Sr₃Fe₂O₇ powder was synthesized by a solid-state reaction [26] . Stoichiometric amounts of α-Fe₂O₃ (Kojundo Chemical Lab., 99.9%) and strontium carbonate (SrCO₃, Kojundo Chemical Lab., 99.9%) were ground in a ball mill and calcinated in air at 900˚C for 9 h. The resulting powder was pressed into pellets and sintered in air at 1300˚C for 24 h. When Sr_2.6La_0.4Fe₂O₇ and Sr₃FeTiO₇ were prepared, stoichiometric amounts of lanthanum oxide (La₂O₃, Kanto Chemical, 98.0%) and titanium dioxide (TiO₂, Kanto Chemical, 98.0%) were added to the starting materials as reported in the literature [27] [28] . All chemical reagents were used without further purification.

Electrodes were prepared using a spray deposition method as reported previously [25] . Briefly, 300 mg of the synthesized powder sample was suspend- ed in 200 mL of ethanol. The suspension was sprayed onto a clean conducting glass substrate (FTO-coated glass, resistance: 20 Ω/square; SPD Laboratory Inc.) at 170˚C using an automatic spray gun (Lumina, ST-6; Fuso Seiki Co., Ltd.). After coating, the resultant transparent black film was calcinated at 500 ^oC in air for 2 h.

The crystal structures of the electrocatalysts were analyzed by X-ray diffraction (XRD) using a PW-1700 X-ray diffractometer (PANalytical) with monochromatic Cu Kα radiation. XRD patterns were recorded from 15˚ to 80˚ in 2θ with a step size of 0.02˚ and a scan rate of 0.25˚/min. Scanning electron microscopy (SEM) inspection was performed using a scanning electron microscope (JSM- 6500F, JEOL).

Polarization curves were obtained with a commercial potentiost at and potential programmer (HZ-5000, Hokuto Denko). A platinum wire was used as a counter electrode. All potentials were measured against a silver/silver chloride reference electrode (Ag/AgCl/KCl(sat.)) and converted to the standard hydrogen electrode (SHE) reference scale using the equation U(versus SHE) = U(versus Ag/AgCl/ KCl (sat.)) + 0.197. The electrolyte solution used for all experiments was 0.1 M sodium sulfate (Na₂SO₄) aqueous solution, which was prepared from highly pure Milli-Q water (18 MΩ∙cm) and Na₂SO₄ (Kanto Kagaku, 99.0%). The pH was adjusted to 7 using 0.1 M sulfuric acid (H₂SO₄, Kanto Kagaku, 96.0%) and 0.1 M sodium hydroxide (NaOH, Kanto Kagaku, 97.0%). No agent for pH buffering was added to the electrolyte solution to avoid effects from the adsorption of multivalent anions. Prior to the measurement, the electrolyte was maintained at a certain temperature and bubbled with argon gas for at least 15 min. Polarization curves were measured by sweeping the electrode potential from the rest potential to 1.5 V at a scan rate of 10 mV/s and the concentration of O₂ dissolved in the electrolyte was monitored during the potential sweep using a needle-type O₂ microsensor (Microx TX3-trace, PreSens). The current density was normalized to the geometric surface area of the electrode.

Figure 1 shows the XRD patterns of Sr₃Fe₂O₇ and its substituted materials Sr_2.6La_0.4Fe₂O₇ and Sr₃FeTiO₇. All materials exhibited XRD patterns indexed to the tetragonal space group I4/mmm as reported in the literature [26] [27] [28] , and no peaks assignable to other crystal phases were detected. The peak position of the prepared Sr₃Fe₂O₇ (Figure 1(a)) matched with a reference data (Figure 1(b), ICSD no. 163173). In contrast, the intensity of the diffraction peaks corresponding to (00h) planes was particularly intense for our prepared Sr₃Fe₂O₇. This can be understood by the preferred orientation of the Sr₃Fe₂O₇ particles which have plate-like shapes (Figure 2(a)) owing to its two-dimensional (2-D)

layered crystal structure composed of stacked rock salt and perovskite layers with the sequence of SrO-(SrFeO₃)₂. For Sr_2.6La_0.4Fe₂O₇ and Sr₃FeTiO₇, shifts of the diffraction peaks to lower angles were observed (Figure 1(e)), confirming that cationic substitution had taken place. The peak shift was larger for Sr₃FeTiO₇ as the expansion of the crystal lattice has been reported to be more prominent for Sr₃FeTiO₇ than Sr_2.6La_0.4Fe₂O₇ [27] [28] .

Figure 2 shows SEM images of the prepared film electrodes. The Sr₃Fe₂O₇ particles had a diameter ranging from 1 μm to 8 μm (Figure 2(a)). In contrast, the particle sizes of Sr_2.6La_0.4Fe₂O₇ and Sr₃FeTiO₇ were approximately from 0.3 μm to 1.2 μm (Figure 2(b) and Figure 2(c)). All the samples were uniformly deposited on the electrodes. From the cross-section image in Figure 2(d), the thickness of the deposited film was found to be approximately 500 nm.

Figure 3 shows polarization curves of a Sr₃Fe₂O₇ film electrode measured at pH 7. Irrespective of the temperature, we observed simultaneous increases in the anodic current and O₂ concentration while neither of them was observed at this potential by using a bare FTO electrode (data not shown). In contrast to the typical polarization curves for OER, Sr₃Fe₂O₇ showed a slight decrease in the anodic current upon sweeping the electrode potential. This decrease is because a part of Sr₃Fe₂O₇ transformed to Sr₃Fe₂(OH)₁₂ during the measurements by intercalation of water molecules between its two rock-salt-type SrO layers [29] . However, because this transformation causes no anodic current and O₂ formation, we can consider that the observed results indicate that the OER was electrocatalyzed on Sr₃Fe₂O₇.

Notably, when the temperature of the electrolyte was increased from 30˚C to 70˚C, the anodic current showed a negative shift of the onset potential of approximately 200 mV. Since the onset potential for O₂ formation was similarly

shifted, these results indicate that the OER activity of Sr₃Fe₂O₇ is improved by increasing the temperature. As demonstrated in a solid oxide electrolysis cell (SOEC), the OER is thermodynamically more favorable at a high temperature and the standard potential of the OER becomes more negative with increasing temperature owing to a decrease in the Gibbs free energy required for the OER [30] . However, the potential shift due to the change in the Gibbs free energy is estimated to be at most only 40 mV at 70˚C because of the small temperature difference between 30˚C and 70˚C. Thus, the observed improvement of the OER activity should originate from a temperature-dependent property of Sr₃Fe₂O₇.

As described in the introduction, Fe⁴⁺ is considered to play an important role in the OER on Fe-based catalysts; however, Fe⁴⁺ is unstable against CD and rapidly disappears in usual [26] [28] [29] [31] [32] [33] According to the literature, the efficiency of CD is closely related to the electronic bandwidth of σ* bonding composed of Fe-3d and O-2pσ* orbitals, and CD occurs when the bandwidth of σ* bonding is narrow [31] [32] . For Fe-based perovskite com- pounds, the bandwidth broadens at high temperatures because with increasing temperature, the Fe-O-Fe bond angle increases and the electronic interaction between Fe and O strengthens [31] [32] . Kuzushita et al. investigated the temperature dependence of the Fe⁴⁺ stability by performing Mössbauer measure- ments and found that there is a critical temperature for the CD of Sr₃Fe₂O₇ at 70˚C ± 10˚C, indicating that CD is suppressed above 70˚C [26] . Therefore, the observed enhancement of OER activity at 70˚C is considered to be due to the suppression of the CD of Fe⁴⁺.

To examine the validity of the interpretation that the stabilization of Fe⁴⁺ leads to the enhancement of the OER activity of Sr₃Fe₂O₇, we also investigated the effect of Fe⁴⁺ stability on the OER activity using the metal-substituted materials. Figure 4(a) shows the polarization curves of Sr_2.6La_0.4Fe₂O₇ measured at 30˚C and 70˚C. Unlike Sr₃Fe₂O₇, the anodic current initiated to increase from essentially the same potential for both temperatures, which is consistent with the

fact that the substitution of Sr with La suppresses CD at room temperature and that Fe⁴⁺ in Sr_2.6La_0.4Fe₂O₇ is stable at both 30˚C and 70˚C [27] . When the OER was conducted with Sr₃FeTiO₇, in which Fe⁴⁺ is stably contained at room temperature [28] , the onset potential was independent of temperature. Thus, the stabilization of Fe⁴⁺ is an effective means of enhancing the OER activity of Fe oxide. As shown in Figure 4(a) and Figure 4(b), a higher current density was observed at 70˚C.

Although the reason for this is unclear, it is assumed to be due to the greater convection of the electrolyte. Notably, the onset potentials observed with these substituted materials were almost the same as that observed for Sr₃Fe₂O₇ at 70˚C, indicating that the efficient formation of Fe⁴⁺ on Sr₃Fe₂O₇ derivatives enables the initiation of the OER around this potential. From a comparison of the polariza- tion curves (Figure 3, Figure 4(a) and Figure 4(b)), the current density of Sr₃Fe₂O₇ at 70˚C was larger than those of metal substituted derivatives at the same temperature. This is likely to be due to higher concentration of Fe⁴⁺ in Sr₃Fe₂O₇.

From the above results, the OER activity of Sr₃Fe₂O₇ derivatives is considered to be determined by the efficiency of Fe⁴⁺ formation, and the suppression of CD was found to be effective for designing active OER catalysts. CD is known to take place not only with Fe⁴⁺ but also with other first-row transition-metal ions [31] [32] . Previously, one of the authors (T. T.) showed that the CD of Mn³⁺ is the primary origin of the pH-dependent OER activity of MnO₂ and succeeded in enhancing OER activity under a neutral pH by suppressing CD [34] [35] [36] . Thus, by analogy with Fe⁴⁺ and Mn³⁺, the stabilization of other first-row tran- sition-metal ions by suppressing CD is likely to be a promising approach for the development of active OER catalysts made from abundant elements. Fe ions introduced in layer-structured metal (hydr) oxides have been reported to form Fe⁴⁺ without causing CD [37] [38] . The application of Fe-doped layered materials to the OER is currently underway in our laboratory.