Dimensionality junctional Janus nanocrystals with heterogeneous stacking fault–lined interface achieve tandem electroreduction of nitrate

Abstract

Electrocatalytic nitrate reduction reaction (NO₃RR) holds substantial potential to close nitrogen cycle, while the selectivity toward ammonia in NO₃RR is still too low to meet practical applications. Here, we report precise synthesis of a class of Ag-Cu dimensionality junctional Janus nanocrystals (DJJNs) with heterogeneous stacking fault–lined interface for tandem electrocatalytic reduction of nitrate to ammonia. Site-selective growth of Cu with a three-dimensional structure on one edge of two-dimensional Ag triangular nanoplates can be realized via controlling reduction kinetics of the Cu precursor. Quantitative analysis of correlations between the composition and the interface length of Ag-Cu DJJNs and their NO₃RR selectivity toward NH₃ reveals that the optimal composition of Ag/Cu nanocatalysts as well as the dimensionality junctional interface plays a key role in selectivity achievement. Density functional theory calculations reveal that the Ag/Cu(111) interface can work in a tandem manner to catalyze NO₃RR to form NH₃ over Ag-Cu DJJNs. This study offers a synthetic design strategy for advancing tandem catalysts toward NO₃RR.

Engineered Ag-Cu nanocrystals achieve tandem catalysis for sustainable ammonia production from nitrate.

INTRODUCTION

The increasing anthropogenic activities such as extensive use of nitrogen-based fertilizers and industrial chemicals have made nitrogen oxides prevalent contaminants in the aquatic system, disrupting the biogeochemical nitrogen cycle (1, 2). Among these species, nitrate is attracting increasing attention as a starting material for transformation into less harmful or value-added products such as nitrogen gas and ammonia with a more sustainable approach. In particular, the electrochemical nitrate reduction to ammonia is investigated owing to its mechanistic complexity as an eight-electron and nine-proton process (3–5). As such, this reaction serves as a model system for studying advanced catalyst design principles in complicated reactions that involve selectivity control, multiple reaction intermediates, and multistep proton-coupled electron transfer (6, 7).

The complexity of nitrate reduction to ammonia results in multiple competing reactions and undesirable by-products, leading to the central challenge of improving selectivity to increase the yield rate (YR) of ammonia. As nitrite is a well-known intermediate in the catalytic conversion of nitrate to either nitrogen or ammonia, the successful key in identifying an ideal electrocatalyst lies in finding a material with active sites capable of facilitating both nitrate reduction reaction (NO₃RR) to nitrite and nitrite reduction reaction (NO₂RR) to ammonia. Thanks to the efforts of many groups, a range of transition metals, including Fe, Co, Cu, Pd, and Ag, have been identified with diverse NO₃RR activity and selectivity (8–15). To date, electrocatalysts with two or more components based on judicious selection of metals for highly efficient NO₃RR have been well developed. Established strategies include alloy engineering to tune electronic structures for optimal intermediate adsorption energy (16–18), single atom isolation to maximize material utilization efficiency (19, 20), and facet engineering to expose optimal active sites (21). Yet, catalyst designs with a focus on spatially separating reaction processes to maximize the selectivity of each catalytic step are still scarce.

Recently, bimetallic Janus heterostructure design for tandem catalysis has emerged as an effective strategy to selectively produce desirable products, particularly from multielectron transfer reactions such as carbon dioxide reduction reaction and NO₃RR (22–27). By positioning two metal components into proximity, this design enables a catalytic relay pathway of NO₃RR that spatially separates the rate-determining step of nitrate deoxygenation to nitrite and the subsequent reaction of hydrogenation to ammonia, thereby maximizing selectivity within one tandem structure. However, current reports on tandem catalyst designs typically either lack precisely tunable domains or a well-defined interfacial structure (28). In the case of NO₃RR on a bimetallic catalyst, some 3d transition metals like Cu facilitate the initial reduction of nitrate to nitrite, while the presence of noble metals like Pd plays a key role in the subsequent reduction of nitrite to *NO and/or *N₂O. Spectroscopic studies have shown that surfaces favoring *N₂O formation are crucial for nitrogen (N₂) production, whereas surfaces that favor *NO formation are essential for ammonia (NH₃) generation (21, 29–31). As such, specific combinations of metals and structural design of catalyst should favor the preferential accumulation of *NO or *N₂O on the surface, thereby facilitating the generation of target products.

The dimension of nanoscale metal catalysts intrinsically defines their catalytic nature via altering their electronic structures and mass transfer processes over the surface (32–35). As such, metal catalysts with different nanoscale dimensionalities including zero-dimensional (0D) nanoparticles (NPs) (36), 1D nanorods or nanowires (34), 2D nanosheets or nanoplates (NPLs) (35), and 3D nanoframe (37) have been intensively reported. The Janus structure of two components with different dimensionality and, thus, a dimensionality junctional interface (DJI) offers a surface structure of tandem catalyst that may realize unique mechanisms for NO₃RR/NO₂RR, providing a design strategy of NO₃RR catalysts for efficient ammonia production. Here, we synthesized dimensionality junctional Janus nanocrystals (DJJNs) of Ag-Cu with a DJI via a site-selective seed-mediated growth of 3D Cu polyhedron on one edge of 2D Ag triangular NPLs. By controlling the amount of Ag seeds and reduction kinetics of the Cu precursor, three kinds of Ag-Cu DJJNs with Cu domains of different sizes can be acquired. As a result of the synthetic tunability, we were able to investigate the role of the Ag and Cu nanodomains with different DJI lengths in promoting NH₃ generation from NO₃RR. Traditionally, a spillover-mediated catalytic pathway, in which the difference in electron affinities between two metal domains results in a favorable electron transfer that promotes catalytic selectivity, is widely adopted in tandem design. On the basis of electrochemical results and theoretical modeling, here, we introduce an interface-mediated tandem catalytic pathway induced by the presence of a DJI that can lower the dissociation energy of key intermediate *NO₂ to favor NH₃ formation, aiding in the spillover-mediated mechanism. The fundamental understanding of structure-property relationships will aid in establishing design principles for advanced NO₃RR tandem catalysts with multiple components and surface properties, providing a pathway to overcome existing scaling relationships and achieve superior catalytic performance.

RESULTS

Synthesis and characterization of Ag-Cu DJJNs

Figure 1A depicts a schematic of proposed electrocatalytic NO₃RR pathways over the surface of Janus nanocrystals with a dimensionality junctional surface and interface, in which NO₃⁻ can be adsorbed at different surface regions of the Janus nanocrystals in principle: (i) domain with 3D morphology, (ii) DJI between the two domains, and (iii) domain with 2D morphology. After the adsorption of NO₃⁻, it can be converted into different intermediates, which will further diffuse across the surface to the region that favors the elementary steps of NO₃RR toward NH₃ by tandem catalysis of the multistep reaction. To generate Janus nanocrystals with a DJI, site-selective growth of Cu on one edge of Ag NPLs was conducted. As a highly asymmetric object, the surface of an Ag NPL is covered by 2 equiv of facets as the basal planes with stacking faults across the NPL and exposed on edges (38, 39). Figure S1 shows a typical transmission electron microscopy (TEM) image of the as-prepared Ag NPLs, which had an average edge length of 120 ± 10 nm. As shown in Fig. 1B, with Ag NPLs serving as the seeds, we were able to obtain Ag-Cu DJJN Janus nanocrystals by controlling the surfactant and the adding speed of CuCl₂·2H₂O and ascorbic acid. The epitaxial growth of Cu on the stacking fault–lined edge of the Ag NPLs results in 3D Cu polyhedral nanocrystals with twin boundaries connected to the 2D Ag seeds with rounded corners. Because of the large lattice mismatch between Ag and Cu, a heterogeneous stacking fault–lined interface with different domains divided by twin boundaries can be generated.

Fig. 1. Synthesis and structural characterization of Ag-Cu DJJNs.

(A) Schematic representation of potential reaction sites over the surface of DJJN classified by adsorption sites of NO₃⁻. (B) Schematic illustration for the synthesis of Ag-Cu DJJNs via site-selective growth of Cu on one edge of the Ag NPLs with a heterogeneous stacking fault–lined interface. The Ag and Cu domains are shown in blue and green colors, respectively. The red lines indicate the stacking faults on the nanocrystal. (C) TEM images of Ag-Cu DJJNs. (D) BF-STEM of an individual Ag-Cu DJJN. The red dashed lines indicate the twin boundary on the Cu component. (E) HAADF-STEM image and energy-dispersive x-ray spectroscopy elemental mapping of Ag-Cu DJJNs. (F) High-resolution HAADF-STEM image of the Ag/Cu interface in Ag-Cu DJJNs. The blue- and green-colored atoms represent Ag and Cu, respectively. The red lines indicate the stacking faults on the nanocrystal. (G to I) Fast Fourier transform patterns of the selected Ag domain (G), Cu domain (H), and Ag/Cu interface (I) in (F). (J) Integrated pixel intensities of Cu and Ag taken from the yellow dashed rectangles denoted with j in (F). (K) Geometric phase analysis for the lattice strain distribution taken from the selected areas in (F).

TEM images of Ag-Cu DJJNs reveal that the Cu domains grow on one of the six equal edges of Ag NPLs (Fig. 1C and fig. S2). A lateral high-angle annular bright-field scanning TEM image of an individual Ag-Cu DJJN clearly indicates the thicknesses of Cu and Ag domains as well as the presence of a twin boundary in the Cu component (Fig. 1D). The high-angle annular dark-field scanning TEM (HAADF-STEM) image and corresponding energy-dispersive x-ray spectroscopy elemental mapping of Ag-Cu DJJN further prove that Cu selectively grows on one edge of Ag NPLs, leading to the formation of well-defined Ag-Cu Janus nanostructures (Fig. 1E). Figure 1F shows a representative atomic resolution HAADF-STEM image of the Ag/Cu interface in Ag-Cu DJJNs, indicating the epitaxial growth of the Cu domains on Ag NPLs. Selected-area fast Fourier transform patterns of the Ag domain, Cu domain, and Ag/Cu interface in Fig. 1F all match well with the typical electron diffraction patterns of the face-centered cubic (fcc) phase along the [110] of the zone axis (Fig. 1, G to I), revealing two sets of fcc lattices with different lattice parameters corresponding to Ag and Cu. The lattice mismatch can also be confirmed by the Moire fringe, which can be observed at the Ag/Cu interface (fig. S3). Figure 1J demonstrates the integrated pixel intensities of the marked Ag and Cu domains along the directions denoted in Fig. 1F. The average Ag-Ag and Cu-Cu distances were measured to be 2.55 and 2.27 Å, respectively, indicating slight lattice expansion of the Ag domain around the Ag/Cu interface. Moreover, the ε_xx and ε_xy components of the strain tensor executed by the geometric phase analysis further show the presence of tensile and compressive strains in the Ag domain and Cu domains, respectively (Fig. 1K).

Composition and DJI tunability of Ag-Cu DJJNs

Such a class of Janus nanocrystals with two domains and a DJI that may deliver exclusive tandem catalysis pathways for enhancement in both the selectivity and activity of multistep NO₃RR toward NH₃ motivates us to seek out the quantitative impact of the two factors on the performance of NO₃RR at the nanoscale. To do so, we synthesized Ag-Cu DJJNs with variable sizes of the Cu domain to control the interfacial length. By simply changing the amount of the Ag seeds involved in the synthesis, three kinds of Ag-Cu DJJNs with different Ag/Cu atomic ratios and lengths of DJI, including Ag₆₅-Cu₃₅ DJJNs, Ag₄₅-Cu₅₅ DJJNs, and Ag₁₅-Cu₈₅ DJJNs, were obtained (figs. S4 and S5). Specifically, at a fixed amount of 1.28 mg for CuCl₂·2H₂O, we obtained Ag-Cu DJJNs with DJI lengths of 32 ± 3, 53 ± 6, and 89 ± 9 nm when adding 0.2, 0.5, and 1 ml of Ag seeds, respectively (Fig. 2, A to C). As a control sample for the catalytic performance comparison, pure Cu polyhedral NPs with a size range from 50 to 200 nm were also synthesized without adding Ag seeds (fig. S6). X-ray diffraction (XRD) was also performed to characterize the structure of as-synthesized Ag-Cu DJJNs. As shown in fig. S7, the characteristic peaks of Ag and Cu in three kinds of Ag-Cu DJJNs are all well consistent with those of pure Ag NPLs and pure Cu NPs, respectively, which further confirms the successful formation of a segregated pure metallic Cu and Ag phase in the nanocrystals and excludes the possibility of a AgCu alloy phase. Both Ag domains and Cu domains in the obtained Ag-Cu DJJNs adopt the standard fcc phase with near zero valence, matching well with the atomic resolution HAADF-STEM result. Enlarged XRD patterns reveal the increasing shift of characteristic peaks of Ag-Cu DJJNs to low angles when increasing the size of the Cu domain, indicating a lattice expansion of both Ag and Cu domains with an enlarged interfacial region and contact area (Fig. 2D and table S1).

Fig. 2. Characterization and comparison of different Ag-Cu DJJNs.

(A to C) Schematic and TEM images of three kinds of Ag-Cu DJJNs: (A) Ag₆₅-Cu₃₅ DJJNs, (B) Ag₄₅-Cu₅₅ DJJNs, and (C) Ag₁₅-Cu₈₅ DJJNs. Scale bars, 100 nm. The yellow lines mark the DJI on the nanocrystals. (D and E) Enlarged XRD patterns (D) and UV-vis spectra (E) of Ag-Cu DJJNs, Ag NPLs, and Cu NPs. Blue dashed lines: JCPDS no. 04-0783 (Ag). Green dashed lines: JCPDS no. 04-0836 (Cu). (F and G) XPS spectra of Ag 3d (F) and Cu 2p (G) for Ag-Cu DJJNs, Ag NPLs, and Cu NPs.

The unique structural features of the DJJNs are reflected in their optical properties as well, which endow their colloidal dispersions with distinctive colors in water (fig. S8). As shown in the ultraviolet-visible (UV-vis) spectra (Fig. 2E), the bimetallic Ag-Cu DJJNs exhibit dampened peaks with respect to the characteristic surface plasmon resonance peak of monometallic Ag NPLs (~329 and ~794 nm) and Cu NPs (~368, ~464, and ~585 nm), respectively. This intrinsic change of optical properties might be attributed to the rounded corner and reduced edge length of Ag NPLs during the seed-mediated growth of Ag-Cu DJJNs as well as a charge transfer between Ag and Cu domains. To confirm the charge transfer process, we further analyze the surface states of as-synthesized Ag-Cu DJJNs, Ag NPLs, and Cu NPs via x-ray photoelectron spectroscopy (XPS). As the size of the Cu domain increases, the binding energies of the Ag 3d core levels in the Ag-Cu DJJNs shift to higher binding energy (Fig. 2F), which is consistent with the previous investigations by Buonsanti and colleagues (23) and unambiguously verifies the existence of electron transfer from the Cu domain to the Ag domain via the Ag/Cu interface. For the corresponding XPS spectra of Cu 2p (Fig. 2G), they demonstrate a consistent peak shift with the electron transfer between the Ag domains and the Cu domains. Besides, the appearance of weak satellite peaks in Ag 3d and Cu 2p XPS spectra indicates slight oxidation of Ag and Cu in some samples in air (23, 27). With the increase in contact interface from Ag₆₅-Cu₃₅ DJJNs to Ag₁₅-Cu₈₅ DJJNs, the valence of Ag decreases as more electrons can transfer from Cu to Ag via a larger interface. These observations suggest tunable electron transfer from Cu to Au domains via the DJI.

Electrocatalytic tandem NO₃RR via Ag-Cu DJJNs

The catalytic activity and selectivity of NPs are highly dependent on their physicochemical attributes, such as the size, shape, composition, and surface structure (33, 40, 41). To identify the effect arising solely from structural changes, it is important to keep other parameters consistent. The availability of Ag/Cu nanocatalysts in a similar reaction environment involving the same surface ligands enabled us to investigate how the unique dimensionality junctional Janus structure promotes NO₃RR. The NO₃RR performance of three as-prepared kinds of Ag-Cu DJJNs was investigated in a homemade H-type cell under ambient conditions. All the applied potentials in this work were recorded with solution resistance correction. For comparison, the NO₃RR performance of Ag NPLs, Cu NPs, and the physical mixture of Ag NPLs and Cu NPs (denoted as “Ag NPLs + Cu NPs”) was also tested under the same conditions. As NO₂⁻ is a well-known intermediate present in the NO₃RR to NH₃, a catalyst enabling tandem NO₃RR that contains active sites for both the rate-determining step of NO₃⁻ to NO₂⁻ and selectivity-determining step of NO₂⁻ to subsequent products is highly favorable. As such, the electrochemical activities for the two reactions were first measured through linear sweeping voltammetry (LSV) (fig. S9). The current density is normalized by the electrode area with the same mass loading of metals. Blank LSV in the absence of reactants (either NO₃⁻ or NO₂⁻) was measured in a 0.5 M Na₂SO₄ electrolyte. All the Ag/Cu catalysts showed almost identical blank LSV curves with low current densities, except Ag NPLs + Cu NPs, indicating well-suppressed hydrogen evolution reaction (HER) under the testing condition. The electrochemical NO₃RR activity of all Ag/Cu catalysts was evaluated in the presence of NO₃⁻ ions (100 mM) in a 0.5 M Na₂SO₄ electrolyte. Because of the low HER activity, the current arising from LSV is primarily attributed to NO₃RR. The activity between −0.8 V_RHE and 0 V_RHE demonstrates an increased current with respect to blank LSV. At −0.8 V_RHE, the current densities were 9.43 mA cm⁻² for Ag NPLs, 19.14 mA cm⁻² for Cu NPs, 13.23 mA cm⁻² for Ag NPLs + Cu NPs, 11.53 mA cm⁻² for Ag₆₅-Cu₃₅ DJJNs, 11.10 mA cm⁻² for Ag₄₅-Cu₅₅ DJJNs, and 20.25 mA cm⁻² for Ag₁₅-Cu₈₅ DJJNs (Fig. 3A). The electrochemical activity for NO₂RR was also measured through LSV with and without NO₂⁻ ions (fig. S9). The electrochemical NO₂RR activities of all shape-controlled Ag/Cu in the presence of NO₂⁻ ions (100 mM) were measured in a 0.5 M Na₂SO₄ electrolyte. NO₂RR occurred after the onset potential of NO₃RR on all catalysts, indicating lower kinetics of NO₂RR. At −0.8 V_RHE, the current densities were 0.9 mA cm⁻² for Ag NPLs, 7.25 mA cm⁻² for Cu NPs, 3.35 mA cm⁻² for Ag NPLs + Cu NPs, 2.65 mA cm⁻² for Ag₆₅-Cu₃₅ DJJNs, 5.52 mA cm⁻² for Ag₄₅-Cu₅₅ DJJNs, and 1.85 mA cm⁻² for Ag₁₅-Cu₈₅ DJJNs. As summarized in Fig. 3A, the pure Ag catalyst shows the highest selectivity of NO₃RR, whereas the addition of the Cu domain results in the different activity and selectivity of NO₂RR, indicating that different interfacial areas between Cu and Ag help regulate tandem catalysis of the NO₂⁻ intermediate. It is noteworthy that the measured LSV current of NO₃RR is larger than that of NO₂RR even though the E°(NO₃⁻/NO₂⁻) is positive. In practice, the LSV currents are governed by (i) interfacial adsorption/activation barriers, (ii) surface coverage/poisoning, and (iii) mass transport—not E° alone. As such, standard potentials simply indicate that the reaction is thermodynamically allowed, but they do not imply higher current on a given catalyst at a given potential.

Fig. 3. Electrocatalytic NO₃RR performance of Ag-Cu DJJNs.

(A) Current density for NO₃RR, NO₃RR, and HER collected at −0.8 V_RHE for all catalysts. (B and C) FE of NO₂⁻ (B) and NH₃ (C) obtained on Ag NPLs, Cu NPs, Ag NPLs + Cu NPs mixture, and Ag-Cu DJJNs at different applied potentials. (D and E) FE of NO₂⁻ (as pointed by the black arrow) and NH₃ (as pointed by the blue arrow) with corresponding YR (as pointed by the red arrow) collected at −0.8 V_RHE as the function of molar percentage of Cu (D) and length of DJI (E). h, hour. (F) FE of NO₂⁻ and NH₃ with corresponding YR obtained on Ag-Cu DJJNs and Ag-Cu nanodimers at −0.8 V_RHE. (G) Schematic illustration of proposed NO₃RR mechanisms on Ag-Cu DJJNs, including spillover-mediated tandem catalysis and interface-mediated tandem catalysis induced by the presence of a DJI.

To elucidate the factors that determine the selectivity of Ag/Cu catalysts, we measured the faradaic efficiency (FE) of NO₂⁻ and NH₃ in NO₃RR at different potentials. Chronoamperometry measurements were performed at various potentials for subsequential qualification of products (fig. S10). Colorimetric methods based on the indophenol blue spectrophotometric method and Griess test were used to quantify the yield of NH₃ and NO₂⁻, respectively (figs. S11 to S13). The FEs of NO₂⁻ and NH₃ are shown in Fig. 3 (B and C, respectively). The lowest FEs of NO₂⁻ and the highest FEs of NH₃ can be delivered by Ag NPLs + Cu NPs (2.59%) and Ag₁₅-Cu₈₅ DJJNs (74.77%), respectively, indicating the necessity of Ag and Cu domains for tandem catalysis of NO₂⁻and the interfacial region that can mediate electron transfer for NH₃ formation. The FE of NO₂⁻ is determined by the difference between the rate of its formation and the rate of its consumption. Therefore, reduced FE(NO₂⁻) on DJJNs should not be interpreted as suppressed NO₃RR; rather, combined with increased FE(NH₃), it indicates accelerated interfacial NO₂⁻-derived intermediate conversion enabled by the Ag/Cu DJI. Combining with the XPS observation that the electron transfers from Cu to Ag, the spillover-mediated mechanism can be suggested to contribute to NO₃RR. Meanwhile, H₂ can be detected at all potentials as the major by-product on Ag₁₅-Cu₈₅ DJJNs (fig. S14). We also carried out control experiments at −0.6 V versus reversible hydrogen electrode (RHE) in a 0.5 M Na₂SO₄ solution without NaNO₃. As shown in fig. S15 (A and B), NH₃ was barely detected in the electrolyte, excluding the N contaminants from other sources including the catalyst, electrolyte, and the environment. ¹⁵N isotope labeling experiments were conducted to further verify that the produced NH₃ was resulting from the feeding NO₃⁻ electrolyte (figs. S15 and S16). After electrolysis at −0.8 V versus RHE, triple coupling and doublet peaks corresponding to ¹⁴NH₄⁺ and ¹⁵NH₄⁺ were detected in the ¹H nuclear magnetic resonance (NMR) spectra of the electrolytes containing ¹⁴NO₃⁻ and ¹⁵NO₃⁻, respectively (fig. S16A), proving that the produced NH₃ was derived from NO₃RR. The FE and YR were very close to those determined by colorimetric methods (fig. S16B), confirming that the produced NH₃ is mainly from NO₃RR.

To quantify the correlation between the composition and interface length of Ag/Cu catalysts and their NO₃RR selectivity toward NO₂⁻ and NH₃, the change of FEs and YRs of NH₃ as the function of molar percentage of Cu and length of DJI is demonstrated in fig. S17. To exclude size effects, we measured the double-layer capacitances of all samples to normalize their FEs and YRs (fig. S18). At −0.8 V_RHE, the FE and YR of NH₃ deliver a volcano-type trend as the molar percentage of Cu increases, indicating that an optimal composition of Ag/Cu catalysts is needed to deliver the highest selectivity (Fig. 3D). In terms of the length of DJI, both the YR and FE of NH₃ increase linearly with the highest values of 60.93 mg hour⁻¹ mg⁻¹ and 74.77% achieved by Ag₁₅-Cu₈₅ DJJNs, suggesting that the extension of DJI promotes the selectivity of NO₃RR toward NH₃ (Fig. 3E). As a control experiment, we also synthesized conventional Ag-Cu nanodimers without a DJI and compared NO₃RR selectivity between Ag₄₅-Cu₅₅ DJJNs and Ag-Cu nanodimers with a similar composition but a different interface structure (figs. S19 to S21). As shown in Fig. 3F, the Ag-Cu nanodimers delivered a lower NO₃RR selectivity relative to that of Ag₄₅-Cu₅₅ DJJNs, further revealing the importance of the DJI in NO₃RR toward NH₃. On the basis of the above observations, two reaction pathways including conventional spillover-mediated tandem catalysis and interface-mediated tandem catalysis induced by the presence of a DJI are proposed, as schematically illustrated in Fig. 3G. In addition, the selectivity and YR of the tandem NO₃RR performance of Ag-Cu DJJNs are also comparable to many reported Ag- and Cu-based NO₃RR catalysts (tables S2 and S3), highlighting the effect of a Janus catalyst with a dimension junctional interface that can promote tandem catalysis pathways of NO₃RR over other catalysts with similar compositions but different Janus or other structures.

The stability of NO₃RR performance was also evaluated on Ag-Cu DJJNs by conducting successive 20 electrolysis cycles at a fixed potential. As shown in fig. S22, the FE of NH₃ exhibited slight fluctuations around 84.8%, over the consecutive electrolysis cycles. After the stability test, the catalysts were detached from the glassy carbon electrode by sonication and characterized by bright-field TEM and XPS. As shown in fig. S23, both the Ag and Cu morphologies were preserved after the long-term stability test. Moreover, XPS analysis demonstrates increased oxidation of both Ag and Cu (fig. S24). The long-term stability of the catalyst was investigated by a chronoamperometry test for 100 hours at −0.8 V versus RHE. As demonstrated in fig. S22, the current density went through a slight decrease in the first 24 hours, after which it became quite stable over days, demonstrating the reasonable current stability realized by the catalyst. The reaction process and mechanism of NO₃RR on the Ag-Cu DJJNs were also investigated by operando synchrotron-radiation Fourier transform infrared spectroscopy. As shown in fig. S25, infrared signals in the range of 2200 to 1000 cm⁻¹ under the applied potential between −0.6 and −0.8 V versus RHE over Ag-Cu DJJNs were collected and analyzed. Specifically, the results verify decreased NO₂⁻ and increased *NH₂OH with the length of the DJI, indicating the positive effect of the DJI for the reduction and hydrogenation of NO₂⁻ in the process of NO₃RR.

According to previous studies of NO₃RR, the catalytic activity is determined by the first step of NO₃⁻ adsorption and conversion to NO₂⁻, while the selectivity of NO₂⁻ toward NH₃ is controlled by the favorability of *NO₂ desorption versus dissociation to *NO (26). Thus, theoretical simulation based on density functional theory (DFT) calculations was further conducted to study the difference in the energetic favorability of these intermediate steps between Ag(111) and Cu(111). Accordingly, three surfaces, including Ag(111), Cu(111), and Ag/Cu(111), were built with models of five atomic layers (figs. S26 to S28) to compare the adsorption and dissociation energies of *NO₃ and *NO₂ (table S4 and Fig. 4). Figure 4D shows that NO₃⁻ adsorbs more favorably on Cu(111) relative to Ag(111) and Ag/Cu(111) by 0.82 and 0.19 eV, respectively, implying that NO₃⁻ will adsorb more readily on the Cu domain of Ag-Cu DJJNs. Besides, the barrier for *NO₃ dissociation on Cu(111) is 0.24 eV lower relative to Ag(111) and 0.04 eV higher relative to Ag/Cu(111). Thus, the apparent barrier for NO₃⁻ adsorption and dissociation is the lowest on Cu(111), in agreement with the experimental observation that Cu NPs have higher NO₃RR activity relative to Ag NPLs (Fig. 3A) and the proposed tandem catalysis mechanism that the *NO₃ → *NO₂ step occurs on the Cu domain of Ag-Cu DJJNs (Fig. 3F). Turning to the NO₂⁻ desorption step versus the dissociation step, we find that *NO₂ desorption is more favorable on Ag(111) relative to Cu(111) and Ag/Cu(111) by 0.26 and 0.19 eV, respectively, implying that NO₂⁻ will desorb more readily from the Ag(111) surface before it can dissociate and form ammonia (Fig. 4H). This result is also in agreement with the experimental observation that Ag NPLs have the lowest NO₂RR activity (Fig. 3A). The activation barrier for *NO₂ dissociation to *NO is lower on Ag/Cu(111) relative to Ag(111) and Cu(111) by 0.17 and 0.04 eV, respectively, providing a theoretical basis for interface-mediated tandem catalysis in which the *NO₂ → *NO step occurs on the DJI of Ag-Cu DJJNs (Fig. 3F). This explains the high activity and selectivity toward ammonia achieved by Ag-Cu DJJNs that expose Ag(111), Cu(111) facets, and the DJI that can work in tandem to catalyze the full reaction to form ammonia.

Fig. 4. Theoretical calculations of electrocatalytic NO₃RR on Ag-Cu DJJNs.

(A to C) Side and top views of NO₃⁻ adsorption on Ag(111) (A), Cu(111) (B), and Ag/Cu(111) (C). (D) Energy of NO₃⁻ adsorption at pH 0 and the activation barrier for *NO₃ → *NO₂ plotted against the applied potential. (E to G) Side and top views of the transition state of NO₂⁻ adsorption on Ag(111) (E), Cu(111) (F), and Ag/Cu(111) (G). (H) Energy of NO₂⁻ desorption at pH 0 and the activation barrier for *NO₂ → *NO plotted against the applied potential. The red, blue, dusty blue, and brown spheres represent O, N, Ag, and Cu, respectively. (I and J) Projected density of states (PDOS) comparison of Cu 3d (I) and Ag 4d (J) in Cu(111), Ag(111), Ag₆₅/Cu₃₅(111), Ag₄₅/Cu₅₅(111), and Ag₁₅/Cu₈₅(111). (K) Reaction trend of different NO₃RR pathways on Cu(111), Ag(111), Ag₄₅/Cu₅₅(111), and Ag₄₅/Cu₅₅(111) with a twin boundary.

To better understand the NO₃RR performance of Ag-Cu DJJNs, we have applied the DFT calculations to investigate the electronic structures and reaction trends. We compared the electronic structures in the five different surfaces including Cu(111), Ag(111), Ag₆₅/Cu₃₅(111), Ag₄₅/Cu₅₅(111), and Ag₁₅/Cu₈₅(111). Cu 3d orbitals display an upshifting trend with increasing content of Cu (Fig. 4I). Meanwhile, Ag exhibits an opposite trend (Fig. 4J). Compared to pristine Ag, all the Ag 4d orbitals in tandem catalysts demonstrate a much higher position, which indicates the higher electron valence in the structure, consistent with the XPS results (Fig. 2F). Ag 4d orbitals show an overall lower position than Cu 3d orbitals, serving as the electron reservoir during NO₃RR. The d-band center has been calculated for different surfaces as an electronic structure descriptor for rationalizing the selectivity of the catalysts. As summarized in fig. S29, to guarantee efficient NO₃RR to generate the desired NH₃, an appropriate d-band center is also important. Therefore, Ag₁₅-Cu₈₅ DJJNs with a median d-band center display the highest FE for NH₃. The reaction pathways of NO₃RR were studied with a deoxidation-to-hydrogenation process simulated by DFT. The adsorption patterns of each intermediate on three surfaces along the NO₃RR steps were optimized before thermodynamic calculations. Figure 4K and fig. S30 exhibit the reaction pathways and free energies of each reaction state calculated from the grand-canonical DFT approach at the potential of 0 V versus RHE (table S5 for free energy source data). Along the reaction pathways, the first protonation of *NO₂ to *NO is thermodynamically uphill, consistent with the consensus that *NO₂ → *NO is the potential-determining step for NO₃RR on Ag- and Cu-based catalysts. The free energy increase of this step is substantially reduced from Ag₄₅/Cu₅₅(111) with a twin boundary to Ag₄₅/Cu₅₅(111), Cu(111), and Ag(111), which clearly demonstrates the effect of the DJI and twin boundary on facilitating the formation of *NO for energetically driving NO₃RR.

DISCUSSION

We have successfully synthesized three kinds of Ag-Cu DJJNs via site-selective growth of Cu on one of three equal edges of Ag NPLs. The obtained Ag-Cu DJJNs demonstrate improved NO₃RR performance toward NH₃. Compared with pure Ag and Cu counterparts, Ag₁₅-Cu₈₅ DJJNs show an NH₃ FE of 83.37% and a YR of 60.93 mg hour⁻¹ mg⁻¹ toward NH₃ production, owing to an extended interfacial region between Ag and Cu with different dimensionalities that facilitate electron transfer. DFT simulations further demonstrate that Cu(111) facets exhibit faster NO₃RR kinetics because of lower *NO₃ adsorption and dissociation barriers, while Ag/Cu(111) facets exhibit higher selectivity to NH₃ from NO₂⁻ because of the lowest *NO₂ dissociation barrier on Ag/Cu(111) than on Ag(111) and Cu(111). On the basis of these electrochemical analysis and theoretical simulations, we propose that two mechanisms of tandem catalysis, including spillover-mediated and interface-mediated tandem catalysis, collectively contribute to NO₃RR over the surface of Ag-Cu DJJNs. This work not only provides an alternative avenue for the controlled synthesis of Janus nanostructures but also sheds light on the rational design of NO₃RR electrocatalysts for efficiently transforming nitrate into NH₃.

MATERIALS AND METHODS

Chemicals and materials

Silver nitrate (AgNO₃; 99.99%), sodium borohydride (NaBH₄; 99%), sodium citrate tribasic dihydrate (Na₃C₆H₅O₇·2H₂O; ≥99.0%), polyvinylpyrrolidone [PVP; M_w (weight-average molecular weight) ≈ 29,000], PVP (M_w ≈ 55,000), l-ascorbic acid (AA; 99%), citric acid (99%), cupric chloride dihydrate (CuCl₂·2H₂O, 99.99%), 1-hexadecylamine (HDA; 90%), and isopropyl alcohol (≥99.5%) were purchased from Sigma-Aldrich. Sodium sulfate (Na₂SO₄; ≥99.0%), sodium nitrate (NaNO₃; ≥99.0%), and ethylene glycol (EG; >99%) were purchased from Thermo Fisher Scientific. All chemical reagents and materials were used as received without further purification. All aqueous solutions were prepared using deionized (DI) water with a resistivity of 18.2 MΩ·cm at room temperature.

Synthesis of Ag NPLs

NPL seeds were first synthesized using a modified chemical reduction method on the basis of previous reports. Seed-mediated growth was then performed using a previously reported method developed by Yin and colleagues (39). In a typical experiment, the as-synthesized seeds were collected by centrifugation for 14 min at 15,000 rpm and then resuspended in a solution comprising DI water (20 ml), sodium citrate tribasic dihydrate (75 mM, 250 μl), and AA (0.1 M, 750 μl). The NPL solution was stirred rapidly, after which growth was initiated by dropwise adding a silver growth solution containing AgNO₃ (1 mM, 40 ml), citric acid (0.1 M, 250 μl), and sodium citrate tribasic dihydrate (1.5 mM, 200 μl) using a syringe pump with an injection rate of 0.2 ml min⁻¹. After completion of titration, the solid products were collected by centrifugation, washed once with water and twice with ethanol, then redispersed into 5 ml of DI water, and stored in a refrigerator for further use.

Synthesis of Ag NPs

In a standard synthetic approach, 24.75 ml of aqueous solution containing AgNO₃ (0.05 M, 50 μl) and PVP (M_w ≈ 29,000, 17.5 mM, 0.1 ml) was vigorously stirred at room temperature in air. NaBH₄ (100 mM, 250 μl) was rapidly injected into this mixture to initiate the reduction, immediately leading to a light-yellow solution. After ∼30 min, the colloidal solution turned to a deep yellow because of the formation of small silver NPs. The solid products were collected by centrifugation, washed once with water and twice with ethanol, then redispersed into 5 ml of DI water, and stored in a refrigerator for further use.

Synthesis of Ag-Cu DJJNs

To synthesize Ag₄₅-Cu₅₅ DJJNs, 0.5 ml of the Ag NPL stock solution (the solvent was changed from DI water to EG before use) was first added into 3 ml of EG solution containing 90 mg of PVP, followed by intensive stirring for 3 hours, denoted as solution A. During this period, 1.28 mg of CuCl₂·2H₂O, 20 mg of HDA, and 3 ml of EG were sequentially added into another vial and then were stirred at 45°C for 2 hours, denoted as solution B. After finishing the stirring, the vial containing solution A was transferred to an oil bath at 140°C. Then, 2 ml of solution B and 2 ml of AA solution (25 mM, in EG) were simultaneously added into solution A using a syringe pump with an injection rate of 0.2 ml min⁻¹. After the completion of titration, the mixture was left to stir for an additional 20 min. Then, the resulting solid products were collected by centrifugation, washed once with hexane and twice with ethanol, then dispersed in hexane, and stored in a refrigerator for further use. For the synthesis of Ag₆₅-Cu₃₅ and Ag₁₅-Cu₈₅ DJJNs, the amount of Ag NPL stock solution was changed to 1 and 0.2 ml, respectively, while the amount of EG was adjusted accordingly to keep the total volume of solution A at 3 ml. All other parameters remained the same.

Synthesis of Cu NPs and Ag-Cu nanodimers

For the synthesis of Cu NPs, the total volume of solution A and other parameters were kept the same without adding Ag NPL stock solution into solution A. For the synthesis of Ag-Cu nanodimers, 0.5 ml of the Ag NP stock solution (the solvent was changed from DI water to EG before use) was added into solution A, while the total volume of solution A and other parameters were kept the same.

Characterization

TEM images were taken using a Hitachi S5500 microscope operated at 30 kV by drop-casting and drying the dispersions on carbon-coated copper grids under ambient conditions. High-resolution TEM, HAADF-STEM, and energy-dispersive x-ray mapping analyses were performed using a JEOL NEOARM probe-corrected transmission electron microscope with aberration correction. The metal contents in the as-obtained catalysts were determined using a THERMO VG PQ ExCell quadrupole-based inductively coupled plasma mass spectrometer. XRD patterns were obtained using a Rigaku MiniFlex 600 Diffractometer operating at 40 mA and 45 kV using Cu Kα1 radiation (λ = 1.541 Å) with 10° min⁻¹ from 20° to 90°. XPS measurements were carried out on a VersaProbe4 X-ray Photoelectron Spectrometer.

Electrochemical measurements

Electrochemical measurements were carried out using a BioLogic potentiostat with a VMP-3 model. All measurements were performed in a gas-tight H-cell using a three-electrode system and an ion-exchange membrane (Nafion 212) at room temperature. A mini HydroFlex reference electrode (ET070, Edaq Inc.) and a platinum wire were used as reference and counter electrodes, respectively. The potentials of the reference electrode were calibrated using a home-built RHE. Before testing, the Nafion 212 membrane was pretreated by sequentially heating it in a 5% H₂O₂ aqueous solution at 80°C (1 hour) and DI water at 80°C (1 hour), followed by treatment in 0.5 M H₂SO₄ (1 hour) and DI water (1 hour). To prepare the working electrode, the suspension of catalysts [Ag/Cu (1 mg ml⁻¹)] was obtained by dispersing the catalyst in a mixture of 490 μl of DI water, 490 μl of isopropanol, and 20 μl of Nafion solution (5 wt %) under ultrasonication for 30 min. Twenty microliters of this suspension was deposited on a precleaned glassy carbon electrode (ALS Co., 1.5 cm by 1 cm) and allowed to dry under ambient conditions to achieve a Ag + Cu mass loading of ~20 μg cm⁻².

Before each recording, the working electrode was cycled in a 0.5 M Na₂SO₄ solution within the intended LSV range between 0.4 and −1.0 V_RHE at a scanning rate of 50 mV s⁻¹ to clean and stabilize the catalyst surface. Current densities were normalized to the geometric area of the working electrode (~1 cm²). For electrocatalytic NO₃RR, the cathode compartment was filled with a 40-ml solution of 100 mM NaNO₃ and 0.5 M Na₂SO₄, while the anode compartment was filled with 40 ml of 0.5 M Na₂SO₄ solution. The LSV curves were recorded by sweeping the potential from 0.4 to −1.0 V_RHE at a rate of 10 mV s⁻¹. The current density was normalized to the mass of Ag + Cu.

Determination of NO₂⁻ concentration using the Griess test

After chronoamperometry tests, NO₂⁻ in the electrolyte was determined by the Griess test. The Griess agent was prepared by dissolving N-(1-naphthyl)ethyldiamine dihydrochloride (800 mg), sulfonamide (40 mg), and H₃PO₄ (2 ml, 85%) into 10 ml of DI water. Before the test, the NO₂⁻-containing electrolyte was diluted so that the UV absorbance value does not exceed 3 [arbitrary units (a.u.)]. Then, diluted solutions (2 ml) were mixed with the Griess agent (40 μl) and rested for 10 min at room temperature. A UV-vis spectrophotometer was used to record the absorption spectrum, and the absorbance at ~540 nm was used to determine the concentration of generated NO₂⁻. The calibration curve was obtained by using a set of NaNO₂ aqueous solutions with increasing concentrations (0.2, 0.4, 1, 2, 4, and 6 μg/ml) on a logarithmic scale as standard samples (fig. S8B) with the same testing operation.

Determination of NH₃ concentration using the indophenol blue method

The concentration of produced ammonia was spectrophotometrically determined by the indophenol blue method with modification. One milliliter of post-NO₃RR electrolyte was first extracted from the electrochemical reaction vessel. Then, 1 ml of 1.0 M NaOH solution containing 5 wt % salicylic acid and 5 wt % sodium citrate was added, followed by the addition of 500 μl of 0.05 M NaClO and 100 μl of 1 wt % C₅FeN₆Na₂O (sodium nitroferricyanide) aqueous solution. After 2 hours of chromogenic reaction at room temperature, the absorption spectrum was measured using a UV-vis spectrophotometer. The formation of indophenol blue was determined at a wavelength of 655 nm. The calibration curve was obtained by using various NH₄NO₃ aqueous solutions as the standard samples (fig. S11) in the same operation.

Calculation of the FE and NH₃ YR

FEs and mass/area–normalized YRs of NH₃ were calculated as follows

$$\mathit{\text{FE}}\left({\text{NH}}_3\right)=(8F\times C\times V\times n)/Q$$ (1)

$$\mathit{\text{FE}}\left({{\text{NO}}_2}^{-}\right)=(2F\times C\times V\times n)/Q$$ (2)

$$\mathit{\text{FE}}\left(\text{others}\right)=1-\mathit{\text{FE}}\left({\text{NH}}_3\right)-\mathit{\text{FE}}\left({{\text{NO}}_2}^{-}\right)$$ (3)

$$\mathit{\text{YR}}\left({\text{NH}}_3\right)=(C\times V\times n)/(t\times m)$$ (4)

where F is the Faraday constant (96,485 C/mol), C is the measured NH₃ concentration, V is the volume of the electrolyte, Q is the total charge passed through the electrode, n is the dilution factor, and m is the mass of metals on the working electrode.

Quantification of NH₃ concentration using ¹H NMR spectroscopy

An isotope-labeled tracer experiment was conducted to further quantify the produced NH₃ using ¹H NMR spectroscopy with a 100 mM ¹⁵NO₃⁻ solution as the nitrogen source (15, 31). After the chronoamperometry test, 200 μl of electrolyte was extracted from the electrochemical reaction vessel and then mixed with 10 μl of maleic acid (3.6 mM, as the internal standard) aqueous solution, 10 μl of H₂SO₄ aqueous solution (4 M), and 380 μl of dimethyl sulfoxide-d₆. After adding tetramethylsilane as the reference, the mixture was sealed into an NMR tube (5 mm in diameter, 400 MHz). The water peak was suppressed using a presaturation method. The concentration of produced NH₃ can be quantified by calculating the integral area (I) of the vinylic singlets for maleic acid and the typical triplet for ammonium on the basis of the following equation

$$C_{{\text{NH}}_4^{+}}=\frac{I_{{\text{NH}}_4^{+}}/H_{{\text{NH}}_4^{+}}}{I_{\mathrm{s}}/H_{\mathrm{s}}}\times C_{\mathrm{s}}$$ (5)

where $C_{{\text{NH}}_4^{+}}$ and $C_{\mathrm{s}}$ are the concentrations of NH₃ and maleic acid in NMR tubes, respectively; $H_{{\text{NH}}_4^{+}}$ and $H_{\mathrm{s}}$ are the numbers of protons for ammonium and maleic acid, respectively; and $I_{{\text{NH}}_4^{+}}$ and $I_{\mathrm{s}}$ are the integrals of ¹H NMR peaks for NH₃ and maleic acid, respectively. The isotopic labeling experiment was performed to verify the origin of NH₃ using 0.5 M Na₂SO₄ containing Na¹⁵NO₃ as the electrolyte in the same chronoamperometry test, as described above.

Determination of H₂ yield using gas chromatography

The H₂ yield was analyzed by an online gas chromatograph equipped with a thermal conductivity detector and a flame ionization detector (Shanghai Ramiin GC 2060).

DFT calculation

DFT calculations were conducted with the Perdew-Burke-Ernzerhof functional using the DMOL code, and the plane-wave pseudopotential for the spin-polarized generalized gradient approximation was used in this study. The empirical DFT-D method was applied using the Grimme scheme for dispersion corrections. The structural models of 3 × 3 × 2 supercells for Ag(111), Cu(111), and Ag/Cu(111) were built with the lattice constants of 4.09, 3.61, and 3.94 Å, respectively. We also introduced a vacuum layer of 20 Å between two neighboring slab models. The process of geometry optimization approached the convergence criteria for the energy and force on each atom reducing below 1 × 10⁻⁶ eV and 0.01 eV Å⁻¹, respectively. The real-space global orbital cutoff radius was set to 4.7 Å with a 3 × 3 × 1 k-point grid for all calculations.

The conductor-like screening model (COSMO) was used to introduce the treatment of solvation effects. The free energy (ΔG) calculations of each elementary step were based on the standard hydrogen electrode model, which can be determined as

$$\mathrm{\Delta }G=\mathrm{\Delta }E+\mathrm{\Delta }{E}_{\text{ZPE}}-T\mathrm{\Delta }S–\mathit{neU}$$ (6)

where ΔE is reaction energy differences obtained directly from DFT calculations, ΔE_ZPE is the zero-point energy (ZPE) difference between the adsorbed and gas phase molecules, T is the temperature of 298.15 K, ΔS is the change in entropy of products and reactants, e is the elementary charge of an electron, n is the number of proton-electron pairs transferred, and U is the applied potential. Here, U = 0 V for free energy diagrams demonstrated in the paper. The free energy changes (∆G) of each elementary reaction step of nitrate reduction on catalysts were simulated according to the following reactions (* represents the adsorption site)

$${{\text{NO}}_3}^{-}+{}^{\ast }\to {}^{\ast }{\text{NO}}_3+{\mathrm{e}}^{-}$$ (7)

$${}^{\ast }{\text{NO}}_3+{\mathrm{H}}_2\mathrm{O}+{\text{2e}}^{-}\to {}^{\ast }{\text{NO}}_2+{\text{2OH}}^{-}$$ (8)

$${}^{\ast }{\text{NO}}_2+{\mathrm{H}}_2\mathrm{O}+{\text{2e}}^{-}\to {}^{\ast }\text{NO}+{\text{2OH}}^{-}$$ (9)

$${}^{\ast }\text{NO}+{\mathrm{H}}_2\mathrm{O}+{\mathrm{e}}^{-}\to {}^{\ast }\text{NHO}+{\text{OH}}^{-}$$ (10)

$${}^{\ast }\text{NHO}+{\mathrm{H}}_2\mathrm{O}+{\mathrm{e}}^{-}\to {}^{\ast }{\text{NH}}_2\mathrm{O}+{\text{OH}}^{-}$$ (11)

$${}^{\ast }{\text{NH}}_2\mathrm{O}+{\mathrm{H}}_2\mathrm{O}+{\mathrm{e}}^{-}\to {}^{\ast }{\text{NH}}_2\text{OH}+{\text{OH}}^{-}$$ (12)

$${}^{\ast }{\text{NH}}_2\text{OH}+{\mathrm{e}}^{-}\to {}^{\ast }{\text{NH}}_2+{\text{OH}}^{-}$$ (13)

$${}^{\ast }{\text{NH}}_2+{\mathrm{H}}_2\mathrm{O}+{\mathrm{e}}^{-}\to {}^{\ast }{\text{NH}}_3+{\text{OH}}^{-}$$ (14)

$${}^{\ast }{\text{NH}}_3\to {\text{NH}}_3 {+}^{\ast }$$ (15)

Supplementary Materials

This PDF file includes:

Figs. S1 to S30

Tables S1 to S5

References