Constructing Asymmetric Sn‐Cu‐C Interface via Defective Carbon Trapped Atomic Clusters for Efficient Neutral Nitrate Reduction

Abstract

Multi‐atom cluster (MACs) catalysts have recently attracted significant research interest for their potential to catalyze multi‐electron reactions through cooperative interactions among adjacent active sites. However, the controllable synthesis of MACs and the electrocatalytic mechanism understanding of their synergistic effects remain challenging. Herein, we develop a defect engineering strategy to anchor bimetallic SnCu atomic clusters at defective graphene (SnCu‐DG) via carbon defect‐mediated atomic trapping, wherein edge defects act as confined reactors for cluster nucleation. Taking nitrate reduction as an example, the SnCu‐DG catalyst achieves a high NH₃ Faradaic efficiency (99.5%) at neutral electrolyte condition, accompanied by a record intrinsic activity of 2.61 × 10⁻¹⁷ mmol h⁻¹ site_Cu ⁻¹, surpassing Cu‐DG and SnCu‐G counterparts by 16.0‐ and 7.8‐fold, respectively. X‐ray adsorption spectra and theoretical calculations reveal the electrons transfer between Cu and carbon defect sites while Sn incorporation intensifies asymmetric charge polarization across the Sn‐Cu‐C interface. This dual modulation collaboratively optimizes the catalytic microenvironment, simultaneously enhancing *NO₂ ⁻ adsorption, accelerating water dissociation kinetics, and breaking the intrinsic linear scaling between intermediate adsorption and hydrogenation.

This work constructs asymmetric Sn‐Cu‐C interface via anchoring SnCu atomic clusters on carbon defects. Such as‐synthesized SnCu‐DG catalyst achieves a high NH₃ Faradaic efficiency (99.5%) and a record intrinsic activity of 2.61 × 10⁻¹⁷ mmol h⁻¹ site_Cu ⁻¹ at neutral electrolyte condition. This work reveals that the electronic asymmetry enhanced by carbon defects is the key for multi‐step electrocatalysis.

1. Introduction

The widespread contamination of groundwater by nitrate pollutants, resulting from agricultural runoff and industrial effluents, presents critical challenges to aquatic ecosystems and public health.^{[ 1 ]} While conventional biological denitrification are widely used, their effectiveness is hindered by sluggish reaction kinetics and operational complexities.^{[ 2} , ^{3 ]} Electrochemical nitrate reduction (NO₃RR) has emerged as a pivotal technology not only for mitigating pervasive nitrate pollution but also for enabling the sustainable production of valuable ammonia—a critical fertilizer and carbon‐free energy carrier—thus addressing dual imperatives of environmental remediation and resource recovery. This approach offers transformative potential for industrial applications, including decentralized treatment of nitrate‐laden agricultural/industrial wastewater and on‐demand green ammonia synthesis for fertilizers or hydrogen storage.^{[ 4} , ⁵ , ⁶ , ^{7 ]} However, the product selectivity of NO₃RR is fundamentally constrained by its intricate eight‐electron, eight‐proton coupled transfer pathway, which involves multiple intermediates (*NO₃ ⁻ → *NO₂ ⁻ → *NO → *NH₂ → *NH₃) with varying adsorption energetics.^{[ 7} , ⁸ , ⁹ , ^{10 ]} Currently, NO₃RR catalyst design faces interconnected challenges, such as, (i) scaling relations hindering multi‐step energetics optimization; (ii) ambiguous structure‐activity correlations in multi‐metallic systems; (ii) potential dynamic structural evolution during reaction impacts long term stability. Compounding these challenges, the competing hydrogen evolution reaction (HER) drastically restrains reaction efficiency.^{[ 11} , ^{12 ]} These limitations underscore the urgent need for designing advanced catalysts with tailored active sites capable of: (1) enhancing nitrate adsorption affinity, (2) activating inert N‐O bonds, and (3) stabilizing critical nitrogenous intermediates (e.g., *NO₂ ⁻, *NO) while suppressing HER.

In recent years, the strategic development of multi‐active‐site catalysts has gained prominence in addressing complex multi‐electron transfer processes.^{[ 13} , ¹⁴ , ^{15 ]} Multi‐atom cluster (MAC) catalysts – bridging the gap between single‐atom and nanoparticle systems – provide distinct advantages due to their high atomic utilization efficiency and unsaturated coordination environments.^{[ 16} , ¹⁷ , ¹⁸ , ¹⁹ , ^{20 ]} Moreover, the atomic ensembles in MACs enable cooperative interactions among adjacent metal sites and substrate, overcoming the scaling relations in single‐atom catalysts, realizing simultaneously reactant activation and intermediate stabilization through multi‐active sites and strong metal‐support interactions (SMSI).^{[ 19} , ²¹ , ²² , ²³ , ^{24 ]} Currently, carbon‐based substrates with engineered defects have emerged as ideal platforms for synthesis of metallic clusters, as vacancy defects not only serve as anchoring sites to suppress metal aggregation but also modulate the electronic structure of supported clusters through interfacial charge transfer.^{[ 25} , ²⁶ , ²⁷ , ^{28 ]} Pioneering studies have demonstrated the potential of carbon defects (single vacancy and C‐585 defect) in controlling the formation of mono‐ and bimetallic sites via strong coordination and confinement effect.^{[ 24} , ²⁹ , ³⁰ , ^{31 ]} However, achieving precise control over MAC nucleation kinetics in larger defects (nanopore with edge defects) remains particularly challenging. Generally, high‐temperature pyrolysis tends to result in random migration and uncontrolled metal aggregation on insufficient defective carbon substrates, favoring thermodynamically stable nanoparticles over atomic clusters.^{[ 32} , ^{33 ]} Furthermore, elucidating structure‐activity relationships for MACs in NO₃RR is complicated due to the complexity of metal‐metal/metal‐carbon substrate interaction and multi‐electron pathways, particularly in multi‐metallic systems.^{[ 16} , ^{34 ]} These challenges highlight the critical need for innovative strategies to regulate cluster nucleation processes and systematically deconvolute MAC contributions to multi‐electron catalysis.

Herein, we present a defect engineering strategy to anchor bimetallic SnCu atomic clusters at graphene edge sites (SnCu‐DG) using defect‐mediated atomic trapping. By utilizing nanopore with edge carbon defects as spatially confined nanoreactors, we achieve controlled cluster nucleation while suppressing metal diffusion and aggregation. Atomic‐resolution microscopy confirms the targeted growth of SnCu clusters at edge defect regions. The optimized SnCu‐DG catalyst demonstrates high NO₃RR performance, achieving a NH₃ Faradaic efficiency with 99.5% at −0.8 V vs. RHE and a record intrinsic activity of 2.61 × 10⁻¹⁷ mmol h⁻¹ site_Cu ⁻¹, surpassing Cu‐DG and SnCu‐G controls by 16.0‐ and 7.8‐fold, respectively. Combining X‐ray absorption spectroscopy (XAS) and density functional theory (DFT) calculations, we reveal that carbon defects synergize with atomic Sn incorporation to induce charge redistribution within the Sn‐Cu‐C interface. This creates a polarized microenvironment where the electron‐donating capacity of Cu sites near to Sn is reduced, while distal Cu sites exhibit enhanced electron donation. Notably, the interfacial charge transfer mediated by carbon defect amplifies this electronic asymmetry, optimizing the adsorption of NO₃ ⁻ at specific Cu sites. Operando Fourier transform infrared spectroscopy (FTIR) and mechanistic DFT analyses further demonstrate that the engineered microenvironment simultaneously strengthens *NO₂ ⁻ adsorption, suppresses HER, and accelerates water dissociation kinetics via carbon defects. This work establishes defect‐mediated confinement as a paradigm for precision MAC design and unveils electronic asymmetry enhanced by carbon defects as a key principle for multi‐step electrocatalysis.

2. Results and Discussion

The SnCu atomic clusters on defective graphene (SnCu‐DG) were synthesized via a chemical etching, impregnation and pyrolysis process. Considering the strong capture capacity of defect sites in carbon, we first introduced sufficient defects on DGO via a facile chemical etching method (see Experimental Section, Figure 1a). Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images of the DGO indicate that the DGO exhibits obvious porous structures on the carbon skeleton, where the O‐groups are uniformly distributed (Figure S1, Supporting Information). The atomic‐level structure of defects on DGO was further observed using an Integrated Differential Phase Contrast scanning transmission electron microscopy (iDPC‐STEM). Clearly, partial upper graphene layers were etched, exposing numerous edges modified with O species (atoms depicted in bright white), which may provide efficient anchor sites for metal species (Figure 1b, and Figure S2, Supporting Information). The powder X‐ray diffraction (PXRD) patterns of GO and DGO show a consistent characteristic peak, while DGO possesses a marginally smaller d‐spacing (higher 2θ degree), which may be attributed to the decrease of in‐plane O‐groups (Figure S2, Supporting Information). The Raman spectra exhibit an enhanced I_d/I_g value of DGO (1.09) compared to that of GO (1.01), quantitatively demonstrating an increased defect density after chemical etching of GO (Figure 1c). Notably, the distribution of O‐group types is also changing with the increase of defect density. The characteristic peaks of C─O─C (849 cm⁻¹) and C─O (1227 cm⁻¹) on DGO weaken in contrast to those of GO, as shown in Fourier Transform Infrared (FTIR) spectra (Figure S3, Supporting Information).^{[ 35} , ^{36 ]} The types and concentrations of O‐groups in GO and DGO samples were also investigated by X‐ray photoelectron spectroscopy (XPS). As shown, the oxygen concentration of DGO is similar to that of GO, which helps exclude the interference of oxygen concentrations concerning the pyrolysis mechanism (Figure S4a,b, Supporting Information). As stated in our previous report, the introduction of defects may potentially control the types of O‐groups.^{[ 35 ]} Therefore, the high‐resolution O 1s spectra of the samples were analyzed. Clearly, the ratio of O‐groups at in‐plane sites (C─O─C, C─O) decreases with the increase in defect density of DGO, indicating the existence of abundant C═O or COOH groups modifying carbon edges on DGO (Figure S4c,d, Supporting Information).

Figure 1.

a) Schematic diagram of the synthesis processes of SnCu‐DG catalyst. b) iDPC‐STEM image of DGO. c) Raman spectra of GO and DGO. d) O‐groups percentage of GO and DGO. e) The formation energy of Sn or Cu atoms at different sites of defective graphene oxide (DGO). f) High angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM) image of SnCu‐DG (inset figure: size distribution of SnCu clusters). g) Coloring HAADF‐STEM image of SnCu‐DG.

Subsequently, the metal salts (Cu and Sn) were loaded onto GO and DGO using the impregnation method, and their structures were analyzed after pyrolysis, named SnCu‐G and SnCu‐DG. For comparison, Cu‐DG was synthesized using the same method but without the addition of Sn salts. The inductively coupled plasma mass spectrometry (ICP‐MS) was first employed to identify the concentration of Cu and Sn in these three samples. As shown, the three samples exhibit similar mass loading of Cu; meanwhile, the Sn mass loading of SnCu‐G and SnCu‐DG was also controlled at the same level (Table S1, Supporting Information), excluding the interference of metal concentrations during pyrolysis. The XRD patterns of the pyrolytic samples indicate that SnCu‐G exhibits a prominent characteristic peak of metallic Sn, in contrast to Cu‐DG and SnCu‐DG, indicating a more intense and uneven aggregation of metal species on the GO substrate (Figure S5, Supporting Information). Additional STEM images of samples further confirmed the above result. In comparison to the aggregated particles of SnCu‐G (Figures S8, Supporting Information), the Cu‐DG and SnCu‐DG exhibit an amorphous structure and a uniform size distribution of atomic clusters (Figure 1f and Figures S6 and S7, Supporting Information), indicating that the introduced defects and O‐groups could significantly affect the migration and aggregation of metal species during pyrolysis. To gain an in‐depth understanding of the formation mechanism of SnCu clusters, density functional theory (DFT) calculations were employed. Theoretically, the O‐groups at edge sites of DGO could induce different coordination structures (e.g., the chelate coordination of M‐O2/M‐C2 modes) compared with the O‐groups in in‐plane sites (M‐O mode) (Figure S9, Supporting Information). DFT calculation results indicate that the thermal stability of M‐O2/M‐C2 sites is higher than that of M‐O sites for both Cu and Sn, which could construct potential nucleation centers and targeted migration pathways for subsequent aggregation of metal atoms, corresponding to our experimental results (Figure 1e). Notably, the HAADF‐STEM images of SnCu‐DG show that the SnCu clusters are primarily dispersed along the edge of carbons with a small and narrow average size of 1.61 ± 0.55 nm (Figure 1f,g and Figure S6, Supporting Information), further confirming that the strong rivet effect of edge defects is conducive to capturing metal species, which, in turn, serves as the nucleation center for the formation of SnCu clusters. Combining experimental and DFT calculation results, it is clear that the introduced O‐groups modifying edge defects influence the targeted migration and aggregation of metal species by constructing multiple coordination structures with varying thermal stability, resulting in the formation of SnCu atomic clusters along the defect edges.

The chemical states of Cu and SnCu clusters on graphene and defective graphene were investigated using X‐ray absorption spectroscopy (XAS) to reveal the impacts of carbon defects on the valences and electronic structures of atomic clusters. The white line peak and the normalized first derivative of the Cu K edge X‐ray absorption near‐edge spectra (XANES) suggested the presence of metallic Cu and weakly oxidized Cu in Cu‐DG and SnCu‐DG (Figure 2a,b).^{[ 37} , ^{38 ]} The characteristic peaks of Cu species among catalysts are quite closed. Therefore, to further evaluate the value and electrocatalytic behavior of metal species in catalysts, we tested the CV curves of all catalysts (Figure S10, Supporting Information). As shown, the CV profiles within the potential region (≈0–0.9 V vs. RHE) irrelevant to nitrate reduction exhibit the redox behavior of the metal species. Notably, all three catalysts display distinct reduction peaks associated with Cu species. Among them, SnCu‐DG shows the highest reduction potential, indicating a greater presence of low‐valence Cu species (e. g. Cu⁺).^{[ 39 ]} The extended X‐ray absorption fine structure (EXAFS) spectra of the samples were further examined. The first coordination sphere peak of Cu in Cu‐DG and SnCu‐DG was located at ≈1.61 Å which could be attributed to Cu–C(O) scattering paths, higher than that of SnCu‐G (1.51 Å). Moreover, the Cu‐Cu peak in Cu‐DG is slightly higher than that in the reference Cu foil, which may be induced by the unsaturated coordination of Cu‐DG with the carbon edges. After the introduction of Sn atoms, a slightly red shift was observed in SnCu‐DG due to the formation of Sn‐Cu bonds. The normalized first derivative of XANES of samples and references were depicted in Figure 2c. A slight difference between the normalized absorbance at 8981 eV of SnCu‐DG and that of Cu‐DG might be attributed to the introduction of Sn atoms, which induced a charge redistribution on the Sn‐Cu‐C interface. The valence state of Sn in samples was further investigated. The absorption edge of Sn K edge XANES spectra of SnCu‐DG shifts toward higher energies compared to that of SnCu‐G, suggesting a higher average Sn valence state in the SnCu‐DG. Notably, the formation energy of Sn‐C2 is much lower than that of Cu‐C2, indicating that the formation of Sn‐C coordination is significantly easier. Therefore, the prominent peak at ≈1.66 Å in the EXAFS spectra of SnCu‐DG can be attributed to the coordination between Sn and carbon edges. In comparison, the Sn species in SnCu‐G exhibit a more complex valence state. A peak at ≈1.90 Å was observed, which is close to the Sn‐O bond in the SnO, which is due to the surface oxidation of Sn.^{[ 39 ]} A small peak at ≈2.69 Å was observed in SnCu‐DG, which is shorter than the Sn‐Sn bond (≈2.78 Å) of the Sn foil sample and could be assigned to Sn‐Cu bond. The Cu and Sn K‐edge EXAFS oscillations of the materials were analyzed using wavelet transform (WT) analysis (Figure 2f, Figures S11 and S12, Supporting Information). The SnCu‐DG exhibits an obvious signal in the range of ≈5–10 Å⁻¹, similar to that of the reference Cu foil, which can be assigned to the Cu‐Cu or Cu‐Sn bonds. Moreover, the wavelet‐transformed Sn K‐edge EXAFS spectra show a blue‐shifted Sn‐Cu signal at the peak of 7.4 Å⁻¹ in SnCu‐DG, in contrast to the Sn‐Sn signal at the peak of 8.5 Å⁻¹ in the Sn foil. Clearly, the valence states and microenvironment of Sn and Cu atoms in SnCu‐G are more complex than those in SnCu‐DG due to the random migration and aggregation of SnCu particles on graphene.

Figure 2.

a) Cu K edge XANES spectra, b) Cu K edge EXAFS spectra in R space and c) derivative of XANES spectra of Cu‐DG, SnCu‐G, SnCu‐DG, Cu foil, Cu₂O, CuO, respectively. d) Sn K edge XANES spectra and e) Sn K edge EXAFS spectra in R space Sn foil, SnO, SnO₂, SnCu‐G, SnCu‐DG, respectively. f) Wavelet transformed EXAFS spectra for Cu foil, Sn foil, SnCu‐DG, respectively. g) Net Bader charges of the atoms in the Cu‐DG, SnCu‐G, SnCu‐DG (Blue spheres: Cu; Green: Sn; Brown spheres: C).

To gain an in‐depth understanding of the potential charge interactions among Sn, Cu, and carbon defects, the net Bader charges of the atoms in Cu‐DG, SnCu‐G, and SnCu‐DG were investigated (Figure 2g). Three models, as shown in Figures S13–S15, Supporting Information, were established and optimized using ab initio molecular dynamics (AIMD) simulations based on the above STEM and XAS results.^{[ 40 ]} The Cu cluster and SnCu cluster on the DG support were anchored to the carbon edges to form Cu‐DG and SnCu‐DG, while the SnCu cluster was distributed on the in‐plane carbon to serve as SnCu‐G. As shown, the charge distribution between the cluster and carbon substrate can be manipulated after introducing the Sn atom. Compared with Cu‐DG, the introduction of Sn doping in the SnCu‐DG not only donates charges to the edge carbon atoms but also donates charges to the adjacent Cu atoms, thereby weakening their charge transfer ability to the carbon atoms. Additionally, the charge transfer capacity from the Cu sites distant from the Sn site to the edge carbon atoms was enhanced, which exacerbated the asymmetrical distribution of charges among the Cu sites, resulting in the formation of an asymmetrical Sn‐Cu‐C interface. Moreover, the charge transfer ability of Sn at defect edges is significantly stronger than that in the graphene plane, with an increased charge value from 0.284 (SnCu‐G) to 0.569 (SnCu‐DG), which also corresponds to the higher valence state of Sn in SnCu‐DG. These results clearly demonstrate that controlling the location of metal species on carbon, particularly at the edge sites, would induce a SMSI than at the in‐plane sites. Moreover, an asymmetrical distribution of charges among metal sites could be manipulated based on the SMSI by introducing Sn atoms. Predictably, such tunable and SMSI would be an effective way to optimize the reaction pathways and adsorption energy of intermediates, which may be beneficial for electrocatalytic reactions involving multiple electron steps.^{[ 41 ]}

While the neutral pH environment inherently aligns with the typical conditions of major nitrate‐contaminated water sources (e.g., groundwater, agricultural runoff), the kinetically sluggish nitrate reduction reaction (NO₃RR) under these conditions presents a significant challenge. Thus, developing catalysts capable of operating efficiently at neutral pH is paramount for practical application. As a proof‐of‐concept reaction, neutral NO₃RR was used to investigate the value of the carbon defect‐induced SMSI for multiple electron electrocatalytic reactions. The NO₃RR performance of samples was evaluated in an airtight H‐type cell with a neutral electrolyte of 0.1 M K₂SO₄ and 0.5 M KNO₃ solution. The linear sweep voltammetry (LSV) curves of SnCu‐DG exhibit a more pronounced difference in the 0.1 M K₂SO₄ solution compared to Cu‐DG and SnCu‐G, indicating a stronger ability to suppress the competing reaction (HER) (Figure 3a). The possible gas or liquid products are quantified using gas chromatography (GC) and colorimetric methods after plotting their standard calibration curves. As shown, the SnCu‐DG exhibits significantly higher NH₄ ⁺ faradaic efficiencies (FEs) than that of Cu‐DG and SnCu‐G in the range over −0.75 to −0.95 V, peaking at 99.5% FE at −0.8 V, comparable to state‐of‐the‐art NO₃RR catalysts (Figure 3b,g). In contrast, the Cu‐DG and SnCu‐G catalysts show similar and low NH₄ ⁺ FEs ≈40%–55%, but they demonstrate higher FEs for NO₂ ^⁻ products than those of SnCu‐DG (Figure 3c, Figures S16–S19, Supporting Information). Such results indicate that the combination of carbon defects and trace Sn doping on Cu clusters is the key to achieving high NH₄ ⁺ FEs. Moreover, the FEs of H₂ are remarkably suppressed after introducing Sn into Cu‐DG, which also applies to SnCu‐G, demonstrating the strong effect of Sn doping in restraining HER (Figure 3d, Figures S20–S22, Supporting Information). As a result, the SnCu‐DG possesses a high overall NH₄ ⁺ yield rate (Figure 3e), which is ≈6.5 and 15 times higher than that of Cu‐DG and SnCu‐G (at 0.8V), respectively. Moreover, SnCu‐DG exhibits much higher intrinsic activity at Cu sites, with a normalized yield rate of 2.61 × 10⁻¹⁷ mmol h⁻¹ site_Cu ⁻¹, compared to Cu‐DG of 3.36 × 10⁻¹⁸ and SnCu‐G of 1.63 × 10⁻¹⁸ mmol h⁻¹ site_Cu ⁻¹ (Figure 3f). The normalized yield rate of Cu site in SnCu‐DG is more than 1 order of magnitude than the reported Cu‐based catalysts (Table S4, Supporting Information), demonstrating the superiority of MAC in catalyzing multi‐step reactions.

Figure 3.

a) LSV polarization curves of Cu‐DG, SnCu‐G, SnCu‐DG catalysts in HER and NO₃RR b) NH₄ ⁺ faradaic efficiencies, c) NO₂ ⁻ faradaic efficiencies d) H₂ faradaic efficiencies and e) NH₄ ⁺ production rate of Cu‐DG, SnCu‐G, SnCu‐DG catalysts in 0.1 M K₂SO₄ electrolyte with 0.5 M KNO₃. f) Normalized NH₄ ⁺ production rate of Cu‐DG, SnCu‐G, SnCu‐DG catalysts. g) Performance comparison of SnCu‐DG with reported advanced electrocatalysts. h) Stability test of SnCu‐DG for 5 electrolysis cycles at −0.80 V vs RHE.

We further investigated the influence of the Sn content on the electrocatalytic performance of NO₃RR by systematically adjusting the precursor feed ratios. Specifically, we varied the Cu/Sn molar ratio to control the elemental composition in the synthesized samples, named Sn_LCu‐DG and Sn_HCu‐DG, corresponding to the decreased and increased Sn content, respectively (Table S5, Supporting Information). Electrochemical evaluations revealed that the linear sweep voltammetry (LSV) curves of Sn_HCu‐DG and Sn_LCu‐DG are similar, with only a slight positive shift observed for the higher Sn content sample (Figure S23a, Supporting Information). Further analyses of NH₄ ⁺ and NO₂ ⁻ products via colorimetric methods demonstrated that both decreasing (NH₄ ⁺ faradaic efficiencies 77.77%) and increasing (NH₄ ⁺ faradaic efficiencies 76.68%) the Sn content from an optimal level led to a reduction in NH₄ ⁺ selectivity (Figure S23b–d, Supporting Information). These results suggest that an appropriate Sn/Cu molar ratio is critical for constructing effective Sn–Cu–C asymmetric interfaces. The optimal Sn doping enhances the catalytic activity and selectivity for ammonia production by finely tuning the catalyst's electronic structure and intermediate stabilization, whereas excess or insufficient Sn impairs performance. Moreover, to exclude the key contribution of potential single atom sites toward NO₃RR, Sn‐DG was synthesized as a comparative sample (Figure S24a,b, Supporting Information), ensuring that all catalysts—Sn‐DG, Cu‐DG, and SnCu‐DG—were prepared under identical conditions and on the same substrate. This consistency helps to eliminate the effects of synthesis procedures and carbon defect influences on nitrate reduction performance. As shown, the catalytic performance results revealed that both Sn‐DG and Cu‐DG exhibit significantly lower selectivity for ammonia compared to SnCu‐DG (Figure S24c–f). This difference indicates that the enhanced activity of SnCu‐DG is primarily attributable to the synergistic effect of the SnCu clusters rather than single atom sites. The stability of SnCu‐DG was also investigated at 0.8 V for intermittent, maintaining a high NH₄ ⁺ FE (> 90%) after several cycles, which suggests excellent stability (Figure 3h). We also systematically evaluated the electrochemical stability and reversibility of SnCu‐DG, SnCu‐G, and Cu‐DG catalysts at continuous work conditions. The electrocatalytic results demonstrate that SnCu‐DG exhibits a higher catalytic current density and maintains excellent stability over prolonged operation, compared with control groups (Figure S25a–c, Supporting Information). Specifically, after 48 h of continuous I‐T test, SnCu‐DG retained 93.99% Faradaic efficiency for NH₄⁺ production (Figures S25d, Supporting Information). These findings confirm the robust electrochemical stability and reversibility of SnCu‐DG under reaction conditions.

Although the synergistic effect of carbon defects and SnCu atomic clusters was demonstrated to be the reason for the high NO₃RR FEs of SnCu‐DG, the intrinsic contribution and role of each factor during electrocatalysis remain unclear. Experimentally, we first employed in situ attenuated total reflectance infrared (ATR‐IR) spectrometry to track the electrocatalytic processes of SnCu‐DG (Figure 4a) and SnCu‐G (Figure 4b), aiming to clarify the function of carbon defects. As shown, these two in situ spectra exhibit similar C background signals, including the sp²‐hybridized C = C (in‐plane stretching, ≈1500–1600 cm⁻¹) and carboxyl (COOH and/or H₂O) (≈1600–1750 cm⁻¹).^{[ 42 ]} Meanwhile, the peaks at 1275 cm⁻¹ were observed in both SnCu‐DG and SnCu‐G, which can be assigned to the *NO₂ ⁻ signal.^{[ 43 ]} Notably, the *NO₂ ⁻ signal can be observed at a wide potential range in SnCu‐DG, while it only appeared at high potentials (>−0.75 V) in SnCu‐G.^{[ 44} , ^{45 ]} carboxyl (COOH ≈1600–1750 cm⁻¹) or ketones (C═O, ≈1750–1850 cm⁻¹), Moreover, the *NO₂ ⁻ signal is enhanced in SnCu‐G with increasing applied potentials, exhibiting a positive correlation with its NH₄ ⁺ faradaic efficiency experimentally. These phenomena indicate that the introduction of carbon defects in SnCu‐DG can enhance the adsorption ability of *NO₂ ⁻ intermediates, particularly at the low applied potential range (≈−0.3–−0.75 V), which is conducive to the subsequent hydrogenation steps. Unlike SnCu‐DG, there are obvious signal changes in SnCu‐G, which may be related to the potential‐dependent water dissociation behaviors.^{[ 46 ]} Clearly, SnCu‐G exhibits an obvious H₂O signal at ≈3363 cm⁻¹ at the applied range of ≈−0.3–−0.75 V, while it disappeared at higher applied potential. In contrast, a much weaker H₂O signal could be observed in the in situ FTIR spectra of SnCu‐DG. Moreover, new signals were generated at ≈1750 and ≈1800 cm⁻² in SnCu‐G after the H₂O signal disappeared. Those signals could be assigned to the which may be raised from the structure reconstruction or surface functionalization of carbon induced by the *OH of water dissociation.^{[ 42 ]} For comparison, these signals can be found in SnCu‐DG at a lower applied potential range, demonstrating its stronger ability for water dissociation. Combined with the DFT and experimental results, it is clear that the SMSI effect between SnCu clusters and carbon defects in SnCu‐DG could not only enhance the adsorption capacity for *NO₂ ⁻ but also facilitate the water dissociation to provide sufficient activated *H, thereby promoting the formation of NH₄ ⁺.

Figure 4.

In situ attenuated total reflectance infrared (ATR‐IR) spectra of a) SnCu‐G and b) SnCu‐DG at various potentials.

To gain insights into the SMSI effect of SnCu‐DG on NO₃RR, density functional theory (DFT) calculations were further employed to elucidate the effects of the carbon defect and Sn doping, respectively. As we discussed above, net Bader charges results demonstrate that the introduction of carbon defects could induce a SMSI while Sn doping initiates an asymmetrical distribution of charges among the surrounding Cu sites. Moreover, the redistribution of charges in SnCu‐DG was further demonstrated by the differential charge density distribution (Figure S26, Supporting Information). Therefore, the free energies of these three models during the NO₃RR process were further calculated to investigate the effects of defects and Sn doping in regulating the reaction pathway and adsorption energies of intermediates (Figures S27–S29, Supporting Information). As shown in Figure 5a, the hydrogenation of *NO₂ ⁻ is the rate‐determining step (RDS) of NO₃RR in both Cu‐DG and SnCu‐G, with the free energy changes of 0.82 eV and 0.69 eV, respectively. However, the hydrogenation step of *NO₂ ⁻ in SnCu‐DG is spontaneous, meanwhile, the RDS is changed to the step of the formation of *NH₂ from *NH₂+OH intermediates, which is favorable to restrain the formation of NO₂ ⁻ products. Notably, the *NH₂OH intermediates tend to exist in the SnCu‐G, while it would split to *NH₂ and *OH in the Cu‐DG and SnCu‐DG, indicating that the reaction pathway changes when using defective carbon support. Considering that the competitive adsorption of NO₃ ⁻ and NO₂ ⁻ on the surface of catalysts is the key to the subsequent hydrogenation or desorption of *NO₂ ⁻, the difference (ΔG_NO3‐ – ΔG_NO2‐) between the free‐energy changes (ΔG) of NO₃ ⁻ and NO₂ ⁻ for the samples were then calculated. A higher ΔG_NO3‐ – ΔG_NO2‐ signifies a stronger NO₂ ⁻ adsorption energy than that of NO₃ ⁻ adsorption energy. As shown, SnCu‐G exhibits a much higher value of ΔG_NO3‐ – ΔG_NO2‐ than that of Cu‐DG, and it could be further enhanced by introducing carbon defects (SnCu‐DG), demonstrating that the Sn doping is critical for enhancing *NO₂ ⁻ adsorption and lowering the difficulty of *NO₂ ⁻ hydrogenation (Figure 5b and Table S6, Supporting Information). It is worth noting that, despite SnCu‐G exhibiting a lower RDS and higher ΔG_NO3‐ – ΔG_NO2‐ value, the experimental FEs of NO₂ ⁻ products are still high and similar to that of Cu‐DG, which may be caused by an insufficient supply of activated *H, as shown by the in situ FTIR results. Therefore, the energy barriers of water dissociation on the catalysts were investigated. Compared to SnCu‐G (0.55 eV), the Cu‐DG (0.17 eV) and SnCu‐DG (0.16 eV) exhibit much lower water dissociation energies, reconfirming a possible insufficient supply of protons on SnCu‐G, particularly at the low applied potential range (Figure 5c). Therefore, although the higher ΔG_HER – ΔG_NO3RR value of SnCu‐G implies that it favors the NO₃RR route over the HER, the high‐water dissociation energy restrains it's *NO₂ ⁻ hydrogenation, resulting in the high NO₂ ⁻ FEs (Figure 5d and Table S7, Supporting Information). These results demonstrated that 1) the synergy between carbon defects and Sn doping not only reduces the barrier of the RDS but also improves the adsorption ability of *NO₂ ⁻ intermediates competing with the high concentration NO₃ ⁻; 2) the carbon defects can potentially change the reaction pathway of NO₃RR; 3) the introduction of carbon defects can dramatically facilitate water dissociation to provide sufficient activated *H for hydrogenation in electrocatalysis.

Figure 5.

a) Free energy diagrams for the NO₃RR to NH₃ on Cu‐DG, SnCu‐G, and SnCu‐DG. b) The differentiation between the free energy changes of NO₃ ⁻ and NO₂ ⁻ in Cu‐DG, SnCu‐G, and SnCu‐DG. c) The energy barriers of H₂O dissociation of Cu‐DG, SnCu‐G, and SnCu‐DG. d) The differentiation between the free energy changes of HER and NO₃RR in Cu‐DG, SnCu‐G, and SnCu‐DG.

3. Conclusion

In summary, this work develop a defect‐mediated atomic trapping strategy for synthesizing SnCu atomic clusters anchored at edge defect sites of defective graphene, achieving exceptional performance in NO₃RR with a high neutral NH₃ Faradaic efficiency (99.5%) and high Cu‐site activity. By utilizing carbon defects as nucleation sites and nanoscale reactors, we successfully tackle the challenges of metal aggregation inherent to conventional pyrolysis strategy. Operando FTIR and DFT studies reveal that the Sn‐Cu‐C interface induces asymmetric charge polarization, enabling the simultaneous optimization of *NO₂ ⁻ adsorption and hydrogenation kinetics. The principles unveiled through carbon defect‐enabled cluster assembly and interfacial charge engineering open transformative avenues for advancing complex electrocatalytic processes, ranging from sustainable nitrogen‐cycle interventions to energy‐intensive redox transformations.

4. Experimental Section

Experimental Subheading: Experimental Details. References are superscripted and appear after the punctuation.^{[ 6 ]}

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Wu Q., Han Y., Wu L., et al. “Constructing Asymmetric Sn‐Cu‐C Interface via Defective Carbon Trapped Atomic Clusters for Efficient Neutral Nitrate Reduction.” Adv. Mater. 37, no. 36 (2025): 37, 2505743. 10.1002/adma.202505743

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.