B‐Site‐Metal Exsolution on Perovskite Oxides Activates Alkaline Water Oxidation via the Lattice Oxygen Mechanism

Effective electrocatalysts are crucial for facilitating the oxygen evolution reaction (OER), the anodic reaction of water electrolysis for renewable green hydrogen production. Perovskite oxides are a group of potential catalysts featuring the lattice oxygen mechanism (LOM) for OER, where O2 formation commences via a lattice oxygen redox process. The LOM pathway breaks the thermodynamic limitation of the adsorbate evolution mechanism (AEM) and achieves a high intrinsic activity. However, perovskite oxides often suffer high OER overpotentials due to the insufficient activation of the LOM pathway. Typically, the overpotential exceeds 300 mV at 10 mA cm−2. This greatly impedes the practical applications of perovskite oxide based OER catalysts. Here, it is demonstrated that the B‐site‐metal exsolution of a La0.6Sr0.4Fe0.8Ni0.2O3‐δ perovskite increases the activity of LOM by a factor of 3.8 at 400 mV overpotential. The activated LOM pathway leads to a 36‐mV reduction in the overpotential at 10 mA cm−2 (from 310 mV to 274 mV) and a 2× increase in the turnover frequency (TOF) at 450 mV overpotential. A membrane electrode assembly (MEA) water electrolyzer equipped with this LSFN‐based catalyst offers 1 A cm−2 current density at 2.46 V and 24‐h operation stability.


Introduction
Renewable electricity powered water electrolysis produces green H 2 with net-zero carbon emission, [1,2] yet large-scale commercialization of this technology is hampered by high production costs. [3]OER, the anodic reaction in the water electrolysis process, is responsible for the low economic feasibility as it accounts for more than 60% of the total overpotential and thus the energy consumption required to drive water electrolysis at industriallyrelevant current densities. [4]OER is a kinetically sluggish DOI: 10.1002/admi.202300760process with a 4-electron transfer.Therefore, the key to reducing the energy cost of water electrolysis is to devise highly active OER catalysts.
OER can be carried out via the adsorbate evolution mechanism (AEM) and lattice oxygen mechanism (LOM) pathways.The former is described using 4 concerted proton-coupled electron transfer (CPCET) reactions evolving only absorbed O-containing species, [5] while the latter involves the mediation of lattice oxygen in the catalyst. [6]According to previous findings, the O-O bond formation in the LOM pathway was found to have a lower reaction energy compared to the AEM. [7]This difference in reaction energies leads to a higher activity for the OER in the LOM pathway.The existence of LOM has been discovered on perovskite oxide based OER catalysts, likely due to the O defects in their lattices (Figure 1). [8]evertheless, perovskite oxides still suffer high OER overpotentials, typically exceeding 300 mV at 10 mA cm −2 . [9,10]hus, accelerating LOM is crucial to reducing the OER overpotential and enhancing activity.One way to implement this idea is to exsolve the B-site metals in perovskite oxides to form more O vacancies than the pristine counterpart.Consequently, this process promotes the rapid adsorption of *OH species at the O vacancy site, [11] resulting in activated lattice oxygen formation.This, in turn, enhances the reactivity of the low-valence metal (LOM) species.(Figure 1).
Hence, in this work, we used a 10% H 2 /Ar reductant to induce the B site metal exsolution in the perovskite oxide lattice and increase the concentration of O vacancies.We used La 0.6 Sr 0.4 Fe 0.8 Ni 0.2 O 3- (LSFN) as an example to demonstrate our approach.The H 2 /Ar-treated LSFN (denoted as H-LSFN) offers 10 mA cm −2 in 1 M KOH at 274 mV overpotential, a Tafel slope of 59 mV dec −1 , and a turnover frequency (TOF) of 12 s −1 .These performance metrics represent a 2× increase in the activity of the OER compared to the pristine LSFN material.In an MEA water electrolyzer using the H-LSFN catalyst and 6 M KOH, we reported a 2.46 V full-cell potential at 1 A cm −2 and stable water electrolysis for 24 h with a degradation rate as low as 6 mV h −1 .The X-ray photoelectron spectroscopy (XPS) study, the H + reaction order analysis, and the H/D kinetic isotope effect (KIE) investigation revealed that the concentration of surface O species increased by a factor of 1.3 due to increased O vacancies, culminating in augmentation of the LOM activity by a factor of at least 3.8 at a 400-mV overpotential.

Material Characterizations
The H 2 /Ar treatment partially converts the LSFN to a Ruddlesden-Popper (RP)-type perovskite structure (SrLaFeO 4 ), [12] and FeNi 3 alloy was produced due to B-site-metal exsolution (Figure 2a).Scanning electron microscopy (SEM) images confirm the formation of nanoparticles on perovskite surfaces (Figure 2b,c).The exsolved nanoparticles have an average diameter of ≈30 nm. Figure 2d shows the high-resolution transmission electron microscopy (HRTEM) image of LSFN, revealing lattice spacings of ≈2.74 Å and 1.94 Å for the (110) and (220) planes, respectively. [13]For H-LSFN, a lattice spacing of ≈1.45 Å in correspondence to the (200) plane of the FeNi 3 alloy and 2.90 Å corresponding to the (103) plane of SrLaFeO 4 were seen (Figure 2e). [12]Energy dispersive X-ray (EDX) elemental mapping images confirmed the uniform element distribution in LSFN and the phase segregation of Ni and Fe elements in H-LSFN due to the exsolution of FeNi 3 nanoparticles (Figure 2f,g; Figure S1a,b, Supporting Information).The same material process method was also applied to La 0.6 Sr 0.4 Cr 0.8 Ni 0.2 O 3- (LSCN), and the Ni exsolution form of nanoparticles was confirmed by SEM, HRTEM, and XRD (Figure S3, Supporting Information).Using XPS, we found stable La and Sr oxidation states after H 2 reduction (Figure S4a,b, Supporting Information), while the formation of metallic Fe and Ni was seen in H-LSFN (Figure S4c,d, Supporting Information).

Electrocatalytic Performance
We then compared the OER activity of H-LSFN and the pristine LSFN in 1 M KOH.The results of linear sweep voltammetry (LSV) obtained using H-LSFN show overpotentials of 274, 340, and 400 mV at 10 ( 10 ), 100 ( 100 ), and 300 mA cm −2 ( 300 ), respectively (Figure 3a).By contrast, at the overpotential of 310 and 400 mV, the pristine LSFN offers 10 and 80 mA cm −2 , respectively (Figure 3a).The LSV curves without iR correction agree with the corrected results (Figure S5, Supporting Information).This translates to a 3.8× improvement in the OER activity at 400 mV on the H-LSFN catalyst compared to pristine LSFN.The contribution of the Ni foam (NF) substrate to OER activity was excluded as it shows a much higher  10 of 400 mV than H-LSFN.We also observed a ≈30-mV less  10 on H 2 /Ar-treated LSCN (H-LSCN) perovskite than the pristine LSCN (Figure S6, Supporting Information), demonstrating the universality of enhancing the OER activity by B-site-metal exsolution.16][17][18][19] The Tafel slope of H-LSFN was 59 mV dec −1 , which is lower than LSFN (90 mV dec −1 ).This suggests that the exsolution of B site metal in H-LSFN causes a shift in the rate-determining step (RDS) from the first electron transfer step to the following elementary reaction.We also compared the Tafel slopes with the reported literature and observed that H-LSFN exhibits a lower Tafel slope (Table S1, Supporting Information).The formation of NiFe (oxy)hydroxide on the surface after OER was confirmed by the elemental mapping results (Figure S7, Supporting Information), which agrees with the literature. [20,21]he electrochemical impedance spectroscopy (EIS) results agree with that of the LSV curves.A multi-component model was successfully fitted to the EIS patterns of all samples, which showed characteristic semicircular patterns.These components include a resistance associated with the solution (R s ), a constant phase element, and a charge transfer resistance (R ct ) (Figure S8, Supporting Information). [22]The Nyquist plots exhibited a change in the EIS shape at 1.5 V versus RHE, close to the  10 on H-LSFN and LSFN (Figure S8a,b, Supporting Information).We ascertained that the R ct of OER on H-LSNF is 1.2 ohm at 1.6 V versus RHE (Figure S8c,d, Supporting Information).However, the pristine LSFN has a higher R ct of 2.3 ohm at the same applied potentials (Figure S8c,d, Supporting Information).This reduction in R ct agrees with the observed faster OER kinetics on H-LSFN than the pristine counterpart, presumably due to enhanced LOM.
To evaluate the intrinsic OER activity, we normalized the OER LSV curves of H-LSFN and pristine LSFN using their electrochemical active surface areas (ECSAs) determined by doublelayer capacitance (C dl ) measurements (Figure S9, Supporting Information).At 350 mV overpotential, the specific activity of H-LSFN (20.11 mA cm −2 ) is ≈3 times as high as that of LSFN (6.71 mA cm −2 ), implying an enhanced intrinsic activity of the catalyst in addition to an increase in active sites (Figure S10, Supporting Information).Although the C dl estimated from EIS results is different from that shown in Figure S9 (Supporting Information), it does not change the conclusion that H-LSFN has a higher ECSA-normalized OER activity than LSFN.TOF, which represents the intrinsic activity of a catalyst in terms of the number of reaction products produced per active site per unit of time, [23] was also calculated based on the quantity of Ni sites (Figure 3d).At 450 mV overpotential, the TOF of H-LSFN was 12.0 s −1 , 2.1 times as high as that of LSFN (5.6 s −1 ), confirming that H-LSFN has greater intrinsic OER activity.Taken together, we concluded that the B-site metal exsolution results in faster kinetics and enhanced intrinsic activity of OER on H-LSFN.
The implementation of the LSFN and H-LSFN catalyst for industrially-relevant water electrolysis applications was demonstrated in an MEA reactor using Pt/C as the cathode and LSFN or H-LSFN as the anode (Figure 3e).The potentials of H-LSFN at 10 mA cm -2 are 1.35 V and 1.14 V at 50 °C in 1 M KOH and 6 M KOH, respectively (Figure S11, Supporting Information).The full-cell potentials of H-LSFN stood at 2.76 V in 1 M KOH electrolyte and 2.46 V in 6 M KOH at 50 °C at 1 A cm −2 (Figure S11, Supporting Information).In 6 M KOH, we demonstrated 24-h stability at 50 °C and 1 A cm −2 .The decay rate of the OER was ≈6 mV h −1 (Figure 3f).Inductively coupled plasma atomic emission spectroscopy (ICP-OES) confirmed the leaching of Ni and Fe after OER (Table S2, Supporting Information), which was known to cause performance degradation. [24]However, H-LSFN has much less Ni and Fe losses at 1 A cm −2 than LSFN.For example, the Fe loss is ≈24% for H-LSFN and ≈49% for LSFN, indicating that H-LSFN has higher stability of the catalyst structure than LSFN.

Mechanistic Studies
To gain insights into the origin of enhanced OER activity on H-LSFN, we investigated the mechanism by measuring the reaction order and the proton/deuteron (H/D) KIE.Unlike the AEM pathway, where the is coupled to electron transfer, the proton and electron transfer processes are decoupled in the LOM pathway. [25,26]By examining the order of the proton reaction ( RHE ), we can have access to the level of interdependence between the electron transfer and proton transfer processes.This can be performed using the following formula: [27,28] RHE where i the current of the OER obtained under varying pH conditions.As shown in Figure 4a, we obtained the OER LSV curves on H-LSFN and LSFN at different pH conditions, respectively.The presence of the non-concerted proton and electron transfer (NCPET) procedure is an important indicator of the LOM mechanism, according to the literature. [25,29,30]Both current densities decreased with increasing pH, suggesting the presence of NCPET processes. [27]H-LSFN has a greater  RHE than LSFN (1.0 versus 0.8), indicating higher proton-electron decoupling for H-LSFN and a larger contribution from the LOM pathway (Figure 4b). [30]We did the same experiment for LSCN (Figure S12, Supporting Information) and confirmed that H-LSCN has a greater  RHE than LSCN (0.32 versus 0.18).Given the LOM pathway involves the dynamic lattice oxygen reconstruction on the catalyst surface, this process is often accompanied by the generation of negatively charged oxygen-containing species. [31]The tetramethylammonium cation (TMA + ) can bind these oxygen-containing components and markedly suppress the OER activity. [32]Therefore, to further verify the OER on H-LSFN and LSFN indeed proceed via the LOM pathway, we performed the test in 1 M tetramethylammonium hydroxide (TMAOH) solutions.A reduction of OER activity was confirmed by the increase in overpotentials for the same current densities and the Tafel slopes (Figure S13, Supporting Information).The above re-sults demonstrate the presence of LOM and reveal that the exsolution of B-site metals increases the contribution of LOM to the OER.
To further investigate the impact of B-site metal exsolution on the LOM mechanism, we performed KIE tests by comparing the OER activities of H-LSFN and LSFN in H 2 O and D 2 O. [33] We found lower current densities at the same overpotentials in D 2 O than in H 2 O (Figure 4c).We then analyzed the H/D KIE on OER within the activation-controlled regime, where the Tafel slopes are similar in H 2 O and D 2 O and the enhancement of current densities across this region is near constant.Our quantification confirms an average current densities decrease of ≈34% on H-LSFN and ≈21% on LSFN at the same overpotentials after switching the electrolyte solvent from H 2 O to D 2 O (Figure S14, Supporting Information).At 1.55 V, D 2 O reduces the OER activity by 33% (Figure 4d).By contrast, at the same potential, the OER activity on pristine LSFN only decreases by 20% (Figure 4d).This agrees with the pH dependence study, where proton transfer exerts a stronger impact on H-LSFN than the pristine LSFN.Similarly, a 31% reduction in OER activity at 1.65 V on H-LSCN was confirmed, ≈1.5 times as high as that on the pristine LSCN (≈20%) (Figure S15,S16, Supporting Information).These results also link the enhanced LOM activity on H-LSFN to B-site metal exsolution.
The relationship between the enhanced LOM activity and Bsite-metal exsolution can be further rationalized by the formation of more oxygen vacancies on H-LSFN than the pristine LSFN.According to the literature, [34][35][36] the oxygen vacancy concentration increases upon the B-site metal exsolution in perovskite.To investigate the changes in oxygen vacancy concentration, O 1s XPS spectra were obtained.(Figure 4e,f).The O 1s spectrum was deconvoluted into two distinct peaks, representing the lattice oxygen species (O lat , ≈529.8 eV) and surface oxygen species (O surf , ≈531.8 eV). [37]The fraction of surface O species increases from 57% to 74% (Figure S17, Supporting Information), equal to a factor of 1.3.The concentration of surface O species increased by a factor of 1.3.The rise in the surface oxygen species concentration can be directly linked to the presence of surface oxygen vacancies. [38]As the surface oxygen vacancies tend to be occupied by hydroxyl species (i.e., OH − ) upon exposure to air, [38,39] an increase in the peak corresponding to surface oxygen species indicates a higher oxygen vacancy concentration at the surface.These results imply that more surface oxygen vacancies were produced on H-LSFN due to the B-site metal exsolution.The increase in the concentration of oxygen vacancies () on H-LSFN compared to LSFN was further quantified iodometric titrations (Table S3, Supporting Information). [40]o further validate the increase in oxygen vacancy concentration caused by B-site metal exsolving, the oxygen intercalation in H-LSFN and LSFN was investigated in an Ar-saturated 1 M KOH. Figure S18 (Supporting Information) shows the oxidation and reduction peaks corresponding to the insertion and extraction of oxygen ions.It is clear that the current density in the regime of intercalation (1.1-1.5 V versus RHE) increases from 9.4 mA cm −2 to 11.9 mA cm −2 with the exsolution of B-site metals, suggesting a facilitated oxygen intercalation and an increasing oxygen vacancy concentration. [40]ccording to the common pseudocapacitive Nernst Equation: [40][41][42] Where E is the measured potential for the intercalation of oxygen, R is the constant of the gas (8.3145J K −1 mol −1 ),  represents the occupancy rate of lattice vacancies that can be filled, F is the constant of Faraday (96485 C mol −1 ), T represents the temperature at which the measurements are taken, E 0 is the standard potential for intercalating oxygen ions.The position of the intercalation redox peak shifts from 1.44 V to 1.47 V, also indicating the increase in the oxygen vacancy concentration in H-LSFN.

Conclusion
In summary, we demonstrated that the B-site metals exsolution in perovskite oxides increases the concentration of O vacancies, resulting in improved OER performance via promoting the LOM pathway.As an example, H-LSFN has a  10 of 274 mV, a Tafel slope of 59 mV dec −1 , and a TOF of 12.0 s −1 , resulting in a 3.8-fold increase at 400 mV overpotential compared to pristine LSFN in 1 M KOH.In an MEA water electrolyzer using the H-LSFN catalyst and a 6 M KOH electrolyte, we reported stable water electrolysis operation for 24 h and a current density of 1 A cm −2 at a full cell potential of 2.46 V, with degradation rates as low as 6 mV h −1 .The pH dependence, H/D KIE, and XPS results together revealed that H-LSFN performs OER via LOM.The B-site metal exsolution generates more oxygen vacancies, facilitating the adsorption of *OH, increasing the decoupling of proton and electron transfer, and thus enhancing the reaction rate via the LOM pathway.We confirmed the universality of our strategy on LSCN catalysts.This finding contributes to advancing efficient and stable perovskite catalysts for OER applications.

Experimental Section
Synthesis of the Samples: The synthesis of LSFN and LSCN oxides was performed through a sol-gel combustion method.For synthesizing LSFN, A stoichiometric ratio of Sr(NO 3 ) 2 , Ni(NO 3 ) 2 •6H 2 O, Fe(NO 3 ) 3 , and La(NO 3 ) 3 •6H 2 O were dissolved in deionized water (DI water).The metal ion: Ethylene diamine tetraacetic acid (EDTA):Citric acid (CA) molar ratio was controlled at 1:1:1.5.The pH of the solution was adjusted to 7-8 using ammonia.After stirring and gelation at 80 °C, the resultant gel was calcined at 350 °C for 4 h to form the precursor powder.To obtain LSFN powder, the precursor powder was fired in air at 950 °C with a ramping rate of 5 °C min −1 for 5 h.For LSCN, the firing temperature was 1100 °C.The H-LSFN and H-LSCN were made by reducing the as-made LSFN and LSCN at 800 °C in 10% H 2 /Ar atmosphere for 3 h.
Material Characterizations: TEM measurements were performed using a TALOS 200X transmission electron microscope.XRD spectra were obtained with a PANalytical X-ray diffractometer.XPS were collected using a ULTRA DLD XPS spectrometer.SEM images were captured by a Zeiss G500 scanning electron microscope.Iodometric titrations were performed according to the reference procedure. [40]2 ml of 2 M KI solution was added to a flask containing 50-80 mg of perovskite.After 30 min, 5 ml of 4 M HCl was added to dissolve the perovskite.The solution was then titrated to a light golden color using a 0.1 M Na 2 S 2 O 3 solution pre-standardized with 0.017 M K 2 Cr 2 O 7 .Starch was then added, and the solution was titrated until clear to reach the endpoint.An AVIO-200 Inductively Coupled Plasma Optical Emission Spectrometer (ICP-OES) was used to study the changes in nickel and iron.30 mg sample was dissolved in 10 ml of 0.5% nitric acid, then diluted to 100 ml with 0.5% nitric acid and mixed well.Standard quality assurance procedures were used, including analyzing initial and continuing calibration checks and blanks, duplicate samples, and preparation of blanks (Blank).
Electrochemical Measurements: To prepare a uniform ink, 2 mg of catalyst and 10 μl of a 5 wt% Nafion solution were mixed in water and ethanol (1:4 v/v, 0.5 ml).The mixture was sonicated for 30 min.The catalyst ink was applied to a 1 × 1 cm 2 Ni foam with 2 mg cm −2 mass loading.A standard three-electrode system using a 1 M KOH electrolyte was utilized for the electrochemical measurements.The reference electrode was Hg/HgO (1 M KOH), and the counter electrode was a Pt wire.
Before each experiment, the catalyst was activated using cyclic voltammetry (CV) with a voltage range of 0.1 to 1 V (versus Hg/HgO) at a scan rate of 10 mV s −1 .All measured values were standardized to the reference of the RHE by applying the equation below: E (V vs. RHE) = E (V vs. Hg∕HgO) + 0.925 + 0.0591 × pH (3) The iR correction was performed using the following equation: Where i and R s correspond to the current and the solution resistance (R s ).Electrochemical impedance spectroscopy (EIS) measurements were conducted after each OER measurement to determine the value of R s .
The Tafel slopes were determined by calculating the slopes of the linear interval of the overpotential versus the logarithm curves.The turnover frequency (TOF) was determined utilizing the equation provided below.
Where j is the current density in specified potential, n denotes the number of electrons transferred during the reaction (For OER, n is equal to 4).A refers to the electrode geometric area, m is the moles of active metal in the catalyst, and F is the constant of Faraday (96485 C mol −1 ).0.03 M, 0.1 M, 0.3 M, and 1.0 M KOH were employed to investigate the pH dependence and the coupling degree between proton and electron transfer processes of the perovskite catalyst towards OER.All the above data were collected using DH7002 potentiostat (Jiangsu Donghua Analytical Instruments Co., Ltd).

Figure 1 .
Figure 1.The schematic illustrates the impact of B-site metal exsolution on the OER on perovskite oxide catalysts.The formation of more oxygen vacancies accelerates OER via the LOM pathway.

Figure 2 .
Figure 2. The morphology and structure of H-LSFN and LSFN.a) XRD diffraction patterns of H-LSFN and LSFN samples.SEM images of b) LSFN and c) H-LSFN.HRTEM images of d) LSFN and e) exsolved nanoparticles.f) EDX elemental mapping of LSFN and g) H-LSFN powders.

Figure 3 .
Figure 3. Electrocatalytic OER performance on H-LSFN and LSFN.a) Polarization curves of H-LSFN and LSFN.b) Comparison of  10 of H-LSFN to the previously reported perovskite OER catalysts.c) Tafel slopes of OER on LSFN and H-LSFN.d) TOF of OER on LSFN and H-LSFN at different overpotentials.e) The schematic of the MEA equipped with Ni foam supported H-LSFN (H-LSFN@NF) anode and a Pt/C cathode.f) The stability of OER on H-LSFN and LSFN in the MEA at 1 A cm −2 .

Figure 4 .
Figure 4.The OER mechanism on H-LSFN.a) The pH dependency of OER activities on LSFN and H-LSFN.b) pH-dependent plot of the logarithm of OER current at 1.6 V after iR correction.The proton reaction order  RHE is indicated by the number in the graph.c) OER LSV curves obtained on LSFN and H-LSFN in 1 M KOH solution.The solvents are H 2 O and D 2 O. d) Comparison of current densities at 1.55 V versus RHE in H 2 O and D 2 O. e) and f) The O 1s XPS spectra of H-LSFN and the pristine LSFN.