Effect of metal substitution on the intrinsic activity of iridium-based oxides for the oxygen evolution reaction

Abstract

Iridium-based oxides are among the most promising catalysts for the acidic oxygen evolution reaction (OER) owing to their high catalytic activity and stability. Substituting iridium with earth-abundant elements could lower costs and potentially boost its intrinsic activity even further; however, no unambiguous structure–activity relationships describing the physical origins of the effect of the substituent for this class of electrocatalysts have been established. In this work, we utilized a series of IrO_x(:M) nanoparticle catalysts to correlate their in situ structural changes with intrinsic OER activity. We observe that IrO_x(:M) with M = W and In feature a significantly higher Ir-mass-normalized OER activity than IrO_x, however the activity enhancements have a different origin. While the increased activity of IrO_x : In stems from a higher number of electrochemically active iridium centers (due to the leaching of indium), IrO_x : W features a higher intrinsic OER activity compared to IrO_x, due to electronic effects of W on neighboring Ir/O sites. Furthermore, operando electrochemical mass spectrometry experiments and density functional theory (DFT) calculations revealed that the enhanced OER activity of IrO_x(:M) does not originate from a promotion of the lattice oxygen coupling mechanism, but is instead associated with a facilitated conventional adsorbate evolution mechanism.

---

Introduction

Iridium oxide is a benchmark anode electrocatalyst in proton-exchange membrane water electrolyzers (PEMWEs) owing to its high catalytic activity and electrochemical stability,^1–10 outperforming other catalyst materials including ruthenium oxide.^11,12 However, the scarcity of iridium is a considerable issue preventing the large scale implementation of PEMWEs.¹³ The use of Ir-based mixed metal oxides in which Ir is partially substituted by another metal is a commonly employed strategy to reduce the precious metal content.^14–19 In addition, the partial substitution of Ir might provide an avenue to enhance the intrinsic activity of the Ir sites.

However, the instability of many Ir-containing complex oxides under OER conditions is a considerable challenge.^15,20–22 For example, although Ir-containing oxides, such as SrIrO₃ or La₂LiIrO₆, have a high surface-area-normalized OER activity,^20,21 they have also exhibited an appreciable degree of reconstruction resulting in the exsolution of IrO_y(OH)_z·nH₂O (also referred to as IrO_x) or crystalline IrO₂ on their surface.^23–29 Such complex transformations under OER conditions often lead to ill-defined, defective motifs that complicate an accurate description of the electronic and geometric structure of the active site (i.e., the oxidation state and coordination number of iridium, Ir–O/OH bond length, etc.). Owing to this inherent complexity and structural dynamics of Ir-based systems, there has been a large number of studies aiming to explore the origins of their high OER activity and to formulate structure–activity correlations with the overarching goal to provide the tool to guide the design of highly active and stable, Ir-based (mixed) oxide electrocatalysts.^{15,20,23–25,30–32} It was found that the structure of IrO_x under OER conditions is strongly influenced by the precursor composition, however the relationship between the precursor structure and the resulting catalyst phase remains poorly understood.^17,21,27,33

A key challenge in understanding the impact of dopants/substituents on the structure and OER activity of Ir-based catalysts is that most studies report electrode surface area- or mass-normalized OER activities without accounting for the fraction of catalytically- or redox-active IrO_x sites. Further, improvements in the catalytic activity of Ir-based catalysts have frequently been attributed, often without unequivocal experimental evidence, to electronic modulation of iridium by the substituent M. Indeed, in many cases, only the structure of the precursor materials was assessed, i.e., the structures that may not exist under actual reaction conditions, as the true active state of the catalysts could be a monometallic IrO_x phase rather than the proposed bimetallic phases. Such misrepresentations have contributed to the recurring appearance of erroneous structure–activity relationships in the recent literature.

Turning to the OER reaction mechanism of (doped) iridium oxides, a number of experimental and computational studies have argued that OER on iridium oxides might involve lattice oxygen sites.^8,34–36 However, the estimation of the fraction of oxygen evolved from the lattice with respect to the total amount of O₂ formed during OER proved to be an experimentally challenging task making the assignment of the dominating reaction pathway in many previous studies rather speculative.

Hence, the aim of this work is to elucidate and generalize the effect of substituents on the intrinsic activity of IrO_x derived from a series of (bi)metallic Ir-based precursors, assess in situ changes in the electronic and local structure of iridium sites under OER conditions, and probe the OER mechanism of IrO_x(:M) materials using electrochemical mass spectrometry (EC-MS) experiments assisted by ¹⁸O-labelling. To this end, we prepared a series of monodispersed (sub-3.5 nm in diameter) bimetallic Ir–M (M = Cr, Mo, W, In) nanoparticles that serve as model catalyst precursors. We observed that there is a significant enhancement in the surface area- and iridium-mass-normalized OER activity for M = W or In (for various Ir_mIn_n and Ir_xW_y compositions). Ex situ XPS of the electrochemically cycled materials show that tungsten is present in the IrO_x lattice in a Wⁿ⁺ state, whereas indium is completely dissolved yielding an In-free IrO_x phase. In situ XAS experiments at the Ir L₃-edge allowed us to quantify the changes in the oxidation state and coordination of Ir sites during OER and to estimate the OER turnover frequency of the in situ formed IrO_x(:M) structures/motifs. Operando electrochemical mass spectrometry (EC-MS) studies using ¹⁸O isotope labelling indicated that the lattice oxygen mediated mechanism (LOM) contributes to OER on all of the tested IrO_x(:M) catalysts, however the contribution of the LOM to the overall O₂ formation rate is rather small and hence cannot explain the significant differences in the intrinsic catalytic activity of the IrO_x(:M) catalysts. Instead, we established that in case of the most active catalyst, i.e., IrO_x : W, the incorporation of tungsten increases the covalency of the Ir–O bonds that results in a decrease of the energy of the potential-limiting step in the conventional adsorbate-evolution mechanism (AEM). Broadly, our work targets to consolidate prior studies, providing a comprehensive perspective on the effect of substituents on the catalytic activity of iridium oxide, while also underscoring persistent challenges in the interpretation of experimental data that have led to a recurring appearance of erroneous structure–activity relationships in the recent literature.

Results and discussion

Average structural characterization of nanoparticles

Ir–M (M = Cr, Mo, W, In) nanoparticles were synthesized via a 1-octadecene- and oleylamine-assisted thermal decomposition of acetylacetonate- or chloride-based metal precursors (further details are provided in the SI).³⁷ To stabilize the Ir^(4−δ)+ state, group VI transition metals were chosen as substituents. On the other hand, indium was selected as a substituent as the incorporation of In³⁺ is expected to yield the opposite effect, i.e., to generate Ir^(4+δ)+ sites (or, alternatively, generate oxygen vacancies). We expected that our choices of M will allow us to assess critically the findings of previous studies that have linked the presence of Ir^(4+δ)+ or Ir^(4−δ)+ species to an enhanced activity of various IrO_x(:M) phases.^23,24,38,39

In the following sections, the composition of the nanoparticles will be described using the nominal Ir : M ratio used in the synthesis. The exact Ir : M composition of the nanoparticles, as determined by inductively coupled plasma optical emission spectroscopy (ICP-OES) is tabulated in Table S1. TEM and HR-STEM images of the monodispersed metallic Ir–M nanoparticles are shown in Fig. 1a, b and S1. The electron diffraction patterns of pristine Ir, Ir₇₅W₂₅, and Ir₇₅In₂₅ nanoparticles demonstrate the presence of a metallic Ir phase in all catalysts (Fig. 1c), which is in agreement with the powder X-ray diffraction patterns of the as-synthesized nanoparticles (Fig. 1d).

Fig. 1 (a) Bright-field transmission electron microscopy (TEM) images and the corresponding histograms of particle size distribution, (b) high-angular annular dark field scanning high-resolution transmission electron microscopy (HAADF-HRSTEM) images and the corresponding Fast Fourier Transforms (FFTs) obtained from single particles enclosed in squared areas on the HAADF-HRSTEM images, (c) azimuthally averaged intensity profiles obtained from the selected area electron diffraction (SAED) patterns of pristine Ir, Ir₇₅W₂₅ and Ir₇₅In₂₅ nanoparticles. The patterns can be indexed in the space group Fm3̄m (metallic Ir). (d) powder X-ray diffraction (XRD) patterns of the pristine Ir, Ir–W, and Ir–In nanoparticles (vertical lines correspond to peak positions of Ir reference [ICDD 01-071-4659]). Additional bright field TEM images of Ir–M (M = Cr, Mo, W, and In) nanoparticles are shown on Fig. S1.

Overall electrochemical activity trends

The OER activity measurements of the series of Ir and Ir–M nanoparticles were performed in 0.1 M HClO₄ (see SI).⁴⁰ In general, all Ir–M catalysts exhibited a gradual increase of the OER current over the first ca. 10 CV cycles, followed by its stabilization. In the following, the OER current measured after 10 CV cycles, was used as the metric of the OER activity.

A comparison of the Ir-mass-normalized and geometric OER activities of Ir and Ir–M (M = Cr, Mo, W, In) is presented in Fig. 2a and S2. Polarization curves based on the geometric area of the electrodes for the Ir–M series are shown in Fig. S3. Among the different catalysts tested, Ir–W and Ir–In stand out due to their very high OER activity, exceeding appreciably the performance of monometallic Ir. Hence, in the following sections we will focus on rationalizing the high OER activity of the Ir–W and Ir–In based catalysts by characterizing in detail their structural and electronic properties.

Fig. 2 (a) Ir mass-normalized polarization curves of Ir, Ir₇₅W₂₅, and Ir₇₅In₂₅ nanoparticles. (b) OER activity at 1.53 V_RHE for different Ir : M ratios of the Ir–W and Ir–In families. Error bars represent the standard deviations for at least seven independent measurements. (c) Chronoamperometric curves of Ir, Ir₇₅W₂₅, and Ir₇₅In₂₅ nanoparticles at 1.55 V_RHE. (d) Polarization curves of Ir, Ir₇₅W₂₅, and Ir₇₅In₂₅ using OER currents normalized by the electric-double-layer-derived electrochemical surface area or the total catalysts' geometric surface area (determined by TEM measurements). The catalyst loading was 10 μg_catalyst cm⁻² (except for the chronoamperometry experiments in which 50 μg_catalyst cm⁻² was used). The electrolyte was O₂-saturated 0.1 M HClO₄.

Focusing on the series of Ir–W and Ir–In catalysts (i.e., varying Ir : M ratios), we observed that in particular Ir₇₅W₂₅ and Ir₇₅In₂₅ show a very high Ir mass-normalized OER activity of ∼1200 A g_Ir⁻¹ (at 1.53 V_RHE), a value that significantly exceeds the activity of a catalyst derived from monometallic Ir nanoparticles (ca. 550 A g_Ir⁻¹ at 1.53 V_RHE, Fig. 2b). It is important to note that the Ir mass-normalized OER activity of the monometallic Ir nanoparticles reported here is on par or even exceeds the activity of other reported IrO_x or Ir-based mixed oxides, including those commonly used as reference materials in the literature (Fig. S4).

Despite the similar Ir-mass-normalized OER activity of Ir₇₅W₂₅ and Ir₇₅In₂₅, the electrochemical stability of Ir₇₅In₂₅ was significantly inferior to that of Ir₇₅W₂₅ and monometallic Ir. Fig. S5 shows the first 10 cycles of cyclic voltammetry (CV) experiments of the Ir–In series for varying ratios of Ir : In. We observe a gradual decrease of the anodic current with cycle number and with increasing In content (in 0.1 M HClO₄). Under alkaline conditions (0.1 M KOH), there was an even faster decrease in the OER current over the first 2–3 cycles for In-rich compositions (Fig. S6). The relatively low electrocatalytic stability of the Ir–In family is also observed in Fig. 2c and can be rationalized by the high solubility of indium by forming In³⁺ or InO₂⁻ species (in acidic and alkaline media, respectively).⁴¹ Interestingly, our observation differs from some previous works that report a stable and pH-independent OER activity of Ir–In based catalysts.⁴²

Although the Ir-mass-normalized OER activity (A g_Ir⁻¹) is a frequently used metric to evaluate the activity of noble metal-based catalysts (Fig. S4), it fails to describe accurately the intrinsic activity of a catalyst (i.e., the activity of an individual catalytic site). In this context, a more appropriate normalization would be based on the electrochemically active surface area (ECSA, derived from double-layer capacitance measurements; see Fig. S7) or on the geometric surface area of the nanoparticles (e.g., as calculated from TEM measurements). When normalizing the OER currents by the surface area (electrochemically active surface area or geometric surface area determined by TEM), Ir–W outperforms notably the monometallic Ir catalyst, while the area-normalized OER activities of Ir–In and monometallic Ir are very similar (Fig. 2d). This comparison (Fig. 2a, d and S2) indicates that the higher Ir-mass-normalized OER current of Ir–In with respect to monometallic Ir is due to a larger number of exposed surface sites (i.e., a larger electrochemically active surface area) rather than a higher intrinsic activity of the catalytically active sites (see further discussion below).

Assessment of the participation of lattice oxygen in the oxygen evolution reaction

For OER, two distinct reaction mechanisms are typically discussed, viz. the lattice oxygen mediated (LOM) mechanism and the (conventional) adsorbate evolution mechanism (AEM). LOM involves the coupling of neighboring oxygen sites within the oxide lattice as the O–O bond formation step, as opposed to a nucleophilic attack of H₂O/OH⁻ species on the terminal metal atoms in the AEM. It has been proposed that LOM can result in lower OER overpotentials than the AEM pathway,^43,44 hence, a large number of studies have aimed at activating the LOM by modifying the electronic structure of Ir-based catalysts. However, very few works have demonstrated experimentally that the LOM (coupling and release of lattice oxygens) indeed takes place.

To address this limitation, we investigate the involvement of lattice oxygen in the OER catalyzed by IrO_x(:M) using operando chip-based electrochemical mass spectrometry (EC-MS, experimental details are provided in the SI), which enables a fully quantitative analysis of the evolved gases.⁴⁵ In these experiments, first the metallic Ir–M (M = W and In) precursors were oxidized electrochemically in 0.1 M HClO₄–¹⁸O to yield ¹⁸O-enriched catalysts. After the ¹⁸O-isotope-labelling step, the electrode was rinsed thoroughly with unlabeled H₂O, followed by cycling in unlabeled 0.1 M HClO₄ while detecting simultaneously the gas-phase products by mass spectrometry. Fig. 3a compares the evolution of the products m/z = 32 (³²O₂), 34 (¹⁶O¹⁸O), and 36 (³⁶O₂) during the first CV cycle as a function of time.

Fig. 3 (a) EC-MS profiles of the current, applied bias and MS ion currents for m/z = 32, 34, and 36 species as a function of time during the first OER cycle; (b) composition of the evolved O₂ species over ten OER cycles for ¹⁸O-labeled Ir, Ir₇₅W₂₅, and Ir₇₅In₂₅. The dashed lines interpolate the steady state fractions of ³²O₂ and ³⁴O₂ (¹⁶O¹⁸O) when using non-labelled water (H₂O–¹⁶O). Electrolyte: 0.1 M HClO₄ in H₂O–¹⁶O. Loading: 10 μg_catalyst (on a gold disk).

From Fig. 3a we observe that there is a significant production of ³⁴O₂ (¹⁶O¹⁸O) which can be formed via the combination of two oxygen atoms from the electrolyte (due to the natural abundance of ¹⁸O in water), via the combination of O atoms from the electrolyte and the oxide lattice or via coupling of two lattice O atoms, and ³⁶O₂ (¹⁸O¹⁸O formed mainly from two lattice oxygen atoms) on Ir-, Ir₇₅W₂₅-, and Ir₇₅In₂₅-derived catalysts. These measurements confirm that there is some contribution of the LOM to the overall O₂ production during OER. Fig. 3b and S8 plot the fraction of ³²O₂, ³⁴O₂ and ³⁶O₂ in the O₂ evolved over the first 10 CV cycles for ¹⁸O-labelled Ir, Ir₇₅W₂₅, and Ir₇₅In₂₅. For all materials tested, the fraction of ³⁴O₂ gradually decreased with increasing cycle number reaching a steady-state value of ∼0.55% after 5 cycles, which is close to 0.4%, i.e., the expected value based on the theoretical fraction of ³⁴O₂ evolved from water considering the natural ¹⁸O abundance of 0.2%.⁴⁶ In addition, the MS signal due to ³⁶O₂ also decreased with cycle number, falling below the detection limit after ca. 5 cycles, consistent with the replenishment of lattice oxygen vacancies (generated in the LOM pathway) with ¹⁶O from the unlabeled electrolyte.

Hence, the main conclusions derived from our EC-MS measurements can be summarized as follows:

(1) Previous works have reported the evolution of ³⁶O₂ from anodically oxidized ¹⁸O-labelled Ir films, while crystalline rutile IrO₂ did not show any evolution of ³⁶O₂.^34,35 Considering that we observed the evolution of ³⁶O₂ in all of the catalysts tested, we hypothesize that the oxidized surface of all Ir–M (M = Ir, W, In) nanoparticles under OER conditions was composed of defective, hydrous IrO_x (formed also under OER conditions on a metallic Ir film) rather than of a well-defined crystalline rutile IrO₂ phase.

(2) While we observe that the OER proceeds to some extent via the LOM on IrO_x(:M), the quantitative analysis of the EC-MS data indicates that the fraction of O₂ formed via the LOM (compared to the overall amount of O₂ evolved) is rather small (Fig. S9).

(3) While we observe small differences in the fractions of ³⁴O₂ and ³⁶O₂ in the O₂ evolved from Ir, Ir₇₅W₂₅ and Ir₇₅In₂₅, a quantification of the different isotopes in the product (Fig. S9) indicates that the differences in the intrinsic activities observed for these three catalysts are likely not due to differences in the dominating OER mechanism (further support is provided by DFT calculations, see below).

In situ X-ray absorption spectroscopy

Next, to obtain more insight into the structure of the catalytically active sites, we used in situ Ir L₃ edge XAS to probe changes in the Ir oxidation state and Ir coordination environment in the IrO_x(:M) series (M = Ir, W, In) under OER conditions (we note that XAS probes the average electronic and geometric local structure of both bulk and surface Ir sites). In the in situ XAS experiments, the electrodes were first cycled between 0.85 V_RHE and 1.65 V_RHE (50 CV scans) to oxidize metallic iridium yielding the active IrO_x phase. Spectra collected at OCV upon immersion of the electrodes into the electrolyte are referred to as “OCV” in Fig. 4a–c, while the data collected on electrodes that have undergone 50 CV cycles are labelled “OCV (cycled)”. Subsequently, chronoamperometry steps at potentials ranging from 1.16 V_RHE to 1.56 V_RHE were performed. The in situ X-ray absorption near edge structure (XANES) and the magnitude of the Fourier transform (FT) of the extended X-ray absorption fine structure (EXAFS) data of Ir, Ir₇₅W₂₅, Ir₇₅In₂₅ and Ir₅₀In₅₀ are shown in Fig. 4a and b. The XANES data of further Ir–W compositions are presented in Fig. S10. The white line (WL) peak position (fitted using one Lorentzian component) was selected as the parameter to estimate the oxidation state of iridium in the catalysts (Fig. S11), instead of using the integrated intensity (see discussion in the SI).

Fig. 4 In situ (a) XANES and (b) non-phase-shift corrected k²-weighted magnitude of the FT EXAFS, at different stages of CV cycling and during chronoamperometric steps of Ir, Ir₇₅W₂₅, Ir₇₅In₂₅, and Ir₅₀In₅₀. (c) Ir L₃-edge WL position, Ir⁰ fraction derived from linear combination fitting (references: Ir powder, Ir(acac)₃ and IrO₂), fitted Ir–M and Ir–O coordination numbers, fitted Ir–O interatomic distance of Ir, Ir₇₅W₂₅, Ir₇₅In₂₅ and Ir₅₀In₅₀ as a function of the applied potential in 0.1 M HClO₄ using a static cell. The bottom baseline of the first panel of (c) indicates the peak position of the fitted metallic Ir reference (11 220.72 eV). All spectra were acquired in fluorescence mode, except of that of the references Ir, Ir(acac)₃, and IrO₂. Error bars represent the uncertainties in the fit variables. The EXAFS fitting parameters can be found in Table S2. Loading: 2 mg_catalyst cm⁻² (on carbon paper).

Evolution of the fraction of IrO_x and the average oxidation state of Ir under OER conditions

When comparing the XANES spectra of Ir, Ir₇₅W₂₅, Ir₇₅In₂₅, Ir₅₀In₅₀ before and after cycling, we observe a shift in the peak position of the white line (WL) to higher energies, along with an increase in the intensity of the WL (Fig. 4a–c and S11). This observation indicates that the average oxidation state of iridium in Ir, Ir–W and Ir–In catalysts increases with CV cycling which we attribute to the formation of oxidized Ir species (i.e., IrO_x) during the initial 50 CV cycles (consistent with the ex situ X-ray photoelectron Ir 4f core spectra (XPS) of the materials, Fig. S12).

The shift in the WL energy position during the chronoamperometric steps can be due to (i) a further oxidation of the remaining metallic Ir⁰ core to generate additional IrO_x species and/or (ii) an increase of the oxidation state of Ir in the already formed IrO_x phase. To disentangle these two potential contributions, we monitor the fraction of metallic Ir⁰ of Ir, Ir₇₅W₂₅, Ir₇₅In₂₅ and Ir₅₀In₅₀ after cycling and during the chronoamperometric steps using linear combination fitting (LCF) analysis (details in the SI). While determining the exact oxidation state of Ir in the IrO_x phase by XANES is very challenging due to the overlapping features of Ir species of different oxidation states, our analysis, nonetheless, allowed us to follow the evolution of the non-Ir⁰ (i.e., IrO_x) component as a function of the applied potential. LCF analysis revealed that the fraction of Ir⁰ decreased during cycling and, to a lesser extent, during the step chronoamperometry experiments, independent of the catalyst formulation.

During the chronoamperometric steps, the decrease in the fraction of Ir⁰ was most pronounced in Ir–In (in particular for Ir₅₀In₅₀) and less noticeable in Ir and Ir₇₅W₂₅ (Fig. 4a and c). Specifically, between the state “OCV (cycled)” and “1.56 V_RHE”, the fraction of Ir⁰ decreased from 0.57 to 0.42 for Ir, from 0.44 to 0.30 for Ir₇₅W₂₅, and from 0.45 to 0.16 for Ir₇₅In₂₅. For Ir₅₀In₅₀, the fraction of Ir⁰ reduced to zero indicating the complete oxidation of metallic Ir during the step chronoamperometric experiments. We speculate that the oxidation of a larger fraction of the original Ir⁰ phase in Ir–In-based catalysts is due to (i) the smaller size of the Ir–In nanoparticles (when compared to the Ir and Ir–W samples), resulting in a higher fraction of surface Ir sites exposed to the electrolyte and (ii) due to the rapid dissolution of In in acidic electrolytes (consistent with the Pourbaix diagram for In–H₂O⁴¹), as also confirmed by ex situ XPS that showed the absence of In in the cycled material (Fig. S13).

In contrast to the Ir–In series in which In was leached out completely, W remained as a mixture of +4, +5 and +6 states (as probed by the ex situ XPS studies) in the structure of the Ir–W catalysts. Furthermore, our W L₃-edge EXAFS data suggest that W has a local environment and R(W–O) that are distinct from that of WO₂ or WO₃ (Fig. S14) (note that due to the restriction of the k-range to ca. 9 Å, the structural analysis was confined to the W–O coordination shell). While the XAS data do not allow an unambiguous determination of the exact geometry of the W sites in IrO_x : W, our observations support a structural model in which (i) WO_x is dispersed within the IrO_x matrix (in the bulk or at the surface of IrO_x), and (ii) there is an electronic interaction between W and Ir (hence pointing to a Ir–O–W bonding). Considering all experimental data acquired, we therefore argue that the incorporation of W into the IrO_x crystal lattice represents the most likely structural model, which was therefore used in the DFT calculations (see below).

Evolution of the local structure of the catalysts as probed by EXAFS

Next, we performed EXAFS fittings to determine the average coordination numbers (CN) and interatomic distances (R) of the Ir–O and Ir–M (M = Ir, W, or In) coordination shells (Fig. 4c, S15, S16, and Table S2; see also Fig. S17 for in situ k²-weighted EXAFS data in k-space of Ir, Ir–W and Ir–In catalysts). Comparing the structure of the catalysts before and after cycling, we find that the CN of the Ir–M shell decreased for all materials tested (see Table S2). In addition, after the chronoamperometric steps up to 1.56 V_RHE, the CN(Ir–M) decreased even further to 7.5(4) in Ir, to 6.2(5) in Ir₇₅W₂₅ and to 0.9(5) Ir₇₅In₂₅ (Fig. 4c). In Ir₅₀In₅₀ the magnitude of the Ir–M shell was negligible after the chronoamperometric steps up to 1.56 V_RHE, consistent with LCF analysis that showed the absence of a metallic Ir⁰ phase. Furthermore, the decrease of the CN(Ir–M) is accompanied by an increase in the corresponding CN(Ir–O) (see Table S2), which is consistent with the reduction of the fraction of Ir⁰ and the formation of an IrO_x phase during CV cycling and the subsequent chronoamperometric steps.

Turning now to the fitted interatomic distances, the R(Ir–M) in the Ir and Ir–W catalysts studied here, fluctuate around 2.7 Å (Fig. S15) throughout the duration of the in situ experiments, in good agreement with R(Ir–Ir) = 2.704(5) Å in the reference Ir powder (Sigma-Aldrich). This observation implies that in the Ir and Ir–W catalysts a metallic iridium core is preserved to some extent during the in situ electrochemical experiment.

Previous reports probing the structure of the active iridium oxide/(oxy)hydroxide phases report a shortening of R(Ir–O) at increasing OER potentials and link the decrease in R(Ir–O) with an increase in the oxidation state of iridium.⁴⁷ At “OCV (cycled)”, the shorter R(Ir–O) of the IrO_x phase formed in the monometallic Ir precursor (R(Ir–O) = 1.958(8) Å) when compared to the Ir⁴⁺ reference IrO₂ (R(Ir–O) = 1.983 Å) suggests a higher (average) oxidation state of the iridium sites of the IrO_x phase in the monometallic Ir catalyst (i.e., Ir^(4+δ)+) compared to crystalline IrO₂.⁴⁸ Indeed, when comparing R(Ir–O) of the IrO_x phase obtained from the monometallic Ir precursor after the chronoamperometric steps (1.94(1) Å at 1.56 V_RHE) and an anodically oxidized IrO_x film (1.94 Å at 1.0 V_Ag/AgCI),⁴⁷ suggests that the oxidation state of Ir in these two materials is very similar (i.e., ca. Ir^4.5+). On the other hand, R(Ir–O) in IrO_x : W formed at the “OCV(cycled)” stage from Ir₇₅W₂₅ is close to that of crystalline IrO₂ (Fig. 4c).

In the Ir–In system, R(Ir–O) at “OCV (cycled)” is larger than the respective values of the monometallic Ir-derived catalysts which we attribute to a lower oxidation state of Ir in the IrO_x phases derived from Ir–In precursors (i.e., Ir⁴⁺ or Ir^(4−δ)+). Application of a bias (up to 1.56 V_RHE) resulted in insignificant changes in the mean oxidation state of the IrO_x phase derived from Ir₇₅In₂₅ and the formation of Ir⁴⁺O_x(OH)_y from the Ir₅₀In₅₀ precursor (bottom panel of Fig. 4c). We speculate that the relatively low oxidation state of iridium in the catalysts formed from the Ir–W and Ir–In precursors is caused by, respectively, (i) the substitution of Ir with high-valence W resulting in a reduction of Ir and (ii) the loss of indium which facilitates the formation of a defective (i.e., oxygen vacancy-rich) IrO_x phase.

The structural difference between the IrO_x phase formed in Ir–In (in particular for higher In contents) and crystalline IrO₂ is also seen qualitatively in the FT EXAFS data. While the crystalline IrO₂ reference shows a peak at ca. 3.5 Å (without phase-shift correction, Fig. 4b) which is expected for a higher-order Ir–Ir coordination in crystalline Ir oxides, this peak is absent in the FT EXAFS of the Ir–In system both at “OCV (cycled)” and during the chronoamperometric steps. These observations suggest that IrO_x derived from Ir₇₅In₂₅ and Ir₅₀In₅₀ nanoparticles (upon loss of In from the lattice) lacks the long-range order that is characteristic of rutile IrO₂.

To conclude, in situ XAS analysis allowed us (i) to probe changes in the average oxidation state of iridium, (ii) determine the fraction of Ir⁰ converted into IrO_x at the various steps of the in situ experiment, and (iii) determine the local environment of the iridium sites in the in situ formed IrO_x(:M) phases. We observed that at OER potentials iridium in the IrO_x phase obtained in the monometallic Ir system has an oxidation state ∼+4.5, whereas the oxidation state of iridium is ∼+4 in IrO_x : W. We attribute this effect to the presence of high oxidation state W in the IrO_x structure. In the Ir–In system, the leaching of In resulted in the complete oxidation of metallic iridium forming a defective IrO_x phase that lacks long range order and in which the oxidation state of Ir at OER potentials does not exceed +4.

Intrinsic activity of the IrO_x phase formed in situ from the different bimetallic precursors

Next, we compare the intrinsic OER activities of the Ir, Ir–In and Ir–W catalysts which we define as the OER current normalized by the mass of the electrochemically active sites (i.e., the mass of in situ formed IrO_x), determined by LCF analysis of the data collected at OCV after the chronoamperometric steps (see also Fig. S18). Using LCF analysis, at the end of the in situ electrochemical experiment the following fractions of IrO_x were determined: 50% (Ir), 65% (Ir₇₅W₂₅), 81% (Ir₇₅In₂₅), and 100% (Ir₅₀In₅₀). Calculating the IrO_x-mass-normalized OER currents, we find that the activities of Ir, Ir₇₅In₂₅ and Ir₅₀In₅₀ are very similar (ca. 400 A g_Ir→IrOx⁻¹ at 1.50 V_RHE, Fig. 5a). However, the IrO_x-mass-normalized OER activity of Ir₇₅W₂₅ is much higher, i.e., 650 A g_Ir→IrOx⁻¹ at 1.50 V_RHE. It is worth noting that the OER activity of the monometallic Ir catalyst studied here is comparable or even exceeding the Ir-mass-normalized activity when compared to previously reported iridium oxides/alloys (Fig. S4). Therefore, the ∼60% higher intrinsic OER activity of Ir–W compared to monometallic Ir represents a significant improvement.

Fig. 5 (a) Intrinsic OER activity at 1.50 V_RHE normalized by the total Ir mass loading or the LCF-derived fraction of IrO_x. (b) Comparison of the turnover frequency (TOF) of the catalyst series studied here with selected, previously reported Ir-based catalysts in acid conditions at different overpotentials. The error bars for Ir₇₅W₂₅ catalyst correspond to uncertainty in the fraction of IrO₂-component in LCF analysis. Empty bars: TOF_surface (based on catalyst surface area except for W_0.99Ir₀₀₁O_3–δ and IrO₂ (plasma-treated)³¹ for which the TOFs are based on the reported amount of Ir atoms at the catalyst surface); filled bars: TOF_bulk (based on total Ir loading). TOF_EA indexed by straight line (−) was calculated using the fraction of IrO_x present after the in situ XAS experiments. Some TOFs from literature were re-calculated using the reported particle size, Ir content, catalyst loading, Brunauer–Emmett–Teller (BET) or TEM-determined surface area.^32,49–51 (NCs: nanoparticles. NS: nanosheets). The exact values are listed in Table S3.

Quantifying the fraction of IrO_x also allowed us to estimate the turnover frequency (TOF) of the IrO_x : M catalysts (further details in the SI, Table S3). In the following we calculate and report three differently defined TOFs:

(1) TOF_surface: calculated by normalizing the OER current to the total geometric surface area of the nanoparticles determined by TEM (serving as an upper boundary);

(2) TOF_bulk: calculated by normalizing the OER current to the total loading of Ir on the electrode (serving as a lower boundary);

(3) TOF_EA: calculated by normalizing the OER current to the mass of the electrochemically active iridium (i.e., the amount of IrO_x determined by LCF analysis).

The following values of TOF_EA were determined: 0.18 s⁻¹ (Ir), 0.34 s⁻¹ (Ir₇₅W₂₅), and 0.20 s⁻¹ (Ir₇₅In₂₅) at 1.50 V_RHE and 0.57 s⁻¹ (Ir), 1.05 s⁻¹ (Ir₇₅W₂₅), 0.59 s⁻¹ (Ir₇₅In₂₅) at 1.53 V_RHE (Fig. 5b). It is noticeable that TOF_EA of the catalysts that are derived from monometallic Ir and Ir₇₅In₂₅ nanoparticles are very similar implying that the structural differences between the IrO_x phases formed from the Ir and Ir–In precursors (specifically, the lack of a long-range order for the latter) have a rather small impact on the intrinsic activity of the catalytically active iridium sites (Fig. 2c). Hence, we can also conclude that the markedly higher geometric activity (mA cm_disk⁻²) and total Ir mass activity (A g_Ir⁻¹) of the Ir–In catalysts, relative to monometallic Ir, arise exclusively from the larger surface area and the greater fraction of Ir converted into IrO_x in the Ir–In system.

Remarkably, the TOF_EA of Ir₇₅W₂₅ is comparable to that of two-dimensional iridium oxide (0.15 s⁻¹ at 1.50 V_RHE) and Ir_0.1Ta_0.9O_2.45 (1.5 ± 0.8 s⁻¹ at 1.53 V_RHE, fraction of the catalytically active sites was estimated by ¹³CO stripping measurements), i.e., placing it among the best performing OER catalysts (Fig. 5b and Table S3).^2,52 Even when using the TOF values that do not account for the fraction of the electrochemically active IrO_x (i.e., TOF_bulk: 0.22 s⁻¹ at 1.50 V_RHE and 0.68 s⁻¹ at 1.53 V_RHE), Ir₇₅W₂₅ exceeds the performance of many state-of-the-art Ir-based electrocatalysts (Fig. 5b and Table S3),^{2,6,31,49–56} indicative that the addition of W yields an exceptionally active OER catalyst.

Origin of the enhanced activity of IrO_x : W catalysts: theoretical considerations

The combined EC-MS and in situ XAS studies indicate that the enhanced OER activity of IrO_x : W does not originate from a switch of the reaction mechanism from the AEM to the LOM pathway. To add a theoretical support and rationalize this observation, we performed DFT calculations of the energetics of different OER pathways (i.e., the conventional AEM, LOM and I2M [coupling of two oxo-moieties bound to neighboring metal sites] pathways) on both IrO₂ and W-substituted IrO₂ (110) surfaces (Fig. 6a). The schematics of the proposed reaction mechanisms and the respective computed Gibbs free energy diagrams are shown in Fig. 6b, S19, S20, and Table S4. While we recognize that the ideal rutile IrO₂ model does not explicitly reproduce the structural complexity of the catalytically active hydrous IrO_x phase, it nonetheless provides a valid framework to assess the impact of tungsten incorporation on the energetics of OER intermediates in the IrO_x system.

Fig. 6 (a) Top view of the (110) termination of unsubstituted IrO₂, and IrO₂ containing one or two W atoms replacing iridium sites on the surface. These models were used in the DFT calculations. Highlighted are the bridging lattice oxygen (O_lat) and coordinatively unsaturated (O_cus) oxygen sites. (b) Schematic representation of the adsorbate evolution mechanism (AEM) of the OER considered in the DFT calculations and the corresponding Gibbs free energy diagram of the four elementary steps computed for each model shown on panel (a).

We observed that the potential-determining step of the AEM over IrO₂(:W) is the step *OOH → * + O₂; this is consistent with previous theoretical OER studies on iridium oxides.^36,57 The incorporation of W in the IrO₂ lattice decreases the free energy difference of this step (i.e., it becomes less endothermic) resulting in turn in a decrease of the theoretical OER overpotential in both W-containing models. Consistent with our measurements and previous literature reports,³⁵ our calculations indicate that the contribution of lattice oxygen via LOM is rather small for IrO₂ as it requires significantly larger OER overpotentials as compared to the AEM; similarly, I2M pathway was also found to be not feasible for the studied model systems (Fig. S19, S20 and Table S4).

To probe the effect of tungsten substitution on the electronic properties of IrO₂ : W, we calculated the electronic band structure and the Bader charges for the three models shown on Fig. 6a, S21 and Table S5. We found that the substitution of iridium with tungsten resulted in a decrease of the Bader charges on the Ir sites (i.e., from 1.88 for unsubstituted IrO₂ to 1.82–1.83 for W-substituted IrO₂), consistent with the in situ XAS observed reduction of the average oxidation state of iridium in Ir₇₅W₂₅. Such redistribution of charges, induced by the inductive effect of W⁵⁸ renders Ir–O bond more covalent that is reflected in a reduced energy gap between the Ir 5d and O 2p band centers for the W-substituted models, when compared to unsubstituted IrO₂ (Table S5). Therefore, a significantly higher intrinsic activity (i.e., TOF) of IrO_x : W derived from Ir–W precursors, relative to monometallic IrO_x, can be primarily ascribed to a W-induced electronic effect on adjacent O/Ir sites. This change in covalency of the Ir–O bond affects the adsorption energies of the O-containing OER intermediates and hence the overall OER kinetics.

Conclusions

This work demonstrates that substituting Ir with W or In yields OER catalysts with significantly higher OER activity than bulk IrO₂ or previously reported Ir-based catalysts. Using a series of complementary characterization techniques including in situ XAS and operando EC-MS analysis we rationalize the origin of the different (intrinsic) OER activities of IrO_x derived from different Ir(–M) precursors. The higher total iridium mass normalized OER activity of Ir–In compared to Ir originates from a higher fraction of the electrochemically (catalytically) active sites. On the other hand, IrO_x : W derived from Ir–W precursors exhibits a significantly higher TOF (i.e., intrinsic activity) than the monometallic IrO_x phase. This is attributed to a W-induced electronic effect on neighboring O/Ir sites. In all Ir-based catalysts studied here, EC-MS experiments confirm the participation of lattice oxygen atoms in OER, however there was no noticeable difference in the extent of lattice oxygen evolution between the different IrO_x : M catalysts. DFT calculations attribute the increase of the OER activity upon partial substitution of Ir by W to an increased covalency of the Ir–O bonds resulting in a decrease of the energy of the potential-limiting step in the AEM pathway rather than an activation of the LOM. Overall, the work reported here introduces novel, highly active OER catalysts, rationalizes the origin of their improved activity and demonstrates the importance of an accurate quantitative assessment of the intrinsic activity of IrO_x-based catalysts to yield robust structure–activity correlations that can aid the rational design of new compositions for acidic water electrolysis.

Author contributions

Y. X. and Y.-H. W. synthesized materials and performed electrochemical measurements. Y.-H. W. performed XPS characterizations. D. P. performed HR-STEM imaging and EDX analysis. A. K. collected TEM images and performed ICP-OES anaylsis. P. C. and V. A. performed DFT calculations. Y. X., Y.-H. W. and D. A. K. performed XAS. Y.-H. W and P. M. A. process XAS data analysis with input from C. S. and V. C. D. A. K. and C. R. M. supervised the research. The manuscript was edited by all authors.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d5sc06242j.

Acknowledgements

The authors acknowledge the Swiss Federal Laboratories for Materials Science and Technology (EMPA) for access to the XPS facility, the Diamond Light Source for access to the synchrotron radiation facilities at the B18 beamlines, and ScopeM (ETH Zürich) for access to the electron microscopy facilities. Harshit Mehta and Arthur Thevenot are acknowledged for participation in the optimization of the synthesis methods. Valentina Vogt and Alexander Oing are thanked for their participation in synchrotron XAS experiments. Y. X. acknowledges the support provided by Prof. D. Tilley. The authors are grateful to Prof. Dimos Poulikakos for his continuous support and encouragement. D. P. acknowledges the financial support of the Swiss National Science Foundation (SNSF) within the Swiss Postdoctoral Fellowship (grant no. TMPFP2_224646). The Swiss National Science Foundation (grant no. 200021_196943/1) and the Swiss Office of Energy (grant no. S1/502898–01) are acknowledged for financial support. This work was created as part of NCCR Catalysis (grant no. 180544), a National Centre of Competence in Research funded by the Swiss National Science Foundation.

References

1. T. Reier, M. Oezaslan and P. Strasser, ACS Catal., 2012, 2, 1765–1772.
2. Y.-R. Zheng, J. Vernieres, Z. Wang, K. Zhang, D. Hochfilzer, K. Krempl, T.-W. Liao, F. Presel, T. Altantzis, J. Fatermans, S. B. Scott, N. M. Secher, C. Moon, P. Liu, S. Bals, S. Van Aert, A. Cao, M. Anand, J. K. Nørskov, J. Kibsgaard and I. Chorkendorff, Nat. Energy, 2021, 7, 55–64.
3. S. Cherevko, S. Geiger, O. Kasian, N. Kulyk, J.-P. Grote, A. Savan, B. R. Shrestha, S. Merzlikin, B. Breitbach, A. Ludwig and K. J. J. Mayrhofer, Catal. Today, 2016, 262, 170–180.
4. D. Y. Kuo, J. K. Kawasaki, J. N. Nelson, J. Kloppenburg, G. Hautier, K. M. Shen, D. G. Schlom and J. Suntivich, J. Am. Chem. Soc., 2017, 139, 3473–3479.
5. H. N. Nong, L. J. Falling, A. Bergmann, M. Klingenhof, H. P. Tran, C. Spori, R. Mom, J. Timoshenko, G. Zichittella, A. Knop-Gericke, S. Piccinin, J. Perez-Ramirez, B. R. Cuenya, R. Schlogl, P. Strasser, D. Teschner and T. E. Jones, Nature, 2020, 587, 408–413.
6. H. N. Nong, T. Reier, H.-S. Oh, M. Gliech, P. Paciok, T. H. T. Vu, D. Teschner, M. Heggen, V. Petkov, R. Schlögl, T. Jones and P. Strasser, Nat. Catal., 2018, 1, 841–851.
7. Y. Lee, J. Suntivich, K. J. May, E. E. Perry and Y. Shao-Horn, J. Phys. Chem. Lett., 2012, 3, 399–404.
8. S. B. Scott, J. E. Sorensen, R. R. Rao, C. Moon, J. Kibsgaard, Y. Shao-Horn and I. Chorkendorff, Energy Environ. Sci., 2022, 15, 1988–2001.
9. C. Liang, R. R. Rao, K. L. Svane, J. H. L. Hadden, B. Moss, S. B. Scott, M. Sachs, J. Murawski, A. M. Frandsen, D. J. Riley, M. P. Ryan, J. Rossmeisl, J. R. Durrant and I. E. L. Stephens, Nat. Catal., 2024, 7, 763–775.
10. D. F. Abbott, D. Lebedev, K. Waltar, M. Povia, M. Nachtegaal, E. Fabbri, C. Copéret and T. J. Schmidt, Chem. Mater., 2016, 28, 6591–6604.
11. S. Geiger, O. Kasian, B. R. Shrestha, A. M. Mingers, K. J. J. Mayrhofer and S. Cherevko, J. Electrochem. Soc., 2016, 163, F3132–F3138.
12. S. Cherevko, A. R. Zeradjanin, A. A. Topalov, N. Kulyk, I. Katsounaros and K. J. J. Mayrhofer, ChemCatChem, 2014, 6, 2219–2223.
13. O. Schmidt, A. Gambhir, I. Staffell, A. Hawkes, J. Nelson and S. Few, Int. J. Hydrogen Energy, 2017, 42, 30470–30492.
14. L. C. Seitz, C. F. Dickens, K. Nishio, Y. Hikita, J. Montoya, A. Doyle, C. Kirk, A. Vojvodic, H. Y. Hwang and J. K. Norskov, Science, 2016, 353, 1011–1014.
15. O. Diaz-Morales, S. Raaijman, R. Kortlever, P. J. Kooyman, T. Wezendonk, J. Gascon, W. T. Fu and M. T. Koper, Nat. Commun., 2016, 7, 12363.
16. D. Lebedev, M. Povia, K. Waltar, P. M. Abdala, I. E. Castelli, E. Fabbri, M. V. Blanco, A. Fedorov, C. Copéret and N. Marzari, Chem. Mater., 2017, 29, 5182–5191.
17. G. Wan, J. W. Freeland, J. Kloppenburg, G. Petretto, J. N. Nelson, D.-Y. Kuo, C.-J. Sun, J. Wen, J. T. Diulus, G. S. Herman, Y. Dong, R. Kou, J. Sun, S. Chen, K. M. Shen, D. G. Schlom, G.-M. Rignanese, G. Hautier, D. D. Fong, Z. Feng, H. Zhou and J. Suntivich, Sci. Adv., 2021, 7, eabc7323.
18. R. Zhang, N. Dubouis, M. Ben Osman, W. Yin, M. T. Sougrati, D. A. D. Corte, D. Giaume and A. Grimaud, Angew Chem. Int. Ed. Engl., 2019, 58, 4571–4575.
19. S. Geiger, O. Kasian, M. Ledendecker, E. Pizzutilo, A. M. Mingers, W. T. Fu, O. Diaz-Morales, Z. Li, T. Oellers, L. Fruchter, A. Ludwig, K. J. J. Mayrhofer, M. T. M. Koper and S. Cherevko, Nat. Catal., 2018, 1, 508–515.
20. L. C. Seitz, C. F. Dickens, K. Nishio, Y. Hikita, J. Montoya, A. Doyle, C. Kirk, A. Vojvodic, H. Y. Hwang, J. K. Norskov and T. F. Jaramillo, Science, 2016, 353, 1011–1014.
21. A. Grimaud, A. Demortière, M. Saubanère, W. Dachraoui, M. Duchamp, M.-L. Doublet and J.-M. Tarascon, Nat. Energy, 2016, 2, 16189.
22. C. W. Song, J. Lim, H. B. Bae and S.-Y. Chung, Energy Environ. Sci., 2020, 13, 4178–4188.
23. A. Minguzzi, O. Lugaresi, E. Achilli, C. Locatelli, A. Vertova, P. Ghigna and S. Rondinini, Chem. Sci., 2014, 5, 3591–3597.
24. V. Pfeifer, T. E. Jones, J. J. Velasco Velez, C. Massue, M. T. Greiner, R. Arrigo, D. Teschner, F. Girgsdies, M. Scherzer, J. Allan, M. Hashagen, G. Weinberg, S. Piccinin, M. Havecker, A. Knop-Gericke and R. Schlögl, Phys. Chem. Chem. Phys., 2016, 18, 2292–2296.
25. Z. Pavlovic, C. Ranjan, M. van Gastel and R. Schlogl, Chem. Commun., 2017, 53, 12414–12417.
26. E. Willinger, C. Massue, R. Schlogl and M. G. Willinger, J. Am. Chem. Soc., 2017, 139, 12093–12101.
27. C. Daiane Ferreira da Silva, F. Claudel, V. Martin, R. Chattot, S. Abbou, K. Kumar, I. Jiménez-Morales, S. Cavaliere, D. Jones, J. Rozière, L. Solà-Hernandez, C. Beauger, M. Faustini, J. Peron, B. Gilles, T. Encinas, L. Piccolo, F. H. Barros de Lima, L. Dubau and F. Maillard, ACS Catal., 2021, 11, 4107–4116.
28. V. Pfeifer, T. E. Jones, J. J. Velasco Velez, R. Arrigo, S. Piccinin, M. Havecker, A. Knop-Gericke and R. Schlogl, Chem. Sci., 2017, 8, 2143–2149.
29. B. Lu, C. Wahl, R. dos Reis, J. Edgington, X. K. Lu, R. Li, M. E. Sweers, B. Ruggiero, G. T. K. K. Gunasooriya, V. Dravid and L. C. Seitz, Nat. Catal., 2024, 7, 868–877.
30. P. Lettenmeier, L. Wang, U. Golla-Schindler, P. Gazdzicki, N. A. Canas, M. Handl, R. Hiesgen, S. S. Hosseiny, A. S. Gago and K. A. Friedrich, Angew Chem. Int. Ed. Engl., 2016, 55, 742–746.
31. S. Kumari, B. P. Ajayi, B. Kumar, J. B. Jasinski, M. K. Sunkara and J. M. Spurgeon, Energy Environ. Sci., 2017, 10, 2432–2440.
32. Y. Pi, Q. Shao, P. Wang, J. Guo and X. Huang, Adv. Funct. Mater., 2017, 27, 1700886.
33. O. Kasian, J. P. Grote, S. Geiger, S. Cherevko and K. J. J. Mayrhofer, Angew Chem. Int. Ed. Engl., 2018, 57, 2488–2491.
34. S. Fierro, T. Nagel, H. Baltruschat and C. Comninellis, Electrochem. Commun., 2007, 9, 1969–1974.
35. O. Kasian, S. Geiger, T. Li, J.-P. Grote, K. Schweinar, S. Zhang, C. Scheu, D. Raabe, S. Cherevko, B. Gault and K. J. J. Mayrhofer, Energy Environ. Sci., 2019, 12, 3548–3555.
36. A. Zagalskaya and V. Alexandrov, ACS Catal., 2020, 10, 3650–3657.
37. L. Fu, P. Cai, G. Cheng and W. Luo, Sustainable Energy Fuels, 2017, 1, 1199–1203.
38. J. J. Velasco-Velez, E. A. Carbonio, C. H. Chuang, C. J. Hsu, J. F. Lee, R. Arrigo, M. Havecker, R. Wang, M. Plodinec, F. R. Wang, A. Centeno, A. Zurutuza, L. J. Falling, R. V. Mom, S. Hofmann, R. Schlogl, A. Knop-Gericke and T. E. Jones, J. Am. Chem. Soc., 2021, 143, 12524–12534.
39. J. Ruiz Esquius, D. J. Morgan, G. Algara Siller, D. Gianolio, M. Aramini, L. Lahn, O. Kasian, S. A. Kondrat, R. Schlogl, G. J. Hutchings, R. Arrigo and S. J. Freakley, J. Am. Chem. Soc., 2023, 145, 6398–6409.
40. J. A. Arminio-Ravelo, A. W. Jensen, K. D. Jensen, J. Quinson and M. Escudero-Escribano, ChemPhysChem, 2019, 20, 2956–2963.
41. M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, National Association of Corrosion Engineers, 1974.
42. C. He, C. Ma, J. Xia, H. Zhang, S. Han, Y. Tian, A. L. Wang, X. Meng, W. Cao and Q. Lu, Adv. Funct. Mater., 2024, 34, 2311683.
43. 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.
44. A. Grimaud, O. Diaz-Morales, B. Han, W. T. Hong, Y. L. Lee, L. Giordano, K. A. Stoerzinger, M. T. M. Koper and Y. Shao-Horn, Nat. Chem., 2017, 9, 457–465.
45. D. B. Trimarco, S. B. Scott, A. H. Thilsted, J. Y. Pan, T. Pedersen, O. Hansen, I. Chorkendorff and P. C. K. Vesborg, Electrochim. Acta, 2018, 268, 520–530.
46. S. B. Scott, J. Kibsgaard, P. C. K. Vesborg and I. Chorkendorff, Electrochim. Acta, 2021, 374.
47. M. Hüppauff and B. Lengeler, J. Electrochem. Soc., 1993, 140, 598.
48. R. K. Pittkowski, S. Punke, A. S. Anker, A. Bornet, N. P. L. Magnard, N. Schlegel, L. G. Graversen, J. Quinson, A. Dworzak, M. Oezaslan, J. J. K. Kirkensgaard, M. Mirolo, J. Drnec, M. Arenz and K. M. O. Jensen, J. Am. Chem. Soc., 2024, 146, 27517–27527.
49. H. Guo, Z. Fang, H. Li, D. Fernandez, G. Henkelman, S. M. Humphrey and G. Yu, ACS Nano, 2019, 13, 13225–13234.
50. H. Luo, F. Lin, Q. Zhang, D. Wang, K. Wang, L. Gu, M. Luo, F. Lv and S. Guo, J. Am. Chem. Soc., 2024, 146, 19327–19336.
51. Y. Chen, H. Li, J. Wang, Y. Du, S. Xi, Y. Sun, M. Sherburne, J. W. Ager 3rd, A. C. Fisher and Z. J. Xu, Nat. Commun., 2019, 10, 572.
52. Q. Dang, H. Lin, Z. Fan, L. Ma, Q. Shao, Y. Ji, F. Zheng, S. Geng, S. Z. Yang, N. Kong, W. Zhu, Y. Li, F. Liao, X. Huang and M. Shao, Nat. Commun., 2021, 12, 6007.
53. J. Gao, C. Q. Xu, S. F. Hung, W. Liu, W. Cai, Z. Zeng, C. Jia, H. M. Chen, H. Xiao, J. Li, Y. Huang and B. Liu, J. Am. Chem. Soc., 2019, 141, 3014–3023.
54. Y. Pi, Y. Xu, L. Li, T. Sun, B. Huang, L. Bu, Y. Ma, Z. Hu, C. W. Pao and X. Huang, Adv. Funct. Mater., 2020, 30, 2004375.
55. L. Zu, X. Qian, S. Zhao, Q. Liang, Y. E. Chen, M. Liu, B. J. Su, K. H. Wu, L. Qu, L. Duan, H. Zhan, J. Y. Zhang, C. Li, W. Li, J. Y. Juang, J. Zhu, D. Li, A. Yu and D. Zhao, J. Am. Chem. Soc., 2022, 144, 2208–2217.
56. G. Wu, X. Zheng, P. Cui, H. Jiang, X. Wang, Y. Qu, W. Chen, Y. Lin, H. Li, X. Han, Y. Hu, P. Liu, Q. Zhang, J. Ge, Y. Yao, R. Sun, Y. Wu, L. Gu, X. Hong and Y. Li, Nat. Commun., 2019, 10, 4855.
57. J. A. Gauthier, C. F. Dickens, L. D. Chen, A. D. Doyle and J. K. Nørskov, J. Phys. Chem. C, 2017, 121, 11455–11463.
58. D. A. Kuznetsov, B. Han, Y. Yu, R. R. Rao, J. Hwang, Y. Roman-Leshkov and Y. Shao-Horn, Joule, 2018, 2, 225–244.

---

Footnote

† These authors contribute equally.

---