Nitrate (NO₃^-), as a common pollutant, is widely distributed in surface and ground water due to the improper use of N-containing fertilizers and the extensive discharge of industrial wastewater. The in vivo conversion of NO₃^- to NO₂^- can cause serious health problems such as blue baby syndrome, and even cancer [1,2]. Traditional biological, physical and chemical technologies such as biological denitrification, reverse osmosis, ion exchange and chemical reduction have been used for the treatment of nitrate for decades [3], but the harsh reaction conditions, complex operation process and high treatment cost have hindered their widespread application [4].

On the other hand, ammonia (NH₃) is a multifunctional compound, which can be used as an indispensable chemical feedstock for common chemical industry products [5–7]. Its high energy density (4.32 kWh/L) makes it a promising candidate for the next generation of carbon-free energy carriers [8]. The production of ammonia by traditional Haber-Bosch technology requires high temperature (400-600 ℃) and high pressure (20-40 MPa), resulting in high energy consumption and costs [9]. Meanwhile, H₂, as one of the raw materials of the Haber-Bosch process, is mainly derived from fossil fuels. As the inevitable emission of a large amount of carbon dioxide in the process of hydrogen preparation from natural gas [10], new ammonia synthesis methods are being actively explored to replace the traditional Haber-Bosch process. In recent years, electrocatalytic reduction of N₂ to ammonia has received great attention. However, the NH₃ Faraday efficiency and selectivity are severely limited by the low solubility and strong triple bond (941 kJ/mol) of N₂ [11,12]. In contrast, the electrocatalytic reduction of NO₃^- to ammonia appears to be an extremely fascinating method for ammonia synthesis, benefiting from the high solubility and low N=O bond energy of nitrate (204 kJ/mol) [13]. Electrochemical nitrate reduction reaction (NITRR) plays the role of “two birds-one stone”, which can realize the transformation of nitrate pollutants into value-added ammonia chemicals. It is well-known that the electrocatalytic reduction of NO₃^- to ammonia is a complex eight-electron and nine-proton transfer process, which inevitably generates NO₂^-, N₂ and other byproducts, resulting in low selectivity of ammonia. Furthermore, achieving the ideal yield rate and Faraday efficiency towards desired NH₃ has proved to be kinetically challenging, since hydrogen evolution reaction (HER) typically behaves as a predominant competitive reaction at more negative potentials. Therefore, electrocatalytic NITRR urgently requires high-performance electrocatalysts to improve the selectivity of ammonia and inhibit the evolution of hydrogen. In addition, compared with HER, OER and other electrocatalytic fields, there is still an urgent need to develop new and efficient electrocatalysts for nitrate reduction and further explore the mechanism of ammonia production.

In recent years, a series of strategies have been adopted to develop NITRR electrocatalysts with high activity and selectivity, among which vacancy engineering has proved to be an effective strategy. For example, oxygen vacancies (OVs) in TiO₂ nanotubes could capture nitrate and weaken the N-O bonding, thus improving NITRR activity [14]. Recently, it was reported that OVs and in-plane heterojunctions in Cu/CuO_x could synergistically regulate the electronic structure of active sites, thereby promoting the adsorption and further activation of nitrate [15]. OVs-rich oxides have been widely reported in the field of NITRR, but nitrides with abundant nitrogen vacancies (NVs) need to be further explored. Zhang et al. found that NVs could induce novel electronic states at the Fermi level, which is beneficial to nitrate adsorption and activation [16]. Therefore, it can be considered that the NVs in nitrides can effectively catalyze nitrate reduction. Up to now, the methods of creating vacancies generally include reduction, corrosion, heat treatment and plasma treatment [17]. It is critical to discover a strategy that can induce NVs more quickly and conveniently without additional processing.

In this work, Co₄N grown on Ni foam (NF), named as Co₄N/NF, was successfully synthesized and first used for nitrate reduction reaction. For comparison, two other catalytic materials, denoted as CoP/NF and Co₉S₈/NF, were fabricated using the same precursors. The Faradaic efficiency, NH₃ yield rate, and NH₃ selectivity of Co₄N/NF (95.4%, 0.2475 mmol h⁻¹ cm⁻², and 99.4%) and CoP/NF (92.2%, 0.1823 mmol h⁻¹ cm⁻², 95.2%) are obviously higher than those of Co₉S₈/NF (49.4%, 0.0157 mmol h⁻¹ cm⁻², 62.5%) at −0.44 V versus the reversible hydrogen electrode (RHE). A series of experimental results unveil the reason why Co₄N/NF and CoP/NF perform better than Co₉S₈/NF, and confirm that the activity of NITRR is enhanced by the nonmetallic vacancies caused by the leaching of nonmetallic species under the condition of in-situ electrochemical reduction. Density functional theory (DFT) calculations suggest that the excellent performance of Co₄N/NF is closely correlated with NVs. It can be concluded that Co₄N with abundant NVs can reduce the NITRR reaction energy and H₂ generation by optimizing the adsorption energy of reactants/intermediates and enhancing the interaction between *H and the surface of Co₄N, thus achieving higher NITRR activity.

Co₄N/NF was synthesized by a two-step method. Typically, a Co(OH)F precursor grown on NF was synthesized by a hydrothermal method. Then, the as-synthesized precursor was annealed with urea at 400 ℃ for 3 h in Ar atmosphere (Fig. S1 in Supporting information). The crystal structures of the precursor and the target catalyst were investigated by X-ray diffraction (XRD). Fig. S2a (Supporting information) and Fig. 1a present the XRD patterns of Co(OH)F and Co₄N samples, respectively. The characteristic diffraction peaks previously indexed to Co(OH)F disappear and new peaks at 44.2°, 51.5° and 75.8° emerge after calcination. Such new diffraction peaks are perfectly matched with the (111), (200) and (220) crystal planes of Co₄N. The microstructure was characterized by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Fig. 1b shows the SEM image of the synthesized Co(OH)F/NF precursor. It can be seen that the whole surface of NF is densely and uniformly coated with flower-like Co(OH)F. For Co₄N/NF, the flower-like structure survives after nitridation but with rough surface (Fig. 1c). The TEM image of Co₄N (Fig. 1d) displays the morphology of the edge of a single micron-sized flower-like structure. The high-resolution TEM (HRTEM) image reveals the lattice fringes with the interplanar spacing of 0.204 nm, which can be indexed to the (111) plane of Co₄N (Fig. 1e). The corresponding selected area electron diffraction (SAED) pattern is also identified as the (111) plane of Co₄N (Fig. 1f). The high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and energy-dispersive X-ray spectroscopy (EDX) elemental mapping images are shown in Fig. 1g, indicating that Co and N elements present a uniform distribution in Co₄N. X-ray photoelectron spectroscopy (XPS) was investigated to further explore the chemical bonds and electronic states of Co₄N/NF and the results are displayed in Fig. S2b (Supporting information) and Figs. 1h and i. Fig. 1h shows the XPS Co 2p spectrum of Co₄N/NF, it can be deconvoluted into multiple types of peaks. The peaks of about 780.3 and 796.1 eV are ascribed to the Co-Co bonds, while two peaks centered at 781.6 and 797.3 eV should be attributed to the Co-N bonds [18]. The results demonstrate that there are two kinds of coordination environments of Co species in Co₄N. As shown in Fig. 1i, the high-resolution N 1s spectrum at 398.8 eV can be compatible with the Co-N bonds [19]. Therefore, the above results clearly confirm the successful synthesis of Co₄N material.

To investigate the NITRR performance of the Co₄N/NF electrocatalyst, an H-type electrolytic cell containing 0.5 mol/L K₂SO₄ and 200 ppm NO₃^--N was used. Co₄N/NF, platinum gauze, and saturated Ag/AgCl electrode were used as the working, counter, and reference electrodes, respectively. For convenience of comparison, all potentials reported in this work were converted to the RHE according to the following formula, E(vs. RHE) = E(vs. Ag/AgCl) + 0.059 × pH + 0.197. The electrochemical nitrate reduction was tested at various potentials ranging from −0.29 V to −0.59 V vs. RHE. The reactant and products including NO₃^-, NH₃ and NO₂^- were detected by colorimetric methods, and the corresponding ultraviolet-visible (UV-vis) adsorption spectra with concentration-absorbance calibration curves are plotted in Figs. S3-S5 (Supporting information). Fig. 2a shows the linear sweep voltammetry (LSV) curves of different electrodes (NF, Co(OH)F/NF, Co₄N/NF) with (solid line) and without (dash line) NO₃^- at a scan rate of 5 mV/s. The blank NF performs relatively low current density both in the presence and absence of NO₃^-, proving that it has almost no NITRR activity and HER activity at low overpotential. The current density of Co(OH)F/NF shows no significant increase in 0.5 mol/L K₂SO₄ electrolyte with 200 ppm NO₃^--N, indicating its poor NITRR activity. Compared with the substrate and the precursor, the significantly increased current density of Co₄N/NF in the electrolyte with nitrate compared to without nitrate at the same overpotential, implying the superior catalytic activity of Co₄N/NF toward NITRR. The reduction peak in the figure may be the reduction peak of Co unrelated to nitrate reduction, as it appears on the LSV curve in blank K₂SO₄ electrolyte. This peak disappears in the presence of NO₃^-, possibly due to the large current density response of Co₄N to nitrate reduction, covering the peak. To further investigate the catalytic performance of Co₄N/NF, chronoamperometry (CA) tests were performed at various potentials (−0.29 ~ −0.59 V vs. RHE). As shown in Fig. 2b and Fig. S6 (Supporting information), with the decrease of cathodic potential, the Faradaic efficiency and selectivity of ammonia display a volcanic shape curve while the NH₃ yield rate gradually increases and tends to be stable. The change trend of NO₃^- conversion rate is similar to that of NH₃ yield rate. The Faradaic efficiency and selectivity of ammonia reached their respective maximum at −0.44 V vs. RHE, so it was determined as the optimal potential. At −0.44 V vs. RHE, the Faradaic efficiency, NH₃ yield rate, conversion rate of NO₃^-, and NH₃ selectivity of Co₄N/NF are 95.4%, 0.2475 mmol h⁻¹ cm⁻², 89.5%, and 99.4%, respectively. The yield rate of NH₃ is much higher than that of Co(OH)F/NF (0.0169 mmol h⁻¹ cm⁻²) and NF (0.0041 mmol h⁻¹ cm⁻²) (Fig. S7 in Supporting information), demonstrating that Co₄N/NF catalyzes the conversion of NO₃^- to NH₃ effectively. The Faraday efficiency and selectivity of Co₄N/NF are superior to or comparable to most other reported NITRR catalysts or vacancy engineering related catalysts (Tables S1 and S2 in Supporting information). The time-dependent concentration curves of NO₃^-, NH₃ and NO₂^- were recorded at −0.44 V vs. RHE. With the extension of reduction time, the concentration of NO₃^- gradually decreases, and the concentration of NH₃ displays a completely opposite trend, indicating that ammonia is produced by the reduction of nitrate (Fig. 2c). Meanwhile, the concentration of NO₂^- is extremely low throughout the reaction process, suggesting that the production of NO₂^- as a byproduct is suppressed, which improves the selectivity of NH₃. As illustrated in Figs. S8 and S9 (Supporting information) and Fig. 2d, we also performed the consecutive recycling tests at −0.44 V vs. RHE to evaluate the durability of the prepared catalyst. Fig. S8 shows the chronoamperometric curves, the current density recovers immediately when the electrolyte is refreshed. After seven consecutive cycles, the Faradaic efficiency, NH₃ yield rate, conversion rate of NO₃^-, and NH₃ selectivity can remain stable, proving the outstanding stability of Co₄N/NF under neutral condition (Fig. 2d and Fig. S9). We also investigated the effect of some ions (CO₃^2-, Na⁺, Cl^-, ClO₄^-, HCO₃^-, Li⁺) commonly found in actual nitrate wastewater on the NITRR activity of Co₄N/NF. As shown in Figs. S10 and S11 (Supporting information), our catalyst can still maintain high NITRR activity when these ions coexist. In order to eliminate external interference, the ammonia yield of Co₄N/NF was determined after electroreduction in a nitrate-free blank electrolyte. As shown in Fig. 2e, no ammonia can be detected in the absence of NO₃^-, indicating that the detected ammonia originates from the added nitrate. We also conducted isotope labeling experiments to strengthen the result described above [20–22]. As shown in Fig. 2f, when the reactant is K¹⁴NO₃, the ¹H NMR spectra show representative triple peaks of ¹⁴NH₄⁺, while typical double peaks of ¹⁵NH₄⁺ can be detected if the reactant is replaced by K¹⁵NO₃ [23,24]. No significant ¹⁴NH₄⁺ can be observed when K¹⁵NO₃ is used, confirming that the produced ammonia is completely evolved from the electrochemical reduction of NO₃^- rather than external interference.

A series of comparative experiments before and after NITRR tests were conducted to reveal the origin of the superior catalytic performance of Co₄N/NF. The electrochemical active surface areas (ECSA) of Co₄N and Co(OH)F can be obtained by the double-layer capacitances (Figs. S12a-c in Supporting information). It is found that the ECSA value of Co₄N is 395 cm⁻², about 20 times that of Co(OH)F (20 cm⁻²), which facilitates the reduction of nitrate. After NITRR tests, the morphology remains roughly unchanged. Interestingly, the surfaces of Co₄N/NF become rougher and completely covered by nanosheets after NO₃^- reduction reaction, thus exposing more active sites (Fig. 3a and Fig. S13 in Supporting information). The electronic structure of electrocatalysts affects their intrinsic activity. XPS analysis was performed to probe the changes of Co₄N before and after NITRR tests. As presented in Fig. 3b, the peaks in the low and high binding energy zones can be attributed to Co 2p_3/2 and Co 2p_1/2, respectively. After NITRR tests, the binding energy of Co 2p appears significantly negative shifts. Simultaneously, the high-resolution Co 2p and N 1s spectra suggest a significant decrease of the N/Co atomic ratio from 0.381 to 0.129 after NITRR, indicating the decreased coordination number of Co centers due to the leaching of nitrogen species (Fig. 3c). As shown in Fig. S14 (Supporting information), no other diffraction peaks are detected except the characteristic peaks of Co₄N, demonstrating that the major phase is still Co₄N during NITRR process. Thus, although about two-thirds of the nitrogen is leached from the Co₄N, the crystalline framework is not damaged and no phase change occurs. Nitrogen leaching results in a large number of NVs, as confirmed by Electron paramagnetic resonance (EPR) tests. According to Fig. 3d, the EPR signal at ~2.0022 is detected in Co₄N after NITRR, proving that NVs are introduced by in-situ electrochemical reduction, that is, nitrogen dissolution induced abundant nitrogen vacancies [25]. Ex situ X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine-structure (EXAFS) spectra were carried out to further determine the fine structure of Co₄N and elucidate the catalytic mechanism. As exhibited in Fig. 3e, the XANES spectra show that the absorption threshold position of Co₄N after NITRR tests moves towards lower energy compared to that of Co₄N before NITRR tests, implying a slight decrease in the oxidation state of Co. The EXAFS spectra of the Co₄N catalyst before and after NITRR tests are shown in Fig. 3f. For the EXAFS spectrum of pre-test Co₄N, the two peaks at approximately 1.5 and 2.2 Å can be attributed to Co-N and Co-Co bonds, respectively [26]. It can be seen that the intensity of the Co-N shell decreases after NITRR tests, indicating that the coordination of Co centers is not saturated and the formation of NVs due to the dissolution of nitrogen species [11]. The results of XANES and EXAFS jointly clarify that the oxidation state and coordination number of Co centers decrease because of the leaching of nitrogen, which is in good agreement with the XPS conclusion. Fig. S15 (Supporting information) shows the Faraday efficiency-cycle number curve of Co₄N/NF at −0.44 V vs. RHE by 9 cycles. It is found that with the increase of reaction time, Faraday efficiency gradually increases and eventually tends to be stable, indicating a positive correlation between Faraday efficiency and nitrogen leaching within a certain time range at the beginning. It also indicates from the side that the ultra-high selectivity of Co₄N/NF electrocatalyst is attributable to the nitrogen vacancy generated by in-situ nitrogen leaching. The above results enable us to infer that the superior NITRR activity can be attributed to the following two aspects. First, the original large-size morphology is transformed into small-size nanosheets during the NITRR process, and more active sites are exposed. Second, the introduction of NVs owing to nitrogen leaching, which may be the main reason for the superior catalytic performance.

CoP/NF and Co₉S₈/NF were synthesized to further verify the hypothesis that potential-driven leaching of nonmetals into the electrolyte enhances nitrate reduction activity under in-situ electrochemical reduction conditions. The SEM images of CoP/NF and Co₉S₈/NF are presented in Fig. S16 (Supporting information). Fig. S17 (Supporting information) shows the XRD patterns of CoP and Co₉S₈, displaying no impurity peaks except the peaks of CoP and Co₉S₈ respectively, indicating the successful synthesis of the target catalysts. Fig. S18 (Supporting information) presents the XPS spectra of CoP/NF. Specifically, the Co 2p peaks at the binding energies of 778.9 and 793.9 eV should be assigned to the Co-P bonds, the peaks at 782.3 and 798.4 eV are ascribed to the Co-O bonds, which are derived from surface oxidation [27]. Meanwhile, the high-resolution P 2p spectrum displays three obvious P 2p_3/2, P 2p_1/2 and P-O signals, which is consistent with the results of Co 2p spectrum. As depicted in Fig. S19 (Supporting information), the high-resolution Co 2p and S 2p spectra of Co₉S₈/NF are fitted. The Co 2p spectrum can be fitted as Co³⁺ and Co²⁺, the peaks at 778.4 and 793.3 eV are assigned to Co³⁺, while the peaks at 781.1 and 797.1 eV belong to Co²⁺ [28]. The S 2p spectrum is decomposed into three types of peaks, which are metal-sulfur bonds (S 2p_3/2 161.5 eV, S 2p_1/2 162.5 eV), surface sulphur ions (163.7 eV), and surface oxidized sulfate species (168.6 eV) [29]. The NITRR properties of CoP/NF and Co₉S₈/NF were also evaluated under the same conditions as the Co₄N/NF electrocatalyst described above. As revealed by the LSV curves in Fig. S20 (Supporting information), both CoP/NF and Co₉S₈/NF exhibit increased current density in the presence of NO₃^-. However, the current density of CoP/NF is significantly higher than that of Co₉S₈/NF at the same cathodic overpotential. The Faraday efficiencies, NH₃ yield rate, conversion rate of NO₃^-, and NH₃ selectivity of CoP/NF and Co₉S₈/NF at different potentials are recorded in Fig. S21 (Supporting information). The Faradaic efficiency presents a volcanic shape curve while the NH₃ yield rate, conversion rate of NO₃^- and NH₃ selectivity gradually increase. It can be clearly observed that the Faradaic efficiency (92.2%), NH₃ yield rate (0.1823 mmol h⁻¹ cm⁻²), conversion rate of NO₃^- (71.2%), and NH₃ selectivity (95.2%) of CoP/NF are apparently higher than those of Co₉S₈/NF (49.4%, 0.0157 mmol h⁻¹ cm⁻², 9.1%, and 62.5%) at −0.44 V vs. RHE. The superior nitrate reduction activity may be due to the changes of CoP/NF revealed by XPS before and after NITRR tests similar to those of Co₄N/NF. As shown in Fig. S22 (Supporting information), the binding energy of Co 2p appears obviously negative shifts while the P/Co atomic ratio decreases from 3.381 to 0.324 after NITRR. Interestingly, with regard to Co₉S₈/NF (Fig. S23 in Supporting information), the Co 2p and S 2p XPS spectra show no obvious change after NITRR tests (S/Co atomic ratio increases from 1.603 to 1.676). XAS further reflected the changes of oxidation state and coordination situation of Co and the dissolution status of P and S species in CoP/NF and Co₉S₈/NF. Fig. S24a (Supporting information) presents the Co K-edge XANES spectra of CoP and post-test CoP, and it is observed that the absorption threshold position of CoP shifts negatively after NITRR tests, suggesting a slightly reduced oxidation state of Co in CoP. The EXAFS spectra of CoP before and after NITRR tests are shown in Fig. S24b (Supporting information), the prominent peak at about 1.8 Å can be ascribed to the Co-P peak [30,31]. After NITRR tests, the intensity of the Co-P peak decreases, indicating that the coordination number of Co decreases because of the leaching of P atoms, which is a strong proof of the formation of P vacancies. However, we can clearly find that the cobalt K-edge XANES spectra of Co₉S₈ before and after NITRR tests almost completely overlap, which manifests the oxidation state of Co₉S₈ remains unchanged during the catalytic process (Fig. S25a in Supporting information). In addition, the main peak located at 1.85 Å is well matched with Co-S [32], and it is worth mentioning that the coordination environment of Co remains stable and unchanged, which matches well with the results of XPS (Fig. S25b in Supporting information). The above discussion strongly proves that the excellent NITRR activities of Co₄N/NF and CoP/NF originate from the nonmetallic vacancies induced by the leaching of N and P atoms, respectively. On the contrary, the poor activity of Co₉S₈/NF is due to the difficulty in breaking the Co-S bond during the electrochemical reduction, resulting in almost no leaching of sulfur species and difficult formation of S vacancies. In conclusion, nonmetallic vacancies induced by in-situ nonmetal leaching can effectively improve the performance of NITRR.

To further illustrate the role of nonmetallic vacancies, we artificially introduced sulfur vacancies for Co₉S₈/NF to generate V_S-Co₉S₈/NF by plasma engraving. As illustrated in Fig. S26 (Supporting information), a stronger EPR signal at g = 2.0034 in V_S-Co₉S₈ is detected after plasma engraving, indicating that more sulfur vacancies are introduced into V_S-Co₉S₈/NF compared with Co₉S₈/NF [33,34]. As shown in Fig. S27a (Supporting information), the LSV curve of V_S-Co₉S₈/NF shows an obvious increase of current density compared with Co₉S₈/NF in the presence of NO₃^-. With the negative shift of the cathode overpotential, the Faraday efficiency and selectivity gradually increase until the potential reaches −0.44 V vs. RHE, and then drop slowly, while the change trend of NH₃ yield rate is the same as that of NO₃^- conversion rate (Figs. S27b and c in Supporting information). The Faraday efficiency and selectivity reach their respective maximum values (83.5%, 83.2%) at −0.44 V vs. RHE. V_S-Co₉S₈/NF shows excellent electrochemical NITRR performance, much higher than Co₉S₈/NF (Fig. S27d in Supporting information). The above results confirm that the sulfur vacancies in V_S-Co₉S₈/NF contribute to the improvement of nitrate reduction activity, which further explains the role of nonmetallic vacancies in NITRR.

DFT calculations were conducted to rationalize the nonmetallic vacancies enhanced mechanism caused by nonmetal leaching under in-situ electrochemical reduction. The (111) surface of Co₄N was selected for DFT calculations based on investigations of XRD, HRTEM and SEAD. The combined results of these three characterizations indicate that the (111) surface appears as the main exposed surface. Co₄N without NVs (marked as WNV) and with single nitrogen vacancy (SNV) were considered as models since NVs were detected after NITRR tests. Six possible NITRR pathways on Co₄N were considered based on previous reports (Fig. S28 in Supporting information) [35–37]. The optimized models of different intermediates involved in the six reaction pathways are presented in Figs. S29 and S30 (Supporting information). The Gibbs free energies of all the intermediates were calculated to determine the optimal reaction pathway. On WNV, pathway 6 (Fig. S28) is thermodynamically more favorable than the other pathways owing to a lower ΔG from *NO to *NHO (0.83 eV) compared to *NO to *NOH (1.55 eV), simultaneously, *NHO hydrogenation to form *NH₂O is exothermic (Fig. S31a in Supporting information). On SNV, pathway 3 (Fig. S28) is preferred due to the exothermic properties of *NOH reduction to *N (Fig. S31b in Supporting information). The respective optimal pathways are plotted in Fig. 4a for better comparison. Nitrate reduction to ammonia is essentially a series of deoxidation and hydrogenation steps. On SNV, the ΔG of NO₃^- adsorption process is 0.03 eV, which is significantly smaller than that on WNV (0.53 eV), indicating that the presence of nitrogen vacancy contributes to the activation and further reduction of NO₃^-. For NITRR, the hydrogenation step on *NO intermediate plays a crucial role, which can be divided into N-side and O-side. On WNV, the free energy change of the key hydrogenation step (*NO to *NHO) is 0.83 eV, much higher than that of *NO to *NOH (0.32 eV) on SNV, suggesting that *NO can be reduced on SNV more easily. The desorption of NH₃ can be regarded as another important NITRR step. Note that the desorption energy on SNV (0.35 eV) is much lower than that on WNV (0.68 eV), indicating that the desorption process of NH₃ is more likely to occur on the surface of Co₄N with abundant NVs. This is further illustrated by the charge density difference maps of NH₃ adsorbed on the surface of WNV and SNV. As shown in Fig. 4b, an increased electron density between NH₃ and SNV is observed, indicating that the interaction between NH₃ and the surface model is weakened, which is conducive to NH₃ escaping from the catalyst surface [38]. Besides, the potential-determining step (PDS) of WNV is *NO to *NHO (0.83 eV), and the free energy change is also higher than that on SNV (*NH₃ to NH₃, 0.35 eV). The high activity of Co₄N catalyst can be explained by the reduced ΔG of the PDS in principle. On SNV, the free energy changes for the release of possible byproducts NO₂, HNO₂ and NO are 1.63, 0.54 and 2.58, respectively, demonstrating that these byproducts are quite difficult to form, thus achieving high NITRR selectivity. In addition, the energy of the HER was also investigated to further interpret the high performance of the Co₄N catalyst (Fig. 4c). The energy required for hydrogen evolution on SNV is 0.84 eV, whereas the adsorption energy of hydrogen atom on the surface of WNV is 0.61 eV. The stronger adsorption of H on the surface of SNV reduces the HER activity, thus inhibiting the production of byproduct H₂. Based on the previously reported literature [39], we infer that *H can facilitate the hydrogenation of *NO to *NOH, thus contributing to the selective conversion of NO₃^- to NH₃, which also explains the difference in potential determination steps between SNV and WNV. The combined results of experiments and theoretical calculations verify that the leaching of nitrogen contributes to the catalytic performance of Co₄N, thus the ultra-high faradaic efficiency of 95.4% and selectivity of 99.4% are obtained.

In conclusion, we developed a high-performance Co₄N/NF electrocatalyst using a facile two-step progress. The Co₄N/NF electrocatalyst displays outstanding activity and selectivity for electrocatalytic reduction of nitrate to ammonia under benign environmental conditions. Such catalyst is demonstrated to exhibit the best catalytic performance at −0.44 V vs. RHE with faradaic efficiency and NH₃ selectivity of 95.4% and 99.4%, respectively. A series of comparative experiments manifest that the N atoms are leached from Co₄N and then NVs are introduced, which are responsible for the distinguished performance of ammonia production. Subsequently, DFT calculations were performed to further verify the vacancies enhanced mechanism proposed above. It is found that the NVs on Co₄N can adjust the adsorption intensity of reactant/intermediates to optimize the reaction pathway. Moreover, NVs can modulate the adsorption energy of hydrogen atoms, and *H is more likely to be generated on the surface of SNV, which is beneficial to the production of ammonia. This study highlights a novel in-situ nonmetal leaching strategy to design Co-based NITRR electrocatalysts with high NH₃ Faraday efficiency and selectivity, and provides in-depth insights into the vacancy-enhanced NITRR mechanism.

There are no conflicts to declare.

We acknowledge the financial supports from National Natural Science Foundation of China (Nos. 91741105, 22006120), Program for Innovation Team Building at Institutions of Higher Education in Chongqing (No. CXTDX201601011), and Chongqing Municipal Natural Science Foundation (No. cstc2018jcyjAX0625).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2023.108341.