which was used as the sulfur source, into the rGO layer. Fig. S4 shows the quantitative analysis of the EDS mapping data of ECA-CNS. The Ni[thin space (1/6-em)]:[thin space (1/6-em)]Co ratio (1.7[thin space (1/6-em)]:[thin space (1/6-em)]1) was well comparable with the initial value. Also, the transition metal to sulfur ratio was about 1[thin space (1/6-em)]:[thin space (1/6-em)]2 indicating the existence of the CoxNi1−xS2 phase after the ECA process. The oxygen content was relatively high possibly due to partial oxidation of the CoxNi1−xS2 phase or oxygenous functional groups in rGO. Notably, HR-TEM micrographs and FFT patterns acquired from different areas reveal that amorphous phases were created within crystalline ECA-CNS nanodots (Fig. 1(h and i)). The ECA-CNS nanodots exhibited a lattice fringe d-spacing of 3.22 Å, indicating a (111) plane of the CoxNi1−xS2 phase (space group, Pa[3 with combining macron]) (Fig. 1j). This indicates that the initial crystal structure of the CNS nanoparticles was retained during ECA treatment, while the regions within the ECA-CNS nanodots are partially oxidized in amorphous phases. In ECA-CNS, partial amorphization of nanodot-decorated CNS nanoparticles significantly increases the density of crystalline/amorphous phase boundaries to levels surpassing those of only surface oxidized TMSs.

Time-dependent TEM microscopy was used to show the morphological changes of CNS. Fig. 2 shows the intermediate products obtained at different reaction stages. Octagonal structures were observed after 2 h (Fig. 2a), indicating that the corners of the nanoprisms may have possessed a greater defect density. As the reaction continued, further etching occurs towards the inner side to hollow out the nanoparticle (Fig. 2b). The inner part becomes almost empty after 15 h (Fig. 2c), while CNS nanoparticles remained along the outer surface. Subsequent ECA treatment collapsed the hollow nanostructure of the CNS nanoparticles and deposited them onto the rGO supports, as illustrated in Scheme 1(d).

image file: c8ta10142f-f2.tif
Fig. 2 TEM images of CNS-rGO products obtained by hydrothermal processing at different reaction times, (a) 2 h, (b) 7 h, and (c) 15 h, illustrating the structural evolution of the exterior and interior of the CNS-rGO hollow nanoparticle. The corresponding elemental mapping images for Co, Ni, and S are also shown at the right side.

The incorporation of N into the rGO support was confirmed by the high-resolution C (1s) and N (1s) XPS spectra of ECA-CNS (Fig. S5(a and b)), which support the results of EDS mapping (Fig. 1g). The C (1s) spectrum of ECA-CNS was deconvoluted into three peaks with binding energies of 284.6, 285.2, and 286.0 eV. These were assigned to sp2-hybridized carbon, sp2-C bonded to N, and C–OH, respectively.43–45 In addition, the high-resolution N (1s) spectrum showed that N in ECA-CNS consisted primarily of quaternary nitrogen. The S (2p) spectrum contained four peaks at binding energies of 162.5 (S–Co), 164.0 (C–S–C), and 168.5 and 169.0 (C–SOn–C) eV, indicating the formation of chemical bonds between Co, C, and S atoms (Fig. S5(c)).46–48 The Co (2p) spectrum of ECA-CNS shows the main binding energies of the Co 2p3/2 and Co 2p1/2 peaks at 780.8 and 796.7 eV which correspond to the peaks for Co2+, revealing the existence of CoO or Co(OH)2 formed during ECA treatment (Fig. S5(d)). Similarly, for the Ni (2p) spectrum (Fig. S5(e)), the major peaks of Ni 2p3/2 and Ni 2p1/2 were observed at around 857.5 and 874.2 eV, respectively, along with the satellite peaks, indicating the presence of surface oxide or hydroxide species.25,37

3.4 Electrochemical property characterization

The electrocatalytic activity of CNS-rGO was investigated using a typical three-electrode system with a rotating disk electrode (RDE) in basic media (1 M KOH aqueous solution). For comparison, CS-rGO, NS-rGO, Ni–Co precursor, and Ni2CoS4 were also tested under the same conditions. The XRD patterns and SEM micrographs of CS-rGO, NS-rGO, and Ni2CoS4 are shown in Fig. S6, indicating similar crystal structures and morphologies to those of CNS-rGO. Linear sweep voltammetry (LSV) was performed on all samples at a scan rate of 5 mV s−1. All potentials were iR-compensated and referenced to a hydrogen electrode (RHE). The electrolyte was bubbled with O2 gas for 10 min prior to acquiring any electrochemical data and the electrode was initially subjected to continuous potential cycling at a scan rate of 100 mVs−1 between 0.95 and 1.65 V for 40 cycles. During these measurements, the RDE was continuously rotated at 2000 rpm to avoid the accumulation of gas bubbles on the electrode. The LSV plot of the samples for the OER in 1.0 M KOH are shown in Fig. 3a. For CNS-rGO and NS-rGO, the anodic peaks centered at about 1.35 and 1.37 V are associated with the reversible redox process, Co(OH)2/Ni(OH)2 + OH ↔ CoOOH/NiOOH + H2O + e.49–51 CNS-rGO exhibited the highest OER activity among the tested samples. The overpotential (η) required to deliver a current density of 10 mA cm−2 (η10) is often used as a reference to evaluate electrocatalytic performance in terms of solar-to-fuel conversion.6 The CNS-rGO required an η10 of 320 mV, while CS-rGO, NS-rGO, Ni–Co precursor, and Ni2CoS4 needed η10 of 340, 360, 330, and 340 mV, respectively. These results show that CNS-rGO had a much higher electrocatalytic activity than CS-rGO, NS-rGO, Ni–Co precursor, and Ni2CoS4. The OER activity of rGO without transition metal sulfides was also measured for the comparison purpose, and negligible OER activity was observed (Fig. S7), demonstrating that a significant OER activity is originated from the TMSs rather than the rGO supports. Fig. S8 shows the effects of Ni–Co precursor loading amounts on the OER activity of CNS-rGO. The XRD patterns of catalysts prepared with different loading amounts (20–80 mg) of the Ni–Co precursor are also presented in Fig. S9. The results show that the crystal structure of CoxNi1−xS2 was well preserved regardless of the loading amount. A loading of 40 mg Ni–Co precursor resulted in the highest OER activity. Therefore, this sample was used in all subsequent electrochemical analyses.
image file: c8ta10142f-f3.tif
Fig. 3 (a) LSV curves (scan rate 0.5 mV s−1) and (b) the corresponding Tafel plots of the CNS-rGO, CS-rGO, NS-rGO, Ni–Co precursor, and Ni2CoS4 catalysts loaded on the rotating disc electrode (2000 rpm) in an O2-saturated 1.0 M KOH solution. (c) The current difference between the anodic and cathodic sweeps as a function of the scan rate. The dashed lines are a linear fitting of the data; the slopes of the fitted lines were used to calculate the electrochemical effective surface area. (d and e) LSV curves (scan rate 0.5 mV s−1) and the corresponding Tafel plots of CNS-rGO (black curve) and ECA-CNS (red curve). (f–h) Current densities, mass activities, and TOFs at η = 340 mV. (i) Chronoamperometry curve of CNS-NGA under an applied voltage of 1.6 VRHE in a 1 M KOH solution.

Fig. 3b shows that the Tafel slope derived from the data was 52 mV dec−1 for CNS-rGO, which is much lower than those of CS-rGO (87 mV dec−1), NS-rGO (73 mV dec−1), and Ni2CoS4 (189 mV dec−1). The Tafel slope is a useful indicator of reaction kinetics; a smaller Tafel slope corresponds to faster kinetics for electrochemical reactions involving charge transfer. Thus, the reaction kinetics of the OER on CNS-rGO were much faster than those on other samples, indicating a synergetic effect of Ni and Co on OER activity. The Tafel slope of the Ni–Co precursor was comparable to that of CNS-rGO, while the Tafel slope of Ni2CoS4 (189 mV dec−1) was significantly lower than that of CNS-rGO. This implies that the presence of Ni and Co hydroxides is also important for boosting the OER. Electrochemical impedance spectroscopy (EIS) showed that CNS-rGO has a much smaller charge-transfer resistance (Rct ∼ 16 Ω, estimated using Randles equivalent circuit model) than the other catalysts (Fig. S10), indicating that more facile electrical transport occurs on CNS-rGO. The electrochemically effective surface area (ECSA) of the catalysts based on their double-layer capacitance (Cdl) was estimated using cyclic voltammetry (CV) at different scan rates. To obtain the capacitive currents related only to double-layer charging, scan-rate dependent CV was performed between 1.41 and 1.46 V, where redox processes do not occur (Fig. S11). The Δj = jajc at 1.435 V is plotted against the scan rate, where the linear slope is twice of the Cdl.52,53Fig. 3c shows that the linear slope followed the sequence of CNS-rGO > CS-rGO > NS-rGO > Ni2CoS4 > Ni–Co precursor, implying that the high OER activity of CNS-rGO may also be related to its large ECSA.

The ECA treatment of CNS-rGO was achieved by applying an oxidation potential of 1.55 V in 1.0 M KOH solution for 1 h. A representative current profile during ECA is presented in Fig. S12, which clearly shows a continuous increase in current during the activation process. Fig. 3d shows the OER performance of CNS-rGO before and after activation. ECA-CNS exhibited a significantly lower η10 (290 mV) than CNS-rGO (η10 = 320 mV). ECA treatments with different times (30 min and 2 h) were also performed to study the relationship between the time of ECA and the OER performance. Note that the sample which is electrochemically activated for 30 min shows slight enhancement in OER performance, while 2 h activation results in water oxidation activity comparable with that of the sample activated for 1 h. The number of crystalline and amorphous boundaries may attain the maximum value during ECA, indicating that an activation time longer than 1 h would not change the water oxidation activity. Thus, the optimal ECA time for CNS-rGO was determined to be 1 h. In addition, the Tafel slope of ECA-CNS decreased from 52 to 46 mV dec−1 after activation (Fig. 3e), indicating enhanced charge transfer and activity for the OER on ECA-CNS due to the high density of crystalline/amorphous phase boundaries. To more quantitatively define the OER activity of the catalysts, mass activity can be calculated as,

mass activity = j/c(1)
where j is the current density at a given potential and c is the loading amount of catalyst on the electrode, which is equal to 0.285 mg cm−2 in this work. Fig. 3f shows the current density of the catalysts at an overpotential of 340 mV. The current density of ECA-CNS (60 mA cm−2) was much greater than those of the other catalysts: six-fold higher than those of CS-rGO, NS-rGO, and the Ni–Co precursor, and three-fold higher than that of CNS-rGO. The corresponding mass activities of the catalysts are presented in Fig. 3g, which shows a trend similar to that of current density. Most notably, the mass activity of ECA-CNS reached 217 A g−1 at the given potential. Also, turnover frequency (TOF) is another important quantitative metric for electrocatalysts and can be calculated by,
TOF = j × A/(4 × F × n)(2)
where j is the current density at the given overpotential, A is the area of the electrode, F is the Faraday constant (96[thin space (1/6-em)]485 C mol−1), and n is the number of moles of active sites of the metal (i.e., Co or Ni) in the catalysts, which can be calculated from the XPS results. The data in Fig. 3h show that the TOF was as high as 0.21 s−1 for ECA-CNS at an overpotential of 340 mV, which is much higher than those of CNS-rGO (0.07 s−1), Ni–Co precursor (0.04 s−1), CS-rGO (0.03 s−1), and NS-rGO (0.02 s−1). Remarkably, the TOF of ECA-CNS was about three-fold greater than that of CNS-rGO, demonstrating the significance of ECA for enhancing the OER activity. Based on the results from the activity metric analysis, one can conclude that ECA-CNS exhibits much better OER activity than the other catalysts tested in this work as well as the benchmark RuO2 catalyst; the electrocatalytic activity of RuO2 for the OER, including LSV and ECSA, is summarized in Fig. S13 and S14. A rotating ring disk electrode (RRDE) was employed to determine the OER faradaic efficiency (FE) of ECA-CNS (Fig. S15). A ring current of 0.095 mA was detected when a constant current of 0.48 mA was applied to the disk electrode for generating O2, indicating that an OER FE of ∼99.9% is achieved for ECA-CNS (see the ESI for the calculation method). Thus, the observed catalytic oxidation current for ECA-CNS can be significantly attributed to the oxygen evolution with high FE. Interestingly, the EIS data show negligible differences between the Rct values of ECA-CNS (ca. 19 Ω) and CNS-rGO (ca. 17 Ω) (Fig. S16). Moreover, as shown in Fig. S17 and S18, the estimated Cdl of ECA-CNS (21.28 mF cm−2) is smaller than that of CNS-rGO (42.94 mF cm−2). These results strongly suggest that the improved OER activity of ECA-CNS cannot be attributed solely to changes in charge transfer conductivity and the high electrochemically active surface area. Our data suggest that the high density of crystalline/amorphous phase boundaries formed during ECA treatment can be responsible for the exceptionally high OER activity of ECA-CNS. The OER activity of ECA-CNS rivals or outperforms that of other state-of-the-art OER catalysts in alkaline media (Table S1, ESI).

Besides good activity, long-term stability is also important for practical applications of electrocatalysts. The electrocatalytic stability of ECA-CNS was evaluated using chronoamperometric measurements at an overpotential of 340 mV. After 10 h of the OER, the current density of ECA-CNS decreased by less than 5% of its initial value (Fig. 3i), demonstrating good stability in aqueous alkaline media. Additionally, the stability of ECA-CNS was tested by continuous cycling in a potential window of 1.15–1.55 V for 1000 cycles. CV curves were almost comparable before and after 1000 cycles (Fig. S19), indicating the good stability of ECA-CNS in alkaline media. The structure and composition of ECA-CNS after the long-term OER test were further studied. Fig. S20 shows TEM images for ECA-CNS after the long-term OER test, confirming that the original morphological and compositional features of ECA-CNS electrocatalysts, such as CNS nanodots deposited on rGO sheets, homogeneous elemental distribution, and partial oxidation within CNS nanodots, were well retained. Fig. S21 shows quantitative data of EDS mapping results on ECA-CNS after the long-term OER test, indicating a large increase of oxygen content due to surface oxidation of TMSs during the prolonged OER in alkaline media. XRD data for the post-OER sample are also shown in Fig. S22. No significant change was observed in XRD patterns for ECA-CNS before and after the long-term OER test. The reduced peak intensity may be attributed to the formation of amorphous phases. These results indicate that the crystal structure of CNS is well retained during the OER. The surface or interior amorphous hydroxide layers may be able to effectively protect crystalline CNS from further oxidation under OER conditions. Overall, electrochemical characterization shows that ECA-CNS is a high-performance electrocatalyst for the OER with superior activity in alkaline aqueous media, demonstrating that engineering of the crystalline/amorphous phase boundary can be a promising design strategy for boosting the electrocatalytic properties of TMSs.

3.5 Effects of phase control and partial oxidation on OER performance

To gain more accurate structural information about the local bonding nature of CoxNi1−xS2, synchrotron-based hard X-ray absorption near edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) spectroscopy were performed on CoxNi1−xS2 (space group Pa[3 with combining macron]) and also on Ni2CoS4 (space group Fd[3 with combining macron]m) as the control sample. Fig. 4(a and b) show the XANES spectra measured at the Co and Ni K-edges, respectively, along with 2+ (Co2+/Ni2+) and 3+ (Co3+/Ni3+) standards. The spectra show two kinds of features i.e., pre and main edge peaks. The inset of Fig. 4a shows pre-edge features where CoxNi1−xS2 shows less intense peaks which are related to the less distorted octahedral symmetry of Co–S interactions. The absorption energy of the elements is usually considered at the middle (0.5 height of the normalized spectrum) of the main edge peak. The XANES spectra of both elements show similar absorption energy positions and they cross each other almost at the middle of the absorption edge. However, the lower and upper parts of the absorption edge of both spectra differ which is related to the arrangement of the crystal geometry. Fig. 4b shows the XANES spectra of the Ni K-edge of CoxNi1−xS2, Ni2CoS4, and standard materials. The pre-edge of Ni2CoS4 shows relatively high intensity which is due to the fact that Ni atoms in Ni2CoS4 are bonded with S atoms in both tetrahedral and octahedral coordination as shown in the crystal structure representation (Fig. S23). The energy position of the main absorption edge of both materials is almost similar as they cross each other at the middle, indicating the similar oxidation state of Ni atoms but different local structure.
image file: c8ta10142f-f4.tif
Fig. 4 XANES of CoxNi1−xS2 and Ni2CoS4 at (a) the Co K-edge and (b) the Ni K-edge plotted with reference spectra. EXAFS of CoxNi1−xS2 and Ni2CoS4 at (c) the Co K-edge and (d) the Ni K-edge.

The local structure around Co and Ni atoms in both materials was investigated using EXAFS spectroscopy and the Fourier transform curves were fitted which provides quantitative structural information. The fitted EXAFS spectra of CoxNi1−xS2 and Ni2CoS4 are shown in Fig. 4(c and d) for the Co and Ni K-edge, respectively, and the calculated parameters such as coordination number (N), bond length (R), and Debye–Waller factor (σ2) are listed in Tables 1 and 2. The first peak at around 1.6 Å shown with a dotted rectangle in Fig. 4c is assigned to the Co–S interaction and the calculated parameters show that Co is bonded with the 6 nearest S atoms to form octahedral symmetry with average distances of 2.2995 and 2.2756 Å for CoxNi1−xS2 and Ni2CoS4, respectively. The second dotted rectangular region includes peaks of Co–M (M = Co, Ni) and Co–S interactions where the fitted parameters are listed in Table 1. The Co atom in CoxNi1−xS2 shows much lower coordination numbers than Ni2CoS4. Fig. 4d shows the fitted EXAFS spectra for the Ni K-edge and the first peak at around 1.6 Å corresponds to the Ni–S interaction. The fitted parameters of the first peak show much higher coordination numbers of Ni–S for Ni2CoS4 than for CoxNi1−xS2. Ni is bonded with S in both octahedral and tetrahedral symmetry in Ni2CoS4, while Ni in CoxNi1−xS2 shows only octahedral symmetry. Ni and Co atoms are generally considered as catalytic active centers in Co–Ni–S compounds.54,55 The under-coordinated Ni and Co atoms in the CoxNi1−xS2 phase can be beneficial for enhancing catalytic activity.

Table 1 Structural parameters obtained from the Co K-edge EXAFS spectra
Material Path Coordination number ΔE (eV) Bond length (Å) Debye–Waller factor (σ2 × 10−3) Å2
CoxNi1−xS2 Co–S 6 ± 0.5 0.312 ± 1.3 2.2995 (±0.053) 1.3 (±0.93)
Co–S 4 ± 1.1 0.312 ± 1.3 3.9805 (±0.081) 2.1 (±0.33)
Co–M 4 ± 0.8 0.312 ± 1.3 4.2351 (±0.091) 3.2 (±0.24)
Ni2CoS4 Co–S 6 ± 0.5 0.398 ± 1.1 2.2756 (±0.011) 2.3 (±0.06)
Co–S 8 ± 1.2 0.398 ± 1.1 3.9879 (±0.021) 3.8 (±0.91)
Co–M 6 ± 2.1 0.398 ± 1.1 3.3413 (±0.039) 5.1 (±2.09)
Co–M 4 ± 1.4 0.398 ± 1.1 3.9205 (±0.164) 2.1 (±0.19)

Table 2 Structural parameters obtained from the Ni K-edge EXAFS spectra
Material Path Coordination number ΔE (eV) Bond length (Å) Debye–Waller factor (σ2 × 10−3) Å2
CoxNi1−xS2 Ni–S 6 ± 1.1 0.215 ± 0.4 2.4095 (±0.013) 3.2 (±1.91)
Ni–S 4 ± 0.8 0.215 ± 0.4 4.0521 (±0.011) 1.3 (±1.09)
Ni–M 4 ± 1.3 0.215 ± 0.4 4.2509 (±0.021) 4.4 (±2.21)
Ni2CoS4 Ni–S 10 ± 2.3 0.833 ± 1.2 2.2039 (±0.031) 3.5 (±3.31)
Ni–S 16 ± 4.1 0.833 ± 1.2 3.9355 (±0.101) 1.9 (±2.22)
Ni–M 6 ± 1.9 0.833 ± 1.2 3.3365 (±0.023) 2.2 (±1.32)
Ni–M 14 ± 1.3 0.833 ± 1.2 3.9515 (± 0.101) 4.2 (± 1.33)

The adsorption energies of OH, O, OOH, and OO were calculated to obtain the OER energy landscapes for the CoxNi1−xS2 and spinel Ni2CoS4 phases using density functional theory (DFT) analysis. The Gibbs free energies of the reactions (ΔG) were obtained using the following equation:

ΔG = ΔE + ΔZPE − TΔS + ΔGU,(3)
where ΔE is the calculated total energy difference, ΔZPE is the zero-point energy correction, TΔS is entropy, and ΔGU is the free energy term. CoxNi1−xS2 was modelled with a composition of Co5Ni9S28 following our synthesis conditions. The calculated free energy diagrams for OH, O, OOH, and OO adsorptions on the CoxNi1−xS2 and Ni2CoS4 systems with U = 1.23 V are presented in Fig. 5. To obtain the free energy diagrams for the energetically favored adsorption sites, we fitted numerous adsorption energy calculations for different sites with different angles of adsorbates on substrates. The reaction steps (I to IV) corresponding to OH, O, OOH, and OO adsorption on CoxNi1−xS2 and Ni2CoS4 substrates are described as follows:
(I) H2O(1) → OHads + H+ + e(4)
(II) OHads → Oads + H+ + e(5)
(III) O + H2O(1) → HOOads + H+ + e(6)
(IV) HOO → O2(g) + H+ + e(7)

image file: c8ta10142f-f5.tif
Fig. 5 Calculated OER energy landscapes for CoxNi1−xS2, Ni2CoS4, and CNSO (surface oxidized CNS) with U = 1.23 V. The inset shows the crystal structure of the CoxNi1−xS2 (space group Pa[3 with combining macron]), Ni2CoS4 (space group Fd[3 with combining macron]m), and CNSO phases, where blue, grey, yellow, and red filled circles are cobalt, nickel, sulfur, and oxygen atoms, respectively.

On the CoxNi1−xS2 surface, reaction step I is spontaneous, as indicated by the negative energy change, while a substantial energy barrier of 1.8 eV is present for Ni2CoS4. Reaction step III has the highest energy barriers of 2.1 and 2.6 eV for the CoxNi1−xS2 and Ni2CoS4 phases, respectively, and acts as the rate determining step (RDS) for the OER. In addition, reaction step IV occurs with a negative energy change on the surface of the CoxNi1−xS2 phase; however an energy barrier of 0.7 eV still needs to be overcome in the Ni2CoS4 phase. Hence, the spontaneous reaction in step I and step IV and the significantly reduced energy barrier for the RDS in the CoxNi1−xS2 phase are responsible for the excellent electrocatalytic OER performance of CNS. In addition, we modeled the partially oxidized surface of CNS (CNSO), where approximately 0.17 mol fraction of S is substituted by O, and obtained the OER energy diagrams (Fig. 5) in order to study the effect of partial oxidation of CNS on the OER. In the CNSO system, relatively huge negative energy drops were found when the reaction steps change from I to II as well as II to IV. This indicates that the OER pathway can be parallelized by the partially oxidized phase at the crystalline–amorphous interface. The CoxNi1−xS2 phase is the most important active phase for the OER due to possessing the lowest energy barrier for the RDS while the CNSO phase can co-assist to facilitate the reaction kinetics. Thus, the high number of crystalline–amorphous interfaces in ECA-CNS should be responsible for the superior electrocatalytic water oxidation properties.

Fast OER reaction kinetics in the CoxNi1−xS2 phase may be attributed to the under-coordinated Ni and Co atoms as revealed by EXAFS analysis, resulting in unsaturated electronic configurations at the surface. Thus, adsorbents or intermediates for the OER can be easily formed within each step, facilitating the overall OER process. Recently, we demonstrated that for the Co–Ni–S system, Ni and S atoms play crucial roles in the OER process for reaction steps I and IV, respectively.56 As seen in the EXAFS spectra for the Ni K-edge (Table 2), the Ni–S bond exhibits a pronounced under-coordinated bonding nature compared to that in the Ni2CoS4 spinel counterpart. Hence, the electron-deficient Ni and S atoms in the CoxNi1−xS2 phase can enhance catalytic activities and charge transfers at both sites, resulting in favorable reaction kinetics for steps I and IV. Notably, the CoxNi1−xS2 phase has been rarely studied as an electrocatalyst for water oxidation, because spinel phases (e.g., Ni2CoS4) are often obtained for the Co–Ni–S system when conventional synthetic processes are applied. However, using Ni–Co hydroxide nanoprisms as precursors for the hydrothermal reaction in weak acidic solvent (i.e., GO solution, pH = 4.7) can result in the formation of the CoxNi1−xS2 phase crystallized in a Pa[3 with combining macron] structure, which has much better OER performance than spinel phases. Thus, the unsaturated electronic configurations of Ni and Co atoms in the CoxNi1−xS2 phase originating from its crystalline structure (space group Pa[3 with combining macron]) as well as the high density of crystalline/amorphous phase boundaries which can parallelize the OER pathway are mainly responsible for the excellent water oxidation properties of ECA-CNS.

4. Conclusions

In summary, we have shown that high densities of crystalline/amorphous phase boundaries can significantly boost the oxygen evolution reaction (OER) activity of cobalt nickel sulfide (CoxNi1−xS2, CNS) crystallized in a cubic Pa[3 with combining macron] structure. Hydrothermal processing of Ni–Co hydroxide nanoprisms yielded CNS nanoparticles with a unique hollow structure. A subsequent electrochemical activation (ECA) process collapsed the hollow CNS nanoparticles onto reduced graphene oxide (rGO) sheets, resulting in few tens of nanometer-sized CNS nanodots separated by thin amorphous layers. The final ECA-CNS catalyst showed excellent OER performance in alkaline media, affording a current density of 10 mA cm−2 at an overpotential of 290 mV, a mass activity of 217 A g−1, a TOF of 0.21 s−1 at an overpotential of 340 mV, and a Tafel slope of 46 mV dec−1. High density crystalline/amorphous interfaces combined with unsaturated electronic configurations of Ni and Co atoms in ECA-CNS are mainly responsible for the excellent OER properties.

Author contributions

H. H., T. S., and S. M. conceived the project and designed the experiments. Y. R. H. and H. H. designed the experiments and prepared the samples. C. W. S., K. M. K, S. M, and H. C. participated in interpreting and analyzing the data. G. A and K. Y. C designed and analyzed the X-ray absorption spectroscopy experiments. All authors commented on the manuscript. H. H. wrote the manuscript.

Conflicts of interest

There are no conflicts to declare.


This work was supported by "Human Resources Program in Energy Technology (No. 20174030201460)" of the Korea Institute of Energy Technology Evaluation and Planning (KETEP), a granted financial resource from the Ministry of Trade, Industry & Energy, Republic of Korea.


  1. Y. P. Liu, G. T. Yu, G. D. Li, Y. H. Sun, T. Asefa, W. Chen and X. X. Zou, Angew. Chem., Int. Ed., 2015, 127, 10902–10907 CrossRef.
  2. R. Subbaraman, D. Tripkovic, K. C. Chang, D. Strmcnik, A. P. Paulikas, P. Hirunsit, M. Chan, J. Greeley, V. Stamenkovic and N. M. Markovic, Nat. Mater., 2012, 11, 550–557 CrossRef CAS PubMed.
  3. F. Barbir, Sol. Energy, 2005, 78, 661–669 CrossRef CAS.
  4. K. Zeng and D. K. Zhang, Prog. Energy Combust. Sci., 2010, 36, 307–326 CrossRef CAS.
  5. T. Reier, M. Oezaslan and P. Strasser, ACS Catal., 2012, 2, 1765–1772 CrossRef CAS.
  6. C. C. L. McCrory, S. H. Jung, J. C. Peters and T. F. Jaramillo, J. Am. Chem. Soc., 2013, 135, 16977–16987 CrossRef CAS PubMed.
  7. Y. Jiao, Y. Zheng, M. T. Jaroniec and S. Z. Qiao, Chem. Soc. Rev., 2015, 44, 2060–2086 RSC.
  8. Z. Peng, D. S. Jia, A. M. Al-Enizi, A. A. Elzatahry and G. F. Zheng, Adv. Energy Mater., 2015, 5, 1402031 CrossRef.
  9. A. Q. Zhao, J. Masa, W. Xia, A. Maljusch, M. G. Willinger, G. Clavel, K. P. Xie, R. Schlogl, W. Schuhmann and M. Muhlert, J. Am. Chem. Soc., 2014, 136, 7551–7554 CrossRef CAS PubMed.
  10. J. Masa, W. Xia, I. Sinev, A. Q. Zhao, Z. Y. Sun, S. Grutzke, P. Weide, M. Muhler and W. Schuhmann, Angew. Chem., Int. Ed., 2014, 53, 8508–8512 CrossRef CAS PubMed.
  11. R. Frydendal, E. A. Paoli, I. Chorkendorff, J. Rossmeisl and I. E. L. Stephens, Adv. Energy Mater., 2015, 5, 1500991 CrossRef.
  12. S. Cobo, J. Heidkamp, P. A. Jacques, J. Fize, V. Fourmond, L. Guetaz, B. Jousselme, V. Ivanova, H. Dau, S. Palacin, M. Fontecave and V. Artero, Nat. Mater., 2012, 11, 802–807 CrossRef CAS PubMed.
  13. F. Y. Cheng, J. A. Shen, B. Peng, Y. D. Pan, Z. L. Tao and J. Chen, Nat. Chem., 2011, 3, 79–84 CrossRef CAS PubMed.
  14. J. W. Nai, H. J. Yin, T. T. You, L. R. Zheng, J. Zhang, P. X. Wang, Z. Jin, Y. Tian, J. Z. Liu, Z. Y. Tang and L. Guo, Adv. Energy Mater., 2015, 5, 1401880 CrossRef.
  15. A. Sivanantham and S. Shanmugam, Appl. Catal., B, 2017, 203, 485–493 CrossRef CAS.
  16. M. Roca-Ayats, E. Herreros, G. Garcia, M. A. Pena and M. V. Martinez-Huerta, Appl. Catal., B, 2016, 183, 53–60 CrossRef CAS.
  17. A. Pendashteh, J. Palma, M. Anderson and R. Marcilla, Appl. Catal., B, 2017, 201, 241–252 CrossRef CAS.
  18. M. Sun, H. J. Liu, J. H. Qu and J. H. Li, Adv. Energy Mater., 2016, 6, 1600087 CrossRef.
  19. E. J. Popczun, J. R. McKone, C. G. Read, A. J. Biacchi, A. M. Wiltrout, N. S. Lewis and R. E. Schaak, J. Am. Chem. Soc., 2013, 135, 9267–9270 CrossRef CAS PubMed.
  20. E. J. Popczun, C. G. Read, C. W. Roske, N. S. Lewis and R. E. Schaak, Angew. Chem., Int. Ed., 2014, 53, 5427–5430 CrossRef CAS PubMed.
  21. X. Long, G. X. Li, Z. L. Wang, H. Y. Zhu, T. Zhang, S. Xiao, W. Y. Guo and S. H. Yang, J. Am. Chem. Soc., 2015, 137, 11900–11903 CrossRef CAS PubMed.
  22. W. X. Zhu, X. Y. Yue, W. T. Zhang, S. X. Yu, Y. H. Zhang, J. Wang and J. L. Wang, Chem. Commun., 2016, 52, 1486–1489 RSC.
  23. H. Q. Zhou, F. Yu, Y. F. Huang, J. Y. Sun, Z. Zhu, R. J. Nielsen, R. He, J. M. Bao, W. A. Goddard, S. Chen and Z. F. Ren, Nat. Commun., 2016, 7, 12765 CrossRef CAS PubMed.
  24. H. Vrubel and X. L. Hu, Angew. Chem., Int. Ed., 2012, 51, 12703–12706 CrossRef CAS PubMed.
  25. J. Masa, P. Weide, D. Peeters, I. Sinev, W. Xia, Z. Y. Sun, C. Somsen, M. Muhler and W. Schuhmann, Adv. Energy Mater., 2016, 6, 1502313 CrossRef.
  26. Y. B. Luan, L. Q. Jing, Y. Xie, X. J. Sun, Y. J. Feng and H. G. Fu, ACS Catal., 2013, 3, 1378–1385 CrossRef CAS.
  27. N. Roy, Y. Sohn and D. Pradhan, ACS Nano, 2013, 7, 2532–2540 CrossRef CAS PubMed.
  28. P. T. Wang, K. Z. Jiang, G. M. Wang, J. L. Yao and X. Q. Huang, Angew. Chem., Int. Ed., 2016, 55, 12859–12863 CrossRef CAS PubMed.
  29. S. Chatman, P. Zarzycki and K. M. Rosso, ACS Appl. Mater. Interfaces, 2015, 7, 1550–1559 CrossRef CAS PubMed.
  30. C. Wan, Y. N. Regmi and B. M. Leonard, Angew. Chem., Int. Ed., 2014, 53, 6407–6410 CrossRef CAS PubMed.
  31. L. L. Feng, G. T. Yu, Y. Y. Wu, G. D. Li, H. Li, Y. H. Sun, T. Asefa, W. Chen and X. X. Zou, J. Am. Chem. Soc., 2015, 137, 14023–14026 CrossRef CAS PubMed.
  32. J. Kibsgaard, Z. B. Chen, B. N. Reinecke and T. F. Jaramillo, Nat. Mater., 2012, 11, 963–969 CrossRef CAS PubMed.
  33. J. Nsanzimana, Y. Peng, Y. Y. Xu, L. Thia, C. Wang, B. Y. Xia and X. Wang, Adv. Energy Mater., 2018, 8, 1701475 CrossRef.
  34. J. Nsanzimana, V. Reddu, Y. Peng, Z. Huang, C. Wang and X. Wang, Chem.–Eur. J., 2018, 24, 1–11 CrossRef PubMed.
  35. P. He, X. Y. Yu and X. W. Lou, Angew. Chem., Int. Ed., 2017, 56, 3897–3900 CrossRef CAS PubMed.
  36. X. Zou, Y. P. Liu, G. D. Li, Y. Y. Wu, D. P. Liu, W. Li, H. W. Li, D. J. Wang, Y. Zhang and X. X. Zou, Adv. Mater., 2017, 29, 1700404 CrossRef PubMed.
  37. J. Masa, I. Sinev, H. Mistry, E. Ventosa, M. de la Mata, J. Arbiol, M. Muhler, B. R. Cuenya and W. Schuhmann, Adv. Energy Mater., 2017, 7, 1700381 CrossRef.
  38. Y. P. Liu, Q. J. Li, R. Si, G. D. Li, W. Li, D. P. Liu, D. J. Wang, L. Sun, Y. Zhang and X. X. Zou, Adv. Mater., 2017, 29, 1606200 CrossRef PubMed.
  39. L. Yu, L. Zhang, H. B. Wu and X. W. Lou, Angew. Chem., Int. Ed., 2014, 53, 3711–3714 CrossRef CAS PubMed.
  40. G. J. Xiao, Y. Zeng, Y. Y. Jiang, J. J. Ning, W. T. Zheng, B. B. Liu, X. D. Chen, G. T. Zou and B. Zou, Small, 2013, 9, 793–799 CrossRef CAS PubMed.
  41. H. L. Cao, X. F. Qian, C. Wang, X. D. Ma, J. Yin and Z. K. Zhu, J. Am. Chem. Soc., 2005, 127, 16024–16025 CrossRef CAS PubMed.
  42. X. Z. Zhou, X. Huang, X. Y. Qi, S. X. Wu, C. Xue, F. Y. C. Boey, Q. Y. Yan, P. Chen and H. Zhang, J. Phys. Chem. C, 2009, 113, 10842–10846 CrossRef CAS.
  43. D. C. Wei, Y. Q. Liu, Y. Wang, H. L. Zhang, L. P. Huang and G. Yu, Nano Lett., 2009, 9, 1752–1758 CrossRef CAS PubMed.
  44. Y. Li, Y. Zhao, H. H. Cheng, Y. Hu, G. Q. Shi, L. M. Dai and L. T. Qu, J. Am. Chem. Soc., 2012, 134, 15–18 CrossRef CAS PubMed.
  45. X. Wang, X. Q. Cao, L. Bourgeois, H. Guan, S. M. Chen, Y. T. Zhong, D. M. Tang, H. Q. Li, T. Y. Zhai, L. Li, Y. Bando and D. Golberg, Adv. Funct. Mater., 2012, 22, 2682–2690 CrossRef CAS.
  46. X. Wang, J. Wang, D. L. Wang, S. O. Dou, Z. L. Ma, J. H. Wu, L. Tao, A. L. Shen, C. B. Ouyang, Q. H. Liu and S. Y. Wang, Chem. Commun., 2014, 50, 4839–4842 RSC.
  47. Z. Yang, Z. Yao, G. F. Li, G. Y. Fang, H. G. Nie, Z. Liu, X. M. Zhou, X. Chen and S. M. Huang, ACS Nano, 2012, 6, 205–211 CrossRef CAS PubMed.
  48. Y. Ito, W. T. Cong, T. Fujita, Z. Tang and M. W. Chen, Angew. Chem., Int. Ed., 2015, 54, 2131–2136 CrossRef CAS PubMed.
  49. J. W. Nai, Y. Lu, L. Yu, X. Wang and X. W. Lou, Adv. Mater., 2017, 29, 1703870 CrossRef PubMed.
  50. J. H. Kim, D. H. Youn, K. Kawashima, J. Lin, H. Lim and C. B. Mullins, Appl. Catal., B, 2018, 225, 1–7 CrossRef CAS.
  51. J. F. Chang, Q. Lv, G. Q. Li, J. J. Ge, C. P. Liu and W. Xing, Appl. Catal., B, 2017, 204, 486–496 CrossRef CAS.
  52. T. R. Zhan, X. L. Liu, S. S. Lu and W. G. Hou, Appl. Catal., B, 2017, 205, 551–558 CrossRef CAS.
  53. F. Song and X. L. Hu, J. Am. Chem. Soc., 2014, 136, 16481–1648446 CrossRef CAS PubMed; H. C. Zhang, Y. J. Li, G. X. Zhang, T. H. Xu, P. B. Wan and X. M. Sun, J. Mater. Chem. A, 2015, 3, 6306–6310 RSC.
  54. H. C. Zhang, Y. J. Li, G. X. Zhang, T. H. Xu, P. B. Wan and X. M. Sun, J. Mater. Chem. A, 2015, 3(12), 6306–6310 RSC.
  55. M. Pavlov, P. E. M. Siegbahn, M. R. A. Blomberg and R. H. Crabtree, J. Am. Chem. Soc., 1998, 120, 548–555 CrossRef CAS.
  56. H. Han, K. M. Kim, H. Choi, G. Ali, K. Y. Chung, Y. R. Hong, J. Choi, J. Kwon, S. W. Lee, J. W. Lee, J. H. Ryu, T. Song and S. Mhin, ACS Catal., 2018, 8(5), 4091–4102 CrossRef CAS.


Electronic supplementary information (ESI) available. See DOI: 10.1039/c8ta10142f
These authors contributed equally to this work.

This journal is © The Royal Society of Chemistry 2019