Synergistic Effect between Fe⁴⁺ and Co⁴⁺ on Oxygen Evolution Reaction Catalysis for CaFe_1−xCo_xO₃

Abstract

Chemical substitution is an effective way to improve electrocatalytic properties in transition metal oxides. We investigate the synergistic effect between Fe⁴⁺ and Co⁴⁺ ions on the catalytic activity for oxygen evolution reaction (OER) in the Fe–Co-mixed perovskite oxide CaFe_1−xCo_xO₃. The OER activity of CaFe_1−xCo_xO₃ is substantially increased by small amounts of Co (Fe) doping into CaFeO₃ (CaCoO₃), leading to the superiority compared to the pure Fe and Co perovskite oxides. The x dependences of the OER overpotential and specific activity for CaFe_1−xCo_xO₃ (0.05 ≦ x ≦ 0.95) are expressed by constant offset from the weighted average between CaFeO₃ and CaCoO₃, which can be interpreted to be the synergistic effect between Fe⁴⁺ and Co⁴⁺ ions on OER activity. The absence of the optimum x for the highest activity for CaFe_1−xCo_xO₃ contrasts with the volcano-like plots reported in various mixed-metal oxides. First-principle calculations using the special quasirandom structure models on CaFe_1−xCo_xO₃ (x = 0.03–0.5) demonstrate that about half the amount of Fe⁴⁺ is electronically activated to possess smaller charge-transfer energies, corroborating the enhancement of catalytic activity in CaFe_1−xCo_xO₃. These findings provide new insight into the synergistic effects in complex transition metal oxide catalysts.

1. Introduction

An electrochemical water splitting using renewable energy is a promising energy-conversion process to achieve sustainable development because of the potential for the hydrogen generation source.^1–3) The anodic reaction of water electrolysis, oxygen evolution reaction (OER), is accompanied by intrinsically high overpotential leading to colossal energy loss in a large-scale operation. Hence, highly active and cost-effective electrocatalysts to decrease OER overpotentials are desired. Precious metals (Ru, Ir) and their compounds are intrinsically active for OER are currently used as OER catalysts,^4–6) but substitution by earth-abundant elements is an essential subject to realize large-scale practical use.

3d transition metal oxides crystallizing in perovskite and spinel structures have been widely investigated as promising OER catalysts owing to their earth abundance and lower cost, in addition to the flexibility in crystal and electronic structures. To achieve the rational design of OER catalysts, several descriptors based on the electronic states for OER activity are proposed: e_g orbital occupancy in transition metal ions,⁷⁾ O 2p band center relative to Fermi energy,⁸⁾ and charge-transfer energy.^9–12) High-valence ions of late 3d metals such as Fe⁴⁺, Co⁴⁺, and Ni³⁺ have smaller charge-transfer energies, exhibiting high OER activity than lower-valence and early 3d metal ions.^13–17) Despite severe synthesis conditions for high-valence metal oxides (typically high-pressure of several GPa), their electronically activated properties are attractive to investigate the chemical substitution effect on electrocatalytic activity. We previously demonstrated the synergistic effect of Fe–Co mixing on OER catalytic activity in the perovskite and related oxides containing equal amounts of Fe and Co ions.¹⁸⁾ The electronic origin for the enhancement of OER activity of Fe⁴⁺–Co⁴⁺ mixed perovskite oxide CaFe_0.5Co_0.5O₃ (see the schematic crystal structure in Fig. 1(a), which was drawn by using the VESTA-3 program¹⁹⁾) was investigated by density-functional theory (DFT) calculation of randomly distributed Fe and Co ions with a special quasirandom structure (SQS) model. About half of Fe ions in CaFe_0.5Co_0.5O₃ possess smaller charge-transfer energies than the pure CaFeO₃, contributing to the OER activity as electronically favorable states. Since, in general, the OER activities are strongly dependent on the composition for mixed-metal oxides, the detailed information about the x dependence of the OER activity for CaFe_1−xCo_xO₃ (0 ≦ x ≦ 1) may provide further insight into the synergistic effects.

Fig. 1

(a) Schematic of the crystal structure of CaFe_1−xCo_xO₃. (b) XRD patterns for CaFe_1−xCo_xO₃. The reflection indices, hkl, for primary Bragg reflections are demonstrated. The asterisks represent the Bragg reflections from the impurity phase(s).

In the present study, we investigated the OER catalytic activities of CaFe_1−xCo_xO₃ at various x values in the whole composition range. Soft X-ray absorption spectroscopy (XAS) analyses indicated that Fe⁴⁺ and Co⁴⁺ valence states are retained in the entire composition. The x dependences of OER overpotential and specific activity for CaFe_1−xCo_xO₃ (0.05 ≦ x ≦ 0.95) were interpreted by constant offset from the weighted average between CaFeO₃ and CaCoO₃, demonstrating a new type of synergistic effect between multiple transition metal ions. The activation of Fe ions in CaFe_1−xCo_xO₃ at various x values was revealed by the theoretical calculation using the SQS models.

2. Experimental

The precursors with nominal compositions of CaFe_1−xCo_xO_3−δ (x = 0, 0.01, 0.05, 0.1, 0.2, 0.4, 0.5, 0.6, 0.8, 0.9, 0.95, 0.99, 1; δ ∼ 0.5) were synthesized by the complex polymerized method,²⁰⁾ as well as the previous study on CaFe_0.5Co_0.5O₃.¹⁸⁾ A stoichiometric mixture, CaCO₃ (99.95%), Fe(NO₃)₃·9H₂O (99.9%), and Co(NO₃)₂·6H₂O (99.5%) in a molar ratio of 1:(1 − x):x, was dissolved in nitric acid at room temperature. A five-fold excess of citric acid and a one-fold excess of 1,2-ethanediol were added to the solution while stirring. The solution was heated up to about 423 K using a hot stirrer and kept at this temperature for 1 h. The dried powder was fired using a furnace at 673 K for 1.5 h and then kept at 1273 K for 10 h in air with occasional grindings to form a precursor.

The precursor was treated under high-temperature and high-pressure conditions to obtain a stoichiometric sample of CaFe_1−xCo_xO₃. The precursor was mixed with an oxidizing agent KClO₄ (99.5%) at a molar ratio of 1:∼0.4 to compensate for the oxygen vacancy. The mixture was charged into a platinum capsule with a 4 mm diameter and a 7 mm height. The capsule was put into a pressure-transmitting medium made of MgO–CoO composite with an edge length of 18 mm. A graphite furnace and a ZrO₂ thermal insulator were installed into the pressure-transmitting medium. The pressure-transmitting medium was put into cubic WC anvils with 12 mm truncations and then treated at high-pressure and high-temperature conditions (8 GPa and 1173 K) using a Walker-type high-pressure apparatus. The obtained powder was washed with deionized water several times to remove the KCl byproduct. The final product was about 150 mg for each synthesis.

Powder X-ray diffraction (XRD) data were collected using a Rigaku Ultima IV diffractometer with a Cu Kα radiation. The XRD data were analyzed using the Rietveld refinement program RIETAN-FP.²¹⁾ Scanning electron microscopy (SEM) images were collected using TM3030 (Hitachi High-Technologies). The elemental mapping was obtained using JSM-6010LA (JEOL Ltd.). The specific surface area (S_BET) was estimated by the Brunauer-Emmett-Teller (BET) analysis of Kr gas adsorption data (BELSORP-max, MicrotracBEL). The powder sample with about 15 mg was treated at 473 K in a vacuum for 1 h in advance. Soft XAS data at L_2,3-edge of Co and Fe were collected in the total electron yield method at the BL12 beamline of the SAGA Light Source. The sample powder was loaded on an indium plate and installed into the sample stage.

OER catalytic activities were evaluated using a rotating disk electrode system in the same manner as the previous study.¹⁸⁾ A 5 mass% proton-type Nafion suspension, 0.1 M KOH aqueous solution, and tetrahydrofuran (THF) were mixed at a 2:1:97 volume ratio. The catalyst ink was prepared by mixing 5 mg of catalyst, 1 mg of acetylene black, and 1 mL of the above THF-Nafion solution. A 6.4 µL of catalyst ink was taken with stirring and dropped cast onto the glassy-carbon disk (4 mm in diameter). Electrochemical measurements were performed using a rotating-disk electrode system (RRDE-3 A, BAS Inc.) with a bipotentiostat (Model-2325, BAS Inc.). A Pt wire was used as the counter electrode. A Hg/HgO electrode (International Chemistry Co., Ltd.) filled with a 0.1 M KOH aqueous solution (Nacalai Tesque, Inc.) was adopted as the reference electrode. At least three electrochemical measurements for each sample were conducted under O₂ saturation at room temperature. The equilibrium potential of the O₂/H₂O redox couple was fixed to 0.304 V versus Hg/HgO in the present conditions. The cyclic voltammogram (CV) was obtained by scanning the disk potential between 0.3 and 0.9 V versus Hg/HgO at a scan rate of 10 mV s⁻¹ and disk rotation rate of 1600 rpm. The anodic and cathodic scans for CV data were averaged to compensate for the capacitive effect. The disk potentials are represented in those versus reversible hydrogen electrode (RHE). The current density was normalized by the BET surface area of the metal oxide catalyst (in the unit of mA cm⁻²_oxide). IR-compensation for the CV data was conducted using the uncompensated resistance (R ∼ 43 Ω), which was determined in advance.

The SQS models were adopted to investigate the electronic structures of CaFe_1−xCo_xO₃. The structure models with a 2 × 2 × 2 supercell, where 32 Fe and Co atoms are included, were constructed with 16 selected compositions: (atomic ratio of Fe:Co) = 32:0 (x = 0), 31:1 (x = 0.03), 30:2 (x = 0.06), 29:3 (x = 0.09), 26:6 (x = 0.19), 23:9 (x = 0.28), 20:12 (x = 0.38), 17:15 (x = 0.47), 16:16 (x = 0.5), 14:18 (x = 0.56), 11:21 (x = 0.66), 8:24 (x = 0.75), 5:27 (x = 0.84), 2:30 (x = 0.94), 1:31 (x = 0.97), and 0:32 (x = 1). The Warren-Cowley parameter, α_Fe(R), which is defined as: α_Fe(R) = 1 − P_Co(R)/x, was used to evaluate the short-range ordering of Fe ions. Here P_Co(R) is the probability that Co atoms exist in the distance from a certain Fe atom, and x is the Co content in CaFe_1−xCo_xO₃. The locations of Fe and Co ions in each SQS model were determined by minimizing α_Fe(R) with respect to Fe and Co configurations. When the distribution of Fe and Co atoms is completely random, P_Co(R) becomes x, resulting α_Fe(R) = 0. We confirmed that enough small values [α_Fe(R) < 0.1] were obtained for all the models compared with the layered Fe–Co ordering model of CaFe_0.5Co_0.5O₃ [α_Fe(R) = 0.33] (see Fig. 2).

Fig. 2

Warren-Cowley parameter α_Fe(R) as a function of x calculated from the SQS models of CaFe_1−xCo_xO₃.

The spin-polarized DFT calculations were conducted using the SQS models. The ferromagnetic structures were adopted for all the oxides. The Perdew-Burke-Ernzerhof (PBE) functional based on the generalized gradient approximation (GGA)²²⁾ was used as implemented in the Vienna Ab-initio Simulation Package (VASP).^23–25) Radial cutoff of 2.3 Bohr and valence electrons of 4s²3d⁶ (Fe) and 4s²3d⁷ (Co) were adopted for PAW potentials. The Hubbard U correction was used to treat the strong on-site Coulombic interactions on the localized 3d electrons.²⁶⁾ The U_eff parameters were set to 3.9 and 3.3 eV for Fe and Co, respectively, which were chosen to reproduce the experimental oxidation enthalpy.^27,28) The valence wavefunctions were expanded by a plane-wave basis set with a cutoff energy of 500 eV. The 2 × 2 × 2 Monkhorst-Pack k-point mesh was used for Brillouin zone integration. The lattice constants and internal coordinates were optimized until the total energy difference, and residual forces converged to less than 10⁻⁵ eV and 10⁻² eV Å⁻¹, respectively. The band centers of oxygen 2p band (ε_2p)⁸⁾ and the unoccupied 3d band of transition metals (ε_3d-un)^11,12) were computed using the same procedure, as described in our previous paper.¹⁸⁾

3. Results and Discussion

Figure 1(b) shows the XRD patterns of CaFe_1−xCo_xO₃. For all the compositions, the main phases were identified as the GdFeO₃-type orthorhombic perovskite structure with the space group of Pnma, where no cationic ordering between Fe and Co ions were realized. Although CaCoO₃ contained a small amount of unidentified impurity phase(s), the other compositions crystallized in single phases without impurity. There was no peak splitting implying the phase separation with different compositions. Figure 3 demonstrates the x dependence of the lattice constants (a, b, c) and unit-cell volume (V) obtained from the Rietveld refinement of XRD data. The values at the end members (x = 0, 1) agree with the previous reports.^29,30) Assuming the steady spin configurations for Fe⁴⁺ (high spin, total spin quantum number S = 5/2) and Co⁴⁺ (intermediate spin, S = 3/2) in the whole composition range,^29,30) it is reasonable that the unit-cell volume almost monotonically decreased with x increased, as expected from that smaller ionic radius of Co⁴⁺ than that of Fe⁴⁺.³¹⁾ The absence of discontinuity excludes the possibility of substantial electronic-state transitions such as intermetallic charge transfer (e.g., Fe⁴⁺ + Co⁴⁺ → Fe^(4+δ)+ + Co^(4−δ)+) and spin-state transition of Co ions.

Fig. 3

Lattice constants (a, b, c) and unit cell volume (V) versus x in CaFe_1−xCo_xO₃ determined by the Rietveld refinement of the XRD data.

Figure 4 displays the SEM images of CaFe_1−xCo_xO₃. The particle sizes of each sample were distributed between ∼1 and ∼10 µm, and the specific surface areas (S_BET) ranged between ∼0.6 and ∼1.2 m²g⁻¹ (see Table 1). These observations confirm no significant difference in morphology in the samples obtained in the same synthesis conditions. We confirmed that there were no substantial segregations in the elemental mapping (Fig. 5).

Fig. 4

SEM images of CaFe_1−xCo_xO₃.

Table 1 Specific surface areas (S_BET), overpotentials (η), and specific activities (SA) for CaFe_1−xCo_xO₃.

Fig. 5

Elemental mappings of CaFe_1−xCo_xO₃.

Figure 6(a) shows the Fe L_2,3-edge XAS data of CaFe_1−xCo_xO₃. The absorption positions at L₃-edge (2p_3/2-3d transition, ∼707 eV) and L₂-edge (2p_1/2-3d transition, ∼721 eV) were almost identical for all CaFe_1−xCo_xO₃ samples. Hence, the tetravalent states for Fe ions were retained in all the compositions unlike the Fe³⁺ spectrum of LaFeO₃ with a different spectral shape.³²⁾ Figure 6(b) displays the Co L_2,3-edge XAS data of CaFe_1−xCo_xO₃. Together with the positive chemical shift by ∼0.5 eV compared to the Co³⁺ reference of LaCoO₃, almost identical spectral shapes and peak positions of L₃-edge (∼778 eV) and L₂-edge (∼792 eV) for CaFe_1−xCo_xO₃ indicate that the Co⁴⁺ valence state is retained in all x range of CaFe_1−xCo_xO₃. The above XAS analyses assure that CaFe⁴⁺_1−xCo⁴⁺_xO₃ valence states are robust. Accordingly, we excluded the possibility of essential electronic state transition which may affect the catalytic activity.

Fig. 6

XAS profiles of (a) Fe and (b) Co K-edges for CaFe_1−xCo_xO₃ and the references (LaFeO₃ and LaCoO₃).

Figure 7(a) displays the linear sweep voltammograms for CaFe_1−xCo_xO₃ on the Fe-rich side (0 ≦ x ≦ 0.5). The OER onset potential (E_onset) was defined by the potential that the current density exceeds 0.05 mA cm⁻²_oxide. The Fe perovskite (x = 0) exhibited intrinsically high activity exceeding RuO₂, as reported previously.¹⁵⁾ The E_onset shifted lower side even for the tiny amount of Co-doping (x = 0.01). The onset potential further shifted at x = 0.05 and retained almost the same values in the samples with more Co substitution for Fe (x ≧ 0.05). This is clear evidence that both lightly and heavily doped Co ions in CaFeO₃ efficiently enhance the OER activity. Figure 7(b) shows the linear sweep voltammograms on the Co-rich side (0.5 ≦ x ≦ 1). The light Fe-doping (x = 0.99) exhibited a slight shift of E_onset by −10 mV from the pure CaCoO₃ (x = 1) as well as light Co-doing on the Fe-rich side (x = 0.01). When x decreased from x = 0.99 to 0.5, the onset potential gradually shifted to the lower side, indicating that the Fe-doping into CaCoO₃ also enhances the OER activity.

Fig. 7

Linear sweep voltammograms for CaFe_1−xCo_xO₃ of (a) x = 0–0.5 and (b) x = 0.5–1 measured in 0.1 M KOH aqueous solution (pH ∼13) at room temperature. The dashed lines represent the current density of 0.05 mA cm⁻²_oxide. The data for RuO₂ is adopted from Ref. 15).

Figure 8(a) depicts the OER overpotential (η) as a function of x, where η values were calculated from the difference between E_onset and the equilibrium potential (1.23 V vs. RHE): η (V) = E_onset − 1.23, together with the numerical data listed in Table 1. The η values in the substantial Co/Fe-doping (0.05 ≦ x ≦ 0.95) were smaller than the parent compounds of CaFeO₃ and CaCoO₃, respectively. The x dependence of η obtained from linear fitting in the range of 0.05 ≦ x ≦ 0.95 was represented as η (mV) = 32(9)x + 284(5). Compared with the interpolation line between x = 0 and 1 based on the simple mixing rule, the offset of −55 mV can be considered the emergence of the synergistic effect between host and guest metal ions. Figure 8(b) displays the x dependence of the specific activity at 1.6 V vs. RHE. The very lightly doped samples of x = 0.01 and 0.99 demonstrated higher activities than pure Fe and Co oxides, respectively. In the compositions with substantial amounts of Co/Fe doping (0.05 ≦ x ≦ 0.95), almost monotonic enhancement in the specific activity was represented as log[SA/mA cm⁻²_oxide] = −0.79(9)x + 0.09(5), corresponding to the offset of about fivefold from the interpolation between x = 0 and 1. The mutual tendencies in the overpotential and specific activity confirm that the synergistic effects between Fe⁴⁺ and Co⁴⁺ on CaFe_1−xCo_xO₃ system (0.05 ≦ x ≦ 0.95) can be interpreted as the constant shifts from the values expected by the simple mixing rule, except for near-end compositions of x = 0.01 and 0.99.

Fig. 8

(a) OER overpotential (η) and (b) specific activity (SA) at 1.6 V vs. RHE as a function of x for CaFe_1−xCo_xO₃. The red lines represent the result of linear fitting between x = 0.05 and 0.95, whereas the black lines display the weighted average between pure CaFeO₃ (x = 0) and CaCoO₃ (x = 1) (black squares). The markers for x = 0.01 and 0.09 (black circles) are distinguished from those for 0.05 ≦ x ≦ 0.95 (red circles).

The above experimental results revealed that the OER catalytic activity for CaFeO₃ and CaCoO₃ is efficiently increased by only one atomic% of doping of Co and Fe ions, respectively. The OER activities of CaFe_1−xCo_xO₃ exceeded the weighted average between CaFeO₃ and CaCoO₃, as represented by the offsets of overpotential and specific activity. These observations indicate that the synergistic effects between Fe and Co ions are efficiently induced for both Fe-host/Co-guest (x < 0.5) and Co-host/Fe-guest (x > 0.5). Since the synergistic effect on CaFe_0.5Co_0.5O₃ was evaluated by local charge-transfer energies in the previous study.¹⁸⁾ We herein discuss the electronic-state evolution in CaFe_1−xCo_xO₃ based on the DFT calculation. Figure 9 illustrates the total and partial DOS in CaFe_1−xCo_xO₃. In all the compositions, the DOSs exhibited metallic features without band gaps. The DOSs near Fermi energy consisted of transition metal 3d and oxygen 2p bands, whereas Ca 4s bands were located far above Fermi energy (∼8 eV). The overall DOSs were hardly changed by Co-substitution on the Fe-rich side (x < 0.5) and vice versa.

Fig. 9

DOSs of CaFe_1−xCo_xO₃ obtained from DFT calculation.

Figure 10 shows the x dependence of the bulk charge-transfer energy (Δ_bulk) calculated from the difference between band centers of transition metal unoccupied 3d and oxygen 2p bands, in which partial DOSs of Fe, Co, and O atoms at different crystallographic sites were consolidated. A monotonic decrease in Δ_bulk from 4.70 eV (x = 0) to 4.14 eV (x = 1) indicates that the average electronic states change within the weighted average between CaFeO₃ and CaCoO₃. This tendency cannot explain the substantial enhancement of catalytic activity in the CaFe_1−xCo_xO₃ solid solutions. Hence, we investigated the local charge-transfer energies (Δ_Fe–O), which were calculated from the difference between band centers of Fe 3d and O 2p for each FeO₆ octahedra, as well as the previous study in CaFe_0.5Co_0.5O₃.¹⁸⁾ Figure 11 displays the histograms of Δ_Fe–O in the Fe-rich composition (0 ≦ x ≦ 0.5). For pure CaFeO₃ (x = 0), almost all the Δ_Fe–O values converged into Δ_bulk (4.71 eV). When one of the 32 Fe atoms in the supercell was replaced by a Co atom (x = 0.03), Δ_Fe–O values split into two groups of around 4.60 eV and 4.75 eV, the former of which contributes to the enhancement of OER activity. The local electronic-state transformations at all the Fe atoms support the experimental result that the OER activity is increased by the small amount of Co-doping (x = 0.01). When the amount of Co atoms increased, the histogram gradually changed to a broad single peak at the lower energy side (∼4.65 eV) at x = 0.09–0.19, resulting in further broadening for x = 0.28–0.5. The broadened histograms indicate that heavily doped Co atoms efficiently activate a substantial amount of Fe ions.

Fig. 10

The bulk charge-transfer energy (Δ_bulk) as a function of x for CaFe_1−xCo_xO₃.

Fig. 11

Histograms of Δ_Fe–O for CaFe_1−xCo_xO₃ (x ≦ 0.5) obtained from the DFT calculations with the SQS models. The data for x = 0.5 were adopted from Ref. 18).

We summarize various synergistic effects on OER activity for the transition metal oxide catalysts. Figure 12(a) depicts the schematic of x dependence of the OER activity obeying a simple-mixing rule, which is utilized to predict wide-ranging properties in composite materials. To our knowledge, such a simple behavior has not been reported in the transition metal oxide OER catalysts. Still, it can be utilized as a starting model to discuss the synergistic effects. The volcano-like x-dependence is widely observed in the mixed-metal oxide catalysts (Fig. 12(b)). The maximum activity is obtained at a particular x (Fig. 12(c)),^7,33–35) where the structural and electronic factors dominating OER activity are optimized. Since the enhancement of OER activity cannot be expected from the simple mixtures, it can be considered a “positive” synergistic effect among the constituent transition metals. The inverse volcano-like plot is observed in a selected example like LaMn_1−xFe_xO₃.¹¹⁾ Such degradation of OER activity is interpreted as a “negative” synergistic effect, although the detailed mechanism has not been elucidated yet. The coexistence of positive and negative synergistic effects exhibiting complex dependence (Fig. 12(d)) has been reported in the quadruple perovskite CaCu₃Fe_4−xCo_xO₁₂ and magnetoplumbite BaFe_12−xCo_xO₁₉.^36,37) The negative synergy is derived from the doping of relatively less active Co³⁺ ions due to intermetallic charge transfer between Cu and Fe/Co. The x dependences of overpotential and specific activity from the weighted average between x = 0 and 1 for CaFe_1−xCo_xO₃ are classified as the constant-offset synergy (Fig. 12(e)), in which the simple mixing rule remains as a baseline. Although it is notable that such a simple model has not been obtained, we propose that the native electronic states form the basis of the solid-solution catalysts, excluding the possibility of drastic enhancement of OER catalytic activity at a sporadic composition.

Fig. 12

Schematics of typical OER activity of the mixed-metal oxides with various synergistic effects: (a) simple mixing rule, (b) volcano, (c) inverse volcano, (d) coexistence of positive and negative synergies, and (e) constant offset.

4. Conclusions

In summary, we investigated the OER activity in Fe–Co mixed oxide CaFe_1−xCo_xO₃ at various x values. CaFe_1−xCo_xO₃ samples were successfully obtained by using the high-pressure method. The X-ray absorption spectroscopy revealed that the Fe⁴⁺ and Co⁴⁺ states were retained in the entire x range. One atomic% of Co/Fe-doping into CaFeO₃/CaCoO₃ efficiently increased the OER activity. The x dependence of the OER overpotential and specific activity is systematically offset from the weighted average of CaFeO₃ and CaCoO₃, proposing a new type of synergistic effect in solid-solution OER catalysts in contrast to the widely reported volcano-like plots. The DFT calculation using the SQS models revealed that the doped Co ions activate about half the amount of the host Fe ions with smaller local charge-transfer energies. The present findings provide new insight into the synergistic effects in complex transition metal oxide catalysts.

Acknowledgments

The X-ray absorption experiment was performed at SAGA Light Source (proposal numbers 2002004F and 2007067F). This work was supported by JSPS KAKENHI (grant number JP20H02825, JP22H04497, and JP22H04512), Takahashi Industrial and Economic Research Foundation, and JFE 21st Century Foundation.

REFERENCES

• 1)   J.  Song,  C.  Wei,  Z.F.  Huang,  C.  Liu,  L.  Zeng,  X.  Wang and  Z.J.  Xu: Chem. Soc. Rev. 49 (2020) 2196–2214. doi:10.1039/C9CS00607A
• 2)   Z.Y.  Yu,  Y.  Duan,  X.Y.  Feng,  X.  Yu,  M.R.  Gao and  S.H.  Yu: Adv. Mater. 33 (2021) 2007100. doi:10.1002/adma.202007100
• 3)   C.X.  Zhao,  J.N.  Liu,  J.  Wang,  D.  Ren,  B.Q.  Li and  Q.  Zhang: Chem. Soc. Rev. 50 (2021) 7745–7778. doi:10.1039/D1CS00135C
• 4)   J.  Yu,  Q.  He,  G.  Yang,  W.  Zhou,  Z.  Shao and  M.  Ni: ACS Catal. 9 (2019) 9973–10011. doi:10.1021/acscatal.9b02457
• 5)   Y.  Lee,  J.  Suntivich,  K.J.  May,  E.E.  Perry and  Y.  Shao-Horn: J. Phys. Chem. Lett. 3 (2012) 399–404. doi:10.1021/jz2016507
• 6)   Y.  Ping,  R.J.  Nielsen and  W.A.  Goddard: J. Am. Chem. Soc. 139 (2017) 149–155. doi:10.1021/jacs.6b07557
• 7)   J.  Suntivich,  K.J.  May,  H.A.  Gasteiger,  J.B.  Goodenough and  Y.  Shao-Horn: Science 334 (2011) 1383–1385. doi:10.1126/science.1212858
• 8)   A.  Grimaud,  K.J.  May,  C.E.  Carlton,  Y.L.  Lee,  M.  Risch,  W.T.  Hong,  J.  Zhou and  Y.  Shao-Horn: Nat. Commun. 4 (2013) 2439. doi:10.1038/ncomms3439
• 9)   W.T.  Hong,  K.A.  Stoerzinger,  Y.-L.L.  Lee,  L.  Giordano,  A.  Grimaud,  A.M.  Johnson,  J.  Hwang,  E.J.  Crumlin,  W.  Yang and  Y.  Shao-Horn: Energy Environ. Sci. 10 (2017) 2190–2200. doi:10.1039/C7EE02052J
• 10)   H.  Li,  Y.  Chen,  S.  Xi,  J.  Wang,  S.  Sun,  Y.  Sun,  Y.  Du and  Z.J.  Xu: Chem. Mater. 30 (2018) 4313–4320. doi:10.1021/acs.chemmater.8b01321
• 11)   I.  Yamada,  A.  Takamatsu,  K.  Asai,  T.  Shirakawa,  H.  Ohzuku,  A.  Seno,  T.  Uchimura,  H.  Fujii,  S.  Kawaguchi,  K.  Wada,  H.  Ikeno and  S.  Yagi: J. Phys. Chem. C 122 (2018) 27885–27892. doi:10.1021/acs.jpcc.8b09287
• 12)   I.  Yamada,  A.  Takamatsu,  K.  Asai,  H.  Ohzuku,  T.  Shirakawa,  T.  Uchimura,  S.  Kawaguchi,  H.  Tsukasaki,  S.  Mori,  K.  Wada,  H.  Ikeno and  S.  Yagi: ACS Appl. Energy Mater. 1 (2018) 3711–3721. doi:10.1021/acsaem.8b00511
• 13)   X.  Li,  H.  Wang,  Z.  Cui,  Y.  Li,  S.  Xin,  J.  Zhou,  Y.  Long,  C.  Jin and  J.B.  Goodenough: Sci. Adv. 5 (2019) eaav6262. doi:10.1126/sciadv.aav6262
• 14)   H.  Sun,  X.  Xu,  Y.  Song,  W.  Zhou and  Z.  Shao: Adv. Funct. Mater. 31 (2021) 2009779. doi:10.1002/adfm.202009779
• 15)   S.  Yagi,  I.  Yamada,  H.  Tsukasaki,  A.  Seno,  M.  Murakami,  H.  Fujii,  H.  Chen,  N.  Umezawa,  H.  Abe,  N.  Nishiyama and  S.  Mori: Nat. Commun. 6 (2015) 8249. doi:10.1038/ncomms9249
• 16)   M.  Kinoshita,  I.  Yamada,  S.  Kawaguchi,  K.  Oka and  S.  Yagi: Mater. Trans. 61 (2020) 1507–1509. doi:10.2320/matertrans.MT-MN2019032
• 17)   S.  Yagi,  K.  Wada,  J.  Yuuki,  W.  Liu and  I.  Yamada: Mater. Trans. 61 (2020) 1698–1702. doi:10.2320/matertrans.MT-M2020147
• 18)   I.  Yamada,  M.  Kinoshita,  S.  Oda,  H.  Tsukasaki,  S.  Kawaguchi,  K.  Oka,  S.  Mori,  H.  Ikeno and  S.  Yagi: Chem. Mater. 32 (2020) 3893–3903. doi:10.1021/acs.chemmater.0c00061
• 19)   K.  Momma and  F.  Izumi: J. Appl. Cryst. 44 (2011) 1272–1276. doi:10.1107/S0021889811038970
• 20)   M.  Kakihana: J. Sol-Gel Sci. Technol. 6 (1996) 7–55. doi:10.1007/BF00402588
• 21)   F.  Izumi and  K.  Momma: Solid State Phenom. 130 (2007) 15–20. doi:10.4028/www.scientific.net/SSP.130.15
• 22)   J.P.  Perdew,  K.  Burke and  M.  Ernzerhof: Phys. Rev. Lett. 77 (1996) 3865–3868. doi:10.1103/PhysRevLett.77.3865
• 23)   P.E.  Blochl: Phys. Rev. B 50 (1994) 17953–17979. doi:10.1103/PhysRevB.50.17953
• 24)   G.  Kresse and  J.  Furthmüller: Phys. Rev. B 54 (1996) 11169–11186. doi:10.1103/PhysRevB.54.11169
• 25)   G.  Kresse: Comput. Mater. Sci. 6 (1996) 15–50. doi:10.1016/0927-0256(96)00008-0
• 26)   S.L.  Dudarev,  G.A.  Botton,  S.Y.  Savrasov,  C.J.  Humphreys and  A.P.  Sutton: Phys. Rev. B 57 (1998) 1505–1509. doi:10.1103/PhysRevB.57.1505
• 27)   Y.  Wang and  G.  Guo: Phys. Rev. B 73 (2006) 064424. doi:10.1103/PhysRevB.73.064424
• 28)   Z.  Hu and  H.  Metiu: J. Phys. Chem. C 115 (2011) 5841–5845. doi:10.1021/jp111350u
• 29)   P.M.  Woodward,  D.E.  Cox,  E.  Moshopoulou,  A.W.  Sleight and  S.  Morimoto: Phys. Rev. B 62 (2000) 844–855. doi:10.1103/PhysRevB.62.844
• 30)   T.  Osaka,  H.  Takahashi,  H.  Sagayama,  Y.  Yamasaki and  S.  Ishiwata: Phys. Rev. B 95 (2017) 224440. doi:10.1103/PhysRevB.95.224440
• 31)   R.D.  Shannon: Acta Crystallogr. Sec. A 32 (1976) 751–767. doi:10.1107/S0567739476001551
• 32)   A.E.  Bocquet,  A.  Fujimori,  T.  Mizokawa,  T.  Saitoh,  H.  Namatame,  S.  Suga,  N.  Kimizuka,  Y.  Takeda and  M.  Takano: Phys. Rev. B 45 (1992) 1561–1570. doi:10.1103/PhysRevB.45.1561
• 33)   Y.  Zhou,  S.  Sun,  J.  Song,  S.  Xi,  B.  Chen,  Y.  Du,  A.C.  Fisher,  F.  Cheng,  X.  Wang,  H.  Zhang and  Z.J.  Xu: Adv. Mater. 30 (2018) 1802912. doi:10.1002/adma.201802912
• 34)   R.P.  Forslund,  W.G.  Hardin,  X.  Rong,  A.M.  Abakumov,  D.  Filimonov,  C.T.  Alexander,  J.T.  Mefford,  H.  Iyer,  A.M.  Kolpak,  K.P.  Johnston and  K.J.  Stevenson: Nat. Commun. 9 (2018) 3150. doi:10.1038/s41467-018-05600-y
• 35)   H.  Li,  S.  Sun,  S.  Xi,  Y.  Chen,  T.  Wang,  Y.  Du,  M.  Sherburne,  J.W.  Ager,  A.C.  Fisher and  Z.J.  Xu: Chem. Mater. 30 (2018) 6839–6848. doi:10.1021/acs.chemmater.8b02871
• 36)   T.  Uyama,  Y.  Okazaki,  S.  Kawaguchi and  I.  Yamada: ACS Appl. Energy Mater. 5 (2022) 214–226. doi:10.1021/acsaem.1c02754
• 37)   I.  Yamada,  F.  Toda,  S.  Kawaguchi and  S.  Yagi: ACS Appl. Energy Mater. 5 (2022) 5995–6002. doi:10.1021/acsaem.2c00394