Modulating Ru-Co bond lengths in Ru₁Co single-atom alloys through crystal phase engineering for electrocatalytic nitrate-to-ammonia conversion

Abstract

Single atom alloys (SAAs) with maximum atomic efficiency and uniform active sites show great promise for heterogeneous catalytic applications. Meanwhile, crystal phase engineering has granered significant interest due to tailored atomic arrangements and coordination environments. However, the crystal phase engineering of SAAs remains challenging owing to high surface energy and complex phase transition dynamics. Herein, Ru₁Co SAAs with tunable crystal phases (hexagonal-close-packed (hcp), face-centered-cubic (fcc), and hcp/fcc structure) are successfully synthesized via controlled phase transitions. These SAAs exhibit distinct crystal phase-dependent performance towards nitrate reduction reaction (NO₃RR), where hcp-Ru₁Co outperforms its counterparts with a NH₃ Faradaic efficiency of 96.78% at 0 V vs. reversible hydrogen electrode and long-term stability exceeding 1200 h. Mechanistic investigations reveal that the hcp configurations enables shorter Ru-Co distances, stronger interatomic interactions, and more positive surface potential compared to hcp/fcc-Ru₁Co and fcc-Ru₁Co, which enhances the NO₃⁻ adsorption, reduces the free energy barrier, and suppresses competitive hydrogen evolution.

Introduction

Noble metal-based nanomaterials have attracted considerable attention in electrocatalysis due to their unique catalytic properties^1,2. Theoretically, modulating the crystal structure can tailor the coordination geometry and electronic configuration of active sites, thereby affecting the adsorption/desorption behavior of reactants and intermediates, which ultimately regulates catalytic performance^3,4. However, the inherent characteristics of metallic bonds (their unsaturated and non-directional nature) tend to drive metal catalysts toward conventional crystal structures^5,6. For instance, Ru typically forms a hexagonal-close-packed (hcp) structure while Pt adopts a face-centered-cubic (fcc) arrangement. This structural persistence presents significant challenges in both controlling crystal phase modifications and establishing clear structure–property relationships in metallic catalytic systems.

To minimize noble metal usage while enhancing catalytic performance, the atomic dispersion of noble metal atoms within a host metal matrix to form single-atom-alloys (SAAs) has emerged as an effective strategy^7,8,9. SAAs can realize the maximum utilization of metal atoms, uniform active sites, peculiar geometric and electronic structures, which collectively contribute to superior catalytic properties^10,11,12,13. The well-defined active centers in SAAs further facilitate fundamental investigations of structure–activity relationships and catalytic mechanisms. Nevertheless, constructing SAAs with tailored crystal structures remains challenging due to the high surface energy of isolated metal atoms and the lack of effective synthetic methods for precise crystal phase control at the nanoscale. Notably, successful demonstrations of crystal phase engineering in SAAs have not yet been reported.

Herein, we successfully synthesize Ru₁Co SAAs with distinct crystal phases (i.e., hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co) through a solvothermal method followed by thermal annealing. These obtained Ru₁Co SAAs are employed as electrocatalysts for nitrate reduction reaction (NO₃RR) to ammonia (NH₃). Remarkably, hcp-Ru₁Co demonstrates competitive performance with a NH₃ Faradaic efficiency (${{{\rm{FE}}}}_{{{{\rm{NH}}}}_{3}}$) of 96.78% at 0 V vs. reversible hydrogen electrode (RHE). The energy consumption (EC) and NH₃ production cost for hcp-Ru₁Co are calculated as 19.04 kWh kg⁻¹ and 0.57 $_USD kg⁻¹, respectively. Additionally, hcp-Ru₁Co maintains operational stability for over 1200 h. Combined experimental and theoretical analyses reveal that hcp-Ru₁Co possesses reduced Ru–Co distances, enhanced Ru–Co interatomic interactions, and more positive surface potential compared to hcp/fcc-Ru₁Co and fcc-Ru₁Co, collectively promoting NO₃⁻ adsorption, lowering the free energy barrier, and suppressing hydrogen evolution reaction (HER). A prototype Zn-NO₃⁻ battery employing hcp-Ru₁Co as the cathode catalyst demonstrates high electrochemical performance.

Results

Synthesis and structural characterizations

The Ru₁Co SAAs with different crystal phases are successfully synthesized via a solvothermal method followed by thermal annealing (see the “Methods” section for experimental details and Fig. 1a for schematic illustration). Specifically, CoCl₂·6H₂O and RuCl₃·xH₂O precursors are dissolved in 1,4-butanediol and heated at 230 °C for 20 min. Subsequent annealing at 300, 500, and 700 °C for 30 min yields hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co, respectively.

Fig. 1: Scheme and XRD characterization.

a Scheme for the formation of Ru₁Co SAAs with different crystal phases. b–d The Rietveld refinement of XRD patterns of hcp-Ru₁Co (b), hcp/fcc-Ru₁Co (c), and fcc-Ru₁Co (d). Source data for this figure are provided as a Source Data file.

The crystal phases of synthesized samples are characterized by X-ray diffraction (XRD). As shown in Fig. 1b, d, the diffraction peaks of hcp-Ru₁Co match well with those of hcp-Co (JCPDS No. 05-0727), while fcc-Ru₁Co exhibits the diffraction peaks well indexed to fcc-Co (JCPDS No. 15-0806). Notably, the diffraction peaks of both hcp-Ru₁Co and fcc-Ru₁Co exhibit slightly negative shifts compared to those of hcp-Co and fcc-Co, indicating the successful incorporation of Ru with a large atomic radius into the Co matrix. The XRD pattern of hcp/fcc-Ru₁Co shows the coexistence of diffraction peaks from both phases (Fig. 1c), verifying the successful synthesis of hcp/fcc-Ru₁Co. Rietveld refinement profiles demonstrate high phase purity in all Ru₁Co SAAs. Inductively coupled plasma-optical emission spectroscopy (ICP-OES) analysis confirms consistent Co/Ru atomic ratios of 99.25/0.75 in all Ru₁Co SAAs (Supplementary Table 1).

The morphology of hcp-Ru₁Co is investigated through transmission electron microscopy (TEM) and high-angle annular dark-field scanning TEM (HAADF-STEM), which reveals a homogeneous nano-urchin structure (Fig. 2a, Supplementary Figs. 1 and 2). Atomic-scale characterization via aberration-corrected HAADF-STEM (AC-HAADF-STEM) utilizes Z-contrast imaging, where Ru atoms (Z = 44) exhibit brighter atomic-scale contrast compared to Co (Z = 27). As shown in Fig. 2b, isolated Ru atoms are observed without detectable clusters or nanoparticles, confirming the atomic dispersion of Ru in the Co matrix. Three-dimensional (3D) surface plots converted from the cyan square in Fig. 2b are analyzed. High-intensity spots represent Ru atoms, while low-intensity spots correspond to Co atoms. The isolated Ru spots in Fig. 2c provide additional evidence of Ru atomic dispersion. The atomic arrangement of hcp-Ru₁Co is further revealed by high-resolution AC-HAADF-STEM (Fig. 2d). An enlarged view from the cyan square in Fig. 2d shows the atomic columns along the [110] zone axis (Fig. 2e), closely matching the simulated atomic model of hcp-Co (Fig. 2g). The corresponding fast Fourier transform (FFT) pattern (Fig. 2f) aligns with the simulated electron diffraction pattern of hcp-Co (Fig. 2h). Interplanar spacings measured from the intensity profiles (converted from orange rectangles in Fig. 2d) are 0.21 and 0.19 nm, corresponding to the (110) and (111) planes of hcp-Ru₁Co, respectively (Fig. 2i, j). Notably, the presence of isolated peaks with relatively high intensity further confirms the atomic dispersion of Ru. HAADF-STEM image and corresponding elemental mappings demonstrate the uniform distribution of Ru and Co in the nano-urchin structure (Fig. 2k–n). Energy-dispersive X-ray spectroscopy (EDS) reveals the Ru/Co atomic ratio of 99.24/0.76 (Supplementary Fig. 3), consistent with ICP-OES results.

Fig. 2: Characterization of hcp-Ru₁Co.

a TEM image of hcp-Ru₁Co. b HAADF-STEM image of hcp-Ru₁Co. c 3D surface plots converted from the cyan square in (b). d High-resolution AC-HAADF-STEM image of hcp-Ru₁Co. e Enlarged AC-HAADF-STEM image taken from the cyan square in (d). f The corresponding FFT pattern. g and h The simulated atomic model (g) and the electron diffraction pattern (h) of hcp-Co viewed along the [110] zone axis. i and j Intensity profiles from the orange rectangles in (d). k–n HAADF-STEM image (k) and corresponding elemental mappings (l–n) of hcp-Ru₁Co. Source data for this figure are provided as a Source Data file.

The structural characteristics of fcc-Ru₁Co are systematically investigated. TEM image confirms the preserved nano-urchin morphology (Fig. 3a and Supplementary Fig. 4). Atomic dispersion of Ru in the Co matrix is verified by HAADF-STEM (Fig. 3b) and 3D surface plot analysis (Fig. 3c). An enlarged AC-HAADF-STEM image (Fig. 3e) from the purple square in Fig. 3d shows the atomic columns along the [110] zone axis, matching the simulated atomic model of fcc-Co (Fig. 3g). The corresponding FFT pattern (Fig. 3f) aligns with the simulated diffraction pattern of fcc-Co (Fig. 3h). Interplanar spacing measurements yield 0.21 nm, corresponding to the (111) plane of fcc-Ru₁Co, with isolated Ru peaks confirming atomic dispersion (Fig. 3i, j). For hcp/fcc-Ru₁Co, TEM image reveals the conserved nano-urchin structure (Fig. 3k and Supplementary Fig. 5). HAADF-STEM image (Fig. 3l) and 3D surface plots (Fig. 3m) confirm the atomic dispersion of Ru. High-resolution AC-HAADF-STEM image reveals two types of atomic arrangements with distinct heterophase boundary (Fig. 3n). FFT patterns (Fig. 3o and p) from the selected purple and cyan squares in Fig. 3n, align with the simulated diffraction patterns of fcc-Co along the [110] zone axis (Fig. 3q) and hcp-Co along the [111] zone axis (Fig. 3r), respectively. FFT pattern (Supplementary Fig. 6b) from the boundary region (orange square) in Supplementary Fig. 6a, reveals the overlapped electron diffraction dots along the [110] zone axis of fcc-Co ([110]_fcc) and the [111] zone axis of hcp-Co ([111]_hcp) (Supplementary Fig. 6c). The intensity profiles from the orange rectangles in Fig. 3n reveal the interplanar spacings of 0.21 and 0.20 nm, assigned to the (111) plane of fcc-Ru₁Co and (011) plane of hcp-Ru₁Co, respectively (Fig. 3s, t). The existence of the isolated maximum peaks further confirms Ru atomic dispersion.

Fig. 3: Characterization of fcc-Ru₁Co and hcp/fcc-Ru₁Co.

a TEM image of fcc-Ru₁Co. b AC-HAADF-STEM image of fcc-Ru₁Co. c 3D surface plots converted from the purple square in (b). d High-resolution AC-HAADF-STEM image of fcc-Ru₁Co. e Enlarged HAADF-STEM image extracted from the purple square in (d). f The corresponding FFT pattern. g and h The simulated atomic model (g) and the electron diffraction pattern (h) of fcc-Co oriented along the [110] zone axis. i and j Intensity profiles from the orange rectangles in (d). k TEM image of hcp/fcc-Ru₁Co. l AC-HAADF-STEM image of hcp/fcc-Ru₁Co. m 3D surface plots extracted from the blue square in (h). n High-resolution AC-HAADF-STEM image of hcp/fcc-Ru₁Co. o and p FFT patterns extracted from the squares in (n). q and r The simulated electron diffraction patterns oriented along the [110]_fcc and [111]_hcp zone axes. s and t Intensity profiles from the rectangles in (n). Source data for this figure are provided as a Source Data file.

The electronic structure and coordination environment of Ru₁Co SAAs with different crystal phases are investigated by X-ray photoelectron spectroscopy (XPS) and X-ray absorption spectroscopy (XAS). XPS analysis confirms that metallic Co and Ru are dominant in all Ru₁Co SAAs (Supplementary Fig. 7). The Ru 3p XPS spectrum of hcp-Ru₁Co exhibits peaks at 461.38 and 483.56 eV, assigned to 3p_3/2 and 3p_1/2 of Ru⁰, respectively¹⁴. A positive shift in the binding energy of Ru⁰ 3p_3/2 peak is observed for hcp/fcc-Ru₁Co (461.79 eV) and fcc-Ru₁Co (462.01 eV) compared to hcp-Ru₁Co (461.38 eV), suggesting reduced electron density on Ru in hcp-Ru₁Co. The observed oxides are attributed to the easy oxidation characters of Co and Ru in air, in accordance with previous reports^15,16. Figure 4a displays the normalized Co K-edge X-ray absorption near edge structure (XANES) spectra of Ru₁Co SAAs and references (i.e., Co foil and Co₃O₄). The absorption edges of all Ru₁Co SAAs are close to Co foil but with slight shifts towards higher energy, indicating the metallic characteristics of Co with reduced electron density. This is attributed to the lower electronegativity of Co (1.9) compared to Ru (2.2), resulting in electron transfer from Co to Ru¹⁷. This directional electron transfer is further corroborated by the decreased valence state of Ru in hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co relative to Ru foil, as revealed in the normalized Ru K-edge XANES spectra (Fig. 4c). Notably, the Co K-edge adsorption of hcp-Ru₁Co positively shifts compared to fcc-Ru₁Co and hcp/fcc-Ru₁Co (inset of Fig. 4a), with the opposite trend observed for the Ru K-edge adsorption (inset of Fig. 4c), suggesting a stronger electronic interaction between Co and Ru in hcp-Ru₁Co.

Fig. 4: Chemical state and coordination environment.

a The normalized Co K-edge XANES spectra of hcp-Ru₁Co, hcp/fcc-Ru₁Co, fcc-Ru₁Co, Co foil, and Co₃O₄. b The Co K-edge FT-EXAFS spectra of hcp-Ru₁Co, hcp/fcc-Ru₁Co, fcc-Ru₁Co, Co foil, and Co₃O₄. c The normalized Ru K-edge XANES spectra of hcp-Ru₁Co, hcp/fcc-Ru₁Co, fcc-Ru₁Co, Ru foil, and RuO₂. d The Ru K-edge FT-EXAFS spectra of hcp-Ru₁Co, hcp/fcc-Ru₁Co, fcc-Ru₁Co, Ru foil, and RuO₂. e The Ru K-edge EXAFS spectra of hcp-Ru₁Co, hcp/fcc-Ru₁Co, fcc-Ru₁Co, Ru foil, and RuO₂. f-i The WT-EXAFS contour plots of hcp-Ru₁Co (f), hcp/fcc-Ru₁Co (g), fcc-Ru₁Co (h), and Ru foil (i) at Ru K-edge. j In situ CO-DRIFTS spectra of hcp-Ru₁Co at room temperature. Source data for this figure are provided as a Source Data file.

Extended X-ray absorption fine structure (EXAFS) spectroscopy is performed to probe the local atomic configurations. The Co K-edge Fourier transformed EXAFS (FT-EXAFS) spectra of Ru₁Co SAAs display a prominent peak at 2.17 Å, matching the characteristic Co–Co bond observed in Co foil (Fig. 4b). This consistent peak position results from the low Ru concent and the predominant Co–Co bond in Ru₁Co SAAs. The Ru K-edge FT-EXAFS spectrum of Ru foil shows a peak at 2.35 Å, corresponding to the Ru–Ru metallic bond (Fig. 4d). In contrast, hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co exhibit the prominent peaks at 2.10, 2.13, and 2.17 Å, respectively. The significant differences in peak positions between Ru foil and Ru₁Co SAAs suggest that the prominent peak of Ru₁Co SAAs is not contributed by Ru–Ru metallic bond, but Ru–Co metallic bond. The absence of Ru–Ru bond further confirms the atomic dispersion of Ru in Ru₁Co SAAs. Notably, the Ru–Co bond length in hcp-Ru₁Co is shorter than that in fcc-Ru₁Co and hcp/fcc-Ru₁Co, suggesting enhanced interatomic interactions between Ru and Co in hcp-Ru₁Co, consistent with the XANES results.

The least-squares EXAFS fitting curves and fitting parameters are provided in Supplementary Figs. 8–17 and Tables 2 and 3. The Co K-edge EXAFS oscillation functions reveal minor variations in both oscillation frequencies and amplitudes for hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co, indicating similar coordination environment for Co species (Supplementary Fig. 18). In contrast, the Ru K-edge EXAFS oscillation functions of Ru₁Co SAAs exhibit significant differences, reflecting substantial variations induced by crystal phase differences (Fig. 4e). The Co K-edge wavelet-transformed EXAFS (WT-EXAFS) contour plot of Co foil shows a local maximum at k = 7.21 Å⁻¹ and R = 2.17 Å, similar to that of hcp-Ru₁Co (k = 7.25 Å⁻¹, R = 2.17 Å), hcp/fcc-Ru₁Co (k = 7.42 Å⁻¹, R = 2.17 Å), and fcc-Ru₁Co (k = 7.59 Å⁻¹, R = 2.17 Å) (Supplementary Fig. 19). However, the Ru K-edge WT-EXAFS contour plot reveals that hcp-Ru₁Co exhibits a local maximum at k = 7.69 Å⁻¹ and R = 2.10 Å, which is significantly different from Ru foil (k = 9.52 Å⁻¹, R = 2.35 Å) and assigned to Ru–Co bond (Fig. 4f, i). Additionally, the lower R value in hcp-Ru₁Co (2.10 Å) in comparison with hcp/fcc-Ru₁Co (2.13 Å) and fcc-Ru₁Co (2.17 Å) further confirms the shorter Ru–Co bond in hcp-Ru₁Co (Fig. 4g, h). These results collectively confirm that hcp-Ru₁Co possesses a shorter Ru–Co bond and thus stronger interatomic interaction compared to hcp/fcc-Ru₁Co and fcc-Ru₁Co.

To further confirm the atomic dispersion of Ru in Ru₁Co SAAs with different crystal phases, in situ CO diffuse reflectance infrared Fourier transform spectroscopy (CO-DRIFTS) is performed (Fig. 4j and Supplementary Fig. 20). Following CO adsorption at room temperature, two broad peaks at 2170 and 2121 cm⁻¹ are observed, corresponding to the mononuclear (Ru^δ+(CO)) and tricarbonyl species adsorbed on single Ru atom (Ru^δ+(CO)₃), respectively^18,19,20. The absence of peaks below 2000 cm⁻¹, which is the fingerprint of bridge-bonded CO, confirms the absence of Ru clusters or nanoparticles^21,22.

Electrocatalytic NO₃RR performance

The electrocatalytic NO₃RR performance is evaluated using an H-type cell. Ru₁Co SAAs supported on Ni foam (NF) serve as the working electrode, with Ag/AgCl (saturated KCl) and Pt wire as the reference and counter electrodes, respectively. The electrolyte contains 0.5 M Na₂SO₄ and 0.1 M NaOH with 200 ppm NO₃⁻-N. All potentials are referenced to the RHE scale. Figure 5a displays the linear sweep voltammetry (LSV) curves of Ru₁Co SAAs in electrolyte with/without 200 ppm NO₃⁻-N. Apparently, upon the addition of NO₃⁻, significant enhancement in current density is observed, indicating the occurrence of NO₃RR. Notably, hcp-Ru₁Co demonstrates the highest current density in the presence of NO₃⁻ and the lowest current density in the absence of NO₃⁻, indicating superior NO₃RR activity and suppressed hydrogen evolution reaction (HER) (Supplementary Fig. 21).

Fig. 5: Electrocatalytic NO₃RR performance.

a LSV curves of hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co at a scan rate of 5 mV s⁻¹ in an electrolyte containing 0.5 M Na₂SO₄ and 0.1 M NaOH with and without 200 ppm NO₃⁻-N at room temperature (25 °C). b ${{\mbox{FE}}}_{{{\mbox{NH}}}_{3}}$ of hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co at different potentials. c Comparison of ${Y}_{{{\mbox{NH}}}_{3}}$, ${{\mbox{FE}}}_{{{\mbox{NH}}}_{3}}$, ${C}_{{{\mbox{NO}}}_{3}^{-}}$, ${S}_{{{\mbox{NH}}}_{3}}$, ${j}_{{{\mbox{NH}}}_{3}}$, and ${{\mbox{EE}}}_{{{\mbox{NH}}}_{3}}$ over hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co at 0 V. d NH₃ production cost of hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co at different potentials. e E_a of hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co at 0 V. f ${Y}_{{{\mbox{NH}}}_{3}}$ of hcp-Ru₁Co at 0 V obtained by colorimetric and ¹H NMR methods (the insets are UV–Vis absorption and ¹H NMR spectra of post-electrolyzed solution over hcp-Ru₁Co at 0 V). g Chronopotentiometry curve of hcp-Ru₁Co in an H-type continuous-flow cell at 20 mA cm⁻² with corresponding ${Y}_{{{\mbox{NH}}}_{3}}$ and ${{\mbox{FE}}}_{{{\mbox{NH}}}_{3}}$. h In situ Raman spectra of hcp-Ru₁Co during chronopotentiometry test at different potentials in an electrolyte containing 0.5 M Na₂SO₄ and 0.1 M NaOH with 200 ppm NO₃⁻-N. All potentials are referenced to the RHE scale without iR compensation. Catalyst loading amount for all NO₃RR tests is 3 mg cm⁻². Error bars represent the standard deviations from three independent measurements. Source data for this figure are provided as a Source Data file.

The NO₃RR performance of Ru₁Co SAAs is systematically investigated through chronoamperometric tests at different potentials. The concentrations of NO₃⁻-N and NH₄⁺-N are quantified by colorimetric methods via UV–Vis spectrophotometer (Supplementary Figs. 22 and 23). As shown in Fig. 5b and Supplementary Fig. 24, hcp-Ru₁Co exhibits superior performance with maximum NH₃ yield rate (${Y}_{{{{\rm{NH}}}}_{3}}$) of 1.33 ± 0.06 mg h⁻¹ cm⁻² at −0.1 V and NH₃ Faradaic efficiency (${{{\rm{FE}}}}_{{{{\rm{NH}}}}_{3}}$) of 96.78 ± 1.97% at 0 V, significantly outperforming hcp/fcc-Ru₁Co (0.90 ± 0.06 mg h⁻¹ cm⁻², 84.41 ± 3.70%) and fcc-Ru₁Co (0.39 ± 0.05 mg h⁻¹ cm⁻², 56.94 ± 2.60%). The minimal concentrations of by-products (i.e., NO₂⁻, N₂H₄, N₂, and H₂) confirm the high NH₃ selectivity of hcp-Ru₁Co (Supplementary Figs. 25–31). At 0 V, hcp-Ru₁Co demonstrates higher values of ${Y}_{{{{\rm{NH}}}}_{3}}$, ${{{\rm{FE}}}}_{{{{\rm{NH}}}}_{3}}$, NO₃⁻ conversion (${C}_{{{{\rm{NO}}}}_{3}^{-}}$), NH₃ selectivity (${S}_{{{{\rm{NH}}}}_{3}}$), half-cell energy efficiencies of NH₃ (${{{\rm{EE}}}}_{{{{\rm{NH}}}}_{3}}$) and NH₃ partial current density (${j}_{{{{\rm{NH}}}}_{3}}$) compared to hcp/fcc-Ru₁Co and fcc-Ru₁Co (Fig. 5c and Supplementary Figs. 32 and 33). Notably, under high-concentration conditions (1400 ppm NO₃⁻-N), hcp-Ru₁Co delivers ${Y}_{{{{\rm{NH}}}}_{3}}$ of 11.65 ± 0.46 mg h⁻¹ cm⁻² at −0.1 V and ${{{\rm{FE}}}}_{{{{\rm{NH}}}}_{3}}$ of 100% at 0 V (Supplementary Fig. 34). These metrics position it among the most active metal-based catalysts for NO₃RR (Supplementary Table 4).

EC and NH₃ production cost analyses are explored only considering the price of renewable electricity (0.03 $_USD kWh⁻¹, full levelized cost of electricity for utility solar power according to the announcement by the US Department of Energy for the 2030 target)^23,24,25. As shown in Supplementary Fig. 35 and Fig. 5d, EC and NH₃ production cost of hcp-Ru₁Co are calculated as 19.04 ± 1.10 kWh kg⁻¹ and 0.57 ± 0.03 $_USD kg⁻¹ at 0.1 V, respectively, significantly lower than those of hcp/fcc-Ru₁Co (37.0 ± 0.37 kWh kg⁻¹, 1.11 ± 0.01 $_USD kg⁻¹) and fcc-Ru₁Co (288.78 ± 0.50 kWh kg⁻¹, 8.66 ± 0.02 $_USD kg⁻¹). Notably, the EC values of hcp-Ru₁Co at all potentials are significantly lower than commercial benchmarks (1–1.5 $ kg⁻¹)²⁶, highlighting its industrial viability. Temperature-dependent ${Y}_{{{{\rm{NH}}}}_{3}}$ measurements reveal the apparent activation energy (E_a) of 3.49 kJ mol⁻¹ for hcp-Ru₁Co, significantly lower than that of hcp/fcc-Ru₁Co (14.22 kJ mol⁻¹) and fcc-Ru₁Co (25.55 kJ mol⁻¹), confirming the accelerated NO₃RR kinetics (Fig. 5e and Supplementary Fig. 36). In addition, hcp-Ru₁Co possesses a larger electrochemically active surface area (ECSA) compared to hcp/fcc-Ru₁Co and fcc-Ru₁Co (Supplementary Fig. 37, Table 5), consistent with the N₂ adsorption–desorption results (Supplementary Fig. 38). ${Y}_{{{{\rm{NH}}}}_{3}}$ values normalized to ECSA and Brunauer–Emmett–Teller (BET) surface area indicate the superior intrinsic activity of hcp-Ru₁Co (Supplementary Figs. 39 and 40). Electrochemical impedance spectroscopy (EIS) reveals reduced charge transfer impedance for hcp-Ru₁Co, attributed to its distinctive hcp structure and large ECSA (Supplementary Fig. 41).

To verify that the produced NH₃ originates from the NO₃RR process catalyzed by hcp-Ru₁Co, control experiments are carried out. Negligible NH₃ is detected in solutions of 0.5 M Na₂SO₄ and 0.1 M NaOH at 0 V or in solutions of 0.5 M Na₂SO₄ and 0.1 M NaOH with 200 ppm NO₃⁻-N at open circuit potential (OCP), and using pure NF substrate as the working electrode (Supplementary Fig. 42). ¹⁵N isotope labeling experiments employing ¹⁵NO₃⁻ as nitrogen source yield characteristic ¹⁵NH₄⁺ peaks at δ = 6.91 and 7.06 ppm in ¹H nuclear magnetic resonance (NMR) spectrum (Supplementary Fig. 43), confirming NH₃ generation through NO₃RR (Supplementary Fig. 44). ${Y}_{{{{\rm{NH}}}}_{3}}$ values obtained from ¹H NMR method match those from colorimetric analysis (Fig. 5f), validating both analytical methods.

Durability is a pivotal aspect in evaluating the performance of electrocatalysts. As depicted in Supplementary Fig. 45, negligible changes in ${Y}_{{{{\rm{NH}}}}_{3}}$ and ${{{\rm{FE}}}}_{{{{\rm{NH}}}}_{3}}$ of hcp-Ru₁Co during 12 consecutive recycling tests demonstrate its robust electrochemical stability. Chronopotentiometric test under continuous electrolyte flow (2 mL min⁻¹, Supplementary Fig. 46) maintains stable ${Y}_{{{{\rm{NH}}}}_{3}}$ and ${{{\rm{FE}}}}_{{{{\rm{NH}}}}_{3}}$ over 1200 h of prolonged operation (Fig. 5g). Post-stability characterizations confirm the retention of nano-urchin morphology, atomic Ru dispersion (Supplementary Fig. 47), crystal structure (Supplementary Fig. 48), and chemical states (Supplementary Fig. 49). In situ Raman spectroscopy during NO₃RR only detects the vibration mode of SO₄²⁻ (981 cm⁻¹) from electrolyte (Fig. 5h, Supplementary Fig. 50)²⁷. No additional Raman peaks are observed at all applied potentials, suggesting that hcp-Ru₁Co does not transform into oxide during NO₃RR. Furthermore, hcp-Ru₁Co exhibits competitive performance across a broad range of NO₃⁻-N concentrations (100–500 ppm), suggesting its wide applicability (Supplementary Fig. 51). As current density increases from 50 to 500 mA cm⁻², ${Y}_{{{{\rm{NH}}}}_{3}}$ of hcp-Ru₁Co progressively rises, attaining a maximum value of 25.52 mg h⁻¹ cm⁻² at 500 mA cm⁻² (Supplementary Fig. 52).

Mechanistic studies

Recent studies have highlighted the critical role of catalyst surface potential in regulating NO₃⁻ adsorption, where elevated surface potentials significantly enhance NO₃⁻ adsorption and improve NO₃RR performance^28,29,30. To explore the surface potential of Ru₁Co SAAs, Kelvin probe force microscopy (KPFM) is conducted (Fig. 6a). The average surface potential of hcp-Ru₁Co achieved by contact potential difference is 155 mV, substantially exceeding that of hcp/fcc-Ru₁Co (100 mV) and fcc-Ru₁Co (44 mV) (Fig. 6b). This enhanced surface potential of hcp-Ru₁Co, attributed to its unique hcp structure, facilitates NO₃⁻ adsorption and thereby accelerates the reaction kinetics. To quantitatively assess NO₃⁻ adsorption behavior, Ru₁Co SAAs are immersed in an electrolyte containing 200 ppm NO₃⁻-N and agitated for 30 min, the NO₃⁻ adsorption capacity (q_e) is determined by measuring the concentration of NO₃⁻ in solution before and after agitation. As shown in Supplementary Fig. 53, hcp-Ru₁Co achieves a q_e value of 1.91 ± 0.05 μmol mg_cat.⁻¹, significantly surpassing hcp/fcc-Ru₁Co (1.20 ± 0.03 μmol mg_cat.⁻¹) and fcc-Ru₁Co (0.77 ± 0.02 μmol mg_cat.⁻¹). Furthermore, Fourier transform infrared (FT-IR) analysis of Ru₁Co SAAs with absorbed NO₃⁻ further confirms the enhanced NO₃⁻ adsorption on hcp-Ru₁Co, evidenced by intensified absorption at 1350 cm⁻¹ corresponding to anti-symmetric stretching vibrations of NO₃⁻ (Supplementary Fig. 54)³¹. These findings collectively demonstrate that hcp-Ru₁Co possesses a stronger NO₃⁻ adsorption, contributing to its enhanced NO₃RR performance.

Fig. 6: Mechanistic studies.

a Surface potential distribution of hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co. b The corresponding surface potentials. c In situ ATR-SEIRAS spectra of hcp-Ru₁Co at different potentials. d Online DEMS spectra of hcp-Ru₁Co during the NO₃RR process. e The PDOS profiles of hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co. The vertical dashed lines indicate the position of the d-band center. f and g The optimized structure models of hcp-Ru₁Co (f) and fcc-Ru₁Co (g). h and i 2D deformation charge density difference maps of hcp-Ru₁Co (h) and fcc-Ru₁Co (i). j The charge density difference of *NO₃ adsorbed on hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co (electron accumulation and depletion are indicated by yellow and cyan regions, respectively). k The free energy diagrams of NO₃RR over hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co. Source data for this figure are provided as a Source Data file.

To probe into the intermediates formed during the NO₃RR process, in situ attenuated total reflection surface-enhanced infrared absorption spectroscopy (ATR-SEIRAS, Supplementary Fig. 55) and online differential electrochemical mass spectrometry (DEMS, Supplementary Fig. 56) are carried out. In situ ATR-SEIRAS spectra of hcp-Ru₁Co (Fig. 6c) display characteristic peaks at 1595, 1483, 1438, 1235, and 1168 cm⁻¹, attributed to the vibrations of *NO, *NH₃, *NH₂, *NO₂, and *NH₂OH, respectively^32,33. The peak at 1668 cm^-1 is assigned to the bending vibration of the adsorbed H₂O³⁴. Online DEMS analysis during chronoamperometry test at 0 V reveals the mass signals of mass-to-charge ratios (m/z) of 46, 33, 32, 31, 30, and 17, attributed to *NO₂, *NH₂OH, *NHOH, *NHO, *NO, and *NH₃, respectively (Fig. 6d and Supplementary Fig. 57). Based on these detected intermediates, the NO₃RR pathway is deduced, involving the following key intermediates: *NO₃ → *NO₂ → *NO → *NHO → *NHOH → *NH₂OH → *NH₂ → *NH₃.

To gain preliminary insights into the enhanced NO₃RR performance of hcp-Ru₁Co, density functional theory (DFT) calculations are performed. The optimized structure models are provided in Supplementary Fig. 58 and Supplementary Data 1. Partial density of state (PDOS) profiles reveal that hcp-Ru₁Co exhibits a d-band center at −1.03 eV, positioned closer to the Fermi level compared to hcp/fcc-Ru₁Co (−1.33 eV) and fcc-Ru₁Co (−1.36 eV) (Fig. 6e). The upshift of the d-band center for hcp-Ru₁Co could theoretically favor the adsorption of intermediates. Additionally, the effect of the crystal structure on Ru–Co interatomic interaction is explored. As shown in the structure models (Fig. 6f, g), hcp-Ru₁Co possesses a relatively shorter Ru–Co bond length (2.53 Å) than fcc-Ru₁Co (2.55 Å), consistent with XAS findings. 2D deformation charge density maps and Bader charge analysis demonstrate electrons transfer from Co to Ru (Fig. 6h, i), with hcp-Ru₁Co showing a slightly larger charge transfer, indicating an enhanced Ru–Co interatomic interaction in hcp-Ru₁Co.

Gibbs free energy of ${{{\rm {NO}}}}_{3}^{-}$ adsorption $\left(\Delta {G}_{*{{{\rm {NO}}}}_{3}}\right)$ reveals that hcp-Ru₁Co exhibits a more negative $\Delta {G}_{*{{{\rm {NO}}}}_{3}}$ of −0.74 eV compared to hcp/fcc-Ru₁Co (−0.53 eV) and fcc-Ru₁Co (−0.05 eV) (Supplementary Fig. 59, Supplementary Data 1). This suggests a stronger NO₃⁻ adsorption on hcp-Ru₁Co, which aligns with the results of NO₃⁻ adsorption capacity and FT-IR experiments in Supplementary Figs. 53 and 54. Charge density difference maps and Bader charge analysis demonstrate that 0.82 electrons are transferred from hcp-Ru₁Co to *NO₃, exceeding the amount transferred on hcp/fcc-Ru₁Co (0.80 electrons) and fcc-Ru₁Co (0.74 electrons) (Fig. 6j). Free energy diagrams identify the hydrogenation of *NO (*NO → *NHO) as the potential-determining step (PDS) for Ru₁Co SAAs (Fig. 6k, Supplementary Figs. 60–62, Supplementary Data 1). Notably, hcp-Ru₁Co exhibits a lower energy barrier of PDS (0.35 eV) compared to hcp/fcc-Ru₁Co (0.39 eV) and fcc-Ru₁Co (0.60 eV). Considering that HER is a primary competitive reaction, the free energy of *H adsorption is calculated (Supplementary Figs. 63, 64, Supplementary Data 1). The free energy required to generate by-product H₂ for hcp-Ru₁Co is 0.91 eV, higher than that for hcp/fcc-Ru₁Co (0.51 eV) and fcc-Ru₁Co (0.09 eV). This suggests that hcp-Ru₁Co could potentially suppress the competing HER.

The performance of the Zn­-NO₃ ⁻ battery

The Zn-NO₃⁻ battery offers dual functionality by generating electricity while converting NO₃⁻ pollutants into valuable NH₃³⁵. The Zn-NO₃⁻ battery is constructed by anchoring hcp-Ru₁Co on NF as the cathode, paired with a zinc plate anode (Supplementary Fig. 65). The hcp-Ru₁Co assembled Zn-NO₃⁻ battery exhibits a stable OCP of 1.405 V vs. Zn/Zn²⁺ (Supplementary Fig. 66). Two such batteries connected in series can power a bulb of 1.5 V (Supplementary Fig. 67). Additionally, the hcp-Ru₁Co assembled Zn-NO₃⁻ battery delivers a high power-density of 6.55 mW cm⁻², surpassing most recently reported Zn-NO₃⁻ batteries (Supplementary Fig. 68 and Table 6). At current densities of 2, 4, and 6 mA cm⁻², the specific capacities of this battery are 576.81, 616.84, and 668.51 mAh g⁻¹, respectively (Supplementary Fig. 69). Consequently, the corresponding energy densities under these conditions are calculated as 470.03, 448.13, and 416.04 mWh g⁻¹. The battery demonstrates high rate-capability and stable operation at current densities ranging from 1 to 20 mA cm⁻² (Supplementary Fig. 70). Continuous discharge–charge cycling at 2 mA cm⁻² suggests its robust stability (Supplementary Fig. 71). Additionally, it achieves an ${Y}_{{{{\rm{NH}}}}_{3}}$ of 1.72 ± 0.02 mg h⁻¹ cm⁻² at 50 mA cm⁻² and a ${{{\rm{FE}}}}_{{{{\rm{NH}}}}_{3}}$ of 90.93 ± 0.80% at 40 mA cm⁻² (Supplementary Fig. 72). Minimal fluctuations in both ${Y}_{{{{\rm{NH}}}}_{3}}$ and ${{{\rm{FE}}}}_{{{{\rm{NH}}}}_{3}}$ during consecutive recycling test at 40 mA cm⁻² confirm robust long-term stability (Supplementary Fig. 73).

Discussion

In summary, Ru₁Co SAAs with distinct crystal phases, i.e., hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co, are successfully synthesized through precise control of the phase transition process and exhibit crystal phase-dependent performance in NO₃RR to NH₃. Notably, hcp-Ru₁Co demonstrates superior catalytic performance with ${{{\rm{FE}}}}_{{{{\rm{NH}}}}_{3}}$ of 96.78 ± 1.97% at 0 V, EC of 19.04 ± 1.10 kWh kg⁻¹, NH₃ production cost of 0.57 ± 0.03 $_USD kg⁻¹, and stable operation exceeding 1200 h, significantly outperforming hcp/fcc-Ru₁Co and fcc-Ru₁Co. Insightful experiments and theoretical calculations reveal that hcp-Ru₁Co is featured with shorter Ru–Co interatomic distances, stronger Ru–Co interatomic interactions, and more positive surface potential, which enhance the adsorption capacity of NO₃⁻, reduces the free energy barrier of PDS, and suppresses the HER. Furthermore, hcp-Ru₁Co, when integrated as the cathode electrocatalyst into Zn-NO₃⁻ battery, demonstrates its significant potential for application in advanced energy devices. This study not only achieves precise control over the crystal phase of SAAs but also provides comprehensive insights into the relationship between the crystal structure of SAAs and catalytic performance.

Methods

Chemicals and reagents

Cobalt chloride hexahydrate (CoCl₂·6H₂O, 99.0 wt%), ruthenium(III) chloride hydrate (RuCl₃·xH₂O, 99.95 wt%), 1,4-butanediol (98 wt%), sodium sulfate anhydrous (Na₂SO₄, 99.0 wt%), deuterium oxide (D₂O, 99.9 at% D), sodium hypochlorite solution (NaClO, available chlorine 5.5–6.5 wt%), ammonium chloride-¹⁵N (¹⁵NH₄Cl, 98.5 wt%), sodium nitrate–¹⁵N (Na¹⁵NO₃, 98.5 wt%), maleic acid (C₄H₄O₄, 98.0 wt%), salicylic acid (C₇H₆O₃, 99.5 wt%), sodium citrate dihydrate (C₆H₅Na₃O₇·2H₂O, 99.0 wt%), N-(1-naphthyl) ethylenediamine dihydrochloride (C₁₂H₁₄N₂·2HCl, 98 wt%), phosphoric acid (H₃PO₄, 85 wt%), sulfanilamide (C₆H₈N₂O₂S, 99.5 wt%), and nitroferricyanide (III) dihydrate (Na₂Fe(CN)₅NO·2H₂O, 99.0 wt%) were obtained from Shanghai Macklin Biochemical Co., Ltd. Sodium hydroxide (NaOH, ≥96.0 wt%), sulfuric acid (H₂SO₄, 95.0–98.0 wt%), hydrogen chloride (HCl, 36.0–38.0 wt%), and ammonium chloride (NH₄Cl, ≥99.5 wt%) were purchased from Sinopharm Chemical Reagent Co., Ltd. Ethanol (C₂H₅OH, ≥99.7 wt%) was purchased from Tianjin Fuyu Fine Chemical Co., Ltd. Acetone (C₃H₆O, ≥99 wt%) was purchased from Yantai Far East Fine Chemical Co., Ltd. Nickel foam (NF, thickness: 2 mm) was purchased from Tianjin EVS Chemical Technology Co., Ltd. Nafion solution (D520, 5 wt%, Dupont) was purchased from Suzhou Sinero Technology Co., Ltd. Nafion 211 membrane (diameter: 3 cm, thickness: 25.4 μm) was purchased from SCI Materials Hub. All reagents were utilized without further purification. Deionized water with a resistivity of 18.2 MΩ was employed throughout all experimental procedures.

Synthesis of Ru₁Co SAAs

0.08 M CoCl₂·6H₂O and 0.15 M NaOH were dissolved in 40 mL of 1,4-butanediol to form solution A. 0.015 g RuCl₃·xH₂O were dissolved in 5 mL of 1,4-butanediol to form solution B. Subsequently, solution B was gradually introduced into solution A under vigorous stirring for 30 min. The mixed solution was then heated to 230 °C for 20 min. The resulting product was washed with ethanol and acetone several times, and dried under vacuum at 60 °C for 3 h. Finally, the precursor powder was annealed at 300, 500, and 700 °C for 30 min at N₂ atmosphere with a heating rate of 5 °C min⁻¹ to form hcp-Ru₁Co, hcp/fcc-Ru₁Co and fcc-Ru₁Co, respectively.

Fabrication of the working electrode

The working electrode was fabricated by immobilizing Ru₁Co SAAs onto the pretreated nickel form (NF) substrate. In detail, NF was sequentially pretreated by sonication in HCl solution (3 M), followed by thorough rinsing with ethanol and deionized water to remove surface impurities. Subsequently, the synthesized sample (hcp-Ru₁Co, hcp/fcc-Ru₁Co, and fcc-Ru₁Co; 20 mg each) was dispersed in a mixed solution consisting of 950 μL ethanol and 50 μL Nafion under sonication for 0.5 h to form a homogeneous ink. Finally, the ink (300 µL) was dropped onto the surface of the pretreated NF (1 × 1 cm²), and dried in air for 30 min to form the working electrode. The mass loading of the catalyst was calculated to be 3 mg cm⁻².

Physicochemical characterizations

XRD patterns were acquired using a Rigaku SmartLab powder diffractometer with Cu Kα radiation (λ = 1.5418 Å). TEM images were recorded using a HITACHI H-8100 electron microscope with an operation voltage of 200 kV. AC-HAADF-STEM was conducted on ThermoFisher Scientific Themis microscope (300 kV) equipped with image and probe correctors, using a beam current of 2 pA, a convergence angle of 25 mrad, and a collection angle range of 55–220 mrad conditions. ICP-OES was carried out on an Agilent 5110 system. XPS measurements were performed on an ESCALABMK II spectrometer equipped with Mg Kα excitation. The C1s peak at 284.8 eV was utilized for calibration. X-ray absorption spectroscopy (XAS) was carried out at the 1W1B beamline of the Shanghai Synchrotron Radiation Facility. In situ CO-DRIFTS experiments were performed on a Bruker Tensor II infrared spectrometer. UV–Vis absorbance spectra were collected using a SHIMADZU UV-1900 spectrophotometer. ¹H NMR spectra were recorded on a Bruker Avance 500 MHz spectrometer. DEMS measurements were conducted on a QAS 100 system (Linglu Instruments Co. Ltd). FT-IR spectroscopy was performed using a Perkin-Elmer spectrometer. In-situ ATR-SEIRAS spectra were collected using an INVENIO S FT-IR spectrometer equipped with a liquid nitrogen-cooled mercury cadmium telluride (MCT) detector. Raman spectra were obtained using a LabRAM HR Evolution (Horiba) with a 532 nm excitation laser. KPFM images were acquired on a BRUKER Dimension Icon with ScanAsyst system. N₂ adsorption–desorption isotherms were measured on a Quantachrome NOVA-3000 instrument.

Electrochemical measurements

Electrochemical experiments were conducted in a H-type cell with a three-electrode configuration by CHI 760E electrochemical workstation (CH Instruments Inc., China). The cathodic and anodic chambers are separated by a Nafion 211 membrane (diameter: 3 cm, thickness: 25.4 μm, DuPont). Prior to use, Nafion membrane underwent sequential pretreatment. First, the membrane was immersed in a 5% H₂O₂ solution and maintained at 80 °C for 1 h to eliminate organic contaminants. Subsequently, the membrane was repeatedly rinsed with deionized water and immersed in deionized water at 80 °C for 1 h to ensure the complete removal of any residual H₂O₂. After pretreatment, the membrane was stored in deionized water under ambient conditions.

A saturated Ag/AgCl electrode and a Pt wire served as the reference and counter electrode, while nickel foam anchored with the prepared Ru₁Co SAAs (mass loading: 3 mg cm⁻²) was used as the working electrode. All potentials were referenced to the RHE scale without iR compensation using Eq. (1):

$${E}_{{{\rm {RHE}}}}={E}_{{{\rm {Ag}}}/{{\rm {AgCl}}}}+0.197+0.059\times {{\rm {pH}}}$$

(1)

where the pH of electrolyte was measured to be 12.98 ± 0.12 using a pH meter (Mettler Toledo).

The electrolyte containing 0.5 M Na₂SO₄, 0.1 M NaOH and 200 ppm NO₃⁻-N, was prepared prior to the NO₃RR tests by dissolving 71.02 g of Na₂SO₄ (0.5 mol), 4.00 g of NaOH (0.1 mol), and 1.21 g of NaNO₃ in 1 L of deionized water within a volumetric flask. The solution was stored at room temperature (25 °C) for subsequent use. The electrolyte volume in cathodic and anodic chamber is 12.5 mL.

LSV test was performed at a scan rate of 5 mV s⁻¹ at a potential range from 0.2 to −0.3 V vs. RHE at room temperature (25 °C). Potentiostatic tests were conducted at different potentials for 1 h, with a stirring rate of 1500 r.p.m. at room temperature (25 °C). After electrolysis, the electrolyte was analyzed by UV–Vis spectrophotometry. Electrochemical double-layer capacitance (C_dl) was assessed using cyclic voltammetry in a non-Faradaic region (−0.45 to −0.55 V vs. Ag/AgCl) at different scan rates of 10, 20, 30, 40, 50, and 60 mV s⁻¹. Electrochemical impedance spectroscopy (EIS) tests were conducted in a frequency range of 0.01 Hz–100 kHz at open circuit potential (−0.036 V vs. RHE) in an electrolyte containing 0.5 M Na₂SO₄, 0.1 M NaOH, and 200 ppm of NO₃⁻-N.

Determination of NO₃ ⁻-N concentration

NO₃⁻-N concentration was determined using UV–Vis spectrophotometry. Initially, 100 μL of cathodic electrolyte were extracted and diluted to a total volume of 10 mL. Then, 200 μL of HCl (1 M) were incorporated to the above solution. Using UV–Vis spectrophotometry, the absorption intensities at wavelengths of 220 nm (A_220nm) and 275 nm (A_275nm) were recorded. The absorbance value (A) was determined employing the equation: A = A_220nm−2 × A_275nm. The UV–Vis absorbance spectra of standard NO₃⁻-N solutions with known concentrations were measured for calibration, resulting in the calibration equation: y = 0.110x + 0.031 with an R² value of 0.999.

Determination of NH₄ ⁺-N concentration

The indophenol blue method was employed to quantify NH₄⁺-N concentration. Initially, 40 μL of cathodic electrolyte were obtained and diluted to a total volume of 4 mL. Subsequently, 4 mL of chromogenic agent composed of 1.0 M NaOH, 0.36 M C₇H₆O₃, and 0.17 M C₆H₅Na₃O₇·2H₂O, 2 mL of NaClO (0.05 M), and 0.4 mL of complexing agent (1 wt% Na₂[Fe(NO)(CN)₅]·2H₂O) were sequentially introduced into the aforementioned solution. Following 1 h’s incubation, UV–Vis absorbance spectra were acquired, and the absorbance at 655 nm was recorded. The UV–Vis spectra of standard NH₄⁺-N solutions with known concentrations were measured for calibration, resulting in the calibration equation: y = 0.459x + 0.032 with an R² value of 0.998.

Determination of NO₂ ⁻-N concentration

Initially, 50 μL of cathodic electrolyte were extracted and diluted to a total volume of 5 mL. Subsequently, 0.1 mL of color reagent composed of 12.9 mM C₁₂H₁₄N₂·2HCl, 0.38 M C₆H₈N₂O₂S and 2.87 M H₃PO₄, were added to the above solution. Following 10 min’s incubation, UV–Vis absorbance spectra were acquired, and the absorbance at 510 nm was recorded. The UV–Vis spectra of standard NO₂⁻-N solutions with known concentrations were measured for calibration, resulting in the calibration equation: y = 0.370x + 0.001 with an R² value of 0.998.

Determination of N₂H₄-N concentration

N₂H₄-N concentration was determined by the Watt and Chrisp method. Initially, a colorimetric reagent was prepared by dissolving 5.99 g C₉H₁₁NO (0.01 mol) in a mixture of HCl (30 mL) and ethanol (300 mL). Subsequently, 5 mL of cathodic electrolyte were extracted and combined with 5 mL of the prepared color reagent. Following 10 min’s incubation at room temperature, UV–Vis absorbance spectra were acquired, and the absorbance at 455 nm was recorded. The UV–Vis spectra of standard N₂H₄-N solutions with known concentrations were tested for calibration, resulting in the calibration equation: y = 0.773x + 0.026 with an R² value of 0.997.

Performance evaluation

To calculate NO₃⁻ conversion (${C}_{{{{\rm{NO}}}}_{3}^{-}}$), Eq. (2) is used as follows:

$${C}_{{{{\rm {NO}}}}_{3}^{-}}=\frac{\Delta {c}_{{{{\rm {NO}}}}_{3}^{-}-{\rm {N}}}}{{C}_{0}}\times 100\%$$

(2)

where ${c}_{0}$ represents the initial concentration of NO₃⁻-N in electrolyte, ${\Delta c}_{{{{\rm{NO}}}}_{3}^{-}-{{\rm{N}}}}$ stands for the variation in the concentration of NO₃⁻-N before and after electrolysis.

To calculate NH₃ selectivity (${S}_{{{{\rm{NH}}}}_{3}}$), Eq. (3) is used as follows:

$${S}_{{{{\rm {NH}}}}_{3}}=\frac{{c}_{{{{\rm {NH}}}}_{3}-{\rm {N}}}}{\Delta {c}_{{{{\rm {NO}}}}_{3}^{-}-{\rm {N}}}}\times 100\%$$

(3)

where ${\Delta c}_{{{{\rm{NO}}}}_{3}^{-}-{{\rm{N}}}}$ is the variation in the concentration of NO₃⁻-N before and after electrolysis, ${c}_{{{{\rm{NH}}}}_{3}-{{\rm{N}}}}$ stands for the concentration of ${{{\rm{NH}}}}_{3}-{{\rm{N}}}$ after electrolysis.

To calculate NH₃ yield rate (${Y}_{{{{\rm{NH}}}}_{3}}$), Eq. (4) is employed as below:

$${Y}_{{{{\rm {NH}}}}_{3}}=\frac{{c}_{{{{\rm {NH}}}}_{3}}\times V}{t\times S}$$

(4)

where ${c}_{{{{\rm{NH}}}}_{3}}$ stands for the concentration of NH₃ after electrolysis, V is the cathodic electrolyte volume, t represents the reduction time, S is the geometric area of the working electrode.

To calculate NH₃ Faradaic efficiency (${{{\rm{FE}}}}_{{{{\rm{NH}}}}_{3}}$), Eq. (5) is used as follows:

$${{{\rm {FE}}}}_{{{{\rm {NH}}}}_{3}}=\frac{8\times {{{\rm{F}}}}\times {c}_{{{{\rm {NH}}}}_{3}}\times V}{17\times Q}\times 100\%$$

(5)

where F stands for the Faradaic constant (96,485 C mol⁻¹), ${c}_{{{{\rm{NH}}}}_{3}}$ is the concentration of NH₃ after electrolysis, V is the cathodic electrolyte volume, and Q is the total charge that has passed through the electrode.

To calculate NO₂⁻ yield rate (${Y}_{{{{\rm{NO}}}}_{2}^{-}}$), Eq. (6) is employed as follows:

$${Y}_{{{{\rm {NO}}}}_{2}^{-}}=\frac{{c}_{{{{\rm {NO}}}}_{2}^{-}}\times V}{t\times S}$$

(6)

where ${c}_{{{{\rm{NO}}}}_{2}^{-}}$ stands for the concentration of NO₂⁻ after electrolysis, V represents the cathodic electrolyte volume, t is the reduction time in h, and S is the geometric area of the working electrode.

To calculate NO₂⁻ selectivity (${S}_{{{{\rm{NO}}}}_{2}^{-}}$), Eq. (7) is used as follows:

$${S}_{{{{\rm {NO}}}}_{2}^{-}}=\frac{{c}_{{{{\rm {NO}}}}_{2}^{-}-N}}{\Delta {c}_{{{{\rm {NO}}}}_{3}^{-}-N}}\times 100\%$$

(7)

where ${\Delta c}_{{{{\rm{NO}}}}_{3}^{-}-{{\rm{N}}}}$ is the variation in the concentration of NO₃⁻-N before and after electrolysis, ${c}_{{{{\rm{NO}}}}_{2}^{-}-{{\rm{N}}}}$ is the concentration of NO₂⁻-N after electrolysis.

To calculate NO₂⁻ Faradaic efficiency (${{{\rm{FE}}}}_{{{{\rm{NO}}}}_{2}^{-}}$), Eq. (8) is used as follows:

$${{{\rm {FE}}}}_{{{{\rm {NO}}}}_{2}^{-}}=\frac{2\times {{{\rm{F}}}}\times {c}_{{{{\rm {NO}}}}_{2}^{-}}\times V}{46\times Q}\times 100\%$$

(8)

where F stands for the Faradaic constant, ${c}_{{{{\rm{NO}}}}_{2}^{-}}$ is the concentration of NO₂⁻ after electrolysis, V is the cathodic electrolyte volume, Q is the total charge that has passed through the electrode.

N₂ selectivity (${S}_{{{{\rm{N}}}}_{2}}$) is obtained by the subtraction method (Eq. 9), as only the N-containing products of NO₂⁻, N₂, and NH₃ can be detected during the NO₃RR process.

$${S}_{{{\rm {N}}}_{2}}=\frac{\Delta {c}_{{{{\rm {NO}}}}_{3}^{-}-N}-{c}_{{{{\rm {NO}}}}_{2}^{-}-N}-{{c}_{{{{\rm {NH}}}}_{3}}}_{-N}}{\Delta {c}_{{{{\rm {NO}}}}_{3}^{-}-N}}$$

(9)

where ${\Delta c}_{{{{\rm{NO}}}}_{3}^{-}-{{\rm{N}}}}$ is the variation in the concentration of NO₃⁻-N before and after electrolysis, ${c}_{{{{\rm{NO}}}}_{2}^{-}-{{\rm{N}}}}$ is the concentration of NO₂⁻-N after electrolysis, ${c}_{{{{\rm{NH}}}}_{3}-{{\rm{N}}}}$ is the concentration of ${{{\rm{NH}}}}_{3}-{{\rm{N}}}$ after electrolysis.

To calculate N₂ Faradaic efficiency (${{{\rm{FE}}}}_{{{{\rm{N}}}}_{2}}$), Eq. (10) is used as follows:

$${{{\rm {FE}}}}_{{{\rm {N}}}_{2}}=\frac{5\times {{{\rm{F}}}}\times {S}_{{{\rm {N}}}_{2}}\times\Delta {c}_{{{\rm {{NO}}}}_{3}^{-}-{\rm {N}}}\times V}{14\times Q}\times 100\%$$

(10)

where F is the Faradaic constant (96485 C mol⁻¹), ${\Delta c}_{{{{\rm{NO}}}}_{3}^{-}-{{\rm{N}}}}$ is the variation in the concentration of NO₃⁻-N before and after electrolysis, V is the cathodic electrolyte volume, Q is the total charge that has passed through the electrode.

H₂ was detected by the gas chromatograph (GC, A91 Plus PANNA) equipped with a thermal conductivity detector (TCD). The gas effluent from the cathodic chamber was directly delivered into the gas sampling loop of GC. H₂ Faradaic efficiency (${{{\rm{FE}}}}_{{{{\rm{H}}}}_{2}}$) is calculated using the following Eq. (11)³⁶:

$${{{\rm {FE}}}}_{{\rm {{H}}}_{2}}=\frac{2\times {{{\rm{F}}}}\times {{V}}_{{{\rm {H}}}_{2}}\times {G}\times {t}\times {{P}}_{0}}{{R}\times {T}\times {Q}\times {10}^{6}}\times 100\%$$

(11)

where F is the Faradaic constant (96,485 C mol⁻¹), ${V}_{{{{\rm{H}}}}_{2}}$ is the volume concentration of H₂ in the exhausted gas from cathodic chamber at a given sampling time (vol%), G is total gas flow rate at room temperature and ambient pressure (20 L min⁻¹), $t$ is the electrolysis time, ${P}_{0}$ is the gas pressure (1.013$\times {10}^{5}$ Pa), R is the ideal gas constant (8.314 J mol⁻¹ K⁻¹), T is the temperature (298 K), and Q is the total charge that has passed through the electrode during the NO₃RR process.

To calculate NH₃ partial current density (${j}_{{{{\rm{NH}}}}_{3}}$), Eq. (12) is employed as follows:

$${j}_{{{\rm {{NH}}}}_{3}}=\frac{{{{\rm {FE}}}}_{{{\rm {{NH}}}}_{3}}\times Q}{t\times S}$$

(12)

where Q is the total charge that has passed through the electrode, t is the reduction time, and S stands for the geometric area of the working electrode.

The half-cell energy efficiency (${{{\rm{EE}}}}_{{{{\rm{NH}}}}_{3}}$) determined by the ratio of fuel energy to the electrical power applied, is calculated using Eq. (13) as follows:

$${{{\rm {EE}}}}_{{{{\rm {NH}}}}_{3}}=\frac{\left(1.23-{E}_{{{{\rm {NH}}}}_{3}}^{0}\right)\times {{{\rm {FE}}}}_{{{{\rm {NH}}}}_{3}}}{1.23-E}$$

(13)

where ${E}_{{{{\rm{NH}}}}_{3}}^{0}$ is the equilibrium potential of the NO₃RR process (0.69 V), E stands for the applied potential.

To calculate energy consumption (EC), Eq. 14 is employed as follows:

$${{{\rm{EC}}}}=\frac{\left(1.23-E\right)\times i\times t}{m}$$

(14)

where E stands for the applied potential, i represents the current, t is the reduction time, $m$ is the mass of produced NH₃ after electrolysis.

To calculate NH₃ production cost, Eq. (15) is employed as below:

$${{{\rm {NH}}}}_{3}{{\rm {production}}\; {\rm {cost}}}=0.03\times {{\rm {EC}}}$$

(15)

where EC stands for the energy consumption, 0.03 ($_USD kWh⁻¹) is the full levelized cost of electricity (LCOE) for utility solar power according to the recent announcement by the US Department of Energy (DOE) for the 2030 target^23,37. Note that this is a simplified calculation method that solely considers the electrical price without taking into account the capital costs and Ohmic losses.

Calculation of ECSA and ECSA-normalized ${Y}_{{{{\rm{NH}}}}_{3}}$

To calculate ECSA, Eq. (16) is used as follows:

$${{\rm {ECSA}}}=\frac{{C}_{{{\rm {dl}}}}}{{C}_{{\rm {s}}}}$$

(16)

where C_s refers to the specific capacitance of a flat metallic surface³⁸, with a value of 40 μF cm⁻². C_dl is the double-layer capacitance, which can be derived from cyclic voltammetry (CV) performed within the non-Faradaic potential range at different scan rates (10–60 mV s⁻¹). The calculation of C_dl is based on Eq. (17) as follows:

$${C}_{{{\rm {dl}}}}=\frac{\Delta i}{2\times v}=\frac{{i}_{{\rm {a}}}-{i}_{{\rm {c}}}}{2\times v}$$

(17)

where i_a and i_c represent the anodic and cathodic current at a certain potential, respectively, ν is the scan rate of CV.

ECSA-normalized ${Y}_{{{{\rm{NH}}}}_{3}}$ is calculated using Eq. (18) as follows:

$${{\rm {ECSA}}}-{{\rm {normalized}}}{Y}_{{{{\rm {NH}}}}_{3}}=\frac{{c}_{{{{\rm {NH}}}}_{3}}\times V}{t\times {{\rm {ECSA}}}}$$

(18)

where V stands for the cathodic electrolyte volume, t is the reduction time, ${c}_{{{{\rm{NH}}}}_{3}}$ is the concentration of NH₃ after electrolysis.

Calculation of BET surface area-normalized ${Y}_{{{{\rm{NH}}}}_{3}}$

${Y}_{{{{\rm{NH}}}}_{3}}$ is normalized to BET surface area according to the following Eq. (19):

$${{\rm {BET}}\; {\rm {surface}}\; {\rm {area}}}-{\rm {{normalized}}}{Y}_{{{{\rm {NH}}}}_{3}}=\frac{{c}_{{{\rm {{NH}}}}_{3}}\times V}{t\times {{\rm {BET}}\; {\rm {surface}}\; {\rm {area}}}}$$

(19)

where V stands for the cathodic electrolyte volume, t represents the reduction time, ${c}_{{{{\rm{NH}}}}_{3}}$ is the concentration of NH₃ after electrolysis. BET surface area is obtained by N₂ adsorption-desorption method.

In situ Raman measurement

Gamry Refence 3000 coupled with an iRaman instrument equipped with a 532 nm laser was employed for in situ Raman measurement. A self-made cell featuring a three-electrode configuration was employed. The electrolyte contained 0.5 M Na₂SO₄, 0.1 M NaOH, and 200 ppm NO₃⁻-N. All Raman spectra were acquired during chronoamperometric experiments.

In situ ATR-SEIRAS measurement

For in situ ATR-SEIRAS measurements, hcp-Ru₁Co was supported on an Au-coated Si crystal substrate as the working electrode, Ag/AgCl and Pt foil functioned as the reference and counter electrodes, respectively. During chronoamperometry experiments at various potentials, spectral data were collected. The electrolyte contained 0.5 M Na₂SO₄, 0.1 M NaOH, and 200 ppm NO₃⁻-N.

Online DEMS measurement

For online DEMS measurements, a peristaltic pump was utilized to continuously flow the electrolyte (0.5 M Na₂SO₄, 0.1 M NaOH, and 200 ppm NO₃⁻-N) into a custom-made electrochemical cell, in which hcp-Ru₁Co loaded on glassy carbon electrode served as the working electrode, while Pt wire and Ag/AgCl were used as the counter and the reference electrodes, respectively. Ar gas was bubbled into the electrolyte constantly before and during the DEMS measurement. The gaseous products generated from the working electrode were brought into a mass spectrometer through a hydrophobic polytetrafluoroethylene (PTFE) membrane (porosity ≥50%, pore diameter ≤20 nm). Chronoamperometry test at 0 V was carried out for 150 s. Once the test was completed and mass signals recovered to their initial levels, the subsequent cycle was initiated under identical conditions. Following four cycles, the measurement was finished.

Isotope labeling experiments

Isotopic labeling experiments utilized Na¹⁵NO₃ (98.5 wt%) as the nitrogen source. The post-electrolysis electrolyte was extracted and adjusted to pH 4 using H₂SO₄. For ¹H NMR test, 1 mL of D₂O with 4.4 mg of C₄H₄O₄ was incorporated into 10 mL of the above solution. The ¹H NMR spectra of standard ¹⁵NH₄⁺ solution with known concentrations were measured for calibration, resulting in the calibration equation: y = 0.016x−0.034 with an R² value of 0.997.

NO₃ ^- adsorption and FT-IR experiments

To ascertain the adsorption capacities of NO₃⁻, Ru₁Co SAAs (10 mg) were immersed in 10 mL of an electrolyte comprising 0.5 M Na₂SO₄, 0.1 M NaOH, and 200 ppm NO₃⁻-N. After stirring for 30 min, the adsorption capacity (q_e) is determined using Eq. (20) as follows:

$${q}_{\rm {{e}}}=\frac{\Delta {c}_{{{{\rm {NO}}}}_{3}^{-}-{\rm {N}}}\times V}{14\times m}$$

(20)

where q_e stands for the adsorption capacity, ${\Delta c}_{{{{\rm{NO}}}}_{3}^{-}-{{\rm{N}}}}$ is the variation in the concentration of NO₃⁻-N before and after stirring, V is the electrolyte volume, m is the mass loading amount of Ru₁Co SAAs.

The Ru₁Co SAAs with absorbed NO₃⁻ underwent sequential washing with deionized water, centrifugation, and drying. Finally, they were analyzed using FT-IR spectroscopy.

Computational details

Density functional theory (DFT) calculations were carried out using the Vienna ab-initio Simulation Package software^39,40. The generalized gradient approximation (GGA) with the Perdew–Burke–Ernzerhof (PBE) exchange-correlation function was used, and projector augmented wave (PAW) pseudo-potentials were used to model ion-electron interactions^41,42,43. A plane-wave cutoff energy basis set with a cutoff energy of 450 eV was applied. To account for van der Waals forces, the DFT-D3 method was employed^44,45. Structural relaxations utilized a Monkhoest–Pack k-point grid of 2 × 2 × 1, with convergence thresholds set to an energy tolerance of 10⁻⁵ eV and force below 0.05 eV Å⁻¹. A 15 Å vacuum region was introduced along the z-axis. The free energy change (ΔG) of the reaction pathway was determined using Eq. 21 as follows:

$$\Delta G=\Delta E+\Delta {{\rm {ZPE}}}-T\Delta S$$

(21)

where ΔE, ∆ZPE, and ∆S are the variations in the reaction energy, zero-point energy change, and entropy change, respectively, T stands for the temperature.

Rechargeable Zn-NO₃ ⁻ battery measurements

For the rechargeable Zn-NO₃⁻ battery tests, an H-cell setup with a Nafion 211 membrane as the separator was used. The anode compartment contained 2 M KOH as the electrolyte, while the cathode compartment utilized a solution of 0.5 M Na₂SO₄, 0.1 M NaOH, and 200 ppm NO₃⁻-N. A polished Zn plate with a geometric area of 0.5 cm² served as the anode, and the cathode employed NF anchored with hcp-Ru₁Co.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.