Self-enhanced localized alkalinity at the encapsulated Cu catalyst for superb electrocatalytic nitrate/nitrite reduction to NH₃ in neutral electrolyte

Abstract

The electrocatalytic nitrate/nitrite reduction reaction (eNO_x⁻RR) to ammonia (NH₃) is thermodynamically more favorable than the eye-catching nitrogen (N₂) electroreduction. To date, the high eNO_x⁻RR-to-NH₃ activity is limited to strong alkaline electrolytes but cannot be achieved in economic and sustainable neutral/near-neutral electrolytes. Here, we construct a copper (Cu) catalyst encapsulated inside the hydrophilic hierarchical nitrogen-doped carbon nanocages (Cu@hNCNC). During eNO_x⁻RR, the hNCNC shell hinders the diffusion of generated OH⁻ ions outward, thus creating a self-enhanced local high pH environment around the inside Cu nanoparticles. Consequently, the Cu@hNCNC catalyst exhibits an excellent eNO_x⁻RR-to-NH₃ activity in the neutral electrolyte, equivalent to the Cu catalyst immobilized on the outer surface of hNCNC (Cu/hNCNC) in strong alkaline electrolyte, with much better stability for the former. The optimal NH₃ yield rate reaches 4.0 moles per hour per gram with a high Faradaic efficiency of 99.7%. The strong-alkalinity-free advantage facilitates the practicability of Cu@hNCNC catalyst as demonstrated in a coupled plasma-driven N₂ oxidization with eNO_x⁻RR-to-NH₃.

A confined catalyst gets rid of the dependence on strong alkaline electrolyte in electroreduction of nitrate/nitrite to ammonia.

INTRODUCTION

Ammonia (NH₃) is one of the most important fundamental feedstocks in the world, which has wide applications in chemical and fertilizer industries and is also a promising energy carrier of the hydrogen economy (1, 2). The Haber-Bosch synthesis of ammonia is an energy-, carbon-, and capital-intensive process, which accounts for 1 to 2% of global energy consumption and is also poorly compatible with distributed agriculture (3, 4). Electrocatalytic nitrogen reduction reaction (eNRR) to NH₃ powered by renewable electricity is a potential alternative to the Haber-Bosch process (5–7). However, the strong N ≡ N bonding and the poor solubility of N₂ in electrolytes are unfavorable in thermodynamics and kinetics for NH₃ production (8–10). In contrast, the NH₃ production via electrocatalytic nitrate/nitrite reduction reaction (eNO_x⁻RR) is more favorable in thermodynamics and kinetics, demonstrating the superior NH₃ yield rate and Faradaic efficiency (FE) to eNRR in aqueous electrolytes (11–13). In addition, the eNO_x⁻RR-to-NH₃ provides a win-win opportunity for distributed fertilizer synthesis and nitrogen pollution control in water with renewable electricity (fig. S1) (14–17).

The eNO_x⁻RR is accompanied by the competitive hydrogen evolution reaction (HER) due to their overlapping reaction potentials, and they both are pH-sensitive (18, 19). Usually, the strong alkaline condition could accelerate the reaction kinetics of the eNO_x⁻RR and suppress the HER (20–23). Hence, most of the excellent eNO_x⁻RR-to-NH₃ performances to date have been achieved in strong alkaline electrolytes (24–26). For example, for copper catalysts, theoretical and experimental studies verified the suitable adsorption of intermediates (e.g., ^*NOH), the favorable kinetics, and the suppressed HER in alkaline conditions, in comparison with the cases in acidic or neutral conditions, leading to better eNO_x⁻RR-to-NH₃ selectivity and activity for the former (22, 23). However, the use of strong alkaline electrolytes is accompanied by many drawbacks unfavorable for practical applications such as the side reaction with ubiquitous CO₂, high corrosion to equipment, and high cost. Therefore, it is desirable to develop highly efficient eNO_x⁻RR-to-NH₃ catalysts suitable for neutral/near-neutral electrolytes.

As is known, the catalytic performance highly depends on the local chemical environment around the catalyst (27–32). eNO_x⁻RR-to-NH₃ is accompanied by the OH⁻ formation via the reaction of NO_x⁻ + (x + 3)H₂O + (2x + 2)e⁻ → NH₃ + (2x + 3)OH⁻ (12). In principle, this process could increase the local pH around the catalytic sites and inhibit the competing HER, thus enhancing the eNO_x⁻RR-to-NH₃ selectivity and activity. However, for conventional catalysts, usually, the in situ formed OH⁻ will rapidly diffuse into the bulk electrolyte driven by the concentration gradient, leading to the slow enhancement of local pH and catalytic performance thereof. Increasing the surface roughness of catalysts, e.g., by nanosizing the particles or constructing surface microstructures, could decelerate the OH⁻ diffusion and enhance the local pH to some extent (19, 33), which is still not enough due to the open structures, especially under stirring or flowing conditions. Hence, how to effectively confine the OH⁻ around the catalytic sites to increase the local pH is a key yet challenging topic.

In recent years, we have developed hierarchical carbon nanocages (hCNC) featuring three-dimensional (3D) porous structure, high conductivity, large specific surface area, and convenient doping, which can facilitate the mass/charge synergetic transfer, thus becoming an advanced platform for energy conversion and storage (34–36). This unique platform, especially the large cavities of carbon nanocages and abundant microchannels across the shells, provides an ideal “semi-closed” environment to encapsulate the eNO_x⁻RR-to-NH₃ active species (e.g., Cu, the most promising catalyst with low-cost and inferior HER competition) inside and prevents the in situ generated OH⁻ from escaping out of the nanocages, as illustrated in Fig. 1. With this strategy, in this study, we encapsulated Cu nanoparticles inside the hydrophilic hierarchical nitrogen-doped carbon nanocages (hNCNC), and the so-constructed confined catalyst of Cu@hNCNC demonstrates a self-enhanced localized high pH environment around the inside Cu nanoparticles. Consequently, the Cu@hNCNC catalyst exhibits an excellent eNO_x⁻RR-to-NH₃ activity even in the neutral electrolyte, comparable to the Cu catalyst immobilized on the outer surface of hNCNC (Cu/hNCNC) with an “open” environment in the strong alkaline electrolyte, with much better stability for the former. In the neutral electrolyte with 1 M NO_x⁻, Cu@hNCNC catalyst exhibits an excellent NH₃ yield of up to 4.0 mol hour⁻¹ g⁻¹ with a high FE of 99.7%. The strong-alkalinity-free advantage suggests the potential application, and the practicability of the Cu@hNCNC catalyst is demonstrated by the coupled plasma-driven N₂ oxidization with eNO_x⁻RR-to-NH₃ for NH₃ synthesis.

Fig. 1. Schematic illustration of the self-enhanced localized alkalinity within the cavities of hNCNC for Cu@hNCNC catalyst during eNO_x⁻RR.

(A) Cu nanoparticles are encapsulated inside the hierarchical nitrogen-doped carbon nanocages (hNCNC). (B) The in situ generated OH⁻ ions accumulate within the cavities of hNCNC due to the hindered escaping, leading to the self-enhanced local high OH⁻ concentration (C_OH−) inside the nanocages. The C_OH− outside the hNCNC increases slowly due to the hindered OH⁻ escaping from the inside and the bulk of neutral electrolyte outside.

RESULTS

Structural characterization and pH-dependent eNO₃⁻RR-to-NH₃ performances

The hydrophilic hNCNC with an N content of 8.6 atomic % is used to ensure the convenient entrance of the aqueous electrolyte into the cavities of nanocages (fig. S2). Cu@hNCNC and Cu/hNCNC catalysts, with Cu nanoparticles inside and outside the hNCNC respectively, were constructed to evaluate the effect of the catalytic microenvironment on the eNO₃⁻RR-to-NH₃ performance, as shown in Fig. 2. The Cu nanoparticles were encapsulated inside the hNCNC via vacuum filling or immobilized on the outer surface of hNCNC by microwave-assisted ethylene glycol (EG) reduction (see the “Materials synthesis” section in Materials and Methods). Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) characterizations show that, for Cu@hNCNC, most cavities of hNCNC are filled with Cu nanoparticles of ~10.7 nm in average size without observable Cu nanoparticles on the outer surface of hNCNC (Fig. 2, A and B, and fig. S3). The lattice fringes with the interplanar distance of 2.1 Å can be assigned to the (111) planes of metallic Cu (JCPDS no. 04-0836) (Fig. 2B). In addition to the Cu nanoparticles, there is still enough space inside the nanocages for catalytic reactions. For Cu/hNCNC, numerous high-dispersed Cu nanoparticles of ~9.8 nm are immobilized on the outer surface of hNCNC (Fig. 2, C and D, and fig. S3). The mass loadings of Cu nanoparticles in Cu@hNCNC and Cu/hNCNC are nearly identical, about 20 wt %, as revealed by the thermogravimetric analysis (fig. S4). The two catalysts show similar resistances (fig. S5).

Fig. 2. Structural characterizations of Cu@hNCNC and Cu/hNCNC.

(A and B) Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images of Cu@hNCNC. Inset in (A) is the low-magnification view of Cu@hNCNC, and inset in (B) is the typical high-resolution TEM image of Cu nanoparticles. (C and D) SEM and TEM images of Cu/hNCNC. (E) Operando Raman spectra. (F and G) Operando x-ray absorption spectroscopy analysis with the first derivative of x-ray absorption near edge structure (F) and k³-weighted Fourier transformed (FT) extended x-ray absorption fine structure spectra of Cu K-edge (G). Note: The data of Cu, Cu₂O, and CuO in (E) and (F) are presented for comparison. a.u., arbitrary units.

The Cu@hNCNC and Cu/hNCNC maintain the hierarchical 3D porous structure of the hNCNC support (Fig. 2, A to D, and fig. S3, A to C). All the hNCNC, Cu@hNCNC, and Cu/hNCNC display the IV-type N₂ adsorption-desorption isotherms with obvious hysteresis, showing the coexisting micro-meso-macropores (fig. S6). For Cu@hNCNC, the mesoporous distribution shifts to the small-sized side in comparison with the case for Cu/hNCNC and hNCNC as expected since partial inside cavities of hNCNC are occupied by the Cu nanoparticles. Such an integrated architecture of hierarchical carbon scaffold and interconnected multiscale pores is beneficial to the synergic charge/mass transfer during the electrocatalysis, favorable for the reaction kinetics (34–36). The micropores across the nanocage wall provide the diffusion channels of reactants into the nanocages for electrocatalytic reaction, meanwhile hinders the diffusion of products from the inside to outside of the nanocages, i.e., the “blocking effect” (figs. S6 and S7 and table S1) (37). As a result, the OH⁻ concentrations (C_OH−) inside the nanocages are much higher than those outside the nanocages, thus significantly affecting the catalytic performance (Fig. 1).

Operando Raman and operando x-ray absorption spectroscopy (XAS) were conducted to characterize the electronic structure and chemical state of Cu species in Cu@hNCNC and Cu/hNCNC in analogy to the pretreatment condition before electrochemical tests of eNO_x⁻RR (figs. S8 and S9). In the Raman spectra of the two catalysts, the initially sharp Raman bands (ascribed to Cu₂O and/or CuO) disappear after the pretreatment at an applied potential of −1.5 V (Fig. 2E). For XAS analysis, the energy of the maximum peak in the first derivative of x-ray absorption near edge structure spectrum is proportional to the valence state of the element. At open circuit potential (OCP), the valence states of Cu in the two catalysts are close to Cu²⁺, which are reduced to Cu⁰ at −1.5 V (Fig. 2F). The extended x-ray absorption fine structure (EXAFS) spectra show that both catalysts feature the Cu-O bonds at OCP, while only the Cu-Cu bond at −1.5 V (Fig. 2G). The fitting results of the EXAFS spectra exhibit similar coordination numbers and bond lengths of Cu-Cu for Cu@hNCNC and Cu/hNCNC after pretreatment, indicating the highly similar structures (table S2). The x-ray photoelectron spectroscopy (XPS) analysis is in high agreement with these operando results (fig. S10). The preceding characterization results indicate that the Cu species is reduced to metallic Cu after pretreatment and the two catalysts have identical electronic structures and chemical states.

pH-dependent eNO₃⁻RR-to-NH₃ performances

The Cu@hNCNC and Cu/hNCNC catalysts have similar active surface areas (fig. S11). The electrochemical performances of the two catalysts were evaluated by the glassy carbon electrode (GCE) with a diameter of 3 mm (working area of 0.07 cm²) (fig. S12). All potentials were versus the Ag/AgCl electrode directly, unless explicitly stated. The linear sweep voltammetry (LSV) curves obtained in the electrolytes with 50 mM NO₃⁻ exhibit much larger current densities for Cu@hNCNC and Cu/hNCNC than those for hNCNC in all pH, with ~0.5-V positive shift of the onset potential, indicating the catalytic function of Cu species to eNO₃⁻RR (fig. S13). The eNO₃⁻RR performance was first evaluated in potentiostatic mode. The dissolved products including NH₃ and NO₂⁻ were quantitatively measured using the colorimetric method (figs. S14 and S15). The eNO₃⁻RR-to-NH₃ performance of Cu@hNCNC and Cu/hNCNC are directly compared taking into account almost the same content and small contribution of the hNCNC support (figs. S4 and S16). In all the electrolytes with pH 7 to 13, the FEs of NH₃ for the two catalysts display “volcano-like” evolutions with increasing the applied potential, reaching the maximum at −1.5 V (fig. S17). A similar performance is obtained on the larger GCE with a working area of 1 cm² (fig. S18). The decays of FEs at −1.6 V are attributed to the increased HER and the decreased catalyst-electrolyte interface thereof due to the coverage of H₂ bubbles (fig. S19). Hence, the optimal potential of −1.5 V was taken for evaluating the eNO₃⁻RR-to-NH₃ performance.

For the Cu/hNCNC, the FE and yield rate of NH₃ increase with increasing pH from 7 to 12, and then markedly decrease at pH 13, with the maximum FE (92%) and NH₃ yield rate (3.4 mol hour⁻¹ g⁻¹) at pH 12 (Fig. 3, A and B). In contrast, the Cu@hNCNC reaches the FE of 90% and the NH₃ yield rate of ~3 mol hour⁻¹ g⁻¹ even at pH 7, close to the optimal performance of the Cu/hNCNC at pH 12 (Fig. 3, A and B). Moreover, the Cu@hNCNC presents a high NH₃ yield rate of 3.5 mol hour⁻¹ g⁻¹ in the pH range of 11 to 12, with the top-level NH₃ partial current densities (j_NH3) (Fig. 3B and table S3). For the cases of pH ≤ 11, Cu@hNCNC also shows better suppression of HER than Cu/hNCNC, as learned by the lower H₂ FE (FE_H2) and H₂ partial current density (j_H2) for the former (Fig. 3C and fig. S20). The FE_H2 of the two catalysts almost drop to 1% at pH 12 and 13, indicating the suppressed HER at high pH (20–23, 38). The decrease of FE_NH3 at pH 13 could be ascribed to the competitive adsorption of OH⁻ on the Cu surface, which results in the decrease of adsorption capacities for NO₃⁻ and thereby the sluggish eNO₃⁻RR kinetics (Fig. 3 and figs. S21 and S22). The kinetics of eNO₃⁻RR is deduced on the basis of the time-dependent concentration change of NO₃⁻ in the electrolyte, which obeys pseudo–first-order kinetics (Fig. 3D and fig. S23). The apparent rate constant (k) for Cu@hNCNC at pH 7 is close to that for Cu/hNCNC at pH 12, indicating a comparable reaction rate. Hence, the Cu@hNCNC catalyst can efficiently express its excellent activity toward eNO₃⁻RR-to-NH₃ in neutral/near-neutral electrolytes, which avoids the unfavorable strong alkaline electrolytes. It is noteworthy that the Cu@hCNC catalyst by encapsulating Cu inside hydrophobic hCNC shows a much lower FE and NH₃ yield rate than Cu@hNCNC catalysts, indicating the crucial role of hydrophilicity of hNCNC from N-doping to ensure the full access of the electrolyte to the encapsulated catalyst (figs. S2 and S24).

Fig. 3. eNO₃⁻RR-to-NH₃ performances of Cu@hNCNC and Cu/hNCNC.

(A to C) Evolutions of NH₃ FE (A), NH₃ yield rate and NH₃ partial current density (B), and H₂ FE (C) with pH of 7 to 13 at −1.5 V (versus Ag/AgCl). (D) Linearized pseudo–first-order kinetic profiles. Note: The error bars in (A) to (C) represent the data range obtained in the repeated tests. The mixture of 0.5 M Na₂SO₄ and 50 mM NaNO₃ was used as the electrolyte.

The catalytic stability at pH 7 and 12 was evaluated at −1.5 V for eight consecutive cycles. The FE and yield rate of NH₃ show negligible changes for Cu@hNCNC while presenting obvious decay after three cycles for Cu/hNCNC. The TEM characterizations show that, after cycling eight times, the Cu nanoparticles are still encapsulated inside hNCNC for Cu@hNCNC while agglomerated severely for Cu/hNCNC (fig. S25). As expected, after testing in the electrolyte for 1 hour, the Cu concentration in the electrolyte for the case of Cu@hNCNC is obviously lower than that for Cu/hNCNC (fig. S26). The superb stability of Cu@hNCNC is attributed to the confinement of hNCNC, which protects the active-Cu species from dissolution and agglomeration (fig. S25).

Evaluation and theoretical simulation of local pH

To get insight into the excellent eNO₃⁻RR-to-NH₃ performance of Cu@hNCNC in neutral/near-neutral electrolytes, we monitored the pH evolution at the electrode surface during the eNO₃⁻RR to deduce the local pH situation inside hNCNC, in combination with finite element simulations, as shown in Fig. 4. A pH meter was placed against the working electrode surface to monitor the pH change, with or without stirring (Fig. 4, A and B). In the potentiostatic test, without stirring, the C_OH− at the surface of the Cu@hNCNC electrode is lower than that of the Cu/hNCNC electrode in the initial stage (~15 min), and reversely later (Fig. 4C). The lower C_OH− for the former in the initial stage results from the retarded OH⁻ ions outward diffusion due to the confinement of hNCNC (the red arrows for Cu@hNCNC in Fig. 4B), and the later reversal results from the more generated OH⁻ ions for the former due to the higher activity of Cu resulting from the local high pH within the cavities of hNCNC. With stirring, the OH⁻ diffusion is accelerated (the red and green arrows in Fig. 4B), leading to the unobvious difference of the C_OH− for Cu@hNCNC and Cu/hNCNC electrodes in the initial stage (~8 min). The subsequent increasing difference of the C_OH− results from the more generated OH⁻ ions for the former due to the higher local pH within the cavities of hNCNC (Fig. 4C and fig. S27). The obvious lower C_OH− with stirring than without stirring is attributed to the more rapid diffusion and dilution of OH⁻ ions into the bulk electrolyte. The pH evolution in a periodic intermittent eNO₃⁻RR with six cycles further confirms these analyses and results (fig. S28).

Fig. 4. Evaluation and theoretical simulation of the local pH for Cu@hNCNC and Cu/hNCNC catalysts.

(A) Configuratioin neutral/near-neutraln for measuring the pH value at the electrode surface (the contact area in the highlighted orange box). (B) Schematic illustration for OH⁻ generation, diffusion, and pH detection. The red, green, and black arrows represent the OH⁻ diffusion out of the cavities of hNCNC, in-between the hNCNC sheets toward the electrode surface, and to the bulk electrolyte, respectively. The black dashed line and the blue arc represent the electrode surface and the pH meter, respectively. (C) The detection of C_OH− at electrode surface during eNO₃⁻RR at −1.5 V (potentiostatic mode) without and with stirring. (D) eNO₃⁻RR-to-NH₃ performances in different intervals without stirring at 80 mA cm⁻² (galvanostatic mode). (E) C_OH− maps after eNO₃⁻RR for 30 min simulated with finite element method (taken from fig. S35 for the electrolyte with an initial pH of 7). Insets at the bottom are the enlarged rectangular areas demonstrating the spatial distribution of C_OH−. The black dots denote different distances from the Cu surface (1 to 9 nm). (F) The simulated spatial distributions of C_OH−. Note: The eNO₃⁻RR was performed in a 30-ml H-type electrolytic cell. The electrolyte is composed of 0.5 M Na₂SO₄ and 50 mM NaNO₃ with an initial pH of 7. The catalyst layer has an area of 0.5 cm × 0.4 cm, with a mass loading of 0.12 mg cm⁻². The stirring rate is 300 rpm.

The preceding experimental results clearly indicate the OH⁻ accumulation within the cavities of hNCNC for Cu@hNCNC due to the hindered outward diffusion of OH⁻ by the nanocage shell with abundant micropores. Taking the galvanostatic test to control the current density in eNO₃⁻RR (fig. S29), using the rotating ring-disk electrode (RRDE) to monitor the pH evolution on the electrode surface (fig. S30), or using phosphate-buffered (PB) solution to exclude the interference of pH fluctuation (figs. S31 to S32), the experimental results also indicate the delayed pH increase on the electrode surface of Cu@hNCNC, which further confirm the “blocking effect” of hNCNC and the local high pH environment around the encapsulated Cu during eNO₃⁻RR thereof. Such a nanocage geometry with self-enhanced local alkalinity makes the Cu@hNCNC an ideal catalyst for eNO₃⁻RR in neutral/near-neutral electrolytes, which is difficult to realize for the supported catalysts, e.g., the Cu/hNCNC, with an open environment (figs. S27 to S30 and S33). Actually, for eNO₃⁻RR-to-NH₃ with an initial pH of 7, the Cu@hNCNC demonstrates superior performances to the Cu/hNCNC in different reaction intervals (Fig. 4D and fig. S34). For Cu@hNCNC, the local high pH leads to high performance in the whole reaction period with a slightly increasing FE and yield rate of NH₃ in the first two intervals. For Cu/hNCNC, the initial low pH leads to low performance in the early period. The FE and yield rate of NH₃ gradually increase with increasing pH during eNO₃⁻RR-to-NH₃ and are close to those of Cu@hNCNC in the later period (approximately the interval of 20 to 30 min) due to the similar pH level. These results demonstrate the strong connection of the catalytic activity with the local pH environment.

The blocking effect of hNCNC is further elucidated by finite element simulations with two 2D models corresponding to Cu@hNCNC and Cu/hNCNC (fig. S35). After eNO₃⁻RR of 30 min, the C_OH− inside hNCNC for Cu@hNCNC is much higher than that outside with a sharp gradient of C_OH− across the shell, while the C_OH− for Cu/hNCNC shows a gradual decrease with increasing the distance from Cu surface (Fig. 4E). Specifically, the C_OH− inside hNCNC is approximately five times higher than that outside for the former (Fig. 4F). This is the effect of a single nanocage in bulk electrolyte, and the C_OH− enhancement should be much strengthened in practical situation with hierarchical porous structure.

On the basis of the above results, the picture for Cu@hNCNC in eNO₃⁻RR could be well figured out. With the initial pH <12, when eNO₃⁻RR is started, OH⁻ is generated. The self-enhanced localized pH inside the hNCNC in turn increases the eNO₃⁻RR activity. Such “positive feedback” leads to the quick increase of the local pH to ~12. The further increase of the local pH (over 12) leads to the drop of the eNO₃⁻RR activity (Fig. 3), i.e., the OH⁻ generation slows down quickly and cannot offset the amount of outward diffusion. Such “negative feedback” results in a stable local high pH of around 12 eventually. This picture is highly consistent with the results of pH measurements and theoretical simulations. For eNO₂⁻RR-to-NH₃, the Cu@hNCNC also exhibited a high FE (94.0%) and yield rate of NH₃ (2.9 mol hour⁻¹ g⁻¹) at −1.5 V in neutral electrolyte, higher than the corresponding data for Cu/hNCNC (fig. S36).

eNO_x⁻RR-to-NH₃ performance of Cu@hNCNC catalyst in neutral electrolyte

The preceding experimental and theoretical results demonstrate the excellent performance of Cu@hNCNC catalyst for both eNO₃⁻RR- and eNO₂⁻RR-to-NH₃ in neutral electrolytes due to the self-enhanced localized high pH around Cu nanoparticles inside hNCNC. Hence, we have further evaluated the eNO_x⁻RR-to-NH₃ performance of Cu@hNCNC in neutral electrolyte with the fixed NO₃⁻/NO₂⁻ concentration ratio of 1:3. With increasing the concentration of NO_x⁻, the current density increases accordingly in the range of 0.1 to 1000 mM, followed by a slight fluctuation in 1000 to 2000 mM (Fig. 5A and fig. S37A). Similarly, both the FE and yield rate of NH₃ monotonously increase in the range of 0.1 to 1000 mM, reaching the maximum of 99.7% and 4.0 mol hour⁻¹ g⁻¹, respectively, and then remain at the high level of >99% and >3.8 mol hour⁻¹ g⁻¹ in 1000 to 2000 mM. The increase of the NH₃ yield rate follows the positive- and zero-order reactions with the low (≤1000 mM) and high (≥1000 mM) concentrations of NO_x⁻, respectively (Fig. 5B and fig. S37B) (39). Importantly, in neutral/near-neutral electrolytes, Cu@hNCNC exhibits high catalytic activity in a wide range of NO_x⁻ concentration, which is suitable for various application scenarios, e.g., the eNO₃⁻RR-to-NH₃ from the wastewater with low NO_x⁻ concentrations, and the industrial-scale NH₃ synthesis by the coupled plasma-driven N₂ oxidization with eNO_x⁻RR-to-NH₃ with tunable NO_x⁻ concentrations. The durability of Cu@hNCNC catalyst in eNO_x⁻RR was evaluated by 50 electrocatalysis cycles in 1000 mM NO_x⁻, which remains the high FE and yield rate of NH₃ with a small fluctuation, indicating the excellent electrocatalytic stability of Cu@hNCNC. In contrast, Cu/hNCNC in strong alkaline electrolytes only presents a comparable yield rate of NH₃ in the initial few cycles, with gradual decay in the following cycles (Fig. 5C and fig. S38). The excellent performance of Cu@hNCNC reaches the top level in neutral electrolytes, which also outperforms those of most non-noble metal electrocatalysts in strong alkaline electrolytes (Fig. 5D and table S3) (15, 20, 21, 24–26, 39–58). Hence, the Cu@hNCNC catalyst can efficiently express its high activity toward eNO_x⁻RR-to-NH₃ in neutral/near-neutral electrolytes with high stability, which avoids the use of unfavorable strong alkaline electrolytes.

Fig. 5. The eNO_x⁻RR-to-NH₃ performance of Cu@hNCNC with different concentrations of NO_x⁻ at −1.5 V in the neutral electrolyte (pH 7).

(A) i-t curves. (B) FE and yield rate of NH₃. (C) Electrocatalytic stability of Cu@hNCNC (red) in 1000 mM NO_x⁻. The corresponding data of Cu/hNCNC (blue) in strong alkaline electrolytes (pH 12) are also presented for comparison. (D) Comparison of the eNO_x⁻RR-to-NH₃ performance of Cu@hNCNC with those of typical electrocatalysts in literature. Note: The NO₃⁻/NO₂⁻ ratio in NO_x⁻ is fixed at 1:3.

Practicability of Cu@hNCNC catalyst in eNO_x⁻RR-to-NH₃

The preceding results exhibit the excellent eNO_x⁻RR-to-NH₃ performance of Cu@hNCNC in neutral electrolytes due to the self-enhanced high local pH around Cu nanoparticles within the cavities of hNCNC. This result indicates the great potential for sustainable NH₃ synthesis as demonstrated by coupling plasma-driven N₂ oxidization with eNO_x⁻RR-to-NH₃ in Fig. 6 (fig. S39). The NO_x gas was generated by the dielectric barrier discharge (DBD) in the air and converted into NO₃⁻/NO₂⁻ in the PB-containing absorber (pH 7), which was then exchanged into the electrolyzer for electroreduction to NH₃ (Fig. 6A). The concentration of NO_x⁻ increases linearly with the plasma operation time and reaches 42.2 mM after 30 min of plasma treatment (Fig. 6B). The subsequent eNO_x⁻RR-to-NH₃ exhibits the high FE (90.5%) and yield rate of NH₃ (2.44 mol hour⁻¹ g⁻¹) for Cu@hNCNC, much better than the corresponding 74.3% and 1.80 mol hour⁻¹ g⁻¹ for Cu/hNCNC, with larger current density for the former (Fig. 6, C and D). This performance of Cu@hNCNC in the coupled system in the neutral condition is comparable to the corresponding ones in literature in the strong alkaline electrolytes, suggesting the potential applications (table S4).

Fig. 6. Demonstration of sustainable NH₃ synthesis by coupling plasma-driven N₂ oxidization with eNO_x⁻RR-to-NH₃.

(A) Schematic illustration of the coupled reaction system. (B) Plots of NO_x⁻ concentration versus plasma operation time. (C) i-t curves. (D) FE and yield rate of NH₃ after 30-min reaction. Note: The air flow rate was 2000 sccm. PB (0.2 M; pH 7) with 0.3 M Na₂SO₄ was used as the absorption solution. The applied potential is −1.6 V (fig. S31).

The self-enhanced local high pH inside hNCNC was also observed in the CO₂ electroreduction reaction for Cu@hNCNC relative to Cu/hNCNC and in the oxygen reduction reaction for Pt@hNCNC relative to Pt/hNCNC (fig. S40). These results indicate the general validity of regulating the local pH environment via carbon nanocage confinement, which provides a new approach to enhance catalytic performances via microenvironment engineering.

DISCUSSION

In summary, we have designed and constructed the confined Cu@hNCNC catalyst by encapsulating Cu nanoparticles inside the hydrophilic hNCNC. Such Cu@hNCNC catalyst can prevent the in situ generated OH⁻ ions during eNO_x⁻RR-to-NH₃ from escaping out of nanocages and thus lead to a self-enhanced localized high pH environment around the Cu nanoparticles, which is confirmed by the experimental results and finite element simulations. Consequently, the Cu@hNCNC catalyst with a semi-closed environment exhibits an outstanding eNO_x⁻RR-to-NH₃ activity even in the neutral electrolyte, equivalent to the Cu/hNCNC catalyst with an open environment in the strong alkaline electrolyte. An excellent NH₃ yield rate of 4.0 mol hour⁻¹ g⁻¹ with FE of 99.7% is achieved in the neutral electrolytes, which outperforms those of most non-noble metal electrocatalysts in strong alkaline electrolytes. The confinement also effectively inhibits the loss and agglomeration of Cu species during eNO_x⁻RR, leading to the excellent stability of Cu@hNCNC catalyst much superior to Cu/hNCNC. The practicability of Cu@hNCNC catalyst is demonstrated by a coupled plasma-driven N₂ oxidization with the eNO_x⁻RR-to-NH₃ process. The integrated advantages of high activity, high FE, and high stability with alkali-free electrolytes suggest an economic and sustainable alternative for ammonia synthesis. Taking into account the unique architecture and properties of the hierarchical carbon nanocages, the convenient strategy in this study, i.e., encapsulating catalytic active species inside the nanocages for achieving a local high pH environment, could be a general way for microenvironment engineering to explore advanced catalysts for some key reactions.

MATERIALS AND METHODS

Materials synthesis

The hCNC and hNCNC were prepared by the in situ MgO template method at 800°C with benzene and pyridine precursor, respectively (59, 60).

The Cu@hNCNC was prepared by a vacuum-filling method (61). In a typical procedure, hNCNC (50 mg) was evacuated to ~10 Pa in a sealed flask, and then, 30 ml of 0.4 M Cu(NO₃)₂ solution was quickly injected into the flask, followed by magnetic stirring for 2 hours. After that, the sample was filtrated, freeze-dried, and then washed three times with deionized water to remove the Cu(NO₃)₂ outside the hNCNC. After the second vacuum-drying process, the sample was reduced to 50 sccm of H₂/Ar (5%/95%) at 220°C for 1 hour, leading to the formation of Cu@hNCNC.

The Cu/hNCNC was synthesized by a microwave-assisted EG reduction method (62). Typically, 20 mg of hNCNC was dispersed ultrasonically into 50 ml of EG, and then 1 ml of 0.4 M Cu(NO₃)₂/EG and 4 ml of 0.2 M NaOH/EG solution was successively added to the suspension and this mixture was stirred for 4 hours. The suspension was irradiated in a microwave oven at 700 W for 40 s followed by aging for 5 min. After filtering, washing with ethanol and water, and drying at 70°C, the Cu/hNCNC was obtained by the reduction in 50 sccm of H₂/Ar (5%/95%) at 220°C for 1 hour. Similarly, the Cu@hCNC and Cu/hCNC were prepared by replacing hNCNC with hCNC for comparison.

Characterization

The morphology and microstructure of the samples were examined by SEM (Hitachi, S-4800, operating at 10 kV) and TEM (JEM-2100, operating at 200 kV). The crystalline structure was determined by an x-ray diffractometer (Bruker, D8 Advance A25, Cu K_a1, λ = 0.154056 nm). The surface composition and element valence were analyzed by XPS (ULVAC-PHI INC, PHI 5000 VersaProbe, Al Kα). N₂ adsorption-desorption isotherms were obtained on Thermo Fisher Scientific Surfer Gas Adsorption Porosimeter at 77 K. hNCNC was degassed at 300°C, while Cu@hNCNC and Cu/hNCNC were degassed at 180°C. The pore size distribution was calculated from the corresponding adsorption branch of N₂ isotherm by the Barrett-Joyner-Halenda method for mesopore and the Horvath-Kawazoe method for micropore. The thermogravimetry (Netzsch, STA449F3) analysis was carried out in 20 vol % O₂-containing Ar of 20 sccm at a rate of 10.0°C min⁻¹. The concentration of Cu dissolved in the electrolyte was determined by inductively coupled plasma optical emission spectrometry (SHIMADZU, ICPE-9810) and the linear correlation of concentration intensity was calibrated using a series of standard copper sulfate solutions with ultrapure water as the solvent. pH was measured by a microelectrode (STMICRO5) on a pH meter (OHAUS, STARTER 3100) and an RRDE (Pine Instrument Co.).

Electrochemical measurements

Five milligrams of catalysts (Cu@hNCNC, Cu/hNCNC, or hNCNC) was dispersed ultrasonically into a mixed solution including 0.8 ml of ultrapure water (18.2 megohms·cm), 0.2 ml of ethanol, and 150 μl of 5% Nafion solution, followed by stirring overnight. Two microliters of the catalyst ink was dropped onto the GCE (3-mm diameter) for natural drying at room temperature. The catalyst mass loading on the working electrode was 0.12 mg cm⁻². Ag/AgCl electrode (3 M KCl) and platinum sheet (1 cm × 1 cm) were used as the reference and counter electrode, respectively. The 0.5 M Na₂SO₄ solution was used as the electrolyte which contained 50 mM NaNO₃ to provide NO₃⁻, with different pH values (7 to 13) adjusted by NaOH or H₂SO₄ solution.

The electrochemical measurements were conducted on a VMP3 electrochemical workstation (BioLogic). The eNO₃⁻RR experiments were performed in the potentiostatic mode using a three-electrode system in a 10-ml H-type electrolytic cell separated by a pretreated proton exchange membrane (Nafion 117). All potentials were versus the Ag/AgCl electrode directly, unless explicitly stated. Under stable pH conditions, the Ag/AgCl potential can be converted to a reversible hydrogen electrode (RHE) scale as follows: E_RHE = E_Ag/AgCl + 0.0592 × pH + 0.209. Ar gas (20 sccm) flowed through the cathodic compartment for 15 min before each electrochemical test and continued in the whole eNO₃⁻RR experiment. Before the first electrochemical measurement, the working electrodes with Cu-based catalysts were pretreated with two potentiostatic cycles of eNO₃⁻RR at −1.5 V for 0.5 hours per cycle. The LSV tests were performed at a scan rate of 20 mV s⁻¹. The chronoamperometric tests were conducted at different potentials for 0.5 hours with a rotation rate of 300 rpm, and the cyclic stability tests were conducted at −1.5 V for 0.5 hours per cycle. Electrochemical impedance spectroscopy measurements were conducted at OCP, and the frequency scan range was from 0.1 Hz to 100 kHz. iR compensation was not conducted in all electrochemical results. The absorption solution of 0.05 M H₂SO₄ was used to collect the evaporated NH₃ in outflowing gas from the cathodic compartment. The NH₃ amounts in the cathodic and anodic compartment electrolytes and in the H₂SO₄ solution were detected and quantified, respectively. The sum was used to calculate the NH₃ yield rate and FE.

Practically, NO₃⁻ and NO₂⁻ coexist in most cases, e.g., in the wastewater, or the NO_x⁻ produced by plasma-driven N₂ oxidization. Hence, the NO₃⁻/NO₂⁻ mixture was used as the raw material to evaluate the potential applications of the catalysts. The NO₃⁻/NO₂⁻ ratio of 1:3 was selected by referring to that in the NO_x⁻ produced by plasma-driven N₂ oxidization in our experiment (Fig. 6B).

Operando Raman test

A single-cell system was used in the operando Raman test. Catalyst-coated GCE was used as the working electrode. Ag/AgCl electrode (3 M KCl) and platinum wire were used as the reference and counter electrode, respectively. The solution of 0.5 M Na₂SO₄ and 50 mM NaNO₃ (pH 7) was used as the electrolyte. The electrolyte was first bubbled by argon for 30 min. Then, it is injected into the electrolytic cell with a circulating flow rate of 10 sccm. eNO₃⁻RR was performed at −1.5 V, and the corresponding Raman spectra were measured using a Horiba JY Evolution spectrometer with a 633-nm laser source.

Operando XAS test

A flow cell system was used in the operando XAS test. The catalyst was coated on the A side of a carbon paper, with the A side facing the chamber and the B side facing outside. The carbon paper was sealed on the outer wall of the flow cell, allowing the x-ray to irradiate directly onto the working electrode. Ag/AgCl electrode (3 M KCl) and nickel foam were used as the reference and counter electrode, respectively. Nafion 117 proton exchange membrane was used as the separator. The solution with 0.5 M Na₂SO₄ and 50 mM NaNO₃ (pH 7) was used as the electrolyte. The flow rate of electrolytes in both the cathode and anode chambers was 10 sccm. eNO₃⁻RR was performed at an OCP of −1.5 V, and the corresponding XAS spectrum at the Cu K-edge was obtained on the BL11B beamline at the Shanghai Synchrotron Radiation Facility. The XAS data were analyzed by using ATHENA.

Determination of products

The possible gas products for eNO₃⁻RR, including H₂, N₂, and O₂, were determined qualitatively by online mass spectrometry (AMETEK DYCOR) and H₂ was quantified by online gas chromatography with a thermal conductivity detector (LU GEN ANALYSIS GC-9860). The liquid ones were quantified by the colorimetry method. The ultraviolet-visible (UV-vis) absorbance spectra were measured on a high-speed spectrometer (PG2000, ideaoptics).

Quantification of NH₃

The NH₃ produced by eNO₃⁻RR was quantified by the indophenol blue method. First, a specific amount of electrolyte in the two compartments (cathodic and anodic) and absorption solution was respectively diluted to 2 ml, to ensure the NH₃ concertation in the detection range. Next, 2 ml of 1 M NaOH solution containing salicylic acid (5 wt %) and sodium citrate (5 wt %) was added and mixed thoroughly, and then 1 ml of 0.05 M sodium hypochlorite and 0.2 ml of sodium nitroprusside (1 wt %) were added. After sitting for 1 hour, the absorption intensity at a wavelength of 653 nm was recorded. The concentration-absorbance curve was calibrated using a series of standard ammonium chloride solutions with the electrolyte (0.5 M Na₂SO₄ and 50 mM NaNO₃) as the solvent.

Quantification of NO₂⁻

A mixture of p-aminobenzenesulfonamide (4 g), N-(1-naphthyl)ethylenediamine dihydrochloride (0.2 g), ultrapure water (50 ml), and phosphoric acid (10 ml, ρ = 1.70 g ml⁻¹) was used as a color reagent. A specific amount of electrolyte was taken out from the cathodic compartment and added into 5 ml of ultrapure water to dilute the detection range. Next, 0.1 ml of color reagent was added to the aforementioned solution and mixed uniformly. After sitting for 20 min, the absorption intensity at a wavelength of 540 nm was recorded. The concentration-absorbance curve was calibrated using a series of standard sodium nitrite solutions with ultrapure water as the solvent.

Quantification of NO₃⁻

A certain amount of electrolyte in the cathodic compartment was diluted to 5 ml to match the range of calibration curves for NO₃⁻. Then, 0.1 ml of HCl (1 M) and 0.01 ml of sulfamic acid solution (0.8 wt %) were added successively and let sit for 30 min. The absorption intensities at 220 and 275 nm were recorded by UV-vis absorption spectroscopy. The final absorbance value was calculated by the following equation: A = A_220nm − 2A_275nm. The concentration of NO₃⁻ was calibrated using a series of standard NaNO₃ solutions with different concentrations.

Calculation of yield rate and FE of NH₃ and NO₂⁻

The yield rate of NH₃ and NO₂⁻ was calculated as

$${\text{NH}}_3 \text{yield rate}=(C_{\text{NH3}}\times V)/(M_{\text{NH3}}\times t\times m_{\text{cat}.})$$

$${{\text{NO}}_2}^{-} \text{yield rate}=(C_{\text{NO2}-}\times V)/(M_{\text{NO2}-}\times t\times m_{\text{cat}.})$$

The FE of NH₃ and NO₂⁻ was calculated as

$${\text{FE}}_{\text{NO3}-\text{to NH3}}=(8\times F\times C_{\text{NH3}}\times V)/(M_{\text{NH3}}\times Q)\times 100\%$$

$${\text{FE}}_{\text{NO2}-\text{to NH3}}=(6\times F\times C_{\text{NH3}}\times V)/(M_{\text{NH3}}\times Q)\times 100\%$$

$${\text{FE}}_{\text{NO3}-\text{to NO2}-}=(2\times F\times C_{\text{NO2}-}\times V)/(M_{\text{NO2}-}\times Q)\times 100\%$$

where C_NH3 and C_NO2− are the concentrations of NH₃ (aq) and NO₂⁻ (aq), respectively; V is the volume of electrolyte in the cathodic or anodic compartment (10 ml); M_NH3 and M_NO2− are the molar masses of NH₃ (17 g mol⁻¹) and NO₂⁻ (46 g mol⁻¹), respectively; t is the electrolysis time (0.5 hours); m_cat. is the catalyst mass on GCE; F is the Faradaic constant (96,485 C mol⁻¹); and Q is the total charge passing the electrode calculated by evaluating the integral over the i-t curves.

For the calculation of FE_NH3 in eNO_x⁻RR, both NO₃⁻ and NO₂⁻ concentrations were quantified before and after the electrolysis. The amount of NO₂⁻ converted from NO₃⁻ was negligible because it was far less than that in the electrolyte. The C_{NH3(from NO3−)} and C_{NH3(from NO2−)} were calculated as

$$C_{\text{NH3}(\text{from NO3}-)}=(C_{\text{NH3}}\times △C_{\text{NO3}-})/(△C_{\text{NO3}-}+△C_{\text{NO2}-})$$

$$C_{\text{NH3}(\text{from NO2}-)}=(C_{\text{NH3}}\times △C_{\text{NO2}-})/(△C_{\text{NO3}-}+△C_{\text{NO2}-})$$

and the FE_NH3 here was calculated as

$${\text{FE}}_{\text{NH3}}={\text{FE}}_{\text{NH3}(\text{from NO3}-)}+{\text{FE}}_{\text{NH3}(\text{from NO2}-)}$$

Calculation of the partial current density for NH₃

The partial current density for NH₃ (j_NH3) was calculated as

$$j_{\text{NH3}}={\text{FE}}_{\text{NH3}}\times \iint \mathit{\text{didt}}/t$$

where FE_NH3 is the FE of NH₃, i is the current during the chronoamperometric tests at given potentials, and t is the electrolysis time.

Calculation of turnover frequency

The turnover frequency (TOF) was calculated as

$${\text{TOF}}_{\text{Cu}@\text{hNCNC}}=J_{\mathrm{k}}/(\mathrm{\alpha }\times n_{\text{Cu}}\times e)$$

where J_k represents the kinetic current (in amperes) calculated by evaluating the integral over the i-t curves in Fig. 6C, α is the number of transfer electrons (8), n_Cu is the number of active Cu, and e is an elementary charge (1.6 × 10⁻¹⁹ C). n_Cu is estimated from the radius of Cu nanoparticle (~5 nm from fig. S3), density (8.96 g cm⁻³) (63), and surface atomic concentration (1.4 × 10¹⁹ m⁻²) (64).

NMR determination of NH₃

The produced NH₄⁺ was quantitatively determined by ¹H nuclear magnetic resonance (¹H-NMR). Dimethyl sulfoxide (DMSO)–d6 was used as the solvent and C₄H₄O₄ as the internal standard. The calibration curve was made by a series of standard NH₄Cl solutions with known concentrations. The standard NH₄Cl solutions were prepared in 0.01 M HCl containing 50 mM NaNO₃. The solution (0.5 ml) was mixed with 0.1 ml of DMSO-d6 containing 0.04 wt % C₄H₄O₄ (20 mg C₄H₄O₄ in 50 g of DMSO-d6). The mixture was tested by a Bruker Avance III 400 MHz spectrometer at room temperature. The calibration curve was achieved using the peak area ratio between NH₄⁺ and C₄H₄O₄. For testing the produced NH₄⁺ from eNO₃⁻RR, the pH of the obtained electrolyte was adjusted to weak acid by 0.01 M HCl before the test. Then, the testing processes are the same as that for making the calibration curve. The amount of produced NH₄⁺ can be calculated from the peak area using the calibration curve.

Isotope labeling experiments

To clarify the N source of NH₃, Na¹⁵NO₃ was used as the feeding ¹⁵N source. Then, the produced ¹⁵NH₄⁺ is determined by the same procedure as that in the NMR determination of the NH₃ section.

pH detection

pH detection with a pH meter

The catalyst ink (5.6 μl) was dropped onto a double side of a carbon paper (0.5 cm × 0.4 cm), followed by natural drying at room temperature. The catalyst mass loading on the carbon paper (working electrode) was 0.12 mg cm⁻². The eNO₃⁻RR experiments were performed using a three-electrode system with the same conditions as the electrochemical measurements above, except for the use of a 30-ml H-type electrolytic cell with controlled stirring. The eNO₃⁻RR was carried out in the potentiostatic mode at −1.5 V or in the galvanostatic mode at 80 mA cm⁻² for 30 min. The pH on the working electrode surface during eNO₃⁻RR was directly detected with a pH meter. For the pH detection in the flow cell system, the electrolyte of 500 ml and the carbon paper of 0.5 cm × 1 cm with the same catalyst loading of 0.12 mg cm⁻² were used, and the pH meter was immersed in the bulk electrolyte.

pH detection with an RRDE

The pH on the electrode surface was measured using the RRDE technique, using a disk electrode (0.2475 cm²) with a Pt ring electrode (0.1866 cm²) (65). The potential of the Pt ring is sensitive to pH change and serves as an indicator for monitoring pH evolution on the surface of the disk electrode. A three-electrode system was used, including the disk electrode in RRDE, Ag/AgCl electrode (3 M KCl), and graphite rod, serving as the working, reference, and counter electrodes, respectively. The pH dependence of the OCP (E_OCP) was then evaluated in H₂-saturated electrolytes including 0.5 M Na₂SO₄ and 50 mM NaNO₃ using the Pt ring electrode. The E_OCP of the Pt ring electrode indicates the equilibrium potential of H⁺/H₂, varying with electrolyte pH

$$E_{\text{SHE}}=-2.303\times R\times T/F\times \text{pH}$$

where R, T, and F are the gas constant, absolute temperature, and Faraday constant, respectively.

The pH measurements were conducted in both single and H-type cells with an H₂-saturated electrolytes including 0.5 M Na₂SO₄ and 50 mM NaNO₃. The catalysts were coated on the disk electrode (working electrode) with a mass loading of 0.12 mg cm⁻². The working electrode rotated at a speed of 0 or 1600 rpm. A galvanostatic test with a current density of 80 mA cm⁻² for 30 min was performed on the disk electrode to obtain a constant number of electrons. E_OCP was simultaneously recorded on the Pt ring electrode. The pH on the disk electrode coated by the catalyst can be deduced from the pH on the Pt ring electrode by the following equation (65)

$$c_{\text{rt},\mathrm{H}+}-c_{\text{rt},\text{OH}-}=N_{\mathrm{D}}(c_{\mathrm{d},\mathrm{H}+}-c_{\mathrm{d},\text{OH}-})+(1-N_{\mathrm{D}})(c_{\infty ,\mathrm{H}+}-c_{\infty ,\text{OH}-})$$

where c_rt,H+ and c_rt,OH− are the concentrations of H⁺ and OH⁻ on the Pt ring electrode, respectively; c_d,H+ and c_d,OH− are the concentrations of H⁺ and OH⁻ on the disk electrode, respectively; c_∞,H+ and c_∞,OH− are the concentrations of H⁺ and OH⁻ in the bulk electrolyte recorded by a pH meter, respectively; and N_D = 0.37 is the collection efficiency of the Pt ring electrode (66).

Plasma electrocatalysis experiments

The plasma electrocatalysis reaction system consists of three parts: a plasma system, an absorber, and an H-type electrolyzer. A DBD system was used to activate and dissociate N₂ and O₂ molecules to produce NO_x. The experimental setup consists of a plasma reactor, a power supply, and an electrical measurement system. The plasma reactor is composed of a quartz chamber and two metal plates that serve as the high-voltage and ground electrodes. The discharge was driven by a plasma generator with a power supply (CTP-2000 K, Nanjing Suman Plasma Technology Co., Ltd). The applied voltage is 80 V. Standard air was fed into the plasma reactor at a constant gas flow of 2000 sccm. The outlet gas was fed into a cylindrical absorber containing electrolytes with 0.2 M PB and 0.3 M Na₂SO₄ (pH 7). This electrolyte was exchanged for the 10-ml H-type electrolyzer with the same electrolyte through a reverse double pump system. The electrochemical measurements are similar to those used in the eNO₃⁻RR, with an optimum potential of −1.6 V (fig. S31). The yield of NO_x⁻ and NH₃ was quantized by UV-vis spectrophotometer measurements.

Theoretical simulation

Theoretical simulations were performed with the COMSOL Multiphysics 5.6 model. The Transport of Diluted Species module was used to estimate the C_OH− near the electrode surface. Two 2D models of the Cu nanoparticles in confined and free states were designed, corresponding to the Cu@hNCNC and Cu/hNCNC (fig. S35). The OH⁻ is constantly generated on the Cu nanoparticle with a surface reaction rate of 3 mol m⁻² s⁻¹, which is estimated by the measured pH on the surface of electrodes in fig. S27. The generated OH⁻ then diffuses to the boundary of the simulation box due to the fixed C_OH− of 10⁻⁷ M (pH 7). The OH⁻ diffusion coefficient is 5.293 × 10⁻⁹ m² s⁻¹, and the simulation box is 40 nm × 40 nm with an initial C_OH− of 10⁻⁷ M (pH 7) inside (67). All the simulations were equilibrated with a time range of 30 min, which is long enough to get the stationary states.

Supplementary Materials

This PDF file includes:

Figs. S1 to S40

Tables S1 to S4

References