Transition metal ions regulated oxygen evolution reaction performance of Ni-based hydroxides hierarchical nanoarrays

Nickel-based hydroxide hierarchical nanoarrays (NiyM(OH)x HNAs M = Fe or Zn) are doped with non-noble transition metals to create nanostructures and regulate their activities for the oxygen evolution reaction. Catalytic performance in these materials depends on their chemical composition and the presence of nanostructures. These novel hierarchical nanostructures contain small secondary nanosheets that are grown on the primary nanowire arrays, providing a higher surface area and more efficient mass transport for electrochemical reactions. The activities of the NiyM(OH)x HNAs for the oxygen evolution reaction (OER) followed the order of Ni2.2Fe(OH)x > Ni(OH)2 > Ni2.1Zn(OH)x, and these trends are supported by density functional theory (DFT) calculations. The Fe-doped nickel hydroxide hierarchical nanoarrays (Ni2.2Fe(OH)x HNAs), which had an appropriate elemental composition and hierarchical nanostructures, achieve the lowest onset overpotential of 234 mV and the smallest Tafel slope of 64.3 mV dec−1. The specific activity, which is normalized to the Brunauer–Emmett–Teller (BET) surface area of the catalyst, of the Ni2.2Fe(OH)x HNAs is 1.15 mA cm−2BET at an overpotential of 350 mV. This is ~4-times higher than that of Ni(OH)2. These values are also superior to those of a commercial IrOx electrocatalyst.

Scientific RepoRts | 7:46154 | DOI: 10.1038/srep46154 particularly those containing nanoparticles. Previous studies have shown that the OER activity of an ultra-thin γ -CoOOH nanosheet is 20-times higher than that of bulk CoOOH and 2.4-times higher than that of an IrO 2 electrocatalyst 26 . Well-aligned nanowire arrays have been used as highly effective electrodes, because of their intrinsic advantages [27][28][29] . Moreover, ordinary electrodes have relatively poor stabilities because of contact between the substrate and electrocatalysts during the electrocatalytic of OER, especially at large current densities. Therefore, hierarchically architectures can be constructed on conductive metal substrates to form high-performance nanocatalysts electrode 30,31 . However, no systematic studies have been performed that combine theoretical and experimental characterizations of the relationship between the doping of transition metals and the OER activities of Ni-based hydroxide nanoarrays.
In this work, the nanostructured morphology and OER activity of Ni-based hydroxide hierarchical nanoarrays (Ni y M(OH) x HNAs, M = Fe or Zn) were modified using two non-noble transition metals (Fe and Zn) as dopants. A systematic experimental and theoretical study of the effect of transition-metal doping on the nanostructure and OER activity of nickel-based catalysts is presented in this work. The intrinsic OER activity trends of the Ni y M(OH) x HNAs followed Ni 2.2 Fe(OH) x > Ni(OH) 2 > Ni 2.1 Zn(OH) x . Theoretical and experimental results were in good agreement. The trends were explained in terms of the surface areas and compositions of active sites, providing potential insights for the future design of more efficient water-splitting catalysts.

Results and Discussion
Ni y M(OH) x HNAs, where M = Fe and Zn, were fabricated by dipping Cu foam substrates coated with one-dimensional (1D) Cu 2 O nanowire arrays into an aqueous solution containing metal chloride salts and sodium hyposulfite using a solution-phase cation exchange method at room temperature. During the cation exchange process, the Cu 2 O nanowires were etched by S 2 O 3 2− , releasing OH − . During this process Ni y M(OH) x HNAs precipitated, these new Ni y M(OH) x HNAs structures inherited the geometry of the Cu 2 O template. Secondary Ni y M(OH) x HNAs nanostructures also formed depending on the solubility of the products and the pH of the reaction system. As illustrated in Fig. 1, the secondary nanostructures of the Ni y M(OH) x HNAs were regulated during this process. Low magnification SEM images of the Ni y M(OH) x HNAs revealed that the surface of the Cu foam substrate was completely covered with vertically aligned nanoarrays (Fig. 2). The inset to Fig. 2a showed the morphology of the Ni(OH) 2 HNAs, which inherited the shape of the 1D Cu 2 O nanowire arrays (Fig. S1) along the axial direction. After doping Ni(OH) 2 with transition metals, the surfaces of the nanowires became rougher, and their morphologies markedly changed into hierarchical structures with secondary nanosheets grown on the primary nanowire arrays (see insets to Fig. 2b,c). The degree of surface roughness on the Ni y M(OH) x HNAs followed the order of Ni(OH) 2 < Ni 2.1 Zn(OH) x < Ni 2.2 Fe(OH) x , indicating a marked increasement in surface area when the appropriate elements were used as dopants. Transmission electron microscopy (TEM) images of the Ni y M(OH) x HNAs further revealed the presence of secondary nanosheets (Figs 3 2 HNAs using Fe and Zn as dopants. When doping with the transition metals Fe and Zn, the surfaces of the HNAs become rougher and more highly amorphous. Figure S5 shows X-ray diffraction (XRD) patterns for the Cu 2 O nanowire arrays and the Co y   observed by XRD did not conflict with the crystal structures obtained from HRTEM, because the faint crystal lattice and weak FFT pattern of the Ni(OH) 2 HNAs indicated a low crystallinity 32,33 .
The Ni/M atomic ratios of the Ni y M(OH) x HNAs were determined with inductively coupled plasma (ICP) emission spectrometry. The ratios in the HNAs were similar to the reactant ratios (Table S1), indicating that the Ni/M ratios in the hydroxides were similar to those in the precursors. The surface compositions and valence states of the as-prepared Ni y M(OH) x HNAs were investigated by X-ray photoelectron spectroscopy (XPS), and the results are shown in Fig 34 . Compared to the un-doped Ni(OH) 2 HNAs, the Ni 2p peaks of the M-doped samples were shifted to more positive energies (Ni 2.1 Zn(OH) x < Ni 2.2 Fe(OH) x ), suggesting that the oxidation of Ni 2+ was favored when Fe and Zn were added. This effect was strongest with Fe 35 . Additional evidence for the presence of Ni 2+ was observed in the two intense shakeup satellite peaks (861.8 eV and 880.0 eV) 16 . The Zn 2p XPS spectrum for the Ni 2.1 Zn(OH) x HNAs contained 2p 3/2 and 2p 1/2 doublets, which are characteristic of Zn 2+ (1022.7 eV and 1045.7 eV) 34 . Fe 2p 3/2 and Fe 2p 1/2 spin-orbital splitting for the Ni 2.2 Fe(OH) x HNAs was deconvolved into four peaks, indicating the coexistence of Fe 2+ (711.5 eV and 723.7 eV) and Fe 3+ (716.0 eV and 726.3 eV) in the Ni 2.2 Fe(OH) x HNAs 36,37 . The O 1s spectrum of the Ni(OH) 2 HNAs was fit to a peak at a binding energy of 530.9 eV, which was assigned to the oxygen in hydroxide. The O 1s spectra of the Ni 2.2 Fe(OH) x and Ni 2.1 Zn(OH) x HNAs were fit with two peaks at binding energies of 530.1 eV and 531 eV, revealing the presence of lattice and hydroxide oxygens, respectively 16 . These results confirm that strong electron interactions occurred between Ni and both Fe and Zn in the Ni y M(OH) x HNAs.
The effect of doping the Ni y M(OH) x HNAs on their electrochemical behaviors were investigated using cyclic voltammetry (CV) in 1 M KOH. As shown in Fig. 5a 2 and NiOOH, and the positive shifts in redox potential were caused by adding the dopants (Fe or Zn) to the electrodes 19,38 .
The effect of doping the Ni y M(OH) x HNAs on their OER catalytic activities were tested with linear sweep voltammetry (LSV) in 1 M KOH (Fig. 5b). The Cu 2 O nanoarrays exhibited a negligible catalytic activity, while the OER current of the Ni 2.2 Fe(OH) x HNAs was much higher than those of the other electrodes. The Ni 2.2 Fe(OH) x HNAs had a low OER onset overpotential (η ) of 234 mV, which was more negative than the η of the Ni(OH) 2 HNAs (254 mV) and commercial IrO x electrocatalyst (248 mV). The high catalytic activity of the Ni 2.2 Fe(OH) x HNAs was also indicated by its ability to support a given current density (j) at a lower η than the other electrodes. At j = 100 mA cm −2 , the as-prepared Ni 2.2 Fe(OH) x HNAs required an η of 298 mV, which was 125 mV and 177 mV less than the η values of the Ni(OH) 2 HNAs (423 mV) and IrO x (375 mV), respectively. Therefore, intrinsic activities were compared at η = 350 mV. The intrinsic activity of the Ni 2.2 Fe(OH) x HNAs was 16-and 5-times higher than those of the Ni(OH) 2 HNAs and IrO x , respectively, revealing strong interactions between Ni and Fe during OER catalysis. Meanwhile, the Ni 2.1 Zn(OH) x HNAs exhibits a more positive onset η of 276 mV, and required a high η of 410 mV to achieve a current density of 100 mA cm −2 . The intrinsic activity of the Ni 2.1 Zn(OH) x HNAs (25.8 mA cm −2 ) was similar to that of the Ni(OH) 2 HNAs. These results suggest that unfavorable interactions occurred between Ni and Zn. The excellent OER activities of the catalysts were attributed to their increased surface areas and specific activities (active sites per unit area).
The Brunauer-Emmett-Teller (BET) surface area measurements were performed to confirm the mesoporous nature of the Ni y M(OH) x HNAs. Nitrogen adsorption-desorption curves revealed a Type IV isotherm (Fig. S8). Additionally, a H3-type hysteresis loop was observed, providing further evidence of nanosheet aggregation 39 . As shown in Fig. 5c, the Ni(OH) 2 HNAs (73.2 cm 2 g −1 ) had a smaller BET surface area than the Ni 2.1 Zn(OH) x (105.6 cm 2 g −1 ) and Ni 2.2 Fe(OH) x (155.6 cm 2 g −1 ) HNAs. These results confirmed observations of increased surface areas in the SEM (Fig. 2) and TEM (Figs 3 and S2) images. Specific activity (current per BET area) is a measure of the density of active sites on the surface of a catalyst. Figure 5d shows LSV curves after normalizing the measured currents to the catalysts' BET surface areas. The specific activity of the Ni 2.2 Fe(OH) x HNAs was 1.15 mA cm −2 BET at η = 350 mV, which was 4-and 8-times higher than those of the Ni(OH) 2 HNAs (0.22 mA cm −2 BET ) and Ni 2.1 Zn(OH) x HNAs (0.13 mA cm −2 BET ), respectively. These results indicated that adding Fe indeed resulted in more active sites, while doping with Zn decreased the number of active sites. Kinetic analyses were performed using LSV to generate Tafel plots and electrochemical impedance spectra (EIS). As shown in Fig. 5e, the resulting Tafel slope of the Ni 2.2 Fe(OH) x HNAs was 64.3 mV dec −1 , which was much lower than that of the Ni(OH) 2 HNAs (123.4 mV dec −1 ), Ni 2.1 Zn(OH) x HNAs (107.2 mV dec −1 ), and IrO x (113.3 mV dec −1 ). Tafel slopes were used to probe the OER mechanisms of the catalysts. Efficient electron and mass transport result in lower Tafel slopes. The EIS was performed in oxygen-saturated 1.0 M KOH (Fig. S6). ZSimpWin 3.5 (Zolartron Analytical) was used to fit the resistance values, as shown in Table S2. As shown in the inset to Fig. S6, all of the Ni y M(OH) x HNAs electrodes were fitted using the same equivalent circuit, which contained three components: solution resistance (R s ), charge-transfer resistance (R ct ), and constant-phase resistance (R cp ). The Ni 2.2 Fe(OH) x HNAs had an R ct of 1.7 Ω, which was much lower than that of the Ni 2.1 Zn(OH) x (21.4 Ω) and Ni(OH) 2 (23.8 Ω) HNAs. These results indicated that OER kinetics were enhanced for the Ni 2.2 Fe(OH) x HNAs electrode. These EIS measurements were consistent with the findings from LSV.
Turnover frequency (TOF) is an intrinsic property of a catalyst and an important indicator of catalyst performance. The TOF of the Ni 2.2 Fe(OH) x HNAs was much higher than that of Ni 2.1 Zn(OH) x and Ni(OH) 2 HNAs (Fig. S7). Moreover, at η = 350 mV, the TOF of the Ni 2.2 Fe(OH) x HNAs was at least 15-, 21-, and 3-times as those of the Ni(OH) 2 HNAs (0.011 s -1 ), the Ni 2.1 Zn(OH) x HNAs (0.008 s -1 ), and IrO x (0.05 s -1 ), respectively. These TOF values further verified the superior catalytic performance of the Ni 2.2 Fe(OH) x HNAs for the OER. This improved performance resulted from strong interactions between Ni and Fe and the presence of more exposed catalytically active sites.
Long-term stability is also important for catalysts that are to be used for practical applications. As shown in Fig. 5f Table 1, and especially the overall performance of Ni 2.2 Fe(OH) x surpasses most reported typical Co-based electrocatalysts for water oxidation under alkaline solution (Table S3).
To better understand the catalytic activities that resulted from doping with Fe and Zn, the binding energies of oxygen on the Ni y M(OH) x catalysts were investigated using DFT calculations. The binding energy of oxygen serves as a reliable measure of the activity of a catalyst toward the OER. Smaller values of binding energy of oxygen(E o ) at a reaction site correspond to higher activities for the OER 41,42 . Hydroxide clusters were constructed based on a model of Ni 2 M(OH) 6 . The oxygen binding energies of the hydroxide clusters increased in the order of Ni 2 Zn(OH) 6 > Ni 3 (OH) 6 > Ni 2 Fe(OH) 6 (Fig. 6b). The reactivities of the amorphous hydroxides followed the order of Ni 2 Fe(OH) 6 > Ni 3 (OH) 6 > Ni 2 Zn(OH) 6 , which was in good agreement with the experimental results.

Conclusion
Nanostructures were generated to regulate the OER activities of the Ni y M(OH) x HNAs using various non-noble transition metals as dopants. According to both experimental and DFT-based theoretical analyses, Fe and Zn had opposite effects on the catalyst's activity. Fe was an effective dopant, while Zn decreased the OER activity of the catalyst. Hierarchical nanostructures allowed efficient charge transfer and a sufficient surface area for active sites. A 3D porous Cu foam not only provided a large surface area and stable anchoring sites for nanoarrays but also acted as an efficient electron collector. Because of the hierarchical nanostructures, the appropriate elemental composition of the catalyst, and the presence of a multifunctional 3D conductive substrate, the Ni 2.2 Fe(OH) x HNAs exhibited an enhanced OER activity. The Ni 2.2 Fe(OH) x HNAs had a low onset η of 234 mV and a small Tafel slope of 64.3 mV dec −1 . They also exhibited an excellent long-term stability for over 20 h in an alkaline electrolyte. The Ni 2.2 Fe(OH) x HNAs also had a superior activity compared to that of a commercial IrO x catalyst, and these Ni 2.2 Fe(OH) x HNAs were prepared using an extremely simple method. Their activities were comparable to those of other NiFe hydroxides obtained through more labor-intensive procedures. This study provides significant new guidelines for and a broader understanding of the use of dopants to improve catalytic activity.  Cu foam (100 pores per inch, 98% porosity, and ~1.5 mm thick) was cut into squares (2.0 cm × 2.0 cm), and cleaned in Mill-Q water and ethanol before use. The Cu foam was then anodized in 0.4 M H 2 C 2 O 4 for 20 min at 36 V, using a graphite plate cathode. Electro-oxidation was performed with a potentiostat (CHI760D, CH Instruments) in a three-electrode configuration consisting of an anodized Cu foam working electrode, a Pt gauze counter electrode, and an Ag/AgCl reference electrode 43 . Cyclic voltammetry was performed in the potential range from − 0.3 V to 0.1 V at a scan rate of 1 mV s −1 in 1 M KOH for the in situ growth of Cu 2 O nanowire arrays on the Cu foam 43,44 .

Methods
Ni y M(OH) x HNAs were fabricated using the Cu 2 O nanowire arrays as sacrificial templates 33  Structural characterization. Scanning electron microscopy (SEM) was performed with a ZEISS MERLIN scanning electron microscope. Microstructural investigations were performed with a JEOL JEM-2100 and Tecnai G2 Spirit TWIN. X-ray diffraction (XRD) patterns were recorded using a Rigaku Ultima IV. The valence states of elements were measured with X-ray photoelectron spectroscopy (XPS, PHI 5000 VersaProbe). All of the spectra were normalized to the C 1 s binding energy at 284.8 eV. Ni/M atomic ratios were measured with a VISTA-MPX ICP-OES. BET measurements were performed on a Quadrasorb SI analyzer at 77 K. Electrochemical Measurements. Electrochemical measurements were performed in O 2 -saturated 1 M KOH with an electrochemical analyzer (CHI760D, CH Instruments), using a three-electrode configuration with an Hg/HgO (1 M KOH) reference electrode that contained a double salt bridge and a platinum mesh counter electrode. The Ni y Fe 1-y (OH) x HNAs (0.5 cm × 0.5 cm) on Cu foams were used as working electrodes. All polarization measurements were performed at a scan rate of 5 mV s −1 . Potentials are reported in terms of the reversible hydrogen electrode (RHE), using: E (RHE) = E (Hg/HgO) + 0.098 V + 0.0591 V × pH. All CV measurements were compensated for iR drop by 75%. Stability was measured using the controlled potential electrolysis method. The EIS was performed with a Princeton PMC 1000 electrochemical workstation in the frequency range of 10 −2 Hz− 10 4 Hz at an amplitude of 5 mV. All electrochemical tests were performed at 25 °C.
Turnover frequency (TOF) was calculated as: TOF = (j × a)/(4 × n × F), where j is the current density at a given potential, a is the surface area of the electrode (0.25 cm 2 for the Cu foam electrode), 4 is the number of electrons transferred in the OER, n is the number of moles of all metal ions available for the OER (including Ni and M), and F is Faraday's constant (96485 C mol −1 ).

DFT Calculations.
Since the Ni y M(OH) x HNAs were predominantly amorphous (the Ni(OH) 2 HNAs had a hexagonal Ni(OH) 2 phase and the Ni 2 M(OH) x HNAs had an amorphous phase), cluster rather than slab model was chosen for the DFT simulation. Hydroxide clusters were first built based on the model of Ni 2 M(OH) 6 45 , as shown in Fig. 6a. All reported DFT calculations with the Hubbard U (DFT + U) calculations were performed at the Perdew-Burke-Ernzenhof/Generalized Gradient Approximation (PBE/GGA) 46 level using the spin-dependent formulation of the hybrid Gaussian and the plane waves method. The calculations were implemented with the open-source CP2K/QUICKSTEP 47,48 code. For a better description of the Ni and Fe 3d electrons, the Hubbard effective terms U eff (Ni) = 5.96 eV and U eff (Fe) = 5.3 eV were added to the PBE functional 49,50 . Electrons in the outer most shells of the atoms were treated as being in their valence states. The Kohn-Sham orbitals of the valence electrons were expanded in molecularly optimized Gaussian basis sets of double-ζ plus polarization quality (MOLOPT-SR-DZVP) 51 . Ionic cores were represented by norm-conserving Goedecker-Teter-Hutter 52-54 pseudopotentials. The auxiliary plane wave basis set was truncated with a 500 Ry kinetic energy cut off.