Screening transition metal electrodes for achieving near 100% selectivity to urea via electroreduction of NO₃⁻ and CO₂ at 100 mA cm⁻² current density

Abstract

The current method to synthesize urea is highly energy-intensive and has a massive carbon footprint. The electrochemical synthesis of urea from NO₃⁻ and CO₂ is an attractive and sustainable approach, as renewable energy can be used to synthesize green urea under ambient conditions by utilizing waste NO₃⁻ and CO₂ from the air or flue gas. In this work, we conducted a thorough catalytic screening of various metal-based catalysts. When the Ag GDE was used as a working electrode, ∼100% urea faradaic efficiency and ∼−100 mA cm⁻² of urea current density were observed at −1.2 V vs. RHE. FTIR analysis further confirmed the formation of urea and the presence of *CO intermediates. Through DFT studies, excellent kinetics and selectivity toward urea on Ag were explained by a combination of the facile first and second C–N bond-formation steps and an endergonic (ΔG > 1.5 eV) formamide (HCONH₂) formation from *CONH₂ step.

---

Introduction

Urea is one of the major nitrogenous fertilizers (70%) and the primary source of nitrogen supply for plants.^1,2 Apart from fertilizers, urea is used for the synthesis of resins, such as melamine and urea-formaldehyde resin,³ and as a diesel exhaust fluid that is used in diesel engines for selective catalytic reduction of NOx.^4,5 Urea is industrially synthesized via the Bosch–Meiser process, which involves the conversion of CO₂ and NH₃ at a high pressure of 110 atm and elevated temperatures between 160 and 180 °C.⁶ NH₃, a primary raw material for urea production, is manufactured through the Haber–Bosch process at high temperatures (400–500 °C) and high pressures (100–200 atm) utilizing H₂ from the steam reforming process.⁷ Furthermore, 48% of the globally produced NH₃ is used for urea manufacturing, and the overall process of making urea is highly energy-intensive and has a massive carbon footprint.⁸ Decarbonizing urea manufacturing by synthesizing urea using renewable energy and readily available feedstocks via an electrochemical approach is highly desired. The electrochemical synthesis of urea from N₂ and CO₂ is very attractive, but the process is very challenging because of stable N≡N bonds (941 kJ mol⁻¹) and lower solubility of N₂ in an aqueous medium.^9,10 In contrast, N=O has lower bond dissociation energy (204 kJ mol⁻¹) in comparison with N≡N (941 kJ mol⁻¹),⁹ and hence, it is less challenging to activate NO₃⁻ in comparison to N₂.¹¹ Industrially, NO₃⁻ is manufactured from NH_3, and electrochemical conversion of NO₃⁻ into NH₃ would make sense only in the context of recycling waste NO₃⁻.^12,13 Synthesizing urea from NO₃⁻ and CO₂ has a major environmental benefit over utilizing NH₃ and CO₂ as the existing processes consume 21–29 billion gigajoules of energy globally per year and generate 0.7–2.3 tons of CO₂ emissions per ton equivalent of urea produced.¹⁴

NO₃⁻ is a major source of pollutants in agricultural run-off water, industrial processing plants, and ammunition waste.¹⁵ ANSOL (65% ammonium nitrate solution) is a major waste stream produced by the Holston Army Ammunition Plant at a rate of 10 million pounds per year. ANSOL is extremely hazardous and unsafe to store, and the utilization of ANSOL in value-added products is one of the problems faced by the Strategic Environmental Research and Development Program (SERDP) of the US Department of Defense. ANSOL is usually sold to mining industries, but there is an inconsistent demand, and the existing methods involve the thermal degradation of ANSOL, which is inefficient and not environmentally friendly. NH₃ can be recovered from ANSOL by stripping it after shifting the pH of the solution toward the alkaline side. This results in a large concentration NO₃⁻ stream that is environmentally hazardous and needs to be treated.

In this study, we focused on co-reducing NO₃⁻ and CO₂ electrochemically to synthesize urea. The direct electrochemical co-reduction of NO₃⁻ and CO₂ is attractive as it allows the production of green urea decentralized with lower capital costs. Also, this study would help us provide insights into electrochemical C–N coupling, which has not yet been explored extensively in the literature. Understanding electrochemical C–N coupling would enable the selective synthesis of chemicals such as urea, methyl amine, acetamide, and benzamide, which are used as precursors in the synthesis of several commercial drugs, such as analgesics, antiemetics, and antipsychotics. The objective of the current study was to achieve the selective electrochemical synthesis of urea from NO₃⁻ and CO₂.

Electrochemical CO₂ reduction reactions (CO₂RRs) have been thoroughly investigated in the literature.¹⁶ Cu is the only catalyst for producing C₂ products like C₂H₄.¹⁷ Ag,¹⁸ Au,¹⁹ and Zn²⁰ are prominent catalysts to produce CO. Electrochemical NO₃⁻ reduction to NH₃ (NRN) has been investigated on a wide variety of catalysts, including Cu,^21,22 Pd,²³ Fe,²⁴ Ti,¹⁵ and Co,^25,26 and among these, cobalt, in particular, has demonstrated good activity for the electrochemical synthesis of NH₃ from NO₃⁻ in alkaline media.^25,26 Moreover, Ru-based catalysts²⁷ have shown high efficiency for NH₃ synthesis from NO₃⁻, enabling the reaction to occur at much more positive potentials, around 0 V vs. RHE. While cobalt can effectively promote the conversion of nitrate to nitrite at relatively positive potentials, it is less efficient for deeper nitrate reduction, which requires potentials well below 0 V vs. RHE. Early works by Shibata et al.^28–32 reported the electrochemical reduction of NO₂⁻/NO₃⁻ and CO₂ on various transition metal-based gas-diffusion electrodes. Zn exhibited the highest urea current efficiency of 35% at −1.75 V vs. SHE from the electrochemical reduction of NO₃⁻ and CO₂. The authors also proposed from experimental evidence that an NH₃-like precursor formed from NO₂⁻ and CO-like precursors is essential for urea synthesis. However, their works did not clearly state several things, such as the experimental procedure or electrolyte composition, and they did not have a rigorous urea quantification procedure. This work, therefore, requires a thorough reinvestigation to get clearer insights. Since then, there have been various investigations in this field utilizing different approaches and methodologies to further understand urea synthesis^33–37 from NO₃⁻ and CO₂, as well as from NO₂⁻ and CO.³⁸ Feng et al.³⁹ reported a 12.2% urea faradaic efficiency (FE) at −1.1 V vs. RHE from the electrochemical reduction of NO₂⁻ and CO₂ on Te-doped Pd nanocrystals. Leverett et al.⁴⁰ used a single-atom Cu catalyst for the electrochemical reduction of NO₃⁻ and CO₂ to urea with a 28% urea FE at −0.9 V vs. RHE. Meng et al.⁴¹ synthesized ZnO porous nanosheets for the electrosynthesis of urea from NO₂⁻ and CO₂ with a urea FE of 23.26% at −0.79 V vs. RHE. Lv et al.⁴² reported a very high urea FE of 53.4% from the electrochemical reduction of NO₃⁻ and CO₂ on an In(OH)₃-based catalyst.⁴³ A detailed summary of the urea FEs and current densities obtained on some prominent catalysts reported in the literature is given in the ESI (Table S1).† A thorough catalyst screening to selectively synthesize urea is not available in the existing literature. In an alkaline medium, the cathodic, anodic, and overall reactions are:

Cathode: 2NO₃⁻ + CO₂ + 11H₂O + 16e⁻ → NH₂CONH₂ + 18OH⁻ E⁰ = 0.78 V vs. RHE

Anode: 16OH⁻ → 4O₂ + 8H₂O + 16e⁻ E⁰ = 1.23 V vs. RHE

Overall: 2NO₃⁻ + CO₂ + 3H₂O → NH₂CONH₂ + 4O₂ + 2OH⁻ E⁰ = 0.45 V vs. RHE

In this work, we report a near ∼100% selectivity for urea from the electrochemical reduction of NO₃⁻ and CO₂ on Ag catalysts. The rest of the article is organized following the work, in which: a detailed experimental catalyst screening was first performed on the prominent catalysts; electrochemical urea synthesis was performed on the Ag catalyst by varying the applied potential, and a mechanism is proposed; the effects of varying the concentrations of NO₃⁻ and CO₂ on the urea FE and urea current density were then studied; the stability of the Ag was assessed by performing an electrochemical urea synthesis reaction (USR) for a period of 9 h; Ag was characterized using XRD and XPS pre- and post-electrolysis followed by operando FTIR studies; and DFT calculations were done to understand the reaction pathways for urea formation on Ag catalysts. This work aimed to overcome the challenges in urea production by establishing an electrocatalytic system that could utilize a wide range of CO₂ and nitrate feedstocks to produce urea in a single, sustainable, and more energy-efficient process, as shown in Fig. S24.†

Experimental section

All the details regarding the materials and chemicals used, such as manufacturer, part number, and purity, are given in the ESI (Tables S2 and S3).†

Electrochemical experiments

All the electrochemical urea synthesis experiments were conducted in a custom-made 3D-printed flow cell (Fig. S3†). Pt was used as the counter electrode, Ag/AgCl/sat. KCl was used as the reference electrode, and 25% HNO₃ was used as the anolyte. The catholyte and anolyte were separated by a Nafion 117 membrane. The Nafion 117 membrane was pre-treated by soaking it in 3% H₂O₂ at 80 °C for 1 h, followed by soaking in deionized (DI) H₂O at 80 °C for 2 h, and then by 0.5 M H₂SO₄ at 80 °C for 1 h. The membrane was rinsed between each step using DI H₂O and the final treated membrane was stored in the DI H₂O. The anolyte volume used was 5 ml and it was kept stationary at the anolyte side. The catholyte volume used was 30 ml, and it flowed from a reservoir to the catholyte side using a peristaltic pump at a rate of 40 ml min⁻¹. The electrodes were polished before the experiments for 10 min using a fine polishing pad (Pike Technologies – Crystal Polishing Kit #162-400) with a ceria-based polishing compound (Pike Technologies – Polishing Compound #162-4014) followed by sonicating the electrodes in DI water for 10 min, after which they were dried in Ar.

For the catalyst screening, and for assessing the effect of the applied potential and the stability studies, a catholyte solution of 0.1 M KNO₃ and 0.1 M KHCO₃ was used. For assessing the effect of the concentration of NO₃⁻, the following solution concentrations were used: 1 M KNO₃ + 0.1 M KHCO₃, 0.5 M KNO₃ + 0.1 M KHCO₃, 0.1 M KNO₃ + 0.1 M KHCO₃, 0.01 M KNO₃ + 0.1 M KHCO₃, and 0.001 M KNO₃ + 0.1 M KHCO₃. The solution was sparged with CO₂ using a sparger for 15 min, such that the solution was equilibrated with CO₂. The pH of the solution was measured before and after electrolysis by using a pH probe. Gas products were collected during the experiment using a gas bag by flowing Ar at a rate of 30 sccm for a period of 30 min. The liquid products remained in the catholyte, and hence, the catholyte was collected in a vial post-electrolysis. The possible gas products were CO, CH₄, H₂, and N₂. The gas products were quantified using gas chromatography (GC) (SRI Multiple Gas Analyzer). The possible liquid products were NH₃, urea, NO₂⁻, and HCOOH. NH₃ and urea were quantified using colorimetric techniques, while NO₂⁻ was quantified by ion-exchange chromatography (IC) (Metrohm), and the rest of the products were quantified by high-performance liquid chromatography (HPLC) (Agilent 1200 HPLC).

The electrolysis was carried out for 1 h using a potentiostat (Biologic SP 300). Potentio-electrochemical impedance spectroscopy (PEIS) was performed before all the experiments to measure the electrolyte resistance between the Luggin capillary of the reference electrode and the surface of the cathode. PEIS was performed by setting a single sine wave mode scanned from 100 kHz to 30 Hz by measuring 10 points per decade, at an amplitude of 20 mA with 3 measures per frequency, and the scan was repeated once. The working electrode voltage was set to 0 V vs. open circuit, while the voltage range was set to −10 V to 10 V, and the current range was set to Auto. Also, 85% of the uncompensated IR drop was compensated through positive feedback using Biologic EC-Lab software, and the other 15% of the uncompensated IR drop was compensated manually during the calculations. LSV was performed at a scan rate of 5 mV s⁻¹ from 0 to −3 V vs. RHE, with an acquisition time of 0.05 s. The voltage range was set to −10 V to 10 V, and the current range was set to Auto. CA was performed with an acquisition time of 0.1 s, with the voltage range of −10 V to 10 V and the current range set to 1 A. For the stability studies, similar settings were used as that for CA but with an acquisition time of 1 s.

Electrochemical 48 h stability tests

The use of 25% HNO₃ in our 9 h test was intended to reduce the overpotential in the anodic chamber. We hypothesized that a pH difference between the chambers could cause substantial variation, potentially affecting long-term product stability. Therefore, for the 48 h stability experiment, we used the same electrolyte in both the anolyte and catholyte chambers to maintain consistent conditions and ensure reliable results. A Glass H-cell (Pike Technologies) was used to conduct the 48 h stability experiment. The cell was equipped with a silver planar electrode as the working electrode, with copper tape serving as the current collector. The electrolyte solution, consisting of 0.1 M KNO₃ and 0.1 M KHCO₃, was used as both the catholyte and anolyte. The two chambers of the H-cell were separated by an Excellion membrane. The experiment was conducted at ambient temperature, with a platinum electrode serving as the counter electrode. Also, an Ag/AgCl reference electrode was placed in the catholyte solution.

Colorimetric quantification of the products

NH₃ was quantified by the Indophenol blue method.⁴⁴ Here, to 1 ml of the electrolyte sample, 1 ml of KOH solution, 500 μL of phenol nitroprusside solution, and 500 μL of sodium hypochlorite solution were added and the resulting solution was incubated in the dark for half an hour. The sample changed color from colorless to blue. The sample was scanned for absorbance as a function of the wavelength from 400 to 800 nm using a visible spectrometer (Genesys 30 Visible Spectrometer). The maximum absorbance was observed at 632 nm, and hence 632 nm was chosen to measure the absorbances and quantify NH₃ in further experiments. Calibration graphs (absorbance vs. concentration of NH₃) were prepared for different concentrations of NH₃ in the electrolyte. Separate calibration graphs were prepared when the concentration of the electrolyte was changed, as the absorbance was sensitive to the pH of the solution. The NH₃ calibration graphs for different electrolyte compositions are provided in Fig. S4 of the ESI.†

Urea was quantified by the diacetylmonoxime method.⁴⁵ Here, to 1 ml of the electrolyte sample, 1 ml of acid-ferric solution and 2 ml of monoxime-carbazide solution were added. The resulting sample was heated at 100 °C with constant stirring for 5 min followed by cooling at room temperature for 5 min. The sample changed color from colorless to pink. The sample was scanned for absorbance as a function of the wavelength from 400 to 800 nm using a visible spectrometer. The maximum absorbance was observed at 525 nm, and hence, 525 nm was chosen to measure the absorbances and quantify urea in further experiments. Calibration graphs (absorbance vs. concentration of urea) were prepared for different concentrations of urea in the electrolyte. Separate calibration graphs were prepared when the concentration of the electrolyte was changed for improved accuracy, as it was observed that the absorbance was sensitive to the electrolyte solution. The urea calibration graphs for different electrolyte compositions are provided in Fig. S5 of the ESI.† In the presence of NH₃, formamide, methyl amine, and acetamide, the diacetylmonoxime method did not provide a color change, and it was selective for urea (Fig. S10†).

Possible sources of error in the colorimetric quantification of the products

The colorimetric quantification of urea using the diacetyl monoxime (DAMO) method is prone to several sources of error, including human and experimental errors. Matrix effects, such as the presence of complex electrolytes or high salt concentrations, can influence absorbance measurements through light scattering or interference with the chromogenic reaction, leading to inaccuracies. Additionally, small timing differences between absorbance measurements of the same solution can introduce variability.

To minimize these errors, we conducted each urea synthesis experiment in triplicate. After each electrochemical reaction, the post-electrolyte solution was collected, and the urea concentration was measured three times to account for any variability in absorbance readings. Recognizing the sensitivity of electrochemical experiments, each experiment was repeated three times to address potential errors arising from electrochemical factors. Furthermore, whenever the salt concentrations were altered, new calibration curves were generated to account for matrix effects and ensure accurate quantification.

One source of error we observed was the pH change in the post-electrolyte after an hour of reaction, which occurred due to the generation of OH⁻ ions. This resulted in an increase in pH, which could affect the colorimetric method and lead to slight errors in urea quantification, potentially causing overestimation. The pH change could also alter the absorbance measurements, which may affect the faradaic efficiency (FE) calculations.

ESI Fig. S4, F2, and E2† highlight this effect. Specifically, Fig. F2 and H2† compare 0.1 M KNO₃ + 1 M KHCO₃ with 0.1 M KNO₃ + 0.01 M KHCO₃. As shown in these graphs, the slopes and intercepts differed significantly when the bicarbonate concentrations changed, which in turn affected the pH and absorbance. While we have addressed most potential sources of error, this pH shift remains a contributing factor to the observed overestimation of the FE.

NMR quantification

For the ¹H NMR tests, dimethylsulfoxide-d6 (DMSO-d₆) was adopted as a deuterated reagent. First, 570 μL of extracted electrolyte without postprocessing was mixed well with 30 μL of 10 mM acetone prepared in DMSO-d₆. Then, the liquid was transferred into an NMR tube for the test. The measurements were carried out on a Bruker 500 MHz AVANCE NEO spectrometer equipped with a cryoprobe. The presented data are the accumulated results of 64 scans. The water resonance was suppressed with the excitation sculpting method using a 3 ms 180° shaped pulse centered at 4.612 ppm. The perfect-echo variant was chosen to reduce J-modulation for the samples analyzed at 500 MHz. A total of 1024 transient scans were recorded with an interscan delay of 1 s. Overall, 64 000 complex points were acquired for each free induction decay with an acquisition time of 3.4 s. The processed spectra were zero-filled to 64 000 real points, and an exponential apodization function with lb = 0.3 Hz was applied before the Fourier transformation. DMSO (3 vol%; 99.9%, Sigma-Aldrich) was added for deuterium locking and referencing. The results are shown in Fig. S15.†

FE calculations

Calibration curves for both the NMR and UV-vis techniques were generated and are provided in the ESI.† These calibration plots were used to determine the concentration and moles of urea present in the post-reaction electrolyte samples.

To calculate the urea faradaic efficiency (FE), the urea partial current density was first determined. The total moles of urea produced in the reaction were calculated from the calibration data. The partial current density for urea formation (in mA cm⁻²) was then calculated using the following equation:

where n is the number of electrons transferred during urea formation (16 electrons for urea, 6 electrons for ammonia), F is Faraday's constant (96 485C mol⁻¹), and t is the reaction time in seconds.

The total urea current density was then divided by the total applied current density to calculate the faradaic efficiency:

This method accounts for the total charge passed in the electrochemical reaction and allows for the calculation of the urea FE using both NMR and UV-vis quantification techniques.

Catalyst characterization

XPS was performed using a Kratos Axis-165 system to analyze the surface composition and the oxidation states of the Ag catalyst before and after electrolysis. Our samples required minimal preparation; specifically, we only attached them to the holder using carbon tape. We conducted surface analysis for the Ag planar electrode and Ag GDE, without performing depth profiling. Calibration was achieved using the carbon peak in the XPS, with a pass energy of approximately 200 eV and a pressure below 10⁻⁸ mbar. Carbon tape was used for the electrical contact. The instrument produced monochromatized Al Kα radiation at 12 kV and 10 mA. A survey scan was conducted between binding energies of 0 and 900 eV with a resolution of 1 eV. Following the survey scan, high-resolution scans were conducted between 560 and 620 eV to identify the Ag 3p peaks, between 360 and 380 eV to identify the Ag 3d peaks, and between 520 and 540 eV to identify the O 1 s peak with a resolution of 0.1 eV. To minimize the noise, 5 sweeps were performed for the high-resolution scans.

XRD was performed on the Ag electrode before and after electrolysis using a Bruker D8 Discover X-ray diffractometer using Cu Kα radiation (λ = 1.5418 Å) generated at 40 kV and 40 mA. Kβ coming from the Cu radiation was filtered using Ni filters. The diffractometer had parallel beam optics and a 0.5° parallel slit analyzer. On the primary side, Göbel mirror was used and on the detector side a LYNXEYE detector, which had 196 channels, each having a channel width of 14.4 mm. The detector slit used was 1.2 mm. A two-theta scan was performed to get the offset of the beam with the sample holder in place using a primary rotary absorbance value of 73.88, following which an external offset correction was made. The sample was placed on the sample holder, and a Z scan was performed to locate the sample edge with an auto primary rotary absorbance. The angular offset of the sample was found by performing a rocking scan with a primary rotary absorbance of 73.88, and flatness correction was made. Finally, two-theta/theta scans were coupled from 10° to 90° with a step size of 0.02° to obtain the XRD spectra. The primary rotary absorbance was set to auto mode. Postprocessing was performed using Diffrac Suite Eva software, and the background noise was subtracted. The data were scanned with the ICDD and the peaks were identified.

FTIR experiments were conducted similar to in previously established studies on nitrate reduction mechanisms.²⁵ The experiments were performed on a Bruker Invenio-S infrared spectrometer. A custom-made electrochemical cell was mounted on top of a 60°-face angled Ge crystal setup on a ATR VeeMax-III variable angle accessory (Pike Tech.). To enhance the metal wettability of the Ge crystal and the conductivity of the substrate, an IR transparent indium-tin-oxide (ITO) layer of 100 nm was sputter-coated over it, using an EMS Quorum 150TS plus sputter-coater. Silver (Ag) was sputter-coated on top of this ITO layer with a thickness of 2 nm. After subtracting the background of the base electrolyte, a potential of −0.1 vs. RHE was applied and the spectra were acquired at different time stamps with a resolution of 2 cm⁻¹ averaged over 10 scans. A liquid N₂-cooled mid-band mercury cadmium telluride (MCT) detector was used while conducting these measurements. The schematic of the setup used is provided in the ESI (Fig. S6).† For ex situ experiments, we used a ZnSe crystal mounted on a heating plate for our analysis. The temperature was set to 110 °C to evaporate all the water, and solid residues were crystallized on the ZnSe crystal, which was then used for analysis in the attenuated total reflectance (ATR) mode.

A Xenemetrix Ex-Calibur EX-2600 system was used for the XRF analysis. This instrument uses a Rh X-ray source, operated here at 20 keV and 10 μA. For the analysis, 200 μL of the solution was deposited onto a piece of filter paper and allowed to dry. Once completely dried, the filter paper was placed into the XRF instrument for analysis. A qualitative survey scan was performed using a standard XRF instrument with a rhodium (Rh) source.

DFT methods

All the density functional theory (DFT) calculations were performed using the Vienna Ab initio Simulation Package (VASP)^46–48 interfaced with the Atomic Simulation Environment (ASE).⁴⁹ DFT calculations in conjunction with the computational hydrogen electrode (CHE) model⁵⁰ were used to determine the intermediate adsorption energies of the urea and formamide formation pathways. Core electrons were described using PAW pseudopotentials,⁵¹ and valence electrons were expanded as planewaves up to a kinetic energy cutoff of 500 eV. The electron exchange and correlation interactions were accounted for using the revised Perdew–Burke–Ernzerhof (RPBE) exchange–correlation functional by Hammer and Nørskov.⁵² Solvation effects were incorporated using a continuum solvation model as implemented in VASP (VASPsol).^53,54 A (3 × 3 × 1) Monkhorst–Pack⁵⁵k-point mesh was used to sample the Brillouin zone. Adsorbate binding energies were calculated using a 3 × 3 × 3 supercell with the bottom two layers fixed and the top layer free to relax with the adsorbate. Geometries were considered to be optimized after the maximum force on each unconstrained atom fell below 0.05 eV Å⁻¹. The transition states for the two C–N bond-formation reactions were estimated using a constrained one-dimensional bond-length scan at constant potential, as described and benchmarked in our previous publications.^56,57 Briefly, the bond length was varied over a series of 25 images spanning the initial state (e.g., *CO + *NO) to the final state (e.g., *CONO). For each image, the C–N bond length was constrained, and the geometry was optimized at constant potential with respect to this constraint. The adsorption-free energies, ΔG, of the reaction intermediates, were calculated using the expression (ΔG_ads = ΔE_ads + ΔZPE − TΔS), where ΔE is the difference in electronic energies of the adsorbed species, ΔZPE is the difference in zero-point energies, and ΔS is the change in entropy of the adsorbed species with respect to the catalyst surface.⁵⁸ The ZPE and entropies (S) were calculated using the harmonic oscillator approximation, which assumes that adsorbed molecules vibrate harmonically and have only vibrational degrees of freedom. The reaction mechanisms for the various elementary steps and further details of the simulation cell can be found in the ESI (Fig. S5 and S6).† The highest energy image was then refined to a true transition state using the improved dimer method as implemented in VASP Transition State Tools.^59,60

Results and discussions

Catalysts screening

Different metallic catalysts were screened for the electrochemical synthesis of urea from NO₃⁻ and CO₂. The electrochemical reactions were performed in a custom-made flow cell, where a solution of 0.1 M KNO₃ and 0.1 M KHCO₃ equilibrated with CO₂ was used as the catholyte. Chronoamperometry was performed by applying a constant potential of −1 V vs. RHE for a period of 1 h. The detailed experimental procedure is given in the methods section. In total, 17 different catalysts were chosen for the study, including the prominent electrochemical CO₂ reduction catalysts,⁶¹ such as Cu, Zn, Ag, Au, Sn, In, Re, and Pb, and with catalysts active for electrochemical NO₃⁻ reduction⁶² to NH₃, such as Co, Ni, Fe, and Pd. The catholyte was tested for urea and NH₃ post-electrolysis using colorimetric methods. The detailed procedure for the quantification methods for urea and NH₃ is provided in the methods section. The catalyst that exhibited the highest urea FE and the highest urea current density (CD) was the preferred one.

Fig. 1 denotes the urea FE and urea CD for the different catalysts. Bi and Re did not show any activity for urea. Pt and Ir showed minimal activity for urea synthesis. Pt, Ir, and Re were good hydrogen evolution reaction (HER) catalysts.⁶³ The activity of Pt toward the HER was significantly suppressed in the presence of nitrates. For Re and Ir, the suppression of HER activity was less than that of Pt. Catalysts that exhibited good activity for electrochemical NO₃⁻ reduction to NH₃, such as Fe, Ni, and Pd, showed much less activity for urea. NH₃ was observed in significant amounts for these catalysts. Co was the best catalyst for electrochemical NO₃⁻ reduction to NH₃ (ref. 25), and it was also active for electrochemical CO₂ reduction to CO and HCOOH.⁶⁴ Co showed a good urea CD (∼−5 mA cm⁻²) but its urea FE was much less (∼30%), and NH₃ was the dominant product when Co was used. Cu was active for both electrochemical CO₂ reduction and electrochemical NO₃⁻ reduction to NH₃ and showed good activity for urea, with a 70% urea FE, but the urea current density (∼−4 mA cm⁻²) was less than that of Co.

Fig. 1 Catalysts screening: urea faradaic efficiency and current density for different catalysts at −1 V vs. RHE.

In, Zn, and Sn were found to be active for electrochemical CO₂ reduction to HCOOH and CO, and they showed higher urea FEs (>80% for In and Sn, and >60% for Zn). Ag and Au (prominent catalysts for electrochemical CO₂ reduction to CO) showed enhanced activities for urea synthesis with >95% urea FEs, with Ag exhibiting the highest urea current density (∼−8 mA cm⁻²) and ∼100% urea FE. Pb, which was the most active catalyst for electrochemical CO₂ to HCOOH, showed no activity toward urea. We hypothesize that the reaction intermediates for CO formation are also key intermediates for urea synthesis. Also, Pb did not show activity toward electrochemical CO formation, which strengthens our hypothesis. We believe that In, Sn, and Zn, which were active for urea synthesis, primarily facilitated CO₂ reduction with HCOOH as the dominant product. In contrast, Ag and Au, which also showed activity for urea, mainly produced CO during CO₂ reduction. The CO₂ reduction mechanisms differ between these two groups, leading us to hypothesize that C–N coupling may also proceed via distinct mechanisms. From the above study, we observed that Ag exhibited the highest urea FE and urea CD, and hence Ag was chosen for further evaluations to improve the catalytic activity and understand the urea formation mechanism. Understanding the mechanism of urea formation on catalysts that reduce CO₂ to HCOOH is beyond the scope of this study and is a potential future work.

Electrochemical measurements

Fig. 2A denotes the linear sweep voltammetry (LSV) profiles for different catholyte solutions when Ag was used as the cathode. Three electrolyte solutions were considered for the study, namely 0.1 M KHCO₃, 0.1 M KNO₃, and 0.1 M KHCO₃ + 0.1 M KNO₃. The first and the last solutions were equilibrated with CO₂ before the study, whereas 0.1 M KNO₃ was not equilibrated with CO₂. A detailed description of the potentiostat settings used to perform the LSV and other experiments in this section is given in the methods section. The possible faradaic reactions when 0.1 M KHCO₃ was used as the electrolyte were the CO₂ reduction reaction (CO₂RR) to CO and HER, NO₃⁻ reduction reaction (NORR) to NO₂⁻, NH₃, and N₂ when 0.1 M KNO₃ was used as the electrolyte, and the urea synthesis reaction (USR) along with the HER, CO₂RR to CO, and NORR when a solution of 0.1 M KHCO₃ and 0.1 M KNO₃ was used as the electrolyte. The slope of the LSV profile was steeper when the solution of 0.1 M KHCO₃ and 0.1 M KNO₃ was used in comparison to the individual electrolyte solutions, indicating the possible USR. The experimentally measured onset potential for the NORR when 0.1 M KNO₃ was used was 0.04 V and for the CO₂RR when 0.1 M KHCO₃ was used was −0.42 V. For the solution containing both 0.1 M KHCO₃ and 0.1 M KNO₃, the onset potential was measured to be −0.22 V. A zoomed-in figure denoting the onset potentials is provided in the ESI (Fig. S1).†

Fig. 2 Electrochemical measurements: all experiments are performed at room temperature with a 1 cm² electrode area. In all the cells, the flowrate of the electrolyte was maintained at 30 ml min⁻¹. (A) Linear sweep voltammetry (LSV) profiles for the Ag cathode using different catholytes, such as 0.1 M KHCO₃ (equilibrated with CO₂), 0.1 M KNO₃ and a solution of 0.1 M KHCO₃ and 0.1 M KNO₃ (equilibrated with CO₂). (B) Urea faradaic efficiency and current density as a function of applied potential using an Ag cathode and a solution of 0.1 M KNO₃ and 0.1 M KHCO₃ equilibrated with CO₂ as the catholyte for a run time of 1 hour. (Yellow symbols indicate urea current density in all the figures). (C) Urea, NH₃, and CO faradaic efficiencies and urea current density as a function of the concentration of NO₃⁻ after the 1 hour electrochemical experiment. The solution contained 0.1 M KHCO₃ and the concentration of NO₃⁻ was varied as 1, 0.5, 0.1, 0.01, and 0.001 M, which is equilibrated with CO₂. (D) Urea faradaic efficiency and current density as a function of time showing the stability of Ag for electrochemical urea synthesis. The solution was not constantly sparged with CO₂. (E) Urea faradaic efficiency and current density as a function of applied potential using an Ag GDE and a solution of 1 M KNO₃ and 0.1 M KHCO₃ equilibrated with CO₂ as the catholyte, with an experimental run time of 30 min.

The effect of the applied potential on the electrochemical USR was studied to understand the potential dependence on the selectivity of urea. A solution of 0.1 M KNO₃ and 0.1 M KHCO₃ equilibrated with CO₂ was chosen as the electrolyte, and the applied potential was varied from −0.6 to −1.5 V vs. RHE. Fig. 2B denotes the urea FE and urea current density as a function of the applied potential. As the applied potential increased in the negative direction, the urea CD increased linearly. For all the applied potentials, the urea FE remained close to 100%. This indicated that the Ag catalyst was very selective for the electrochemical USR for the electrolyte concentration of 0.1 M KNO₃ and 0.1 M KHCO₃. The selectivities of the electrochemical CO₂RR products, such as CO, CH₃OH, and C₂H₄, were potentially driven, and the selectivities dropped drastically when the applied potential was changed, even by as little as 0.2 V vs. RHE, due to the competing HER.¹⁷ It has been reported in the literature that the HER is drastically suppressed even in the presence of small amounts of NO_x.⁶⁵ The presence of concentrated amounts of nitrates in our system also suppressed the HER and hence the urea selectivity remained constant (∼100%) in the studied potential range. At higher overpotentials (electrochemical urea synthesis), the urea FE dropped significantly, and the NH₃ FE increased due to the overreduction of NO₃⁻. The urea and NH₃ performances at high overpotentials are provided in Fig. S2 of the ESI.†

The effect of the concentration of NO₃⁻ on the selectivity of urea was investigated. The concentration of bicarbonates was kept constant in the electrolyte (0.1 M KHCO₃) and the concentration of NO₃⁻ (1, 0.5, 0.1, 0.01, and 0.001 M KNO₃) was varied. For all the above cases, the electrolyte was equilibrated with CO_2, and chronoamperometry was performed at −1 V vs. RHE. Fig. 2C denotes the FEs of the products and urea current densities as the concentration of NO₃⁻ was changed by fixing the concentration of bicarbonate. As the concentration of NO₃⁻ increased, the urea current density increased as a function of the concentration of NO₃⁻. At lower concentrations of NO₃⁻, CO was observed to a great extent. As the concentration of NO₃⁻ increased, NH₃ was observed along with a decreasing concentration of CO, indicating that the NORR was preferred over the CO₂RR. Beyond 0.1 M NO₃⁻, only urea was observed, and other by-products, such as CO, H₂, and NH₃, were not observed. This indicates that the concentration of NO₃⁻ is a key parameter in deciding the selectivity of urea apart from the applied potential.

The stability of the Ag toward the electrochemical USR was studied for a period of 9 h by performing chronoamperometry at −1 V vs. RHE using a solution of 0.1 M KNO₃ and 0.1 M KHCO₃ equilibrated with CO₂ as the catholyte. The products were sampled every 1 h, and Fig. 2D denotes the urea FE and urea CD as a function of time. A constant urea FE of ∼100% was observed, while the urea CD remained constant throughout the study period, indicating that Ag was stable for the electrochemical USR. Additionally, we performed a long-term stability test with a 48 h experiment. Given the extended duration, we replaced the anolyte with the same electrolyte used in the catholyte (0.1 M KNO₃ + 0.1 M KHCO₃), instead of HNO₃, to prevent the migration of H⁺ from the anolyte over time. The performance remained consistent, with the system continuously producing urea at a faradaic efficiency of nearly 90%, as shown in Fig. S18.† So far, all the analyses were performed on planar Ag. To improve the urea CD, 10 nm Ag was sputter-coated on carbon paper, which acted as a gas-diffusion electrode (GDE), and the reaction was carried out using 1 M KNO₃ and 0.1 M KHCO₃ equilibrated with CO₂. The urea CD improved drastically in comparison with planar Ag, although an order of magnitude improvement was not observed as the NO₃⁻ was still in the liquid phase, and only the CO₂ concentration was improved in the gas phase. Fig. 2E denotes the effect of the applied potential when the Ag GDE was used as the electrode. At −1.25 V vs. RHE, a ∼100% urea FE and ∼−100 mA cm⁻² urea current density were observed, which are the highest so far reported in the literature.

The urea quantification results obtained through the UV-vis analysis were validated using ¹H NMR spectroscopy. The ¹H NMR spectrum of the post-electrolysis sample displayed a peak at a chemical shift of 5.5 ppm, corresponding to urea, with an estimated FE of approximately 98%. To benchmark the NMR technique against UV-vis spectrometry, a parity plot was generated, showing a slope close to 1, indicating a strong correlation between the two methods (as illustrated in Fig. S15D†). Additionally, to confirm that the urea originated from the nitrate rather than contamination, an isotope-labeled nitrate (¹⁵KNO₃) was used in the electrolyte for urea synthesis with a Ag planar electrode, yielding an FE of 88%. The isotope peaks in the ¹H NMR spectrum appeared at 5.17 and 5.47 ppm, with a coupling constant of 180 Hz, further validating the formation of urea in our electrochemical setup. The detailed NMR spectra and calibration graphs are provided in the ESI.† We also performed an extended stability test for a period of 48 h, as shown in Fig. S18.†

Next, X-ray diffraction (XRD) spectroscopy and X-ray photoelectron spectroscopy (XPS)⁶⁶ were performed on a Ag catalyst pre and post-electrolysis to understand the facets present on the Ag and its oxidation states. A detailed description of the experimental methods used to perform the XRD and XPS is given in the methods section.

Fig. 3A denotes the XRD spectrum for the Ag catalyst pre-electrolysis. Peaks were observed at 2θ locations corresponding to 38.11°, 44.23°, 64.42°, 77.32°, and 81.55°. The spectra matched identically with metallic Ag as per the International Centre for Diffraction Data (ICDD No. 04-0783). The prominent facet observed on the Ag used for the electrochemical USR was the 111 facet. Fig. 3B denotes the XRD spectrum for the Ag catalyst post-electrolysis. A slight noise could be observed in the data, but the peaks corresponding to the 2θ locations matched with the metallic Ag, indicating that the facets were preserved, and there was no structural reorganization of the catalyst post-reaction.

Fig. 3C denotes the spectrum obtained from XPS performed on the Ag pre-electrolysis. Two prominent peaks could be observed at binding energies corresponding to 374. 5 and 368.5 eV. These peak locations corresponded to Ag 3d_3/2 and Ag 3d_5/2. The location of Ag 3d_5/2 at 368.5 eV and the difference in binding energies between Ag 3d_5/2 and Ag 3d_5/2 (6 eV) indicated that Ag was in its metallic state.⁶⁷ A negligible shift (∼0.12 eV) in the Ag 3d_5/2 binding energy peak was observed for the Ag post-electrolysis, as denoted by Fig. 3D, and the catalyst still remained in its metallic state, indicating that there was no change in the oxidation state of the catalyst after performing 1 h electrochemical USR. No significant change was observed on the Ag GDE before and after electrolysis based on the scanning electron microscopy (SEM) images (Fig. S11†). XRF analysis was also performed to determine if Ag leaching occurred during extended periods of cell operation, as shown in Fig. S23.† The results indicated that most the observed peaks could be attributed to the Rh source and the sample holder, including prominent Rh-Kα and Rh-Kβ lines. However, no silver (Ag) peaks were identified in the qualitative survey scan, despite the analysis specifically targeting Ag detection. This suggests that no Ag leaching occurred under the reductive conditions in the electrolyte during the 48 h stability test. Following the characterization of the Ag planar electrode, the Ag gas-diffusion electrode (Ag GDE) was also analyzed using XPS. As illustrated in Fig. S26 and S27,† the silver in the Ag GDE remained in its metallic state after electrolysis. Additionally, Auger spectra for both pre-and post-electrolysis were collected, revealing the presence of AgM₄N₄₅N₄₅ and AgM₅N₄₅N₄₅ in both conditions, as shown in Fig. S28.†⁶⁶

Fig. 3 Catalyst characterization: X-ray diffraction (XRD) spectra for planar Ag: (A) pre-electrolysis, and (B) post-electrolysis. X-ray photoelectron spectra (XPS) for planar Ag: (C) pre-electrolysis, and (D) post-electrolysis.

In situ FTIR studies

The electrochemical co-reduction of CO₂ and NO₃⁻ was performed on a Ag electrode at −1 V vs. RHE, and operando attenuated total reflectance surface-enhanced infrared absorption spectra were obtained as a function of time. A mixed solution of 0.1 M KNO₃ and 0.1 M KHCO₃ equilibrated with CO₂ was used as the electrolyte. Background subtraction was performed using the blank electrolyte before acquiring the FTIR spectra. The details of the experimental procedure are provided in the methods section. Several peaks appeared corresponding to different functional groups, and Fig. 4A denotes the obtained FTIR spectra. The peaks were identified by considering the FTIR database and from the Sadler Handbook of infrared spectra. Further, ex situ experiments were conducted on the reactants and products to better analyze the spectra and identify the peaks (Fig. S7† shows the FTIR spectra of the reactants and products). Further details on the obtained spectra are provided in the ESI Section.†

Fig. 4 Attenuated total reflectance surface-enhanced infrared absorption spectra (ATR-SEIRAS): (A) in situ FTIR spectra obtained using ATR-SEIRAS for the electrochemical co-reduction of NO₃⁻ and CO₂ on planar Ag at −1 V vs. RHE at 1 min, 4 min, 8 min, and 16 min. (B) C–N stretching at 1487 cm⁻¹. (C) C=O stretching at 1640 cm ⁻¹. (D) C–O stretching at 1942 cm⁻¹. (E) N–H stretching at 3384 cm⁻¹.

The most prominent peak appeared at a peak location corresponding to 1487 cm⁻¹ in the C–N stretching region, as shown in Fig. 4B. As there was no significant increase in peak intensities in the C–H stretching region, we could rule out the possibilities of methyl amine and formamide. Hence, the C–N stretching peak could be attributed to urea formation. There was an increase in the intensity of the C=O stretching bands corresponding to urea between wavenumbers 1500 and 1750 cm⁻¹, as shown Fig. 4C. Both these signature peaks indicated the formation of urea. The *CO adsorption peak was observed in the in situ FTIR analysis at around ∼1942 cm⁻¹ during the electrochemical urea synthesis reaction, as shown in Fig. 4D. A similar observation has been reported in the literature.² The intensity of the peak increased as a function of time and dropped down denoting that CO was a key intermediate in the electrochemical urea synthesis reaction. The intensity of the peaks between wavenumbers 3500 and 3100 cm⁻¹ increased as a function of time, corresponding to the N–H stretching of urea, as shown in Fig. 4E.

The in situ electrochemical analysis provided strong evidence for the formation of urea (in addition to the colorimetric results), and CO* intermediates were observed. The in situ data were consistent with the DFT calculations performed for urea formation on the Ag electrode, which is discussed in the following section.

DFT investigation of electrocatalytic C–N coupling

DFT calculations were performed to investigate the pathways toward the electrochemical synthesis of urea and formamide on Ag (100), (111), and (110), beginning with the two reacting species CO₂ and NO₃⁻. A schematic of one feasible proposed pathway on the low-index facets of Ag is given in Fig. 5.

Fig. 5 One of the several feasible pathways for the electrochemical synthesis of urea via the co-reduction of CO₂ and NO₃⁻ over an Ag catalyst.

To arrive at this proposed pathway, we calculated the free energy of adsorption of each species at potentials between 0 V and −1.5 V vs. RHE, as shown in Fig. 6A. As in previous works,^19,25 the adsorption of NO₃⁻ anions is considered via the reaction given in eqn (1):

NO₃⁻(aq.) + H₂O + e⁻ + * → *NO₂ + 2OH⁻ (1)

Fig. 6 DFT results: (A) adsorption energies of NO₃⁻ and CO₂. (B) C–N bond-formation barrier on Ag (100). (C) Free energy diagram for CO₂ and nitrate co-reduction at 0 V vs. RHE.

We remark here that the initial adsorption of nitrate is poorly understood; the reaction NO₃⁻ + * → *NO₃ + e⁻ is an oxidation reaction that will be unfavorable under reducing conditions. One may also consider the reductive adsorption of nitric acid, HNO₃. Given the dissociation equilibrium coefficient for nitric acid (∼0.4 eV),^19,68 the availability of this species is likely vanishingly small (∼1 in 10⁷ based on a simple Boltzmann distribution) in an aqueous electrolyte. However, NO₃⁻ reduction to NH₃ is well documented in the literature, and so we assume a concerted two proton-electron transfer event occurs, leading to adsorbed nitrite (*NO₂). We found this step to be very exergonic, even close to 0 V vs. RHE, suggesting the spontaneous formation of nitrite under the reaction conditions. The adsorption of CO₂ was similarly assumed to occur via a concerted coupled proton-electron transfer step, as in eqn (2), as reported in previous works:^69–71

CO₂(g) + H₂O + e⁻ + * → *COOH + OH⁻ (2)

The reaction free energy was calculated to be positive (≈+1.26 eV) at 0 V vs. RHE, indicating that CO₂ adsorption is unfavorable at moderately reducing potentials, consistent with reports from previous studies suggesting that CO₂ adsorption is rate-limiting during electrochemical CO₂ reduction on Ag.^70,71 On Ag (100), the free energy of the coupled CO₂ and protonation will be more favorable as the potential becomes more reducing, becoming spontaneous at −1.26 V vs. RHE based on the computational hydrogen electrode model.⁵⁰ This value coincides with experimental reports at −1.25 V vs. RHE, at which a ∼100% urea FE and ∼−100 mA cm⁻² urea current density were observed. We note that CO₂ adsorption will occur at less reducing potentials on more reactive, high-index facets, possibly explaining the experimental observation of the urea current at less reducing potentials. Also, as shown in eqn (3), a concerted coupled proton-electron transfer step followed by an Eley–Rideal-like C–N bond coupling step can also occur between adsorbed nitrogen species (*NO) and CO₂ to produce *NOCOOH. The reaction free energy for this step was calculated to be +0.37 eV at 0 V vs. RHE, becoming spontaneous at −0.37 V vs. RHE on Ag (100). This value was closer to −0.22 V vs. RHE, which was the experimentally observed onset potential for urea synthesis

*NO + CO₂(g) + H₂O + e⁻ → *NOCOOH + OH⁻ (3)

Based on our prior analysis finding that NO₃⁻ reduction on weaker nitrogen binding catalysts (such as Ag) is limited by the protonation of *NO to form *NOH²⁵ and that the main product of CO₂ reduction on Ag is CO,^70,72 we propose that the first C–N bond formation for the co-reduction of NO₃⁻ and CO₂ will occur between adsorbed CO and NO (*CO + *NO → *CONO). As shown in Fig. 6B, DFT was used to calculate the first C–N bond-formation barrier on Ag at a constant potential (see methods section for further details). The barrier calculations for the coupling reactions of NO with CO₂, *COOH, and *CO on other low-index facets of Ag are given in the ESI.† We report the barrierless coupling of *CO and *NO on Ag (100), and very low barriers on other low-index facets of Ag. The coupling of *NO with *COOH similarly exhibited very low barriers, effectively instantaneous on the timescale of the reaction turnover, suggesting that the dominant coupling mechanism will be driven by the steady-state availability of the reactants, rather than being limited by an activation barrier. In the absence of a full microkinetic model of the process, it is not possible to conclusively comment on the predominant mechanism of the first C–N bond-formation step, only that it occurs without a significant activation barrier. Given prior reports of *CO being the predominant surface species for CO₂ reduction on Ag,^72,73 in addition to our previously published analysis of nitrate reduction on transition metals,⁷⁴ we hypothesize that the coupling mechanism *CO and *NO may be the predominant pathway for the first C–N bond-formation step.

To probe the reaction pathway beyond the first C–N bond-formation step, all possible reaction pathways were investigated, and the free energy of the various intermediates was calculated (see Fig. S8 and S9 in the ESI† for further details) under the reaction conditions, with the free energy diagram for the most thermodynamically favorable pathway shown in Fig. 6C.

Our calculations reveal that the second C–N bond formation may occur after a series of protonation steps (*CONH₂ + *NO → *CONH₂NO, ΔG < 0). We found that the additional protonation of adsorbed *CONH₂ to produce HCONH₂ (formamide) was thermodynamically unfavorable (ΔG > 1.5 eV at 0 V vs. RHE), consistent with the lack of formamide detected by our experimental efforts. The second C–N bond-formation barrier for this reaction on Ag (100) was also calculated. Our results suggest that the second C–N bond-formation barrier, which we found to be 0.56–0.62 eV, was likely not rate-determining given the reported current density toward urea. The free energy diagram, shown at 0 V vs. RHE, illustrated that the energies of all the elementary steps in this reaction were exergonic, except for the protonation of adsorbed CON (*CON + H⁺ + e⁻ → *CONH). However, as this step is a reduction step, it becomes more favorable at a more cathodic potential. From our DFT results, we hypothesize that urea synthesis from the co-reduction of CO₂ and nitrate on Ag is limited by either (i) CO₂ adsorption on the catalyst surface, or (ii) the mass transport of reactants from the bulk electrolyte to the reaction plane. The urea current density was drastically improved when the GDE configuration is used, which supports this hypothesis.

Summary and conclusions

Electrochemical C–N coupling to synthesize urea from NO₃⁻ and CO₂ has not yet been thoroughly investigated in the literature, and so in this work, we performed a thorough catalyst screening and found that Ag is selective, active, and stable for the electrochemical synthesis of urea. CO₂ transport to the Ag surface and adsorption are the limiting factors, as in the case of the electrochemical CO₂RR. It was demonstrated that the GDE configuration can overcome the limitations and a very high urea current density of ∼−100 mA cm⁻² and a urea faradaic efficiency of ∼100% were observed at −1.25 V vs. RHE, which are the highest so far reported in the literature. The concentration of NO₃⁻ plays a key role in the selectivity, and a minimum concentration of 0.1 M KNO₃ was required for ∼100% selectivity, while at lower concentrations of NO₃⁻, the CO₂RR becomes dominant instead of the USR. Our DFT calculations revealed that on Ag (100), a combination of facile first and second C–N bond-formation steps and an endergonic formamide formation step from *CONH₂ results in its improved kinetics and selectivity toward urea synthesis. We believe that the current work could potentially lead to further studies on electrochemical C–N coupling for obtaining high-value-added chemicals.

Data availability

The data supporting this article have been included as part of the ESI.† Correspondence and requests for materials should be addressed to Meenesh R. Singh or Joseph A. Gauthier.

Author contributions

N. C. K. and I. G. prepared and characterized the catalysts under the direction of M. R. S. The electrocatalytic experiments were performed by N. C. K. and I. G. guided by M. R. S. The analysis of liquid products was performed by N. C. K. and I. G., guided by M. R. S. NMR experiments were performed by S. C. under the guidance of K. G. FTIR was performed by R. R. B. under the supervision of M. R. S. Theoretical calculations were performed by S. A. O., directed by J. A. G. The manuscript was composed by N. C. K., I. G., S. A. O., R. R. B., J. A. G., and M. R. S. M. R. S. devised the project, conceptualized the main ideas, and worked on almost all the technical aspects.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

This material is based on work performed in the Materials and Systems Engineering Laboratory at the University of Illinois Chicago (UIC) and Gauthier's Lab at Texas Tech University in collaboration with Glusac's Lab (UIC). M. R. S. acknowledges funding from UIC. J. A. G. acknowledges support from the National Science Foundation (NSF) Engineering Research Center for Advancing Sustainable and Distributed Fertilizer Production (CASFER) under Grant No. 2133576. S. A. O. and J. A. G. gratefully acknowledge support from The Welch Foundation under Grant Number D-2188-20240404.

References

1. P. M. Glibert, J. Harrison, C. Heil and S. Seitzinger, Biogeochemistry, 2006, 77, 441–463.
2. D. Milani, A. Kiani, N. Haque, S. Giddey and P. Feron, Sustain. Chem. Clim. Action., 2022, 100008.
3. A. H. Conner, Polymeric Materials Encyclopedia, 1996, p. 11.
4. S. Dammalapati, P. Aghalayam and N. Kaisare, Ind. Eng. Chem. Res., 2019, 58, 20247–20258.
5. N. Balaji, P. Aghalayam and N. S. Kaisare, Ind. Eng. Chem. Res., 2018, 57, 6853–6862.
6. B. Carl and M. Wilhelm, Process of manufacturing urea, US Pat., BASF SE, US1429483, 1922.
7. L. Wang, M. Xia, H. Wang, K. Huang, C. Qian, C. T. Maravelias and G. A. Ozin, Joule, 2018, 2, 1055–1074.
8. S. Giddey, S. Badwal, C. Munnings and M. Dolan, ACS Sustain. Chem. Eng., 2017, 5, 10231–10239.
9. G.-F. Chen, Y. Yuan, H. Jiang, S.-Y. Ren, L.-X. Ding, L. Ma, T. Wu, J. Lu and H. Wang, Nat. Energy, 2020, 5, 605–613.
10. N. C. Kani, A. Prajapati, B. A. Collins, J. D. Goodpaster and M. R. Singh, ACS Catal., 2020, 10, 14592–14603.
11. N. C. Kani, A. Prajapati and M. R. Singh, ACS ES&T Eng., 2022, 2, 1080–1087.
12. N. C. Kani, N. H. L. Nguyen, K. Markel, R. R. Bhawnani, B. Shindel, K. Sharma, S. Kim, V. P. Dravid, V. Berry and J. A. Gauthier, Adv. Energy Mater., 2023, 13, 2204236.
13. L. Barrera and R. Bala Chandran, ACS Sustain. Chem. Eng., 2021, 9, 3688–3701.
14. M. Xia, C. Mao, A. Gu, A. A. Tountas, C. Qiu, T. E. Wood, Y. F. Li, U. Ulmer, Y. Xu and C. J. Viasus, Angew. Chem., Int. Ed., 2022, 61, e202110158.
15. J. M. McEnaney, S. J. Blair, A. C. Nielander, J. A. Schwalbe, D. M. Koshy, M. Cargnello and T. F. Jaramillo, ACS Sustain. Chem. Eng., 2020, 8, 2672–2681.
16. S. Nitopi, E. Bertheussen, S. B. Scott, X. Liu, A. K. Engstfeld, S. Horch, B. Seger, I. E. Stephens, K. Chan and C. Hahn, Chem. Rev., 2019, 119, 7610–7672.
17. A. Prajapati, N. C. Kani, J. A. Gauthier, R. Sartape, J. Xie, I. Bessa, M. T. Galante, S. L. Leung, M. H. Andrade and R. T. Somich, Cell Rep. Phys. Sci., 2022, 3, 101053.
18. E. L. Clark, S. Ringe, M. Tang, A. Walton, C. Hahn, T. F. Jaramillo, K. Chan and A. T. Bell, ACS Catal., 2019, 9, 4006–4014.
19. J.-X. Liu, D. Richards, N. Singh and B. R. Goldsmith, ACS Catal., 2019, 9, 7052–7064.
20. B. Qin, Y. Li, H. Fu, H. Wang, S. Chen, Z. Liu and F. Peng, ACS Appl. Mater. Interfaces, 2018, 10, 20530–20539.
21. Y. Wang, W. Zhou, R. Jia, Y. Yu and B. Zhang, Angew. Chem., Int. Ed., 2020, 59, 5350–5354.
22. L. Barrera, R. Silcox, K. Giammalvo, E. Brower, E. Isip and R. Bala Chandran, ACS Catal., 2023, 13, 4178–4192.
23. J. Lim, C.-Y. Liu, J. Park, Y.-H. Liu, T. P. Senftle, S. W. Lee and M. C. Hatzell, ACS Catal., 2021, 11, 7568–7577.
24. P. Li, Z. Jin, Z. Fang and G. Yu, Energy Environ. Sci., 2021, 14, 3522–3531.
25. N. C. Kani, J. A. Gauthier, A. Prajapati, J. Edgington, I. Bordawekar, W. Shields, M. Shields, L. C. Seitz, A. R. Singh and M. R. Singh, Energy Environ. Sci., 2021, 14(12), 6349–6359.
26. X. Deng, Y. Yang, L. Wang, X. Z. Fu and J. L. Luo, Advanced Science, 2021, 8, 2004523.
27. S. Han, H. Li, T. Li, F. Chen, R. Yang, Y. Yu and B. Zhang, Nat. Catal., 2023, 6, 402–414.
28. M. Shibata and N. Furuya, Electrochim. Acta, 2003, 48, 3953–3958.
29. M. Shibata, K. Yoshida and N. Furuya, J. Electrochem. Soc., 1998, 145, 2348.
30. M. Shibata, K. Yoshida and N. Furuya, J. Electrochem. Soc., 1998, 145, 595.
31. M. Shibata and N. Furuya, J. Electroanal. Chem., 2001, 507, 177–184.
32. M. Shibata, K. Yoshida and N. Furuya, J. Electroanal. Chem., 1998, 442, 67–72.
33. Y. Luo, K. Xie, P. Ou, C. Lavallais, T. Peng, Z. Chen, Z. Zhang, N. Wang, X.-Y. Li, I. Grigioni, B. Liu, D. Sinton, J. B. Dunn and E. H. Sargent, Nat. Catal., 2023, 6, 939–948.
34. X. Huang, Y. Li, S. Xie, Q. Zhao, B. Zhang, Z. Zhang, H. Sheng and J. Zhao, Angew. Chem., Int. Ed., 2024, e202403980.
35. J. Geng, S. Ji, M. Jin, C. Zhang, M. Xu, G. Wang, C. Liang and H. Zhang, Angew. Chem., Int. Ed., 2023, 62, e202210958.
36. M. Jiang, M. Zhu, M. Wang, Y. He, X. Luo, C. Wu, L. Zhang and Z. Jin, ACS Nano, 2023, 17, 3209–3224.
37. Y. Zhao, Y. Ding, W. Li, C. Liu, Y. Li, Z. Zhao, Y. Shan, F. Li, L. Sun and F. Li, Nat. Commun., 2023, 14, 4491.
38. H. Wan, X. Wang, L. Tan, M. Filippi, P. Strasser, J. Rossmeisl and A. Bagger, ACS Catal., 2023, 13, 1926–1933.
39. Y. Feng, H. Yang, Y. Zhang, X. Huang, L. Li, T. Cheng and Q. Shao, Nano Lett., 2020, 20, 8282–8289.
40. J. Leverett, T. Tran-Phu, J. A. Yuwono, P. Kumar, C. Kim, Q. Zhai, C. Han, J. Qu, J. Cairney and A. N. Simonov, Adv. Energy Mater., 2022, 12, 2201500.
41. N. Meng, Y. Huang, Y. Liu, Y. Yu and B. Zhang, Cell Rep. Phys. Sci., 2021, 2, 100378.
42. C. Lv, L. Zhong, H. Liu, Z. Fang, C. Yan, M. Chen, Y. Kong, C. Lee, D. Liu and S. Li, Nat Sustainability, 2021, 4, 868–876.
43. L. Zhengyi, Z. Peng, Z. Min, J. Hao, L. Hu, L. Shengqi, Z. Heng, Y. Song and Z. Zehui, Appl. Catal., B, 2023, 338, 122962.
44. S. Z. Andersen, V. Čolić, S. Yang, J. A. Schwalbe, A. C. Nielander, J. M. McEnaney, K. Enemark-Rasmussen, J. G. Baker, A. R. Singh and B. A. Rohr, Nature, 2019, 570, 504–508.
45. C. Chen, X. Zhu, X. Wen, Y. Zhou, L. Zhou, H. Li, L. Tao, Q. Li, S. Du and T. Liu, Nat. Chem., 2020, 12, 717–724.
46. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169.
47. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
48. G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1993, 47, 558.
49. A. H. Larsen, J. J. Mortensen, J. Blomqvist, I. E. Castelli, R. Christensen, M. Dułak, J. Friis, M. N. Groves, B. Hammer and C. Hargus, J. Phys.: Condens. Matter, 2017, 29, 273002.
50. J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard and H. Jonsson, J. Phys. Chem. B, 2004, 108, 17886–17892.
51. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758.
52. B. Hammer, L. B. Hansen and J. K. Nørskov, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 7413.
53. K. Mathew, R. Sundararaman, K. Letchworth-Weaver, T. Arias and R. G. Hennig, J. Chem. Phys., 2014, 140, 084106.
54. K. Mathew, V. C. Kolluru, S. Mula, S. N. Steinmann and R. G. Hennig, J. Chem. Phys., 2019, 151, 234101.
55. H. J. Monkhorst and J. D. Pack, Phys. Rev. B: Condens. Matter Mater. Phys., 1976, 13, 5188.
56. J. A. Gauthier, Z. Lin, M. Head-Gordon and A. T. Bell, ACS Energy Lett., 2022, 7, 1679–1686.
57. J. A. Gauthier, J. H. Stenlid, F. Abild-Pedersen, M. Head-Gordon and A. T. Bell, ACS Energy Lett., 2021, 6, 3252–3260.
58. J. A. Zamora Zeledón, M. B. Stevens, G. K. K. Gunasooriya, A. Gallo, A. T. Landers, M. E. Kreider, C. Hahn, J. K. Nørskov and T. F. Jaramillo, Nat. Commun., 2021, 12, 620.
59. A. Heyden, A. T. Bell and F. J. Keil, J. Chem. Phys., 2005, 123, 224101.
60. G. Henkelman and H. Jónsson, J. Chem. Phys., 1999, 111, 7010–7022.
61. Y. i. Hori, Modern Aspects of Electrochemistry, 2008, pp. 89–189.
62. X. Fu, J. Zhang and Y. Kang, Chem Catal., 2022, 2(10), 2590–2613.
63. J. Greeley, T. F. Jaramillo, J. Bonde, I. Chorkendorff and J. K. Nørskov, Nat. Mater., 2006, 5, 909–913.
64. M. Usman, M. Humayun, M. D. Garba, L. Ullah, Z. Zeb, A. Helal, M. H. Suliman, B. Y. Alfaifi, N. Iqbal and M. Abdinejad, Nanomaterials, 2021, 11, 2029.
65. B. H. Ko, B. Hasa, H. Shin, E. Jeng, S. Overa, W. Chen and F. Jiao, Nat. Commun., 2020, 11, 1–9.
66. A. M. Ferraria, A. P. Carapeto and A. M. Botelho do Rego, Vacuum, 2012, 86, 1988–1991.
67. K. D. Bomben, J. F. Moulder, P. E. Sobol and W. F. Stickle, Handbook of X-ray photoelectron spectroscopy, Perkin-Elmer Corporation, 1992.
68. F. Calle-Vallejo, M. Huang, J. B. Henry, M. T. Koper and A. S. Bandarenka, Phys. Chem. Chem. Phys., 2013, 15, 3196–3202.
69. H.-C. Mi, C. Yi, M.-R. Gao, M. Yu, S. Liu and J.-L. Luo, ACS Appl. Mater. Interfaces, 2022, 14, 43257–43264.
70. J. Hussain, H. Jónsson and E. Skúlason, ACS Catal., 2018, 8, 5240–5249.
71. W. Deng, P. Zhang, B. Seger and J. Gong, Nat. Commun., 2022, 13, 803.
72. M. R. Singh, J. D. Goodpaster, A. Z. Weber, M. Head-Gordon and A. T. Bell, Proc. Natl. Acad. Sci. U. S. A., 2017, 114, E8812–E8821.
73. J. Hussain, H. Jonsson and E. Skúlason, ACS Catal., 2018, 8, 5240–5249.
74. N. C. Kani, J. A. Gauthier, A. Prajapati, J. Edgington, I. Bordawekar, W. Shields, M. Shields, L. C. Seitz, A. R. Singh and M. R. Singh, Energy Environ. Sci., 2021, 14, 6349–6359.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4se00841c

‡ Authors contributed equally.

---