Understanding dissolution mechanism of oxide perovskites and activity tuning for O₂ evolution by surface doping

Abstract

The oxygen evolution reaction (OER) is the major bottleneck in electrochemical water splitting. Development of efficient electrocatalysts from earth-abundant materials is essential. This study reports a design principle for tuning the catalytic efficiency and stability of perovskite oxides (ABO₃), focusing on SrMnO₃, SrFeO₃, SrCoO₃, and SrNiO₃ through B-site surface doping with Mn, Fe, Co, and Ni metal ions. Owing to multiple active sites and tunable oxidation states, these oxide surfaces exhibit a considerable reduction in thermodynamic overpotential (η_TD) upon doping. The dissolution free energy calculations are performed across a wide pH window (acidic to alkaline) and oxidative potentials to determine the Pourbaix stability and surface dissolution pathways. We find that SrCoO₃ and SrNiO₃ show the highest stability on doping with different transition metal atoms. Reaction free energy analysis indicates that the lattice oxygen mechanism generally outperforms the adsorbate evolution mechanism on the undoped surfaces. Surface doping results in a wide variation in the preferred reaction pathway. Mn-doped SrNiO₃ emerges as the most active system with η_TD of 0.34 V, while Fe doping enhances the activity of SrCoO₃ the most. Crystal orbital Hamilton population and electronic structure analyses show that modulation of the d-band center and surface structure deformation upon doping, influences the catalytic activity and stability. Using the activity descriptor G_max(η) at an applied η of 0.3 V, the corresponding rate-determining steps (RDS) are identified. While most sites follow a single-step RDS, Fe- and Co-doped SrNiO₃ exhibit multistep contributions. On varying the doping percentages of Fe on the SrCoO₃ surface, we find that 25% surface doping shows the highest catalytic activity for OER.

---

Introduction

With the depletion of fossil fuels and the enhancement of environmental pollution, the demand for clean and sustainable energy sources has surged significantly in recent years. Renewable energy generation from solar, wind, and hydro power has shown significant advantages.¹ However, these renewable energy options often face challenges due to seasonal natural cycles and regional variability. To address these limitations, there has been growing interest over the past few decades in developing energy storage and conversion technologies to store energy intermittently and use it as needed.^2–5

The water-splitting reaction could allow for the generation of renewable fuels.^6–8 Electrochemical water-splitting reactions involve two half-cell reactions: the hydrogen evolution reaction (HER: 2H⁺ + 2e⁻ → H₂) at the cathode and the oxygen evolution reaction (OER: 2H₂O → O₂ +4H⁺+ 4e⁻) at the anode.^3,9–11 H₂ serves as a clean fuel but while the formation of H₂ is more important, the OER at the anodic half-cell is the primary bottleneck of the process. The high overpotential associated with the OER reduces the overall reaction efficiency.^12–16

An efficient electrocatalyst could reduce the overpotential for O₂ evolution. In the past decades, increasing attention has been given to developing electrocatalysts for OER. While developing efficient catalysts, three crucial factors must be considered: activity, stability, and cost-effectiveness.^1,17–20 The Pt-group metal-based electrocatalysts, such as iridium oxide (IrO₂) or ruthenium oxide (RuO₂), are found with the highest efficiency in an acidic medium for the OER.^21–23 However, due to their expensive nature, these catalysts would be unsuitable for industrial-scale utilization. The other catalysts with good activity for OER face several inherent challenges. The catalytic surfaces undergo surface reactions under applied potentials and solvent pH, resulting in the gradual degradation of the catalytic sites.^24–29 This leads to a reduction in the catalytic efficiency under operating conditions over time. Inspired by the recent advancements in developing effective synthetic strategies and understanding catalytic efficiencies of transition metal-based catalysts, tremendous research attention has been drawn to investigate the OER activity of the transition metal oxide surfaces.^30–34

Transition metal oxides have emerged as a promising alternative to traditional Pt-group metal-based electrocatalysts for OER. Among these, perovskite oxides, with the general formula ABO₃, offer unique advantages.^35–41 Typically, the A-site is occupied by non-transition metal ions, whereas the B-site hosts transition metal ions. The A-site shows the 12-fold coordination with oxygen, and the B-site has a 6-fold coordination.^34,35 The oxidation states of the transition metal site in perovskite oxides could be further tuned by varying the A-site metal ions and the creation of the O vacancies.^42–44 Furthermore, in recent studies, precise synthetic techniques have been developed for precisely doping the B-site metal ions, with multiple transition metals with varying oxidation states. This flexibility could allow for precisely modulating the catalytic properties.^{32,32,34,45–53}

In addition to catalytic activity, stability is a crucial factor in evaluating the efficiency of an electrocatalyst.^26–29 The perovskite oxide catalysts show a performance degradation often linked with the surface oxidation and aqueous dissolution.^9,54 To enhance the surface activity and stability under the OER environment, strategies such as doping with different transition metals have been employed for spinel and antispinel oxide surfaces.⁵⁵ Bimetallic and multi-metallic spinel oxide surfaces show multiple active sites with tunable oxidation states.^56–63 Furthermore, the presence of different metal centers with varying electronegativity helps in redistributing the metal–O electron density and improves the resistance to aqueous dissolution under various oxidative electrochemical conditions.^58,64–67

Recent studies have reported several oxide perovskite materials with comparable activity for OER as the Pt-group metal-based catalysts.^{32,32,50–53} The electrochemical stability of the perovskite oxides have been a major issue retarding the electrochemical activity for OER.^9,54 Understanding the aqueous dissolution pathways of the oxide perovskite surfaces will allow devising an effective approach to improve their electrocatalytic activity for OER and unleash a design principle. The OER activity of perovskite oxides is closely linked to their electronic structure and frontier energy states, pinning near the Fermi energy level.^{9,32,34,45,49–53,68} The frontier energy states could also regulate the surface reactivity with the aqueous environment. Therefore, a strategy to modulate the frontier energy states will lead to an optimum strategy to tune both the activity and stability of these surfaces. Inspired by this idea, in this study, we employ a B-site surface doping of the four perovskite oxides surfaces (SrMnO₃, SrFeO₃, SrNiO₃, and SrCoO₃) with different transition metal atoms (Mn, Fe, Ni, and Co).^46,69,70 Modulating the d-band center of the oxide surfaces, we tune the stability and electrocatalytic activity for the OER.⁷¹ The surface Pourbaix analysis has been performed to understand the surface aqueous dissolution reactions and stable surface configurations under the oxidative potentials and varying pH from acidic to alkaline domains. Doping with transition metals is found to enhance the stability and activity of the surfaces of both SrCoO₃ and SrNiO₃. We find that SrCoO₃ with Fe-dopant and SrNiO₃ with Mn-surface dopant, the OER overpotential is considerably reduced compared to the undoped surfaces.

Computational details

The computational calculations are performed with density functional theory (DFT), using the Vienna Ab initio Simulation Package (VASP).^72–74 The electron exchange–correlation effects are treated within the Generalized Gradient Approximation (GGA) using the revised Perdew–Burke–Ernzerhof (RPBE) functional.⁷⁵ The projector augmented-wave (PAW) method⁷⁶ is employed to describe the electron–ion interactions, and a plane-wave basis set with an energy cutoff of 400 eV is used. The Brillouin zone is sampled using a 2 × 2 × 1 Monkhorst–Pack k-point mesh.⁷⁷

The electronic and ionic convergence criteria are set to 10⁻⁶ eV for energy and 10⁻³ eV Å⁻¹ for forces, respectively. To account for non-covalent interactions between adsorbates and surfaces, van der Waals corrections are included. Solvent effects are incorporated using an implicit solvation model as implemented in the VASPSol package.^78–80 The implicit solvation model is employed to describe the electrochemical interface. In this setup, the perovskite surfaces and adsorbed intermediates are treated quantum mechanically, while the aqueous environment is represented by a polarizable continuum (PC) medium with a dielectric constant of 80. Electrolyte interactions are incorporated through a linearized Poisson–Boltzmann approach, with a Debye screening length of 3 Å. Non-electrostatic contributions to the solvation free energy arising from surface tension are excluded, as they are less significant in interfacial two-dimensional systems.^81,82 Consequently, the TAU parameter, which corresponds to the effective surface tension (eV Å⁻²) in the VASPsol package and accounts for cavity formation free energy, is neglected in the present calculations.

All calculations are performed with spin polarization to account for the unpaired electron and variable spin states of the surface systems. The Atomistic Simulation Environment (ASE) package is used for generating VASP input files and post-processing of the computational results.⁸³ We have considered a 2 × 2 supercell of the perovskite surfaces to carry out the reaction free energy calculations. On the other hand, for the surface Pourbaix analysis, we have considered both 2 × 2 and 2 × 3 supercells of the surfaces to obtain the different surface vacancy structures. In all the simulations, a four-layer surface has been taken, with the bottom two layers kept fixed to mimic the bulk configuration. The upper two layers, along with the adsorbates, are allowed to minimize to obtain the most preferred geometries. A gap of 20 Å is kept along the non-periodic direction to minimize the furious interactions between the adjacent periodic cells. A detailed explanation of the computational method for reaction free energy and surface Pourbaix energy calculations is given in Sections S-I and S-II of the SI, respectively.⁷¹

Results and discussion

We perform the dissolution free energy calculations of the four perovskite oxides (SrMnO₃, SrFeO₃, SrCoO₃, and SrNiO₃), varying the pH from the acidic to alkaline range and the potential in the oxidative region from 1 to 2.0 V vs. SHE to understand the dissolution mechanism.⁹ The corresponding stability is determined under varied electrochemical environments. In this study, we have considered the 001 surface termination with B-site ions on the surfaces. The 001 BO-site terminated (Fig. 1a) surfaces have been identified as the most stable surface terminations in the previous studies.^{13,39,84–86} Furthermore, the 001 surfaces have been analyzed and found to be most active for OER. Therefore, we have built our model surfaces using the 001 BO surface terminations for the stability and catalytic studies.

Fig. 1 (a) A schematic overview of the (001) surface termination of the pristine surface of oxide perovskite, where green spheres denote Sr, blue corresponds to the transition metal (Mn, Fe, Co, or Ni), and red spheres represent oxygen atoms. (b) The projected density of states (PDOS) plot for the (001) surfaces of SrMnO₃, SrFeO₃, SrCoO₃, and SrNiO₃. In these plots, the red regions correspond to the energy states of the O p-bands, while the blue regions represent the metal d-band states. The green horizontal line marks the O p band center with respect to the Fermi energy (black horizontal line).

Electrochemical stability of the pristine and doped 001 surface

Depending upon the positioning of the O-p band center with respect to the Fermi level (E_Fermi), the perovskite oxides show different dissolution pathways.⁹ Fig. 1b shows the projected density of states (PDOS) of the 001 surface of the four perovskite oxides. The red region shows the energy states of the O-p bands. The O-p band center (denoted by the green line) is positioned relatively away from the E_Fermi for all the surfaces (varying within the range 2.99 to 3.50 eV). Therefore, based on previous studies, the dissolution mechanism will primarily involve the leaching of surface B-site transition metal ions.⁵⁴ Furthermore, on the 001 surfaces, the transition metal ions are in direct contact with the aqueous medium compared to the cationic Sr²⁺ ions; therefore, leaching of the surface transition metals will be more feasible than the A-site ion leaching.

The dissolution free energies (ΔG_pbx) for the transition metal ions are computed considering the stable transition metal ion phase separation to metal oxides at different oxidation states, as shown in eqn (1) (Section S-II in the SI gives a detailed overview of the calculation method):

SrMO₃ + zyH₂O → SrM_1−yO₃ + yMO_z + 2zyH⁺ + 2zye⁻ (1)

Here, M corresponds to the transition metal ions, and y denotes the proportion of the metal ions leaching out from the surface. We have compared the stability of the surfaces with respect to the formation of different metal oxide solid phases (MO_z, where z determines the oxidation state of M in MO_z phases) after dissolution. The more negative the value of ΔG_pbx, the greater the possibility of dissolution. The ΔG_pbx values are calculated considering the surface vacancy formation of 17%, 25%, 33%, 50%, and 66%.

The ΔG_pbx for the four perovskite 001 surfaces is shown in Fig. 2(a)–(d) at three different pH levels. The y-axis corresponds to the ΔG_pbx values, and the x-axis denotes the variation in the oxidative potential with reference to SHE (V_SHE). The stable configurations are found to vary considerably depending on the potential ranges at each pH. The black vertical dotted lines give the potential ranges when we found a stability switching of the configuration on varying the pH values.

Fig. 2 (a)–(d) Illustrate the Pourbaix diagrams for SrMnO₃, SrFeO₃, SrCoO₃, and SrNiO₃, respectively, evaluated at pH 1, 7 and 14 under an applied potential range from 1.0 to 2.0 V. These diagrams include various corresponding metal oxides such as MO, MO₂, M₂O₃, and M₃O₄, and consider different metal vacancy compositions of 17%, 25%, 33%, 50% and 66%. Across all compositions and materials, the Pourbaix plots consistently denote enhanced thermodynamic stability associated with a 66%, 25% metal-site vacancy for SrMnO₃, 25% and 50% metal-site vacancy for SrFeO₃, 25%, and pristine surfaces for SrCoO₃ and SrNiO₃. The inset below shows the colour scheme considered to denote the different dissolution products.

On the SrMnO₃ 001 surface (Fig. 2(a)), at pH = 1, the dissolution of the surface Mn atoms up to the 25% surface vacancy formation is found to be most suitable up to oxidative potentials of 1.6 V_SHE. Beyond this potential, the surface leaching of 66% surface atoms is more suitable. Shifting to neutral pH, the most stable surface configuration corresponds to 25% surface vacancy up to 1.8 V_SHE. The formation of 66% surface vacancy is more preferred beyond that potential range. The alkaline pH shows a greater stability of the pristine surface at low oxidative potential (around 1.5 V_SHE), beyond that, the 25% vacancy formation shows more feasibility.

A differing stability trend is observed for the SrFeO₃ surface (Fig. 2(b)). The most stable surface configuration corresponds to the leaching of the surface Fe atoms to 25% at the potential range up to 1.6 V_SHE at pH = 1. The stability of the pristine surface increases in the neutral and alkaline regions at low oxidative potentials. The relative stability of the surfaces with 50% surface vacancy shows a greater preference with the gradual increase in the potential. However, the potential window defining the corresponding stability region for these vacancy formations considerably varies. At pH 7, the pristine surface shows the highest stability up to 1.4 V_SHE, whereas at pH 14, the pristine surface shows the highest stability up to 1.7 V_SHE. Beyond that, the 50% vacancy formation is more preferable at both pH values.

SrCoO₃ and SrNiO₃ show a similar stability ordering of the surfaces (Fig. 2(c) and (d), respectively). At pH 1, the pristine surface shows the highest stability at a lower potential range (up to 1.6 and 1.7 V_SHE for SrCoO₃ and SrNiO₃, respectively). An increase in the potential up to 2 V_SHE, the surface with 25% vacancy becomes the most stable surface configuration. At pH 7 and 14, we found that the pristine surface has the most negative ΔG_pbx for all potential ranges, corresponding to the most stable structural configuration for both SrCoO₃ and SrNiO₃.

Next, we study the effect of doping with the different transition metal atoms on these four perovskite surfaces. We have considered 25% surface doping and calculated the electrochemical stability of the doped surfaces. Mn surface positions in SrMnO₃ have been doped with Fe, Co, and Ni atoms. For SrFeO₃, the doped surfaces have been created by doping the surface Fe atom with Mn, Co, and Ni. On the other hand, Co and Ni atoms on SrCoO₃ and SrNiO₃ surfaces, respectively, are doped with Mn, Fe, and Ni dopants, and Mn, Fe, and Co dopants. The doped surfaces are denoted as M^*(M). Here, M^* corresponds to the dopant metal, and M denotes the host metal. For example, Fe-doped SrMnO₃ surface will be denoted as Fe(Mn).

Fig. 3 shows the two-dimensional stability plots of the doped perovskite surfaces. The Pourbaix energies are calculated considering the most stable surface configuration detected for each surface. In the presence of the surface dopants, the dissolution of the surface metal ions could follow three different pathways depending upon the type of metal leaching (eqn (2)–(4)). The formation of metal oxides from both the host (M) and dopant (M^*) metal atoms of the surface could lead to the generation of surface vacancies. On the other hand, either the host or the dopant metal atom could leach out and form a single phase of the metal oxide. Further details for the computations are provided in Section S-II in the SI.

Fig. 3 The Pourbaix plot with pH on the x-axis and potential on the y-axis. The right-hand scale corresponds to the evaluation of ΔG_pbx. The blue region indicates thermodynamic stability, while the yellow and red regions denote instability. From the plot, it is evident that only SrCoO₃ and SrNiO₃ exhibit a wide stable region, while SrFeO₃ shows a comparatively narrow stable domain. Doped surfaces are denoted as M^*(M), where M^* represents the dopant element, and M denotes the host B-site cation of the perovskite surface. Accordingly, Fe(Mn), Co(Mn), and Ni(Mn) correspond to Fe-, Co-, and Ni-doped SrMnO₃ surfaces, respectively. A similar nomenclature is adopted for the remaining doped surfaces.

On doping the different transition metal dopants, both SrMnO₃ and SrFeO₃, show considerable negative free energies for the creation of the surface vacancies. Both these surfaces will lead to the dissolution of the surface metal atoms. For SrFeO₃, we find a narrow stability region. For Mn- and Co-doped SrFeO₃, the stability region corresponds to a pH value varying between 4.5 and 14, and the potential range between 1 and 1.4 V_SHE. With Ni as the surface dopant, the stability region corresponds to 3.9 to 14 pH and potential 1 to 1.6 V_SHE. On the other hand, SrNiO₃ and SrCoO₃ show a considerable range of stable regions. At acidic pH, the stable region corresponds to lower oxidative potentials. However, at neutral and alkaline pH, the stability region corresponds to both lower and higher oxidative potentials.

To understand the differing stability trend, we analyze the electronic d-band centers of the surface host metals (Fig. 4(a)). For the undoped surfaces, a shift in the d-band center to more negative energies is observed on moving from SrMnO₃ to SrFeO₃ to SrCoO₃ to SrNiO₃. The shaded bars show the d-band center values of the surfaces without the dopant in Fig. 4a. The d-band center for the Mn in SrMnO₃ is −0.59 eV from the Fermi energy, whereas the corresponding values are −1.66 eV for SrFeO₃, −1.75 eV for SrCoO₃, and −2.07 eV for SrNiO₃. A shift in the d-band centers to more negative energies implies lower reactivity of the surfaces with the electrolyte and aqueous molecules. This results in an increasing electrochemical stability of the pristine surfaces across a wide range of pH and potential values of the undoped surfaces from SrMno₃ to SrFeO₃ to SrCoO₃ to SrNiO₃ (Fig. 2a–(d)).

Fig. 4 (a) The electronic d-band centers in eV of surface host atoms in SrMnO₃, SrFeO₃, SrCoO₃, and SrNiO₃, respectively, (from left to right) without and with the different surface dopants. The x-axis represents the different materials, while the y-axis indicates the average d-band center. Shaded bars correspond to the d-band center values of the undoped surfaces. The d-band center values are given relative to the Fermi energy level. (b) The plot of free energy at 25% surface vacancy (along x-axis) on the pristine surfaces with the d-band center (y axis). A linear correlation is obtained with R² for the linear fit of 0.75.

To further elucidate the effect of d-band center positioning with respect to the Fermi level in governing surface stability, we compare the d-band center values with the free energies associated with the formation of 25% surface vacancies on the pristine surfaces. The tendency for surface vacancy formation varies significantly depending on the surface structure, solution pH, and applied electrochemical potential. The direct comparison of individual key dissolution steps across different surfaces is challenging. Nevertheless, our analysis shows that the formation of 25% surface vacancies is thermodynamically feasible for all the surfaces considered. Therefore, we compare the dissolution free energies associated with the formation of 25% surface vacancies with the corresponding d-band center positions. Fig. 4(b) presents this comparative analysis. The resulting linear correlation indicates that as the d-band center shifts to more negative energies, the dissolution process becomes increasingly endothermic. This trend demonstrates that the position of the d-band center plays a crucial role in determining surface stability. Specifically, a less negative d-band center strengthens interactions with solvent molecules, thereby promoting surface dissolution.

A comparison of the average d-band centers of the host metal centers on the doped surfaces (Fig. 4a) shows great similarity to the undoped surfaces. For SrMnO₃, the d-band center values vary within the range −0.54 to −0.81 eV. The doping with Co slightly reduces the d-band center values to the more negative side (−0.81 eV). With SrFeO₃, the d-band center values marginally vary in the presence of the different dopants. The values correspond to −1.60 for Mn(Fe), −1.58 eV for Co(Fe), and −1.49 eV for Ni(Fe). Both the SrCoO₃ and SrNiO₃ doped surfaces show the highest negative values for the d-band centers. The corresponding values vary within the range −1.89 to −2.22 eV for SrCoO₃, and −1.90 eV to −2.18 eV for SrNiO₃. In the case of SrCoO₃, the d-band center values for the doped surfaces are slightly more negative than the undoped surfaces, with the highest negative value (−2.22 eV) obtained for Mn(Co). On the other hand, for SrNiO₃, the d-band center value for Mn(Ni) is only greater than the undoped surface. For both Fe(Ni) and Co(Ni), the d-band center value becomes less negative (−1.93 eV and −1.90 eV, respectively) compared to the undoped surface.

The d-band center of the B-site metal ions has been known to modulate the catalytic activity and choice of the particular reaction pathway. In this study, we find that the relative stability of the perovskite surfaces could also be governed by the alignment of the d-band center relative to the Fermi energy level. A strong correlation between surface dissolution free energy values and the position of the d-band center has been obtained (Fig. 4b). Both SrCoO₃ and SrNiO₃ show good stability in both doped and undoped forms due to the deeper positioning of the d-band center with reference to the Fermi energy level.

Doping influences the catalytic activity of the perovskite surfaces

Having determined the electrochemical stability of the surfaces, we analyze the catalytic activity of the pristine and doped surfaces for OER. The stability and surface configurations of the surfaces will largely modulate the catalytic activity. On the perovskite oxide surfaces, among the different mechanisms, the adsorbate evolution mechanism (AEM) and lattice oxygen mechanism (LOM) compete during the oxidation of water to produce O₂.^13,32 Fig. 5a shows the two mechanisms. AEM (black pathway in Fig. 5a) involves the adsorption of the oxygenated intermediates at the surface metal sites. The evolution of O₂ occurs via the formation of the different intermediates OH*, O*, and OOH*. In contrast, the LOM pathway (red pathway in Fig. 5a) enables lattice oxygen atoms to directly participate in O–O bond formation. The strength of the lattice O bond with the surface transition metal atoms determines the reaction thermodynamics and the overpotential to drive the oxidation reaction of H₂O to molecular O₂. The evolution of the O₂ creates a vacancy on the surface (vacO₂, Fig. 5a). In the next step, the adsorption of H₂O leads to the formation of adsorbed OH* at the vacancy site (vacOH* in Fig. 5a, red pathway). The removal of a proton and an electron from vacOH* regenerates the catalytic site in the fourth electrochemical step.

Fig. 5 (a) The adsorbate evolution mechanism (AEM) and the lattice oxygen mechanism (LOM) are shown. The black pathway represents the AEM route, while the red pathway corresponds to the LOM route. The proton-coupled-electron transfer is shown in each step, and the free energy changes are denoted as ΔG_i values. (b)–(c) The lower panel presents the crystal orbital Hamilton population (COHP) plots and the corresponding integrated crystal orbital Hamilton population (ICOHP) values for metal–O* bond on undoped SrCoO₃ and SrNiO₃ (001) surfaces, respectively. The black vertical line distinguishes bonding from antibonding states; the energy levels on the left of this line correspond to antibonding states, while those on the right represent bonding states. The black horizontal line indicates the Fermi energy.

The reaction free energies of the key OER intermediates on the pristine surfaces of the four perovskite materials are computed to elucidate the relative ordering of catalytic activity and the preferred reaction pathways. The reaction free energy values are calculated using the computational hydrogen electrode method.¹² The computational details are provided in Section S-I in the SI. Fig. S1 presents the free-energy diagrams for the pristine surfaces along their most favorable reaction pathways, while Table ST1 summarizes the corresponding reaction free energies and overpotential values. SrMnO₃ and SrFeO₃ favor the AEM, with calculated overpotentials (η_TD) of 0.94 and 0.7 V, respectively.

On pristine SrCoO₃ and SrNiO₃ surface sites, the AEM is found to have higher η_TD compared to LOM. For LOM, the corresponding η_TD values are 0.51 and 1.06 V, respectively. On the other hand, for AEM, the η_TD values are 0.66 V and 1.16 V (Table ST6, SI), respectively. The formation of the OOH* intermediate shows the highest endothermicity on SrCoO₃ and corresponds to the PDS for the AEM pathway. On the SrNiO₃ surface, the second step, formation of O* intermediate, corresponds to the PDS.

The selectivity of the mechanistic pathways for OER on pristine SrCoO₃ and SrNiO₃ surfaces is in good agreement with previous studies.^{13,16,87–89} The binding strength of the O* intermediate is the determining factor, responsible for the preferential selectivity of the LOM vs. AEM mechanism. To understand the preferences in the mechanistic pathway, we analyse the metal–O bond strength of the O* intermediate and plot the crystal orbital Hamilton population (COHP) density in the bonding and antibonding metal–O bands.⁹⁰ Fig. 5b and c show the COHP plots for the O* intermediates on SrCoO₃ and SrNiO₃. For SrMnO₃ and SrFeO₃ surfaces, the corresponding COHP plots are given in Fig. S3 in the SI. The vertical line in black differentiates the bonding and antibonding energy states; on the left-hand side of each plot, the antibonding energy states are shown, whereas the bonding energy states are on the right-hand side of the vertical line. The black horizontal line denotes the Fermi energy.

Analysing the COHP plots, we find that for the O* intermediate on both SrCoO₃ and SrNiO₃ surfaces, the antibonding metal–O* bands are filled up compared to SrMnO₃ and SrFeO₃ surfaces. This results in a slight elongation of the metal–O* bond, thereby enhancing the possibility of bonding with the lattice O atoms. The occupancy of the antibonding bands destabilizes the O* intermediate in comparison to the , resulting in the preference for LOM (Fig. S3 in the SI). The corresponding antibonding states are less occupied for SrMnO₃, and SrFeO₃ surfaces, and the integrated crystal orbital Hamilton population (ICOHP) values in Fig. S3 (SI) are greater than the values shown in Fig. 5(b) and (c). This indicates greater metal–O* bond strength on SrMnO₃ and SrFeO₃ 001 surfaces and preference for AEM over LOM.¹³

The relative trend in the catalytic activity of the pristine surfaces shows good coherence with previous experimental and theoretical reports.^{13,16,87–89} Comparing the catalytic activity of SrMnO₃, SrFeO₃, SrCoO₃, and SrNiO₃ (Table ST1 in SI), we find that the pristine SrCoO₃ surface is most active for OER.

Since SrCoO₃ and SrNiO₃ show the highest stability, in the proceeding discussion, we will only consider these surfaces with and without the surface dopants to study the electrocatalytic activity. The reaction free energy values for the AEM and LOM pathways on the doped surfaces for SrCoO₃ and SrNiO₃ are shown in Table 1. For comparison, we have given the corresponding values for the undoped surfaces of SrCoO₃ and SrNiO₃. The potential determining steps (PDS) are denoted in bold, and overpotential (η_TD) values are in column 8 in Table 1. For each surface, the catalytic site with the lowest η_TD values is given in Table 1. We denote the active surface sites using the nomenclature M_1M2(Co) and M_1M2(Ni), where M₁ corresponds to the dopant, and M₂ denotes the active site among the dopant or the host. Within the parentheses, we write the host metal atom. The Co_Co(Co) and Ni_Ni(Ni) for SrCoO₃ and SrNiO₃ denotes the undoped surfaces. The reaction free energies for the other sites with higher η_TD values are given in Table ST6 in the SI.

Table 1 Reaction free energies (ΔG) along the preferred OER pathway (AEM/LOM) for the undoped and the doped SrCoO₃ and SrNiO₃ perovskite surfaces. The active sites are shown in Column 2. The ΔG values are shown in Columns 3–6. The potential-determining step (largest ΔG) is highlighted in bold. Column 7 shows the G_max values and the corresponding reaction step as RDS. The corresponding η_TD values and preferred mechanism are denoted in columns 8 and 9, respectively

Material	DopantActive site(host)	ΔG1				Gmax	ηTD	Mechanism
SrCoO3
Undoped	CoCo(Co)	1.74	0.71	1.15	1.33	0.21(ΔG1)	0.51	LOM
Mn-doped	MnMn(Co)	2.08	0.94	1.04	0.86	0.55(ΔG1)	0.85	LOM
Fe-doped	FeFe(Co)	1.25	1.47	1.30	0.90		0.24	LOM
Ni-doped	NiNi(Co)	1.36	0.54	1.64	1.39		0.41	LOM
SrNiO3
Undoped	NiNi(Ni)	1.31	0.19	1.13	2.29		1.06	LOM
Mn-doped	MnNi(Ni)	1.55	0.47	1.13	1.57	0.76(ΔG1)	0.34	LOM
Fe-doped	FeFe(Ni)	1.74	1.62	1.77	−0.21	0.54(ΔG1 + ΔG2 + ΔG3)	0.54	AEM
Co-doped	CoCo(Ni)	1.84	1.55	1.79	−0.25	0.59(ΔG1 + ΔG2 + ΔG3)	0.61	AEM

---

When considering the LOM pathway, we analyse oxygen vacancy formation at all unique oxygen sites surrounding both the host and dopant metal atoms. For the dopant, we identify a single unique oxygen site bridging the host and dopant metal centers. On the otherhand, two distinct lattice oxygen sites associated with the host metal participate in vacancy formation: the oxygen bridging two host metal atoms and the oxygen bridging the host and dopant metal centers (Fig. S2, SI). We have considered all these lattice O to evaluate the minimum energy pathways. In Table ST6 (SI), reactions at the host metal site are distinguished based on the participating lattice oxygen, denoted as [O_H] and [O_D] in column 2. [O_H] corresponds to the oxygen bridging two host metal atoms, whereas [O_D] corresponds to the oxygen bridging the host and dopant metal atoms.

A considerable reduction in the η_TD values has been observed with the surface doping. On SrCoO₃ surfaces, we find that for all the doped surfaces, the active site with the lowest η_TD values corresponds to the doped sites. LOM shows a slightly greater preference compared to the AEM with all the surface dopants. The corresponding to the formation of the intermediate has a lower thermodynamic barrier compared to O* formation. This favors the LOM pathway.

The PDS shows a wide variation depending on the active sites and the surface dopants. The Mn_Mn(Co) show the formation of OH* as having the highest endothermicity. For Fe_Fe(Co), the formation of (step 3 for LOM, Fig. 5a) is the PDS, whereas the Ni_Ni(Co) site shows the highest endothermicity for the formation of the vacOH* intermediate.

The SrNiO₃ surface with the different surface dopants shows a differing trend. Unlike the Ni_Ni site (undoped surface), the AEM is found to be the most preferred pathway with Fe and Co surface doping (Fe_Fe(Ni) and Co_Co(Ni)). Though formation shows more preference than O* formation, on these sites, the regeneration of the catalytic site (the fourth step, ΔG*) has been found with high endothermicity while following the LOM.

Compared to the undoped surface (Ni_Ni(Ni)), both Fe_Fe(Ni) and Co_Co(Ni) sites show an increment in the catalytic activity. The PDS for Fe_Fe(Ni) corresponds to the formation of the vacOH* intermediate with η_TD value of 0.54 V. In contrast, on the Co_Co(Ni) site, the PDS corresponds to the formation of the OH* intermediate. The corresponding η_TD value is 0.61 V. With Mn-doping, we have seen the highest catalytic activity. On both Mn_Ni(Ni) and Mn_Mn(Ni) sites, the η_TD values are considerably lower than the Ni_Ni site (Table 1 and ST6 in the SI). Both sites show higher preference for LOM. The most active site is Mn_Ni (Ni), with a PDS corresponding to (η_TD value of 0.34 V).

Surface oxygen vacancy formation is often associated with significant structural distortion and, in some cases, surface collapse of perovskite oxides. Previous studies have shown that lattice oxygen participation in the oxygen evolution reaction, accompanied by the formation of oxygen vacancies, can accelerate surface degradation and metal dissolution.^91–94

In the present study, however, we do not observe such structural instability for either SrCoO₃ or SrNiO₃ (with or without surface dopants) upon surface O-vacancy formation. Fully relaxed surface geometries exhibit only minimal distortions following lattice oxygen participation in O₂ evolution (Fig. S7 in SI). These results indicate that both SrCoO₃ and SrNiO₃ surfaces maintain structural stability under the considered reaction conditions.

Fig. 6(a) and (b) show the comparison of the reaction free energies on the doped surfaces with the undoped surfaces. The ideal catalysts, with ΔG values of 1.23 eV for each step, are also plotted in these diagrams (black lines). Compared to the undoped surfaces, a considerable reduction in the ΔG value of the PDS is observed for the most active doped surfaces.

Fig. 6 (a) and (b) Free energy diagrams (FEDs) for the most favorable reaction pathways on various doped perovskite surfaces of SrCoO₃ and SrNiO₃. The black lines represent an ideal catalyst surface, with each reaction step assigned an energy of 1.23 eV. The FEDs in purple, red, blue, and green correspond to Mn-, Fe-, Co-, and Ni-doped surfaces, respectively. (a) shows the FEDs for SrCoO₃ with different dopants, and (b) shows those for SrNiO₃ with different dopants. (c) and (d) The ICOHP values for undoped SrCoO₃ and Fe-doped SrCoO₃, respectively. (e) and (f) Show the ICOHP values for undoped SrNiO₃ and Mn-doped SrNiO₃.

The difference between the center of the d-band and p-band has been known to influence the preference of the mechanism in the OER.^{32,52,53,95–97} The decrease in the gap between the d-band and p-band center promotes the increase in the covalency of the M–O bond and could enhance the preference for LOM. In this study, we analyze the d-band centers of the host (M_nd) and dopant (D_nd) B-site ions and determine the difference with the O p-band centers of the different types of lattice O available in the surroundings of these metal ions. Fig. S6 shows the different O types present surrounding the dopant and the host metal center, and Table ST7 in the SI gives the differences in the band centers. For most of the cases, the differences in the d-band and p-band centers are minimal except for SrCoO₃ with the Mn dopant. We have obtained that formation of is more favorable than that of the O* intermediate, corresponding to favoring the lattice O participating in the reaction. However, in most of the cases, formation does not correspond to the potential determining step (PDS). Therefore, the specificity towards the preference of the mechanism is not solely dependent on the values.

To understand the differing trends in the reactivity of the doped surfaces, we analyze the electronic structure of the surfaces. For SrCoO₃, Mn-doping leads to stabilizing the d-bands of the active site to the most negative values (−2.22 eV). This results in the lowest catalytic activity and the highest η_TD values compared to the undoped surface. With the Fe doping to SrCoO₃ surface, the catalytic activity is enhanced the most. The Fe site exhibits the least negative value for the d-band center (−1.46 eV), resulting in the greatest reduction in the endothermicity of the OH* intermediate. This is the possible reason for enhancing the catalytic activity of this site. The Fe_Co(Co) site, though, has an almost similar d-band center as the undoped surface (−1.88 eV and −1.76 eV for Fe_Co(Co) and Co_Co, respectively); nevertheless, an increase in the activity is observed. We find considerable surface structure distortion on Fe doping (Fig. 6d). An increase in the binding strength of Fe with lattice O is observed. The Co–O_lattice has been elongated compared to the undoped surface. We have calculated the ICOHP values to determine the metal–O_lattice bond strength, as shown in Fig. 6(d) and compared to the undoped surface (Fig. 6c). The corresponding ICOHP values are lower than those of the undoped surface, reflecting the lowering of the bond strength. The structural distortions allow the creation of dangling valences on the surface metal sites and increase its activity for OER. An almost similar activity is obtained for the Co_Fe(Co) and Fe_Fe(Co) sites. For the other doped surfaces, we have not observed a significant deviation of the ICOHP values in comparison to the undoped surfaces. Table ST8 in the SI shows these ICOHP values.

Similar structural distortions and alignment of the d-band centers of the SrNiO₃ surface, on doping with the different transition metals, have been found to tune the catalytic activity. While following the LOM mechanism, we find that the vacOH* intermediate shows extensive stability on both the undoped and doped surfaces, except for Mn_Ni(Ni) and Mn_Mn(Ni) sites. This increases the endothermicity of (corresponding to catalytic site regeneration), leading to an enhancement of the η_TD value. Therefore, to enhance the activity of these sites, it is important to reduce the binding affinity of the H to the lattice O site. Therefore, the lattice O–metal bonds need to be strengthened to reduce the endothermicity of the fourth step. Stronger binding of lattice O to the metal atoms will reduce the O–H bond strength in vacOH* and hence the endothermicity of .

The ICOHP values of the lattice O with the metal centers increase on doping with Fe, Co, and Mn in comparison to the undoped SrNiO₃ surface. A maximum increment in the binding strength of the lattice O is observed for the Mn-doped surface. Therefore, for Mn_Ni(Ni) and Mn_Mn(Ni) sites, formation of vacOH* shows higher endothermicity than other doped surfaces. A decrease in binding strength has been observed, providing an advantage for these sites. The thermodynamic barrier of the fourth step is reduced, thereby the η_TD values have been reduced to a great extent for Mn_Ni(Ni) and Mn_Mn(Ni) sites. Fig. 6(f) shows the corresponding ICOHP values for lattice O bonded to Mn_Ni(Ni) and Mn_Mn(Ni) sites. For the other surfaces (doped and undoped SrNiO₃ surfaces), the ICOHP values are shown in Table ST8 in the SI.

To understand the effect of dopant concentration on the OER activity, we perform a systematic activity analysis of the most active doped surface: Fe-doped SrCoO₃ surfaces with Fe concentrations of 17%, 25%, 33%, and 50%. The 25% doping level is achieved by substituting one surface Co atom in a 2 × 2 (001) surface supercell of SrCoO₃. The other doping concentrations 17%, 33%, and 50% are obtained by substituting one, two, and three surface Co atoms, respectively, in a 2 × 3 surface supercell of SrCoO₃.

We have considered both the host and dopant atoms as potential active sites and compared the AEM and LOM pathways across different dopant concentrations. Fig. 7 shows a comparison of the theoretical η_TD values as a function of Fe doping percentage. In all cases, the LOM pathway exhibits lower η_TD values than AEM. At a Fe doping concentration of 17%, the Fe_Co(Co) site exhibits the lowest η_TD. However, as the dopant concentration increases, the Fe_Fe(Co) site becomes the most favorable active site for OER. Among all doping percentages studied, the 25% Fe-doped surface shows the lowest η_TD values for both Fe_Co(Co) and Fe_Fe(Co) sites. Although the difference in the corresponding η_TD values is minimal at the 25% surface doping, the Fe_Fe(Co) site is slightly more favorable.

Fig. 7 The OER activity analysis of the effect of surface Fe concentration on 001 SrCoO₃ surfaces. The considered surface doping concentrations are 17%, 25%, 33%, and 50%. 0% corresponds to the undoped SrCoO₃ surface. The Fe_Fe(Co) site is denoted in blue circular markers, whereas for the Fe_Co(Co) site, orange square markers are used. A maximum activity is observed for the 25% surface doping.

Recent studies have shown that B-site doping of perovskite surfaces can effectively modulate the positions of the O-p band and metal d-band centers. By varying the Mn doping concentration on the SrCoO₃ surface, the electronic structure can be systematically tuned for different applications.⁹⁸ More prominently, Fe doping of SrCoO₃ surfaces has been reported to significantly enhance the O₂ evolution reaction (OER) activity.⁹⁹ Several Sr–Fe–Co–O perovskite compositions have been synthesized by varying the Fe/Co ratio. Sr₂Co_1.5Fe_0.5O_6−δ exhibits the highest OER performance, with an overpotential of 318 mV at 10 mA cm⁻² and a Tafel slope of 44.8 mV dec⁻¹.⁹⁹ These experimental findings are in good agreement with our computational results. In particular, we find that maintaining the Fe dopant concentration at approximately 25% yields the maximum O₂ evolution activity on the Fe-doped SrCoO₃ surface (Fig. 7). Among the different active sites, Fe_Fe(Co) shows the highest activity. Although doping of the other perovskite surfaces has not yet been reported for OER application, our results demonstrate that tuning the dopant concentration on SrNiO₃ and SrFeO₃ surfaces can significantly enhance their catalytic activity for the oxygen evolution reaction.

The kinetics of the elementary electrochemical steps are important in defining catalytic activity. η_TD is evaluated at zero applied potential and corresponds to the elementary step with the largest thermodynamic free energy. However, in multistep electrochemical pathways, this step is not always the rate-determining step (RDS), particularly in cases where the potential-determining step (PDS) differs from the RDS.¹⁰⁰

To address this, Exner and co-workers introduced a potential-dependent reactivity descriptor, G_max(η), which directly identifies the RDS under an applied potential.¹⁰⁰ The descriptor is derived from the free energy changes of the elementary steps (ΔG_i, i = 1, 2, 3, 4) under an applied overpotential. Unlike η_TD, G_max(η) scales linearly with the free energy of the RDS, providing a more accurate reflection of catalytic activity.^101–103

For OER catalysts, values of G_max(η) less than or equal to 1.0 eV at η = 0.3 V are typically associated with good catalytic performance.¹⁰¹ Accordingly, we have used this criterion in our analysis (Table 1) to identify the RDS and evaluate the efficiency of various active sites with different dopants. The specific ΔG_i values that govern the RDS are given in column 7.

The trend in G_max values indicates that, for the majority of active sites obtained in our study, the RDS is governed by a single elementary step. However, the corresponding step determining the RDS shows wide variation depending upon the active sites. The Co_Co(Co), Mn_Mn(Co) sites show ΔG₁ as the RDS. On the other hand, are found to be the RDS for Fe_Fe(Co), Ni_Ni(Co) and Ni_Ni(Ni) respectively. This suggests that the reaction kinetics in these sites are largely dictated by the energetics of the specific step with the largest G_max values.

In contrast, a different pattern emerges for certain dopants, most notably Fe and Co doped to the SrNiO₃ surfaces, where G_max is not confined to a single ΔG_i value. These surfaces instead show the cumulative contribution of multiple steps (e.g., ΔG₁ + ΔG₂ + ΔG₃). The kinetic limitation is distributed over several transitions rather than a localized single barrier on these surface sites. The multistep character in defining the RDS reflects a more complex kinetic landscape, where the reaction energetics of multiple intermediates determine the catalytic activity.

To assess the experimental viability of the doped surfaces, we have calculated the formation energies (E_form) for all perovskites with the different surface dopants using bulk metals as references for the dopants and the pristine surface as the reference for the host perovskite (Fig. S4 in the SI). In Section S-V in the SI, the details for the formation energy of the doped surfaces are shown.

A clear trend is observed in which the E_form becomes progressively less favorable when moving from early to late transition-metal dopants. While all dopants on SrMnO₃ surfaces show endothermic E_form, SrFeO₃ shows exothermic stabilization only for Mn doping. In contrast, most dopants with SrCoO₃ and SrNiO₃ surfaces exhibit exothermic formation energies, indicating enhanced thermodynamic stability. Notably, Fe-doped SrCoO₃ and Mn-doped SrNiO₃ surfaces, which exhibit good catalytic activity, display exothermic formation energies. Therefore, these compositions are experimentally viable.

Conclusions

In this study, we have analyzed the catalytic activity and reaction mechanism for OER on four perovskite oxide surfaces, namely SrMnO₃, SrFeO₃, SrCoO₃, and SrNiO₃. The electrochemical stability and catalytic activity for OER have been analyzed, focusing on the (001) surface termination of the pristine surface, and doped with different transition metal ions at the B-site. We have calculated the dissolution free energies across a wide pH range (acidic to alkaline) and oxidative potentials of 1–2 V vs. SHE to determine the electrochemical stability under these varying conditions. PDOS analysis of the (001) surfaces shows the O-p band center located 2.99–3.50 eV below E_Fermi, indicating that dissolution mainly proceeds through leaching of surface B-site transition metal ions.

Dissolution free energies (ΔG_pbx) are computed for surface vacancy creation of 17–66%. For SrMnO₃, Mn dissolution is most favorable (up to 66%) in acidic media, while neutral and alkaline pH favor only 25% or pristine surfaces depending on potential. SrFeO₃ surfaces are most stable with 50% vacancy formation at acidic pH, whereas pristine surfaces dominate stability at neutral and alkaline pH under low potentials. In SrCoO₃ and SrNiO₃, pristine surfaces are most stable at low potentials across pH 1–14. At higher oxidation potential and acidic pH, 25% vacancy formation is found to be possible. However, at neutral and alkaline pH, the pristine surfaces show extensive stability for the entire potential range.

With 25% surface doping of transition metals (Mn, Fe, Co, and Ni) wide variation in the electrocatalytic activity and stability is observed. Pourbaix free energy (ΔG_pbx) calculations of these doped systems reveal many dissolution pathways depending on whether the host, dopant, or both ions leach. SrMnO₃ do not show stability regions with any of the surface dopants. SrFeO₃ show a narrow stable region at the alkaline pH and low oxidative potential. However, SrCoO₃ and SrNiO₃ show stable regions at the neutral to alkaline pH. The positioning of the metal d-band centers with respect to the Fermi energy strongly influences the stability of the surfaces.

Next, we analyze the catalytic activity of pristine and doped SrCoO₃ and SrNiO₃ surfaces. OER is found to occur via two competing pathways: Adsorbate Evolution Mechanism (AEM) and the Lattice Oxygen Mechanism (LOM). The calculations show that undoped SrCoO₃ and SrNiO₃ favor LOM with η_TD values of 0.51 and 1.06 V, respectively. Analysis of COHP plots reveals progressive weakening of metal–O bonds from SrCoO₃ to SrNiO₃, promoting LOM. Surface doping further tunes catalytic activity: Fe-doping in SrCoO₃ reduces η_TD to as low as 0.24 V at the Fe_Fe(Co) site. In SrNiO₃, Mn-doping significantly lowers η_TD to 0.34 V at Mn_Ni(Ni) site. We found that surface distortion and the alignment of d-band center of the active sites considerably determine the relative activity of the surface sites. On varying the Fe-dopant percentages on SrCoO₃ surface, we find that 25% surface doping shows the highest catalytic activity for OER.

The activity descriptor G_max(η) at the applied η of 0.3 V is used to analyse the rate-determining step (RDS). While most active sites exhibit a single step as the RDS, Fe- and Co-doped SrNiO₃ display a multistep process for determining the reaction rate. The design principle shown in this study to tune the catalytic activity and stability of perovskite oxide surfaces by surface doping will be of great importance for efficient catalyst development for OER.

Author contributions

S. S.: data curation, formal analysis, investigation, methodology, and visualization. N. K.: conceptualization, funding acquisition, methodology, project administration, resources, software, supervision, and visualization. Both authors contributed to writing the original draft, reviewing, and editing.

Conflicts of interest

The authors declare no conflicts of interest.

Data availability

The data related to computations will be available from the authors. Most of the data used to set up the figures has been provided in the manuscript and supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d5ta07376f.

Acknowledgements

N. K. and S. S. acknowledge the Shiv Nadar Institution of Eminence and Shiv Nadar Foundation for financial assistance and computational support. The calculations are performed at the supercomputer (Magus).

Notes and references

1. Z. W. Seh, J. Kibsgaard, C. F. Dickens, I. Chorkendorff, J. K. Nørskov and T. F. Jaramillo, Science, 2017, 355, eaad4998.
2. Y. Zhu, W. Zhou and Z. Shao, Small, 2017, 13, 1603793.
3. L. Han, S. Dong and E. Wang, Adv. Mater., 2016, 28, 9266–9291.
4. B.-J. Kim, D. F. Abbott, X. Cheng, E. Fabbri, M. Nachtegaal, F. Bozza, I. E. Castelli, D. Lebedev, R. Schaublin and C. Copéret, et al., ACS Catal., 2017, 7, 3245–3256.
5. R. S. Haszeldine, Science, 2009, 325, 1647–1652.
6. N.-T. Suen, S.-F. Hung, Q. Quan, N. Zhang, Y.-J. Xu and H. M. Chen, Chem. Soc. Rev., 2017, 46, 337–365.
7. Y. Jin, H. Wang, J. Li, X. Yue, Y. Han, P. K. Shen and Y. Cui, Adv. Mater., 2016, 28, 3785–3790.
8. Y. Zheng, Y. Jiao, Y. Zhu, L. H. Li, Y. Han, Y. Chen, M. Jaroniec and S.-Z. Qiao, J. Am. Chem. Soc., 2016, 138, 16174–16181.
9. B. Han, M. Risch, Y.-L. Lee, C. Ling, H. Jia and Y. Shao-Horn, Phys. Chem. Chem. Phys., 2015, 17, 22576–22580.
10. O. Diaz-Morales, I. Ledezma-Yanez, M. T. Koper and F. Calle-Vallejo, ACS Catal., 2015, 5, 5380–5387.
11. W. Li, H. Tian, L. Ma, Y. Wang, X. Liu and X. Gao, Mater. Adv., 2022, 3, 5598–5644.
12. J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard and H. Jonsson, J. Phys. Chem. B, 2004, 108, 17886–17892.
13. J. S. Yoo, X. Rong, Y. Liu and A. M. Kolpak, ACS Catal., 2018, 8, 4628–4636.
14. H. Li, Y. Chen, J. Z. Y. Seow, C. Liu, A. C. Fisher, J. W. Ager and Z. J. Xu, Small Sci., 2022, 2, 2100048.
15. Y. Zhao, D. P. Adiyeri Saseendran, C. Huang, C. A. Triana, W. R. Marks, H. Chen, H. Zhao and G. R. Patzke, Chem. Rev., 2023, 123, 6257–6358.
16. I. C. Man, H.-Y. Su, F. Calle-Vallejo, H. A. Hansen, J. I. Martínez, N. G. Inoglu, J. Kitchin, T. F. Jaramillo, J. K. Nørskov and J. Rossmeisl, ChemCatChem, 2011, 3, 1159–1165.
17. K. S. Exner, Acc. Chem. Res., 2024, 57, 1336–1345.
18. S. Singh and N. Karmodak, ACS Appl. Energy Mater., 2024, 7, 7854–7863.
19. M. Sokolov and K. S. Exner, Chem Catal., 2024, 4, year.
20. N. Karmodak and O. Andreussi, ACS Energy Lett., 2020, 5, 885–891.
21. N. Danilovic, R. Subbaraman, K.-C. Chang, S. H. Chang, Y. J. Kang, J. Snyder, A. P. Paulikas, D. Strmcnik, Y.-T. Kim and D. Myers, et al., J. Phys. Chem. Lett., 2014, 5, 2474–2478.
22. J. Rossmeisl, A. Logadottir and J. K. Nørskov, Chem. Phys., 2005, 319, 178–184.
23. S. Cherevko, S. Geiger, O. Kasian, N. Kulyk, J.-P. Grote, A. Savan, B. R. Shrestha, S. Merzlikin, B. Breitbach and A. Ludwig, et al., Catal. Today, 2016, 262, 170–180.
24. Y. Zhu, W. Zhou, J. Yu, Y. Chen, M. Liu and Z. Shao, Chem. Mater., 2016, 28, 1691–1697.
25. G. K. K. Gunasooriya and J. K. Nørskov, ACS Energy Lett., 2020, 5, 3778–3787.
26. R. Frydendal, E. A. Paoli, B. P. Knudsen, B. Wickman, P. Malacrida, I. E. Stephens and I. Chorkendorff, ChemElectroChem, 2014, 1, 2075–2081.
27. N. Hodnik, P. Jovanovic, A. Pavlisic, B. Jozinovic, M. Zorko, M. Bele, V. S. Selih, M. Sala, S. Hocevar and M. Gaberscek, J. Phys. Chem. C, 2015, 119, 10140–10147.
28. K. S. Exner, ChemElectroChem, 2021, 8, 46–48.
29. V. M, S. Singh, F. Bononi, O. Andreussi and N. Karmodak, J. Chem. Phys., 2023, 159, 111001.
30. D. A. Kuznetsov, B. Han, Y. Yu, R. R. Rao, J. Hwang, Y. Román-Leshkov and Y. Shao-Horn, Joule, 2018, 2, 225–244.
31. N. Karmodak, L. Bursi and O. Andreussi, J. Phys. Chem. Lett., 2021, 13, 58–65.
32. A. Grimaud, O. Diaz-Morales, B. Han, W. T. Hong, Y.-L. Lee, L. Giordano, K. A. Stoerzinger, M. T. Koper and Y. Shao-Horn, Nat. Chem., 2017, 9, 457–465.
33. Z. Wang, Y.-R. Zheng, I. Chorkendorff and J. K. Nørskov, ACS Energy Lett., 2020, 5, 2905–2908.
34. J. Xu, C. Chen, Z. Han, Y. Yang, J. Li and Q. Deng, Nanomaterials, 2019, 9, 1161.
35. M. De Graef and M. E. McHenry, Structure of Materials: an Introduction to Crystallography, Diffraction and Symmetry, Cambridge University Press, 2012.
36. J. Hwang, R. R. Rao, L. Giordano, Y. Katayama, Y. Yu and Y. Shao-Horn, Science, 2017, 358, 751–756.
37. R. E. Cohen, Nature, 1992, 358, 136–138.
38. K. Huang, R. S. Tichy and J. B. Goodenough, J. Am. Ceram. Soc., 1998, 81, 2565–2575.
39. J. Suntivich, H. A. Gasteiger, N. Yabuuchi, H. Nakanishi, J. B. Goodenough and Y. Shao-Horn, Nat. Chem., 2011, 3, 546–550.
40. E. Fabbri, M. Nachtegaal, T. Binninger, X. Cheng, B.-J. Kim, J. Durst, F. Bozza, T. Graule, R. Schäublin and L. Wiles, et al., Nat. Mater., 2017, 16, 925–931.
41. B.-J. Kim, X. Cheng, D. F. Abbott, E. Fabbri, F. Bozza, T. Graule, I. E. Castelli, L. Wiles, N. Danilovic and K. E. Ayers, et al., Adv. Funct. Mater., 2018, 28, 1804355.
42. W. T. Hong, M. Risch, K. A. Stoerzinger, A. Grimaud, J. Suntivich and Y. Shao-Horn, Energy Environ. Sci., 2015, 8, 1404–1427.
43. H. Over, Y. D. Kim, A. Seitsonen, S. Wendt, E. Lundgren, M. Schmid, P. Varga, A. Morgante and G. Ertl, Science, 2000, 287, 1474–1476.
44. J. Kim, X. Yin, K.-C. Tsao, S. Fang and H. Yang, J. Am. Chem. Soc., 2014, 136, 14646–14649.
45. A. Grimaud, K. J. May, C. E. Carlton, Y.-L. Lee, M. Risch, W. T. Hong, J. Zhou and Y. Shao-Horn, Nat. Commun., 2013, 4, 2439.
46. A. Swarnkar, W. J. Mir and A. Nag, ACS Energy Lett., 2018, 3, 286–289.
47. W. Zhang, H. Liu, Y. Qu, J. Cui, W. Zhang, T. Shi and H.-L. Wang, Adv. Mater., 2024, 36, 2309193.
48. Y. Liang, F. Li, X. Cui, C. Stampfl, S. P. Ringer, X. Yang, J. Huang and R. Zheng, Sci. Adv., 2025, 11, eads7054.
49. Z. Li, P. Wei and G. Wang, Energy Fuels, 2022, 36, 11724–11744.
50. C. E. Beall, E. Fabbri and T. J. Schmidt, ACS Catal., 2021, 11, 3094–3114.
51. S. Sen and T. K. Mandal, ChemCatChem, 2025, e00535.
52. J. T. Mefford, X. Rong, A. M. Abakumov, W. G. Hardin, S. Dai, A. M. Kolpak, K. P. Johnston and K. J. Stevenson, Nat. Commun., 2016, 7, 11053.
53. N. Zhang and Y. Xiong, J. Phys. Chem. C, 2023, 127, 2147–2159.
54. B.-J. Kim, E. Fabbri, M. Borlaf, D. F. Abbott, I. E. Castelli, M. Nachtegaal, T. Graule and T. J. Schmidt, Mater. Adv., 2021, 2, 345–355.
55. L.-B. Liu, C. Yi, H.-C. Mi, S. L. Zhang, X.-Z. Fu, J.-L. Luo and S. Liu, Electrochem. Energy Rev., 2024, 7, 14.
56. Y. Ma, M. Zha, Y. Dong, L. Li and G. Hu, Mater. Res. Express, 2019, 6, 115033.
57. Y.-F. Li and A. Selloni, ACS Catal., 2014, 4, 1148–1153.
58. F. Song, L. Bai, A. Moysiadou, S. Lee, C. Hu, L. Liardet and X. Hu, J. Am. Chem. Soc., 2018, 140, 7748–7759.
59. Z.-D. He, R. Tesch, M. J. Eslamibidgoli, M. H. Eikerling and P. M. Kowalski, Nat. Commun., 2023, 14, 3498.
60. I. Hussain, U. Sajjad, O. J. Kewate, U. Amara, F. Bibi, A. Hanan, D. Potphode, M. Ahmad, M. S. Javed and P. Rosaiah, et al., Mater. Today Phys., 2024, 42, 101382.
61. H. Wang, W. Xu, S. Richins, K. Liaw, L. Yan, M. Zhou and H. Luo, Electrochim. Acta, 2019, 296, 945–953.
62. H. Wang, J. Wang, Y. Pi, Q. Shao, Y. Tan and X. Huang, Angew. Chem., Int. Ed., 2019, 131, 2338–2342.
63. J. Bian, R. Su, Y. Yao, J. Wang, J. Zhou, F. Li, Z. L. Wang and C. Sun, ACS Appl. Energy Mater., 2019, 2, 923–931.
64. M. Bajdich, M. García-Mota, A. Vojvodic, J. K. Nørskov and A. T. Bell, J. Am. Chem. Soc., 2013, 135, 13521–13530.
65. O. N. Avcı, L. Sementa and A. Fortunelli, ACS Catal., 2022, 12, 9058–9073.
66. N. Karmodak and O. Andreussi, J. Phys. Chem. C, 2021, 125, 10397–10405.
67. A. Li, S. Kong, C. Guo, H. Ooka, K. Adachi, D. Hashizume, Q. Jiang, H. Han, J. Xiao and R. Nakamura, Nat. Catal., 2022, 5, 109–118.
68. J. Suntivich, K. J. May, H. A. Gasteiger, J. B. Goodenough and Y. Shao-Horn, Science, 2011, 334, 1383–1385.
69. A. S. Raman, R. Patel and A. Vojvodic, Faraday Discuss., 2021, 229, 75–88.
70. B. Luo, F. Li, K. Xu, Y. Guo, Y. Liu, Z. Xia and J. Z. Zhang, J. Mater. Chem. C, 2019, 7, 2781–2808.
71. The computational details are in the SI.
72. G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1993, 48, 13115–13118.
73. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
74. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
75. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865.
76. S. Maintz, V. L. Deringer, A. L. Tchougréeff and R. Dronskowski, J. Comput. Chem., 2013, 34, 2557–2567.
77. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758.
78. K. Mathew, R. Sundararaman, K. Letchworth-Weaver, T. Arias and R. G. Hennig, J. Chem. Phys., 2014, 140, 084106.
79. K. Mathew, V. Kolluru, S. Mula, S. N. Steinmann and R. G. Hennig, J. Chem. Phys., 2019, 151, 234101.
80. J. A. Gauthier, S. Ringe, C. F. Dickens, A. J. Garza, A. T. Bell, M. Head-Gordon, J. K. Nørskov and K. Chan, ACS Catal., 2018, 9, 920–931.
81. J. A. Gauthier, C. F. Dickens, S. Ringe and K. Chan, ChemPhysChem, 2019, 20, 3074–3080.
82. O. Andreussi, I. Dabo and N. Marzari, J. Chem. Phys., 2012, 136, 064102.
83. A. H. Larsen, J. J. Mortensen, J. Blomqvist, I. E. Castelli, R. Christensen, M. Dułak, J. Friis, M. N. Groves, B. Hammer and C. Hargus, et al., J. Condens. Matter Phys., 2017, 29, 273002.
84. Y. Choi, D. S. Mebane, M.-C. Lin and M. Liu, Chem. Mater., 2007, 19, 1690–1699.
85. R. Evarestov, E. Kotomin, Y. A. Mastrikov, D. Gryaznov, E. Heifets and J. Maier, Phys. Rev. B: Condens. Matter Mater. Phys., 2005, 72, 214411.
86. Y.-L. Lee, J. Kleis, J. Rossmeisl and D. Morgan, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 80, 224101.
87. L. Trotochaud, J. K. Ranney, K. N. Williams and S. W. Boettcher, J. Am. Chem. Soc., 2012, 134, 17253–17261.
88. J. O. Bockris and T. N. Veziroglu, Int. J. Hydrogen Energy, 2007, 32, 1605–1610.
89. M. D. Bhatt and J. Y. Lee, Energy Fuels, 2020, 34, 6634–6695.
90. R. Dronskowski and P. E. Bloechl, J. Phys. Chem., 1993, 97, 8617–8624.
91. S. Geiger, O. Kasian, M. Ledendecker, E. Pizzutilo, A. M. Mingers, W. T. Fu, O. Diaz-Morales, Z. Li, T. Oellers and L. Fruchter, et al., Nat. Catal., 2018, 1, 508–515.
92. O. Kasian, J.-P. Grote, S. Geiger, S. Cherevko and K. J. Mayrhofer, Angew. Chem., Int. Ed., 2018, 57, 2488–2491.
93. C. W. Song, J. Lim, H. B. Bae and S.-Y. Chung, Energy Environ. Sci., 2020, 13, 4178–4188.
94. F.-Y. Chen, Z.-Y. Wu, Z. Adler and H. Wang, Joule, 2021, 5, 1704–1731.
95. N. Zhang and Y. Chai, Energy Environ. Sci., 2021, 14, 4647–4671.
96. J. Zaanen, G. Sawatzky and J. Allen, Phys. Rev. Lett., 1985, 55, 418.
97. A. Grimaud, W. T. Hong, Y. Shao-Horn and J.-M. Tarascon, Nat. Mater., 2016, 15, 121–126.
98. W. Zhang, Y. Wang, Y. Li and X. Zhang, J. Ind. Eng. Chem., 2024, 139, 199–212.
99. S. R. Ede, C. N. Collins, C. D. Posada, G. George, H. Wu, W. D. Ratcliff, Y. Lin, J. Wen, S. Han and Z. Luo, ACS Catal., 2021, 11, 4327–4337.
100. S. Razzaq and K. S. Exner, ACS Catal., 2023, 13, 1740–1758.
101. K. S. Exner, ACS Catal., 2020, 10, 12607–12617.
102. M. Usama, S. Razzaq, C. Hättig, S. N. Steinmann and K. S. Exner, Nat. Commun., 2025, 16, 6137.
103. K. Dhaka and K. S. Exner, J. Catal., 2025, 443, 115970.

---