Sequential-chain coupling over hierarchical click-sites enables highly selective urea electrosynthesis

Abstract

Electrochemical C − N coupling is an appealing approach for sustainable urea synthesis, while it is technically challenging due to the complex reaction mechanism and the spatiotemporal mismatch between C- and N- intermediates. Here, inspired by click chemistry, we design a hierarchical click-site catalyst (Se−InO_x) that enables an efficient sequential-chain coupling pathway for urea electrosynthesis, achieving a urea yield rate of 254.94 mmol h⁻¹ g⁻¹, Faradaic efficiency of 78.61%, >85% N_urea-selectivity and 100% C_urea-selectivity. Mechanistic studies reveal that Se−InO_x as the first click-site can selectively adsorb NO₃⁻ and hydrogenate it to stable *NO₂, while inhibiting CO₂ adsorption at this stage. The surface-anchored *NO₂ then acts as the second click-site to click couple with CO₂, forming the key *CO₂NO₂ intermediate. This sequential-chain coupling strategy effectively resolves the spatiotemporal mismatch between N- and C- intermediates, thereby maximizing the suppression of side-reactions and enhancing C − N coupling selectivity. Techno-economic analysis and scalable synthesis validate the feasibility of this approach, providing a blueprint for high-selectivity multicomponent electrosynthesis.

Introduction

Urea (CO(NH₂)₂), with annnual production exceeding 180 million tons, is an indispensable nitrogen-based fertilizer and a key intermediate in pharmaceuticals and fine chemicals^1,2,3,4. Its industrial production via the Haber−Bosch (350−550 °C, 150−350 bar) and Bosch−Meiser (150−200 °C, 150−250 bar) processes is highly energy-intensive^5,6,7,8, consuming 1–2% of global energy and emitting over 200 million tons of CO₂ annually (equivalent to 1.2 tons of CO₂ emissions per ton of urea, Fig. 1a)^9,10,11. Electrochemical urea synthesis (urea electrosynthesis) from waste-derived nitrate (NO₃⁻) and anthropogenic CO₂ under ambient conditions offers a sustainable alternative to fossil-dependent thermochemical processes, with the potential for net-negative carbon emissions (up to −0.6 tons of CO₂ per ton of urea, Fig. 1b)^{3,8,9,12,13,14,15,16,17,18,19,20,21}. However, this route is technically limited by low activity and selectivity.

Fig. 1: Schematic illustration of sustainable urea electrosynthesis and a mechanistic comparison between traditional electrocatalytic approaches and hierarchical click catalysis strategy.

a Comparison of traditional urea production via the industrial Bosch−Meiser process versus sustainable electrosynthesis driven by renewable electricity. The conventional Bosch−Meiser process operates under harsh conditions, consuming substantial energy and resulting in large CO₂ emissions. In contrast, the electrochemical approach powered by renewable energy provides a sustainable and carbon-negative alternative, directly converting waste-derived nitrate and anthropogenic CO₂ into urea under ambient conditions. b Net CO₂ emissions associated with the Bosch−Meiser process compared to sustainable electrochemical synthesis. c Mechanistic differences between conventional electrocatalytic processes and the hierarchical click catalysis strategy introduced herein. Traditional methods suffer significantly from competitive adsorption of reactants, spatial-temporal mismatches, and undesired side-reactions. Our proposed hierarchical click catalysis approach overcomes these limitations by introducing sequential and non-competitive hierarchical active sites, enabling precise adsorption and sequential coupling of nitrate and CO₂-derived intermediates. This effectively suppresses side-reactions and substantially enhances catalytic activity and selectivity toward C−N coupling to urea. Colour codes: C, grey; N, blue; O, red; H, white.

Recent advances in urea electrosynthesis have demonstrated promising improvements in activity and selectivity through the design of synergistic catalytic strategies^{9,14,15,16,22,23,24,25,26,27} such as bonded Fe−Ni pairs¹⁵, alternating CuWO₄ dual-metal sites¹⁶ and heterostructured Cu−Bi bimetallics²². While these studies underscore the immense promise of sustainable urea production, major challenges remain as follows: (i) a complex 16-electron transfer involving two C−N coupling steps results in broad product distributions^16,17,23; (ii) spatial separation of active sites leads to spatiotemporal mismatches between C- and N- intermediates (Fig. 1c)^19,22; and (iii) insufficient regulation over the cross-adsorption of NO₃⁻, CO₂ and their corresponding N- and C- intermediates causes undesired cross-contamination^14,28. These factors collectively promote undesired CO₂ reduction (CO₂RR) and nitrate reduction (NO₃RR) side-reactions that suppress C − N coupling and limit urea selectivity^9,12,15,18. These challenges underscore the urgent need for new catalytic strategies capable of orchestrating C−N coupling with high precision and selectivity.

Click chemistry, awarded the 2022 Nobel Prize in Chemistry, offers a highly modular and selective framework for assembling complex molecules under mild conditions with high efficiency and minimal by-products^29,30,31. Inspired by its ability to precisely stitch together small molecular units, we propose a hierarchical click-site strategy for urea electrosynthesis, in which two reactants, NO₃⁻ and CO₂, undergo sequential-chain coupling. The first click-site selectively anchors and activates either NO₃⁻ or CO₂ while excluding the other, followed by a second click-site that promotes targeted coupling with the remaining reactant via a nucleophilic attack on the surface-bound intermediate. This hierarchical click-site strategy, inspired by click chemistry, enables precise sequential-chain adsorption of NO₃⁻ and CO₂ on a single active site, addressing key challenges in C−N coupling selectivity by mitigating (i) competitive intermediate adsorption, (ii) spatiotemporal mismatches, and (iii) interfacial cross-contamination. Despite its conceptual promise, this approach remains unexplored in urea electrosynthesis due to two major challenges: (i) the difficulty of achieving selective adsorption of only one feedstock at the first click-site, given the comparable affinity of most catalysts for both NO₃⁻ and CO₂; and (ii) the instability of the second click-site, which is often over-hydrogenated or prematurely desorbed.

To overcome these challenges, we systematically screened representative catalysts from CO₂RR, NO₃RR and C−N coupling, identifying InO_x as a promising base due to its weak CO₂ adsorption via O-terminal (*OCO) configurations, disfavoring *CO formation while preferentially adsorbing and stabilizing nitrate-derived *NO₂ (Supplementary Fig. 1)^{32,33,34,35,36,37}. Guided by molecular orbital theory (Supplementary Fig. 2)³⁸, substituting lattice oxygen with a larger and less electronegative element is expected to increase electron density at the active centers and induce moderate lattice distortion, thereby making CO₂ adsorption thermodynamically unfavorable. Following this rationale, we replaced lattice O with Se, yielding a Se-modulated InO_x catalyst (Se−InO_x) with enhanced *NO₂ stabilization and fully suppressed CO₂ binding. In this system, Se−InO_x (Fig. 1c) serves as the first click site, selectively adsorbing NO₃⁻ and converting it to stable *NO₂ without interacting with CO₂, while suppressing undesired NO₃RR and CO₂RR. The surface-bound *NO₂ then functions as second click site, facilitating the selective nucleophilic attack of CO₂ at the N site of *NO₂ to form the key *CO₂NO₂ intermediate. This sequential-chain coupling aligns C- and N- intermediates both spatially and temporally, enabling precise C−N bond formation while minimizing cross-contamination and by-product generation. Overall, the hierarchical click-site concept enables selective and sustainable urea electrosynthesis, establishing a generalizable framework for high-selectivity multicomponent electrosynthesis.

Results

Design of hierarchical click-site catalysts

Effective implementation of hierarchical click catalysis requires sequential and selective adsorption of CO₂ and NO₃⁻, wherein CO₂ adsorption is selectively suppressed while NO₃⁻ is preferentially adsorbed and subsequently transformed into a stable *NO₂ during the first click step. To identify suitable host for this function, we initially screened several metals commonly employed in CO₂RR. Metals such as Cu, Fe, Zn, and Ag exhibited strong CO₂ adsorption via carbon-terminal (*COO) modes (Fig. 2a), leading to complex intermediates and products, such as methane, methanol, ethylene, and ethanol, etc.^32,33. Conversely, metals such as Bi, Sn, and particularly In, exhibit weaker CO₂ adsorption via the oxygen-terminal (*OCO) mode, predominantly yielding formate (HCOOH)³⁴. Among these, In-based catalysts, represented by In₂O₃, exhibit the weakest CO₂ affinity^35,36,37. Meanwhile, the intrinsically high activation barrier from *NO₂ to *NOOH in NO₃RR favors the stabilization of *NO₂ (Supplementary Fig. 1), positioning In₂O₃ as a promising candidate for hierarchical click-site design. However, pristine In₂O₃ still retains measurable CO₂RR activity, necessitating further electronic modification to completely suppress undesired CO₂ adsorption^36,37. Based on molecular orbital theory (Supplementary Fig. 2)³⁸, enhancing the local electron density around In centers can weaken interactions with the electrophilic CO₂ molecule. To this end, we substituted lattice oxygen with larger and less electronegative anions (S, Se) to modulate the local electronic environment of the In₂O₃. Electron localization function (ELF) analysis confirmed the enhanced electron density around In atoms upon doping with S and Se (Fig. 2b). This result is quantitatively supported by charge redistribution analysis (Fig. 2c and Supplementary Figs. 3 and 4), manifesting as an increase in electron density around In atoms from 1.24 (InO_x) to 1.83 (Se−InO_x). Structural distortion indices (Di) further revealed progressive lattice distortion from InO_x (Di=0.014) to S−InO_x (Di=0.039) and Se−InO_x (Di=0.048, Fig. 2d). These distortions shift electronic states toward the Fermi level (Fig. 2e), weakening intrinsic In−O covalent bonds, thus enabling selective adsorption and minimizing undesired side-reactions.

Fig. 2: Optimization of CO₂ and NO₃⁻ adsorption on catalyst surfaces.

a Comparison of CO₂ adsorption strength across representative metal-based catalysts. b Electron location function maps for InO_x, S−InO_x, and Se−InO_x. c Charge variations at In and doping sites (X=O, S, Se) in InO_x, S−InO_x, and Se−InO_x. d Corresponding crystal structures illustrating lattice distortions, with Di denoting distortion indices. e Effect of atomically dispersed Se on the electronic environment of In sites within Se−InO_x. f, g Comparative CO₂ adsorption energies (f) and ICOHP analysis of In−O_CO2 bond (g) for InO_x, S−InO_x, and Se−InO_x. h, i Comparative NO₃⁻ adsorption energies (h) and corresponding ICOHP analysis of In−O_NO3− bond strengths (i) for InO_x, S−InO_x, and Se−InO_x. j, k Schematic illustrations of CO₂ and NO₃⁻ adsorption behaviors on pristine InO_x and Se−InO_x surfaces. Colour codes: C grey, O red, N blue, H white, In orchid, S yellow, Se light green, active site white bar. Source data are provided as a Source data file.

Density functional theory (DFT) calculations were conducted to validate the modulation of surface adsorption behavior. Substitution of lattice oxygen with S or Se rendered CO₂ adsorption thermodynamically unfavorable, with the adsorption energy shifting from −0.26 eV (InO_x) to +0.17 eV (S−InO_x) and +0.27 eV (Se−InO_x), respectively (Fig. 2f). Integrated crystal orbital Hamilton population (ICOHP) analyses revealed decreased In−O_CO2 bonding strength on Se−InO_x surfaces, confirming the suppression of CO₂ adsorption (Fig. 2g and Supplementary Fig. 5-7). In contrast, NO₃⁻ adsorption progressively strengthened from InO_x to S−InO_x and Se−InO_x, as indicated by increased covalency in In−O_NO3− interactions observed via ICOHP analysis, projected density of states (PDOS), and differential charge density mapping (Fig. 2h, i and Supplementary Figs. 8–10). As expected, the Se−InO_x displayed the strongest *NO₂ adsorption and the highest hydrogenation barrier toward *NOOH formation (Supplementary Fig. 11), effectively preventing the over-reduction to NH₃ by-product. Collectively, these results indicate that Se−InO_x acts as an efficient hierarchical click-site, selectively adsorbing NO₃⁻ and stabilizing it as *NO₂, while completely suppressing CO₂ adsorption and thereby minimizing undesired side-reactions (Fig. 2j, k).

Characterization of hierarchical click-site catalysts

The Se−InO_x catalyst was synthesized via a two-step pyrolysis process. Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images revealed that Se−InO_x consisted of porous nanoparticles with an average diameter of ~30 nm, comparable to InO_x and S−InO_x (Fig. 3a and Supplementary Figs. 12–17). Powder X-ray diffraction (XRD) analysis (Fig. 3a) confirmed retention of the cubic In₂O₃ phase (PDF#71-2195) after S or Se incorporation. High-resolution aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC HAADF-STEM) demonstrated distinct lattice fringes of 0.25 nm indexed to the (040) and (004) planes along the [100] zone axis (Fig. 3b, c), in agreement with simulated HR-TEM and electron diffraction patterns (Supplementary Figs. 18–21). Atomic-resolution STEM-energy dispersive X-ray spectroscopy (STEM-EDS) mapping distinctly revealed atomically dispersed Se uniformly substituting for O atoms around In sites (Fig. 3d), corroborated by X-ray photoelectron spectroscopy (XPS, Supplementary Figs. 22 and 23). Elemental analysis determined the Se content as ~2.92 at%, comparable to metal single-atom concentrations corresponding to a near-stoichiometric atomic dispersion^39,40. Similar substitution levels were observed for S–InO_x (Supplementary Table 1). Complementary integrated differential phase contrast (iDPC) imaging revealed local lattice distortions arising from the substitution of larger-radius Se atoms in place of smaller O atoms within the InO_x lattice (Fig. 3e and Supplementary Figs. 24–26), consistent with previous distortion analysis and further confirmed by geometric phase analysis (GPA) strain (Supplementary Fig. 27)⁴¹. Based on these findings, an atomic model was constructed (Fig. 3f), highlighting how Se substitution at O lattice sites induced lattice rearrangements, modifying the local electronic environment of the In active sites and thereby influencing catalytic performance. We performed X-ray Absorption Near Edge Structure (XANES) and Extended X-ray Absorption Fine Structure (EXAFS) measurements at the In K-edge to examine local coordination and oxidation state changes on the InO_x, S−InO_x, and Se−InO_x catalysts (Supplementary Fig. 28−31 and Supplementary Table 3). The XANES spectra (Supplementary Fig. 30a) show that the absorption edges of three samples lie between those of metallic In and In₂O₃, and are close to that of In₂O₃, suggesting the oxidation state is slightly below +3 in all samples. Notably, with the incorporation of S and Se, a gradual redshift of the edge was observed, indicating a gradual reduction in the In oxidation state. This trend is attributed to the larger atomic radii and lower electronegativities of S and Se, which promote electron localization around In centers and reduce the symmetry of In−O coordination, thereby enhancing electronic back-donation to In atoms. The non-phase-corrected FT-EXAFS spectra (Supplementary Fig. 30b) display a sharp first-shell peak for pristine InO_x, whereas broader and more asymmetric peaks appear in S−InO_x and Se−InO_x, indicating increased local disorder and the presence of additional In−S or In−Se bonds. This interpretation is further supported by wavelet transform (WT) analysis (Supplementary Fig. 31), where the maximum intensity for Se−InO_x (~4.60 Å⁻¹) exceeds that of S−InO_x (~4.20 Å⁻¹) and InOx (~4.00 Å⁻¹), reflecting contributions from heavier scatterers. EXAFS fitting (Supplementary Table 3) reveals that all three samples maintain a similar In coordination number (N ≈ 6), with partial substitution of In−O by In−S or In−Se (N ≈ 0.6) in the doped catalysts. These comparable coordination environments establish a solid basis for subsequent structure-property analysis. Collectively, these findings confirm that S and Se incorporation induces subtle local distortions and partially modulates the In oxidation state, without drastically altering the overall coordination geometry.

Fig. 3: Structural characterization and mechanistic insights into hierarchical click-sites.

a XRD patterns of InO_x, S−InO_x and Se−InO_x; inset of (a) shows a TEM image of Se−InO_x. b, c Aberration-corrected HAADF-STEM images along the [100] direction; inset of (c) shows an intensity line scan for the atoms within the yellow boxed area. d Atomic-level STEM-EDS element mapping images of In + O + Se, In + Se, In + O and HAADF + O + Se. e iDPC−STEM image along the [100] direction; wherein yellow and red circles indicate positions of In and O/Se, respectively, and arrows indicate the distortion. f Graphical representation of InO_x and Se−InO_x, wherein blue, pink, and purple dots represent In, O, and Se atoms, respectively. g In-situ ATR−SEIRAS results on InO_x and Se−InO_x surfaces saturated with CO₂, NO₃⁻ and CO₂ + NO₃⁻ under operando potentials. h Experimental (Exp.) and simulated (Sim.) EPR spectra. i FE_HCOOH and FE_NO2- on InO_x, S−InO_x and Se–InO_x surfaces at 50 mA cm⁻². j Proposed catalytic mechanism of hierarchical click catalysis on the Se−InO_x surface. Colour codes: C grey, O red, N blue, H white, In orchid, Se light green. Source data are provided as a Source data file.

Subsequently, in-situ attenuated total reflection infrared (ATR-SEIRAS) spectroscopy under CO₂-saturated conditions showed pronounced CO₂ adsorption on InO_x but negligible signals on Se−InO_x (Fig. 3g and Supplementary Fig. 32), consistent with theoretical predictions of suppressed CO₂ affinity (Fig. 2f, g). In contrast, under conditions with only NO₃⁻ present, Se–InO_x exhibited markedly enhanced adsorption and stabilization of *NO₂ species. Notably, simultaneous introduction of CO₂ and NO₃⁻ resulted in the detection of a red-shifted CO₂ adsorption peak and a blue-shifted *NO₂ signal on the Se−InO_x surface (Fig. 3g and Supplementary Fig. 33), suggesting direct interaction and electron transfer between adsorbed CO₂ and *NO₂, thus forming the crucial *CO₂NO₂ intermediate. Electron paramagnetic resonance (EPR) spectroscopy confirmed signals for *CO₂NO₂ and *CONH_x intermediates (Fig. 3h), further supporting direct C−N coupling on Se−InO_x. Complementary electrochemical characterization reinforced the spectroscopic findings (Fig. 3i). Based on these results, we propose a hierarchical click-site catalytic mechanism proceeding via a sequential-chain coupling pathway (Fig. 3j): (i) Se doping activates adjacent In sites as first click-sites, which selectively adsorb NO₃⁻ and rapidly convert it into stable *NO₂ without competing CO₂ adsorption. (ii) These *NO₂ species then function as second click-sites, selectively coupling with CO₂ at the N-center to generate the pivotal *CO₂NO₂ intermediate. This sequential-chain design effectively suppresses side-reactions associated with CO₂RR and NO₃RR, thereby overcoming the intrinsic challenges of competitive adsorption and spatiotemporal mismatch (Supplementary Fig. 34).

Enhanced urea selectivity on hierarchical click-site catalysts

The urea electrosynthesis performance of hierarchical click-site catalysts was evaluated using gas-diffusion electrodes loaded with catalysts in a flow electrolysis cell with CO₂-saturated 0.1 M KHCO₃ + 0.1 M KNO₃ electrolyte (Supplementary Fig. 35). Se−InO_x with 2.92 at% Se displayed the highest urea FE, yield rate and J_urea among samples with varying Se contents (1.16, 2.92 and 6.77 at%) and was thus selected for subsequent evaluation (Supplementary Figs. 36–44 and Table 2). As shown in Fig. 4a, b, urea yield rate and FE increased progressively with S and Se incorporation, particularly Se−InO_x achieved a competitive urea yield rate and FE at −0.4 V (vs. RHE) with 69.29 ± 7.48 mmol h⁻¹ g⁻¹ and 62.44 ± 2.85%, surpassing S−InO_x (49.45 ± 6.51 mmol h⁻¹ g⁻¹ and 44.15 ± 3.50%) and significantly outperforming InO_x (14.85 ± 2.11 mmol h⁻¹ g⁻¹ and 10.02 ± 1.12%, Supplementary Tables 4–6). In addition, the S/Se co-doped sample (S_1.16Se_1.64 − InO_x) exhibited intermediate urea electrosynthesis activity between that of S−InO_x and Se−InO_x, further corroborating the electronic modulation effect of Se doping on the In active sites (Supplementary Figs. 45–49 and Table 1). Full product analysis showed similarly low NH₃ and NO₂⁻ yield rates among the three catalysts, while HCOOH production predominated on InO_x but was strongly suppressed on Se−InO_x with enhanced urea performance (Supplementary Figs. 50, 51 and Tables 7–9). These observations, supported by partial current densities (J_product) and Tafel slopes (Supplementary Figs. 52–58), are consistent with theoretical and in-situ spectroscopic evidence showing that Se introduction promotes C−N coupling while suppressing CO₂RR and NO₃RR side-reactions. Control experiments confirmed that urea formation required the co-presence of CO₂ and NO₃⁻ (Fig. 4c), affirming that urea arose from their co-electroreduction. Moreover, Se−InO_x exhibited stable electrochemical performance over 10 consecutive cycles (Fig. 4d), with negligible variation in urea yield rate, J_urea, and FE. Post-reaction characterizations (XRD, HAADF-STEM, STEM mapping, diffraction, and XPS, Supplementary Figs. 59–62 and Table 1) confirmed that the atomic-scale structural integrity, homogeneous elemental distribution, and original electronic configuration after prolonged electrochemical operation of Se−InO_x, suggesting good electrochemical durability and structural stability.

Fig. 4: Urea electrosynthesis performance of Se–InO_x.

a, b Urea yield rates and FEs of Se−InO_x, S−InO_x, and InO_x in CO₂-saturated 0.1 M KHCO₃ + 0.1 M KNO₃ electrolyte. c Comparison of urea yield rates for Se−InO_x under different gas/electrolyte conditions. d Cycling stability test of Se−InO_x over 10 consecutive electrolysis cycles at −0.4 V (vs. RHE). e Effect of NO₃⁻ concentration on the FE distribution of all products. f FEs of all products for Se−InO_x, S−InO_x, and InO_x at −0.4 V (vs. RHE) in CO₂-saturated 0.1 M KHCO₃ + 0.15 M KNO₃ electrolyte. g, h N−selectivity (g) and C−selectivity (h) for Se−InO_x, S−InO_x, and InO_x in CO₂-saturated 0.1 M KHCO₃ + 0.15 M KNO₃ electrolyte. i Comparison of optimal urea yield rates, FEs, J_urea, N_urea−selectivity, and C_urea−selectivity for Se−InO_x, S−InO_x, and InO_x. j Comparison of urea yield rates and FEs of Se−InO_x with other catalysts reported in the literatures. Electrolyte pH: 6.8 ± 0.5. All potentials are without iR compensation. Error bars represent the standard deviation from three independent measurements. Source data are provided as a Source data file.

The first click-site requires sufficient NO₃⁻ to generate adequate *NO₂ on the catalyst surface, while the second click-site necessitates a comparable amount of CO₂ to effectively couple with *NO₂. Further, optimized urea selectivity was achieved by balancing NO₃⁻ and CO₂ availability at the catalytic interface through controlled adjustment of NO₃^– concentrations (c(NO₃⁻)) and CO₂ flow rate. At c(NO₃⁻) below 0.15 M (0.05 and 0.10 M), HCOOH and H₂ were detected as the main by-product alongside urea, with minor amounts of NO₂⁻ and NH₃ (Fig. 4e). Increasing the c(NO₃⁻) to 0.15 M resulted in the highest urea FE of 78.61%, with HCOOH undetected and minimal production NO₂^–, NH₃ and H₂ production (Fig. 4e), indicating an optimal dynamic balance of NO₃⁻ and CO₂ at the catalyst surface. Further increasing c(NO₃⁻) resulted in increased NO₂⁻ and NH₃ formation, suggesting over-hydrogenation to NH₃ by-product, consistent with a concurrent decrease in H₂ evolution. We further evaluated the effect of CO₂ flow rate at fixed c(NO₃⁻) = 0.15 M (Supplementary Fig. 63). At low flow rates (≤40 mL min⁻¹), urea FE decreased markedly, while NO₃RR by-products became dominant, likely due to limited CO₂ accessibility at the catalyst-electrolyte interface. This imbalance led to *NO₂ accumulation, followed by desorption or over-hydrogenation to NH₃. Increasing the flow rate to 80 mL min⁻¹ significantly enhanced urea FE and suppressed NO₂⁻ and NH₃ formation, thereby mitigating side reactions (Supplementary Fig. 63). Notably, HCOOH remained undetectable throughout. These findings define 0.15 M NO₃⁻ and 80 mL min⁻¹ CO₂ as the optimal conditions for achieving high urea selectivity on hierarchical click-site catalysts. Even under optimal interfacial conditions, InO_x favored CO₂RR (FE_HCOOH = 41.28%, C_HCOOH–selectivity = 94.42%, Fig. 4f–h and Supplementary Figs. 64 and 65) over NO₃RR (FE_NH3+NO2- = 20.85%, N_NH3+NO2-–selectivity = 73.38%) and C−N coupling (FE_urea = 19.52%, N_urea−selectivity = 26.62%, C_urea−selectivity = 5.58%). Conversely, S−InO_x and Se−InO_x notably suppressed CO₂RR and NO₃RR side-reactions. Notably, Se−InO_x achieved a urea FE of 78.61%, N_urea−selectivity of 85.32%, and C_urea−selectivity of 100% (Figs. 4f–h and Supplementary Figs. 64 and 65), alongside significantly reduced NH₃ formation (FE_NH3 = 0.53%, N_NH3−selectivity = 0.58%) and completely inhibits HCOOH formation (FE_HCOOH = 0.00%, N_HCOOH−selectivity = 0.00%). Thus, hierarchical click-site engineering endowed Se−InO_x with competitive urea yield rate, FE, J_urea, and N_urea/C_urea−selectivities (Fig. 4i), demonstrating promising performance among reported electrocatalysts (Fig. 4j and Supplementary Table 10).

Mechanistic verification

To elucidate the underlying mechanism of urea electrosynthesis, DFT calculations (Figs. 5a–f, Supplementary Fig. 66, Data 1 and Table 11) and in-situ spectroscopic analyses were conducted (Fig. 5g–j). As shown in Fig. 5b, two major uphill steps dominate the urea formation pathway, i.e., the initial *NO₂ and CO₂ coupling to form *CO₂NO₂ (*NO₂ + CO₂ → *CO₂NO₂), identified as the likely rate-determining step (RDS), and the subsequent protonation of *CO₂NH₂ to *COOHNH₂. Notably, the second coupling step occurred spontaneously. Se incorporation significantly lowered the energy barrier for *CO₂NO₂ formation from 0.98 eV (InO_x) to 0.55 eV (Se−InO_x), and for *CO₂NH₂ protonation from 0.72 to 0.45 eV, suggesting a more favorable energy landscape. All catalysts also showed a thermodynamic preference for *NO₂ + CO₂ → *CO₂NO₂ over *OCHO + H → *OCHOH and *NO₂ + H → *NOOH (Fig. 5d–f and Supplementary Figs. 67–69), with Se−InO_x showing the largest energy gap (0.87 and 2.78 eV), compared to S−InO_x (0.20 and 2.59 eV) and InO_x (0.04 and 1.37 eV), which may help reduce HCOOH and NH₃ by-products. CI-NEB calculations suggest that CO₂ diffusion and subsequent coupling with anchored *NO₂ proceeds with reduced barriers of 0.87, 0.66, and 0.48 eV for InO_x, S−InO_x, and Se−InO_x, respectively (Supplementary Fig. 70). The resulting *CO₂NO₂ intermediates are predicted to exhibit progressively shorter C−N bond lengths (1.81 to 1.61 Å) and higher ICOHP values (2.08 to 3.50; Supplementary Figs. 71a and 72), implying stronger C−N covalent bonding upon Se doping. COHP analysis further reveals an upward shift of the antibonding orbital, which may enhance bond stability (Supplementary Fig. 71b). The negligible role of *NH₂- and *NH₂OH-based coupling with CO₂ highlights the efficiency and specificity of the *NO₂ + CO₂ pathway (Supplementary Fig. 73). Additionally, Se−InO_x showed the lowest H₂O dissociation barrier, which could facilitate proton supply for subsequent hydrogenation of *CO₂NO₂ to urea (Supplementary Figs. 74–80).

Fig. 5: Mechanistic study of C−N coupling via hierarchical click catalytic sites.

Urea electrosynthesis reaction pathway diagrams on S−InO_x (a) and Se−InO_x (c). b Calculated free energy diagram for urea electrosynthesis. Comparison of energy barriers from *NO₂ to *NOOH and *CO₂NO₂ on InO_x (d), S−InO_x (e), and Se−InO_x (f). Electrochemical in-situ ATR-SEIRAS spectra on InO_x (g), S−InO_x (h), and Se−InO_x (i) in CO₂-saturated 0.1 M KHCO₃ + 0.1 M KNO₃ electrolyte. j Electrochemical in-situ EPR spectra on Se−InO_x in CO₂-saturated 0.1 M KHCO₃ + 0.1 M KNO₃ electrolyte (Sim. = simulated). k Schematic representation of hierarchical click-site mechanism: Se-induced tuning of In electronic structure enables selective NO₃⁻ adsorption and stabilization of *NO₂ (1st click-site), followed by spatially guided CO₂ adsorption to form *CO₂NO₂ (2nd click-site), ultimately driving selective C − N coupling toward urea. Colour codes: C grey, O red, N blue, H white, In orchid, S yellow, Se light green. Source data are provided as a Source data file.

In-situ ATR-SEIRAS further confirmed these insights (Fig. 5g–i). On InO_x surface (Fig. 5g), strong *HCOO signals at ~1650 cm⁻¹ was increased with negative potential, consistent with dominant CO₂RR⁴². Concurrently, a *NO₂ signal at ~1200 cm⁻¹ diminished with increasing potential, suggesting poor *NO₂ stability on InO_x⁴³. The absence of *COOHNH₂ and *C−N signals implies that *NO₂ conversion led primarily to NH₃, as evidenced by the increasing *NH₂ signal¹⁸. Therefore, CO₂RR and NO₃RR are independent and dominant on the InO_x surface. For S−InO_x, *HCOO signals are suppressed, while *NH_x/*NH₂ and NH₂OH bands are enhanced, indicating improved NO₃⁻ adsorption and conversion to NH₃ compared to InO_x (Fig. 5h)¹⁷. Notably, the *NO₂ signal decreased at −0.3 V (vs. RHE) and gradually transformed into *CO₂NO₂, accompanied by minor *COOHNH₂ and *C−N signals, suggesting improved urea electrosynthesis. Se−InO_x showed no *HCOO signal, and negligible *NH_x, *NH₂, and NH₂OH signals, indicating complete suppression of CO₂RR and significant inhibition of NO₃RR. Instead, a clear transformation of *NO₂ to *CO₂NO₂ at −0.2 V (vs. RHE) was observed (Fig. 5i), with rapid growth of the *CO₂NO₂ signal and accompanying *C−N and *COOHNH₂ signals, indicating a preference for C−N coupling reactions on Se−InO_x. ^11,44 These findings were further supported by a series of control experiments, demonstrating sequential-chain coupling pathways at hierarchical click-sites (Supplementary Fig. 81).

In-situ EPR spectra further provided additional support for the sequential-chain coupling pathway on Se−InO_x (Fig. 5j). The signal intensity of *CO₂NO₂ increased with applied potential, in agreement with the ATR-SEIRAS findings, confirming the role of *NO₂ and CO₂ in the C−N coupling mechanism. Based on these results, we propose a hierarchical click-site catalysis mechanism via a sequential-chain coupling pathway (Fig. 5k), where Se-induced electronic modulation of In sites selectively activates of NO₃⁻ to form *NO₂ without CO₂ adsorption, constituting the first click step. The adsorbed *NO₂, along with the catalyst, facilitates selective coupling with CO₂ as the second click step. The resulting *CO₂NO₂ undergoes hydrogenation to urea with minimal side-reactions, aided by efficient H₂O dissociation, providing protons. To probe the catalyst behavior under operating conditions, we performed in-situ electrochemical XAS at the In K-edge on the InO_x, S−InO_x, and Se−InO_x catalysts. As shown in Supplementary Figs. 82–87, shifting the applied potential from open-circuit potential (OCP) to −0.6 V (vs. RHE) resulted in only minor edge shifts across all samples, indicating that the In centers retain relatively stable electronic structures during operation. These subtle changes reflect moderate electron transfer between In sites and the adsorbed intermediates involved in C−N coupling, CO₂RR, and NO₃RR, without inducing significant redox transitions. Additionally, FT-EXAFS spectra show that the In−S and In−Se coordination environments in S−InO_x and Se−InO_x remain essentially unchanged under applied potentials (Supplementary Tables 12–14), demonstrating the structural robustness of these heteroatom-doped environments during electrocatalytic operation.

Techno-economic evaluation and scale-up of urea electrosynthesis

To enhance the economic viability of urea electrosynthesis, we increased the electrolyte concentration to CO₂-saturated 1 M KHCO₃ + 0.1 M KNO₃, leveraging the known benefits of high K⁺ concentration in accelerating C−N coupling via improved proton/electron transfer^11,45. This high K⁺ concentration significantly enhanced the urea yield rate and J_urea, achieving peak values of 254.94 mmol h⁻¹ g⁻¹ and 21.87 mA cm⁻² (Fig. 6a and Supplementary Fig. 88), which were 2.7 times those of low K⁺ concentration, respectively. A techno-economic analysis (TEA, Fig. 6b and Supplementary Table 15) showed that the minimum cost of urea production using Se−InO_x is US$1.01 kg⁻¹ at an electricity price of US$0.03 kWh⁻¹, close to market levels but highlighting limited competitiveness at higher electricity costs. However, the production cost could be reduced to as low as US$0.16 kg⁻¹, if electricity cost is reduced to US$0.01 kWh⁻¹, enabled by sustainable energy sources and by-products are valorized⁴⁶. Subsequently, a financial net present value (FNPV) analysis was conducted, amortizing capital and operational costs over a 20-year lifespan. Profitability can only be achieved at electricity prices of US$0.01 kWh⁻¹ and current densities ≥50 mA cm⁻² (Fig. 6c and Supplementary Fig. 89), with payback periods decreasing substantially from 19-years at 50 mA cm⁻² (FNPV = US$82,923) to just 6-years at 100 mA cm⁻² (FNPV = US$7,141,881). Despite the gradual decline in FE at higher current densities, the economic value of by-products ensured cost recovery and profitability.

Fig. 6: Techno-economic analysis and scale-up of urea electrosynthesis.

a Urea yield rate and FE of Se−InO_x under different current densities in CO₂-saturated 1 M KHCO₃ + 0.1 M KNO₃ electrolyte. b The production cost per kg of urea at electricity prices of US$0.03 and US$0.01 kWh⁻¹. c FNPV of Se-InO_x as a function of current density at an electricity price of US$0.01 kWh⁻¹. d Sensitivity analysis of single-variable impact on urea production cost; Bars show the cost change when each parameter is varied by ±25% around the baseline value (in parentheses). Vertical dashed line: baseline cost ($1.01 kg⁻¹_urea). e Cost-reduction roadmap based on continuous tuning of key economic parameters. f Long-term electrolysis performance over 20 h in a 5*5 cm² flow-type electrolyzer using CO₂-saturated 1 M KHCO₃ + 0.1 M KNO₃ electrolyte; the inset of (f) shows the isolated urea product and the corresponding ¹H NMR and ¹³C NMR. Electrolyte pH: 7.8 ± 0.4. Source data are provided as a Source data file.

To provide guidelines for future development, univariate sensitivity analysis (Fig. 6d) identified key cost drivers as cell voltage, electricity price, FE, equipment, and labor cost, while water and maintenance costs had negligible impact. Specifically, reducing electrolyzer voltage to 1.5 V (e.g., via alternative anodic small molecule oxidation reactions) could lower costs by US$0.30 kg⁻¹. Decreasing electricity costs (expansion of renewable energy) to US$0.01 kWh⁻¹ and increasing FE to 85% (catalyst optimization based on the hierarchical click-site strategy) could further lower the cost by US$0.28 and US$0.16 kg⁻¹ (Supplementary Figs. 90–93), respectively. Collectively, these improvements could decrease urea production costs from US$1.01 kg⁻¹ to US$0.24 kg⁻¹ even without considering by-product revenues. Finally, the scalability of Se−InO_x was validated in 20 h continuous operation within a 5 × 5 cm² flow electrolyzer under CO₂-saturated 1 M KHCO₃ + 0.1 M KNO₃ electrolyte, maintaining stable current density (Fig. 6f and Supplementary Fig. 94). The solid product, isolated by concentration, benzene extraction, and subsequent removal of benzene, yielded 1.102 g of pure urea solid, which was confirmed by ¹H and ¹³C NMR characterization (inset of Fig. 6f)²¹. Based on the mass of the isolated urea, the average urea yield rate and FE over 20 h were calculated as 183.70 mmol h⁻¹ g⁻¹ and 35.28%, respectively. Compared to the small-scale setup, the slightly reduced performance is attributed to increased ohmic resistance, mass transport limitations, and unavoidable losses during purification, challenges typical in scaled-up systems. Nonetheless, the Se–InO_x demonstrated robust activity and selectivity under scaled conditions, affirming its promise for practical applications. The life cycle assessment (LCA) shows that when powered by the average U.S. grid (0.12 kg_CO2 MJ⁻¹), the urea electrosynthesis route generates higher greenhouse gas (GHG) emissions (15.76 kg_CO2 kg⁻¹_urea) than the conventional Bosch−Meiser process (1.90 kg_CO2 kg⁻¹_urea), mainly due to the high electricity demand of the electrochemical cell (Supplementary Fig. 95 and Tables 16 and 17). Under low-carbon renewable electricity (0.003 kg_CO2 MJ⁻¹), the GHG intensity drops significantly to 0.63 kg_CO2 kg⁻¹_urea urea, which is lower than that of the conventional route under comparable renewable scenarios (1.81 kg_CO2 kg⁻¹_urea), highlighting the potential for emission reductions within a gate-to-gate boundary. In line with the conservative LCA framework adopted in this study, the results reflect process-level emissions at the plant gate and do not imply permanent CO₂ sequestration or net carbon removal. Nevertheless, the use of renewable electricity markedly enhances the environmental performance of the electrochemical pathway, supporting its promise for sustainable chemical manufacturing.

Discussion

In summary, we have developed a hierarchical click-site catalysis strategy via a sequential-chain coupling pathway to overcome the long-standing challenge of low selectivity in urea electrosynthesis from NO₃⁻ and CO₂. Mechanism studies demonstrate that Se−InO_x, constructed by uniformly dispersed Se on CO₂-inert InO_x surface, serves as the first click-site to selectively adsorb NO₃⁻ and stabilize it as *NO₂. This adsorbed *NO₂ together with the catalyst, then acts as the second click-site to sequential-chain coupling CO₂, forming the key *CO₂NO₂ intermediate, which undergoes rapid hydrogenation to urea. This hierarchical click-site design effectively resolves the spatiotemporal mismatch between N- and C- intermediates, achieving precise C − N coupling while effectively preventing cross-contamination and undesirable by-products. As a result, Se−InO_x achieves a urea yield rate of 254.95 mmol h⁻¹ g⁻¹, a FE of 78.61%, >85% N_urea−selectivity, and 100% C_urea−selectivity, outperforming most previously reported materials. TEA and urea powder production, together with gate-to-gate LCA results under renewable electricity scenarios, suggest the feasibility of scalable, cost-effective, and lower-emission deployment. This work provides a generalizable framework for achieving high selectivity in complex electrochemical transformations and presents a promising route toward sustainable nitrogen and carbon valorization.

Methods

Chemicals

Indium(III) chloride tetrahydrate (InCl₃·4H₂O, 97%), cadmium acetate dihydrate (Cd(Ac)₂·2H₂O, ≥98.0%), 2,5-thiophenedicarboxylic acid (C₆H₄O₄S, 2, 5-TDC, 99%), selenium powder (Se, ~100 mesh, ≥99.5%), thioacetamide (C₂H₅NS, ≥99.0%), potassium nitrate (KNO₃, ≥99.0%), isopropanol (C₃H₈O, 99.5%), diacetylmonoxime (C₄H₇NO₂, ≥99.0%), phosphoric acid (H₃PO₄, >85% in H₂O), sulfuric acid (H₂SO₄, 95%–98%), iron(III) chloride (FeCl₃, ≥99.99%), thiosemicarbazide (TSC, 99%), ethylenediaminetetraacetic acid disodium salt (EDTA, 99%), urease (~1 U/mg), potassium bicarbonate (KHCO₃, ≥99.95% trace metals basis, 99.7–100.5% dry basis), ammonium chloride (NH₄Cl, ≥99.5%), hydrochloric acid (HCl, 37%), para-(dimethylamino)benzaldehyde (C₉H₁₁NO, 99%), salicylic acid (C₇H₆O₃, ≥99.0%), sodum citrate dihydrate (C₆H₅Na₃O₇, ≥99.0%), sodium hypochlorite (NaClO, available chlorine 4.00–4.99%), sodium hydroxide (NaOH, 97%, powder), sodium nitroferricyanide(III) dihydrate (Na₂Fe(CN)₅NO·2H₂O, 99%), ethanol (CH₃CH₂OH) and Nafion 115 were purchased from Sigma-Aldrich (Singapore) and used without further purification unless otherwise noted. High-purity CO₂ (≥99.999%) and Argon (Ar, ≥99.999%) gases were purchased from Air Liquide (Singapore). All aqueous solutions were prepared with high-purity de-ionized water (DI-water, resistance 18 MΩ cm^-1).

Synthesis of InO_x

To synthesize InO_x, an In-based metal-organic framework (In-MOF) was constructed using 2, 5-TDC as the organic linker. In a representative procedure, 0.5 g of InCl₃·4H₂O and 0.5 g of Cd(Ac)₂·2H₂O were dissolved in 200 mL of ethanol and combined with 100 mL of an ethanolic solution containing 1 g of 2,5-TDC. The mixed solution was maintained at 60 °C for 12 h to allow for MOF crystallization. The solid product was collected by centrifugation, sequentially washed with ethanol and deionized water to remove residual reactants, and then dried under vacuum. The resulting In-MOF precursor was thermally treated at 800 °C for 4 h in a muffle furnace. Post-calcination, the material was washed with deionized water and lyophilized to obtain InO_x.

Synthesis of Se-InO_x

Selenium incorporation was achieved by positioning 0.3 mmol of Se powder upstream and 100 mg of InO_x downstream within an alumina crucible. The system was sealed in a tubular furnace under argon and heated to 400 °C at a heating rate of 20 °C min⁻¹ and kept warm for 2 h. This step was repeated three times to promote homogeneous Se dispersion across the surface. The obtained solid was purified via triple deionized water washing and drying. The final sample was labeled Se−InO_x−2.92 (abbreviated as Se−InO_x). Analogous procedures with adjusted selenium amounts (0.1 mmol and 1 mmol) yielded Se−InO_x−1.16 and Se−InO_x−6.77, respectively.

Synthesis of S-InO_x

Similarly, S−InO_x was synthesized by replacing Se powder with C₂H₅NS under the same experimental conditions.

Physical characterization

The morphology of the as-prepared samples was characterized using a FESEM (JEOL 7800 F), TEM. High-resolution imaging was conducted using a Titan G2 60-300 HRTEM equipped with spherical and chromatic aberration correctors, and a FEI Titan G3 20-80 for HAADF-STEM analysis. Elemental composition was determined by ICP-MS (Thermo XSERIES 2), EDS, and elemental mapping (150 ~ 230 Mx at 200 kV on STEM). Crystal structures were identified by XRD (Bruker D8 Advance, 9 kW, 40 kV, 40 mA) using Cu Kα radiation (λ = 1.5418 Å). XPS (Thermo ESCALAB 250XI, energy range: 0−1400 eV) was employed to evaluate surface chemical states. EPR spectral signals were collected using Bruker EMX plus-9./12 spectrometer operating at X-band (9.85 GHz). FTIR spectra were recorded using a Thermo Fisher Nicolet iS 50 spectrometers to detect organic functional groups. UV−vis absorption spectra were obtained with an Agilent Cary 60 spectrophotometer. Additionally, the ¹H NMR spectra were obtained using a superconducting-magnet NMR spectrometer (Bruker AVANCE III HD, 600 MHz).

In-situ ATR-SEIRAS characterization

In-situ ATR-SEIRAS measurements were carried out on a Nicolet™ iS50 FT-IR spectrometer equipped with a specially designed reaction cell and a mercury cadmium telluride (MCT) detector cooled with liquid nitrogen. A 20 mm Au-coated Si hemispherical prism (MTI Corporation) was used both as the catalyst support and the internal reflection element. A Pt foil, an Ag/AgCl electrode (4 M KCl), and a CO₂-saturated 0.1 M KNO₃ + 0.1 M KHCO₃ solution served as the counter electrode, reference electrode, and electrolyte, respectively. Catalyst ink was prepared following the electrochemical protocol and drop-cast on the prism at a loading of 1 mg cm⁻². In-situ ATR-FTIR spectra were recorded while scanning the working electrode potential from OCP to −0.6 V (vs. RHE) at a scan rate of 5 mV s⁻¹. During measurements, pure CO₂ gas (99.999% purity) was continuously flowed into the reaction chamber, and spectral changes were monitored in the range of 4000 − 1000 cm⁻¹.

In-situ EPR characterization^47,48

In-situ EPR measurements were conducted at room temperature using a continuous-wave (CW) Bruker EMX plus-9/12 spectrometer operating at X-band (9.85 GHz). A custom-designed quartz flat cell (Supplementary Fig. 96), comprising four ports, was employed to establish a three-electrode electrochemical configuration. The cell featured a 1 mm-thick flat window, optimized to reduce signal attenuation due to excessive solution thickness. The working electrode consisted of a flexible indium tin oxide (ITO) substrate (0.125 mm, ~10 Ω) loaded with catalyst (1 mg cm⁻²), precisely cut to align with the central window region. A platinum wire served as the counter electrode, while an Ag/AgCl electrode was used as the reference. The electrolyte—comprising 0.1 M KNO₃ and 0.1 M KHCO₃ supplemented with 10 mM DMPO—was purged with CO₂ for 20 min before use and continuously supplied during operation. EPR spectra were recorded using a 100 kHz modulation frequency and 10 dB microwave power, across a 200 G sweep range. Each scan lasted 60 s, and final spectra were averaged from two consecutive measurements for improved resolution. Spin-trapping experiments using 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) were additionally performed on a Bruker EMX Micro EPR spectrometer under identical conditions. Spectral simulations were carried out using the EasySpin software (v6.0.0) integrated in MATLAB R2020a⁴⁹.

The X-ray absorption fine structure (XAFS) and in-situ XAFS characterization

The X-ray absorption fine structure (XAFS) spectra of the samples were collected at BL11B beamline of Shanghai Synchrotron Radiation Facility (SSRF). The SSRF is a third-generation synchrotron radiation light source composed of a 3.5 GeV electron storage ring, a full energy booster, and a 150 MeV linear accelerator. The electron-beam current during testing is 200 mA. The BL11B beamline used a bending magnet as the photon source. The experiments were performed using the double-crystal monochromator (DCM) with a Si (311) crystal. The corresponding reference samples and the samples for ex-situ experiments were collected in transmission mode. The samples for in-situ experiments were collected in fluorescence mode using a Lytle detector.

In-situ XAFS measurement was conducted with a homemade H-type cell with Nafion 115 to separate the anode and the cathode. This cell contains three electrodes with a carbon foil coated by catalyst as the working electrode, Pt mesh as the counter electrode, and Ag/AgCl as the reference electrode, respectively. The catalyst was drop-cast on the carbon foil and allowed to dry. A small window was cut out on the cathode side and sealed with Kapton film to allow fluorescence signals to pass from the electrode to the detector. The electrolyte was obtained by bubbling CO₂ into the 0.1 M KNO₃ and 0.1 M KHCO₃ aqueous solution for 30 min before testing. The XAFS spectra were collected under open-circuit potential and applied potentials. The acquired XAFS data were processed according to the standard procedures using the ATHENA module of Demeter software packages. To obtain the quantitative structural parameters around central atoms, least-squares curve parameter fitting was performed using the ARTEMIS module of Demeter software packages.

The following EXAFS equation was used:

$$\chi (k)={\sum }_{j}\frac{{N}_{j}{S}_{0}^{2}{F}_{j}(k)}{k{R}_{j}^{2}}\cdot exp[-2{k}^{2}{\sigma }_{j}^{2}]\cdot exp\left[\frac{-2{R}_{j}}{\lambda (k)}\right]\cdot sin[2k{R}_{j}+{\phi }_{j}(k)]$$

(1)

Electrochemical characterizations

The catalyst ink was prepared by dispersing 5 mg of the as-synthesized material and 30 µL of 5 wt% Nafion solution into 1 mL of a mixed solvent (isopropanol: deionized water = 4:1, v/v), followed by ultrasonication for 30 min to ensure homogeneity. A 40 µL of the resulting ink was transferred onto a 1.0 × 1.0 cm² GDL, achieving a mass loading of 0.2 mg cm⁻². Urea electrosynthesis from NO₃⁻ and CO₂ was conducted within a three-electrode flow-type electrolytic cell connected to a CHI 760E electrochemical workstation under ambient laboratory conditions (25 ± 1 °C). The synthesized catalyst on GDL served as the working electrode, a GDL as the counter electrode, and a saturated Ag/AgCl/KCl electrode as the reference. A Nafion 115 membrane (thickness ~127 μm, 2 × 2 cm²) was employed as the separator. Prior to use, the Nafion 115 membrane was sequentially treated in 5 wt% H₂O₂ at 80 °C for 3 h, rinsed with deionized water at 80 °C for 1 h, soaked in 0.5 M H₂SO₄ at 80 °C for 2 h and again at 100 °C for 8 h, followed by additional rinsing in deionized water at 80 °C and thorough washing to ensure cleanliness. The electrolyte, consisting of KNO₃ and 0.1 M KHCO₃, was freshly prepared on the day of each experiment and used on the same day. The gas (CO₂ or Ar) and electrolyte (30 mL) were fed at rates of 80 mL and 40 min⁻¹, respectively. Before the electrochemical test, the electrolyte was thoroughly purged with Ar or CO₂ flow for a period of 30 min. During electrolysis, the catholyte and anolyte were directed into their respective chambers, separated by the Nafion membrane, and subsequently routed into individual reservoirs. CO₂ was introduced into the cathode chamber through PTFE tubing, facilitating its interaction with the catalyst surface. The flow of the catholyte and anolyte was maintained using a peristaltic pump. Gas-phase products were analyzed using gas chromatography after flow stabilization. Chronoamperometric measurements were performed at various potentials for 1200 s, while maintaining a constant Ar or CO₂ flow within the electrochemical system. All potentials were converted to the reversible hydrogen electrode (RHE) scale using the formula: E_RHE = E_Ag/AgCl + 0.0592 × pH + 0.197. The Ag/AgCl (saturated KCl) reference electrode was calibrated in H₂-saturated 0.5 M H₂SO₄ using a clean Pt wire as both working and counter electrodes. The open-circuit potential was recorded and taken as 0 V (vs. RHE) after stabilization. The resistance was measured by EIS at open circuit potential (100 kHz–0.1 Hz, 5 mV), giving an average value of 2.7 ± 0.8 Ω. Unless otherwise stated, all reported voltages are without iR compensation. The Tafel slopes were determined using the Tafel equation: η = b log J_urea + a, where η is potential (V), J_urea is partial current density of urea production (mA cm⁻²), and b is the corresponding Tafel slope (mV dec⁻¹). Catalyst durability was assessed over ten successive chronoamperometric cycles at −0.4 V (vs. RHE).

Large-scale electrolysis and urea isolation

Under identical conditions, electrolysis was carried out in 1 M KHCO₃ + 0.1 M KNO₃ electrolyte using a constant current method across a range of 2−100 mA cm⁻². To obtain high-purity urea, a scaled-up electrolysis was performed in a 5 × 5 cm² flow cell continuously for 20 h. The resulting electrolyte was concentrated and subjected to benzene extraction to isolate urea³. After solvent removal, the solid urea product was obtained and confirmed by both H and C NMR spectroscopy.

Determination of urea

Method 1-Diacetylmonoxime (DAMO) colorimetry¹⁷: Briefly, 1 mL of electrolyte was mixed with 2 mL of acid−ferric solution (comprising 100 mL concentrated phosphoric acid, 300 mL concentrated sulfuric acid, 600 mL deionized water, and 100 mg ferric chloride). Then, 1 mL of a DAMO-thiosemicarbazide (TSC) solution (5 g DAMO and 100 mg TSC in 1 L deionized water) was added. The mixture was heated to 100 °C for 20 min and cooled to room temperature. Absorbance was measured at 525 nm using a UV−vis spectrophotometer, and urea concentrations were calculated based on a standard calibration curve. Method 2-Urease enzymatic decomposition¹⁴: 0.5 mL of a urease solution (5 mg mL⁻¹, contained 0.1 g ethylenediaminetetraacetic acid disodium salt (EDTA) and 0.49 g K₂HPO₄ per 100 mL) was added to 4.5 mL of electrolyte, followed by incubation at 40 °C for 40 min. NH₃ concentrations were determined before and after decomposition using the indophenol blue method, and urea concentration was obtained by subtracting the pre-decomposition NH₃ content from the post-reaction result.

Determination of NH₃

The concentration of NH₃ produced in the electrolyte was determined using the indophenol blue method³⁹. After electrolysis, 2 mL of the electrolyte was sequentially mixed with: 2 mL of 1 M NaOH containing 5 wt% sodium citrate and 5 wt% salicylic acid as solution A, 1 mL of 0.05 M NaClO as solution B, and 0.2 mL of 1 wt% sodium nitroprusside (C₅FeN₆Na₂O) as solution C. The solution was kept in the dark at ambient temperature for 2 h, followed by UV−vis absorption measurement in the range from 550 to 750 nm.

Determination of NO₂ ⁻

NO₂⁻ in the electrolyte was quantified using the Griess reagent method¹⁴. A 200 μL aliquot of Griess reagent was added to 5 mL of electrolyte, heated to 100 °C for 1 min, and cooled to room temperature. The absorbance at 540 nm was recorded, and nitrite concentration was obtained by referencing a standard calibration curve.

Determination of other liquid-phase products

Liquid-phase products were analyzed by ¹H NMR spectroscopy using dimethyl sulfoxide (DMSO) as the internal standard. Dimethyl sulfoxide (DMSO) was used as an internal standard, allowing for accurate determination of the liquid product yield rate.

Determination of gas-phase products

Gas-phase products were analyzed using a gas chromatograph (GC) equipped with both a thermal conductivity detector (TCD) and a flame ionization detector (FID), with Ar used as the carrier gas.

Calculaions for yield rate and FE

The detailed yield rates have been calculated by using the following equation:

$$Y({{{\rm{urea}}}})=c({{{\rm{urea}}}})\times V/(t\times A)$$

(2)

$$Y({{{{\rm{NH}}}}}_{3})=c({{{{\rm{NH}}}}}_{3})\times V/(t\times A)$$

(3)

$$Y({{{{\rm{NO}}}}}_{{2}^{-}})=c({{{{\rm{NO}}}}}_{{2}^{-}})\times V/(t\times A)$$

(4)

$$Y({{{\rm{HCOOH}}}})=n({{{\rm{HCOOH}}}})/(t\times A)$$

(5)

The detailed FEs have been calculated as follows:

$${FE}({{{\rm{urea}}}})=16F\times c({{{\rm{urea}}}})\times V/(60.06\times Q)$$

(6)

$${FE}({{{{\rm{NH}}}}}_{3})=8F\times c({{{{\rm{NH}}}}}_{3})\times V/(17\times Q)$$

(7)

$${FE}({{{{\rm{NO}}}}}_{{2}^{-}})=2F\times c({{{{\rm{NO}}}}}_{{2}^{-}})\times V/(47\times Q)$$

(8)

$${FE}({{{\rm{HCOOH}}}})=2F\times n({{{\rm{HCOOH}}}})/Q$$

(9)

where F is the Faradic constant (96485.3 C mol⁻¹), c is concentration, V is the volume of electrolyte, t is the reaction time and A is the catalyst area and Q is the total charge passed through the working electrode.

Calculaions for selectivity

The C-selectivity for CO₂-to-urea and CO₂-to-HCOOH conversions was calculated as follows:

$$C{({{{\rm{C}}}}{\mbox{-}}{{{\rm{product}}}})}_{{{{\rm{selectivity}}}}}=n{({{{\rm{C}}}}{\mbox{-}}{{{\rm{product}}}})}_{{{{\rm{C}}}}}/n{({{{\rm{total}}}})}_{{{{\rm{C}}}}}\times 100\%$$

(10)

where n(C-product)_C is the number of moles of C in C-product, and n(total)_C is the total number of moles of C atoms in the products from CO₂RR.

Similarly, the N−selectivity for NO₃⁻-to-urea, NO₃⁻-to-NO₂^- and NO₃⁻-to-NH₃ conversion was calculated using the following equation:

$$N{({{{\rm{N}}}}{\mbox{-}}{{{\rm{product}}}})}_{{{{\rm{selectivity}}}}}=n{({{{\rm{N}}}}{\mbox{-}}{{{\rm{product}}}})}_{{{{\rm{N}}}}}/n{({{{\rm{total}}}})}_{{{{\rm{N}}}}}\times 100\%$$

(11)

where n(N-product)_N is the number of moles of nitrogen in as-produced urea, and n(total)_N is the total number of moles of N atoms in the products from NO₃RR

DFT calculation details

First-principles calculations based on spin-polarized density functional theory (DFT) were performed using the Vienna Ab initio Simulation Package (VASP)⁵⁰, with the exchange-correlation interactions were described by the Perdew-Burke-Ernzerhof (PBE) functional within the generalized gradient approximation (GGA)⁵¹. To model surface reactions, periodic slab models were constructed with a 15 Å vacuum layer along the non-periodic direction to eliminate spurious interactions between periodic layers. The plane-wave basis set was truncated at a cutoff energy of 450 eV, and Brillouin zone integration was sampled using a Monkhorst-Pack k-point grid of 2 × 2 × 1 for structural optimizations. Convergence thresholds were set to10⁻⁵ eV for total energy and 0.01 eV Å⁻¹ for atomic forces, ensuring high precision in geometry relaxations. The adsorption energy was calculated according to the Gibbs free energy difference between the product and reactant, and Gibbs free energy correction, including entropy and the zero-point energy correction, was taken into consideration at 298.15 K. For the intermediate species of H⁺, its Gibbs free energy was calculated as half of H₂. The averaged length of In-X (X=O, S, Se) and distortion index of polyhedron analyses were implemented to present the change of local structure. The distortion index (D) was calculated as the following equation:

$$D=\,\frac{1}{n} {\sum }_{i=1}^{n}\frac{\left|{l}_{i}\,-\,{l}_{{ave}}\right|}{{l}_{{ave}}}$$

(12)

where ${l}_{i}$ and ${l}_{{ave}}$ represent the distance from the central atom In to the ith coordinating atom (O, or S, or Se) and the average bond length of the polyhedron, respectively. Electron localization function⁵² and Bader charge methods were employed to obtain the distribution and transfer of electron. The crystal orbital Hamilton population (COHP)⁵³ calculations were performed to reveal the chemical bonding and antibonding mechanism. For the dissociation of H₂O, the climbing-image nudged elastic band (CI-NEB)⁵⁴ method was used to obtain the dissociation barrier of H₂O.

Techno-economic evaluation

A comprehensive TEA was performed to assess the feasibility of urea electrosynthesis via the co-reduction of CO₂ and NO₃⁻ under industrially relevant conditions. The assessment included energy consumption estimation based on Faraday’s law, process-scale current and voltage inputs, and material balances for all relevant species (urea, NH₃, HCOOH, CO₂, H₂O, and O₂). Capital costs were calculated based on DOE H₂A guidelines, including stack and balance-of-plant components, with depreciation modeled over a 20-year plant lifetime. Operating costs accounted for electricity, water usage, membrane separation, and labor. Sensitivity analyses were conducted to evaluate the influence of current density, FE_urea, cell voltage, and electricity price on financial net present value (FNPV) and urea production cost. Detailed parameters, assumptions, and numerical results are provided in the Supplementary Information.