Modulating the interfacial solvation structure to promote hydroxyl migration for alkaline hydrogen oxidation

Abstract

Alkaline hydrogen oxidation reaction (HOR) kinetics are fundamentally governed by electrode-electrolyte interfacial water structure and hydrogen-bond (H-bond) networks, yet their dynamic regulatory mechanism remains elusive. Here we show that precise modulation of the P coordination environment of Ir-based catalysts effectively enhances alkaline HOR catalytic performance. A series of Ir-based model catalysts with tunable P-coordination numbers are synthesized on nitrogen-doped carbon (NC) (IrP₂-Ir₂P@NC, IrP₂@NC, Ir@NC). The optimal Ir-P coordination (5.1 ± 0.7) induces controlled partial dissolution of Ir cations, a dynamic phenomenon rarely exploited in alkaline HOR catalysis. This effect reconstructs the interfacial H-bond network via enriched gap-H₂O and strengthened network connectivity, while optimizing hydrogen and hydroxyl adsorption to accelerate the Volmer step. The optimized IrP₂-Ir₂P@NC delivers a mass activity of 1.72 mA μg⁻¹ and a prominent alkaline exchange membrane fuel cell (AEMFC) peak power density of 1.59 W cm⁻². Ab initio molecular dynamics simulations confirm modified solvation structure elevates interfacial H-bond density, thereby elucidating the previously unresolved mechanism by which cation dissolution dynamically templates the H-bond network for enhanced HOR performance.

The kinetics of alkaline hydrogen oxidation reaction depends on H-bond network at catalyst surfaces. Here, the authors tune the P coordination of Ir-based catalysts, which triggers metal leaching, reorganizes interfacial H-bond networks and improves fuel cell performance.

Introduction

Electrochemical energy conversion presents potential solutions to global challenges such as environmental pollution and the energy crisis^1–3. Anion exchange membrane fuel cells (AEMFCs) stand out due to their ability to directly convert hydrogen fuel-an environmentally friendly, zero-carbon energy source with high energy density-into electrical energy under alkaline conditions^2,4–7. Unfortunately, the transition from an acidic to an alkaline medium typically results in a two orders magnitude decrease in the HOR kinetics, which hinders its widespread commercial application^8,9. Consequently, substantial efforts are devoted to exploring efficient catalysts for alkaline HOR. Advancements in-situ characterization have enabled the observation of adsorbed hydroxyl (OH_ad) during the HOR reaction process¹⁰. OH_ad species can be used as descriptors for the HOR reaction process^11–14. However, the interaction between key intermediate hydroxyl groups and interfacial water is often overlooked.

Currently, the Volmer step involving hydroxyl adsorption (OH⁻ - e⁻ → OH_ad) and the subsequent water formation (OH_ad + H_ad → H₂O) are generally considered as the rate-determining step (RDS)^11,15,16. Recent investigations indicate that adsorbed hydroxyl groups are critical determinants of the kinetics of alkaline HOR¹⁷. From a molecular perspective, there exist complex interactions between the OH_ad groups adsorbed over the electrode surface and the adjacent interfacial water molecules^17–19. These interactions are exhibited not only by intermolecular electrostatic gravitational forces and van der Waals forces, but more importantly, the interfacial water molecules can regulate the hydroxyl groups through the formation of hydrogen bond networks^17,20–22. Currently, tuning the electronic structure through alloying, doping, and defect engineering is employed to facilitate the formation of H-OH bonds and promote HOR kinetics^1,2,23–26. However, there are still obvious limitations in the current research on the alkaline HOR field. Initially, most studies focus on optimizing the binding energy of OH_ad by regulating the electronic structure of the electrode, while neglecting the quantitative analysis of the dynamic interactions between adsorbed intermediates (OH_ad, H_ad) and interfacial water molecules²⁷. Second, the influence of dissolved metals on interfacial water and hydrogen bond networks has been overlooked²⁸. Against this background, exploring the interactions between adsorbed intermediates and interfacial water molecules, along with the influence of metal cations on interfacial water and hydrogen bond networks, provides a promising approach to advance high-performance HOR catalysts.

In this work, a series of catalysts (including IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC) were designed. Initially, the adsorption behaviors of different catalysts were investigated via DFT simulations, and it was found that IrP₂-Ir₂P@NC exhibited the optimal adsorption of *H and *OH adsorptions, which reduced the energy barrier for water formation and accelerated the Volmer step. The Ir-P component of these catalysts was characterized by being only affected by the P coordination number, which served to investigate the effect of different P coordination numbers on the alkaline HOR of these Ir-P compounds. Under H₂-O₂ atmosphere, IrP₂-Ir₂P@NC was employed as the anode of the AEMFC, delivering a peak power density of 1.59 W cm⁻². Based on Coulomb’s law, quasi-in-situ XPS and quasi-in situ absorption spectroscopy, it was concluded that an appropriate amount of P coordination weakened the strength of Ir-P bonds, leading to partial dissolution of Ir and thus achieving a solvation enhancement effect. Combined with in-situ characterizations, this solvation enhancement effect was found to help increase the content of gap-H₂O and strengthen the connectivity of the hydrogen bond network, which was beneficial for adjusting the point of zero charge (PZC) and promoting hydroxyl adsorption. AIMD further confirmed that the solvation enhancement effect from partial Ir dissolution strengthened the hydrogen bond network connectivity, thereby effectively improving the catalytic activity of alkaline HOR.

Results

DFT theoretical predictions

Based on our proposed concept, a series of catalytic models with different P coordination configurations (IrP₂-Ir₂P@NC, IrP₂@NC, Ir@NC) were designed (Supplementary Fig. 1 and Supplementary Data 1). The corresponding HOR processes were systematically investigated via density functional theory (DFT) simulations. The Crystal orbital hamiltonian population (COHP) values of different catalysts were calculated separately using the OH adsorption models and CO adsorption models (Supplementary Figs. 2, 3). The results demonstrated that the introduction of an appropriate amount of P effectively improved the OH adsorption capacity and CO poisoning resistance of the catalyst (Fig. 1a–c and Supplementary Figs. 4, 5 and 6). The differential charge analysis via Supplementary Fig. 7 indicated that the electron clouds of IrP₂@NC and Ir@NC were mainly concentrated at the bottom, whereas the electron cloud of IrP₂-Ir₂P@NC was concentrated not only at the crystal interface but also at the bottom, indicating that an appropriate amount of P coordination was helpful for enhancing the intense electronic interaction between P and Ir. Compared with IrP₂@NC and Ir@NC, IrP₂-Ir₂P@NC exhibited a higher occupancy of total density of states (DOS) in proximity to the Fermi level. This indicates that IrP₂-Ir₂P@NC has more electron states accessible for electron occupation and mobility, thereby enhancing the electronic conductivity (Fig. 1d). In Fig. 1e, the d-band centers of IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC were −2.12 eV, −2.23 eV, and −1.93 eV, respectively. This result indicated that appropriate P coordination was beneficial for enhancing hydroxyl adsorption²⁹.

Fig. 1. Theoretical calculation results.

a–c Calculated COHP of IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC. d The density of state (DOS) plots. e PDOS and Ir d-band center values, f Calculated OHBEs for IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC. g CO adsorption energies, h Gibbs free energy profiles of basic HOR on the surfaces. Source data for Fig. 1 are provided as a Source Data file.

Furthermore, the theoretical model of CO-metal bonding revealed the mechanism of electronic structure interaction (Supplementary Fig. 8)^30,31. Partial electrons from the 5σ orbital of the CO molecule transferred to the metal d orbitals, while electrons from the metal d orbitals back-donated to the 2π* antibonding orbital of CO. Thus, the interaction could be reflected by tuning the d-band center and enhancing the OH adsorption capability. The d-band center of IrP₂-Ir₂P@NC lay between those of IrP₂@NC and Ir@NC. This shift originated from charge transfer induced by tuning the number of P coordinations, which synergistically optimized the adsorption strengths of H and OH. Accordingly, the back-donation of electrons from Ir d orbitals to the 2π* antibonding orbital of CO was suppressed, endowing IrP₂-Ir₂P@NC with moderate resistance toward CO poisoning³². The binding affinities of essential intermediates were calculated. Supplementary Figs. 2, 9 display the adsorption sites of IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC. As shown in Supplementary Fig. 10, IrP₂-Ir₂P@NC demonstrated the lowest hydrogen binding energy of −0.81 eV, in contrast to IrP₂@NC (−1.28 eV) and Ir@NC (−1.19 eV), indicating that the HOR process was favorable. Additionally, as derived from Fig. 1f, IrP₂-Ir₂P@NC exhibited the highest hydroxyl binding energy of −1.73 eV. This was consistent with the results obtained from COHP. Therefore, the above results indicated that appropriate P coordination enabled IrP₂-Ir₂P@NC to achieve optimal HBE and OHBE. To further explore the regulatory effect of applied voltage on the electronic structure and adsorption behavior of catalysts, we calculated and analyzed the DOS and OHBE of different catalysts at a potential of 0.1 V, as displayed in Supplementary Figs. 11 and 12. As the applied voltage increased, the charge quantity of the catalyst system gradually decreased, the number of electrons in the bonding state was reduced, and the overall energy band structure shifted downward. Meanwhile, it was observed that the increase in voltage significantly strengthened the hydroxyl adsorption strength of the catalysts, following the order of IrP₂-Ir₂P@NC > Ir@NC > IrP₂@NC. Overall, appropriate P coordination effectively modulated the electronic structure of the catalysts and further enhanced their hydroxyl adsorption capability. Additionally, the adsorption energies of CO* on different catalysts were calculated, and the adsorption sites are shown in Supplementary Fig. 3. The CO* adsorption energies followed the order: IrP₂-Ir₂P@NC (−1.29 eV) < Ir@NC (−1.49 eV) < IrP₂@NC (−1.53 eV), indicating that appropriate P coordination helped tune the electronic structure of Ir, thereby enhancing its CO poisoning resistance (Fig. 1g). This was consistent with the results obtained from the aforementioned COHP and d‑band center analyses. Additionally, the energy profiles of IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC during the alkaline HOR process were calculated, and the theoretical models involved. As expected, IrP₂-Ir₂P@NC exhibited the lowest water formation energy barrier of 1.32 eV, compared to IrP₂@NC (1.69 eV) and Ir@NC (1.58 eV), indicating improved HOR kinetics (Fig. 1h). Notably, DFT adopted simplified computational approximations and could not fully replicate the complex environments of real electrodes, which imposed inherent restrictions on the quantitative interpretation of experimental observations.

Morphological and structural characterizations

Guided by the DFT simulations. IrP₂-Ir₂P@NC catalytic model was rationally constructed (Fig. 2a). Initially, Zn(NO₃)₂·6H₂O and 2-methylimidazole were individually solubilized in methanol. After mixing, ZIF-8 crystals were synthesized by coordination following the mixing process. Subsequently, ZIF-8 was pyrolyzed to convert into nitrogen-doped carbon (NC) with a porous structure (Supplementary Fig. 13). Then, under ultrasonic conditions, Ir species and triphenylphosphine were added to the mixed solution containing NC. Finally, the target catalyst (denoted as IrP₂-Ir₂P@NC) was synthesized via pyrolysis. Initially, the successful preparation of Ir-based phosphides and metallic Ir were confirmed by X-ray diffraction (XRD) patterns. They were consistent with Ir (JCPDS: 87-0715), IrP₂ (JCPDS: 89-3985), and Ir₂P (JCPDS: 77-0299), respectively (Fig. 2b and Supplementary Fig. 14). Raman spectroscopy was used to evaluate the catalysts structural defects. Two bands within the observed range, D (1319 cm⁻¹) and G (1597 cm⁻¹), corresponded to disordered sp³ and sp² hybridized carbon atoms, respectively³³. Compared with IrP₂@NC and Ir@NC, IrP₂-Ir₂P@NC demonstrated a higher intensity ratio (I_D/I_G), demonstrating that the appropriate doping of P loading promoted the generation of more carbon defects (Supplementary Fig. 15)³⁴. The N₂ adsorption-desorption isotherm showed that IrP₂-Ir₂P@NC (719.29 m² g⁻¹) exhibited a larger specific surface area than IrP₂@NC (446.35 m² g⁻¹) and Ir@NC (579.66 m² g⁻¹), which helped expose more active sites and accelerate reaction kinetics (Supplementary Figs. 16, 17 and 18)³⁵.

Fig. 2. Synthesis and characterization of IrP₂-Ir₂P@NC.

a Schematic illustration of the general synthetic process for IrP₂-Ir₂P@NC (black: C; blue: N; light red: Ir; light purple: P). b XRD patterns of IrP₂-Ir₂P@NC, IrP₂@NC, Ir@NC. c AC HAADF-STEM image of IrP₂-Ir₂P@NC (pink: Ir; white: P). d, e Intensity profiles of IrP₂-Ir₂P@NC acquired from (c). f, g IFFT images at different areas of IrP₂-Ir₂P@NC acquired from (Supplementary Fig. 21). h SAED pattern of IrP₂-Ir₂P@NC. i Ir L₃-edge XANES spectra of IrP₂-Ir₂P@NC, IrP₂@NC, Ir@NC, Ir foil and IrO₂. j Determination of Ir oxidation in IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC with reference samples of Ir foil and IrO₂ to ascertain the average oxidation state acquired from (i). k Corresponding Fourier transformed the EXAFS spectrum spectra of the IrP₂-Ir₂P@NC, IrP₂@NC, Ir@NC, Ir foil and IrO₂. l Ir-P coordination number and Ir-P bond length of IrP₂-Ir₂P@NC and IrP₂@NC IFFT images at different areas of IrP₂-Ir₂P@NC acquired from (Supplementary Table 1). Source data for Fig. 2 are provided as a Source Data file.

To further examine the atomic-level characteristics of the catalyst, field-emission scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were used. The NC structure was observed by SEM to exhibit a dodecahedral morphology (Supplementary Fig. 19). TEM was used to observe the morphology of the materials. Loading the catalyst onto NC resulted in the generation of particles. As demonstrated in Supplementary Fig. 20, the average particle diameter of IrP₂-Ir₂P@NC was 11.74 nm. To further investigate the local atomic arrangements, we utilized aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and TEM. AC AADF-STEM was employed to observe lattice fringes with spacings of 0.225 and 0.312 nm corresponding to the (−212) plane of IrP₂ and the (111) plane of Ir₂P, respectively (Fig. 2c–e). Moreover, As illustrated in Fig. 2f, g and Supplementary Fig. 21 illustrate that TEM analysis indicated lattice spacings of 0.233 and 0.199 nm, which are ascribed to the (−202) crystal plane of IrP₂ and the (−202) crystal plane of Ir₂P, respectively. Selected area electron diffraction (SAED) examination revealed diffraction patterns associated with IrP₂ (−102), IrP₂ (−312) and Ir₂P (220) (Fig. 2h). The results validated the effective synthesis of IrP₂-Ir₂P@NC. As depicted in Supplementary Figs. 22, 23 and 24, illustrate lattice fringes with interplanar lengths distances of 0.205 nm, 0.183 nm and 0.141 nm, which were consistent with the (220) plane of Ir₂P, (−113) plane of IrP₂ and the (220) plane of Ir, respectively. These experimental results were in accordance with the XRD data, providing another evidence for the successful synthesis of IrP₂ and Ir₂P. Additionally, the elemental mapping images verified the homogeneous distribution of C, N, P and Ir (Supplementary Figs. 25, 26).

To qualitatively detect the presence of vacancies in the samples, electron paramagnetic resonance (EPR) measurements were performed. Supplementary Fig. 27 illustrated that IrP₂-Ir₂P@NC displayed a signal peak compared to IrP₂@NC and Ir@NC, signifying the existence of P vacancies³⁶. Additionally, X-ray photoelectron spectroscopy (XPS) was used to analyze the surface composition and valence states. As illustrated in Supplementary Fig. 28, in the Ir 4 f spectrum, the peaks at approximately 60.25 eV and 60.94 eV corresponded to the Ir-P and Ir-Ir bonds, respectively^37,38. Compared with other materials, the binding energy of Ir-P bonds in IrP₂-Ir₂P@NC was lower than that in IrP₂@NC but higher than that in Ir₂P@NC. This indicated that with increasing P coordination number, the charge of Ir decreased while its valence state increased slightly. Additionally, the Ir 4 f spectrum of Ir@NC is shown in Supplementary Fig. 29.

To further explore the electronic states and coordination environments of metal species in IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC, synchrotron X-ray absorption spectroscopy (XAS) was employed. Normalized X-ray absorption near-edge spectroscopy (XANES) at the Ir L-edge was performed with Ir foil and IrO₂ as references (Fig. 2i). Based on the scaling relationship between absorption energy and Ir valence state, the average valence state of Ir in IrP₂-Ir₂P@NC was 1.33, lying between 1.76 in IrP₂@NC and 0.45 in Ir@NC. This indicated that the introduction of P promoted electron loss of Ir (Fig. 2j)^39,40. Furthermore, Ir L-edge extended X-ray absorption fine structure (EXAFS) spectra were collected to analyze the local environment of Ir atoms (Fig. 2k). The EXAFS spectra of Ir@NC displayed a distinct peak at 2.42 Å, akin to that of Ir foil, which could be attributed to the Ir-Ir bond. With the introduction of P, IrP₂-Ir₂P@NC and IrP₂@NC exhibited peaks at 2.01 Å and 1.97 Å, respectively, which could be attributed to the Ir-P bond. Specifically, compared with IrP₂@NC, the Ir-P signal peak in IrP₂-Ir₂P@NC exhibited a minor positive shift, signifying a modest elongation of the Ir-P bond in IrP₂-Ir₂P@NC. Additional curve fitting was performed to determine the specific structural parameters around the Ir atoms (Supplementary Figs. 30, 31 and 32). The analysis revealed that the coordination number of Ir-P in IrP₂-Ir₂P@NC was 5.1, which was lower than that in IrP₂@NC (CN = 5.9) (Supplementary Table 1). The distinct coordination environments of IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC were further investigated, aligning with the aforementioned EXAFS results (Supplementary Figs. 33-37). Figure 2l demonstrated that the Ir-P bond length progressively diminished as the coordination number of P in the catalyst increased⁴¹.

HOR performance in alkaline media and catalytic mechanism unraveling

The electrocatalytic performance of the prepared catalysts HOR was evaluated in H₂-saturated 0.1 M KOH using a rotating disk electrode (RDE) in a standard three-electrode system. We initially examined the influence of various reaction conditions on the catalytic performance of IrP₂-Ir₂P@NC. We determined that the catalytic performance of IrP₂-Ir₂P@NC reached a peaked at 800 °C with an Ir concentration of 10.26 wt.% (Supplementary Figs. 38, 39 and Table 2). Figure 3a and Supplementary Fig. 40 demonstrates that, IrP₂-Ir₂P@NC displayed markedly enhanced HOR activity. The measured activity followed the order Ir₂P@NC > Ir₂P@NC > Ir@NC > IrP₂@NC, further indicating that appropriate P coordination contributed to enhanced catalytic activity. In the N₂-saturated control experiment, the anodic current was confirmed to originate from the oxidation of H₂ (Supplementary Fig. 41)⁴². The HOR polarization curves of IrP₂-Ir₂P@NC were examined at different rotation speeds were studied. The current density increased as the rotation speed increased, signifying the mass transfer process of H₂ (Supplementary Fig. 42a)⁴³. According to the Koutecky-Levich equation, the calculated slope of IrP₂-Ir₂P@NC was 5.56 cm² mA⁻¹ s^−1/2, which was closely consistent with the theoretical value of 4.87 cm² mA⁻¹ s^−1/2, indicating a two-electron transfer process (Supplementary Fig. 42b)⁴³. The asymmetric behavior observed from the Tafel plot suggested that IrP₂-Ir₂P@NC aligned with the Heyrovsky-Volmer mechanism, with the Volmer step recognized as the RDS (Fig. 3b)⁴⁴. Figure 3c summarized the j_k values of different catalysts. The j_k value of IrP₂-Ir₂P@NC was 31.56 mA cm⁻², exceeding those of IrP₂@NC (1.22 mA cm⁻²), Ir@NC (3.24 mA cm⁻²) and Pt/C (10.21 mA cm⁻²). IrP₂-Ir₂P@NC demonstrated mass activity (1.72 mA μg⁻¹), surpassing the majority of documented HOR catalysts (Supplementary Fig. 43 and Supplementary Table 3, 4). Subsequently, an accelerated stability test was implemented by chronoamperometry in H₂-saturated 0.1 M KOH. Supplementary Fig. 44 demonstrated that IrP₂-Ir₂P@NC displayed a milder reduction in current density in comparison to IrP₂@NC, Ir@NC and Pt/C, underscoring the acceptable long-term stability of IrP₂-Ir₂P@NC. Furthermore, we analyzed the dissolution amount of metallic Ir combined with the stability trend, as shown in Supplementary Fig. 45. In the initial stage of the reaction, the activity of IrP₂-Ir₂P@NC increased continuously with the dissolution of metallic Ir. The catalytic activity reached its maximum when the Ir dissolution amount reached 6.26 ppb, and then began to decrease gradually. When the dissolution amount reached 19.22 ppb, the catalytic activity of IrP₂-Ir₂P@NC stabilized. To evaluate the CO tolerance of the catalyst, we performed relevant experiments under an atmosphere of H₂/1000 ppm CO. The results showed that the polarization curve of IrP₂-Ir₂P@NC almost overlapped with the initial curve. In contrast, the current density of Pt/C decreased sharply under the same CO concentration, indicating that IrP₂-Ir₂P@NC exhibited moderate resistance to CO poisoning⁴⁵ (Fig. 3d). Furthermore, to further evaluate the CO tolerance of the catalyst, we performed chronoamperometric response measurements. As shown in Supplementary Fig. 46, the current retention of IrP₂-Ir₂P@NC was significantly higher than that of commercial Pt/C, which further confirmed the moderate CO poisoning resistance of IrP₂-Ir₂P@NC.

Fig. 3. Electrocatalytic alkaline HOR performances.

a Polarization curves for HOR over IrP₂-Ir₂P@NC, IrP₂@NC, Ir@NC and commercial Pt/C catalysts. Operation conditions: 0.1 M KOH electrolyte (pH = 12.9 ± 0.01), scan rate of 10 mV s⁻¹, Ir mass loading of 47.59 μg_Ir cm⁻², rotating speed of 1600 rpm, continuous H₂ flow rate of 25 sccm, working electrode: 5 mm-diameter GCE with geometric area of 0.196 cm², test temperature of 25 ± 2 °C. b Tafel plots. c Comparison of j_k for studied electrocatalysts acquired from (b). d HOR performance of IrP₂-Ir₂P@NC and Pt/C in pure H₂ and H₂/1000 ppm CO. e CO stripping curves of IrP₂-Ir₂P@NC, IrP₂@NC, Ir@NC and Pt/C. f Schematic illustration of the AEMFC. g Polarization and power density curves of H₂/O₂ and H₂/Air AEMFCs with IrP₂-Ir₂P@NC (0.40 mg_Ir cm⁻²) as the anode and commercial Pt/C (0.50 mg_Pt cm⁻²) as the cathode. Operation conditions: cell temperature of 80 °C, gas back pressure of 0.2 MPa on both sides of the cell, and fully humidified H₂ and O₂ fed at a flow rate of 1000 sccm. Source data for Fig. 3 are provided as a Source Data file.

To further assess the HOR catalytic activity of the catalyst, we examined it by probing the binding strength of OH. The adsorption behavior of OH_ad was analyzed by CO stripping²⁹. Figure 3e illustrates, IrP₂-Ir₂P@NC (0.66 V) had a greater negative value than IrP₂@NC (0.82 V), Ir@NC (0.79 V), and Pt/C (0.71 V), signifying that IrP₂-Ir₂P@NC had relatively higher OH binding energy (OHBE), conforming to the reported HOR activity^46,47. To investigate the practical application of IrP₂-Ir₂P@NC in fuel cells, we employed IrP₂-Ir₂P@NC as the anode catalyst and commercial Pt/C as the cathode catalyst for an anion exchange membrane fuel cell (AEMFC) (Fig. 3f). Notably, the AEMFC with IrP₂-Ir₂P@NC delivered peak power densities (PPD) of 1.59 W cm⁻² and 1.27 W cm⁻² under H₂-O₂ and H₂-Air atmospheres, respectively, and such performance was competitive against values from recently documented HOR catalysts (Fig. 3g, Supplementary Fig. 47 and Supplementary Table 5, 6). To further verify the practical durability of the catalyst, we performed a long-term AEMFC stability test under H₂/O₂ atmosphere at a voltage of 0.7 V. As shown in Supplementary Fig. 48, IrP₂-Ir₂P@NC operated continuously for 100 h and maintained high stability. In addition, ICP-MS analysis of the electrolyte after the long-term reaction showed that the Ir dissolution amount was approximately 104.3 ppb. This result further demonstrated that appropriate Ir dissolution contributed to long-term and efficient catalysis. To investigate the changes of the catalyst after the long-term stability test, we characterized the post-reaction sample. As displayed in Supplementary Figs. 49, 50 and 51, the structure and composition of the catalyst remained intact without obvious changes, further confirming the structural stability of the catalyst.

Effect of P coordination on Ir-P bond dynamic evolution and origin of HOR activity

The bond strength influenced the interaction between the catalyst and adsorbates, therefore altering the electrocatalytic performance. Coulomb’s law (F∝Qq/r²) indicates that bond strength (F) was related to the atomic valence charges (Q, q) and bond length (r)⁴⁸. Generally, bond length stretching and valence state decrease led to the weakening of bond strength. As shown in Fig. 4a, b, in contrast to IrP₂@NC, the Ir-P bonds in IrP₂-Ir₂P@NC underwent elongation and valence state decrease, resulting in diminished bond strength. Thus, we hypothesized that the Ir-P bonds would be partially dissolved during the electrochemical reaction, thereby enhancing the alkaline HOR activity. To further confirm our hypothesis, we investigated the changes in the content of Ir-P bonds during the reaction via quasi-in-situ XPS. The content of Ir-P bonds decreased gradually with the increase of potential (Fig. 4c, d and Supplementary Fig. 52). Additionally, it could be clearly observed that the Ir ion content of IrP₂-Ir₂P@NC was lower than that of IrP₂@NC. This indicated that an appropriate P coordination number helped promote the dissolution of Ir to form metal cations, thereby achieving a solvation enhancement effect and further improving the catalytic reaction activity, which was consistent with our hypothesis. Meanwhile, due to the enhanced solvation effect at the interface, we speculated that the dissolved Ir ions would exert a significant influence on the interfacial water (Supplementary Fig. 53).

Fig. 4. Investigation of the dynamic evolution of Ir-P bonds via quasi-in situ characterizations.

a The influence of M-O bond length/valence on bond polarizability. b The impact of Ir-P bond length and valence state on bond strength for IrP₂-Ir₂P@NC and IrP₂@NC (red: Ir; purple: P). c Quasi in-situ electrochemical XPS spectra of IrP₂-Ir₂P@NC. d The influence of different voltages on the content of Ir ions in IrP₂-Ir₂P@NC and IrP₂@NC acquired from (c). e Ir L₃-edge XANES spectra of IrP₂-Ir₂P@NC at different voltages. f Corresponding Fourier transformed the EXAFS spectrum spectra of the IrP₂-Ir₂P@NC at different voltages. g Corresponding Fourier transformed the EXAFS spectrum spectra of the IrP₂@NC at different voltages. h The influence of different voltages on the coordination number of Ir-P bonds in IrP₂-Ir₂P@NC and IrP₂@NC The influence of different voltages on the content of Ir ions in IrP₂-Ir₂P@NC and IrP₂@NC acquired from (Supplementary Table 7). i Comparison of Ir-P bond energies between IrP₂-Ir₂P@NC and IrP₂@NC. j Relationship between Ir-P bond energy and ion content in IrP₂-Ir₂P@NC and IrP₂@NC at 40 mV acquired from (i and d). k Wavelet transformation of Ir L₃-edge EXAFS data for IrP₂-Ir₂P@NC at different voltages. Source data for Fig. 4 are provided as a Source Data file.

To further observe the dynamic evolution of the Ir-P bond during the reaction, we conducted quasi-in situ XAS measurements. As shown in Fig. 4e and Supplementary Fig. 54, the Ir valence states of IrP₂-Ir₂P@NC and IrP₂@NC did not change significantly during the reaction, demonstrating that the application of voltage had no obvious effect on the electronic configuration. To further investigate the local environment of Ir atoms, we collected EXAFS spectra at the Ir K-edge under different voltages. As displayed in Fig. 4f, g, IrP₂-Ir₂P@NC and IrP₂@NC exhibited a distinct peak at 2.38 Å and 2.36 Å, respectively, which could be attributed to the Ir-P bond. With the application of voltage, without significant change in the bond length of the Ir-P bond was observed. Meanwhile, IrP₂@NC showed a persistent characteristic peak at 1.92 Å upon voltage application, which could be attributed to the Ir-O bond, indicating its excessive hydroxyl adsorption and limited desorption rate. In contrast, a distinct Ir-O bond peak in IrP₂-Ir₂P@NC was only observed at 0 V. This difference demonstrates that an appropriate amount of P coordination promotes OH transport, thereby enhancing catalytic activity in alkaline HOR. The specific structural parameters around Ir atoms were determined by curve fitting of the EXAFS spectra (Supplementary Figs. 55, 56 and Table 7). The analysis showed that with the application of voltage, the coordination numbers of the Ir-P bonds in both IrP₂-Ir₂P@NC and IrP₂@NC gradually decreased (Fig. 4h). In summary, these findings demonstrate that during the reaction, an appropriate amount of P coordination causes partial dissolution of Ir to form metal cations. This process promotes hydroxyl transport and enhances the solvation effect at the interface. Bond energy is a key parameter reflecting the stability of chemical bonds. To this end, we calculated the Ir-P bond energies in IrP₂-Ir₂P@NC and IrP₂@NC. As shown in Fig. 4i, compared with IrP₂@NC (2.34 eV), the Ir-P bond energy of IrP₂-Ir₂P@NC decreased to 1.60 eV. This further indicated that the Ir-P bonds in IrP₂-Ir₂P@NC were more prone to cleavage, generating metal cations, which thus provided strong support for the aforementioned results (Fig. 4j). Further investigation into the different synergistic environments of IrP₂-Ir₂P@NC and IrP₂@NC showed consistency with the aforementioned EXAFS results (Fig. 4k and Supplementary Fig. 57).

P coordination-mediated Ir ion dissolution and interfacial water network regulation

In order to elucidate the impact of Ir ion dissolution (influenced by different P coordination numbers) on interfacial water, in-situ Raman spectroscopy was utilized to investigate the interfacial water structure. Isotope-labeled in-situ electrochemical Raman spectroscopy utilizing D₂O electrolyte was conducted on IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC under alkaline conditions. Figure 5a–c illustrate that spectra were obtained at different applied potentials in 0.1 M KOH, all exhibiting a broad band from 2000 to 2900 cm⁻¹, attributable to the O-H stretching vibration mode of interfacial water⁴⁹. The O-H stretching peak of interfacial water was resolved into three distinct fractions. The O-H stretching vibration frequency at approximately 2650 cm⁻¹ was ascribed to water molecules distributed in the gap region (gap-H₂O), the frequency at around 2580 cm⁻¹ was linked to water molecules over the gap region (above-gap-H₂O), and the lowest O-H stretching vibration frequency at approximately 2500 cm⁻¹ was associated with water molecules adjacent to the electrode surface (K⋅H₂O)⁵⁰. Figure 5d depicted the potential-dependent area ratios of the peaks extracted from different catalysts, demonstrating the intensity, percentage, and variations of gap-H₂O. As expected, the proportion of gap-H₂O in IrP₂-Ir₂P@NC exceeded that in IrP₂@NC and Ir@NC, indicating the highest water concentration in the gap region. As the applied voltage grew, the proportion of gap-H₂O increased, indicating enhanced connectivity of the hydrogen-bond network in the gap-H₂O region, following the order IrP₂-Ir₂P@NC > Ir@NC > IrP₂@NC. This further suggested that the dissolution of Ir induced by an appropriate amount of P coordination augmented the number of interfacial water molecules, thereby promoting the connectivity of the hydrogen bond network.

Fig. 5. Characterization of the interfacial structures by in-situ Raman.

Normalized in-situ Raman spectra of interfacial water structure on a IrP₂-Ir₂P@NC, b IrP₂@NC and c Ir@NC catalysts were collected. d The corresponding trend of the gap-H₂O content acquired from (a, b and c). e Stark slopes for IrP₂-Ir₂P@NC acquired from (a, b and c). f Zeta potential of IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC. g Capacity against potential curves for IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC were measured. h Schematic illustration of the influence of OH_ad on the interfacial water structure (purple: K; white: H; red: O; light purple: P; light red: Ir). i Schematic diagram of dynamic hydrogen bond network configuration. Dashed lines represent hydrogen bonds (white: H; red: O; light purple: P; light red: Ir). Source data for Fig. 5 are provided as a Source Data file.

The Stark slopes obtained by plotting the vibration frequencies as a function of electrode potential further revealed the contributions of different water components⁵¹. As shown in Fig. 5e and Supplementary Figs. 58, 59 compared with the Stark slopes of above-gap-H₂O and K⋅H₂O in IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC, the Stark slope of gap-H₂O was larger, indicating its preferential consumption during HOR. Additionally, we investigated the hydroxyl adsorption on the catalysts utilizing Zeta potential analysis. Compared with IrP₂@NC (−22.12 eV) and Ir@NC (−22.12 eV), IrP₂-Ir₂P@NC demonstrated a higher negative Zeta potential of −45.69 eV, indicating its stronger hydroxyl adsorption capacity (Fig. 5f). In summary, we concluded that moderate P coordination facilitated an increase in interfacial water content, promoting the connectivity of the hydrogen bonding network and accelerating the occurrence of HOR kinetics. Moreover, the PZC value, an important descriptor of HOR kinetics, was intimately associated with the interfacial configuration of water molecules. Figure 5g illustrates that the PZC of IrP₂-Ir₂P@NC was 0.94 eV, much lower than that of IrP₂@NC (1.14 eV) and Ir@NC (0.98 eV), correlating well with the HOR activity. Meanwhile, the enhanced hydroxyl adsorption in IrP₂-Ir₂P@NC generated a greater amount of adsorbed hydroxyl groups (OH_ad) and dynamically liberated more free water molecules into the gap region, thereby significantly boosting the connectivity of the hydrogen bond network (Fig. 5h). More importantly, the enhanced OH adsorption on IrP₂-Ir₂P@NC provided a greater amount of adsorbed OH_ad and released more free water molecules into the gap region, thereby significantly improving the connectivity of the hydrogen bond network. Figure 5i intuitively depicted the HOR catalytic mechanism of IrP₂-Ir₂P@NC, aiming to facilitate a better understanding of the reaction process.

In-situ surface-enhanced infrared absorption spectroscopy (SEIRAS) was further utilized to explore the interfacial water on the catalyst surface. Figure 6a and Supplementary Figs. 60, 61 illustrate that, in the SEIRAS spectra, the O-H stretching peak of interfacial water was decomposed into three different components. The component with the highest vibrational frequency was assigned to gap-H₂O, the intermediate frequency to above-gap-H₂O, and the lowest frequency to K⋅H₂O. Figure 6b presents the proportions of gap-H₂O for different catalysts at various potentials. Compared to IrP₂@NC and Ir@NC, IrP₂-Ir₂P@NC demonstrated the largest proportion of gap-H₂O, indicating its optimal hydrogen-bond network connectivity and thus competitive HOR activity. This finding was congruent with the in-situ Raman results and the HOR catalytic activity. Furthermore, as illustrated in Fig. 6c, IrP₂-Ir₂P@NC had a lesser redshift in comparison to IrP₂@NC and Ir@NC, signifying its enhancement of hydrogen bond formation, which was consistent with the above results. Supplementary Figs. 62, 63 and 64 illustrate the assessments of the electron localization function (ELF) evaluations for IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC, revealing the strongest electron interaction between IrP₂-Ir₂P@NC and adsorbed OH, succeeded by Ir@NC and subsequently IrP₂@NC⁵². This trend was consistent with the results from CO stripping cyclic voltammetry (CV) and Zeta potential measurements. Therefore, the aforementioned results demonstrated that optimal P coordination enhanced the connectedness of the hydrogen-bond network, while excessive P coordination had the contrary effect, which was consistent with the results obtained from in-situ Raman spectroscopy. To validate the theoretical effects of the interfacial water structure described above, ab initio molecular dynamics (AIMD) simulations were performed. Figure 6d–f illustrated the computed representations of regional water molecules for IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC. It could be observed that the concentration hierarchy of water molecules along the region close to the catalyst interface was detected as IrP₂-Ir₂P@NC > Ir@NC > IrP₂@NC, which was consistent with our aforementioned results. More importantly, the concentration trend reflected by these regional water molecule simulation diagrams was highly consistent with that of the oxygen density diagrams of the corresponding materials (Supplementary Figs. 65, 66 and 67). Furthermore, to investigate the impact of this alteration on the hydrogen bond network, we statistically analyzed the hydrogen bond density. As shown in Fig. 6g, h, the concentration trend of the hydrogen bond network at the interface near the catalyst was IrP₂-Ir₂P@NC > Ir@NC > IrP₂@NC, which matched the trend diagram derived from our aforementioned in-situ characterizations. In summary, these results suggested that optimal P coordination helped promote the dissolution of Ir, thereby achieving a solvation enhancement effect. This alteration enhanced the gap-H₂O content, expedited the transfer of hydroxyl species, and facilitated the connection of the hydrogen bond network, thus improving the catalytic activity of alkaline HOR. Therefore, the above results indicated that appropriate P coordination enabled IrP₂-Ir₂P@NC to achieve optimal OHBE, thereby significantly accelerating the Volmer step and realizing high catalytic activity (Fig. 6i). Furthermore, we investigated the correlations between OHBE, gap-H₂O and the catalytic performance of the catalysts at 50 mV. As shown in Fig. 6j illustrated a positive correlation among OHBE, gap-H₂O content, and catalytic activity, further indicating that an increase in gap-H₂O content was beneficial for enhancing hydroxyl adsorption and further improving the catalytic activity of HOR. In addition, we also investigated the relationship between the water formation energy barrier and the number of hydrogen bonds. As shown in Fig. 6k, the lower the water formation energy barrier, the greater the number of hydrogen bonds. Therefore, we conclude that an increase in gap-H₂O content is conducive to promoting the transport of hydroxyl species, as indicated by the Volmer step (OH⁻ + H_ad → H₂O + e⁻ + *). This change facilitates the occurrence of the RDS, reduces the water formation energy barrier, further increases the number of hydrogen bonds, and significantly promotes the catalytic activity for the alkaline HOR. It is worth noting that the solvation enhancement effect induced by Ir dissolution mentioned above is not only applicable to the present Ir-based catalytic system, but also universally applicable to other noble metal-based catalysts.

Fig. 6. Characterization of the interfacial structures by SEIRAS and AIMD.

a Deconvolution of the O-H stretching vibration peak from 0 V to 0.2 V vs RHE for IrP₂-Ir₂P@NC in 0.1 M KOH. b The corresponding trend of the gap-H₂O content acquired from (a, Supplementary Figs. 61 and 62). c Comparison of the O-H stretching vibration frequencies of gap-H₂O of IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC acquired from (a, Supplementary Figs. 61 and 62). d–f IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC surfaces by AIMD simulations in interfacial on the water density. g, h IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC surfaces by AIMD simulations and Average number of H-bonds in interfacial water network (gray: N; brown: C; white: H; red: O; light purple: P; light red: Ir). i Ir-P coordination number and OHBEs of IrP₂-Ir₂P@NC and IrP₂@NC acquired from (1 f). j The corresponding trends of gap-H₂O content and HOR polarization curves at 50 mV for different catalysts acquired from (1 f). k Variation trends of energy barriers and number of H-bonds for IrP₂-Ir₂P@NC, IrP₂@NC and Ir@NC acquired from (g and 1 h). Source data for Fig. 6 are provided as a Source Data file.

Discussion

In conclusion, this work elucidates that heteroatom coordination number engineering is essential for manipulating the interfacial solvation environment and optimizing alkaline HOR performance. Theoretical simulations confirm IrP₂-Ir₂P@NC structure enables optimal H and OH adsorption to accelerate the Volmer step. A key is that tailoring Ir-based catalysts’ P coordination number precisely induces controlled Ir dissolution in alkaline media, where in-situ Ir cations trigger a solvation enhancement effect (elevating gap- H₂O content and H-bond connectivity). Interestingly, we establish a universal structure-activity relationship among heteroatom coordination number, interfacial solvation state and HOR activity, clarifying the intrinsic link between metal dissolution and catalytic performance at the catalyst-electrolyte interface. More importantly, the AEMFC with IrP₂-Ir₂P@NC achieves a peak power density of 1.59 W cm⁻² under H₂-O₂ atmosphere, and the obtained performance is competitive with data of recently reported alkaline HOR catalysts. This work departs from the conventional focus on composition/structure modulation, puts forward a design principle of “coordination number regulation → solvation effect optimization → catalytic activity enhancement”, and offers fundamental insights for the rational design of electrocatalysts.

Methods

Chemicals and materials

All chemicals and solvents were commercially available and used as received without further purification. Iridium(III) chloride hydrate (IrCl₃, Innochem, Ir > 64%); 2-methylimidazole (C₄H₆N₂, 98%); Analytical grade zinc nitrate hexahydrate (Zn(NO₃)₂·6H₂O, 99%); Triphenylphosphine (C₁₈H₁₅P, McLean, 99%), and potassium hydroxide (KOH) were sourced from Shanghai Aladdin Biochemical Technology Co., Ltd., China. Methanol was purchased from Xilong Scientific Co., Ltd. Nafion® D-521 dispersion was obtained from Alfa Aesar Chemical Co., China. Nitrogen (N₂, 99.999%) was purchased from Guilin Xinhongyi gas Co., Ltd. All experimental procedures were performed using ultrapure water with a resistivity of 18.2 MΩ cm⁻¹.

Preparation of electrolytes

To prepare 0.1 M KOH aqueous electrolyte, 5.611 g of solid potassium hydroxide was fully dissolved in ultrapure water. The resulting solution was quantitatively diluted to a final volume of 1 L volumetric flask and subsequently transferred and sealed in a polypropylene container. The electrolyte was preserved in a dry, well-ventilated environment to avoid composition degradation and performance attenuation during storage.

Synthesis of ZIF-8 and NC Precursor

To synthesize ZIF-8 precursor, 2.79 g of Zn(NO₃)₂·6H₂O, 3.08 g of 2-methylimidazole were dissolved in 75 mL of methanol and sonicated for 5 min, the latter was added to the former solution and heated at 35 °C for 4 h. The resulting ZIF-8 powder was collected by centrifugation, washed several times with methanol and dried at 60 °C overnight. The ZIF-8 powders were heat-treated at 900 °C for 3 h in an N₂ atmosphere at a heating rate of 5 °C min⁻¹ to obtain the final product. The black samples were collected after natural cooling to room temperature, and the final product was obtained and denoted as NC.

Synthesis of IrP₂-Ir₂P@NC, IrP₂@NC, Ir@NC

In a typical synthesis, 30 mg of previously prepared NC support, 0.066 mmol IrCl₃ and 0.15 mmol triphenylphosphine were slowly dissolved with 20 mL mixture solution (H₂O: ethanol = 1:1 v/v). Subsequently, the mixture solution was moved into a 100 mL Teflon-lined autoclave and heated at 160 °C for 9 h. The sample was harvested by centrifugation, washed several times with ethanol, and dried in vacuum oven overnight. The obtained sample was then annealed at 800 °C under N₂ for 2 h to get IrP₂-Ir₂P@NC.

For comparison, the fabrication of IrP₂@NC was analogous to that of IrP₂-Ir₂P@NC, except that the temperature was changed to 950 °C, as explained in the discussion section. Concurrently, Ir@NC can be synthesized without P source.

Synthesis of Ir₂P@NC

Initially, 30 mg of the as-prepared NC support, 0.066 mmol of IrCl₃, and 0.27 mmol of triphenylphosphine were slowly dissolved in 20 mL of a mixed solution (H₂O: ethanol = 1:1 v/v). The mixture was then transferred into a 100 mL Teflon-lined autoclave and heated at 160 °C for 9 h. The product was collected by centrifugation, washed with ethanol several times, and dried in a vacuum oven overnight. The obtained sample was subsequently annealed at 790 °C under H₂/Ar atmosphere for 2 h to obtain Ir₂P@NC.

Physical characterizations

Powder X-ray diffraction (XRD, Rigaku D/Max 2500 V/PC, Japan) was measured to investigate the crystal structure. Scanning electron microscopy (SEM) measurements were performed on the FEI Quanta 200 FEG system. Transmission electron microscopy (TEM) images were acquired on a JEM-2100F electron microscope to characterize the morphology and elemental distribution of the catalysts. Ir L₃-edge X-ray absorption spectra (XAS) were recorded at the 1W1B beamline of Shanghai Synchrotron Radiation Facility. X-ray photoelectron spectroscopy (XPS, Thermo Fisher EscaLab 250XI) was utilized to analyze the chemical state and electronic structure of the catalysts. Inductively coupled plasma mass spectrometry (ICP-MS, PerkinElmer corporation, FLexarNexION300X) was adopted to quantify the metal contents within catalyst samples. The Brunauer-Emmett-Teller (BET) method was used to measure the specific surface area and pore size distribution of the samples. Raman spectra were acquired on a Renishaw in Via spectrometer equipped with a 532 nm visible laser. Electron paramagnetic resonance (EPR) spectra were recorded with a Bruker E500 spectrometer.

Preparation of working electrodes

To prepare the catalyst inks, 3.0 mg catalysts were dispersed in 500 μL of mixed solvents containing water, isopropanol and Nafion® perfluorinated resin solution (5 wt.% in water) (v/v/v = 1/1/0.02) by ultrasonication for 30 min. A glassy carbon (GC) electrode (diameter: 5 mm) was polished using slurry containing 0.05 μm γ-alumina powder, then washed sequentially with ultrapure water and ethanol to acquire a clean electrode surface. After natural air drying of the polished GC substrate, 15.0 μL of the as-prepared ink coated the electrode surface dropwise, followed by ambient air drying prior to all electrochemical tests. ICP-MS data determined the Ir mass loading deposited on the working electrode, with a final Ir loading of 47.59 μg_Ir cm⁻².

Electrochemical measurements

All electrocatalytic tests were conducted on a CHI 760E (Shanghai, China) electrochemical workstation within a standard three-electrode system. A 5 mm-diameter glassy carbon electrode (GCE, disk area: 0.196 cm²) functioned as the working electrode, while a graphite rod and saturated KCl-filled with Ag/AgCl electrode acted as the counter and reference electrodes separately. The Biologic VMP3 multichannel potentiostat recorded electrochemical impedance spectroscopy (EIS) profiles at open-circuit potential with an upper frequency limit of 200 kHz. All electrochemical characterizations maintained a stable thermal condition of 298 ( ± 0.1) K, and every detected potential value converted to the reversible hydrogen electrode (RHE) scale following complete iR compensation.

Electrochemical impedance spectra (EIS) tests were conducted by the AC impedance spectra from 200 kHz to 0.1 kHz with a voltage perturbation of 10 mV. The real part of the resistance at 1 kHz was taken as the uncompensated resistance and was used to obtain the iR-free potential (EiR-free) according to the following equation,

The Biologic VMP3 potentiostat recorded electrochemical impedance spectra (EIS) across a frequency window ranging from 200 kHz down to 0.1 kHz, applying a 10 mV alternating voltage perturbation. The resistive real component extracted at 1 kHz standed for uncompensated resistance, which further calculated iR-corrected potentials (EiR-free) via the listed formula below:

$$E_{iR-\mathrm{free}}=E-{iR}_{\mathrm{u}}$$ 1

where E is the measured potential and i is the corresponding current.

All recorded potential values align with the reversible hydrogen electrode (RHE) scale after standard calibration procedures. Calibration proceeds in H₂-saturated electrolyte with high-purity hydrogen, where a platinum foil serves as the working electrode. Potential transformation complies with the formula listed as follows:

$$\left({\mathrm{E}}_{\mathrm{Ag}/\mathrm{AgCl}}+0.197+0.059\times \mathrm{pH}\right)$$ 2

The kinetic current density was calculated by Koutecky–Levich equation. The measured overall HOR current density (j) can be divided into kinetic current density (j_k) and diffusion current density (j_d) based on the Koutecky–Levich equation.

$${\frac{1}{j}}_d{+\frac{1}{j}}_k=\frac{1}{j}$$ 3

Where j is the measured current density, j_d is the limiting current density and j_k is the kinetic current density.

Mass activity (MA) was normalized by using j_k and the mass active metal dripped onto the RDE.

$$MA=\frac{j_k}{M}$$ 4

CO stripping measurements were initially carried out in CO-saturated 0.1 M KOH solution at constant potential (0.1 V vs RHE) for 10 min to enable CO adsorption onto the metallic surface. Subsequently, the CO gas was changed to the N₂ gas and held for 30 minutes to remove the residual CO in the electrolyte. The CO stripping curve was collected by cyclic voltammetry. The CO stripping curves with different scanning rates were obtained at potential scanned from 0 V to 1.2 V vs RHE.

$${ECSA}_{activemetal}=\frac{Q_{co}}{0.42mC{cm}^{-2}*M}$$ 5

where Q_CO is the total charge of adsorbed CO oxidation, 0.42 mC cm⁻² corresponds to monolayer CO adsorption and M refers to the total loading of active metal on the working electrode.

Anion exchange membrane fuel cell (AEMFC) tests

The as-synthesized target catalyst served as the anode catalyst, and commercial Pt/C was used as the cathode catalyst. For the preparation of catalyst inks, the catalyst and alkaline ionomer dispersion (PiperION A5-HCO₃, Versogen, USA) were ultrasonically dispersed in a mixed solvent of deionized water and isopropanol. The as-prepared anode and cathode catalyst inks were separately sprayed onto two sides of the anion exchange membrane (PiperION A20-CO3, Versogen, 20 μm) to fabricate catalyst-coated membranes (CCMs). Notably, no separate independent pretreatment was performed on the pristine PiperION A20-CO3 membrane prior to catalyst spraying. Instead, full activation of the membrane was synchronously realized together with the coated catalyst layers via unified post-treatment of the entire CCM. All CCMs were immersed in 1 M KOH solution for 3 h for activation, and the fresh solution was replaced every hour. Afterwards, the CCMs were thoroughly rinsed with ultrapure water until the pH of the residual washing liquid reached neutrality. Gaskets, gas diffusion layers, graphite bipolar plates with single flow fields and metal current collectors were sequentially assembled on both sides of the treated CCM to construct a complete AEMFC single cell. The cell performance was evaluated on a Scribner 850e fuel cell test station. During the tests, both the single cell and gas humidifier were maintained at 80 °C with a relative humidity of 100%. The backpressure of H₂ was set to 200 kPa, and the backpressures of O₂ and air were both fixed at 200 kPa. In the H₂-O₂ test mode, the flow rates of H₂ and O₂ were 1.5 L min⁻¹ and 1.5 L min⁻¹, respectively. In the H₂-air test mode, the flow rates of H₂ and air were both 1.5 L min⁻¹.

In-situ surface-enhanced infrared absorption spectroscopy

In-situ Fourier transform infrared (FT-IR) spectroscopy was measured using a Shimadzu IRTracer-100. For the preparation of the catalyst ink, 3 mg of the catalyst was mixed with 245 μL of isopropanol, 245 μL of ultrapure water, and 5 μL of Nafion. Then, the ink was deposited on a Si prism coated with an Au film. A platinum foil electrode and an Ag/AgCl electrode were used as the working electrode, counter electrode, and reference electrode, respectively.

In-situ Raman spectra

In-situ Raman spectra were acquired employing a customized cell (Gaoss Union), equipped with a saturated Ag/AgCl reference electrode and a carbon rod counter electrode in 0.1 M KOH. The evolution of catalyst was tracked by collecting Raman spectra at constant potential ranging from 0 to 0.20 V (vs. RHE).

DFT calculation method

All spin-polarized DFT computations were implemented via the Vienna Ab Initio Simulation Package (VASP)^53,54. To handle exchange-correlation interactions, the Perdew-Burke-Ernzerhof (PBE) functional within the generalized gradient approximation (GGA) framework was adopted for all calculations⁵⁵. Projector augmented-wave (PAW) pseudopotentials were utilized to characterize the electron-ion interactions in all modeled systems⁵⁶. During structural relaxation processes, atomic configurations were fully optimized until the Hellmann-Feynman residual force acting on every single atom fell below 0.02 eV Å⁻¹. Meanwhile, the self-consistent iteration threshold for solving the Kohn-Sham equations was defined at a total energy convergence of 10⁻⁶ eV.

AIMD simulation details

All computational simulations were carried out within the density functional theory (DFT) formalism using the projector augmented-wave (PAW) approach encoded in the Vienna Ab Initio Simulation Package (VASP)^54,57. The Perdew-Burke-Ernzerhof (PBE) functional belonging to the generalized gradient approximation (GGA) was chosen to describe the exchange-correlation potential^55,58. A plane-wave kinetic energy cutoff of 400 eV was assigned for all computational sets. The self-consistent iteration convergence threshold for solving the Kohn-Sham equations was fixed at 10⁻⁵ eV. Full geometric relaxation was conducted for all atomic configurations until the residual Hellmann-Feynman force on each atom dropped below 0.05 eV Å⁻¹. Only the Γ-point was adopted for Brillouin zone integration throughout the simulations. Ab initio molecular dynamics (AIMD) simulations were performed under the NVT canonical ensemble with a constant temperature of 298 K, where the particle count, system volume and temperature were kept unchanged. A timestep of 2 fs(1 fs) was adopted for precise integration of atomic motion equations, and the total simulation trajectory extended up to 10 ps⁵⁹.

Author contributions

B.W. and Y.Z. (Yongfa Zhu) supervised, reviewed and edited the manuscript. X.Y. provided supervision, funding acquisition and editing. C.N. and B.W. conceptualized the project, performed all the experiments, analyzed the results and wrote the original draft. Y.Z. (Yulin Zhao), Y.L., and L.C. conducted the investigation. Z.L., H.L. and T.T. curated the data. X.S., L.Z., B.W. and R.C. performed partial data testing. W.Z. and J.Q. revised the manuscript.

Peer review

Peer review information

Nature Communications thanks Deli Wang and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

X.Y. discloses support for the research of this work from the National Natural Science Foundation of China [grant number 52363028, 21965005], the Natural Science Foundation of Guangxi Province [grant number 2021GXNSFAA076001, 2018GXNSFAA294077], the Guangxi Technology Base and Talent Subject [grant number GUIKE AD23023004, GUIKE AD20297039] and the Innovation Project of Guangxi Graduate Education [grant number YCBZ2024082]. W.Z. discloses support for the research of this work from the Guangdong Basic and Applied Basic Research Foundation [grant number 2025A1515010773]. B.W. discloses support for the research of this work from the National Research Foundation Singapore and A*STAR under the LCER Phase 2 Programme 1st Emerging Technology [grant number U2411D4006].

Data availability

The data supporting the findings of this study are available within the article and its Supplementary Information files. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-75394-x.