Selective electrosynthesis of urea from nitrate and carbon dioxide with low overpotential

Abstract

Reducing the energy cost and CO₂ emission of traditional urea synthesis through alternative strategy is crucial for the sustainable social development. Herein, we report the electrosynthesis of urea from nitrate and CO₂ on Ag/CdO heterostructured catalyst with low overpotential. The optimized Ag_0.07/CdO catalyst demonstrates a high urea yield rate of 111.6 ± 3.1 mmol g_cat.^–1 h^–1 with 50.0 ± 2.7% Faradaic efficiency at −0.15 V vs. RHE in a flow cell, and can further achieve a urea yield rate of 427.0 ± 30.2 mmol g_cat.^–1 h^–1 in an optimized two-electrode cell with stable response for 1000 h. Operando spectroscopy and theoretical calculations indicate that Ag attracts *NO₂ intermediate as well as promotes the reconstruction of CdO into low coordinated CdCO₃ that facilitate the activation of CO₂, thus benefits *CO/*NO₂ adsorption and coupling at low overpotential as well as suppresses the generation of CO and NH₃ by-products.

Introduction

Over the past century, nitrogen fertilizer has fed approximately 27% of the world’s population by nourishing crops^1,2. Urea is the most common nitrogen fertilizer and its nitrogen content is as high as ~46%, which plays a pivotal role in the agricultural field and population growth³. In addition, urea is widely used in the preparation of dyes, resins and pharmaceutical products⁴. The traditional industrial synthesis of urea (Bosch–Meiser process) requires high energy consumption arising from high temperature (~200 °C), high pressure (~210 bar) and a complex synthesis process involving artificial nitrogen fixation to form NH₃^5,6. It is worth noting that the synthetic ammonia industry consuming more than 2% of global energy and more than 50% of synthesized NH₃ is used for urea synthesis in 2020 alone⁷. Therefore, it is imperative to explore sustainable and economical routes for urea synthesis.

In recent decades, electrochemical synthesis of urea has drawn increasing attention since the utilization of clean renewable energy and environmental concern⁸. Various nitrogen sources (e.g., N₂, NO₃⁻ and NO₂⁻) can be coupled with the electrochemical reduction of greenhouse gas CO₂ to produce urea^9,10,11. Among them, nitrate has a relatively low N=O dissociation energy (204 kJ mol^–1)¹², making it more suitable as an N source for urea electrosynthesis. Moreover, nitrate can be obtained from N₂/air using a non-thermal plasma process¹³. The total energy consumption of direct electrosynthesis of urea via N₂-NO₃⁻-urea process can be comparable or even less than the traditional process (Supplementary Note 1), while the CO₂ emissions involved in urea electrosynthesis may be reduced by more than 50% compared with the traditional urea synthesis method⁷, indicating that the electrochemical co-reduction of NO₃⁻ and CO₂ for urea synthesis with renewable energy holds potential as a new alternative pathway. Recently, some metal oxides/hydroxides catalysts (In(OH)₃³, Vo-CeO₂¹¹, CuSiO_X¹⁴, TiO₂/C¹⁵, SrCo_0.39Ru_0.61O_3-δ¹⁶, CuWO₄¹⁷, etc.) and metal catalysts (Zn/Cu⁸, Pd₄Cu₁-FeNi(OH)₂¹⁸, etc.) have been investigated for the electroreduction of CO₂ and NO₃⁻ to produce urea. Despite these achievements, there are still several challenges that need to be addressed due to the complex 16 electron transfer (CO₂ + 2NO₃⁻+ 18H⁺ + 16e⁻ → NH₂CONH₂ + 7H₂O) with multi intermediates involved in this process: (1) The overpotential of urea electrosynthesis is relatively high. Although the thermodynamic potential for the conversion of urea from CO₂ and NO₃⁻ is 0.811 V versus standard hydrogen electrode (0.41 V vs. RHE at pH~6.8)¹⁹, the applied potentials in most reported works are generally more negative than −0.3 V vs. RHE^2,10. (2) The catalysts usually suffer from unsatisfactory yield rate (mostly <60 mmol g_cat.^–1 h^–1) and poor stability (typically between 10 h and 100 h). (3) The high competitiveness of side reactions such as the generation of CO/NH₃ and hydrogen evolution reaction (HER) leads to a low urea selectivity. Therefore, the development of efficient and selective electrocatalysts to generate urea from NO₃⁻ and CO₂ at low overpotentials is crucial. Some previous works suggest that the rate determining step in urea electrosynthesis is the initial C-N coupling^11,14, and promoting the coupling of primary intermediates such as *CO, *NO₂ during CO₂ and NO₃⁻ co-reduction may effectively reduce the overpotential and inhibit the side reactions. Thus constructing heterogeneous catalysts with well-designed tandem active sites that simultaneously facilitates the adsorption and coupling of *CO and *NO₂ would be an efficient approach to achieve the selective production of urea. Among the metal catalysts, Ag-based materials exhibit quite high electrocatalytic activity for the conversion of NO₃⁻²⁰ in addition to CO₂ reduction response. While Cd-based nanomaterials can also catalyze the reduction of CO₂ to produce C-containing adsorption intermediates and generate CO^21,22, which may be beneficial for the C-N coupling in urea electrosynthesis. However, the tandem electrocatalysis of Ag-Cd heterostructured materials for urea synthesis from CO₂ and NO₃⁻ is rarely been explored.

In this study, we achieved efficient and selective urea electrosynthesis through early *CO and *NO₂ coupling at low overpotentials on Ag/CdO heterostructured catalyst. Operando X-ray absorption spectroscopy (XAS), Raman spectroscopy and diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) prove that the Ag component in the synthesized Ag/CdO catalyst not only promotes the reconstruction of CdO into CdCO₃ with oxygen vacancies, which can efficiently adsorb and activate CO₂, but also facilitates the adsorption of electrophilic intermediate *NO₂ during the reduction process. Density functional theory (DFT) calculations indicate that the in situ reconstructed Ag/CdCO₃ catalyst benefits the adsorption and conversion of N-containing and C-containing intermediates with low free energy differences. The C-N coupling on the reconstructed Ag/CdCO₃ sites is possibly spontaneous to form the key intermediate *CONO₂ (−0.89 eV), thus would inhibit the generation of by-products such as NH₃ and CO compared with the step of *NO₂ hydrogenation to *HNO₂ (−0.54 eV) and *CO desorption to CO (0.49 eV). Therefore, the optimized Ag_0.07/CdO catalyst demonstrates a high urea Faradaic efficiency (FE_urea) of 51.2 ± 3.0% even at −0.1 V vs. RHE in an H-cell and a high urea yield rate of 111.6 ± 3.1 mmol g_cat.^–1 h^–1 with a partial current density ~7.2 mA cm^–2 at a positive potential of −0.15 V vs. RHE in a flow cell. Moreover, it can achieve a urea yield rate of 427.0 ± 30.2 mmol g_cat.^–1 h^–1 at a current density of ~60 mA cm^–2 in an optimized two-electrode cell with stable response for 1000 h, which is quite competitive for the reported urea electrosynthesis system^{3,8,9,23,24,25,26}.

Results

Synthesis and structural characterization of electrocatalysts

The Ag/CdO heterostructured catalysts were prepared via a high-temperature calcination (550 °C, 2 h) after chemical precipitation of Ag and Cd salts. According to the Ag content tested by X-ray fluorescence (XRF) in Supplementary Table S1, the prepared sample is defined as Ag_x/CdO (Ag_0.03/CdO, Ag_0.07/CdO and Ag_0.16/CdO, x represents the atomic ratio of Ag in metal elements). Pure Ag and CdO are also prepared for comparison (Supplementary Figs. S1 and S2). The X-ray diffraction (XRD) patterns of the synthesized Ag_x/CdO samples could be assigned to the heterostructure of metallic Ag (cubic phase) and CdO (cubic phase) (Supplementary Fig. S3). The morphology of the synthesized samples manifests as nanoparticles with a size of ~100 nm with clear lattice spacing belonging to Ag and CdO heterostructures according to scanning electron microscopy (SEM, Supplementary Fig. S4) and high-resolution transmission electron microscopy (HR-TEM, Supplementary Fig. S5) test. However, particles tend to agglomerate as the Ag content increases to 14.1 atom%. Additionally, Cd, Ag and O are homogeneously dispersed throughout the nanoparticles in energy dispersive X-ray spectroscopy (EDX) image (Fig. 1a). Aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC-HAADF-STEM) is conducted to further identify the heterostructures in Ag_0.07/CdO catalyst (Fig. 1b–d). The array spacings are 0.270 and 0.145 nm, corresponding to the interplanar spacing of CdO (111) and Ag (220) facets, respectively. Furthermore, Ag, Cd, and O are indicated in the survey spectrum of X-ray photoelectron spectroscopy (XPS) for Ag_0.07/CdO (Supplementary Fig. S6a). The peaks located at 404.5 (Cd²⁺ 3d_5/2) and 411.3 eV (Cd²⁺ 3d_3/2) belonging to Cd 3d (Supplementary Fig. S6b) are negative shift compared to pure CdO (405.0 and 411.8 eV), while the spectrum of Ag 3d (Supplementary Fig. S6c) shows two peaks at 367.7 (Ag 3d_5/2) and 373.7 eV (Ag 3d_3/2) that are lower than metallic Ag (368.0 and 374.0 eV), indicating electron transfer from Ag to Cd^27,28. The coordination environment of the metal sites in Ag_0.07/CdO electrocatalyst is further validated by XAS. Due to the high proportion of CdO in Ag_0.07/CdO, the valence and Cd-Cd/Cd-O bond configuration of Cd in bulk Ag_0.07/CdO are similar to CdO^29,30, which are illustrated by the X-ray absorption near edge structure (XANES, Fig. 1e) data at the Cd K-edge and the Fourier transform extended X-ray absorption fine structure (FT-EXAFS, Fig. 1f). Meanwhile, wavelet transform (WT) contour plots of Ag_0.07/CdO show peaks corresponding to Cd-O and Cd-Cd coordination, which is highly consistent with CdO (Fig. 1g and Supplementary Fig. S7). However, although WT of Ag K-edge coordination environment is metallic Ag with a separate Ag-Ag coordination (Supplementary Fig. S8 and Fig. 1j), XANES and FT-EXAFS (Fig. 1h, i) indicate that the average valence state of Ag in Ag_0.07/CdO is slightly higher than metallic Ag, which is consistent with XPS results. The above multiple results indicate that the synthesized Ag_0.07/CdO catalyst exhibits a heterostructure of Ag and CdO, with electronic interactions between the two components, which could potentially contribute to the electrocatalytic performance of the catalyst³¹.

Fig. 1: Structural and compositional characterization.

a HR-TEM image of Ag_0.07/CdO catalyst and the corresponding elemental mappings of Cd, Ag and O. b AC-HAADF-STEM image of Ag_0.07/CdO catalyst. c The enlarged view and lattice spacing within the blue box in (b). d The enlarged view and lattice spacing within the red box in (b). e Cd K-edge and h Ag K-edge XANES spectra of Ag_0.07/CdO catalyst (The inset shows an enlarged view of the shaded region). Fourier transform of k³-weighted f Cd K-edge and i Ag K-edge EXAFS spectra for the r space. Wavelet transform of the k³-weighted EXAFS data of g Cd K-edge and j Ag K-edge for Ag_0.07/CdO catalyst.

Electrochemical urea synthesis

Urea electrosynthesis test is firstly carried out on an H-type cell with a typical three-electrode configuration under ambient conditions (Supplementary Figs. S9–S28). The linear sweep voltammetry (LSV) tests were performed to evaluate the current response of the prepared electrocatalysts towards CO₂ and NO₃⁻ co-reduction in CO₂ saturated 0.1 M KHCO₃ and 0.05 M KNO₃. As shown in Supplementary Fig. S9, Ag_x/CdO heterostructured catalysts demonstrate higher current densities compared to pure Ag and CdO, indicating a positive response in co-electrolysis of CO₂ and NO₃⁻ toward C-N coupling, especially for Ag_0.07/CdO. In addition, a competitive FE_urea of 51.2 ± 3.0% is observed on Ag_0.07/CdO catalyst at just −0.1 V vs. RHE (Supplementary Figs. S14 and S25–27)^{3,9,32,33,34,35,36}, and the main by-products include H₂ (FE = 11.8 ± 3.5%), NO₂⁻ (FE = 12.2 ± 1.0%), and NH₃ (FE = 13.6 ± 4.3%). The FE_urea is also evaluated by ion chromatography (IC) after urea decomposition via urease method, which is not significantly different from the results of UV-vis and HPLC-MS (Supplementary Fig. S27). Moreover, it is observed that the FE_NO2− on Ag_0.07/CdO is lower than that on Ag_0.03/CdO and Ag_0.16/CdO, which may be due to its efficient utilization of NO₂⁻ as an N-containing intermediate for urea synthesis. Additionally, IC is used to verify FE_NO2− and FE_NH4+ for different catalysts. Similar trends are observed for these sample (Supplementary Fig. S28). Compared to Ag_0.07/CdO, pure Ag and CdO catalysts exhibit poor performance towards urea formation. Specially, pure Ag exhibits almost the highest FEs of NH₃ (31.0%) and NO₂⁻ (49.4%) compared to other catalysts, which means that Ag may tend to activate NO₃⁻ in the co-reduction process of CO₂ and NO₃⁻. To verify the origin of N element, the 0.05 M isotope ¹⁵N-labeled potassium nitrate (K¹⁵NO₃) added 0.1 M KHCO₃ is used as the electrolyte for chronoamperometry measurements (Fig. 2a and Supplementary Figs. S29–31) and the results explicitly confirm that the N source in urea comes from NO₃⁻. The electrochemical surface area (ECSA) of the prepared catalyst is evaluated and as shown in Supplementary Figs. S32 and 33, the ECSA values of the catalyst are very close, which indicates that it is the inherent activity of the catalytic site rather than simple ECSA contributes to C-N coupling. The durability of Ag_0.07/CdO catalyst toward urea electrosynthesis are also examined. As exhibited in Supplementary Fig. S34a, the Ag_0.07/CdO catalyst maintains a relatively stable current density of approximately 1.0 mA cm⁻² for 24 h at −0.1 V vs. RHE in H-cell, indicating good stability. Further, the stability and reproducibility of urea generation is validated in repeated testing. The final FE_urea remains above 50% after 10 h long-term successive runs, demonstrating good reproducibility (Supplementary Figs. S34b and S35). To further understand the structure-performance relationship of Ag_0.07/CdO catalyst in urea electrosynthesis, the same proportion of Ag and CdO are physically mixed and the electrochemical performance is tested in Supplementary Fig. S36a, b. The results show that FE_urea and partial current density (J_urea) are 34.7% and 0.24 mA cm^–2 at −0.1 V vs. RHE, respectively, which are lower than those of Ag_0.07/CdO but significantly higher than that of pure Ag (6.5% ± 1.5%) and CdO (11.6% ± 5.0%), indicating the crucial role of the bifunctional Ag/CdO structure in the electrochemical synthesis of urea. Furthermore, XPS (Supplementary Fig. S36c, d) proves that the physical mixture of Ag and CdO do not undergo significant electron transfer, indicating that the electronic interaction between the two components on Ag_0.07/CdO catalyst may further promote the generation of urea. In addition, control tests including an experiment on replacing CO₂ with argon and an experiment where CO₂ and nitrate are left in the cell without applying the electrochemical process are added and the results show that no urea is produced (Supplementary Figs. S37–39). A flow cell with three electrode system is further applied to the urea electrosynthesis from CO₂ and NO₃⁻ (Fig. 2b and Supplementary Figs. S40–42). In Fig. 2b, the results show that the FE_urea is about 50% and a high urea yield rate is 111.6 ± 3.1 mmol g_cat.^–1 h^–1 at −0.15 V vs. RHE, indicating the efficient electrochemical C-N coupling at a low overpotential in a flow cell. Furthermore, the Ag_0.07/CdO catalyst maintains long stability for 1000 h at a current density of ~10 mA cm^–2 during urea electrosynthesis (Supplementary Fig. S41) and exhibits the lowest overpotential with a high urea yield rate^{3,9,12,14,15,16,17,18,32,36,37,38,39,40,41}.

Fig. 2: Electrosynthesis of urea.

a MS spectra for K¹⁴NO₃ and K¹⁵NO₃ with CO₂ co-reduction, respectively. b FE_urea and urea yield rate at different potentials in a flow cell. c LSV at 10 mV s^–1 scan rate for Ag_0.07/CdO catalyst in a two-electrode cell, d Faradaic efficiency of each product and e Stability test of urea synthesis during 1000 h of electrolysis at −1.6 V in a two-electrode cell. Electrochemical measurements were performed without iR compensation in a flow cell (R = 6.0 ± 0.2 Ω) and a two-electrode cell (R = 0.5 ± 0.1 Ω) at 25 °C. The electrolyte is 1.0 M KOH and 0.05 M KNO₃ (pH = 14.0 ± 0.1) and the flow rate of CO₂ during test was maintained at 20 mL min^–1 by a gas flow meter.

Furthermore, the performance of Ag_0.07/CdO catalyst is significantly improved by optimizing the process conditions (e.g. ionomer, membrane and flow field configurations) in a two-electrode cell (Specific details in “Methods”). As shown in Fig. 2c, d and Supplementary Figs. S43 and 44, Ag_0.07/CdO catalyst can achieve a urea yield rate of 427.0 ± 30.2 mmol g_cat.^–1 h^–1 with a FE_urea of 59.1 ± 6.6% at a current density ~60 mA cm^–2 (−1.6 V). Figure 2e displays that the urea yield rate can maintain around 75% of the initial value and FE_urea is over 50% after 1000 h stability test, which is competitive for the electrosynthesis of urea from CO₂ and NO₃⁻ (Supplementary Table S2).

Mechanism study

To get insights into the high C-N coupling activity of Ag_0.07/CdO, structural characterizations are employed in the catalysts after CO₂ and NO₃⁻ co-reduction. The XRD patterns reveal that Ag maintains the cubic phase, while CdO evolves into CdCO₃ during the reduction process (Supplementary Figs. S45–47). Furthermore, HR-TEM displays the clear lattice spacing belonging to heterostructured Ag and CdCO₃ with the array spacings of 0.272 and 0.236 nm, corresponding to the interplanar spacing of CdCO₃ (006) and Ag (111) facets after 10 h and ~100 h of reduction, respectively (Supplementary Fig. S48), indicating the good stability of the reconstructed heterostructure. XPS results show that the binding energy of Cd for Ag_0.07/CdO is similar to CdCO₃ after reaction (Supplementary Fig. S49a), while the binding energy of Ag in Ag 3d spectrum is transferred to higher energy after co-reduction (Supplementary Fig. S49b), indicating that the Ag component acquires electrons during the reduction process. In order to investigate the effect of in situ reconstructed Ag_0.07/CdCO₃ on C-N coupling, physical mixed Ag_0.07/CdCO₃, named as Ag_0.07/CdCO₃-PM, is prepared, in which CdCO₃ is obtained by precipitation method (Supplementary Figs. S50 and 51). Its FE_urea and J_urea are measured to be 33.5% and 0.33 mA cm^–2 at the optimal applied potential (−0.1 V vs. RHE), which are lower than those of Ag_0.07/CdCO₃ reconstructed from the Ag_0.07/CdO catalyst (Supplementary Fig. S50b, c). However, its FE_urea can still reach over 30%, proving that CdCO₃ is the main catalytic component derived from CdO. XPS results (Supplementary Fig. S50d–f) show that only lattice oxygen (531.2 eV) is observed in Ag_0.07/CdCO₃-PM, while in situ reconstructed Ag_0.07/CdCO₃ exhibits dissociated molecular oxygen adsorbed on oxygen vacancies (534.3 eV)⁴². Therefore, it can be inferred that in situ reconstructed Ag_0.07/CdCO₃ with vacancies would promote C-N coupling.

To further understand the structural evolution mechanism of Ag_0.07/CdO catalyst during co-reduction, operando XAS spectroscopy are performed. As shown in Fig. 3a, the relative absorption edge positions in the Cd K-edge XANES shows a positive shift as the applied potential decreases, suggesting the gradual structural evolution of CdO towards CdCO₃⁴³. Additionally, although the k³-weighted Fourier transform of Cd K-edge EXAFS analysis shows the Cd-Cd and Cd-O bonds always exist during reaction (Fig. 3b), the average Cd-Cd and Cd-O coordination numbers show a decrease in the coordination number of the Cd-O bond from 6 to 5.6, while the coordination number of the Cd-Cd bond decreased from 12 to 6.1 (Supplementary Figs. S52–55 and Supplementary Table S3), further proving that CdO gradually evolved into CdCO₃ during the reduction process. Moreover, the coordination number (5.6) of the Cd-O bond for the reconstructed Ag_0.07/CdCO₃ catalyst is lower than the theoretical coordination number of CdCO₃ (6.0), indicating the possible presence of oxygen vacancies in the in situ reconstructed CdCO₃, which is consistent with XPS. These defects may promote CO₂ adsorption and activation on Cd sites¹¹. In addition, the relative absorption edge positions in the Ag K-edge XANES for Ag_0.07/CdO after reaction shifts slightly negatively towards the position of the Ag foil, indicating that the Ag component acquires electrons during the reduction process, which is consistent with the XPS results (Supplementary Fig. S56). While FT-EXAFS of Ag_0.07/CdO after reaction exhibits a separate Ag-Ag bond and with no significant change in coordination number before and after co-reduction (Supplementary Figs. S57–59 and Supplementary Table S4). The above characterization results indicate that Ag_0.07/CdO reconstructs into Ag_0.07/CdCO₃ with oxygen vacancies during the CO₂ and NO₃⁻ co-reduction. The CdCO₃ with oxygen vacancies would provide a suitable chemical environment for strong binding with CO₂ molecules through the combined action of Cd, C, and O atoms and further generate *CO intermediates⁴⁴. In addition, the Ag component that accumulates electrons during reduction may form local electron-rich regions that attract the adsorption of electrophilic intermediates such as *NO₂²⁰. Therefore, in situ reconstructed Ag_0.07/CdCO₃ with oxygen vacancies would provide tandem dual sites for the adsorption of C-containing and N-containing intermediates and further facilitates C-N coupling to form urea.

Fig. 3: Mechanism study.

a Operando Cd K-edge XANES spectra (The inset shows an enlarged view of the shaded region) and b Fourier transform of k³-weighted Cd K-edge EXAFS spectra for the r space at various potentials (The shaded region highlights the evolving trends in Cd-Cd bonding and coordination-environment). c In situ Raman spectroscopy for Ag_0.07/CdO. Operando DRIFTS for d Ag_0.07/CdO, e CdO and f Ag. g Three-dimensional operando DRIFTS for Ag_0.07/CdO.

To verify the above assumptions, in situ Raman spectroscopy is used to monitor the reaction intermediates on Ag_0.07/CdO, Ag and CdO during reduction (Fig. 3c and Supplementary Fig. S60a, b). Characteristic Raman peaks of major species NO₃⁻ (1048 cm^–1) and carbon paper (1380 and 1600 cm⁻¹) could be observed at open circuit potential (OCP)^14,45. For Ag_0.07/CdO catalyst, new peaks appear at 1189, 1250, 1550 and 2055 cm^–1 due to chelating nitrito (NO₂⁻), nitroxyl (ν (N=O)) and *CO vibration as the applied potential decreases^46,47, indicating that the catalyst effectively activates NO₃⁻ and CO₂. It is worth noting that Raman characteristic peak at 1016 cm^–1 is the symmetrical stretching vibration of the C-N group (ν_s (C-N))^37,46, confirming the coupling of C-N. Compared to Ag_0.07/CdO, pure Ag shows stronger characteristic peaks belonging to chelating nitrito and nitroxyl, while there is no obvious *CO vibration peak, indicating that Ag only facilitates the activation of NO₃⁻ to N-containing intermediates under these potentials. On the contrary, the weak vibrational peak of *CO is exhibited on CdO when it transforms into CdCO₃ at −0.1 V vs. RHE (Supplementary Fig. S60b), while Raman peaks belonging to transition state N-containing species are almost undetectable. The above comparison results indicate that when Ag_0.07/CdO catalyst is applied for co-electrolysis of CO₂ and NO₃⁻, the Ag component tends to adsorb NO₃⁻ and activate it to transition state N-containing intermediates such as *NO₂, *NO, etc., while CdCO₃ species with oxygen vacancies reconstructed from CdO adsorbs CO₂ to produce *CO intermediates. Conveniently, N-containing intermediates are further coupled with adsorbed *CO to efficiently produce urea. In situ Raman spectroscopy further shows the dynamic evolution of catalyst structure. In Supplementary Fig. S60c, d, Raman peaks of CdO (960 cm^–1) could be observed at OCP for CdO and Ag_0.07/CdO catalysts. As the applied potential shifts negatively, the Ag_0.07/CdO catalyst exhibits a significant CdCO₃ (1086 cm^–1)^48,49,50 vibration peak at more positive potential (from −0.05 V vs. RHE) compared with CdO catalyst (−0.1 V vs. RHE), indicating that the presence of Ag may promote the formation of CdCO₃. The possible reason for the facilitated non-faradic reconstruction of CdO with the presence of Ag is the enrichment of local CO₃^2– during the reduction process (Supplementary Fig. S61). Since Ag/CdO demonstrates more obvious current for the reduction of NO₃⁻ and CO₂ compared with CdO, more local protons would be consumed during the reaction, resulting in the quick enrichment of surface CO₃^2– and the promoted formation of CdCO₃ at higher potential.

In order to further reveal the urea formation mechanism on Ag_0.07/CdO, operando diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) is conducted during urea synthesis. The infrared signals of Ag_0.07/CdO, Ag and CdO catalysts are collected from 1000 to 2200 cm^–1 from −0.05 to −0.25 V vs. RHE (Fig. 3d–g). For Ag_0.07/CdO, a vibrational peak located at ~1210 to 1250 and 1516 cm^–1 are assigned to *NO₂ and *NH, respectively, and the two bands at around 1150 and 1623 cm⁻¹ could be assigned to the wagging and bending modes of –NH₂, suggesting that NO₃⁻ is efficiently activated^23,51. Additionally, the vibrational peaks located at 1317 and 1395 cm^–1 correspond to *CO and ν_s (OCO). Specially, the band at around 1476 cm^–1 is ascribed to the ν_s (C-N) and the stretching mode of C=O is also observed at ~1680 cm⁻¹, validating the formation of urea¹⁴. However, the peak of N-containing intermediates on CdO is quite weak. While pure Ag mainly exhibits stronger vibrational peaks belonging to N-containing intermediates (*NO₂, *NH, *NH₂, *NH₃) and the peak intensity of C-containing intermediates is weaker compared to Ag_0.07/CdO. This is consistent with the experimental results of high FE_NO2− and FE_NH3 on the Ag catalyst. Therefore, it can be fully assumed that Ag_0.07/CdO catalyst exhibits good urea synthesis performance due to the abundant N-containing and C-containing intermediates provided by the Ag and in situ reconstructed CdCO₃, creating sufficient collision opportunities for C-N coupling at a low overpotential.

Since the N-containing intermediates are quite complex, different N sources are added to the electrolyte for electrolysis to better identify the specific pathway of C-N coupling during urea synthesis. Supplementary Fig. S62 shows that FE_urea is 28.7% at −0.1 V vs. RHE when the N source is NO₂⁻, while FE_urea decreases to 5.4% when replacing with NH₂OH, and almost no urea generation is found in the addition of N₂H₄·H₂O. This alteration elucidates that *NO₂ may serve as the key N-containing intermediate for C-N coupling with *CO and subsequently generate urea. However, although NO₂⁻ as an N source may provide some *NO₂ for C-N coupling, the enhanced generation of NH₃ reduces its selectivity of urea (Supplementary Figs. S63 and 64). In addition, CO is also used to replace CO₂ for urea electrosynthesis in the presence of NO₃⁻. As shown in Supplementary Fig. S65, when the C source is replaced with CO, the FE_urea reaches 55.6% and the partial current density of urea increases to 1.33 mA cm^–2 (CO₂ as C source: 0.59 mA cm^–2) at −0.1 V in H-cell, indicating that that *CO is the key intermediate for C-N coupling. Therefore, we further conclude that the electrosynthesis of urea from NO₃⁻ and CO₂ may require the construction of binary active sites that can effectively activate NO₃⁻ and CO₂ simultaneously to provide sufficient *NO₂ and *CO intermediates.

Although the difficulty of simulating precise materials under catalytic conditions, DFT calculations are carried out on Ag (111), CdCO₃ (006), Ag (111)/CdCO₃ (006) and Ag (111)/CdCO₃ (006) with oxygen vacancies (Ag/CdCO₃-Ov) surfaces to further reveal the detailed reaction pathways in urea electrosynthesis (Supplementary Data 1 and Supplementary Figs. S66–73). As shown in Supplementary Figs. S66 and 67, the adsorption energies of NO₃⁻ are calculated on Ag, CdCO₃, Ag/CdCO₃-Ov-Ag site and Ag/CdCO₃-Ov-Cd site, respectively. Ag and Ag/CdCO₃-Ov-Ag site exhibit lower adsorption free energy (−1.02 eV and −1.08 eV, respectively) compared to CdCO₃ and Ag/CdCO₃-Ov-Cd site (−0.86 eV and −0.89 eV, respectively), implying that NO₃⁻ adsorbs more easily on Ag sites. In Fig. 4a, Ag exhibits lower free energy difference (−2.07 eV) compared to CdCO₃ (−1.18 eV) for the conversion of NO₃⁻ to *NO₂, indicating that the *NO₂ intermediate is more easily generated at the Ag site compared to CdCO₃ during the reduction of NO₃⁻ to *NO₂. This is also consistent with the experimental results of Ag catalyst to produce NO₂⁻, and confirms the results of Raman and DRIFTS. Ag/CdCO₃-Ov and Ag/CdCO₃ exhibits favorable free energy differences (−1.92 and −1.85 eV, respectively) for the formation of *NO₂ intermediates due to the presence of Ag, which would provide abundant N-containing intermediates for subsequent C-N coupling. In addition, the generation of *CO₂ on CdCO₃ seems to be spontaneous compared to Ag according to the calculated free energy difference, which means that CO₂ is preferentially adsorbed on CdCO₃. Specifically, it is worth emphasizing that the hydrogenation of *CO₂ is the potential determining step and Ag/CdCO₃-Ov (0.39 eV) reveals the lowest free energy difference compared to Ag/CdCO₃ (0.49 eV), CdCO₃ (1.19 eV) and Ag (1.35 eV), indicating that Ag_0.07/CdCO₃ with oxygen vacancies would promote the activation of CO₂ and further facilitate the C-N coupling. Therefore, a lower overpotential is observed on Ag/CdCO₃ in urea electrosynthesis. Subsequently, the *COOH is spontaneously reduced to *CO and C-N coupling occurs with the bonding of *NO₂ and *CO intermediates. Additionally, the free energy difference of forming *CONO₂ (−0.89 eV) on Ag/CdCO₃ surface is beneficial for early C-N coupling compared with the free energy difference of *NO₂ hydrogenation to *HNO₂ (−0.54 eV) and *CO desorption to CO (0.49 eV) in Fig. 4b, which may explain the high selectivity of urea production. As shown in Fig. 4c, the occurrence of the first C-N coupling has an activation energy barrier on Ag/CdCO₃ and Ag/CdCO₃-Ov. However, Ag/CdCO₃-Ov (0.561 eV) displays a lower activation energy barrier than Ag/CdCO₃, which may accelerate the reaction. In addition, the activation energy barrier of C-N coupling is found to be further decreased (0.442 eV) under a higher *CO coverage (Supplementary Figs. S74 and 75), indicating that the reaction rate of C-N coupling is facilitated, which is consistent with the experimental results (Supplementary Fig. S65). These results theoretically demonstrate that the in situ reconstructed Ag/CdCO₃ catalyst benefits the adsorption and coupling of N-containing and C-containing intermediates at low overpotentials and suppress the generation of by-products such as CO, NO₂⁻, etc.

Fig. 4: DFT calculations.

a Free-energy diagram for urea production on Ag (111), CdCO₃ (006), Ag (111)/CdCO₃ (006) and Ag (111)/CdCO₃ (006)-Ov. b Free-energy for *NO₂ and *CO reaction paths. c The initial, transition, and final states during the *CONO₂ formation of Ag/CdCO₃ and Ag/CdCO₃-Ov are presented.

Discussion

In summary, we synthesized an Ag/CdO heterostructured catalyst to promote the co-activation of CO₂ and NO₃⁻ and achieved efficient and selective urea electrosynthesis at low overpotentials. The optimized Ag_0.07/CdO catalyst displays a high urea yield rate of 111.6 ± 3.1 mmol g_cat.^–1 h^–1 with a Faradaic efficiency of 50.0 ± 2.7% at a positive potential of −0.15 V vs. RHE in a flow cell, and can further achieve stable response (1000 h) with a competitive urea yield rate of 427.0 ± 30.2 mmol g_cat.^–1 h^–1 at a current density of ~60 mA cm^–2 in a two-electrode cell with optimized membrane and flow field architecture. Operando XAS, Raman and DRIFTS prove that the Ag component in the synthesized Ag/CdO catalyst not only promotes the reconstruction of CdO component into CdCO₃ with oxygen vacancies, which can efficiently adsorb and activate CO₂, but also attracts the electrophilic intermediate *NO₂ during the reduction process. DFT calculations indicate that the in situ reconstructed Ag/CdCO₃ catalyst benefits the adsorption and coupling of N-containing and C-containing intermediates at low overpotential and suppresses the generation of by-products such as CO and NO₂⁻. We believe this work can provide guidance for the design of efficient urea electrosynthesis catalysts via CO₂ and inorganic nitrogen sources co-reduction and help enclose the cycle of carbon and nitrogen economy.

Methods

Materials

Cadmium acetate (Cd(CH₃COO)₂·2H₂O, Innochem, 99.9%), Silver nitrate (AgNO₃, Chengdu Kelong Co. Ltd., ≥99.8%), Potassium nitrate (KNO₃, Aladdin, 99.0%), Potassium nitrite (KNO₂, Innochem, 97%), Sodium nitrofenicyanide dihydrate (C₅H₄FeN₆Na₂O₃, Macklin, 99.0%), Sulfanilic acid (C₆H₇NO₃S, Innochem, 99%), N-(1-naphthyl) ethylenediamine dihydrochloride (C₁₂H₁₄N₂·2HCl, Macklin, 98%), Iron chloride anhydrous (FeCl₃, Sinopharm chemical reagent Co. Ltd., ≥97.0%), Sodium citrate (C₆H₅Na₃O₇, Sinopharm chemical reagent Co. Ltd., 98%), Thiosemicarbazide (CH₅N₃S, Macklin, 98.5%), Diacetylmonoxime (C₄H₇NO₂, Macklin, 99%), Salicylic acid (C₇H₆O₃, Tianjin Dengfeng Chemical Reagent Co. Ltd., 99.5%), Phosphoric acid (H₃PO₄, Sinopharm chemical reagent Co. Ltd., AR), Sodium hypochlorite (NaClO, Macklin, active chlorine ≥30%), ¹⁵N-Potassium nitrate (K¹⁵NO₃, Yuanye, abundance 99atom%), Hydroxylamine (NH₂OH, Shanghai Titan Scientific Co. Ltd., 50 wt.% in water), Sodium hydroxide (NaOH, Innochem, 99%), Urease (Yuanye, 1.1U/mg), potassium bicarbonate (KHCO₃, Aladdin, 99.5%), hydrazine hydrate (N₂H₄·H₂O, Sinopharm Group, 85%). Ethanol (AR) is used without any further purification and the Ultra-pure water (18.25 MΩ cm⁻¹) used in all experiments was prepared by ultra-pure purification system (Ulupure).

Synthesis of Ag_x/CdO, Ag and CdO

For Ag_x/CdO, 80 mg Cd(CH₃COO)₂·2H₂O and different quantities of AgNO₃ were added to 25 mL of water in a glass beaker and the solution is continuously stirred. When the solution became clear, 0.1 M NaOH (10 mL) was added to the solution. After stirring for 10 min, the solution was allowed to stand for 12 h. Then it was washed with alcohol and water and centrifuged. The obtained product was dried in a vacuum oven at 100 °C. Finally, high-temperature calcination (550 °C, 2 h) is carried out. And during the synthesis process of pure Ag and CdO, Cd(CH₃COO)₂·2H₂O and AgNO₃ are not added, respectively, and other conditions are the same as Ag_x/CdO.

Preparation of working electrodes

The as-prepared catalysts were deposited on the carbon black support for the co-reduction of CO₂ and NO₃⁻. The electrocatalysts (1 mg) and Vulcan XC-72 carbon powder (4 mg) were dispersed in a mixture of 500 μL isopropyl alcohol and 20 μL of 5% Nafion solution. Ultrasonic treatment of the above mixture for 60 min to obtain a uniform suspended ink. The ink was evenly dripped on both sides of carbon paper (about 1.0 cm × 1.0 cm), and vacuum dried at 60 °C for 1 h to obtain a working electrode. The effective catalyst loading is approximately 0.1 mg.

Characterizations

The morphologies and compositional elements were confirmed by a Field emission scanning electron microscope (FSEM) using the model Sirion 200. Transmission electron microscopy (TEM) was carried out on a Tecnai G2 transmission electron microscope (FEI, USA). High-resolution transmission electron microscope (HR-TEM) characterizations were obtained via a Talos F200X transmission electron microscope (FEI, Netherlands). X-ray diffraction (XRD) patterns were obtained via a DMAX-2400X-ray diffractometer (Rigaku, Japan). Raman measurements were performed using a LabRAM HR Evolution (Horiba Jobin Yvon, France) with 532 nm excitation wavelength. X-ray photoelectron spectroscopy (XPS) was carried out with an Axis-Ultra DLD-600W X-ray photoelectron spectrometer (Shimazu, Japan). The ¹H NMR is performed on a Bruker AscendTM 600 MHz. Operando diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was performed using a Nicolet iS50R spectrometer (Thermo Scientific, USA). X-ray absorption spectroscopy (XAS) was determined by the Shanghai Synchrotron Radiation Facility (SSRF) at the beamline BL11B1. X-ray fluorescence (XRF) patterns were obtained from EAGLE III operated at 40 kV. A UV spectrophotometer with the model TU-1810 was used to obtain absorbance.

Urea electrosynthesis testing in an H-type cell

Urea electrosynthesis tests were performed in a three-electrode two-compartment H-cell (The single-chamber volume is 50 mL) controlled by an electrochemical workstation (CHI660e, Chenhua). Two chambers were separated by a proton exchange membrane (Nafion^®212, Dupont, 50.8 μm). The Nafion^®212 membrane was first immersed in deionized water (18.25 MΩ cm^–1) at 60 ~ 80 °C for 30 min, then transferred to a 3–5 wt.% H₂SO₄ solution and soaked at 70 ~ 80 °C for an additional 30 min. Subsequently, the membrane was boiled in deionized water at 80 °C for 30 min, and finally stored under refrigeration at 4 ~ 8 °C. The working electrode (WE) was a catalyst-loaded carbon paper. An Ag/AgCl electrode (saturated KCl solution) and a graphite rod were used as the reference (RE) and counter electrodes (CE). The CO₂-saturated 0.1 M KHCO₃ + 0.05 M KNO₃ solution (40 mL, pH = 6.9 ± 0.1) was applied as the electrolyte and the flow rate of CO₂ during test was maintained at 20 mL/min by a gas flow meter (AST10-DBC, Asert Instruments (Beijing) Co. Ltd). The electrolyte was prepared by dissolving 10.01 g of KHCO₃ and 5.05 g of KNO₃ together, followed by dilution to 1000 mL in a volumetric flask and subsequent storage. Before the test, CO₂ gas was pumped into the cathode chamber electrolyte for 30 min, discharging the soluble oxygen in the solution and obtaining the CO₂-saturated solution. During the test, the cathode chamber electrolyte was stirred at 600 rpm. All the electrode potentials were measured against Ag/AgCl electrode and converted to the reversible hydrogen electrode (RHE) by E_RHE = E_Ag/AgCl + 0.197 + 0.059 × pH, where pH is the average pH value of electrolyte. All electrochemical measurements in H-cell did not have any iR compensation. And ECSA is estimated via ECSA = C_dl/C_s, Where the double-layer capacitance (Cdl) of the catalysts was estimated by measuring the capacitive current density associated with different scan rates via CV curves in a non-faradaic region in CO₂-saturated 0.1 M KHCO₃ + 0.05 M KNO₃ solution and the value was determined to be half that of the linear fitting slope, and C_s is the specific capacitance of a flat surface with 1 cm² of real surface area.

Urea electrosynthesis testing in a gas diffusion flow cell

Three electrode system: Urea electrosynthesis tests were performed in a gas diffusion flow cell consisting of a gas-diffusion layer (TGP-H-060, 20% PTFE, TORAY) deposited with Ag_0.07/CdO (1.0 cm × 1.0 cm) as the working electrode, nickel foam (1 mm, 1.0 cm × 1.0 cm) as the counter electrode, and Ag/AgCl (saturated KCl solution) as the reference electrode. 1.0 M KOH + 0.05 M KNO₃ solution (200 mL, pH = 14.0 ± 0.1) was used as the electrolyte. The electrolyte was prepared by dissolving 56.1 g of KOH and 5.05 g of KNO₃ together, followed by dilution to 1000 mL in a volumetric flask and subsequent storage. The electrolyte was maintained at 64 mL/min during electrolysis. Preparation of gas diffusion electrode: The electrocatalysts (1 mg) and Vulcan XC-72 carbon powder (4 mg) were dispersed in a mixture of 500 μL isopropyl alcohol and 20 μL of 5 wt% Nafion solution. Ultrasonic treatment of the above mixture for 60 min to obtain a uniform suspended ink. The ink was evenly dripped on one side of the gas-diffusion layer (1.0 cm × 1.0 cm) to obtain a gas-diffusion electrode. The effective catalyst loading is approximately 0.1 mg. Cathode and anode channels were separated by proton exchange membrane (Nafion®212, Dupont, 50.8 μm). In the gas channel, the flow rate of CO₂ during the test was maintained at 20 mL/min. The anodic process is typically the electrochemical oxidation of water to form oxygen in alkaline (i.e., the electrochemical oxygen evolution reaction, OER, 4OH⁻ → O₂ + 2H₂O + 4e⁻) and the potential of the nickel foam counter electrode is about 1.65 V vs. RHE during operation. All electrochemical measurements in the flow cell did not have any iR compensation.

Two-electrode cell (serpentine flow field with thin electrolyte layer of 1 ~ 2 mm): a gas-diffusion layer (TGP-H-060, 20% PTFE, TORAY) deposited with Ag_0.07/CdO (1.0 cm × 2.0 cm) was as the working electrode, NiFe-layered double hydroxide catalyst (1.0 cm × 2.0 cm) was as the counter electrode, 1.0 M KOH + 0.05 M KNO₃ solution (250 mL) was used as the electrolyte and the effective catalyst loading is approximately 0.2 mg cm^–2. Cathode and anode channels were separated by anion-exchange membrane (Sustainion grade RT, 50 μm) and the Nafion binder was replaced with PiperION. The Sustainion grade RT membrane was immersed in a 1 M KOH aqueous solution at 20 ~ 30 °C for at least 24 h, thoroughly rinsed with deionized water, and subsequently prepared for use.

Operando characterizations

Operando Raman spectroscopy was performed on a Renishaw inVia Raman microscope system with in situ Raman gas diffusion H-type electrolytic cell (Gaossunoin, China). The wavelength of the excitation source of the laser is 532 nm (50%). A 50 × long focal length distance objective (Leica) was used for focus. The graphite rod and Ag/AgCl electrode (saturated KCl solution) as the counter and reference electrodes, respectively. The spectrum was collected after 10 min of stable operation for each potential.

Operando diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS): during the testing process, an Ag/AgCl electrode and a carbon rod were used as the reference electrode and the counter electrode, respectively. A CO₂-saturated electrolyte containing 0.1 M KHCO₃ and 0.05 M KNO₃ was used for co-electrolysis of CO₂ and NO₃⁻. The background spectrum was obtained at the open circuit potential before each measurement. Then, absorption spectra at different potentials with a spectral resolution of 4 cm^–1 were collected. In the spectrum, a negative peak indicates the production or increase of a substance or functional group.

Operando X-ray absorption structure (XAS) spectroscopy: the operando Cd K-edge and Ag K-edge XAS spectroscopy were collected on the beamline BL11B1 of the Shanghai Synchrotron Radiation Facility (SSRF). The Ag foil is used to calibrate the energy. During the operando XAS measurements, the method of constant potential is applied. Operando XAS spectroscopy signals were collected in fluorescence mode using an in situ electrochemical cell with polyimide film windows (XAFS-2, Gaossunoin, China). The data reductions of the experimental spectra to normalized XANES and Fourier-transformed radial distribution functions (RDFs) were performed through the standard XAS procedure.

Product analysis

Gaseous products analysis: a gas chromatograph (GC, Shimadzu GC-2014) was applied to quantitative analysis of gaseous products.

The Faraday efficiency of the product is calculated based on the gas chromatographic peak area, and the formula is as follows:

$${{FE}}_{{gas}}=\frac{{{{C}}}_{{product}}\times {{V}}\times 2{{F}}{\rho }_{0}/{{RT}}}{{i}_{{total}}}\times 100\%$$

(1)

where i_total is the current density obtained by direct measurement at this potential, C_product stands for the volume density of gas products based on GC calibration, V is the flow rate of CO₂, F is the Faraday constant (96,485 C/mol), ρ_o is the atmospheric pressure (1.013 bar), and T is the test temperature (298 K), R is the ideal gas constant (8.314 J mol/K).

Urea product analysis: (1) Diacetylmonoxime-thiosemicarbazide (DAMO-TSC) method⁵². For H-cell, 1 mL electrolyte was added into 2 mL acid-ferric solution (10 mL phosphoric acid, 30 mL sulfuric acid, 60 mL deionized water and 10 mg ferric chloride). And then 1 mL DAMO-TSC solution (0.5 g DAMO and 10 mg TSC were dissolved in 100 mL deionized water) was added into the mixture. After that, the solution was heated to 100 °C and maintained for 20 min. UV-Vis absorption spectrum was performed after it was cooled to room temperature, and the absorbance at 525 nm was acquired. A series of standard urea solutions was used to obtain working curves for urea determination. (2) Urease decomposition method¹¹. 0.01 g EDTA-2Na and 0.049 g K₂HPO₄ were dissolved in urease solution (10 mL, 5 mg mL^–1). Then, 2.0 mL electrolyte was added into the above solution (0.2 mL). The final solution was left to stand for 40 min at 37 °C. The produced NH₃ was quantified with the indophenol blue method. For the above two methods of flow cell, neutralize the pH of 1 mL of electrolyte with 1 mL of 0.5 M H₂SO₄, and then take 1 mL of the above solution from this mixed solution and perform the same operation as H-cell. It is worth noting that this is equivalent to diluting the solution twice, so the actual concentration of urea in the electrolyte is twice that obtained from the spectrum.

NH₃ product analysis: Indophenol blue method^53,54. Typically, 2 mL of the electrolyte after electrolysis was mixed with 2 mL of a 1 M NaOH solution containing 5 wt% salicylic acid and 5 wt% sodium citrate. Subsequently, 1 mL of 0.05 M NaClO and 0.2 mL of 1 wt% sodium nitroferrocyanide solution were added. The mixture was then left to stand at room temperature for 2 h. Afterward, the absorbance at 655 nm was measured using UV spectrophotometry. Standard curves of concentration versus absorbance were prepared using ammonium chloride solutions of varying concentrations, with each solution containing the same electrolyte concentration as that used in the test samples.

NO₂⁻ product analysis^6,14: a total of 0.5 g of sulfonic acid was dissolved in a mixture of 90 mL of deionized water and 5 mL of acetic acid, followed by the addition of 5 mg of naphthalene ethylenediamine hydrochloride. The final volume was adjusted to 100 mL. This solution was used as the colorimetric reagent. Subsequently, 1 mL of the electrolyte after electrolysis was mixed with 4 mL of the colorimetric reagent and kept in a dark environment for 15 min. The absorbance at 540 nm was then measured using a UV spectrophotometer.

The FEs for urea, NH₃, and NO₂⁻ were calculated according to formulas:

$${{FE}}_{{urea}}=\frac{16\times {{F}}\times {{C}}\times {{V}}}{60.06\times {{Q}}}\times 100\%$$

(2)

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

(3)

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

(4)

where C is the concentration of generated urea (µg/mL), V is the volume of electrolyte and Q is the total charge passed through the working electrode.

¹H NMR, HPLC-MS and IC characterizations

¹H NMR:

DMSO-d₆ as deuterium reagent: the ¹H NMR sample was prepared by mixing 500 μL of post-reaction solution with 200 μL DMSO-d₆ and 50 μL of maleic acid solution (200 ppm) (1024 scans).

D₂O as deuterium reagent: 500 μL electrolyte mixed with 100 μL DMSO (0.05 μL/mL) and then dissolved in 150 μL D₂O solution, and DMSO was used as the internal standard solution.

HPLC-MS: HPLC-MS (Thermo Scientific Q Exactive) spectroscopy achieves the quantification of urea. A 5 μm ZORBAX SB-AQ column (150 × 4.6 mm, Agilent) suitable for increasing the retention value of polar compounds is performed. The corresponding mobile phase, flow rate, detected wavelength and ion source are acetonitrile-0.1% formic acid aqueous solution (1:99), 0.300 mL min⁻¹, 250 nm, and ESI + MS, respectively.

IC: IC data for ammonia, nitrate, and nitrite are obtained on an ion chromatograph model 881 Compact IC pro. In IC of anion test, the column type and eluent composition are prepared using Metrosep A Supp 5-250/4.0 and 3.2 mM Na₂CO₃/1.0 mM NaHCO₃, respectively, and the corresponding flow rate is 0.700 mL min⁻¹. In IC of cation test, the eluent composition and flow rate are 2.0 mM HNO₃ and 0.900 mL min⁻¹, respectively.

Computational details

DFT calculations were performed using the Vienna Ab-initio Simulation Package (VASP)^55,56. Exchange-correlation effects were treated within the generalized gradient approximation (GGA) using the Perdew-Burke-Ernzerhof (PBE) functional^57,58. Core-valence interactions were described with the projector augmented-wave (PAW) method⁵⁹. A plane-wave energy cutoff of 480 eV was used, and Brillouin-zone integration employed a 3 × 3 × 1 Monkhorst-Pack k-point mesh⁶⁰. A vacuum space of 15 Å was introduced normal to the surface to eliminate periodic interactions. Structures were relaxed to 1.0 × 10^–4 eV in energy and 0.02 eV Å^–1 in forces. When optimizing charged models like NO₃⁻ using VASP, we set the model with a negative charge based on the total number of atoms and the corresponding number of valence electrons. The computational hydrogen electrode (CHE) model was used to calculate the Gibbs free energy change (ΔG) of reaction step:

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

(5)

where ΔE is the electronic energy difference directly obtained from DFT calculations, ΔZPE is the zero-point energy difference, T is the room temperature (298.15 K) and ΔS is the entropy change. ZPE could be obtained after frequency calculation by ref. ⁶¹:

$${{\rm{ZPE}}}=\frac{1}{2}\,\sum {hv}$$

(6)

And the TS values of adsorbed species are calculated according to the vibrational frequencies⁶²:

$${{\rm{TS}}}={k}_{B}T\left[{\sum }_{k}{{\rm{ln}}}\left(\frac{1}{1-{e}^{-{hv}/{k}_{B}T}}\right)+{\sum }_{k}\frac{{hv}}{{k}_{B}T}\frac{1}{({e}^{{hv}/{k}_{B}T}-1)}+1\,\right]$$

(7)

where k_B is the Boltzmann constant, h is the Planck constant and v stands for vibrational frequencies.