Recent progress in electrochemical synthesis of urea through C-N coupling reactions

NO₂^- as nitrogen sources

In 1995, Shibata et al. pioneered the electrochemical synthesis of urea under standard atmosphere and ambient temperature and briefly mentioned the mechanism of urea electrosynthesis^[20]. In their opinions, the simultaneous existence of activated CO (i.e., adsorbed CO) and ammonia intermediate was a necessary condition for the electrocatalytic synthesis of urea. They predicted that the active substances could come from CO₂ and simple ammonia compounds such as NO₃^-/NO₂^-, respectively. However, it had been proved by comparative experiments that substituting NH₃ for NO₂^- or CO for CO₂ could not achieve the electrosynthesis of urea. So it could be concluded that urea was produced while NO₂^- and CO₂ were reduced at the same time; i.e., the NO₂^- was reduced to an ammonia intermediate, which further reacted with adsorbed CO.

Up to now, it is not difficult to conclude from previous reports that the mechanism of urea electrosynthesis mostly depends on the conversion of CO₂ to ^*CO and the conversion of NO₂^- to ^*NH₂ at catalytic sites. Especially in urea electrosynthesis with NO₂^- as raw material, it is a general rule that the two important intermediates (^*CO and ^*NH₂) can form the NH₂CONH₂ structure by C-N coupling [Figure 1A].

Figure 1. The reaction pathway for the electrosynthesis of urea by co-reduction of NO₂^- and CO₂. (A) The pathway for the C-N coupling of ^*NH₂ and ^*CO₂ intermediates; (B) The pathway for the C-N coupling of ^*NH₂ and ^*COOH intermediates.

As shown in the work of Cao et al., the co-reduction of CO₂ and NO₂^- was realized on Cu-TiO₂ catalysts with oxygen-rich vacancies^[21]. The ^*CO was produced by electrocatalytic reduction of CO₂ at the low-valence Cu sites; meanwhile, NO₂^- was converted to ^*NH₂ precursor by a reduction reaction at the active sites of bi-Ti³⁺. Subsequently, the C-N coupling of the two precursors originates ^*NH₂CONH₂ species, which desorbed to form urea.

The overall reactions are expressed as follows:

In order to unravel the reaction mechanism, Feng et al. carried out control experiments via CO₂-temperature-programmed desorption (TPD) and electrochemical CO-stripping measurement [Figure 2A-C], and the density functional theory (DFT) theoretical calculation was also implemented^[22]. The results showed that CO₂ and NO₂^- were reduced to ^*CO and ^*NH₂, and coupled to form ^*CONH₂, which was further converted to urea over the Te-Pd nanocrystals (NCs)/C catalyst [Figure 2D].

Figure 2. (A) CO₂-TPD for Pd NCs and Te-Pd NCs; CO-stripping measurements for (B) Te-Pd NCs/C and (C) Pd NCs/C (0.1 M KHCO₃ solution, the scanning rate is 20 mV·s^-1, the first cycle has a CO absorbed adlayer and the second cycle does not. Dose CO for 30 min before measuring); (D) Schematic diagram for urea electrosynthesis from CO₂RR and NO₂^-RR at Te-Pd NCs. Reprinted with permission from ref^[22]. Copyright 2020 American Chemical Society; (E) In situ ATR-FTIR spectra for ZnO-V under CO₂, NaNO₂, and both; (F) Schema of urea synthesis on ZnO-V. Reprinted with permission from ref^[23]. Copyright 2021 The Authors. Cell Reports Physical Science published by Elsevier. TPD: Temperature-programmed desorption; NCs: nanocrystals; CO₂RR: reduction reaction of CO₂; ATR-FTIR: attenuated total reflection Flourier transformed infrared.

It is worth noting that in the oxygen vacancy-rich ZnO system reported by Meng et al., the abundant oxygen vacancies (OVs) significantly reduced the interfacial charge transfer resistance of the catalyst^[23,24]. This report used in situ attenuated total reflection Flourier transformed infrared (ATR-FTIR) to characterize the reaction process of CO₂ with NO₂^-. As shown in Figure 2E, under the CO₂ alone atmosphere, the signals detected at 1,360 and 1,210 cm^-1 belonged to ^*COOH, and its peak increased with decreasing potentials. In the presence of NO₂^- alone, the signals belonging to NH₃ and NO₂^- were detected at 1,100 and 1,220 cm^-1, respectively. The NO₂^- signal exhibited a rising peak due to its depletion. In the case of co-catalyzed CO₂ and NO₂^-, the ^*COOH signal at 1,210 cm^-1 was offset by the NO₂^- signal at 1,220 cm^-1, and no ^*COOH signal peak was observed at 1,360 cm^-1, suggesting that ^*COOH was consumed during urea synthesis. Moreover, the C–N bond signal was detected at 1,440 cm^-1. The ATR-FTIR results could guide the possible reaction pathways: the O atoms in NO₂^- filled the surface OVs, and then formed ^*NH₂ intermediates through multistep proton-coupled electron transfer (PCET) process [Figure 2E and F]; the CO₂ entered the vacancies was transformed into ^*COOH intermediates; urea was synthesized via subsequent coupling of ^*NH₂ and ^*COOH intermediates [Figure 1B]^[25,26].

NO as nitrogen sources

As mentioned previously, ^*NO is a key intermediate for NO₃^-/NO₂^- electroreduction. Therefore, Huang et al. proposed a urea production strategy, i.e., direct reduction of NO and CO₂ through the prepared Zn nanobelts (Zn NBs) catalysts^[27]. Online differential electrochemical mass spectroscopy (DEMS) measurement was used to detect the intermediates and products in urea synthesis reactions, and the signals attributed to NH₂OH, HNO, NH₃, and CO/N₂ were periodically detected [Figure 3A]. In this case, the signals of NH₃ and N₂ came from the by-products of the reaction process. The ATR-FTIR was also performed to capture the absorbed intermediates [Figure 3B]. When NO and CO₂ were co-reduced, a vibrational band appeared at 1,465 cm^-1, which belonged to the C–N bond.

Figure 3. (A) Online DEMS measurements and (B) In situ ATR-FTIR spectra on Zn NBs; (C and D) Free energy changes to form (C) The first C–N bond and (D) The second on the surface of Zn(101); (E) Scheme of urea synthesis mechanism on the Zn surface with NO and CO₂ as raw materials. Reprinted with permission from ref^[27]. Copyright 2022 American Chemical Society. DEMS: Differential electrochemical mass spectroscopy; ATR-FTIR: attenuated total reflection Flourier transformed infrared; Zn NBs: Zn nanobelts.

The DFT calculations revealed that NO electrochemical reduction followed the nitrogen-alternating pathway^[28]: ^*NO → ^*NHO → ^*NHOH (^*HNOH, equally) → ^*NH₂OH → ^*NH₂ → ^*NH₃, while CO₂-to-CO conversion involved ^*COOH and ^*CO intermediates. The specific reaction pathway could be explained as follows: in the process of conversion from CO₂ to CO, the existence of absorbed NO was capable of further stabilizing ^*COOH and ^*CO intermediates. However, the binding of ^*COOH to nitrogen-related intermediates (^*NO, ^*NHO, ^*NHOH, ^*NH₂OH, and ^*NH₂) was energy unstable. The energy required to form ^*CO-NH₂ was -0.46 eV, and the energy barrier of ^*CONH₂ formation had a moderate Ea value (+0.85 eV), which was kinetically favorable for the formation of C–N bonds [Figure 3C]. Notably, the desorption of ^*CO was the potential-determining step (PDS). The second C-N coupling tended to be achieved by the coupling of ^*CONH₂ and ^*NH₂ according to theoretical calculations [Figure 3D]. Thus, it was deduced that the urea electrosynthesis reaction pathway for co-reduction of NO and CO₂ was realized via the step coupling of ^*CO and ^*NH₂ [Figure 3E].

NO₃^- as nitrogen sources

Although most studies have shown that the C-N coupling process follows the law of the conversions of CO₂ to ^*CO and NO₂^- to ^*NH₂, compared to NO₂^-, the urea electrochemical synthesis with NO₃^- as the nitrogen source shows obvious distinct mechanisms in different catalytic systems. The reaction pathway for the electrosynthesis of urea by co-reduction of NO₃^- and CO₂ is displayed in Figure 4.

Figure 4. The reaction pathway for the electrosynthesis of urea by co-reduction of NO₃^- and CO₂.

In the experiment reported by Lv et al., on the {100} facets of In(OH)₃, the NO₃^- was thermodynamically reduced to ^*NO₂ intermediates, while the protonation process of CO₂ required an additional 0.38eV of energy^[29]. On the two coordinated unsaturated In atoms, the electron was transferred to ^*NO₂. The local “In-O-C-O-In” configuration supported the transfers of electrons to ^*CO₂. The reaction between the two intermediates of ^*NO₂ and ^*CO₂ achieved an early direct C-N coupling. This conclusion was confirmed by a contrast electrocatalysis test. The lower energy barrier formed by ^*CO₂NO₂ (0.35 eV) contributed to the selective production of urea compared to that for protonating ^*NO₂ to ^*HNO₂ (0.62 eV). Further protonation tended to form ^*CO₂NH₂ intermediates; the rate-determining step of urea electrosynthesis occurred when ^*CO₂NH₂ intermediates were protonated to ^*COOHNH₂, which required an increase in free energy of 1.58 eV. Subsequently, the thermodynamically favorable second step of C-N coupling and urea generation occurred. In another report about the indium oxyhydroxide (InOOH) electrocatalyst with defect engineering created by OVs, the authors used advanced operando synchrotron radiation-Fourier transform infrared spectroscopy (SR-FTIR) to further verify the synthesis mechanism of urea and reached the same conclusion as above [Figure 5A and B]^[30]. The pathway of catalytic reaction and C-N coupling is shown in Figure 4A. It is noteworthy that the reaction pathway of the Zn/Cu hybrid catalyst reported by Luo et al. with excellent Faradaic efficiency (FE) of 75% is also consistent with Figure 4A^[31]. They utilized in situ surface-enhanced Raman spectroscopy (SERS) to observe the intermediates on the catalyst during urea synthesis. Under the conditions of electrocatalytic urea synthesis, the signals located at 334~337 cm^-1 belonging to M-OCONH₂, i.e., the characteristic signal of ^*CO₂NH₂ intermediates, were observed in both the electrolyte (1 M KHCO₃ + 1 M KNO₃) and the reference solution (ammonium carbamate, NH₂COONH₄), while they were not detected in the KHCO₃ solution. It confirmed the presence of ^*CO₂NH₂ intermediates during the catalytic process. For single-component Cu catalyst (electrolyte: 1 M KHCO₃ + 1 M KNO₃), these signals were also not detected, indicating that Zn was crucial for the formation of the ^*CO₂NH₂ intermediate. The in situ infrared reflection-absorption spectroscopy (IRRAS) confirmed the critical step in the process where ^*CO₂NH₂ protonation to ^*COOHNH₂ occurred.

Figure 5. (A) Free energy of urea synthesis on the {010} facets at Vo-InOOH and original InOOH; (B) Infrared signals (800 to 1,800 cm^-1) obtained from operando SR-FTIR at potentials of -0.30 to -1.00 V (vs. RHE) on Vo-InOOH during the C-N coupling. Reprinted with permission from ref^[30]. Copyright 2022 American Chemical Society; (C and D) Free energy of reduction (C) ^*NO₃ to ^*NH₂, (D) CO₂ to CO; (E) Free energy of C-N coupling on the surface of Cu@Zn, Zn, and Cu. Reprinted with permission from ref^[32]. Copyright 2022 American Chemical Society. InOOH: Indium oxyhydroxide; SR-FTIR: synchrotron radiation-Fourier transform infrared spectroscopy; RHE: reversible hydrogen electrode.

For works on the electrocatalytic synthesis of urea reported by other groups, the core-shell structure Cu@Zn catalyst designed by Meng et al. was conducive to improving the generation of ^*NH₂ and ^*CO intermediates and accelerating the combination of ^*NH₂ and ^*CO to produce urea [Figure 4D]^[32]. Nevertheless, the DFT results showed that the transition from ^*NO₃ to ^*HNO₃ and the formation of ^*COOH in the reduction of CO₂ to ^*CO were potential limiting steps [Figure 5C-E]. Theoretical calculations by Leverett et al. also showed that the formation of ^*COOH intermediates at the early stage was the rate-determining step in urea production^[33].

Zhang et al. carried out operando SR-FTIR measurement and DFT calculation for bonded Fe-Ni diatomic sites catalyst (B-FeNi-DASC) [Figure 6A-C]^[34]. The theoretical calculation results showed that the applied potential required for driving the CO formation was +0.32 eV at Fe-Ni bimetallic sites, which is much lower than the free energies for the generation of the key intermediate ^*COOH for the monometallic atom catalysts (up to +1.88 eV for Fe-N₄ and +2.31 eV for Ni-N₄), and thus, CO₂ was preferentially reduced to ^*CO intermediate. NO₃^- was reduced to ^*NO, but the direct coupling of ^*CO and ^*NO was disadvantageous in energy (+0.81 eV). So the C-N coupling should follow another path. The ^*NO intermediate might be reduced to ^*NOH or ^*HNO. The formation of ^*HNO (+0.62 eV) was more favorable to thermodynamics than that of ^*NOH (+0.90 eV). Subsequently, ^*NOH was further dissociated from ^*NH by two protonations. At the Fe-Ni bonding site, a thermodynamically spontaneous and dynamically feasible coupling occurred between ^*NH and ^*CO with a corresponding energy barrier of +0.21 eV, which was further supported by operando SR-FTIR results. The vibration intensity of the infrared bands near 1,694 cm^-1 increases with the increase of applied negative potential, and reaches its highest value at around -1.50 V, indicating the formation of the intermediate species ^*NHCO. The next NO would subsequently attach to the Ni site and promptly bind to ^*NHCO, transforming into the important intermediate ^*NHCONO, followed by continuous PCET processes that realize the formation of urea [Figure 4C].

Figure 6. (A) 3D operando SR-FTIR spectra for B-FeNi-DASC; (B) Infrared signal (range 1,500 to 3,750 cm^-1) at different potentials for B-FeNi-DASC during C-N coupling; (C) Free energy for B-FeNi-DASC of urea synthesis. Reprinted with permission from ref^[34]. Copyright 2022 Nature Publishing Group; (D) Geometric structure and free energy of C-N coupling process at Cu₄-CeO₂ (H, C, N, O, Cu, and Ce atoms are labeled as white, gray, blue, red, pink, and yellow balls in the geometric structure, respectively); (E) Operando SR-FTIR spectroscopy at different potentials for Cu₁-CeO₂ during the C-N coupling. Reprinted with permission from ref^[35]. Copyright 2023 Wiley-VCH GmbH. 3D: Three-dimensional; SR-FTIR: synchrotron radiation-Fourier transform infrared spectroscopy; B-FeNi-DASC: bonded Fe-Ni diatomic sites catalyst.

In the latest report, Wei et al. modified copper monoatoms on the CeO₂ vector (Cu₁-CeO₂) to catalyze the synthesis of urea^[35]. DFT calculation results indicated that the ^*OCNO intermediate was formed by the Eley-Raideal (ER) mechanism. For the purpose of synthesizing urea, a second ^*NO needs to be introduced into the system to combine with ^*OCNO [Figure 4B]. The ^*ONCONO was formed exothermic and subsequently reduced to urea by eight electron-proton-transfer steps. The forming of ^*HONCONO was the potential limiting step that required free energy of +0.58 eV [Figure 6D]. The cutting-edge operando SR-FTIR measurements and controlled electrolysis experiments showed that urea was formed from key intermediates ^*OCNO [Figure 6E].

N₂ as nitrogen sources

It is noted that in the case of complex catalyst systems, the C-N coupling process with N₂ as nitrogen sources reaches agreement on key intermediates, focusing on the formation of intermediate ^*NCON. Subsequent hydrogenation processes may have two reaction pathways, depending on the alternative and distal mechanisms[Figure 7].

Figure 7. The reaction pathway for the electrosynthesis of urea by co-reduction of N₂ and CO₂.

On the surface of the Ni₃(BO₃)₂ electrocatalyst with Lewis pairs [frustrated Lewis pairs (FLPs)] designed by Yuan et al., the involved hydroxyl, and adjacent Ni sites served as Lewis base (LB) and acid (LA), respectively^[36]. FLPs could achieve efficient adsorption of N₂ and CO₂ by unique orbital interaction^[37]. The occupied π orbitals of CO₂ contributed electrons to the d empty orbitals on the LA sites while the σ^* empty orbitals of CO₂ received electrons from the LB centers, thereby achieving the activation of CO₂. The interaction between occupied σ orbital on N₂ and the d empty orbital on the LA Ni sites would weaken the N≡N bond, in other words, depleting a large number of electrons in the σ state of N₂ [Figure 8A]. The “donation-acceptance” course at the FLP sites would effectually polarize the reactant molecules, stretching the chemical bond until it eventually broke^[38]. Subsequently, through “σ orbital electron transfer”, the spontaneous C-N coupling of the ^*CO with ^*N=N^* intermediates generated the ^*NCON^* intermediate. The conversion of ^*NCON^* intermediate to ^*NCONH required a free energy of 0.76 eV, which was considered the PDS of the whole reaction. Yuan et al. used the Mott-Schottky Bi-BiVO₄ heterostructure in another experiment^[39]. First, spontaneous charge transferred at the interface of the heterogeneous structure promoted the formation of a space charge region in which N₂/CO₂ was directionally adsorbed via electrostatic interaction^[40]. Then, the ^*CO and ^*N=N^* formed the ^*NCON^* intermediate by electrocatalytic C-N coupling reactions [Figure 8B and C]. Subsequently, the hydrogenation process was more inclined to follow the alternative mechanism to produce urea.

Figure 8. (A) Scheme of the simultaneous activation of N₂ and CO₂ to product urea under frustrated FLPs on Ni₃(BO₃)₂-150 catalysts. Reprinted with permission from ref^[36]. Copyright 2021 The Royal Society of Chemistry; (B) Free energy of urea synthesis by alternate mechanism at Bi-BiVO₄; (C) Mechanism of urea electrosynthesis with the synergistic effect of the Bi-BiVO₄ Mott-Schottky heterostructure. Reprinted with permission from ref^[39]. Copyright 2021 Wiley-VCH GmbH; (D) Free energy of urea synthesis at PdCu/TiO₂. Reprinted with permission from ref^[13]. Copyright 2020 Nature Publishing Group. FLPs: Frustrated Lewis pairs.

The research of Chen et al. supported the same opinions^[13]. The DFT results indicated that the side-on configuration on the PdCu surface favors the feedback of electrons from the d orbitals of Pd and Cu into the π^* orbitals of N₂, lowering the order of the N–N bond. The ^*N=N^* directly coupled with CO formed the urea precursor ^*NCON^*, and this process was thermodynamically and kinetically feasible. In the subsequent hydrogenation step, the ^*NCON^* formed the ^*NCONH structures. The second protonic step might have two reaction pathways, ^*NCONH₂ and ^*NHCONH, depending on the distal and alternating mechanisms^[41]. The third proton-coupling process was the potential limiting step in urea production, requiring +0.64 and +0.78 eV energies for the distal and alternative mechanisms, respectively [Figure 8D].

Above all, the reaction pathways for urea synthesis using NO₂^- or NO as nitrogen sources mostly prefer the C-N coupling of ^*NH₂ and ^*CO intermediates, except for the ZnO-Vo catalyst reported by Meng et al., which followed the coupling of ^*NH₂ and ^*COOH intermediates^[23]. This universal phenomenon can be attributed to the unstable binding of ^*COOH to N-related intermediates (e.g., ^*NO, ^*NHO, ^*NHOH, ^*NH₂OH, and ^*NH₂), while ^*COOH → ^*CO is usually a thermodynamically spontaneous process. Compared to other N sources, the pathways that use NO₃^- as N sources have different positions for the C-N coupling in various catalysts. The possible reason for this phenomenon is that the reduction reaction of NO₃^- (NO₃^-RR) has the largest number of protonation steps, which provides a rich variety of N-related intermediates. In this case, several intermediates (e.g., ^*NO₂, ^*NO, ^*NH, and ^*NH₂, as shown in Figure 4) have the potential to serve as key intermediates for C-N coupling. For the system that uses N₂ as a nitrogen source, the electrocatalytic C-N coupling processes are all realized by the insertion of ^*CO onto ^*N₂ after the elongation of the N≡N bond. The initial coupling of ^*CO to ^*N₂ will benefit from the matching of the molecular orbital symmetry of the two intermediates. Subsequent hydrogenation will follow either the alternative or distal mechanism [Figure 7]. It depends on the energy required for the reaction.

Characterization methods for urea detection and quantification

As discussed in the previous section, urea can be synthesized by the C-N coupling reaction of CO₂ with NO₂^-, NO₃^-, N₂, or NO. Side reactions are easy to occur during the catalytic process due to the complex multistep reaction mechanism. The competition of side reactions with synthesized urea results in low urea yields and FE. The detection of low urea concentrations is susceptible to various factors, leading to inaccurate quantification. Thus, both the influencing factors and detection limits of urea quantification need to be considered in urea synthesis. Currently, urea synthesized via electrocatalysis is primarily detected and quantified through colorimetry [including urease decomposition^[35,42] and diacetyl monoxime (DAMO) method]^[29], ¹H nuclear magnetic resonance (NMR) spectroscopy^[31] and high-performance liquid chromatography-mass spectrometry (HPLC-MS)^[23].

The main properties of electrocatalysts for urea synthesis

To overcome the disadvantage of the traditional urea synthesis process technology with high energy consumption, researchers have designed various kinds of electrocatalysts for the electrochemical synthesis of urea by compositional adjustment, structural design, elemental doping, and so on. And the selection of nitrogen sources, such as NO₃^-, NO₂^-, N₂, and NO, was expanded and co-reduced with CO₂ to achieve efficient electrochemical synthesis of urea. It is well known that FE is a performance index for evaluating urea production. By comparing the existing studies, it is not difficult to find that most of the catalysts that took the construction of bimetallic sites as well as inducing the generation of OVs achieved good FEs. For example, Cu-Ti³⁺ sites reported by Cao et al.^[21], Cu-W double sites by Zhao et al.^[44], Co-Ru double sites in SrCoRuO₃ catalysts by Lv et al.^[45] and the relay reaction of Zn and Cu in three-dimensional (3D) hybrid catalysts reported by Luo et al.^[31]. The alternating and synergistic reaction mechanism of the dual sites reduced the possibility of C-C coupling, while the adsorption function of the OVs avoided the desorption of the intermediates to produce the unwanted by-products. Thus, the selectivity for urea electrosynthesis was improved. Additionally, Saravanakumar et al. constructed a three-phase reaction interface to improve the reaction efficiency by controlling the TiO₂ size to the nanometer level and constructing porous membranes with Nafion^[46]. The In(OH)₃ and InOOH catalysts reported by Lv et al. effectively suppressed the HER and improved the urea selectivity through the n-to-p type transformation of the semiconductor behavior of the catalysts^[29,30]. The main properties of electrocatalysts for the synthesis of urea are shown in Table 1.

Metal system

In 1995, Shibata et al. realized the electrosynthesis of urea by constructing a Cu-loaded gas diffusion electrode for the first time^[20]. The as-prepared catalyst simultaneously electrocatalyzed NO₂^- and CO₂ to urea and exhibited a current efficiency of about 37% at -0.75 V vs. reversible hydrogen electrode (RHE). After decades of development, metal system catalysts have made great progress. For example, Huang et al. used the in situ electrochemical reduction methods to construct the Zn NBs from ZnO nanosheets, and this catalyst successfully synthesized urea by simultaneous reduction of CO and NO, with a urea yield of 15.13 mmol·h^-1·g^-1 and FE of 11.26% at -0.92 V vs. RHE^[27]. The high activity of Zn in urea synthesis was due to the intrinsic activity of metallic Zn and the control of nanostructures of the catalyst. The intrinsic activity could be described as the high selectivity for reducing CO₂ to CO and the inhibition of dimerization of ^*N intermediates during NO reduction reaction (NORR)^[53]. The corresponding results of the characterization test and simulated calculation revealed that ^*CO and ^*NH₂ (the critical intermediates in urea formation) were gained through the co-reduction of CO₂ and NO and coupled to form the C–N bond of urea step-by-step.

Recently, Liu et al. proposed an AuCu self-assembled nanofibers (AuCu SANFs) catalyst with a Boerdijk-Coxeter structure with (111)-dominant facets. The AuCu SANFs with one-dimensional nanostructure were used to couple the CO₂RR with NO₂^- to synthesize urea [Figure 9A-F]^[42]. In AuCu SANFs, abundant defects of AuCu 1D nanowires structure (twin boundaries, stacking faults, and atomic steps, shown in Figure 9D-F) provided plenty of highly active sites for electrocatalytic synthesis of urea^[54], while the Boerdijk-Coxeter structure could bring the generation of lattice strain and modulability of electronic structure^[55]. Combined with the synergistic effect of AuCu bimetal^[56], the AuCu nanocatalyst synthesized urea with a yield of 3,889.6 μg·h^-1·mg^-1_cat. at -1.55 V (vs. Ag/AgCl) and the maximum FE of 24.7%.

Figure 9. (A) SAED and (B) HRTEM for the AuCu self-assembled nanofibers. The illustrations in (B) show the lattice fringes of the square area and FFT pattern; (C-F) HRTEM for AuCu nanowires [red arrows (twin boundaries), blue arrows (stacking faults), and circles (atomic steps)]. Reprinted with permission from ref^[42]. Copyright 2022 Elsevier; (G) Schematic illustration for the synthesis of Cu@Zn catalyst. Reprinted with permission from ref^[32]. Copyright 2022 American Chemical Society; (H) Scheme of Zn/Cu 3D hybrid catalysts on a gas-diffusion layer. Orange and blue denote Cu and Zn, respectively. Reprinted with permission from ref^[31]. Copyright 2023 Nature Publishing Group. SAED: Selected area electron diffraction; HRTEM: high-resolution transmission electron microscope; FFT: fast Fourier transform; 3D: three-dimensional.

Similarly, Meng et al. used an electroreduction method to construct Cu@Zn nanowires with a self-supported core-shell structure, which was able to electrosynthesize urea by carbon dioxide and nitrate contaminants [Figure 9G]^[32]. This self-supported nanostructure was capable of avoiding using any of the polymer binders and increasing the active sites. The difference in work function between Cu and Zn led to the transfer of electrons from Zn to Cu, and the electron-deficient Zn surface would exhibit partial positive charge, which was more conducive to the electroreduction of NO₃^-[57]. The yield and FE of urea on Cu@Zn nanowires were apparently better than those on Zn and Cu catalysts. Combined with the characterization test and theoretical calculation, the results showed that the electron transfer between the Zn shell and the Cu core structure promoted the formation and coupling of ^*CO and ^*NH₂ intermediates to generate C–N bonds, thereby improving electrocatalytic properties.

The latest breakthrough came from Luo et al., who reported a 3D hybrid catalyst with functional synergy between Cu and Zn by spraying an incomplete Zn layer on top of the Cu layer [Figure 9H]^[31]. In situ spectroscopy and theoretical calculations demonstrated that Zn reduced the energy barrier for C-N coupling, and Cu lowered the reaction energy required for protonation to produce urea, realizing a relay electrocatalytic mechanism. By optimizing components and electrochemical test configurations, the urea FE achieved an extraordinary 75% in simulated wastewater containing 1,000 ppm NO₃^- [N]. Hybrid catalysts, consisting of two types of sites - one that promoted C–N bond formation and another facilitated protonation, had the ability to avoid the need to balance the adsorption energies of all the essential steps using a single catalytic site. The Zn/Cu hybrid catalysts realized a relay catalytic mechanism in which both the Zn and Cu sites were catalytically active and performed different functions. In contrast, single-component catalysts usually promote either the formation step of the C–N bond or the protonation step. For example, in the Cu@Zn core-shell catalyst described previously [Figure 9G], the shell layer Zn was catalytically active and its electronic structure was adjusted by the core layer Cu to promote C–N bond formation.

When it comes to the design of a metallic electrocatalyst, it is necessary to consider the intrinsic catalytic activity of the selected metal atoms. Priority should be given to the metal atoms with high CO₂RR and NRR selectivity. Nonetheless, the FE of pure metal catalysts was not universally high. The possible reason was that metal atoms were easy to aggregate, resulting in low atomic utilization and unclear catalytic active sites. Regulating the nanostructure of catalysts has been recognized as a universal method to optimize the catalytic performance of metal electrodes^[58]. For example, the AuCu nanofiber catalyst synthesized by Liu was composed of ultrathin AuCu alloy nanowires. The nanowires arranged in parallel have the unique characteristics of anisotropy and high surface area, which can improve atomic utilization, accelerate electron transport, and avoid agglomeration and dissolution of catalysts. Cu@Zn nanowires developed by Meng had unique self-supported nanostructures that avoided the use of polymer binders and effectively increased the active sites on the catalyst. What is more, the electronic interactions between components would be promoted by metal alloying. On the AuCu SANFs catalyst designed by Liu et al., the binding of Au changed the coordination and electronic structure of Cu, enhancing the adsorption and activation of CO₂ and improving the catalytic selectivity^[42]. Lu et al. took advantage of the difference in work functions between Cu and Zn to construct an electron-deficient Zn surface, facilitating the electrical reduction of NO₃^- and increasing the opportunities for C-N coupling. Thus, the performance of bimetallic catalysts was generally better than that of single metals^[59].

Metallic oxide system

In 2017, through two distinct methods, co-precipitate and microwave, Siva et al. constructed two structurally different FeTiO₃ electrodes as high-efficiency catalysts for CO₂ and NO₂ reduction to urea at a low current density of 10 mA·cm^-2[60]. It was observed that the performance of Fe-doped TiO₂ was more favorable than Cd or Co in improving the conversion of CO₂ and NO₂ to urea. Although the work by Siva et al. only studied the performance of the catalyst from the potential instead of urea synthesis yield and FE, through the comparison of the two catalytic electrodes synthesized by different methods, we learned that the FeTiO₃ catalyst synthesized by microwave possessed a more uniform crystal structure, which provided more adsorption functional groups^[60]. Therefore, microwave synthesized FeTiO₃ showed better cathodic behavior than co-precipitate synthesized FeTiO₃ for urea synthesis.

Saravanakumar et al. constructed the electrocatalyst by drop-casting the P-25 TiO₂ nanoparticles and Nafion solution on the electrode^[46]. TiO₂ had the characteristics of high oxidation/reduction capacity and high structural stability, and could be combined with Nafion to form a stable porous film on the electrode surface, creating a balanced three-phase reaction interface, which was favorable for urea synthesis^[61]. As expected, the prepared TiO₂/Nafion nanocomposite electrode successfully converted CO₂ and NO₃^- to urea with a FE of 40% at a lower overpotential than that reported by similar systems previously.

After the creative work on realizing C-N coupling and producing acetamide with excellent selectivity and efficiency that used ammonia as the N source by Jouny et al. in 2019^[16]. Chen et al. proposed that the co-reduction of nitrogen and carbon dioxide could form the C–N bonds, thus achieving urea synthesis under ambient conditions [Figure 10A-C]^[13]. They demonstrated an approach to construct an electrocatalyst by attaching PdCu alloy nanoparticles to TiO₂ nanosheets to the electrocatalytic coupling of N₂ and CO₂ in water and form urea. Compared to the single-metal-loaded catalyst, Pd₁Cu₁/TiO₂ exhibited higher urea yield and FE, and a decline in the onset potential. The improvement of catalytic performance could be attributed to the optimization of the electronic structure and the strong interaction between the bimetals and the support. TiO₂ nanosheets treated with a reduction atmosphere at elevated temperatures were easy to form OVs. The formation of OVs was capable of narrowing the band gap of TiO₂ without changing the crystalline phase of the materials. Moreover, the comparison experiment proved that the interaction between TiO₂ and metal was stronger than that of carbon support (Carbon Black XC72R). The TiO₂ had been reported to stabilize the intermediates to promote CO₂RR^[62]. However, the N₂ with low solubility and the stable N≡N triple bond (940.95 kJ·mol^-1) resulted in poor selectivity, FE, and yield of urea^[63]. CeO₂ is an n-type semiconductor with high electron density that can stabilize metal dopants. According to this, Wei et al. reported a catalyst loaded with Cu single atoms on a CeO₂ carrier synthesized by wet impregnation and calcination^[35]. A series of characterization proved that in comparison to the high concentration Cu-doped sample H-Cu-CeO₂, the low concentration Cu-doped L-Cu₁-CeO₂ exhibited a nanorod morphology with atomically dispersed atoms, which provided active sites that could be accurately identified. The catalyst reached an average urea yield rate of 52.84 mmol·h^-1·g^-1_cat. at -1.60 V (vs. RHE), which was at the leading level among similar studies reported so far. In order to monitor the structural evolution of the Cu₁-CeO₂ catalyst during the C-N coupling process, this work employed the Operando X-ray absorption spectroscopy (XAS) technique. The XAS results indicated that during the catalytic process, the individual Cu²⁺ was gradually reduced to Cu⁺ and Cu⁰ to form Cu₄ clusters which were the real active sites for electrocatalytic urea synthesis. This process was reversible at open-circuit potential, thus providing the catalyst with superior structural and electrochemical stabilities. Yu et al. attempted to use Ru-doped CeO₂ to increase the OV concentration, and successfully doped Ru into the interior of the CeO₂ lattice through a wet reduction process^[49]. Ru atoms partially replaced the Ce sites, leading to lattice contraction, resulting in local lattice disorder and promoting the formation of OVs. OVs could promote the electroreduction of NO₃^- and CO₂ through enhanced adsorption, and Ru could capture the ^*CO, leading to significant inhibition of HER. The synergistic effect of Ru doping and OVs optimized the adsorption and activation of the reactants and facilitated the C-N coupling kinetics. At a Ru doping ratio of 5%, the catalyst had a high urea yield (20.2 mmol·h^-1·g^-1) and FE (20.1%) at a low overpotential (-0.70 V vs. RHE).

Figure 10. (A) Urea synthesis schematic diagram at Pd₁Cu₁/TiO₂-400; (B) Formation rate of urea at Pd₁Cu₁/TiO₂-400; (C) The current density and FE for each product at different potentials at Pd₁Cu₁/TiO₂-400. Reprinted with permission from ref^[13]. Copyright 2020 Nature Publishing Group; (D and E) O 1s XPS of (D) TiO₂ without doping and (E) Cu-TiO₂; (F and G) Cu 2p XPS of (F) CuO and (G) Cu-TiO₂; (H and I) Ti 2p XPS of (H) TiO₂ without doping and (I) Cu-TiO₂. Reprinted with permission from ref^[21]. Copyright 2020 Elsevier Inc. FE: Faradaic efficiency; XPS: X-ray photoelectron spectroscopy spectra.

As mentioned above, OVs are important structural defects in metal oxides. It is rich in electron density and can act as a catalytic center^[64]. Studies have shown that OVs would induce anatase TiO₂(101) surface to form a pair of Ti³⁺ sites, which exhibited excellent binding and reduction capacity for nitrogen atoms^[65]. Therefore, Cao et al. adulterated Cu in TiO₂ (Cu-TiO₂) to promote the forming of abundant defect sites and OVs [Figure 10D-I]^[21]. The doped Cu acted as the main adsorption site for CO₂ and provided electrons to form ^*CO intermediates. At the same time, the low-valence Cu dopants induced abundant OVs, which were conducive to the formation of exposed bi-Ti³⁺ active sites. Adsorption and reduction of NO₂^- on Ti³⁺ sites elevated the selectivity of NO₂^--to-^*NH₂^[66]. The synergistic interaction of the two sites promoted the formation of ^*CO and ^*NH₂ intermediates and led to efficient C-N coupling subsequently. This Cu-TiO₂ electrocatalyst with rich OVs exhibited a fabulous urea production rate (20.8 μmol·h^-1) at -0.40 V (vs. RHE) with a FE of 43.1%, significantly better than those of TiO₂ without doping (5.91 μmol·h^-1 and 27.3%). After that, Meng et al. designed a self-supported ZnO porous nanosheet with rich OVs (ZnO-V), and used it as an electrocatalyst for efficient urea synthesis with carbon dioxide and nitrite as raw materials^[23]. Due to the presence of surface oxygen-rich vacancies, the urea yield of ZnO-V reached 16.56 mmol·h^-1 at -0.79 V(vs. RHE), obviously better than that of ZnO (7.72 mmol·h^-1). And the FE of ZnO-V achieved 23.26%, which was approximately three times as high as that of ZnO (8.10%). Lv et al. used a sol-gel chemical process and controlled thermal treatment to synthesize SrCoRuO₃ catalysts with Co-Ru double sites^[45]. Meanwhile, the co-doping produced abundant OVs, which modulated the electronic structure of the active sites, realizing efficient urea electrosynthesis. The substitution of Ru sites by Co could distort the SrRuO₃ crystal structure due to the smaller atomic radius of Co. In their work, tunable lattice strain was realized by precise control of Co doping ratio. Ru and Co atoms had been demonstrated to be active metals for NO₃^- and CO₂ activation, respectively. In situ experiments and DFT theoretical calculations indicated that Co and Ru acted as active sites for the formation of ^*CO and ^*NH₂ intermediates, respectively. The OVs modulated the electronic structure of Co and Ru sites, weakening the adsorption of the key intermediates. The synergistic effect of Co-Ru sites and OVs promoted the C-N coupling between ^*CO and ^*NH₂, avoiding the desorption of the intermediates to produce unwanted by-products.

In addition to the introduction of OVs in the electrocatalyst, the urea yield can also be increased by designing the heterojunction on the electrocatalyst. For example, Yuan et al. designed the Bi-BiVO₄ electrocatalyst with Mott-Schottky heterostructures, which reached an excellent yield rate of urea of 5.91 mmol·h^-1·g^-1 at -0.40 V vs. RHE and a FE of 12.55% [Figure 11A-F]^[39]. This might be attributed to the fact that the formed heterostructure structure interface could promote charge transfer spontaneously; the resulting space-charge layer guaranteed the selective adsorption of CO₂, activation of inert N₂, and the exposure of more active sites, contributing to C-N coupling and urea synthesis^[40]. The electrochemical impedance spectroscopy (EIS) showed that the Bi-BiVO₄ had a smaller semicircle and a higher slope than the unhybrid BiVO₄, confirming the formation of the localized nucleophilic and electrophilic regions on the surface of Bi-BiVO₄. Theoretical simulations verified that electron-rich N atoms in N₂ and electron-deficient C atoms in CO₂ would be targeted for adsorption in the electrophilic and nucleophilic regions, respectively. Govindan et al. successfully doped Ru-Pd alloy nanoparticles into two-dimensional (2D) WO₃ and MXene nanosheets to form a Ru-Pd/WO₃/MXene heterostructure^[52]. A bifunctional electrode was constructed for simultaneous cathodic reduction of CO₂ and N₂ to urea and anodic oxidation of water. In the electrochemical reduction process at the cathode, CO₂ was first converted to ^*CO and then directly coupled with N₂ via sequential protonation to produce urea. At -0.60 V (vs. RHE), the urea yield was 227 μg·h^-1·mg^-1_cat. with a FE of 23.7%. Since WO₃ was at a lower conduction band (CB) position, it could adsorb single pairs of N₂ in the active site more efficiently, thus activating the reduction of N₂ to ^*N₂ intermediates. The 2D structure of MXene provided a stable platform for catalytic activity and excellent conductivity. The construction of the Ru-Pd metal active sites in the WO₃/MXene heterostructure increased the solubility of CO₂, which was converted to ^*CO intermediates by the electron-proton transfer process. Therefore, the Ru-Pd/WO₃/MXene catalyst had an efficient C-N coupling capability.

Figure 11. (A) High-resolution Bi 4f and (B) V 2p spectra for BiVO₄ and Bi-BiVO₄; (C) Mott-Schottky diagram for BiVO₄; (D) Diagram of charge transfer process at Bi-BiVO₄; (E) Specific surface area and (F) XPS for BiVO₄ and Bi-BiVO₄. Reprinted with permission from ref^[39]. Copyright 2021 Wiley-VCH GmbH; (G) SEM and (H and I) HRTEM images for Ni₃(BO₃)₂-150 catalysts; (J) XPS survey spectra, (K) high-resolution O 1s and (L) Ni 2p spectra for the original Ni₃(BO₃)₂, Ni₃(BO₃)₂-150, and Ni₃(BO₃)₂-250. Reprinted with permission from ref^[36]. Copyright 2021 The Royal Society of Chemistry. XPS: X-ray photoelectron spectroscopy spectra; SEM: scanning electron microscopy; HRTEM: high-resolution transmission electron microscope.

Nevertheless, the above strategy only focused on the adsorption and activation of reactant molecules, while ignoring the electrocatalytic C-N coupling reaction that occurred during urea synthesis. Therefore, Yuan et al. further synthesized the floral nickel borate [Ni₃(BO₃)₂] by designing unique frustrated FLPs [Figure 11G-L]^[36]. FLPs consisting of LA and LB enabled chemisorption and reaction of gas molecules^[67]. On the Ni₃(BO₃)₂ catalyst, the coordination unsaturated Ni sites and the neighboring surface hydroxyl groups acted as LA and LB sites, respectively. During the reaction, the LA sites coupled with empty orbitals would capture the lone pair electrons from the N atoms in N₂, while the surplus electrons in the LB sites would be donated to the C atoms in CO₂. On the FLPs, the unoccupied orbitals of LA and the non-bonding orbitals of LB would interact with the bonding and antibonding orbitals of the reactant molecules, respectively, which led to the polarization of N₂ and CO₂. After molecular activation via FLPs, the ^*N=N^* intermediate was coupled with the ^*CO intermediate through an σ-orbital carbonylation way to synthesize the ^*NCON^* precursor^[37,68]. The acquired highest yield rate was 9.70 mmol·h^-1·g^-1_cat. and the FE was 20.36% at -0.50 V (vs. RHE).

A new study by Chen et al. doped Sb in amorphous BiO_x clusters, which affected the mode of CO₂ adsorption by modulating the 6p orbital configuration of Bi, improving the possibility of producing urea^[51]. The prepared Sb_xBi_1-xO_y clusters possessed a reaction potential of only -0.30 V (vs. RHE), a yield of 307.97 μg·h^-1·mg^-1_cat., and a FE of 10.9%, which was considerably improved in performance compared to the undoped BiO_x clusters (80 μg·h^-1·mg^-1_cat., FE = 1.3% at -0.40 V vs. RHE). Both in situ experiments and the first principles calculations indicated that Sb was capable of effectively stabilizing the electronic and geometrical structures of amorphous BiO_x clusters. The doping of Sb introduced an OV in the BiO_x system, corresponding to two excess electrons localized in the 6p orbital of Bi, and the low-oxidation state of Bi(II) emerged^[69]. During catalyzing, the specific Bi-O coordination number of Bi(II) decreased and stabilized at ~1.4, and these unsaturated coordination sites had high catalytic activity. As a result, the adsorption geometry of CO₂ on the catalyst surface was changed from O-connected to C-connected, and the preferential adsorption of C atoms promoted the generation of the key ^*CO intermediates of C-N coupling while obstructing the formation of unwanted by-products, such as HCOOH derivatives. Furthermore, the Bi(II) active sites could facilitate C-N coupling by efficiently injecting electrons into N₂ that favorably modified the symmetry of the CO₂ and N₂ frontier orbitals. DFT calculations showed that the ^*N₂ with injected electrons formed a “new-type” highest occupied molecular orbital (HOMO) that matched the symmetry of the lowest unoccupied molecular orbital (LUMO) of ^*CO. Then, the possibility of forming ^*NCON^* by C-N coupling was greatly enhanced.

The catalysts of metal oxide systems are usually doped with metal atoms to achieve the electrocatalytic synthesis of urea. Therefore, the selection of a suitable substrate material should be taken into account primarily. The chosen substrate ought to generate strong interactions with the doped metal, and it is necessary to ensure that the loaded heteroatoms are sufficiently well dispersed. Secondly, designing active sites that combine the abilities of adsorption, synergistic activation and C-N coupling of reactant molecules is desirable. For instance, Zhao et al. utilized the high-valent metals W and Cu to prepare a CuWO₄ catalyst with alternating reaction sites^[44]. An outstanding FE of 70.1% ± 2.4% was achieved at a low overpotential of -0.20 V (vs. RHE). It is well known that Cu is a common CO₂RR catalyst, but consecutive Cu sites on Cu-based catalysts may generate C-C coupling which reduces the synthesis efficiency of urea. High-valent tungsten oxides are beneficial for stabilizing ^*NO₂ intermediates and do not trigger CO₂ reduction reactions. So, the alternating reaction site composed of W and Cu, two active sites with different functions, could effectively balance the adsorption capacity of ^*NO₂ and ^*CO intermediates. In addition, defect engineering has been widely applied to modulate the physical and chemical properties of catalysts. OVs are accessible and significant structural defects in metal oxides that can serve as active sites with abundant electrons. For example, in Cu-TiO₂ catalysts, the doping of low-valent Cu induced abundant OVs, favoring the formation of exposed bi-Ti³⁺ active sites^[21]. ZnO with OVs (ZnO-V) exhibited higher electrochemically active surface area (ECSA) than pure ZnO^[23]. What is more, research by Yuan et al. provides the idea of utilizing heterojunction to induce spontaneous charge transfer and generate space-charge regions^[39]. The engineering of distinct space-charge regions is capable of guiding the accurate adsorption and activation of reactant molecules, which contribute to the thermodynamically feasible coupling of the C–N bond and efficient urea production.

Metal hydroxide system

N-type semiconductor catalysts have been used in the electrosynthesis of urea due to their properties of inhibiting the HER^[70]. Thus, Lv et al. realized urea electrosynthesis via directly coupling NO₃^- with CO₂ on an In(OH)₃ catalyst with {100} facets at ambient conditions^[29]. The synthesized catalyst achieved highly selective urea production with a yield of 533.1 μg·h^-1·mg^-1_cat. and FE of 53.4% at -0.60 V (vs. RHE). It is noted that the selectivity of nitrogen and carbon was 82.9% and nearly 100%, respectively. This study showed that the CO₂ absorbed at the surface, in addition to being the feedstock for the urea synthesis, could also help produce the hole accumulating layer on the catalyst by trapping electrons, resulting in an n-to-p type transformation of the semiconductor behavior of the In(OH)₃ surface. The inhibition of HER was achieved by the transformation of n-p semiconductor properties on the surface of In(OH)₃ electrocatalyst^[71].

On this basis, Lv et al. synthesized the OV-rich indium oxyhydroxide electrocatalyst (Vo-InOOH)^[30]. The DFT calculations indicated that the rate-determining step in the whole urea synthesis process was the ^*CO₂NH₂ protonation. The defect engineering of Vo could reconfigure the electronic structure of the surface in the active site, thereby reducing the energy barrier for converting ^*CO₂NH₂ to ^*COOHNH₂, which was the determining factor of the catalytic efficiency in this system. The constructed Vo-InOOH electrocatalyst showed a remarkable FE of 51.0%, with a yield of 592.5 μg·h^-1·mg^-1_cat..

Since the steps to synthesize urea are sequential protonation processes after the C-N coupling, the inhibition of the HER is of great significance to the overall efficiency. Lv et al. ingeniously utilized the manipulated conversion of semiconductor properties on the indium oxyhydroxide surface to achieve effective inhibition of the HER^[29]. Despite many debates about the mechanism of n-p conversion in semiconductors, the concept of controlling electron transfer still inspires us to design more efficient electrocatalyst surfaces that facilitate C-N coupling reactions.

Nitrogen-doped carbon system

The system that doped heteroatoms (e.g., N) on the carbon substrate had drawn attention as an electrocatalyst due to their unique 2D nanostructure, high surface area, and more exposed active sites^[72]. In 2021, Roy et al. first proposed dual-Si doped graphene-C₆N₆ sheets (Si₂-g-C₆N₆) and used theoretical calculation of DFT to study the process of N₂ and CO₂ coupling^[73]. Bader charge analysis indicated that the charge transferred from the doped surface to the antibonding π^* orbitals of the N₂. The elongation of the N–N bond led to the activation of the N₂. The activated N₂ spontaneously adsorbed the ^*CO intermediate to form a C-N coupling. Moreover, according to DFT calculations, protons were more likely to be adsorbed at the Si active sites instead of the free N centers of the sheet. Thus, the potential of urea production initially was far lower than that of NH₃ synthesis and HER. Similarly, Dutta et al. proposed metal-free electrocatalysts based on dual silicon-doped C₉N₄ nanosheets, which could effectively produce urea with a low overpotential value [Figure 12A-C]^[74]. According to the Bader charge analysis, the presence of diatomic Si on the C₉N₄ substrate generated surplus surface charge density that could be provided to adjacent N atoms from N₂. Thus, the Si sites became the active center for N₂ catalysis, adsorbing N₂ in a side-on manner. Compared to ^*NHH, the density of states images of ^*N(CO)N had more overlap around the Fermi level, illustrating that the interaction between Si and attached N was stronger. Hence, the formation of ^*N(CO)N was more favorable, which meant the synthesis of urea was more favorable. Both above studies had shown that metal-free C_xN_y materials could promote PCET reactions, where CO was inserted into activated N₂ and then directly coupled to C–N bonds to form ^*NCON^*, which was an important intermediate for urea production.

Figure 12. (A) Optimized geometry for the surface of Si₂-g-C₆N₆ (black ball: C atom, blue ball: N atom, golden ball: Si atom, and numerical value shows the bond length, Å); (B) HSE03 band structure and (C) Projected density of state for the Si₂-g-C₆N₆ surface. Reprinted with permission from ref^[73]. Copyright 2021 American Chemical Society; (D and E) XANES spectra at Cu K-edge for Cu single atoms confined in graphene sheets (Cu-GS-800, 900 and 1000), compared with Cu^IIO [Cu(II)] and Cu^I₂O [Cu(I)]; (F) Fourier transformed EXAFS at the Cu K-edge for Cu-GS-800, 900 and 1000. Reprinted with permission from ref^[33]. Copyright 2022 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH. XANES: X-ray absorption near edge structure; EXAFS: extended X-ray absorption fine structure.

Single-atom catalysts loaded on carbon substrates exhibited exceptional electrocatalytic activities because of the unique electronic structure and maximized atom utilization^[75]. Leverett et al. designed a Cu single-atom catalyst (Cu-SAC) in the carbon framework that doped with N, which was able to electrocatalyze both CO₂RR and NO₃^-RR [Figure 12D-F]^[33]. The electrochemical measurements showed that the CO FE of the coordination structure of Cu-N₄ was much higher than that of Cu-N_4-x-C_x. The polarization curves for CO₂RR also demonstrated that the intrinsic activity of CO₂RR at the Cu-N₄ site was significantly higher. Compared with Cu-N₃-C₁ and Cu-N₂-C₂, Cu-N₄ had a lower Gibbs free energy (ΔG) of ^*COOH formation according to the DFT calculations. In addition, Cu-N₄ had almost no decrease in CO FE at more negative potentials, which could be attributed to the enhanced coupling of ^*CO₂ and ^*NO₂, resulting in improved urea yields. The performance of the urea production rate reached 4.3 nmol·s^-1·cm^-2, and the urea FE was 28% at -0.90 V (vs. RHE).

The adjacent sites of bimetallic atoms have the ability to manipulate the geometric and electronic structures of SACs^[76]. Zhang et al. pioneered the bonding of Fe-Ni pairs to prepare bonded Fe-Ni diatomic catalysts (B-FeNi-DACs) [Figure 13]^[34]. The DFT computations demonstrated that in the single-atom doping system, the active center was covered by the ^*NO intermediates, which inhibited the effective capture and activation of CO₂. After the doping of the second metal atom, the applied potential required to drive ^*CO generation was reduced, which indicated that the bimetallic sites were more effective in promoting C-N coupling. The bonded Fe-Ni pairs acted as the active sites for synergistically adsorbing and activating of reactants, promoting the thermodynamically and kinetically of C-N coupling^[77]. At -1.50 V vs. RHE, an excellent urea yield rate of 20.2 mmol·h^-1·g^-1 had been successfully realized with an FE of 17.8%. Identically, the latest research reported by Hou et al. also emphasized the feasibility of urea production through C-N coupling on alloy catalysts^[48]. Considering the excellent performance of nickel-based catalysts in CO₂RR and iron-based catalysts in NO₃^-RR^[78,79], Hou et al. constructed bimetallic FeNi₃ alloy particles loaded with N-doped porous carbon for electrocatalytic urea synthesis by reduction of CO₂ and NO₃^-[48]. The structure of porous carbon promoted electrochemical kinetics, as well as the sufficient exposure of active sites. The prepared electrocatalysts achieved the highest yield of 496.5 μg·h^-1·mg^-1_cat. with a FE of 16.58% at -0.90 V (vs. RHE). Zhu et al. proposed that the MN₃-M’N₄ bimetallic moiety with two dispersed active sites could simultaneously inhibit the separate reduction process after the adsorption of CO₂ and N₂^[80]. They used ab initio molecular dynamics (AIMD) simulations to calculate the thermal stability of 26 homonuclear and 650 heteronuclear bimetallic systems and filtered out 205 structurally stable combinations. And then they picked out 18 M-M’ combinations (Sr-Ir, Ti-Rh, Fe-Os, Y-Co, Y-Rh, Zr-Rh, Sc-Os, Cr-Ir, Mn-Rh, Mn-Ir, Co-Os, Y-Fe, Y-Ru, Zr-Ru, Y-Re, Y-Ir, Zr-Cr, Ru-Co) with high catalytic activity by a five-step high-throughput screening on the basis of NCON, CO and OCOH pathway. Combined with the evaluation of urea synthesis selectivity, the three most superior M-M’ (Fe-Os, Co-Os and Ru-Co) electrocatalysts were finally obtained. Among them, the Ru-Co combination (RuN₃-CoN₄) exhibited a very low limiting potential and excellent urea production performance, which had great potential for practice.

Figure 13. (A) TEM (The scale bar is 100 nm) and (B) Aberration-corrected HAADF-STEM of B-FeNi-DASC. (The scale bar is 2 nm); (C) HAADF-STEM intensity profile with atomic-resolution EELS for the Fe-Ni pair; (D) Fourier transform EXAFS for Fe-SAC, I-FeNi-DASC, and B-FeNi-DASC; (E) Wavelet-transform of Fe element for Fe foil, Fe-SAC, I-FeNi-DASC, and B-FeNi-DASC. Reprinted with permission from ref^[34]. Copyright 2022 Nature Publishing Group. TEM: Transmission electron microscopy; HAADF-STEM: high-angle annular dark-field-scanning transmission electron microscopy; B-FeNi-DASC: bonded Fe-Ni diatomic sites catalyst; EELS: electron energy loss spectroscopy; EXAFS: extended X-ray absorption fine structure; SAC: single-atom catalyst.

Synthesizing monatomic or diatomic catalysts on the carbon nitride substrates was a significant strategy to improve the utilization of atoms^[81,82]. In this system, the coordination structures of active sites become a key factor that affects the catalytic efficiency. The bonded diatomic catalysts not only overcome the limitation that single-atom catalysts selectively adsorb and activate reactants unilaterally, but also increase the possibility of C/N intermediates meeting and coupling to form critical C–N bonds. The construction of diatomic sites has emerged as a viable strategy to achieve synergistic adsorption and activation for electrocatalytic C-N coupling reactions, which improve the yield and FE of urea effectively^[83].

Other system

Doping heteroatoms on metal substrate

Doping heteroatoms are an effective strategy for improving the adsorption energy of intermediates by adjusting the surface electron structure^[87,88]. Feng et al. reported a Te-doped Pd NC catalyst to synthesize urea through coupling CO₂RR with nitrite reduction [Figure 14A-E]^[22]. The doping of Te promoted not only the adsorption of CO₂ to form ^*CO but also the production of NH₃, satisfying the needs of urea formation. Under the optimized flow cell, the urea concentration reached 0.95 wt%, and the N atom efficiency (NE) was 82.3%.

Figure 14. (A) HAADF-STEM with diameter histogram and (B) high-magnification TEM for Te-Pd NCs; (C) PXRD for Pd NCs and Te-Pd NCs. (i and ii) PXRD patterns enlarged from (C); (D) HRTEM for Te-Pd NCs from eectangular region in (B); (E) HAADF-STEM as well as elemental mappings of Te-Pd NCs. Reprinted with permission from ref^[22]. Copyright 2020 American Chemical Society; (F) The charge density difference and the corresponding results for Bader charge analysis of Co-PMDA-2-mbIM; the yellow indicates electron accumulation and the cyan indicates electron depletion, the isosurface value is 0.0025 e·Bohr^-3; (G and H) The DOS and integral DOS for (G) Co-PMDA and (H) Co-PMDA-2-mbIM. Reprinted with permission from ref^[50]. Copyright 2022 The Royal Society of Chemistry. HAADF-STEM: High-angle annular dark-field-scanning transmission electron microscopy; TEM: transmission electron microscopy; NCs: nanocrystals; PXRD: powder X-ray diffraction; HRTEM: high-resolution transmission electron microscope; PMDA: pyrodimethyl dianhydride; 2-mbIM: 2-methylbenzimidazole; DOS: density of states.