Synthesis and characterizations of MNi/NCThe bespoke MNi/NC catalysts series, including RuNi/NC, PtNi/NC, PdNi/NC, CuNi/NC and ZnNi/NC, were readily synthesized using a generalized method involving tandem high-temperature pyrolysis and “inside-out” Kirkendall alloying and hollowing process (Fig. 1a). Initially, nickel nitrate was coordinated with ethylenediaminetetraacetic acid (H4EDTA) to produce the coordination complex [Ni2(EDTA)]n precursor. Subsequent pyrolysis at 1000 °C in an Ar atmosphere, followed by an acid wash, yielded the intermediary Ni nanoparticles encapsulated in N-doped nanocarbons (Ni@NC). Whereafter, Ni@NC was mixed with specific metal salts and co-annealed in a 5 wt% H2-Ar atmosphere at a designated temperature to produce the final MNi/NC catalyst. Notably, during the conversion from Ni@NC to MNi/NC, the bulk of encapsulated Ni nanoparticles migrated outward and reacted with the other metal present on the surface, forming smaller-sized nanoalloys. This process resulted in a hollow interior and abundant pore channels within the carbon nanocage. This unique structural evolution can be reasoned to the intermingling of metals driven by the typical Kirkendall effect, which arises from the relatively faster diffusion rate of Ni atoms than the targeted metal (M) atoms at elevated temperatures31. This culminates in a net outward flow of Ni atoms toward the surface, and the subsequent vacancies coalescence into the Kirkendall voids at the original sites of the Ni nanoparticles.Fig. 1: Schematic synthesis and structural characterizations of MNi/NC (M = Ru, Pt, Pd, Cu, Zn) catalysts.a Schematic illustration of the synthetic process of MNi/NC. b, c TEM images of Ni@NC. d‒m HAADF-STEM images (the orange dashed circles sketch the outlines of the curved carbon nanocages) and EDS mappings of MNi/NC, and AC-STEM images of MNi nanoalloys (insets are the schematical crystallographic atomic arrangements): (d, e) RuNi/NC; (f, g) PtNi/NC; (h, i) PdNi/NC; (j, k) CuNi/NC; (l, m) ZnNi/NC. n XRD patterns of Ni@NC and MNi/NC. o Optical images showing water contact angles of Ni@NC and MNi/NC series.Transmission electron microscope (TEM) analysis of the intermediary Ni@NC reveals the presence of buried Ni nanoparticles with an average size of ca. 10 nm within the onion-like multilayered graphitic nanocarbons (Fig. 1b, c and Supplementary Fig. 2). In contrast, TEM images of MNi/NC clearly illustrate the formation of hollow carbon nanocages, characterized by shell thicknesses ranging from 2 to 3 nm and diameters spanning 5 – 10 nm, with curvature radii varying from 2.1 to 6.0 nm (Supplementary Fig. 3). The external surfaces of the nanocages are intricately adorned with Ni-based bimetallic nanoalloys resembling nanotips protruding from the high-curvature carbon nanocages. High-angle annular dark field (HAADF) scanning transmission electron microscopy (STEM) mapping exhibits the evenly distributed bimetallic elements within these nanoalloys (Fig. 1d‒m and Supplementary Figs. 4–8). And atomic-resolution HAADF imaging using aberration-corrected STEM (AC-STEM) reveals well-matched lattice atomic arrangements corresponding to Ru1.68Ni0.32 (hexagonal), PtNi3 (cubic), Pd0.19Ni0.81 (cubic) and CuNi3 (cubic) and Zn0.08Ni0.92 (cubic), respectively. Furthermore, their metal contents determined by inductively coupled plasma-atomic emission spectrometry (ICP–AES) are presented in Supplementary Table 1. As observed, a discrepancy in the alloy composition is evident for these MNi/NC samples, especially for RuNi/NC, which exhibits a much lower Ni content. This disparity might be attributed to the significant lattice mismatch between hexagonal Ru and cubic Ni, which likely impedes the complete Kirkendall alloying process, resulting in some Ni nanoparticles remaining unalloyed within the graphitic carbon nanocages. Various factors, such as lattice matching, atomic radius, electronic configuration, and thermodynamic compatibility of the metals are posited to collectively influence the kinetics of Kirkendall alloying, and consequently determine the final alloy structure.The definitive phase structures of these nanoalloys were corroborated through X-ray diffraction (XRD) analysis (Fig. 1n). Except for RuNi/NC, which shows diffraction peaks corresponding to hexagonal Ru1.68Ni0.32 and residual cubic Ni, all other MNi/NC samples exhibit similar patterns to Ni@NC and match well with the standard patterns of their respective cubic alloy structures. Notably, a comparison between Ni@NC and the MNi/NC series reveals a significant attenuation of the graphite (002) peak at 26° in the latter, accompanied by the emergence of an amorphous carbon peak at a lower angle. This observation indicates an impaired graphite structure and evidences the Kirkendall alloying process during which intermetallic interdiffusion across the carbon layers occurred. Such Kirkendall alloying is anticipated to engender a strong metal-support interaction that not only upgrades the electronic structures of nanoalloys but also establishes a robust linkage between the nanoalloy and carbon nanocages conducive to sustained electrocatalysis. Besides, after Kirkendall alloying, all MNi/NC samples exhibit enhanced water affinity compared to Ni@NC, as evidenced by their super-hydrophilic characters with low water contact angles <10° (Fig. 1o and Supplementary Fig. 9). The super-hydrophilicity of MNi/NC is primarily attributed to the rich surface oxygen-containing functional groups and defects inherited from Ni@NC formed during the pickling process, and might be further amplified by the increased surface roughness and microporosity resulting from the Kirkendall alloying process. Importantly, this super-hydrophilicity could play a crucial role in enhancing the HER electrocatalytic activity by endowing the material with improved electrolyte accessibility, enhanced mass transport, increased active site availability, stabilized reaction intermediates, and reduced charge transfer resistance.3D electron tomography and X-ray spectroscopic analysis of RuNi/NCTo meticulously dissect the local microstructural attributes, RuNi/NC was selected as a prototypical example from the MNi/NC series for an in-depth analysis. The STEM image presented in Fig. 2a clearly shows the uniform dispersion of RuNi nanoalloys on carbon nanocages, exhibiting a mean diameter of 2 nm (Supplementary Fig. 10). The well-defined lattice atomic arrangement, characteristic of the hexagonal P63/mmc Ru1.68Ni0.32 alloy oriented along the [\(10\bar{1}\)] zone axis, is distinctly portrayed in Fig. 2b. Furthermore, the annular bright field (ABF) STEM image in Fig. 2c clearly delineates that the RuNi nanoalloy is effectively supported by the carbon nanocages featuring continuous multilayer graphite lattice fringes. To resolve the local structure more comprehensively, we performed three-dimensional (3D) electron tomography for RuNi/NC (Supplementary Movie 1). The 3D structural visualization depicts a singular RuNi nanoalloy about 3 nm in size planted on a carbon nanocage with an approximate diameter of 20 nm, as vividly shown in Fig. 2d and Supplementary Fig. 11. The pony-size carbon nanocage, distinguished by its high-curvature surface, renders the RuNi nanoalloy a nanotip-like protrusion, thus inducing a typical tip effect on the RuNi nanoalloy with a locally enhanced electric field32. Additionally, as a result of the outward Kirkendall diffusion of Ni atoms, substantial micropores are observed across the entire structure of the carbon nanocage. This observation is further attested by the high specific surface area (326.3 m2 g−1) and total pore volume (1.18 cm3 g−1) of RuNi/NC, as measured by N2 isothermal sorption (Supplementary Fig. 12). The cross-section depicted in Fig. 2e unveils the apparent hollow cavity within the carbon nanocage composed of multiple graphite layers. Their graphitic structure is verified by Raman spectra (Supplementary Fig. 13), ensuring a highly conductive network favorable for electrocatalysis33,34. Notably, it distinctly discloses that the tip-like RuNi nanoalloy is partially embedded within the shell region of the carbon nanocage. This configuration affirms a robust metal-support interaction, indicative of a promising and enduring framework for electrocatalysis. Such intimate integration between the RuNi nanoalloy and the carbon nanocage is further elucidated by the laminated cross-sections and isosurfaces of RuNi/NC, as depicted in Fig. 2f.Fig. 2: 3D TEM tomography and ex situ synchrotron XAFS study of RuNi/NC.a–c HAADF and AC-STEM images of RuNi/NC. d 3D visualization of a single RuNi/NC nanoparticle, showing the partial embedding of RuNi nanoalloy within the carbon nanocage shell, achieved through electron tomography technique. e Cross-section of RuNi/NC, highlighting the multilayer porous shell and hollow cavity of the carbon nanocage. f Laminated cross-sections and isosurfaces of RuNi/NC. g XANES and (h) FT-EXAFS spectra at the Ni K-edge of Ni@NC, RuNi/NC, Ni foil, and NiO. i XANES and (j) FT-EXAFS spectra at the Ru K-edge of RuNi/NC, Ru foil, and RuO2. Wavelet transform analysis of (k) Ni K-edge and (l) Ru K-edge EXAFS oscillations of RuNi/NC.The local electronic structures of Ni and Ru atoms in RuNi/NC were then investigated by X-ray absorption spectroscopy (XAS). Figure 2g illustrates that the X-ray absorption near edge structure (XANES) spectroscopy reveals nearly coincident Ni K-edges for Ni@NC and Ni foil, indicating the metallic state of Ni in Ni@NC35. Conversely, the Ni K-edge of RuNi/NC is situated between those of Ni foil and NiO, exhibiting a more pronounced white line peak compared to Ni@NC. This suggests that the Ni valence in RuNi/NC is intermediate between 0 and +2, likely due to electron transfer from Ni to the more electronegative Ru in the RuNi nanoalloy, a finding that is consistent with the X-ray photoelectron spectroscopy (XPS) analysis (Supplementary Figs. 14–16). The local coordination environment of Ni atoms in RuNi/NC was probed by Fourier transformed extended X-ray absorption fine structure (FT-EXAFS) spectroscopic curve fitting analysis (Supplementary Fig. 17). The Ni K-edge EXAFS spectrum shows a prominent peak at 2.15 Å, attributed to Ni‒Ru and Ni‒Ni bonds, while a fainter peak at 1.56 Å corresponds to the Ni‒N bond (Fig. 2h, Supplementary Tables 2 and 3)36. Besides, the Ru K-edge XANES spectrum of RuNi/NC aligns with that of Ru foil (Fig. 2i), and the Ru K-edge EXAFS of RuNi/NC reveals a significant peak at 2.33 Å, indicative of Ru–Ni and Ru–Ru bonds (Fig. 2j and Supplementary Table 4)37. The wavelet transformed EXAFS of RuNi/NC further distinguishably demonstrates the local bonding information, prominently featuring a strong peak of Ru–Ni bond (Fig. 2k, l).Electrocatalytic HER performanceThe HER performance of RuNi/NC was assessed in 1 M KOH solution (pH 13.74 ± 0.12) by a three-electrode configuration with a rotating disk electrode (RDE) as the working electrode under ambient conditions. For comparison, Ni@NC, Ru/NC (Supplementary Fig. 18) and commercial 20 wt% Pt/C catalysts were also tested under identical conditions. The linear sweep voltammetry (LSV) curves illustrated in Fig. 3a and Supplementary Fig. 19 demonstrate the high electrocatalytic HER activity of RuNi/NC, achieving a low overpotential (η) of 12 mV to deliver a current density of 10 mA cm‒2. This performance markedly surpasses those of Ni@NC (459 mV), Ru/NC (55 mV) and Pt/C (28 mV), poisoning it among the leading results to date. Additionally, RuNi/NC requires merely −43 and −66 mV to attain the current densities of 50 and 100 mA cm‒2, respectively. The superior HER activity of RuNi/NC is further validated by a 13.6-fold increase in mass activity (5.32 A mgRu+Ni−1, with a mass loading of 0.01 mg cm−2 for Ru and 0.03 mg cm−2 for Ru + Ni) compared to that of Pt/C (0.39 A mgPt−1, with a mass loading of 0.05 mg cm−2 for Pt) at −0.05 V vs. RHE (as per this potential scale hereafter) (Supplementary Fig. 20). Given their similar electrochemical active surface area (ECSA) estimated from the double-layer capacitance (Cdl) of RuNi/NC (14.7 mF cm‒2) and Pt/C (14.5 mF cm‒2) (Supplementary Fig. 21), this result indicates a substantially enhanced intrinsic activity of RuNi nanoalloy active sites compared to Pt. This conclusion can be supported by the significantly higher ECSA-normalized current density of 0.44 mA cm‒2 for RuNi/NC, compared to the 0.12 mA cm‒2 for Pt/C at −0.1 V (Supplementary Fig. 22). To eliminate the mass loading effect and obtain a clearer idea of the intrinsic activity, the turnover frequency (TOF) is calculated for all catalysts (Supplementary Fig. 23). The results demonstrate that RuNi/NC exhibits a significantly higher TOF value than Pt/C across the whole potential range, achieving a TOF of 1.70 s−1 at an overpotential of 100 mV, which is nearly double that of Pt/C (0.89 s−1). The underlying reaction kinetics is analyzed by Tafel plots shown in Fig. 3b, revealing that RuNi/NC possesses a Tafel slope as low as 30.9 mV dec‒1, significantly superior to Pt/C (48.1 mV dec‒1), indicating a rapid HER kinetic essence of RuNi/NC following the highly desired Volmer–Tafel mechanism38. This enhanced reaction rate is further corroborated by electrochemical impedance spectroscopy (EIS) measurements, which confirm the lowest charge transfer resistance of RuNi/NC among the samples (Supplementary Fig. 24). Notably, beyond its high performance in alkaline media, RuNi/NC also demonstrates high HER electrocatalytic activity in 0.5 M H2SO4 (pH 0.33 ± 0.03) and 1 M PBS (pH 7.02 ± 0.02) electrolytes (Supplementary Figs. 25‒28), with impressively low overpotentials of 40 and 56 mV at 10 mA cm‒2, respectively (Fig. 3c and Supplementary Table 5). These results endow RuNi/NC a high-performance pH-universal HER electrocatalyst across various application scenarios.Fig. 3: Electrocatalytic HER performances of RuNi/NC.a LSVs in 1 M KOH. b Tafel plots in 1 M KOH. c Comparison of overpotentials at 10 mA cm‒2 (η10) in 1 M KOH, 0.5 M H2SO4 and 1 M PBS. d Chronoamperometry for stability tests of RuNi/NC and Pt/C at 10 mA cm‒2 in 1 M KOH without iR-compensation. The testing conditions include RDE surface area: 0.196 cm2, catalyst loading: 0.25 mg cm‒2, rotation rate: 1600 rpm, scan rate: 5 mV s‒1, temperature: 25 °C, electrolyte pH: 13.74 ± 0.12 (1 M KOH), 0.33 ± 0.03 (0.5 M H2SO4), and 7.02 ± 0.02 (1 M PBS), electrolyte resistance: 6.8 ± 0.2 Ω (1 M KOH), 6.2 ± 0.2 Ω (0.5 M H2SO4), 12.1 ± 0.1 Ω (1 M PBS). e LSVs of RuNi/NC || RuO2 and Pt/C || RuO2 for overall water electrolysis. f Chronopotentiometry of RuNi/NC || RuO2 for overall water electrolysis at 10 mA cm−2. The testing conditions include carbon paper electrode surface area: 1 cm2, catalyst loading: 0.25 mg cm‒2, cell configuration: single cell, scan rate: 5 mV s‒1, temperature: 25 °C, electrolyte: 1 M KOH (pH 13.74 ± 0.12), and without iR-compensation. g LSVs of RuNi/NC || RuO2 and Pt/C || RuO2 for chlor-alkali electrolysis. h Chronopotentiometry of RuNi/NC || RuO2 for chlor-alkali electrolysis at 10 mA cm‒2. The testing conditions include electrode surface area: 1 cm2 (carbon paper or low-carbon steel), catalyst loading: 0.25 mg cm‒2, cell configuration: H-type cell, membrane: cation exchange membrane, scan rate: 5 mV s‒1, temperature: 90 °C, electrolytes: saturated NaCl (anolyte, pH 7.01 ± 0.03), and 3 M NaOH and 3 M NaCl (catholyte, pH 14.12 ± 0.11), and without iR-compensation.To further examine the durability of RuNi/NC for HER, a chronopotentiometric test was conducted at 10 mA cm‒2 in 1 M KOH. As depicted in Fig. 3d, RuNi/NC exhibits a quite high catalytic stability for up to 1600 h, with a minimal potential fading rate of only 0.036 mV per hour, which is significantly lower than the 0.67 mV per hour observed for Pt/C over a duration of just 130 h under identical conditions. Additionally, the catalytic stability of RuNi/NC was evaluated at higher current densities ranging from 50 to 300 mA cm‒2, revealing similarly high stability over durations of 140 – 200 h (Supplementary Fig. 29). Such striking robustness validates the high structural stability and corrosion resistance of RuNi/NC, which can be attributed to its structural superiority, including robust metal-support interaction and finely tailored electronic structures. To ascertain the actual contribution of RuNi nanoalloys for HER electrocatalysis, a poisoning experiment was further implemented using potassium thiocyanide (KSCN) as a probe reagent due to its strong complexation ability to metal atoms39. As shown in Supplementary Fig. 30, the introduction of 10 mM KSCN into the electrolyte significantly decreased the HER current density on RuNi/NC and increased the overpotential to 125 mV at 10 mA cm‒2, implying the inhibition of RuNi active sites by KSCN molecules. This indicates that the superior HER performance is predominantly attributed to the RuNi nanoalloys. Importantly, the high electrocatalytic HER performance of RuNi/NC exceeds that of the majority of previously reported Ru- and Ni-based leading HER electrocatalysts in both overpotential and stability metrics at a current density of 10 mA cm‒2 in 1 M KOH electrolyte (Supplementary Table 6).Overall water electrolysis and chlor-alkali electrolysis performanceMotivated by the high alkaline HER performance of RuNi/NC, we further explored its practical utility in overall water electrolysis in a 1 M KOH electrolyte, paired with a commercial RuO2 anode in a single cell. As shown in Fig. 3e, LSVs reveal that the cell necessitates only 1.42 V (without iR compensation) to reach a current density of 10 mA cm‒2, significantly outperforming the Pt/C || RuO2 electrolyzer (1.61 V), as well as the majority of previously reported Ru-/Ni-based alkaline water electrolysis HER electrocatalysts (Supplementary Table 7). Moreover, the RuNi/NC || RuO2 electrolyzer demonstrates an impressive capability to sustain a steady current density of 10 mA cm−2 throughout a prolonged test lasting over 40 h with only minimal voltage decay (Fig. 3f). Notably, during the water electrolysis, drastic bubbling was observed at both the RuNi/NC cathode and RuO2 anode, while the H2 bubble flow on RuNi/NC was obviously smoother without any adhesion (Supplementary Fig. 31 and Supplementary Movie 2), highlighting the super-hydrophilic merit of RuNi/NC surface conducive to the rapid liberation of active sites.To further assess the practical applicability of RuNi/NC, we conducted experiments mimicking industrial high-temperature (90 °C) chlor-alkali electrolysis conditions. This process simultaneously generates caustic soda, chlorine gas (Cl2), and H2 through brine electrolysis40,41. The electrochemical cell consists of two chambers separated by a cation-exchange membrane, with the anode chamber containing a saturated NaCl solution (pH 7.01 ± 0.03) and the cathode chamber filled with a mixture of 3 M NaCl and 3 M NaOH (pH 14.12 ± 0.11) (Supplementary Fig. 32). RuNi/NC was used as the cathode HER catalyst, while commercial RuO2 was used as the anode catalyst for the chlorine evolution reaction (ClER, E0 = 1.36 V vs. RHE). As shown in Fig. 3g, it is noteworthy that this RuNi/NC || RuO2 electrode pair achieves the chlor-alkali electrolysis at a current density of 10 mA cm-2 with a low cell voltage of 1.96 V, significantly outperforming the Pt/C || RuO2 combination which requires 2.27 V under identical conditions. Notably, this cell voltage represents by far the lowest reported for chlor-alkali electrolysis (Supplementary Table 8). Moreover, the authenticity of the anodic ClER is confirmed by iodometric analysis, and rotating ring-disk electrode (RRDE) studies demonstrate an average ClER selectivity exceeding 90% at 1.9 V vs. RHE (Supplementary Figs. 33 and 34). Furthermore, durability tests show that the RuNi/NC || RuO2 system maintained stable performance during continuous operation at 10 mA cm-2 for 120 h, with minimal voltage fluctuations, highlighting the robustness of RuNi/NC against brine-induced corrosion (Fig. 3h). These performances position RuNi/NC as a promising candidate for large-scale chlor-alkali electrolysis and other electrochemical processes requiring sustained HER catalysts.Mechanic understanding of the tip-intensified HER activityTo study how local interfacial structural manipulation benefits the catalytic activity, we conducted COMSOL Multiphysics finite element simulations to understand the relationship between nanotip curvature and cation concentration at the electrode-electrolyte interface42,43,44,45. Figure 4a illustrates the increase in electron density at the tip as curvature increases, resulting in a localized enhancement of the electrostatic field. Quantitative analysis show that reducing the curvature radius from 10 to 1 nm leads to a nearly four-fold increase in the local electric field strength of the electrical double layer (EDL), from 0.67 × 105 – 2.6 × 105 kV m‒1 at ‒0.1 VRHE (Supplementary Fig. 35a). This enhancement in electric field strength suggests a significant concentration of hydrated K+ cations in the outer Helmholtz layer within the EDL. Applying the Gouy-Chapman-Stern model, we find that a tip with a 2 nm curvature radius exhibits approximately double the surface K+ concentration compared to a tip with a 4 nm radius (Fig. 4b and Supplementary Fig. 35b–d). Interestingly, our simulations indicate that tip curvature has a more substantial effect on surface K+ concentration than the applied potential (Fig. 4c). These findings support the hypothesis that the tip effect might be leveraged in catalyst design to disrupt the interfacial hydrogen bonding network by inducing K+ enrichment within the EDL. This approach may facilitate the release of trapped OH* intermediates, thereby enhancing HER activity46.Fig. 4: Finite element simulation, DFT calculations and in situ synchrotron X-ray absorption spectroscopic study.a Color maps of electron density distribution on the surfaces of electrodes with different curvature radii of 2, 4, and 6 nm. b Simulated surface density distribution of K+ on a tip electrode with a curvature radius of 2 nm. c The applied potential-dependent local K+ concentration on different curved tips. d Free energy diagrams of H2O adsorption/dissociation on different sites without K+ (denoted as E-field) and with K+ (denoted as K+/E-field). e PDOS of Ru and Ni sites with K+ on RuNi/NC surface toward H2O adsorption. f Free energy diagrams of H* adsorption on different sites without K+ (denoted as E-field) and with K+ (denoted as K+/E-field). g Comparison of H2O dissociation and H* adsorption free energies on different sites. h Schematic illustration of the proposed HER catalytic mechanism on RuNi/NC. i XANES and (j) EXAFS spectra at the Ru K-edge, and (k) XANES and (l) EXAFS spectra at the Ni K-edge of RuNi/NC recorded at different applied voltages during electrocatalytic HER process.To delve deeper into the potential influence of concentrated interfacial K+ cation on regulating the HER elementary steps, including H2O adsorption/dissociation and H* adsorption/recombination, density functional theory (DFT) calculations were performed. The slabs were modeled based on the hexagonal RuNi (110) surface given its thermodynamically favored stability (Supplementary Data 1, and Supplementary Figs. 36 and 37). In the presence of a local electric field with a magnitude set at ‒1 V/Å29,43, two distinct scenarios were constructed: a layer of H2O molecules without K+ (denoted as E-field) and with K+ (denoted as K+/E-field). Subsequently, the intermediate adsorption on both the Ru site and Ni site were compared. The Gibbs free energy change diagram for H2O adsorption and dissociation reveals that, in the absence of interfacial K+, both Ru and Ni sites could readily adsorb H2O (Fig. 4d). However, the introduction of K+ renders a selectively discouraged H2O adsorption on the Ni site, while minimally affecting the Ru site, implying a preference for H2O adsorption on the oxyphilic Ru site. Notably, a spontaneous H2O dissociation only on the Ru site is simulated in the presence of interfacial K+, suggesting the pivotal role of interfacial hydrated K+ in promoting H2O dissociation and thus accelerating HER kinetics. The strong H2O adsorption/dissociation capability on the Ru site is further evidenced by the computed partial density of states (PDOS) of H2O adsorption with K+ and E-field (Fig. 4e), which shows a larger orbit overlap and upshifted d-band center (εd) on Ru site than that on Ni site. Moreover, Fig. 4f illustrates the Gibbs free energy of H* adsorption (ΔGH*), where the presence of K+ and E-field endows Ni site with an optimally favorable ΔGH* approaching zero, implying the dissociated H* intermediates may overflow from Ru site to the adjacent Ni site and preferentially absorb on there. This assertion is further supported by the significantly lower activation free energy calculated for H* migration from a Ru site to an adjacent Ni site compared to that towards another adjacent Ru site (Supplementary Fig. 38). To perceive the impact of interfacial K+ on regulating the adsorption behavior of intermediates on Ru and Ni sites, Fig. 4g presents the specific Gibbs free energies for both H2O dissociation and H* adsorption on Ru and Ni sites. By contrastive analysis, we can deduce that, the tip-intensified HER electrocatalysis of RuNi/NC likely proceed through a cooperative relay between neighboring Ru and Ni sites facilitated by hydrated K+. This hypothesized mechanism entails H2O adsorption and dissociation on Ru site, followed by the migration of dissociated H* and subsequent Tafel-type recombination on Ni site, i.e. an H* spillover-bridged Volmer‒Tafel electrocatalytic process, as schematically depicted in Fig. 4h.To experimentally validate the catalytic mechanism of RuNi/NC and elucidate the specific roles of Ru and Ni sites, in situ XAS measurements were performed under operational HER conditions to scrutinize the local structural evolutions and electronic changes of RuNi/NC (Supplementary Fig. 39). Figure 4i presents the Ru K-edge XANES spectra at varying applied potentials, a slight shift of the absorption edge to the higher energy side under open circuit voltage (OCV) as compared to the ex situ spectrum, indicates an increase in the oxidation state of Ru. This shift is likely due to the adsorption of H2O or OH– on Ru atoms, causing electron delocalization and partial surface atom rearrangement47,48. As the applied potential shifted to –0.025 and –0.05 V, a gradual shift of absorption edge towards lower energy was observed, indicating a reduced oxidation state of Ru subsequent to H2O dissociation. This shift signifies rapid H2O dissociation under the reductive conditions of HER, corroborating the efficacy of Ru in H2O dissociation. Accordingly, the Ru K-edge FT-EXAFS spectra exhibited a noticeable increase in the alloying degree as the applied potential decreased (Fig. 4j), and upon returning to OCV, the Ru–Ni coordination peak recovered, suggesting a highly resilient active sites of RuNi/NC for HER. In contrast, the Ni K-edge XANES spectra exhibited imperceptible absorption edge shift under different applied potentials (Fig. 4k), and the intensity of the first-shell peak in the FT-EXAFS spectra marginally increased at –0.05 V (Fig. 4l), probably reflecting an increased alloying degree under cathodic potential. This insensitivity in electronic structure change implies that Ni atoms may not participate in the redox process during HER, aligning well with the non-faradaic Tafel process involving the recombination of two H* to generate the H2 gas, thus affirming the role of Ni as an H*-adsorption site in HER. The foregoing analysis unequivocally establishes that HER catalysis by RuNi/NC adheres to a relayed Volmer–Tafel mechanism, wherein Ru and Ni respectively facilitate the H2O dissociation-related Volmer process and H* recombination-related Tafel process, which are bridged by the H* spillover in between.
