A nickel iron diselenide-derived efficient oxygen-evolution catalyst

1Sunlight-driven water splitting or carbon dioxide (CO2) reduction to make solar fuels is a promising solution to solar energy storage. Essential to the water splitting and CO2 reduction reactions is the oxygen-evolution reaction (OER). This reaction is kinetically sluggish and demands an efficient electrocatalyst. Although noble metal-based catalysts such as IrO2 and RuO2 exhibit good OER activity, their scarcity and high cost pose great constrains for large-scale applications. Tremendous efforts have been made in recent years to develop non-precious OER catalysts. Despite the progress, a significant overpotential is still required for state-of-the-art catalysts. To reach 10 mA cm−2, a widely used figure of merit equivalent to ∼12% solar to hydrogen efficiency, nearly all non-precious catalysts need an overpotential of more than 250 mV.

2The majority of non-precious OER catalysts are metal oxides and (oxy)hydroxides. Recently, a few non-oxide-based OER catalysts including metal phosphides, sulfides and selenides are reported. Given the limited stability of these compounds under highly oxidative potentials in alkaline solutions, questions have arisen on the nature of the true active species. Indeed, we and others showed that the surfaces of Ni2P and CoP were transformed into metal oxides during catalysis, which were responsible for the catalytic activity. For Ni, Co and Fe sulfides, Chen et al. showed that they were entirely transformed into the corresponding metal oxides during OER. Since metal selenides have similar chemical reactivity to metal sulfides, we were surprised by previous reports which suggested stability of bulk NiSe, Ni3Se2 and CoSe2 materials under OER. Using a post-catalytic analysis, we show here that NiSe is completely converted into nickel hydroxides during OER, indicating that metal oxides or hydroxides are the active and final forms of metal selenides pre-catalysts in OER. This knowledge promoted us to purposely use metal selenides as templating precursors to highly active metal oxide OER catalysts, because methods to produce ultrasmall nanostructured metal selenides are readily available. Following this strategy, we synthesize a hitherto unknown selenide, nickel iron diselenide (NixFe1−xSe2), which upon in situ transformation into oxides, catalyses OER with an overpotential of only 195 mV for a current density of 10 mA cm−2. This is until now the most active single-phase OER catalyst in alkaline solutions. The high activity of this NixFe1−xSe2-derived catalyst is largely due to its desirable nanostructure, inherited from its selenide precursor.

3NiSe was synthesized via a hydrothermal approach using Ni foam as the precursor. The electrocatalytic activity of NiSe towards OER was investigated in 1 M KOH using a three-electrode electrochemical system. Galvanostatic scan at the current density of 10 mA cm−2 was used to activate the catalyst. The activity was then measured by linear sweep voltammetry (LSV) at a scan rate of 1 mV s−1. The overpotentials to reach 10 mA cm−2 was ∼253 mV (Supplementary Fig. 1), in agreement with previous reports. The morphology and composition of the catalyst after catalysis for 12 h were then examined. Transmission electron microscopy (TEM) image showed that single-crystal nanowires of NiSe were converted to polycrystalline particles made of ultrathin nanosheets (Fig. 1a,b). Selected area electron diffraction (SAED) pattern of the sample after OER can be indexed to (111), (103) and (301) planes of α-Ni(OH)2 (Fig. 1b) (space group: P-31m, JCPDS No. 22-0444). The different lattice fringes in high-resolution TEM (HRTEM) images of samples before and after OER confirmed the total conversion of NiSe into Ni(OH)2. Similarly, the elemental mapping of the sample after OER showed that the Se content decreased from 50.2% to 4.0%, but the oxygen content increased from 2.4 to 52.8% during the transformation (Fig. 1c,d). Thus, Se was nearly completely removed while oxygen was incorporated during OER. The above data indicate that NiSe is entirely converted into Ni(OH)2 under OER conditions, which is the active form of the catalyst. Interestingly, the activity of the NiSe-derived Ni(OH)2 is higher than the most active Ni(OH)2 nanoparticles prepared by direct synthesis, which requires 300 mV to reach 10 mA cm−2 (refs, ). This result suggests that metal selenides may serve as templating precursors to metal oxides or hydroxides with superior OER activity than those prepared by other methods. With this in mind, we turn our attention to nickel iron selenides, as NiFeOx is one of the most active OER catalysts.

4Metal selenides normally exist in two forms, mono-selenide (MSe) and diselenides (MSe2). We attempted to synthesize both NixFe1−xSe and NixFe1−xSe2. However, only diselenide NixFe1−xSe2 could be obtained. NiSe2 and FeSe2 are well-known substances, while NixFe1−xSe2 is a hitherto unknown selenide, probably due to a mismatch between the crystal structures of NiSe2 and FeSe2. Both have a cubic structure (space group: Pa-3), but NiSe2 has a cell parameter of a=5.960 Å, while FeSe2 has a parameter of a=5.776 Å. To circumvent this mismatch, a reductive solvothermal process was applied using NiFe-layered double hydroxide (LDH) grown on Ni foam as the precursor. We hypothesized that Ni and Fe were already well dispersed in the crystal lattices of NiFe LDH, so that its selenization might lead to pure-phase NixFe1−xSe2 without phase separation between NiSe2 and FeSe2. This approach indeed worked.

5Powder X-ray diffraction (PXRD) pattern (Fig. 2a) of the as-synthesized NixFe1−xSe2 (x=0.8) indicated the formation of a cubic pyrite-phase metal selenide, similar to NiSe2 (space group: Pa-3, JCPDS No. 88–1711). The unit cell of cubic metal diselenides is shown in the inset in Fig. 2a. No crystalline impurity was detected. The diffraction peaks of NixFe1−xSe2 were shifted to higher angles compared with their counter parts in NiSe2 (upper right inset in Fig. 2a). The crystal parameter of NixFe1−xSe2 is a=5.884 Å, which is between NiSe2 (a=5.960 Å) and FeSe2 (5.776 Å). NixFe1−xSe2 samples with other Ni:Fe rations (x=0.9 and 0.67) was prepared by varying the Ni:Fe atomic ratio of the starting materials (see Methods for details). The formation of NixFe1−xSe2 was supported by X-ray photoelectron spectroscopy (XPS). In both Ni 2p3/2 and Fe 2p3/2 spectra (Fig. 2b and Supplementary Fig. 2), the main peaks have binding energies between those of the corresponding metal and metal oxides. For example, the main peak at 853.5 eV in the Ni 2p3/2 spectra has a binding energy between nickel metal (852.6 eV) and nickel oxides (853.7–854.9 eV). These binding energies are indicative of metal selenides. The Se 3d5/2 binding energy of NixFe1−xSe2 is 55.2 eV, red-shifted from 55.4 eV for elemental Se (Fig. 2c). The minor broad peaks in Fig. 2b,c were attributed to surface impurities NiOx, FeOx and SeO2. This assignment is consistent with the previous finding that the surface of metal selenides is prone to oxidation by air.

6The morphology of as-prepared NixFe1−xSe2 (x=0.8) was first characterized by scanning electron microscopy (SEM). At the microscale, NixFe1−xSe2 inherited the morphology of NiFe LDH (Fig. 3a,b and Supplementary Fig. 3), and after selenization, nanoplates of ∼150 nm in thickness were formed. Higher-magnification images, however, revealed that in contrast to NiFe LDH nanoplates which had a smooth surface (Supplementary Fig. 3b), NixFe1−xSe2 nanoplates were composed of numerous nanoparticles (Fig. 3b). The TEM image further revealed the porous hierarchical morphology of NixFe1−xSe2 (Fig. 3c). The nanoparticles were made of ultrathin nanosheets (Fig. 3d). HRTEM image indicated the single-crystal nature of each nanosheet (Fig. 3e). Lattice fringes were observed, with inter-planar distances of ∼0.245 and ∼0.215 nm, corresponding to the (−112) and (220) planes, respectively. The basal plane of the nanosheet is (111), which was confirmed by fast-Fourier transform image (inset in Fig. 3e). Elemental mapping analysis (Supplementary Fig. 4) showed that Ni, Fe and Se were homogenously distributed, consistent with the formation of single-phase NixFe1−xSe2. The atomic ratio of Se/Ni/Fe is ∼7.3:2.6:1.

7As mentioned above, we expected that NixFe1−xSe2 would be converted to NiFeOx under OER conditions. This NixFe1−xSe2-derived oxide (NixFe1−xSe2-DO; we use ‘DO' to designate the selenide-derived oxide from here on) was in fact our target catalyst. Our hope was that NixFe1−xSe2-DO would inherit the potentially beneficial nanostructure of NixFe1−xSe2. Thus, NixFe1−xSe2-DO was obtained by subjecting NixFe1−xSe2 to galvanostatic scan at the current density of 5 mA cm−2 until a stable potential was reached. SEM images (Fig. 4a,b) showed that NixFe1−xSe2-DO had a similar overall morphology to NixFe1−xSe2. TEM images (Fig. 4c,d) showed the nanosheets of NixFe1−xSe2-DO were only about 1–2 nm in thickness. HRTEM image (Fig. 4e) showed lattice fringes of NixFe1−xSe2-DO, which were different from those of NixFe1−xSe2. The inter-planar distance of 0.261 nm NixFe1−xSe2-DO was indexed to the (101) planes of metal hydroxides (for example, LDH or α-Ni(OH)2).

8Elemental mapping was conducted to examine the change of composition after the in situ transformation. From NixFe1−xSe2 to NixFe1−xSe2-DO, Ni and Fe remained homogeneously distributed, Se was removed, while oxygen was incorporated (Fig. 5a). Based on the energy-dispersive X-ray spectra (EDS), the Se content decreased from 67.1% in NixFe1−xSe2 to 0.7% in NixFe1−xSe2-DO, and the oxygen content increased from undetectable in NixFe1−xSe2 to 52.3% in NixFe1−xSe2-DO (Fig. 5b). The compositional change was further confirmed by high-resolution XPS spectra (Fig. 5c–e). The Se 3d peak nearly vanished while the O 1s peak increased dramatically in intensity. The main peak in the O 1s spectrum was at 530.6 eV, attributed to Ni-OH or Fe-OH (ref. ). On the basis of the XPS spectra, the atomic percentages of O and Se were ∼47.7% and 1.29% in NixFe1−xSe2-DO, in agreement with the EDS results. The Ni 2p3/2 peak was shifted from 853.5 eV in NixFe1−xSe2 to 855.1 eV in NixFe1−xSe2-DO, also consistent with the formation of NiFeOx. Finally, the PXRD pattern of NixFe1−xSe2-DO (Fig. 5f) also indicated the disappearance of NixFe1−xSe2 and the formation of a crystalline NiFeOx phase. Based on previous studies of the electrochemical oxidation of NiSe and MoSe2 in basic solutions, the in situ transformation is proposed to occur via the following pathways:

9The electrocatalytic activity of NixFe1−xSe2-DO, NiFe LDH, NiSe2-DO, NiSe-DO and Ni foam (NF) towards OER in 1 M KOH oxidation was measured and compared (see Methods for details). As shown in Fig. 6a, NixFe1−xSe2-DO is the best catalyst among the five compounds, giving much higher current density (J) at the same overpotential (η). To reach J=10 mA cm−2, NixFe1−xSe2-DO required an overpotential of only 195 mV, which was 49, 46, 58 and 96 mV less than that of NiFe LDH, NiSe2-DO, NiSe-DO and NF, respectively (Fig. 6b). The current density at η=250 mV was 262 mA cm−2, which was 16-, 18-, 29- and 262-fold higher than those of NiFe LDH, NiSe2-DO, NiSe-DO and NF, respectively (Fig. 6c). In fact, the activity of NixFe1−xSe2-DO is superior to other state-of-the-art catalysts (Supplementary Table 1). NiFe LDH has been regarded as the most active OER catalyst in alkaline conditions. Lu et al. reported the overpotential of ∼256 mV for 10 mA cm−2 by NiFe LDH grown on Ni foam. The catalytic activity of NiFe LDH was further improved by applying an improved electrosynthesis method. However, an overpotential of 224 mV for 10 mA cm−2 was still required. The activity of NixFe1−xSe2-DO compares favourably even to the best hybrid catalysts made of NiFeOx and carbon nanomateirals. Ma et al. reported that the NiFe hydroxide/graphene superlattice reached 10 mA cm−2 at an overpotential of 210 mV. Hou et al. reported that exfoliated graphene/Co0.85Se/NiFe LDH composites catalysed OER with an overpotential of 203 mV for 10 mA cm−2. The most active hybrid catalyst, r-GO/NiFe LDH, was reported by by Long et al., which gave 10 mA cm−2 at an overpotential of 195 mV. It was proposed that the activity of these hybrid materials originated from a synergetic effect between NiFeOx and carbon nanomaterials (CNT, r-GO and carbon quantum dots). Thus, our NixFe1−xSe2-DO is until now the most active single-phase catalyst, and even higher activity might be achieved by coupling it to carbon nanomaterials. The influence of Ni:Fe atomic ratio on the OER activity was probed. The highest activity was obtained at x=0.8, while the samples with x=0.9 and 0.67 exhibited only modestly lower activity (Supplementary Fig. 5). This result is consistent with previous reports that that the activity of NiFeOx varied only slightly when the Fe:Ni ratio was changed from 0.1 and 0.55 (refs, ).

10By plotting overpotential against log (J), the kinetic parameters of OER by the five catalysts were calculated (Fig. 6d). The NixFe1−xSe2-DO has a Tafel slope of about 28 mV dec−1, close to that of NiFe LDH (32 mV dec−1), but much smaller than the other catalysts (40–54 mV dec−1). We further compared the electrochemical surface area (ECSA), estimated from their double-layer capacitance (Cdl) of NixFe1−xSe2-DO and NiFe LDH. The ECSA of NixFe1−xSe2-DO is 2.4-fold of the ECSA of NiFe LDH (Fig. 6e) that does not account for the 16-fold higher catalytic activity of the former. We attribute the superior activity of NixFe1−xSe2-DO to its desirable nanostructure. As shown in Fig. 4, NixFe1−xSe2-DO nanoparticles are made of ultrathin nanosheets. If the edges of NiFeOx are the active sites, then a higher number of active sites is expected on NixFe1−xSe2-DO than other forms of NiFeOx. To provide additional support for this hypothesis, the microstructures of NixFe1−xSe2-DO was compared with NiFe LDH, one of the most active forms of NiFeOx (refs, ). SEM images show that NiFe LDH is made of large single-crystalline nanoplates whereas NixFe1−xSe2-DO is made of highly porous nanoplates consisting of ultrathin nanosheets (Supplementary Fig. 6). Thus, NixFe1−xSe2-DO's particle size is much smaller than NiFe LDH. The porosity of NixFe1−xSe2-DO and NiFe LDH were probed by N2 adsorption–desorption measurements. Compared with NiFe LDH, more pores were formed in NixFe1−xSe2-DO (Supplementary Fig. 7). More importantly, an additional sharp peak was observed at a size range of 2.5–4 nm for NixFe1−xSe2-DO, indicating the formation of tiny nanopores. The nanoporosity resulted in a high Brunauer-Emmett-Teller (BET) surface area of 109 m2 g−1 for NixFe1−xSe2-DO, which is 8.3 times higher than that of NiFe LDH. The higher porosity and smaller particle size of NixFe1−xSe2-DO compared with NiFe LDH would lead to higher number of edge sites in the former, which might explain its superior catalytic activity. The molecular origin of this activity is subject to further studies.

11The stability of OER catalysed by the NixFe1−xSe2-DO electrode was tested at a constant current density J of 10 mA cm−2 for 24 h. Figure 6f shows that after an activation period of 0.5 h, the overpotential remained at ∼195 mV during 24 h. The Faradaic efficiency of OER was determined using a fluorescence O2 probe. A quantitative yield was found quantitative during a 6.5 h electrolysis experiment (inset in Fig. 6f).

12In summary, using NiSe as a representative example of metal selenides, we show that metal selenides are converted into metal oxides or hydroxides under OER conditions. These oxides or hydroxides are responsible for the catalytic activity of metal selenides in OER. Taking advantage of this in situ transformation, we prepared for the first time a nickel iron diselenide (NixFe1−xSe2) and used it as a templating precursor to ultrathin nanosheets of the corresponding oxide, NixFe1−xSe2-DO. A current density of 10 mA cm−2 is obtained at an overpotential of only 195 mV in alkaline solutions using NixFe1−xSe2-DO as the catalyst, making it one of the most active OER catalyst reported to date. The templating approach described here might be applicable for the synthesis of other metal oxide-based nanomaterials.

13Both NiSe2 and NixFe1−xSe2 were prepared through two steps: (i) metal hydroxide precursors were grown on nickel foam via a hydrothermal method; and (ii) hydroxide precursors were converted into diselenides via a solvothermal selenization treatment. ‘x' represents the Ni ratio in the total metal elements of the starting solution. Typically, for NixFe1−xSe2 (x=0.8), Ni(NO3)2·6H2O (2.0 mmol, 582 mg), FeSO4·7H2O (0.5 mmol, 139 mg), NH4F (10 mmol, 370 mg) and urea (25 mmol, 1,501 mg) were dissolved in H2O (40 ml). The as-obtained solution together with one piece of nickel foam (5 × 30 mm) were then sealed in a 45 ml Teflon-lined stainless steel autocalve and heated at 120 °C in an electric oven for 16 h. After being washed thoroughly with distilled water and absolute ethanol, the as-prepared nickel foam coated with NiFe hydroxides was submerged into another 45 ml autocalve containing Se (3.75 mmol, 296 mg), NaOH (7.5 mmol, 300 mg), hydrazine (0.14 ml) and dimethylformamide (DMF) (25 ml). After keeping at 180 °C for 1 h, NixFe1−xSe2 was obtained. To synthesize NixFe1−xSe2 (x=1, 0.9 and 0.67) with variable Ni:Fe ratios, Ni(NO3)2·6H2O and FeSO4·7H2O with the desired molar ratio were dissolved in water with a total quantity of 2.5 mmol. When x=1, NiSe2 was obtained.

14PXRD patterns were recorded on an X'Pert Philips diffractometer with monochromatic CuKα radiation (λ=1.540598 Å) and a fast Si-PIN multi-strip detector. As-synthesized NiSe2 and NixFe1−xSe2 were used directly for PXRD measurements. Morphology and microstructure was examined by a Phillips (FEI) XLF-30 FEG SEM and a FEI Tecnai Osiris TEM equipped with high-brightness field emission gun (XFEG). Energy-dispersive X-ray spectroscopy (EDX) mapping images were taken under a scanning TEM modal. Samples for TEM were prepared by drop-drying the samples from their diluted ethanol suspensions onto carbon-coated copper grids. XPS measurements were performed on a PHI5000 VersaProbe II XPS system by Physical Electronics (PHI) with a detection limit of one atomic per cent. Monochromatic X-rays were generated by an Al Kα source (14,867 eV). The diameter of the analysed area is 10 μm. N2 adsorption–desorption measurements were conducted on Micromeritics 3Flex adsorption analyser at 77K. Before the measurements, the samples were degassed at 100 °C under vacuum for 1 h. Pore size distributions were calculated by the Barrett-Joyner-Halenda (BJH) method from the desorption branches of the isotherms.

15Electrochemical characterizations including cyclic voltammetry (CV), LSV and chronopotentiometry were carried out on a Gamry Reference 3000 electrochemical instrument using a three-electrode electrochemical system. A 1 M KOH solution was used as electrolyte, and an Ag/AgCl electrode with saturated KCl filling solution and Pt wire were used as reference and counter electrodes, respectively. For catalyst grown on nickel foam (NiSe2, NixFe1−xSe2 and NiFe-LDH), they were used as work electrodes directly. Hot glue was employed to fix the working area at 0.2 cm−2 (0.5 × 0.4 cm). Before test, the reference electrode was measured against another unused Ag/AgCl reference electrode stored in saturated KCl solution. Calibration of Ag/AgCl reference electrodes was done by measuring the reversible hydrogen electrode (RHE) potential using a Pt electrode under a H2 atmosphere. During the test, Ag/AgCl reference electrode was constructed to a double-junction electrode to minimize contact between KOH and KCl. CVs were performed at a scan rate of 1 mV s−1, and the average of the two potentials at which the current crossed zero was taken to be the thermodynamic potential for the hydrogen electrode reaction, In 1 M KOH electrolytes, Evs.RHE=Evs.Ag/AgCl+1.009 V, and overpotential for OER was η=Evs.RHE−1.23 V=Evs.Ag/AgCl−0.221 V. Ohmic drop correction was performed using the current interrupt method by the potentiostat. Before recording the catalytic activity, catalysts were activated by a chronopotentiometry scan with constant current density of 5–10 mA cm−2 until reaching a stable state. Normally, we set the pretreatment time at 12 h. Following the preconditioning, two cycles of LSVs were measured at a scan rate of 1 mV s−1. Tafel slopes were calculated based on the LSV curves by plotting overpotential against log(current density). Chronopotentiometry measurements were performed to evaluate the long-term stability. The ECSA was determined by measuring the capacitive current associated with double-layer charging from the scan-rate dependence of CVs. For this, the potential window of CVs was 0.2–0.3 V versus Ag/AgCl. The scan rates were 5, 10, 25, 50, 100, 200, 400, 600, 800 and 1,000 mV s−1. The double-layer capacitance (Cdl) was estimated by plotting the ΔJ=(Ja−Jc) at 0.25 V versus Ag/AgCl against the scan rate. The linear slope is twice of the double-layer capacitance Cdl. The measurements of O2 were performed using an Ocean Optics Multifrequency Phase Fluorimeter (MFPF-100) with a FOXY-OR 125 probe (see in Supplementary Methods).

16The data that support the findings of this study are available from the corresponding authors upon request.

17How to cite this article: Xu, X. et al. A nickel iron diselenide derived efficient oxygen evolution catalyst. Nat. Commun. 7:12324 doi: 10.1038/ncomms12324 (2016).