A-Site Cation-Ordering Layered Perovskite EuBa0.5Sr0.5Co2-x Fe x O5+δ as Highly Active and Durable Electrocatalysts for Oxygen Evolution Reaction.

The developments of high-performance and tolerant catalysts may enable more sustainable energy in the future, especially toward water oxidation. Herein, we report A-site cation-ordering layered perovskite EuBa0.5Sr0.5Co2-x Fe x O5+δ (EBSCFx) (x = 0.2-0.6) electrocatalysts. When evaluated for oxygen evolution reaction (OER) in alkaline media, EuBa0.5Sr0.5Co1.6Fe0.4O5+δ (EBSCF0.4) exhibits the best catalytic activity among all of these catalysts, as evidenced by the lowest overpotential of 420 mV at a current density of 10 mA cm-2. Notably, the catalytic activity of EBSCF0.4 is better than that of commercial IrO2 at the overpotential >460 mV. Furthermore, the EBSCF0.4-20RuO2 (involving 20 wt % RuO2) composite catalyst is developed and gives an overpotential as low as 390 mV at 50 mA cm-2, which is even superior to commercial RuO2. For overall water splitting, an electrolysis voltage of merely 1.47 V is achieved at 10 mA cm-2 in an electrolyzer employing EBSCF0.4-20RuO2 as bifunctional catalysts, with exceptional durability for 24 h. Such a performance outperforms state-of-the-art IrO2∥Pt/C and RuO2∥Pt/C couples. According to density functional theory (DFT) calculations, the unique catalytic properties of EBSCF0.4 may benefit from highly active Fe sites with octahedral coordination, and the synergistic effects of Fe and Ru sites in the composite catalyst accelerate the electrochemical water oxidation.

Introduction

Nonrenewable fossil fuels have been becoming less and less available due to their increased consumption in daily life, and the demand for sustainable energy is growing as well. Meanwhile, combustion products from fossil fuel result in environmental pollution, i.e., emission of CO2, SO2, NO, etc. Therefore, searching for renewable and clean energy is imperative, and the decomposed products (H2 and O2) of water have been considered to be potential candidates. Electrochemical oxygen and hydrogen evolution from overall water splitting have been recognized to resolve these problems, but the slow oxygen evolution kinetics greatly affects the efficiency of water electrolysis because of complicated oxidization steps, including continuous multielectron reactions.[1−3] To date, noble-metal-based electrocatalysts have been used for oxygen evolution reaction (OER) and hydrogen evolution reaction (HER), such as IrO2, RuO2, Pt/C, and Pd.[2,4−7] However, their scarcity and high cost limit their extensive utilization, and they have poor durability in alkaline electrolyte solutions.[8−10] Much efforts have been made to explore non-noble metal catalysts or efficient catalysts with less content of noble metals, which will have tremendous durability in alkaline water oxidation.[11−18]

Owing to their low cost, stable structure, and compositional flexibility, much attention has been paid to perovskite oxides with the formula AMO3 (A = lanthanide or alkali earth metal; M = transition metal); they have been widely used in many applications, such as gas catalysis and permeation membrane, solid oxide fuel cells, lithium/zinc–air batteries, and supercapacitors.[19−23] Among various perovskite oxides, SrCoO3- and LaNiO3-based perovskites have been studied in the electrocatalysis field due to their inherent activity for alkaline OER.[24,25] Calle-Vallejo and co-workers showed that SrCoO3 possessed the highest OER activity using bulk thermochemistry computations.[26] Shao et al. successfully prepared some SrCoO3-based perovskite catalysts that had considerable catalytic activity for OER (Table S1).[27−32] On the basis of these results, perovskite oxides could be considered as next-generation high-performance electrocatalysts for OER. In the past decade, A-site cation-ordering layered perovskites with the formula LnBaM2O5+δ (Ln = lanthanide elements) had attracted much interest in high-temperature oxygen reduction reactions (ORRs) due to their adequate mix of electronic and ionic conductivity, high surface oxygen exchange, and bulk diffusion coefficients.[33−36] Most recently, layered perovskites have also been reported to show comparable OER performance with substantial stability in alkaline media (Table S1), e.g., PrBaCo2O5+δ,[37] PrBa0.5Sr0.5Co2O5+δ,[38] PrBa0.5Sr0.5Co2–FeO5+δ,[39] NdBa0.5Sr0.5Co1.5Fe0.5O5+δ,[40] PrBa0.8Ca0.2Co1.5Fe0.5O5+δ,[41] PrBa0.85Ca0.15MnFeO5+δ,[42] and NdBaMn2O5.5.[43] In such cation-ordering layered perovskites, the [Ln–Oδ] and [Ba–O] layers alternately locate along the [001] zone axis, where intrinsic oxygen vacancies exist in the [Ln–Oδ] layers due to the coordinating difference between Ln3+ and Ba2+ ions. The creation of surface oxygen vacancies is usually required to optimize the electrochemical performance of oxide catalysts. For the sake of reducing the cost and improving the performance and durability, novel layered perovskite catalysts for OER are pursued.

In our group, it was found that layered perovskite EuBaCo2O5+δ (EBC) shows a high electrical conductivity of 5.92 × 104 S m–1 at room temperature. When substituting Sr2+ for Ba2+ in EBC, the electrical conductivity is enhanced to 1.31 × 105 S m–1 for EuBa0.5Sr0.5Co2O5+δ (EBSC) (Figure S1), while the crystal structure transforms from an orthorhombic (space group Pmmm) symmetry to a tetragonal one (space group P4mmm). The high conductivity allows fast charge transport at the electrode/catalyst/electrolyte interfaces. To further improve the electrocatalytic activity, Sr and Fe codoped EuBa0.5Sr0.5Co2–FeO5+δ (EBSCFx) (x = 0.2–0.6) perovskites are chose as the research targets. In this work, we report a series of cation-ordering layered perovskites EBSCFx as efficient electrocatalysts for alkaline OER. Benefiting from codoping of Sr and Fe ions, EBSCF0.4 exhibits a comparable OER activity to those of commercial IrO2 and RuO2 at high current densities. Furthermore, the EBSCF0.4–20RuO2 composite catalyst is developed with distinct OER and HER activities for overall water splitting, even outperforming the classic IrO2∥Pt/C and RuO2∥Pt/C couples. Density functional theory (DFT) calculations are also performed to clarify the OER mechanism and to explore the active sites in the layered perovskites.

Results and Discussion

Structural Characteristics of EBSCFx

The layered perovskite structure of EBSCFx was first characterized by powder X-ray diffraction (PXRD). Figure a shows the PXRD patterns of EBC and EBSCFx (x = 0–0.6) samples prepared by a glycine–nitrate combusting route. All of the diffraction peaks can be indexed to a layered perovskite structure, and no impurities were detected in the products. When incorporating the Sr and Fe ions into EBC, the splitting double peaks at about 46–47° disappear (Figure b), suggesting a higher tetragonal symmetry for EBSCFx relative to an orthorhombic symmetry for EBC. On increasing the doping fraction of Fe ions, the diffraction peaks gradually shift to the low angle direction, which is indicative of the lattice expansion. This is in agreement with the ionic radius difference between Co3+ (r = 75 pm) and Fe3+ (r = 78.5 pm) with six coordination environments.[44] Rietveld refinements are performed on the observed profiles to obtain accurate structural information. The refined results also demonstrate an increase in the cell volume of EBSCFx (Table S2). The representative refined profiles of EBC and EBSCF0.4 are shown in Figure c,d, and their crystal structures are schematically illustrated in Figure e. The refinements present an orthorhombic superlattice structure (ap × 2ap × 2ap) for EBC (Pmmm) with a = 3.8810, b = 7.8465, and c = 7.5651 Å, as well as a tetragonal superlattice structure (ap × ap × 2ap) for EBSCF0.4 (P4mmm) with a = 3.8583 and c = 7.5853 Å (Table S2). Typically, the refinement converges to the low reliable factors of Rwp = 5.68%, R = 4.50%, and χ2 = 1.74 for EBSCF0.4. The observed, calculated, and difference profiles together with Bragg diffraction positions of the EBSCFx (x = 0, 0.2, and 0.6) samples are shown in Figure S2. From inductively coupled plasma-optical emission spectrometry (ICP-OES) analysis, it was found that the actual atomic ratios in the samples are quite similar to the stoichiometric formulas (Table S2), indicating that our synthesis method can effectively dope the Sr/Fe ions in each catalyst.

Figure 1

(a) PXRD patterns of EBC and EBSCFx (x = 0–0.6) samples. (b) Magnified PXRD patterns of EBC and EBSCFx (x = 0–0.6) samples in the 2θ range of 45–50°. Refined PXRD profiles of (c) EBC and (d) EBSCF0.4 samples. The crosses represent the observed profile, the line through the crosses is the calculated profile, the vertical markers indicate the Bragg diffraction positions, and the bottom line shows the difference between the observed profile and the calculated one. (e) Schematic representations of crystal structures for EBC and EBSCF0.4.

Transmission electron microscopy (TEM) was employed to analyze the detailed structure and composition of the catalysts. Representative TEM images show that the EBC and EBSCF0.4 samples consist of many irregular particulates with an average size of around 500 nm (Figure a,b). From the high-resolution transmission electron microscopy (HRTEM) image (Figure c), clear fringes can be observed in the EBC with a lattice spacing of 0.267 nm, corresponding to the (022) plane of the orthorhombic layered perovskite. Selected-area electron diffraction (SAED) pattern along the [011] zone axis also confirms the orthorhombic Pmmm structure of EBC (Figure d). As shown in Figure e, the HRTEM image of EBSCF0.4 depicts a lattice spacing of 0.376 nm assigned to the (002) plane of the tetragonal layered perovskite. Figure f displays the corresponding SAED pattern of EBSCF0.4, which promulgates the projection of the [100] zone axis for the tetragonal perovskite (P4mmm). The presence of superlattice reflections is identified in EBC and EBSCF0.4 by fast Fourier transformation (FFT) patterns, further confirming their cation-ordering layered perovskite structure (Figure S3). Energy-dispersive X-ray (EDX) spectroscopy was used to determine the complete Fe doping in the EBSCFx solid solutions. The EDX spectrum of EBSCF0.4 suggests the presence of the starting components (Figure g), and the Cu signal originates from the sample holder for TEM experiments. Scanning transmission electron microscopy (STEM) and EDX elemental mapping images reveal a uniform distribution of Eu, Ba, Sr, Co, and Fe elements in EBSCF0.4 (Figure h,i). According to PXRD and TEM analyses, it is concluded that the Sr/Fe ions are thoroughly substituted for the Ba and Co ions in EBC, respectively.

Figure 2

Typical TEM images of (a) EBC and (b) EBSCF0.4 samples. (c) HRTEM image and (d) SAED pattern of the EBC sample. (e) HRTEM image and (f) SAED pattern of the EBSCF0.4 sample. (g) EDX spectrum of the EBSCF0.4 sample. (h) STEM and (i) EDX elemental mapping images of the EBSCF0.4 sample, involving Eu, Ba, Sr, Co, Fe, and O elements.

Transport Properties of EBSCFx

To understand the transport properties of the EBSCFx catalysts, DFT calculations were performed with the EBC (Pmmm), EBSC (P4mmm), and EBSCF0.4 (P4mmm) (Figure S4). The computed total density of states (TDOS) of all of these perovskites crosses the Fermi level (Figure a–c), indicating that they have an intrinsic metallic nature. The partial density of states (PDOS) depicts the intermixed Co 3d/Fe 3d and O 2p states on the Fermi level (Figure d–f). The antibonding orbital from the strong hybridization of O 2p–Co 3d bands is formed crossing over the EF,[45] which suggests the metallic nature of the layered perovskites reported here. The large overlap level of Co 3d and O 2p states in EBSC implies that the covalency of Co 3d–O 2p becomes stronger due to the introduction of Sr dopants.[46] In the valence band, the peaks below EF result from both t2g and eg orbitals. All eg states are occupied above EF in the conduction band, which indicates that Co3+ and Fe3+ stabilize in high spin states, Co3+ (4t2g2eg) and Fe3+ (3t2g2eg). To elucidate the electronic features, the electrical conductivities of EBC and EBSCFx (x = 0–0.6) were measured at elevated temperatures in air. All perovskite samples show a metal-like conduction behavior over the measuring temperature range. The electrical conductivity of EBSC is much higher than that of EBC, e.g., 5.92 × 104 and 1.31 × 105 S m–1 for EBC and EBSC at room temperature, respectively. The enhanced conductivity may be attributed to the increase in the hole carrier concentration (Co4+). In contrast to Ba2+, the smaller Sr2+ ions have a stronger polarization force, leading to a higher oxygen content and average valence state of Co ions in EBSC. On increasing the doping level, the electrical conductivity of EBSCFx gradually decreases, e.g., 7.76 × 104, 3.46 × 104, and 1.46 × 104 S m–1 for EBSCFx (x = 0.2, 0.4, and 0.6) at room temperature, respectively. The Fe ions would hinder small polaron hopping through the Co–O–Co bonds, which may account for the decreased conductivity. The relatively lower conductivity of EBSCF0.4 can be associated with the decreased TDOS near the Fermi level. The metal-like conduction behavior ensures fast charge transfer kinetics on the catalyst surface, which is beneficial for the water oxidation process.

Figure 3

Computed TDOS of (a) EBC, (b) EBSC, and (c) EBSCF0.4, with Co3+/Fe3+ ions in the high spin state (Co3+: 4t2g2eg and Fe3+: 3t2g2eg). The Fermi level is labeled by the vertical dashed line, and the arrows represent spin-up and spin-down state projection. Computed PDOS of the Co 3d, Fe 3d, and O 2p states in (d) EBC, (e) EBSC, and (f) EBSCF0.4. (g) Temperature dependence of electrical conductivity for the EBC and EBSCFx (x = 0–0.6) catalysts in the temperature range of 30–200 °C in air.

OER Activity of EBSCFx

The electrocatalytic activity of EBSCFx catalysts toward OER is first studied by linear-sweep voltammetry (LSV) in 1.0 M KOH solution (Figure a). As expected, the Sr/Fe doping effectively improves the electrochemical performance of the EBC catalyst. The LSV curves show a higher OER onset potential for EBC (1.61 V vs RHE) than those of EBSC, EBSCF0.2, EBSCF0.4, and EBSCF0.6, as well as commercial IrO2 and RuO2 benchmarks. The contrast of the overpotential values is primarily required at 10 mA cm–2, which is an acceptable standard related to solar fuel cell application. Among all layered perovskite catalysts, EBSCF0.4 exhibits the best OER activity. Notably, the EBSCF0.4 catalyst has a comparably low overpotential (η) of ca. 420 mV at 10 mA cm–2. This value is more negative than those of EBC (η = 520 mV), EBSC (η = 500 mV), EBSCF0.2 (η = 490 mV), and EBSCF0.6 (η = 440 mV), but it is still more positive than those of IrO2 (η = 400 mV) and RuO2 (η = 290 mV). The OER activity of EBSCF0.4 is among that of the best-performing perovskite electrocatalysts (Table S1). Importantly, it is noted that the current density of EBSCF0.4 exceeds that of IrO2 at the overpotential >460 mV and is close to that of RuO2 at large overpotentials. These results demonstrate that the electrochemical performance of EBSCF0.4 can be comparable to the commercial benchmarks at large overpotentials. To further improve the catalytic activity of EBSCF0.4, EBSCF0.4–RuO2 composite catalysts are developed. The LSV curves of the EBSCF0.4–xRuO2 (x = 10, 20, and 30 wt %) composite catalysts were measured to estimate the optimal content of RuO2. As shown in Figure S5, EBSCF0.4–20RuO2 has the lowest η value of 290 mV at 10 mA cm–2, indicating that it is the most efficient composite catalyst for OER. Interestingly, EBSCF0.4–20RuO2 delivers an overpotential of merely 390 mV at a large current density of 50 mA cm–2, which is even superior to that of commercial RuO2 (η = 520 mV) (Figure a). This offers a way for designing highly efficient catalysts with less content of noble metals. The synergistic effects of EBSCF0.4 and RuO2 are supposed to have a substantial impact on the OER activity. The structure of the EBSCF0.4–20RuO2 sample was characterized by PXRD and TEM. The obvious characteristic peaks at 32.63, 33.19, 46.86, 48.32, 58.52, and 59.44° are indexed, respectively, to the (110), (102), (200), (004), (212), and (114) facets of the layered perovskite EBSCF0.4 (Figure S6a). Aside from EBSCF0.4, the representative strong peaks at 28.13 and 35.20° are related to the (110) and (101) facets of tetragonal RuO2 (PDF #43-1027). The TEM image shows that the EBSCF0.4–20RuO2 sample is composed of agglomerated RuO2 nanoparticles on EBSCF0.4 with a size of around 20–50 nm (Figure S6b). The clear lattice fringes with lattice spacings of 0.272 and 0.314 nm correspond to the (102) plane of EBSCF0.4 and the (110) plane of RuO2, respectively (Figure S6c,d). Figure b shows the Tafel plots of all samples, in which the derived Tafel slopes are used to examine the OER kinetics. In particular, EBSCF0.4 has the lowest Tafel slope of 68.3 mV dec–1 among the perovskite catalysts. Such a Tafel slope is close to the expected value (59 mV dec–1) for OER, suggesting an electrochemical pre-equilibrium process followed by a chemical rate-limiting step,[47] such as O–O bond formation between the adsorbed −OH and the active Co/Fe=O site. Nevertheless, EBC gives a larger Tafel slope of 117.8 mV dec–1, which is indicative of a rate-limiting electrochemical process.[48] A smaller Tafel slope is obtained for EBSCF0.4 compared with IrO2 (86.3 mV dec–1) and RuO2 (80.4 mV dec–1). Moreover, EBSCF0.4–20RuO2 shows a slightly decreased Tafel slope of 62.3 mV dec–1, reflecting the fast OER kinetics in EBSCF0.4. The mass activities of the catalysts are compared at 1.7 V vs RHE (Figure S7). Most noticeably, EBSCF0.4 has a higher mass activity (112.7 mA g–1) than that of the IrO2 benchmark (105.4 mA g–1), and the mass activity of EBSCF0.4–20RuO2 reaches as high as 396.5 mA g–1, which is ∼3.8 and 2.0 times higher than those of IrO2 and RuO2, respectively.

Figure 4

(a) LSV curves of the EBC, EBSCFx (x = 0–0.6), EBSCF0.4–20RuO2, IrO2, and RuO2 catalysts toward OER in 1.0 M KOH solution. (b) Tafel plots of the EBC, EBSCFx (x = 0–0.6), EBSCF0.4–20RuO2, IrO2, and RuO2 catalysts. (c) LSV curves of the EBSCF0.4 catalyst before and after 3000 cycles. (d) Chronoamperometric (CA) measurements of the EBSCF0.4 catalyst at 1.65 V vs RHE. (e) LSV curves of commercial RuO2 before and after 3000 cycles. (f) Chronoamperometric measurements of commercial RuO2 at 1.52 V vs RHE.

The excellent durability plays another key factor in practical applications toward electrocatalytic OER, and the continuous LSV cycling and chronoamperometry (CA) curves were tested. Figure c shows the LSV curves of EBSCF0.4 before and after 3000-cycle OER measurements. The catalytic activity of EBSCF0.4 retains the initial one after the OER cycling. It is noted that the overpotential decreases evidently at >50 mA cm–2 even after 3000 cycles, probably due to an activation reaction occurring during the cycling process. The practical operation for EBSCF0.4 is assessed by the CA curve over a period of 10 h (Figure d). When biased potentiostatically at 1.65 V vs RHE, the current density is almost constant in the initial 3 h and then slightly increases. Contrarily, remarkable activity degradation is observed in commercial RuO2 after 3000 cycles (Figure e), and the overpotential increases by about 20 mV at 10 mA cm–2. The CA response curve of RuO2 also undergoes a decrease from 10 to 7.3 mA cm–2 (Figure f). The operating durability of EBC and EBSC was also evaluated for alkaline OER (Figure S8). The initial current density of 10 mA cm–2 was achieved at required potentials, which then stabilized around the vicinity of this value for both EBC and EBSC during the 10 h CA tests. The unique tolerance of EBSCF0.4 may be ascribed to its stronger structural stability. From oxygen temperature-programmed desorption (O2-TPD) curves, the EBSCF0.4 catalyst shows a relatively stable feature at elevated temperatures and the lattice oxygen prefers to be continuously released from RuO2 (Figure S9). Moreover, EBSCF0.4–20RuO2 also has a promoted durability relative to RuO2, as confirmed by the LSV cycling and CA curves (Figure S10). The unique stability of EBSCF0.4 is also supported by PXRD patterns and TEM analysis. No diffraction peaks associated with additional phases (metal hydroxides or oxides) could be found in the PXRD pattern of EBSCF0.4 after 3000-cycle OER measurements, and the peak positions almost remained unchanged, while the diffraction intensity became lower due to less catalyst mass for the electrochemical tests (Figure S11a). Especially for EBSCF0.4–20RuO2, all diffraction peaks reveal the coexistence of two different phases of EBSCF0.4 and RuO2 before and after the OER measurements (Figure S11b). Moreover, the cycling durability of EBSCF0.4 was further probed by HRTEM and corresponding FFT patterns. The perfect crystalline structure is observed for the pristine EBSCF0.4 sample (Figure S12a). After the OER measurements, the clear lattice fringes were found to still extend all the way to the catalyst particle (Figure S12b), indicating that the crystalline nature of EBSCF0.4 is retained even after long-term cycling operation. Notably, EBSCF0.4 retains a very thin amorphous film (<1 nm in thickness) on the surface after the OER measurements. As we know, such an amorphous layer with poor electrical conductivity would deactivate the catalytic sites and prevent the diffusion of electrolyte and generated products, resulting in a suppression of the OER activity.[32,49] If the amorphous layer only separates the catalyst surface from the electrolyte, the electrochemical performance of all catalysts can decrease to the lower level identically, but the results suggest that the amorphous film also reflects the bulk properties to a certain extent. The slight increase in the current density of EBSCF0.4 after the OER measurements may be attributed to the transformation of the covalent bonding network. Similar improvements in the OER stability have been clarified in some perovskite catalysts, such as SrCo0.95P0.05O3−δ,[32] SrFeO3,[49] and Ba0.5Sr0.5Co0.8Fe0.2O3−δ.[50] In a perovskite oxide, A-site alkaline-earth metal ions (Ba2+ and Sr2+) have ionic bonding characteristics, whereas B-site transition-metal ions (Co3+ and Fe3+) are covalently coordinated with environmental oxide ions.[51] The surface of EBSCF0.4 becomes amorphous during some cycles of OER, while the Ba2+ and Sr2+ ions are easily dissolved from the surface region into the electrolyte. Consequently, the covalent Co/Fe–O network in EBSCF0.4 prevents progressive amorphization, accompanied by a considerable increase in the OER current.[49,50] STEM-EDX line-scanning profiles were obtained to check the composition change of EBSCF0.4. For the as-synthesized sample, all metal ions show relatively steady profiles from the surface to the bulk region over a distance of 100 nm (Figure S13a), revealing the uniform element distribution in the pristine catalyst. After the OER measurements, the Sr and Ba ions show more substantial change over the near-surface region in comparison to the other metal ions (Figure S13b), suggesting stronger leaching of Sr2+ ions on the catalyst surface.

The electrochemical impedance spectra were acquired at 1.7 V vs RHE to evaluate the charge transfer resistance (Rct) for OER (Figure a). It is noteworthy that EBSCF0.4 has a much smaller semicircle diameter on the real axis than those of the other perovskite catalysts. This demonstrates that EBSCF0.4 provides a low interfacial charge transfer resistance between the catalyst and electrolyte. As fitted by the corresponding equivalent circuits, the Rct value of 24.5 Ω is obtained for EBSCF0.4, which is ∼1/2 of those for EBC (50.3 Ω) and EBSC (46.3 Ω) and is almost equal to both IrO2 (21.8 Ω) and RuO2 (21.5 Ω). For the measured catalysts, EBSCF0.4–20RuO2 has the lowest Rct of 18.4 Ω, which is consistent with the OER activity. The fitting parameters of impedance spectra are summarized in Table S3. Next, the double-layer capacitance (Cdl) associated with the electrochemical active surface area is measured to clarify the electrocatalytic activity. The cyclic voltammetry (CV) curves were recorded at various scanning rates, in which the capacitances can be derived from the difference between anodic and cathodic currents as a function of the scanning rate. As shown in Figure b, the Cdl values of 3.09, 3.85, 3.91, 4.75, and 3.99 mF cm–2 are obtained for EBC, EBSC, EBSCF0.2, EBSCF0.4, and EBSCF0.6, respectively. Accordingly, more active sites are available on the surface of EBSCF0.4. As expected, EBSCF0.4–20RuO2 has the largest Cdl value of 20.7 mF cm–2 (Figure c). N2 adsorption–desorption isotherms suggest the highest specific-surface area of 2.9 m2 g–1 for EBSCF0.4 (Figure S15). We conclude that the electrochemically active zone is likely to be extended by doping the Fe ions, which boosts the high OER activity of EBSCF0.4. Indeed, EBSCF0.4 shows a lower bulk conductivity than those of EBC and EBSC (Figure g), but it provides the lowest interfacial charge transfer resistance (Figure a), highlighting the importance of other influences on oxygen electrocatalysis kinetics. According to previous reports, the charge transfer resistance of layered perovskites could be reduced with addition of a low conducive FeOOH nanoflake.[38] Zhu et al. presented that the charge transfer process for OER may be improved by increasing the active sites.[28] Thus, it is believed that the charge transfer reaction is affected by not only bulk electrical conductivity but also other factors, such as the surface-active site and oxygen defect aspects. As mentioned above, the higher capacitance and surface area of EBSCF0.4 mean more active sites on the catalyst surface. Moreover, it is noteworthy that abundant surface defects also favor the OER kinetics, as discussed in the following section.

Figure 5

(a) Nyquist plots of impedance spectra for the EBC, EBSCFx (x = 0–0.6), EBSCF0.4–20RuO2, IrO2, and RuO2 catalysts recorded at a potential of 1.7 V vs RHE. (b, c) Differences of capacitive current density (Δj/2 = (ja – jc)/2) taken at 1.15 V (vs RHE) as a function of the scanning rate for the EBC, EBSCFx (x = 0–0.6), EBSCF0.4–20RuO2, IrO2, and RuO2 catalysts. (d) X-ray photoelectron spectroscopy (XPS) spectra of O 1s for the EBC and EBSCFx (x = 0–0.6) samples. (e) XPS spectra of Co 2p for the EBC and EBSCFx (x = 0–0.6) samples. The asterisks represent the shoulder peaks arising from the interaction between Ba2+ and O2– ions. (f) XPS spectra of Fe 2p for the EBSCFx (x = 0–0.6) samples.

XPS Spectra of EBSCFx

The X-ray photoelectron spectroscopy (XPS) technique was used to investigate the surface state of the catalysts. Figure d shows XPS spectra of O 1s for EBC and EBSCFx (x = 0–0.6). Four characteristic peaks can be separated by the deconvolution analysis. The binding energies located at ∼532.2, 530.8, 529.5, and 528.2 eV are related to adsorbed molecular H2O, adsorbed −OH or molecular O2, highly oxidative intermediate oxygen species (O2– and O–), and lattice oxygen (O2–), respectively.[52,53] The molar ratios of each oxygen species for the samples are listed in Table S4. The OER catalysis may benefit from the surface oxygen vacancies, which is intimately associated with the surface O2– and O– species of the catalysts.[54,55] For the perovskite electrocatalysts, the formation of oxygen vacancies gives rise to optimal eg electron occupancy, strong OH– adsorption, and improved electrical conductivity, further enhancing the OER activity.[56] Furthermore, Bao et al. reported that the surface oxygen vacancies can reduce the adsorption energy of molecular H2O, leading to the fast OER rate.[57] In our work, the percentages of O2–/O– species for the perovskite catalysts are in the following order: EBSCF0.4 > EBSCF0.6 > EBSCF0.2 > EBSC > EBC, which is in accordance with the order of the OER activity. Consequently, their OER kinetics may be dominated by oxygen vacancy defects (O22– and O–). From the O2-TPD curves (Figure S16), it is seen that the O2 desorption temperature for EBSCF0.4 is much lower than that of EBSC, reflecting its excellent O2 desorption ability. The efficient activity of EBSCF0.4 can be caused by abundant surface oxygen vacancies and strong oxygen desorption ability. Besides that, the core-level Co 2p and Fe 2p XPS spectra were also characterized. First, the partial oxidation of the Co ions is observed in EBSCFx from XPS spectra of Co 2p (Figure e). Two main lines of Co 2p3/2 and Co 2p1/2 are identified for all perovskite samples, in which the fine structure is noticed by the presence of two shoulder peaks. This phenomenon cannot be explained by a small amount of Co2+. The characteristics of Co2+ ions in an oxygen-coordination environment are a very broad main line and a very strong satellite peak at 786 and 803 eV,[58] and both of them are not observed here. In this case, the shoulder peaks might be attributed to the interaction between Ba2+ cations and O2– anions, as indicated by the similar findings in the other perovskites.[52,59] Although it is a challenge to determine the cobalt valence by the peak-differencing operation due to the overlap between Co 2p and Ba 3d signals, the valence of Co ions is still deduced from the position and shape of the Co 2p main lines and satellite peaks. As can be seen, the Co 2p main lines of EBSCFx shift to the higher binding energy direction. These observations, together with the flatter shapes of the satellite peaks,[52,60,61] are indicative of the presence of high-valence Co ions in EBSC and EBSCFx. According to the previous studies by Grimaud et al., the increase in the Co oxidation state represents an increased number of Co3+ and Co4+ with the reduction of Co2+, leading to the robust OER activity of double perovskites.[62] From the first approximation, Co3+ has an intermediate spin state in the octahedral (O) and square pyramidal (C4) symmetry, and Co4+ stabilizes in a high spin state in the O symmetry. Such assignments make the eg electron occupancy of cobalt ions in double perovskites be ∼1.1–1.3. These estimated values agree with the design principle reported by Suntivich et al.,[27] where the volcano peak of the OER activity for the perovskite catalysts corresponds to the eg occupancy near ∼1.2. Besides the eg filling near unity, molecular H2O is preferentially adsorbed to positively charged cobalt ions, thanks to the strong electrostatic affinity. In other words, the Co3+/Co4+ ions have great adsorption ability for molecular H2O, which is a key initial step of OER. Finally, since the ideal electron filling number in the antibonding orbital is close to 1.0 for the oxide catalysts, a high oxidation state of Fe4+ (3t2g1eg) is usually desired. Figure f shows the XPS spectra of Fe 2p for the EBSCFx (x = 0.2–0.6) samples. Two main lines of Fe 2p3/2 and Fe 2p1/2 clearly state the existence of Fe3+, while the subpeak centered at ∼715.5 eV suggests the high oxidation state of Fe4+. The molar ratio of Fe3+/Fe4+ for EBSCFx is approximately 65:35 on the surface of the catalysts (Table S4). The presence of Fe4+ ions in EBSCFx would make an optimal eg filling configuration, further contributing to the OER activity.

Overall Water Splitting

In this work, the electrolyzer was fabricated by pairing EBSCF0.4 (EBSCF0.4–20RuO2) loaded on Ni foam as both anodic and cathodic catalysts for overall water splitting. In view of IrO2/RuO2 and Pt/C for efficient OER and HER, the integrations of IrO2∥Pt/C and RuO2∥Pt/C were also assembled as references. Figure a shows the polarization curves of these catalyst couples in 1.0 M KOH solution. The strong oxidation peak located at around 1.44 V for the RuO2∥Pt/C couple is ascribed to the oxidation process of Ni2+ to Ni3+ on the Ni foam substrate, which is in accordance with the following reaction equation: Ni2+ + 3OH– → NiOOH + H2O + e–.[63,64] This behavior has been studied in detail for most of the Ni-containing electrocatalysts.[64−67] More intense oxidation peaks for RuO2∥Pt/C than those of other catalyst combinations can be observed, suggesting more active sites in precious metal-based catalysts for the oxidation of Ni atoms. Actually, the IrO2∥Pt/C and RuO2∥Pt/C couples effectively catalyze the water electrolysis. The current density of 10 mA cm–2 is achieved at 1.62 and 1.51 V for IrO2∥Pt/C and RuO2∥Pt/C, respectively. By comparison, EBSCF0.4∥EBSCF0.4 delivers a large cell voltage of 1.73 V at 10 mA cm–2, which is attributed to the weak HER activity of EBSCF0.4 (Figure S17). To overcome this point, EBSCF0.4–20RuO2 is applied as bifunctional electrocatalysts for overall water splitting. Surprisingly, EBSCF0.4–20RuO2∥EBSCF0.4–20RuO2 yields an extremely low cell voltage of 1.47 V at 10 mA cm–2. Such performance outperforms those of IrO2∥Pt/C, RuO2∥Pt/C, and the best-performing bifunctional electrocatalysts reported so far (Table S5). The robust water electrolysis of EBSCF0.4–20RuO2 can be explained by its outstanding OER and HER activities (Figures a and S17). As shown in Figure S17, the highly intrinsic activity of commercial RuO2 should be responsible for the much improved HER performance of EBSCF0.4–20RuO2. In our study, only the LSV curves of EBSCF0.4, EBSCF0.4–20RuO2, Pt/C, and RuO2 are measured to preliminarily explore their HER activity, and a more detailed work would be performed in the future. The electrolyzer with a bifunctional EBSCF0.4 or EBSCF0.4–20RuO2 catalyst presents an exceptional durability in alkaline media (Figure b,c). After the 24 h CA tests, EBSCF0.4∥EBSCF0.4 exhibits nearly ∼100% current retention, whereas IrO2∥Pt/C and RuO2∥Pt/C show a dramatic current decay over the same period (Figure S18). Throughout the durability experiments for EBSCF0.4–20RuO2, both anode and cathode surfaces continuously release massive O2 and H2 bubbles (Figure d and Movie S1). In combination with efficient activity and stability, EBSCF0.4 and EBSCF0.4–20RuO2 hold great potential for water electrolysis application.

Figure 6

(a) Polarization curves of overall water splitting for the EBSCF0.4∥EBSCF0.4, EBSCF0.4–20RuO2∥EBSCF0.4–20RuO2, IrO2∥Pt/C, and RuO2∥Pt/C couples in 1.0 M KOH solution. (b) Chronoamperometric measurements of the EBSCF0.4∥EBSCF0.4 couple at 1.73 V. (c) Chronoamperometric measurements of the EBSCF0.4–20RuO2∥EBSCF0.4–20RuO2 couple at 1.47 V. (d) Photomicrograph of H2 and O2 bubbles released from both the EBSCF0.4–20RuO2 cathode and the EBSCF0.4–20RuO2 anode.

DFT Calculations for the OER Mechanism

Finally, DFT calculations were performed on EBC, EBSCF0.4, and RuO2 to obtain insights into their inherent OER activity. The classic four-step reaction mechanism on transition-metal sites is assumed, which is consistent with the method proposed by Nørskov et al.[68,69] This mechanism on a metal center involves four continuous proton–electron conduction reactions with the HO*, O*, and HOO* intermediates (Figure a). The relatively more stable facets of the oxide catalysts were usually adopted as structural models for the free energy calculation of each reaction step.[68] With respect to the relative stability, the (100) facet is considered to estimate the inherent OER activity of the perovskite catalysts.[68,70] In this work, the free energies of (100), (010), and (001) facets are −138.67, −138.26, and −138.34 eV for EBSCF0.4 from the DFT calculations, respectively. Therefore, the more stable (100) plane is chosen for adsorbing the oxygen-based intermediates. For a direct comparison of computational results, DFT calculations were also performed with the (100) plane of RuO2. The slab models of (100) planes for EBC, EBSCF0.4, and RuO2 with adsorbed intermediates are built for the calculations (Figures S19–S21). Gibbs free energies for the adsorptions of the intermediates on the Co, Fe, and Ru sites are obtained.

Figure 7

(a) Schematic illustration of the classic OER mechanism on a metal (M) ion site. Standard free energy diagrams for OER at zero potential (U = 0) and equilibrium potential (U = 1.23 V) by DFT calculations: (b) octahedral Co site in EBC, (c) octahedral Fe site in EBSCF0.4, and (d) Ru site in RuO2.

Figure b–d shows the free energy diagrams of EBC, EBSCF0.4, and RuO2 under different potentials (U = 0 and 1.23 V). The free energy ΔEHO* of the octahedral Co site in EBC is found to be maximum (1.88 eV) among four reaction steps (Figure b). The formation of the HO* intermediate is determined to be the rate-limiting step, and the theoretical overpotential is estimated to be 0.65 V for OER. For the octahedral Fe site in EBSCF0.4, the ΔEO* – ΔEHO* difference is nearly equal to ΔEHOO* – ΔEO*, indicating that the water oxidation is controlled by both steps (Figure c). The theoretical overpotential of the octahedral Fe site is calculated to be 0.43 V, which is much lower than that of the Ru site in RuO2 (0.78 eV) (Figure d). Such a low overpotential demonstrates that the octahedral Fe site is highly active for OER in EBSCF0.4. Besides for the octahedral Fe site, the free energies on octahedral Co and Sr sites in EBSCF0.4 were also calculated. As shown in Figure S22, both Co and Sr sites present large theoretical overpotentials. Basically, this excludes the possibility that the Co and Sr ions are the main active sites. The high activity of the Fe site is attributed to the high valence state of Fe4+ with strong metal–oxygen covalency, thereby giving rise to the efficient activity of EBSCF0.4. According to previous theoretical studies, the ΔEO* – ΔEHO* difference can effectively describe the OER activity of transition-metal oxides due to the close relation between ΔEHO* and ΔEHOO*.[71,72] For the ideal oxide catalyst, ΔEHOO* and ΔEHO* are predicted to be 3.69 and 1.23 eV, and ΔEO* is in the middle at 2.46 eV.[68] In the case of EBSCF0.4, the octahedral Fe site has an optimal ΔEO* of 2.60 eV (Figure c), and the ΔEHOO* and ΔEHO* of the Ru site are 3.68 and 1.03 eV, respectively (Figure d). These free energies are almost the same as the ideal ones. The theoretical calculations support highly active Fe sites in EBSCF0.4 for the water oxidation, and synergistic effects of optimal free energies on Fe and Ru sites benefit the highly catalytic activity of EBSCF0.4–RuO2.

Conclusions

In summary, layered perovskite EBSCFx (x = 0.2–0.6) oxides have been evaluated as electrocatalysts toward alkaline OER. As far as we know, this is the first time to report novel perovskite EBSCFx for catalyzing water oxidation. Most importantly, the catalytic activity of EBSCF0.4 is even better than that of commercial IrO2 at >20 mA cm–2. The unique electrocatalytic properties of EBSCF0.4 may be attributed to high electrical conductivity, increased double-layer capacitance, abundant surface oxygen species associated with the oxygen vacancies, and highly active Fe sites. Furthermore, the EBSCF0.4–20RuO2 composite catalyst was developed and its performance was found to be superior to that of commercial RuO2. When used as a bifunctional catalyst for practical water electrolysis, EBSCF0.4–20RuO2 delivers an extremely low cell voltage of 1.47 V at 10 mA cm–2 with the successive electrolysis for 24 h. The synergistic effects of EBSCF0.4 and RuO2 are supported by the DFT calculations. This study would open a strategy for designing layered perovskite electrocatalysts and would help develop cost-effective all-in-one catalysts for overall water splitting.

Experimental Section

Material Preparation

The EBC and EBSCFx (x = 0, 0.2, 0.4, and 0.6) samples were successfully synthesized by a glycine–nitrate combusting route. Briefly, stoichiometric amounts of Eu(NO3)3·6H2O (99.9%, Aladdin Industrial Corporation), Ba(NO3)2·6H2O (99.5%, Kemiou Chemical Reagent Corporation), Sr(NO3)2 (99.5%, Tianda Chemical Reagent Corporation), Co(NO3)3·6H2O (99%, Guangfu Fine Chemical Engineering Institute), and Fe(NO3)3·9H2O (98.5%, Tianli Chemical Reagent Corporation) were mixed and dissolved in deionized water with magnetic stirring operation. After that, 1.2 g of glycine was placed into the above nitrate mixture, and the molar ratio of glycine to total metal ions was found to be 4:1. The aqueous mixture was fired on a heating plate, calcined at 1050 °C, and maintained for 12 h in ambient air. The final products were denoted EBC and EBSCFx (x = 0–0.6) depending on different doping fractions.

Characterizations

Powder X-ray diffraction (PXRD) data were collected on a Bruker diffractometer (AXS D8 ADVANCE) operated at 40 kV and 40 mA with Cu Kα radiation at room temperature. The diffraction angle (2θ) varied from 10 to 80° at a step size of 0.02°. Rietveld refinement was performed for the PXRD patterns using Rietica software.

Surface element chemistry was analyzed by X-ray photoelectron spectroscopy (XPS). XPS spectra were performed with a KRATOS spectrometer (AXIS Ultra DLD) with an Al Kα radiation source (1486.6 eV). The base pressure for the experiments was set to 7 × 10–8 Pa. After calibrating the binding energies on the basis of C 1s reference (284.8 eV), XPS spectra were fitted using the XPSPEAK program.

The atomic ratios in the EBSCFx (x = 0.2, 0.4, and 0.6) samples were determined by inductively coupled plasma-optical emission spectrometry (ICP-OES) measurements (Thermo Scientific iCAP6300).

The specific surface areas of the samples were estimated from N2 adsorption–desorption isotherms. The experiments were conducted on an ASAP 2400 system at 77 K. The specific surface area was calculated by the Brunauer–Emmett–Teller (BET) model.

Transmission electron microscopy (TEM) was performed with a field-emission microscope (JEM-2100, JEOL) operated at 200 kV. High-resolution transmission electron microscopy (HRTEM), selected-area electron diffraction (SAED) patterns, scanning transmission electron microscopy (STEM), and corresponding energy-dispersive X-ray (EDX) spectroscopy were conducted on a JEM-2100F microscope (JEOL). The sample powder was ultrasonicated and dispersed in an ethanol solution for several hours and then dropped onto a Cu grid coated with an ultrathin C film.

The structural stability of the catalysts was assessed by oxygen temperature-programmed desorption (O2-TPD). The O2-TPD experiments were performed with a TP 5076 chemisorption apparatus (Tianjin Xian Quan) from 30 to 900 °C under an Ar atmosphere (99.99%). About 0.1 g of the catalyst was raised to 300 °C and kept for 1 h in Ar to eliminate the surface contaminants adsorbed from ambient air, and then it was cooled down to 30 °C. Subsequently, the pretreated sample was linearly heated with a constant rate (10 °C min–1) under a flowing Ar atmosphere.

For electrical conductivity measurements, the as-prepared powder sample was uniaxially pressed to a compact pellet (2.5 × 2.5 × 20 mm3) at 10 MPa and then sintered in air at 1050 °C for 12 h with a constant ramp of 5 °C min–1. The relative density of the sintered pellet was around 90% compared with the theoretical value. The electrical conductivity was measured in air by a standard four-point method between 30 and 200 °C. Four Ag wires were attached onto the compact specimen using a conductive Ag slurry (Sino-Precious Metals Holding Company). A multimeter (Keithley 2700) and a sourcemeter (Keithley 2450) were used to supply the current and to record the resistance, respectively. The electrical conductivity was calculated by the following equation: σ = (1/R) × (l/S), where R is the resistance and l and S represent the length and cross-section area of the specimen.

DFT Calculations

To elucidate the electronic structure and catalytic properties, density functional theory (DFT) calculations were carried out using the Vienna ab initio simulation package (VASP) code, with the projector augmented wave (PAW) potential, the spin-polarized generalized gradient approximation (GGA), and the Perdew–Burke–Ernzerhof (PBE) exchange–correction function. We chose Eu_3, Ba_sv, Sr_sv, Co, Fe, and O PAW pseudopotentials, involving Eu (4d/5s/5p), Ba (5s/5p/6s), Sr (4s/4p/5s), Co (3d/4s), Fe (3d/4s), and O (2s/2p) valence electrons. All slab models had several atomic layers and a vacuum spacing of ∼15 Å, and the Brillouin zone was sampled on an 8 × 4 × 2 Monkhorst–Pack grid to obtain the same density of k-points. The relaxations of the lattice parameters and all internal atom positions were allowed by a conjugate gradient algorithm until the ΔE was below 10–4 eV and the residual force difference was below 0.02 eV Å–1. The energy cutoff was set to be 400 eV.

The Gibbs free energies EHO*, EO*, and EHOO* represent the adsorption energies of HO*, O*, and HOO* intermediates on the surface of the catalysts, respectively. E(*) is the energy of the clean surface, and EH and EH are the energies of gas-phase H2 and H2O molecules, respectively. The Gibbs free energy differences (ΔEHO*, ΔEO*, and ΔEHOO*) can be calculated by the following expressions:[68]

Electrocatalytic Measurements

For the electrocatalytic measurements, the catalyst loaded on an electrode was prepared via a drop-casting technique. In a typical procedure, the catalyst powder (2.0 mg), acetylene black (2.0 mg, MARCLIN Biochemical Corporation), 5 wt % Nafion solution (25 μL, Sigma-Aldrich Fine Chemicals Corporation), isopropyl alcohol (55 μL, FUYU Fine Chemical Engineering Corporation), and 165 μL of deionized water were physically mixed and ultrasonicated for several hours to obtain a catalyst slurry. Afterward, the as-prepared slurry (5 μL) was dropped onto the surface of a prepolished glassy carbon electrode (GCE) with a diameter of 0.5 cm. The mass loadings of total materials and oxide catalyst were 0.4165 and 0.2082 mg cm–2, respectively. The catalyst-coated GCE was subsequently dried at room temperature for electrocatalytic measurements. The commercial IrO2 (99.9%, Energy of Chemical, SAEN Chemical Technology Corporation) and RuO2 (99.9%, Aladdin Biochemical Corporation) benchmarks were also tested for performance comparison. For the EBSCF0.4–xRuO2 (x = 10, 20, and 30 wt %) composite catalysts, 2 mg of the EBSCF0.4 and RuO2 mixture was used to prepare the catalyst electrode through the same procedure.

Electrochemical measurements for oxygen evolution reaction (OER) and hydrogen evolution reaction (HER) were performed with an electrochemical workstation (VERTEX, IVIUM Technologies BV). The standard three-electrode cell with catalyst-coated GCE (work electrode), graphite rod (counter electrode), and Ag/AgCl/3.5 M KCl [reference electrode, 0.197 V vs reversible hydrogen electrode (RHE)] was applied for the measurements, while 1.0 M KOH solution was used as the alkaline media. Before each measurement, high-purity N2 (99.99%) or O2 (99.99%) was purged for 30 min to obtain a N2- or O2-saturated electrolyte solution, and then cyclic voltammetry (CV) curves were measured several times at 50 mV s–1 until a stable state was obtained. Linear-sweep voltammetry (LSV) experiments were carried out at a scanning rate of 2 mV s–1, from which the Tafel slope values could be derived based on the Tafel expressionwhere η, b, j, and jo represent the overpotential, Tafel slope, current density, and exchange current density, respectively. Chronoamperometry (CA) experiments were performed to examine the long-term durability of the catalysts. In this work, all potentials were corrected to compensate for the effect of the electrolyte solution and then converted to the calibrated RHE ones (E vs RHE) by the following equation:where E(NHE) is the potential referenced to the normal hydrogen electrode. The following expression was used to calculate the OER overpotential (η):Electrochemical impedance spectra were acquired at 1.7 V vs RHE for OER in the frequency range of 10–2–105 Hz with amplitude 10 mV. To evaluate the effective electrochemical surface area (ECSA), CV experiments were performed in a non-Faradic current region to obtain the double-layer capacitance (Cdl) values. The CV curves were measured between ∼1.1 and 1.2 V vs RHE at various scanning rates of 40, 60, 80, 100, 120, 140, 160, and 180 mV s–1. The formula Δj/2 = Cdl × v was used to determine the double-layer capacitance, where Δj is the difference between anodic current (ja) and cathodic current (jc) extracted from the current density at 1.15 V vs RHE, and v is the scanning rate for CV. So the slope value of the fitting line for Δj/2 against v represents the Cdl value that was proportional to the ECSA. All electrochemical measurements were carried out at room temperature.

For electrochemical performance concerning the overall water splitting, the EBSCF0.4 or EBSCF0.4–20RuO2 catalyst was loaded on commercial Ni foam (1 × 1 cm2, JIAYISHENG Electronics Corporation) as both the anode and cathode. The electrolyzer with the IrO2 (RuO2) and 20 wt % Pt/C (Pt/C, E-TEK Corporation) couple was also assembled for performance comparison. The mass loading of all catalysts was 0.8 mg cm–2 for the overall water splitting, and the LSV and CA measurements were carried out in 1.0 M KOH.