Analytical-grade ruthenium chloride hydrate (RuCl₃·xH₂O), cobalt(II) chloride hexahydrate (CoCl₂·6H₂O), potassium bicarbonate (KHCO₃), anhydrous ethanol (C₂H₅OH), urea (CO(NH₂)₂), 1 M PBS (pH 7.0), ethylene glycol (C₂H₆O₂), hydrochloric acid (HCl, 36-38 wt%), and potassium hydroxide (KOH) were obtained from Aladdin. NF was obtained from Shanghai Jiaqingyuan Co., Ltd. Each chemical was employed as received, without an additional purification step.

CoRuO_x electrocatalysts were deposited directly on NF by co-precipitation and then calcined. Before deposition, a 1 cm × 2 cm NF piece was sonicated for 10 min in 40 mL of 3 M HCl to strip the native oxide layer. It was subsequently sonicated in 40 mL of anhydrous ethanol for 3 min, washed repeatedly with deionized water until neutral, and dried at ambient temperature.

RuCl₃·xH₂O and CoCl₂·6H₂O were used as the Ru and Co sources, respectively. To prepare Co₁Ru₁(OH)_x, 0.0414 g (0.2 mmol) of RuCl₃·xH₂O, 0.0476 g (0.2 mmol) of CoCl₂·6H₂O, and 0.012 g (0.2 mmol) of urea were dispersed in 40 mL of ethylene glycol by sonication. The same protocol was applied to Co₂Ru₁(OH)_x, Co₃Ru₁(OH)_x, Co₄Ru₁(OH)_x, and Co₅Ru₁(OH)_x. In these syntheses, the Ru precursor was kept at 0.0414 g (0.2 mmol), whereas the mass of CoCl₂·6H₂O was set to 0.0952, 0.143, 0.1903, and 0.238 g to obtain Co/Ru molar ratios of 2:1, 3:1, 4:1, and 5:1. A Co-only solution was prepared by sonicating CoCl₂·6H₂O (0.1903 g, 0.8 mmol) and urea (0.012 g, 0.2 mmol) in 40 mL of ethylene glycol, corresponding to the CoO_x sample. The corresponding Ru-only solution contained RuCl₃·xH₂O (0.0414 g, 0.2 mmol) and urea (0.012 g, 0.2 mmol) in the same volume of ethylene glycol, corresponding to the RuO_x sample.

The cleaned NF was placed vertically in each precursor solution, which was maintained at 95 °C for 12 h. The recovered specimens were alternately rinsed three times with anhydrous ethanol and deionized water to eliminate residual ions and weakly adsorbed material, followed by air drying at room temperature. The hydroxide intermediates were labeled Co₁Ru₁(OH)_x, Co₂Ru₁(OH)_x, Co₃Ru₁(OH)_x, Co₄Ru₁(OH)_x, Co₅Ru₁(OH)_x, Co(OH)_x, and Ru(OH)_x. They were placed in a crucible, heated at 2 °C min^-1 to 300 °C, and calcined for 2 h. The products were denoted Co₁Ru₁O_x, Co₂Ru₁O_x, Co₃Ru₁O_x, Co₄Ru₁O_x, Co₅Ru₁O_x, CoO_x, and RuO_x, respectively. For electrochemical testing, a 1 cm × 1 cm catalyst-coated region was exposed to the electrolyte; the other 1 cm × 1 cm region was insulated and attached to the electrode clip. Reported current densities were normalized to the 1.0 cm² exposed geometric area.

Sample morphology was examined by scanning electron microscopy (SEM, JSM-6701F, JEOL, Japan) at 5 kV. Transmission electron microscopy and high-resolution TEM (TEM/HRTEM, JEM-2100F, JEOL, Japan; 200 kV) were used to resolve the nanoscale and crystallographic features, while associated energy-dispersive X-ray spectroscopy (EDS) maps provided elemental distributions. Surface composition and chemical states were analyzed by X-ray photoelectron spectroscopy (XPS, AXIS ULTRA DLD, Kratos Analytical, UK) using monochromatic Al Kα radiation (hυ = 1,486.6 eV) at 15 kV and 10 mA. Binding energies were referenced to the C 1s signal at 284.8 eV. Raman measurements employed a LabRAM HR Evolution system (HORIBA Scientific, France), a 473 nm laser, and a 600 grooves mm^-1 grating. X-ray absorption fine-structure (XAFS) data spanning 4.5-20 keV were acquired on a SuperXAFS-H3000 benchtop hard X-ray absorption spectrometer (Shanghai Institute of Applied Physics, CAS, China). Fourier-transform infrared spectra were collected with a Vertex 80 Fourier Transform Infrared Spectroscopy (FTIR) instrument (Bruker, Germany).

Electrochemical measurements were performed using a CS350M workstation from CH Instruments. Catalyst-coated NF with a 1 cm² exposed area was the working electrode for HER testing in KOH; Hg/HgO and graphite were used as the reference and counter electrodes, respectively. Except for the potentials monitored directly during durability experiments, all measured potentials were converted to the reversible hydrogen electrode (RHE) scale. For a Hg/HgO electrode filled with 1 M KOH, the conversion followed Ref.^[21]:

(1) $$ \begin{equation} \begin{aligned} E_{{vs. } R H E}=E_{v s . H g / H g O}+E_{H g / H g O}^{\theta}+0.059 \times p H \end{aligned} \end{equation} $$

Potentials measured with an Ag/AgCl reference electrode containing saturated KCl were converted as described in Ref.^[22]:

(2) $$ \begin{equation} \begin{aligned} E_{{vs. } R H E}=E_{v s . A g / A g C l}+E_{A g / A g C l}^{\theta}+0.059 \times p H \end{aligned} \end{equation} $$

The measured pH values were 14, 7, and 8 for 1 M KOH, 1 M PBS, and saturated KHCO₃, respectively.

Linear sweep voltammograms were collected at 5 mV s^-1 with 90% iR compensation. To estimate the electrochemical double-layer capacitance (C_dl), cyclic voltammograms in 1 M KOH were obtained at 10, 20, 40, 60, 80, 100, and 120 mV s^-1. The electrochemical surface area (ECSA) was then determined from C_dl using:

(3) $$ \begin{equation} \begin{aligned} \mathrm{ECSA}=\left(\frac{\mathrm{C}_{\mathrm{dl} \times \mathrm{S}}}{\mathrm{C}_{\mathrm{s}}}\right) \end{aligned} \end{equation} $$

Here, S is the geometric area of the working electrode, and C_s is the specific capacitance assigned to an ideally smooth surface. A C_s value of 40 μF cm^-2, selected from prior electrochemical reports, was used in this study^[23]. For 1 M PBS and saturated KHCO₃, CV data were collected at 20, 40, 60, 80, and 100 mV s^-1.

The durability of Co₄Ru₁O_x was evaluated using a direct-current power supply for 200 h. The starting current density was -100 mA cm^-2 in 1 and 6 M KOH and -10 mA cm^-2 in 1 M PBS. Overall water splitting was evaluated in a membrane electrode assembly (MEA) containing Co₄Ru₁O_x as the cathode and NF as the anode; its linear sweep voltammetry (LSV) curve was recorded at 50 mV s^-1. For Zn-CO₂ testing, Zn foil and Co₄Ru₁O_x served as the anode and cathode, respectively, under continuous CO₂ supply. Polarization data were acquired at 50 mV s^-1, and the overpotentials were derived from RHE-converted potentials.

Alkaline HER activity was first compared for the prepared CoRuO_x electrodes in 1 M KOH. Co₄Ru₁O_x was the most active composition: overpotentials of 78, 127, and 196 mV produced 100, 300, and 700 mA cm^-2, respectively. At 100 and 300 mA cm^-2, these values were lower than those of Co₁Ru₁O_x (262 and 340 mV), Co₅Ru₁O_x (127 and 225 mV), and commercial Pt/C (115 and 250 mV) [Figure 1A and B]. Its advantage over Pt/C increased further at 700 mA cm^-2, supporting its suitability for high-current alkaline operation^[9,24]. The Tafel plot in Figure 1C gave a slope of 21.64 mV dec^-1 for Co₄Ru₁O_x, smaller than those of the other prepared electrodes and Pt/C, and therefore indicative of faster HER kinetics^[25].

Cyclic voltammetry was used to probe the origin of the high HER activity. Co₁Ru₁O_x, Co₂Ru₁O_x, Co₃Ru₁O_x, Co₄Ru₁O_x, Co₅Ru₁O_x, and Pt/C were scanned from 10 to 120 mV s^-1 [Supplementary Figure 1]. As presented in Figure 1D, the C_dl of Co₄Ru₁O_x was 20.9 mF cm^-2, exceeding the values for Co₅Ru₁O_x (3.51 mF cm^-2), Co₁Ru₁O_x (15.44 mF cm^-2), and Pt/C (13.2 mF cm^-2; Supplementary Figure 2). The calculated ECSA values are summarized in Figure 1E. Co₄Ru₁O_x reached 523 cm², compared with 330 cm² for Pt/C, consistent with a larger number of accessible sites. Chronopotentiometry in 1 M KOH was then used to assess durability. At -100 mA cm^-2, the potential of Co₄Ru₁O_x changed negligibly over 200 h [Figure 1F], demonstrating robust alkaline stability.

The polarization response in 6 M KOH was compared with that in 1 M KOH in Figure 1G. Larger overpotentials, especially at high current, were observed in the more concentrated electrolyte. The greater viscosity and lower water activity of 6 M KOH can impede ion diffusion, transport near the electrode, and detachment of gas bubbles^[26,27]. A smaller population of free water molecules may also slow water dissociation, an important kinetic step in alkaline HER^[28,29]. Thus, any conductivity gain at higher KOH concentration is offset by less favorable mass transfer and interfacial reaction conditions. Nevertheless, Co₄Ru₁O_x needed only 121, 162, and 223 mV to reach 100, 300, and 700 mA cm^-2 in 6 M KOH. In addition, operation at -100 mA cm^-2 was maintained for 200 h [Figure 1H]. These observations demonstrate catalytic activity and long-term durability over a wide range of alkaline concentrations.

Morphological, structural, and elemental data for Co₄Ru₁O_x are compiled in Figure 2. Before the durability experiment, the SEM image in Figure 2A shows an interconnected, comparatively uniform porous network. This architecture increases access to reaction sites and supports electrolyte infiltration and reactant transport, which benefit HER^[30]. The pre-test EDS maps [Supplementary Figure 3] show broadly even distributions of Co, Ru, Ni, and O. After 200 h of HER operation, most of the porous skeleton remains visible, although limited reconstruction and small surface particles appear [Figure 2B]. The post-test maps in Supplementary Figure 4 continue to show well-retained distributions of all four elements, with no clear sign of migration. These observations support the structural and compositional durability of Co₄Ru₁O_x. TEM, HRTEM, and local EDS mapping were additionally used to resolve the catalyst structure. Lattice fringes are evident in the TEM image and in enlarged HRTEM views in Figure 2C. Spacings of approximately 0.318 and 0.204 nm correspond to the RuO₂ (110) and Co₃O₄ (400) planes, respectively, confirming RuO_x- and CoO_x-related oxide domains^[31,32]. The TEM image in Supplementary Figure 5 and EDS maps in Figure 2D further demonstrate good dispersion of Co, Ru, Ni, and O in the chosen region. Accordingly, the material combines locally crystalline oxide domains with homogeneous elemental dispersion.

XPS was used to compare the surface states of Co₄Ru₁O_x before and after the 200 h HER test in 1 M KOH [Supplementary Figure 6]. The survey spectra in Supplementary Figure 6A contain signals from Co, Ru, O, C, and Ni in both specimens, showing that the principal surface elements persist after extended operation. In the Co 2p spectra [Supplementary Figure 6B], the initial Co 2p_3/2 and Co 2p_1/2 components occur at 780.9 and 796.9 eV, with satellites at 789.5 and 801.2 eV^[33,34]. Following the durability test, the principal components were observed at 781.1 and 797.3 eV, and the satellites at 787.6 and 801.6 eV. The modest changes in position and line shape indicate that the Co-containing surface species remain largely intact. The O 1s spectra in Supplementary Figure 6C contain surface-oxygen and lattice-oxygen contributions. Their initial positions, 531.9 and 530.9 eV, shift to 531.4 and 530.7 eV after testing^[34,35], while the two-component profile is preserved. This response is consistent with a stable oxygen environment and oxide framework. Supplementary Figure 6D covers the overlapping C 1s/Ru 3d region. Before testing, the 284.8 eV feature arises from C-C/C=C bonds, and the 280.2 and 286.6 eV components are associated with Ru 3d^[36]. Ru-related signals remain readily observable after 200 h without an obvious loss in overall intensity, confirming retention of surface Ru.

The electronic states and short-range coordination of Co and Ru were probed by XPS and XAFS to understand the superior HER activity of Co₄Ru₁O_x. The high-resolution spectra indicate appreciable Co-Ru electronic coupling [Supplementary Figure 7]. Relative to RuO_x, the Ru 3d_5/2 component shifts by about 0.5 eV, from 280.7 to 280.2 eV, in Co₄Ru₁O_x. The Co 2p_3/2 and Co 2p_1/2 components likewise move from 781.4 and 797.5 eV in CoO_x to 780.9 and 796.9 eV in the mixed oxide, decreases of approximately 0.5 and 0.6 eV^[37]. The correlated shifts support electron-density redistribution between the two metals and the formation of modified surface electronic and chemical environments after Ru is introduced^[38].

Differences between the Co K-edge XANES spectra of Co₄Ru₁O_x and CoO_x [Supplementary Figure 8] provide further evidence that the electronic environment of Co is altered. Fourier-transformed EXAFS also reveals a change in local coordination. CoO_x displays a principal contribution near 2.6 Å that is assigned to Co-O-Co coordination^[39], whereas Co₄Ru₁O_x has a feature near 1.9 Å associated with Co-M (M = Co or Ru) scattering and a weaker Co-O-Co signal. Ru addition therefore reconstructs the local environment of Co and electronically tunes the cobalt-oxide framework. Taken together, the XPS and XAFS data show that Ru changes both the electronic state and coordination of Co, producing catalytic centers with favorable interfacial electronic properties. This reconstruction promotes interfacial electron transfer and regulates the interaction between the active sites and hydrogen intermediates, which helps explain the high HER activity of Co₄Ru₁O_x.

Practical HER behavior was examined in a MEA electrolyzer^[40,41]. The cathode consisted of Co₄Ru₁O_x on NF, and uncoated NF was used as the anode; this device is denoted Co₄Ru₁O_x||NF. Figure 3A depicts its cathode-membrane-anode arrangement. Figure 3B shows polarization curves for a 1 × 1 cm²||1 × 1 cm² cell at several temperatures. Increasing the temperature from 30 to 70 °C continuously increased current density, consistent with accelerated interfacial kinetics, ion transport, and gas-bubble removal. At 2.0 V, the current density rose from 0.43 A cm^-2 at 30 °C to 1.59 A cm^-2 at 70 °C [Figure 3C]. During a 200 h chronopotentiometric experiment at 100 mA cm^-2, the cell voltage remained close to 1.8 V with only small variations [Figure 3D]. The stable response demonstrates the promise of Co₄Ru₁O_x||NF for membrane-electrolyzer operation.

Catalyst response in neutral and near-neutral media was assessed in 1 M PBS and saturated KHCO₃ [Figure 4]. PBS measurements used a three-electrode arrangement with graphite and Ag/AgCl as the counter and reference electrodes. In Figure 4A and B, Co₄Ru₁O_x reaches 10, 100, and 300 mA cm^-2 at 16, 217, and 554 mV, whereas Pt/C requires 17, 371, and 667 mV. The two electrodes are comparable at low current, but the mixed oxide needs substantially less overpotential at higher current^[17]. The Tafel slopes are 99.51 mV dec^-1 for Co₄Ru₁O_x and 169.55 mV dec^-1 for Pt/C [Supplementary Figure 9], again favoring the oxide catalyst. CV and C_dl analyses in Supplementary Figures 10-13 give 7.36 mF cm^-2 for Co₄Ru₁O_x and 12.21 mF cm^-2 for Pt/C. Because superior high-current activity is obtained despite the smaller C_dl, the advantage in PBS is attributed primarily to faster reaction kinetics, more effective use of surface sites, and porous-structure-assisted mass transport. The potential remained nearly constant over 200 h at -10 mA cm^-2 [Figure 4C], confirming durability in the phosphate buffer.

Saturated KHCO₃ was also examined in a three-electrode cell using graphite as the counter electrode and Hg/HgO as the reference. In this electrolyte, Co₄Ru₁O_x again outperformed Pt/C. The oxide catalyst required 173, 262, and 325 mV to produce 100, 300, and 500 mA cm^-2, respectively, compared with 212, 388, and 532 mV for Pt/C [Figure 4D and E]. These values demonstrate effective HER operation in a near-neutral bicarbonate medium. Tafel slopes from Supplementary Figure 14 were 51.85 and 53.88 mV dec^-1 for Co₄Ru₁O_x and Pt/C, indicating similar kinetics with a small advantage for the former. Supplementary Figures 15-17 yield C_dl values of 20.41 and 38.51 mF cm^-2, respectively. Thus, the lower overpotentials of Co₄Ru₁O_x cannot be explained by a larger electrochemical area; more efficient active-site utilization and reaction kinetics are the more likely origins.

Device-level utility was investigated by using Co₄Ru₁O_x as the cathode catalyst in a Zn-CO₂ cell^[42,43]. In the mechanism described by Kim et al.^[20], hydration of dissolved CO₂ generates carbonic acid and produces a mildly acidic cathodic environment favorable for HER. Simultaneously, Zn oxidation at the anode supplies electrons, allowing hydrogen and electricity to be generated together. The Co₄Ru₁O_x-based cell attained 9.6 mW cm^-2, vs. 6.6 mW cm^-2 for the Pt/C cell [Figure 4F]. These results extend the favorable neutral and near-neutral response observed in PBS and KHCO₃ to a complete Zn-CO₂ device.

Table 1 benchmarks Co₄Ru₁O_x against recent Ru- and Ru-Co-containing HER electrocatalysts in terms of overpotential, Tafel slope, electrolyte, test current, and durability. In 1 M KOH, Co₄Ru₁O_x combines an overpotential of 78 mV at -100 mA cm^-2, a 21.64 mV dec^-1 Tafel slope, and 200 h of continuous operation. This balance of activity and durability compares favorably with many reported Ru-based and Ru-Co-based materials and is consistent with cooperative Co/Ru oxide effects and a tuned surface electronic environment. Unlike most alkaline-centered reports, the present catalyst also remains active in neutral PBS and bicarbonate media, demonstrating compatibility with distinct proton-transfer environments.

Electrochemical impedance spectroscopy was used to clarify why Co₄Ru₁O_x remains highly active even though its C_dl is lower than that of Pt/C in PBS and saturated KHCO₃ (Nyquist plots, Supplementary Figure 18). The solution resistances (R_s) of Co₄Ru₁O_x and Pt/C are 1.44 and 1.92 Ω in 1 M KOH, 4.06 and 5.83 Ω in 1 M PBS, and 1.52 and 2.25 Ω in saturated KHCO₃, respectively. Although R_s varies with electrolyte conductivity, its consistently smaller value for the mixed oxide indicates lower ohmic losses under each HER condition.

Operando Raman spectra were next collected in 1 M KOH, 1 M PBS, and saturated KHCO₃ to probe the HER interface [Supplementary Figure 19]. Similar potential-dependent changes in the broad O-H stretching envelope near 3,500 cm^-1 appear in all three media, showing that interfacial water activation is retained. The elementary HER sequence is comparable, but its rate is shaped by electrolyte-specific proton donors, conductivity, and interfacial solvation. In KOH, water dissociation precedes formation of adsorbed hydrogen. The near-neutral PBS environment supplies fewer protons and normally slows proton transfer; nevertheless, effective water activation at Co₄Ru₁O_x supports proton delivery and HER^[54]. In saturated KHCO₃, bands near 1,015 and 1,064 cm^-1 are assigned to HCO₃^- and CO₃^2-, respectively^[56,57], evidencing a dynamic interfacial bicarbonate/carbonate equilibrium that buffers local proton transfer. The absence of new bands associated with heavy carbonate or bicarbonate accumulation suggests that these species do not substantially obstruct the active interface. FTIR spectra in Supplementary Figure 20 show hydroxyl-stretching and water-bending bands near 3,404 and 1,630 cm^-1, respectively, for both RuO_x and Co₄Ru₁O_x^[58]. The stronger hydroxyl-related absorption of the mixed oxide indicates greater interaction with interfacial water. Its electrolyte tolerance therefore derives from maintained water activation together with effective charge and proton transport, rather than from identical kinetics in every electrolyte.