Modulating Ni/Ce Ratio in Ni_yCe_100−yO_x Electrocatalysts for Enhanced Water Oxidation

Abstract

Oxygen evolution reaction (OER) is the key reaction for water splitting, which is used for hydrogen production. Oxygen vacancy engineering is an effective method to tune the OER performance, but the direct relationship between the concentration of oxygen vacancy and OER activity is not well understood. Herein, a series of Ni_yCe_100−yO_x with different concentration of oxygen vacancies were successfully synthesized. The larger concentration of oxygen vacancies in Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x result in their lower Tafel slopes, small mass-transfer resistance, and larger electrochemical surface areas of the catalysts, which account for the higher OER activities for these two catalysts. Moreover, with a fixed current density of 10 mA/cm², the potential remains stable at 1.57 V for more than 100 h, indicating the long-term stability of the Ni₇₅Ce₂₅O_x catalyst.

1. Introduction

Hydrogen is considered as one of the clean energy to replace the traditional fossil fuel to solve the environmental and energy problem. Electrocatalytic water splitting is an effective way to produce hydrogen with high purity from intermittent renewable energy (i.e., wind and solar energy). Two half reactions including hydrogen evolution reaction (HER) and oxygen evolution reaction (OER) are at cathode and anode, respectively. The four-electron transfer steps of OER lead to the large sluggish kinetics, resulting in a high overpotential for water splitting [1,2,3,4,5,6,7]. An OER catalyst with high activity is of critical importance to reduce the overpotential so that a small bias voltage can be used and thus increase the efficiency of the OER process.

The OER activity of a catalyst can be enhanced by introducing defects, especially oxygen vacancies [8,9,10,11]. The local electron density distribution will be changed and the oxygen vacancy itself can be the OER active sites. Wang et al. [12] treated Co₃O₄ nanosheets by Ar-plasma to get rich oxygen vacancies. The specific activity of the Co₃O₄ nanosheets with oxygen vacancies was 10 times higher than that of the pristine Co₃O₄. Ceria is a nonstoichiometric material and the oxygen vacancies are easily formed due to the shift between Ce³⁺ and Ce⁴⁺ [13,14,15,16,17]. The oxygen vacancies in Ce-containing catalysts such as CeO₂/Co₃O₄ accelerated the electron transfer, resulting in good OER activity [9]. In our previous work, oxygen vacancies in NiCeO_x catalyst were proved to be the active sites for water oxidation [10]. However, the direct correlation between the concentration of oxygen vacancies and OER activity has not been identified yet.

In this work, a series of Ni_yCe_100−yO_x with different Ni/Ce ratio were synthesized on NF (nickel foam)/NiO substrate with simple dip-coating and annealing methods. The surface NiO obtained from the oxidation of NF can prohibit the diffusion of Ni atoms to the deposited Ni_yCe_100−yO_x so that the Ni/Ce ratio is not modified. Oxygen vacancy defects are formed successfully in all the NF/NiO/Ni_yCe_100−yO_x (simply referred to as Ni_yCe_100−yO_x) catalysts. The concentration of oxygen vacancy defects for Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts are larger than other Ni_yCe_100−yO_x catalysts, resulting in a similar larger electrochemically active surface area and the same lower Tafel slope of 66 mV/decade. The overpotential to achieve a current density of 10 mA/cm² for the Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts are 338 mV and 341 mV, respectively. It is noted that these overpotentials are lower than other Ni_yCe_100−yO_x catalysts. With a fixed current density of 10 mA/cm², the Ni₇₅Ce₂₅O_x catalyst exhibits an ultra-high stability of over 100 h.

2. Materials and Methods

2.1. Sample Synthesis

The synthesis of NF/NiO/Ni_yCe_100−yO_x catalysts. The first step was to prepare the NF/NiO substrate. A 10 × 15 mm² Nickel Foam (NF, >99.99%, MTI Corporation, Richmond, CA, USA) substrate with a thickness of 0.08 mm was firstly cleaned by acetone (99%, Wako) in an ultrasonic bath for 5 min. Then the NF was rinsed by deionized water three times. Subsequently, the NF was dried in air and annealed for 2 h with a heating rate of 2 °C/min at 400 °C in a muffle furnace to obtain NF/NiO substrate. The second step is to prepare the nickel and cerium mixed precursor solution. The precursor solution was prepared by dissolving 0.3 M citric acid and 0.15 M metal ions in 20 mL ethanol. The molar ratio of Ni and Ce ions were 95:5, 90:10, 75:25, 50:50, 25:75, and 10:90 for the catalysts of Ni₉₅Ce₅O_x, Ni₉₀Ce₁₀O_x, Ni₇₅Ce₂₅O_x, Ni₅₀Ce₅₀O_x, Ni₂₅Ce₇₅O_x and Ni₁₀Ce₉₀O_x, respectively. Ce(NO₃)₃ 6H₂O offers the Ce ions and Ni(NO₃)₂ 6H₂O offers the Ni ions. The final step is the preparation of NF/NiO/Ni_yCe_100−yO_x catalysts, hereafter simply referred to as the Ni_yCe_100−yO_x. The prepared precursor solution was deposited onto NF/NiO by dip coating and then annealed in the same manner as the NF/NiO substrate.

2.2. Structural Characterization

A field emission Scanning electron microscope (SEM, JEOL JSM 7600 FA) was used for the measurements of SEM. A diffractometer (Rigaku Co. Ltd., SmartLab, Japan) with Cu Kα radiation (dwelling time = 2 s, incident angle = 0.5°, step size = 0.02°, λ = 1.541 Å) was used for the collecting of grazing incidence X-ray diffraction data. A Renishaw inVia Raman Microscope system was used to acquire the Raman spectra at room temperature (25 °C). A ×100 objective and a 532 nm excitation laser were used. A PHI 5000 VersaProbe (ULVAC-PHI) with an Al Kα X-ray source (1486.6 eV) was used to obtain the X-ray photoelectron spectroscopy (XPS). The pass energies of 117.4 eV and 23.5 eV were used for the electron analyzer to analyze the wide scans and narrow scans, respectively.

2.3. Electrochemical Measurements

A cylindrical glass cell with a standard three-electrode configuration was used for the electrochemical measurements. A Pt wire and a Ag/AgCl electrode were used as the counter electrode and the reference electrode, respectively. The working electrode was Ni_yCe_100−yO_x electrode. A potentiostat (Princeton Applied Research, VersaSTAT 4) was used to perform the electrochemical measurements. The potentials were calibrated against the RHE according to the following equation: (E_RHE = E_Ag/AgCl + 0.059 pH + E⁰_Ag/AgCl), where E_Ag/AgCl is the potential difference measured between the Ag/AgCl electrode and the working electrode, E⁰_Ag/AgCl (0.1976 V at 25 °C) is the standard electrode potential for an Ag/AgCl electrode, pH is the pH of the electrolyte solution, and E_RHE is the calibrated potential. The electrolytes used were saturated with oxygen before and during the OER experiments.

The polarization curves were collected through linear sweep voltammetry (LSV), and the scan rate was 10 mV/s. Controlled-current water electrolysis was performed using a chronopotentiometric technique [16]. The solution resistance R_s (~2 Ω), determined using the electrochemical impedance spectroscopy (EIS) technique [16], was used to correct the iR drop across the solution. Unless otherwise stated, all given potentials are vs. RHE and corrected for the iR drop across the electrolyte. Tafel plots obtained from the steady-state polarization curves with a scan rate of 1 mV/s. Cyclic voltammetry (CV) was used to determine the electrochemical capacitance of the samples presented in this paper [18]. The potential was swept in a range from 0.05 V above the open-circuit potential (OCP) to 0.05V below the OCP in a static solution with five different scan rates: 0.005, 0.01, 0.025, 0.05 and 0.1 V s⁻¹. The working electrode was held for 10 s at each end of the potential sweep before continuing to the next sweep. All experiments were performed at room temperature.

3. Results and Discussion

The surface morphology information of the NF/NiO and Ni_yCe_100−yO_x samples was analyzed by SEM, and the images are shown in Figure 1. Holes with sizes varying from several nanometers to a few hundred nanometers were observed on the surface of NF/NiO. NF/NiO sample was prepared by annealing Ni substrate in the muffle furnace for 2 h with the elevating rate of 2 °C/min, and these holes were from the Ni foam substrate. The surface of NF/NiO consisted of small nanocrystals, as shown in Figure 1b. The Ni_yCe_100−yO_x samples were synthesized by depositing nickel and cerium mixed precursor solutions with different Ni/Ce ratios on NF/NiO and then annealing the samples in air. The deposited layer covers the surface of NF/NiO nanocrystals for the Ni₉₅Ce₅O_x sample, as shown in Figure 1c,d. Other Ni_yCe_100−yO_x catalysts have the similar morphologies with the Ni₉₅Ce₅O_x catalyst, as displayed in Figure S1 of the supporting information.

Grazing incidence XRD measurements were performed to examine the crystal structure of the Ni_yCe_100−yO_x samples, and the results are shown in Figure 2a. Only Ni and NiO related peaks were found and no CeO₂ related peaks were detected for Ni_yCe_100−yO_x samples. It indicates that Ni and Ce mixed uniformly in the top layer and formed an amorphous structure. The surface element information of the Ni_yCe_100−yO_x samples was further analyzed by XPS. As shown in Figure 2b, Ce 3d peaks of CeO₂ and Ni 2p peaks of NiO were observed and indicates that the deposited layers of Ni_yCe_100−yO_x catalysts are indeed composed of Ni and Ce mixed oxides. The curve fitting results of Ce 3d_3/2 for Ni_yCe_100−yO_x samples are shown in Figure 2c. The peak areas of Ce 3d_3/2 for Ni₉₀Ce₁₀O_x, Ni₇₅Ce₂₅O_x, Ni₅₀Ce₅₀O_x, Ni₂₅Ce₇₅O_x and Ni₁₀Ce₉₀O_x samples are 31.7, 55.7, 206.2, 433.0 and 603.3. For Ni_yCe_100−yO_x samples, the peak areas of Ce 3d_3/2 are in proportion with the Ce content, and the higher Ce content will lead to larger peak areas of Ce 3d_3/2. The sequence of the peak areas of Ce 3d_3/2 of Ni_yCe_100−yO_x samples are in accordance with the Ce content in these samples (Table S1). This indicates that the designed Ni/Ce ratios remain constant for the synthesized Ni_yCe_100−yO_x samples.

Two Raman peaks of 224 cm⁻¹ and 563 cm⁻¹ were observed for Ni₇₅Ce₂₅O_x, Ni₅₀Ce₅₀O_x, Ni₂₅Ce₇₅O_x and Ni₁₀Ce₉₀O_x samples, as shown in Figure 3a. For Ni₉₅Ce₅O_x and Ni₉₀Ce₁₀O_x samples, only the peak of 563 cm⁻¹ was observed. Crystalline CeO₂ is known to have a strong F_2g Raman peak at 464 cm⁻¹ related to its fluorite structure [19]. The presence of ions with the oxidation states lower than Ce⁴⁺ in the CeO₂ has been shown to induce a Raman band, known as the D band, from 500 to 700 cm⁻¹ [19,20,21]. This band is associated with the presence of oxygen vacancy defects created in the non-stoichiometric CeO_2-y by the 3+ coordinated ions. In addition to the introduction of the D band, the F_2g band will be weakened and becomes asymmetric and broad [22]. In the Raman spectra of the Ni_yCe_100−yO_x samples, there is no F_2g band, which suggests that there is no crystalline CeO₂ with a fluorite structure in these samples [23]. The broad peak at 563 cm⁻¹ (D band) indicates the formation of oxygen vacancy defects. The oxygen vacancy defects should be related to the presence of Ce³⁺ because of the incorporation of Ni into CeO₂, as suggested by the literature [19,20,23]. Furthermore, the amorphous structure of Ni_yCe_100−yO_x contributes to the broadness of the peak [22,23]. According to the areas of this peak among different Ni_yCe_100−yO_x samples, we can roughly estimate the concentration of oxygen vacancy defects in these catalysts. The peak areas of the D band have been calculated and summarized in Table S2. The peak areas of 563 cm⁻¹ for Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x samples are similar and larger than those of other Ni_yCe_100−yO_x samples, suggesting that these two catalysts own larger concentration of oxygen vacancy defects. The peak at 224 cm⁻¹ is related to Ce-OH vibrations which are resulted from surface defects. Different types of hydroxyl groups generated by the dissociation of surface adsorbed water and doubly bridging hydroxyl groups on reduced cerium oxide are detected in the Raman spectra [24].

XPS was carried out on Ni_yCe_100−yO_x samples to further analyze the oxygen vacancy defects. The Ce 3d peaks of CeO₂ were observed for Ni_yCe_100−yO_x samples, as shown in Figure 3b. The Ce 3d band is composed of ten individual peaks, that are labeled on Figure 3b as v, v″, v‴, u, u″, u‴, v⁰, v’, u⁰ and u’. The v, v″, v‴, u, u″ and u‴ peaks represent the 3d¹⁰4f⁰ state of Ce⁴⁺, and the v⁰, v’, u⁰ and u’ peaks represent the 3d¹⁰4f¹ state of Ce³⁺ [13,25]. The intensities of these peaks increased with the increasing Ce content in Ni_yCe_100−yO_x samples. The concentration of Ce³⁺ and Ce⁴⁺ can be estimated according to the relative areas of the corresponding peaks. The major valence state of Ce was 4+, Ce³⁺ were also detected for Ni_yCe_100−yO_x samples. It is commonly known that the oxygen vacancy defects will be formed with the appearance of Ce³⁺ to maintain electrostatic balance according to Equation (1):

4Ce⁴⁺ + O²⁻ → 4Ce⁴⁺ + 2e⁻/☐ + 0.5O₂ → 2Ce⁴⁺ + 2Ce³⁺ + ☐ + 0.5O₂

(1)

☐ represents the empty position by the removal of O^2– from the lattice (i.e., oxygen vacancy defect). It suggests that the oxygen vacancy defects formed for the Ni_yCe_100−yO_x samples, which is consistent with the Raman results.

To test the electrochemical performance of Ni_yCe_100−yO_x catalysts, the polarization curves of these catalysts (Figure 4a) were obtained using Linear Sweep Voltammetry (LSV). The overpotentials of the catalysts for the current density of 10 mA/cm² are listed in

Table 1. Electrochemically active surface area (ECSA), Tafel slope, mass-transfer resistance (R_CT) and the overpotential (η) for the current density of 10 mA/cm² for each catalyst investigated in 1 M KOH.

Catalyst	ECSA/cm2	Tafel Slope	RCT	η for 10 mA/cm2
Ni95Ce5Ox	307.5 cm2	68 mV/decade	18.2 Ω	351 mV
Ni90Ce10Ox	362.5 cm2	68 mV/decade	12.6 Ω	350 mV
Ni75Ce25Ox	515 cm2	66 mV/decade	9.9 Ω	338 mV
Ni50Ce50Ox	487.5 cm2	66 mV/decade	7.5 Ω	341 mV
Ni25Ce75Ox	267.5 cm2	68 mV/decade	21 Ω	356 mV
Ni10Ce90Ox	177.5 cm2	73 mV/decade	31.8 Ω	363 mV

. The Ni₁₀Ce₉₀O_x catalyst showed the lowest current density for the applied potentials and had the largest overpotential of 363 mV for the current density of 10 mA/cm². The overpotentials for the Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts were 338 mV and 341 mV to obtain the current density of 10 mA/cm², which were lower than that of other samples. The Tafel slope results of Ni_yCe_100−yO_x catalysts were shown in Figure 4b and Table 1. The Tafel slopes of the Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts (66 mV/decade) were close to that of Ni₉₅Ce₅O_x, Ni₉₀Ce₁₀O_x and Ni₂₅Ce₇₅O_x catalysts (68 mV/decade), and were lower than that of the Ni₁₀Ce₉₀O_x catalyst (73 mV/decade). An electrocatalyst with a low Tafel slope will have a small kinetic barrier for electron and mass transfer [2,5]. This indicates that the transfer barriers of electron and mass in the Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts are slightly improved.

The double-layer capacitance can be used to estimate the electrochemically active surface area (ECSA) of each sample. In order to know the double-layer capacitance, we first obtained CV curves of the capacitance current in the non-Faradaic voltage region (a 0.1 V potential range centered on the OCP) for several different scan rates (Figure 5). The rate of change in the current at OCP with respect to the scan rate corresponds to the double-layer capacitance [18]. For this reason, the current at OCP was plotted against the scan rate for the Ni_yCe_100−yO_x catalysts, and a line of best fit was fitted for each catalyst’s data set, as shown in Figure 4c. The double layer capacitance was 12.3 mF, 14.5 mF, 20.6 mF, 19.5 mF, 10.7 mF, and 7.1 mF for the Ni₉₅Ce₅O_x, Ni₉₀Ce₁₀O_x, Ni₇₅Ce₂₅O_x, Ni₅₀Ce₅₀O_x, Ni₂₅Ce₇₅O_x, and Ni₁₀Ce₉₀O_x samples, respectively. The ECSA can be calculated according to the formula ECSA = C_DL/C_s, where a specific capacitance of C_s = 0.040 mF cm⁻² was used in this work [18]. The calculated ECSA values for the Ni_yCe_100−yO_x catalysts as well as other relevant electrochemistry parameters are summarized in Table 1. The ECSAs of the Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts are similar and larger than that of other Ni_yCe_100−yO_x catalysts. This is consistent with the observed current densities, as a larger ECSA means a sample has more active sites and therefore can catalyze more reactions at once and sustain a large current.

The charge transfer resistance (R_CT) of the Ni_yCe_100−yO_x catalysts were obtained from their Nyquist plots, as shown in Figure 4d. As shown in Table 1, the R_CT of the Ni_yCe_100−yO_x catalysts decreased firstly and then increased with the increasing Ce content in the catalysts, and the Ni₅₀Ce₅₀O_x catalyst had the lowest R_CT of 7.5 Ω at an applied bias of 1.55 V vs. RHE. The Ni₇₅Ce₂₅O_x catalyst also showed a low R_CT of 9.9 Ω. However, the R_CT of the Ni₁₀Ce₉₀O_x (31.8 Ω) catalyst was much higher than other Ni_yCe_100−yO_x catalysts. The small mass-transfer resistance of the Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts stands for their favorable OER kinetics.

For the Ce-based catalysts, the oxygen mobility can be promoted by the generated oxygen vacancy defects, resulting an improved ionic conductivity [10]. Also, the oxygen vacancy defects can be act as the OER active sites to catalyze the water oxidation reaction [17]. Therefore, the larger concentration of oxygen vacancy defects results in the lower Tafel slopes, small mass-transfer resistance, and larger ECSAs of the Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts, which account for the higher OER activities for these two catalysts.

Controlled-current water electrolysis (Figure 6) was done to test the long-term performance and stability of the Ni₇₅Ce₂₅O_x catalyst. With a fixed current density of 10 mA/cm², the potential remained stable at 1.57 V vs. RHE for more than 100 h. This indicates that the Ni₇₅Ce₂₅O_x catalyst is very stable during long-term water electrolysis.

4. Conclusions

In summary, a series of NF/NiO/Ni_yCe_100−yO_x catalysts were synthesized through the simple dip-coating and annealing methods. The oxygen vacancy defects are formed successfully in all the Ni_yCe_100−yO_x catalysts, and the concentration of oxygen vacancy defects for Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts are larger than other Ni_yCe_100−yO_x catalysts. This results in the larger electrochemically active surface areas for Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts because of the abundant active sites offered by the defects. The rich oxygen vacancy defects also improve the ionic conductivity so that a lower Tafel slope of 66 mV/decade is obtained for Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts. The improved ionic conductivity also results in the small mass-transfer resistance of Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalysts, which is favorable for their OER kinetics. Therefore, the Ni₇₅Ce₂₅O_x and Ni₅₀Ce₅₀O_x catalyst exhibit higher OER activity than other Ni_yCe_100−yO_x catalysts with the overpotential of 338 mV and 341 mV for the current density of 10 mA/cm². With a fixed current density of 10 mA/cm², the potential remains stable at 1.57 V for more than 100 h, indicating the long-term stability of the Ni₇₅Ce₂₅O_x catalyst.