Amide-engineered copper nitride for shortcut-pathway-locked electroreduction of concentrated hydroxymethylfurfural Authors: Ju Huang, Jianqiu Zhu, Bowen Liu, Chencheng Dai, Zhimin Chen, Hao Wu, Gang Li, Shichao Du, Jian-Qiang Wang, Zhiyu Ren, Zhichuan J Xu, Honggang Fu Published: 2026 ABSTRACT AbstractAchieving high selectivity in the electrocatalytic hydrogenation (ECH) of concentrated 5-hydroxymethylfurfural (HMF) to 2,5-dihydroxymethylfuran (DHMF) remains a challenge due to competitive adsorption at high HMF concentrations, which limits active-site availability and suppresses *H generation. Regulating the balance between *H generation and HMF adsorption is therefore essential for maintaining selectivity under practical conditions. Here we show a surface amide-functionalized copper nitride catalyst (Ami-Cu3N) featuring dual-nitrogen active sites. The catalyst achieves >99% DHMF selectivity and >98% Faradaic efficiency, with a production rate of 63.4 mmol cm-2 h-1 at 500 mA cm-2 under concentrated HMF conditions. Mechanistic studies indicate that the combined roles of lattice nitrogen (promoting water dissociation) and amide nitrogen (modulating HMF adsorption) lower the energy barriers for both *H formation and HMF hydrogenation, facilitating high activity and selectivity. In addition, a coupled electrolysis system enables simultaneous production of value-added products at both electrodes. These findings provide design principles for electrocatalysts that enable selective biomass conversion at high reaction rates.Subject terms: Electrocatalysis, Electrocatalysis, Electrocatalysis Introduction 2,5-Dihydroxymethylfuran (DHMF), a biodegradable, multifunctional diol precursor widely used in synthesizing polyethers, polyurethanes, and polyamides, commands a market price 3–5 times higher than that of 5-hydroxymethylfurfural (HMF) and is experiencing rapidly growing global demand1,2. Conventional DHMF production depends on energy-intensive hydrogenation under high temperature and pressure, which diminishes its economic and environmental appeal3,4. In contrast, electrocatalytic reduction of biomass-derived HMF using green hydrogen, generated from water electrolysis powered by renewable electricity, offers a sustainable route under mild conditions5,6. Nevertheless, at industrial-level concentrations (>100 mM), achieving high-selectivity electrosynthesis of DHMF remains challenging, primarily due to insufficient control over the electrocatalytic hydrogenation (ECH) pathway and the impacts of competing reactions. Direct electroreduction via concerted proton-electron transfer (CPET) suffers from inefficient proton transfer and aggregation of reactive intermediates (e.g., ∙C–OH radicals), which promotes dimerization and undermines selectivity7. In contrast, the Langmuir-Hinshelwood (L-H) mechanism employs surface-adsorbed hydrogen (*H) to hydrogenate adsorbed HMF directly, suppressing dimerization and offering enhanced pathway control. However, under high HMF concentrations relevant to practical operation, the L-H pathway still competes with the hydrogen evolution reaction (HER), as strong competitive adsorption between HMF and water limits H availability at the catalyst surface, thereby reducing hydrogen utilization efficiency8. More critically, at elevated HMF concentrations, maintaining sufficient surface HMF coverage necessitates strong adsorption, while the increased reactant flux intensifies competition for interfacial proton-electron pairs, exacerbating proton-transfer limitations and favoring the re-emergence of CPET-driven side reactions. Therefore, a central catalyst design challenge lies in simultaneously ensuring a sufficient surface H supply and selective HMF adsorption under high reactant concentrations, while suppressing CPET and HER, to enable efficient and selective DHMF production from concentrated HMF feedstocks. Recent efforts to improve ECH performance have focused on increasing the surface coverage of active *H species through strategies such as incorporating electron-rich metals, constructing heterogeneous interfaces, and designing bimetallic sites9,10. While these approaches exhibit promising ECH activity at low HMF concentrations, they often suffer from inadequate selectivity and Faradaic efficiency (FE) under high concentrations of HMF (>100 mM), limiting their practical utility11. An ideal ECH catalyst must simultaneously facilitate *H generation and improve HMF adsorption12. Copper-based catalysts are particularly promising13. They can effectively generate *H via water dissociation (H2O → *H + OH-), comparable to noble metals like Pt or Ru, while their relatively high hydrogen adsorption free energy (ΔG*H > 0) suppresses HER, thereby maintaining a high *H coverage14,15. Further incorporation of nitrogen into Cu to form copper nitrides enhances the metal’s d-electron affinity, which promotes water dissociation and sustains a high *H supply16. Equally critical is the enrichment of HMF at the electrode-electrolyte interface. Molecular modification, especially the introduction of nitrogen-containing functional groups, enables precisely tuning of the catalyst’s electronic structure and local microenvironment, markedly enhancing HMF adsorption17,18. Such modifications generally induce surface charge redistribution, creating localized nucleophilic regions (e.g., Mδ+–Nδ- motifs), which strengthen interaction with the electrophilic carbonyl group of HMF19. Therefore, by rationally designing the nitrogen-coordination environment around Cu centers, an adaptive synergy between *H generation and HMF adsorption can be achieved, leading to high activity and selectivity under industrially relevant high-concentration conditions. Building on these insights, we synthesized a surface amide-functionalized copper nitride catalyst (Ami-Cu3N/CF) via synchronous surficial amide functionalization and nitridation. The lattice N in Cu3N promotes *H generation and coverage, while the surface amide groups enhance HMF adsorption through microenvironment modulation. The adjacent dual-nitrogen sites facilitate rapid *H-HMF binding, achieving dynamic balance during electrocatalytic hydrogenation. Under high-concentration HMF conditions (100–1000 mM), Ami-Cu3N/CF demonstrates high catalytic stability and selectivity, with a FE for DHMF consistently maintained above 90%; the selectivity for DHMF exceeds 99%, thereby showing competitive performance compared to previously reported systems. Mechanistic studies combining experimental and theoretical approaches reveal a synergistic multi-nitrogen effect. Furthermore, a coupled electrolysis system for concurrent DHMF (cathode) and 2,5-furandicarboxylic acid (FDCA; anode) production reduces the cell voltage compared to conventional OER-coupled HMF reduction, demonstrating improved energy efficiency and practical potential. Results Synthesis and structural characterizations of Ami-Cu3N/CF The synthesis and phase evolution pathway of Ami-Cu3N/CF is schematically depicted in Fig. 1a, b. A pretreated Cu foam (CF) substrate was oxidized to form Cu(OH)2 nanowires (Cu(OH)2/CF), which underwent coordination corrosion with BTC to yield a BTC-Cu metal-organic framework (BTC-Cu/CF). Subsequent thermal annealing under NH3 induced simultaneous nitridation and in-situ amide functionalization, producing Ami-Cu3N/CF. The X-ray diffraction (XRD) pattern of Ami-Cu3N/CF (Fig. 1b) shows distinct Cu3N peaks (PDF# 86-2283) alongside those of the metallic Cu substrate, with positions matching those in Cu3N/CF prepared without coordination corrosion. The scanning electron microscope (SEM) image reveals that Ami-Cu3N/CF exhibits an octahedral-derived framework structure derived from BTC-Cu/CF, contrasting the nanowire morphology of its Cu(OH)2/CF precursor (Figs. 1c, d and S1). This octahedral-derived framework features a rugged surface adorned with numerous nanocubes. Conversely, control samples Cu3N/CF (NH3 nitridation) and Cu/CF (H2 reduction), synthesized without BTC’s coordinating-corrosive effect, largely retain a nanowire-like morphology, albeit with rougher or more curved surfaces (Supplementary Fig. 2). The transmission electron microscopy (TEM) image of exfoliated Ami-Cu3N (Fig. 1e) confirms nanocubes (10–50 nm) embedded on the octahedral-derived frameworks. Higher-magnification TEM images show that these nanocubes consist primarily of porous carbon that embeds numerous small nanoparticles, attributed to incomplete decomposition of BTC-Cu organic ligands, as supported by thermogravimetric analysis (Supplementary Fig. 3). The high-resolution TEM image (Fig. 1f, g) shows a spacing of 0.38 nm, corresponding to the (100) interplanar crystal spacing of Cu3N. Furthermore, the energy dispersive spectrometer elemental mapping reveals that C, N, and Cu elements are distributed in Ami-Cu3N nanocubes (Fig. 1h). Ami-Cu3N/CF exhibits the largest Brunauer–Emmett–Teller surface area (2.62 m2 g−1) compared to Cu3N/CF (1.45 m2 g−1) and Cu/CF (0.89 m2 g−1), attributed to BTC-derived porosity, which facilitates reactant diffusion and exposure of active sites (Supplementary Fig. 4)20. Structural construction and characterization. a Schematic illustration of synthetic route for Ami-Cu3N/CF via nanowire-templated growth, ligand-driven MOF conversion, and concurrent nitridation and amidation. b XRD patterns of Ami-Cu3N/CF, BTC-Cu/CF, Cu3N/CF, Cu(OH)2/CF and Cu/CF. Red triangles, black symbols and yellow stars denote Cu3N, Cu(OH)2 and Cu phases, respectively. c, d SEM image of Ami-Cu3N/CF. e TEM image of Ami-Cu3N/CF. f, g high-resolution TEM images of Ami-Cu3N/CF exfoliated from CF. h HAADF-STEM image and corresponding elemental mapping images of Cu (yellow), N (green) and C (red) for Ami-Cu3N/CF exfoliated from CF. Source data for b are provided as a Source Data file. The surface composition and valence states of Ami-Cu3N/CF, Cu3N/CF, and Cu/CF were analyzed by X-ray photoelectron spectroscopy (XPS). High-resolution Cu 2p XPS spectra and corresponding Cu LMM Auger spectra (Fig. 2a, b) collectively indicate that Ami-Cu3N/CF consists predominantly of Cu+ species (Cu 2p3/2: 930–938 eV; Cu 2p1/2: 950–958 eV), with minor Cu2+ contributions21. The Cu 2p peaks of Ami-Cu3N/CF exhibit a slight negative binding energy shift (0.05 eV) compared with those of Cu3N/CF. Cu3N/CF exhibits higher Cu2+ content, whereas Cu/CF is dominated by metallic Cu0. The ubiquitous presence of Cu2+ across all catalysts is attributed to surface passivation from ambient air exposure22. Crucially, the carbonaceous layer derived from amide functionalization mitigates oxidation of Cu+, enabling partial retention of the Cu+ valence state in Ami-Cu3N/CF. The electronic and coordination structures of Ami-Cu3N/CF were characterized using X-ray absorption spectroscopy (XAS). Figure 2c displays the normalized Cu K-edge X-ray absorption near-edge structure (XANES) spectra for Ami-Cu3N, Cu3N, BTC-Cu, and reference samples. The absorption edge energies of Ami-Cu3N and Cu3N lies between those of Cu2O and Cu foil, indicating Cu valence states in them intermediate between +1 and 023. Moreover, Cu K-edge extended X-ray absorption fine structure (EXAFS) spectra (Fig. 2d) reveal similar primary peaks at ∼1.25 and ∼2.2 Å for Ami-Cu3N, Cu3N, and BTC-Cu, corresponding to Cu–N/O and Cu–Cu scattering paths24. Nitridation-driven conversion of BTC-Cu to Ami-Cu3N alters the Cu coordination environment (Supplementary Table 1), yielding a significant reduction in the first-shell coordination number (CN) for Cu–N/O (BTC-Cu: CN = 4.64; Ami-Cu3N: CN = 1.32). Moreover, Ami-Cu3N/CF exhibits further reduced CNs in both first and second coordination shells compared to Cu3N and Mix-Ami-Cu3N, with a long Cu–N–Cu bond length (Supplementary Figs. 5 and 6). This indicates that, in addition to the intrinsic Cu–N and Cu–Cu bonding in Cu3N, the N species from amide groups in the external carbon layer are additionally coordinated to Cu sites in Ami-Cu3N. Wavelet transform EXAFS analysis further confirms this; specifically, the Cu–N–Cu signal in Ami-Cu3N/CF is stronger than that in Cu3N/CF. It maybe benefits from the formation of the –CO–NH–Cu3N linkage, which proceeds via the reaction of the -COOH group in BTC-Cu with NH3 during nitridation, is energetically favorable, as evidenced by a reaction energy (ΔE) of −0.214 eV (Supplementary Fig. 7)25. The above findings indicate that the synthesized Ami-Cu3N exhibits a proximal multi-nitrogen coordination environment. Electronic structure and coordination environment. a Cu 2p XPS spectra and b Cu LMM auger spectra of Ami-Cu3N/CF, Cu3N/CF and Cu/CF. c Cu K-edge XANES spectra and d corresponding Fourier-transformed EXAFS spectra of Ami-Cu3N, Cu3N and BTC-Cu, with Cu foil, Cu2O and CuO as reference compounds. The dashed lines indicate the main scattering features corresponding to Cu–N and Cu–Cu coordination. e Wavelet transform EXAFS spectra of Ami-Cu3N, BTC-Cu, Cu3N and Cu. f N 1s XPS spectra of Ami-Cu3N/CF and Cu3N/CF. g N K-edge XANES spectra of Ami-Cu3N and Cu3N. h XPS valence band spectra of Ami-Cu3N/CF, Cu3N/CF and Cu/CF. Source data for a–h are provided as a Source Data file. The presence of –CO–NH– groups in Ami-Cu3N/CF is confirmed by comparing the Fourier-transform infrared (FTIR) and high-resolution N 1s XPS spectra between Ami-Cu3N/CF and Cu3N/CF. The FTIR spectrum (Supplementary Fig. 8) of Ami-Cu3N/CF highlights the amide II band (1541 cm−1, C–N stretching vibration and N–H in-plane bending of a –CO–NH– group), amide I band (1663 cm−1, C=O stretching vibration in a secondary amide group), and C–N bond (stretching vibration at 1400 cm−1 and flexural vibration at 703 cm−1)26,27. The N 1s XPS spectrum of Ami-Cu3N/CF (Fig. 2f) reveals a distinguished peak at 400 eV, assigned to the amide nitrogen (N–C=O), absent in Cu3N/CF. Both catalysts show peaks at 397.4 eV (Cu–N) and 399.0 eV (N–H). Deconvolution of the O 1s XPS spectrum (Supplementary Fig. 9) yields peaks at 531.2 eV (amide O, –NH–C=O) and 532.6 eV (carboxyl O, HO–C=O). The C 1s XPS spectrum deconvolutes into peaks at 289.0 eV (C=O), 286.7 eV(C–N), 285.5 eV (C–O), and 284.4 eV (C–C)28. The N K-edge XANES spectrum of Ami-Cu3N/CF (Fig. 2g) further corroborates the successful formation of amide bonds. The spectrum shows four distinct peaks: a strong pre-edge peak N1 (~398.6 eV) from M3d–N2p covalency, minor features N2/N3 (~399.4/400.6 eV) from amino groups and nitrogen dopants within carbon layer, and peak N4 (~405 eV) from the N 1s core level to C–N σ* states29–32. This assignment validates the chemical transformation from BTC to the amide-linked product. The XPS valence state spectrum (Fig. 2h) results from the cumulative contributions of the electronic structures of all atoms in the sample. The relative intensities follow the order Ami-Cu3N/CF 98.0% FEDHMF. Ami-Cu3N/CF maintains >90% FEDHMF and >90% HMF conversion across a broad current range (−100 to −600 mA cm−2), indicating improved performance relative to the control catalysts. Full-duration chronopotentiometry (Supplementary Fig. 17) validates that the charge-normalized comparison reflects robust steady-state catalysis rather than transient deactivation. Unlike the decaying control electrodes, Ami-Cu3N/CF sustains a stable potential across all current densities, indicating that its performance is associated with a robust, amide-stabilized active phase. At high cathodic current densities (−600 mA cm−2), a slight decline in FEDHMF (~90%) is observed. Additional concentration-dependent and hydrodynamic control experiments (Supplementary Fig. 18) indicate that interfacial mass transport becomes a limiting factor under these conditions. Enhancing reactant supply via improved convection or concentration effectively mitigates this limitation and restores DHMF selectivity. This combination of high activity at low potential and effective HER suppression represents an important step toward practical implementation. Ami-Cu3N/CF exhibits competitive mass-specific productivity compared to reported catalysts, while maintaining near-quantitative DHMF yield and FE (Supplementary Table 5). To verify that the enhanced performance stems from intrinsic catalytic properties rather than potential-induced bias under galvanostatic control, we conducted potentiostatic evaluations at fixed thermodynamic driving forces (Supplementary Fig. 19). Notably, the matched-potential potentiostatic comparison provides a critical control for excluding protocol-dependent selectivity bias. Even under identical thermodynamic driving forces (−10 to −50 mVRHE), Ami-Cu3N/CF maintains near-quantitative DHMF yield and FE, showing improved performance relative to Cu3N/CF and Cu/CF. This result demonstrates that the enhanced hydrogenation performance of Ami-Cu3N/CF is intrinsic to its amide-engineered interfacial structure rather than an artifact of the galvanostatic testing protocol. The much higher overpotentials required by Cu/CF further exacerbate selectivity loss at high current densities. Under such strongly cathodic conditions, accelerated water reduction leads to increased HER contribution, thereby diverting *H away from productive hydrogenation pathways. In contrast, Ami-Cu3N/CF operates at much milder potentials, where the interfacial structure effectively sustains selective HMF hydrogenation while suppressing parasitic HER. These matched-potential results therefore provide direct evidence that the catalyst-dependent selectivity originates from intrinsic interfacial chemistry rather than from differences in the operating potential imposed by galvanostatic testing. Control experiments using bare Cu foam show HER dominated behavior with minimal DHMF formation, indicating that the high selectivity of Ami-Cu3N/CF originates from the engineered active interface rather than the substrate (Supplementary Fig. 20). The apparent turnover frequency (TOF), normalized to electrochemically accessible Cu sites derived from surface oxidation voltammetry, was used to compare the intrinsic HMF ECH activity of the catalysts (Supplementary Fig. 21). Despite the conservative nature of this normalization (which counts the total accessible Cu reservoir), Ami-Cu3N/CF exhibits a markedly higher TOF (17.2 s−1, at −0.35 VRHE) than Cu3N/CF (10.1 s−1) and Cu/CF (1.5 s−1), confirming its enhanced intrinsic activity associated with nitrogen-modified Cu sites36. High-concentration HMF ECH is crucial for industrial applications. Evaluation across 20–1000 mM HMF demonstrates that, unlike Cu3N/CF and Cu/CF, which lose activity at higher concentrations, Ami-Cu3N/CF maintains nearly constant DHMF yield and FE (≈99%) with a productivity of 63.4 mmol cm−2 h−1 at 1000 mM and >98% HMF conversion (Figs. 3d and S22). At identical current densities, the cathode potential of Ami-Cu3N/CF remains remarkably stable against varying HMF concentrations, in sharp contrast to the significant potential drift observed for Cu3N/CF and Cu/CF (Supplementary Table 3). This invariance confirms that the FE evolution is not an artifact of potential fluctuation but is intrinsically grounded in concentration-dependent kinetics, where higher substrate coverage effectively suppresses the competing HER. Under 200 mA cm−2 constant current (Fig. 3e), HMF depletes quantitatively (>99% conversion) with concurrent DHMF accumulation to maximum yield. In 200 mM electrolyte, near-complete conversion occurs within 3 h (Figs. 3f and S23), and Ami-Cu3N/CF retains >98% DHMF yield and >99% selectivity over 90 cycles (Supplementary Fig. 24), confirming high stability. This performance is competitive with previously reported benchmarks (Supplementary Table 5)2,11,37–57. To further decouple intrinsic catalyst stability from potential-induced degradation, we additionally performed long-term stability tests under matched-potential conditions (Supplementary Fig. 25). Under the same applied half-cell potential, Ami-Cu3N/CF still exhibits more stable performance than Cu3N/CF and Cu/CF, indicating that its enhanced durability is not solely a consequence of operating at milder galvanostatic potentials, but also reflects the intrinsic robustness of Ami-Cu3N/CF. Furthermore, substrate scope studies (Supplementary Figs. 26–28, and Supplementary Table 6) confirm broad applicability for aldehyde electroreduction, achieving enhanced yields, selectivity, and FE toward furfuryl, benzyl, and 4-methoxybenzyl alcohols. The comprehensive performance results indicate that the multi-nitrogen coordination environment of Ami-Cu3N regulates *H and substrate adsorption dynamics, enabling efficient and selective aldehyde hydrogenation. Capitalizing on the bifunctional efficacy of Ami-Cu3N/CF, we further assembled a proof-of-concept paired electrolysis system. By coupling cathodic HMF hydrogenation with anodic HMF oxidation (Supplementary Figs. 29–31), thereby circumventing the energy-intensive oxygen evolution reaction, we illustrate a generalizable strategy for maximizing energy efficiency in biomass valorization. As this configuration serves primarily to demonstrate system-level feasibility rather than intrinsic catalytic metrics, detailed performance data and a preliminary techno-economic analysis are presented in Supplementary Figs. 32–38. Origin of Ami-Cu3N/CF’s high activity for HMF ECH By integrating quasi-in-situ XRD and XPS tests with post-reaction crystalline phase, microstructure, and atomic-level analysis, we established that the catalyst evolves via a partial, self-limited electrochemical reconstruction mechanism (Supplementary Figs. 39–48). Importantly, additional high-bias XPS comparisons further show that this self-limiting behavior is not merely a consequence of the mild operating window, but reflects an intrinsic resistance to deep metallization conferred by amide functionalization. Under reaction-relevant cathodic potentials, the partial reduction of Cu+ to metallic Cu0 is thermodynamically inevitable; thus, the emergence of Cu0 represents an intrinsic response to the reductive environment required for high-rate HMF hydrogenation. Crucially, however, unlike unfunctionalized Cu3N, which suffers from deep reduction, the amide-functionalized catalyst retains a dominant Cu+ character. Quantitative XPS analysis shows that the surface amide coverage (RAmi-N/Cu–N) rapidly reaches a stable plateau, confirming that these groups act as robust ligands to inhibit excessive metallization. XAS evidence further supports this interfacial model. The post-reaction EXAFS spectrum requires a dual-shell fit consisting of Cu–N and Cu–Cu contributions, indicating the coexistence of residual nitride coordination and newly formed metallic Cu species. The fitted Cu–Cu bond length (2.70 Å) is longer than that of bulk Cu metal (2.54 Å) and close to the Cu–Cu distance in the Cu3N lattice (2.73 Å), which is consistent with a closely coupled Cu/Cu3N interfacial structure formed after reconstruction. Complementary N K-edge XANES and Quasi-in-situ FTIR confirm the integrity of the amide linkages (–CO–NH–) during this process (Supplementary Figs. 47 and 48). Time-dependent inductively coupled plasma atomic emission spectrometry analysis also reveals a dynamic reconstruction process characterized by transient Cu dissolution followed by spontaneous re-deposition (Supplementary Fig. 49, and Supplementary Table 9). Ultimately, the bulk electrolyte remains virtually free of dissolved Cu species after stabilization. Taken together, the quasi-in-situ XPS, XAS, and XRD results show that amide functionalization suppresses deep reduction of Cu3N and promotes a self-limited reconstruction toward an Ami-Cu/Cu3N heterostructure. Importantly, the enhanced operational durability of Ami-Cu3N/CF should not be attributed to a single factor alone. Rather, amide functionalization confers both (i) structural stabilization that limits reconstruction depth and prevents full metallization, and (ii) enhanced intrinsic activity that allows the catalyst to operate at substantially milder cathodic potentials. These two effects synergistically contribute to the observed long-term stability. Thus, under operating conditions, the catalyst evolves into a robust amide-functionalized Cu/Cu3N heterostructure, which is inferred from the post-reaction characterizations to represent the steady-state active surface. This experimentally inferred end-state, rather than the dynamic reconstruction pathway itself, was subsequently used as the structural basis for the DFT analysis. To elucidate the DHMF formation pathway over Ami-Cu3N/CF, in-situ FTIR spectroscopy was employed to monitor HMF adsorption and conversion behavior. As shown in Figs. 4a and S50, three key intermediates are identified: *H (around 2053 cm−1), −C=O (near 1650 cm−1), and the furan ring (1014, 1072, 1153, 1453 and 1515 cm−1)58–61. No discernible peaks are observed under open-circuit potential (OCP) conditions, whereas characteristic bands appear upon applying more negative potentials. Hydrogenation of HMF commences at −0.1 VRHE on Ami-Cu3N/CF. As the potential became more negative, the *H band exhibits a blue shift, indicating weakened adsorption caused by competitive HER. The peaks associated with the furan ring remain unchanged across all potentials, indicating negligible adsorption of the furan ring on the catalyst. Cu3N/CF shows a similar trend but with significantly weaker *H and HMF adsorption (Supplementary Fig. 51). During electrolysis at −0.5 VRHE (Fig. 4b, c), *H and C=O intensities first increase then decrease after 250–300 min, corresponding to *H consumption and C=O desorption. These results indicate that active *H species are involved in the hydrogenation process, while HMF adsorbs primarily via the −C=O group, implying that hydrogenation occurs selectively at the carbonyl group. Additionally, in-situ FTIR confirms the enrichment of −COOH groups from HMF on the Ami-Cu3N/CF surface. Mechanistic insights into HMF hydrogenation. a Operando FTIR spectra of Ami-Cu3N/CF recorded at potentials from −0.1 VRHE (OCP) to −1 VRHE. b Time-resolved operando FTIR spectra at −0.5 VRHE. c Evolution of IR signal corresponding to *H, −C=O and furan ring over Ami-Cu3N/CF and Cu3N/CF during the ECH process. d–f 1H NMR spectra of HMF ECH products obtained in PBS/H2O, PBS/H2O with D2O addition, and PBS/D2O, respectively. All the samples were dissolved in DMSO-d6 for tests. g EPR trapping for hydrogen radicals over Ami-Cu3N/CF, Cu3N/CF and Cu/CF. h OCP changes of Ami-Cu3N/CF, Cu3N/CF, and Cu/CF in 0.5 M PBS after injecting 20 mM HMF. No iR correction is applied as no current flows under open-circuit conditions. i Schematic illustration of the proposed HMF hydrogenation mechanism on Ami-Cu3N/CF. Source data for a–h are provided as a Source Data file. To elucidate the hydrogen source in HMF ECH, isotope labeling experiments were conducted. Nuclear magnetic resonance (NMR) spectra (Fig. 4d–f and S52) of the products in PBS/H2O display DHMF signals, while rapid H−D exchange at the −OH group is confirmed by signal disappearance upon D2O addition. The spectrum obtained from PBS/D2O shows deuteration at three hydrogen positions, demonstrating that water serves as the hydrogen source. Quasi-in-situ EPR using 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) as a trap detects ∙H species on Ami-Cu3N/CF, Cu3N/CF, and Cu/CF in electrolyte under potential, confirming water-derived ∙H generation, while no signal is observed in the absence of voltage (Figs. 4g and S53)62,63. CV analysis reveals distinct *H desorption features at 0.56, 0.35, and 0.03 VRHE for Ami-Cu3N/CF, Cu3N/CF, and Cu/CF, respectively (Supplementary Fig. 54). Notably, Ami-Cu3N/CF exhibits the most pronounced anodic peak, indicative of a substantially enriched *H reservoir. To probe the mechanistic role of this reservoir, we introduced tert-butanol (TBA) as a site-blocking agent. The progressive attenuation of the *H-related peak upon TBA addition, coupled with the concomitant suppression of HMF reduction kinetics (Supplementary Fig. 55), provides compelling evidence that surface-adsorbed *H is the indispensable active species driving the hydrogenation, consistent with an L-H pathway (Supplementary Fig. 56). OCP measurements were used to monitor adsorption behavior changes within the Helmholtz layer after HMF introduction. As shown in Fig. 4h, upon injection of HMF, the OCP of Ami-Cu3N/CF drop markedly by 1.31 V, substantially larger than that of Cu3N/CF (1.01 V) or Cu/CF (0.18 V). This indicates stronger HMF adsorption on Ami-Cu3N/CF, suggesting that the amide-functionalized surface facilitates HMF adsorption, a crucial precondition for its efficient subsequent conversion. To further quantify HMF enrichment on the catalyst surface, quartz crystal microbalance measurements were conducted. Quartz crystal microbalance measurements further quantify this enrichment: upon exposure to 20 mM HMF, Ami-Cu3N/CF exhibits approximately two- and twelve-fold higher mass increases than Cu3N/CF and Cu/CF, respectively (Supplementary Fig. 57), confirming the elevated adsorption capacity arising from the amide-functionalized carbon layer. The contact angles for Ami-Cu3N/CF, Cu3N/CF, and Cu/CF are 118.9°, 138.8°, and 147.9°, respectively (Supplementary Fig. 58), further supporting the improved affinity of the amide-functionalized surface toward HMF in aqueous medium. Spectroscopic evidence from concentration-dependent FTIR reveals a systematic red-shift of the amide N–H stretching band upon HMF introduction (Supplementary Fig. 59). This vibrational perturbation confirms the formation of interfacial hydrogen bonds (−NH…O=CHMF), demonstrating that the surface amide groups function as chemical anchors to facilitate the preferential enrichment of HMF. The electrochemical reduction of HMF on Ami-Cu3N/CF is proposed to proceed via the mechanism illustrated in Fig. 4i. Water is activated and dissociated into *H at the Cu/Cu3N heterojunction interface, accompanied by electron transfer; subsequently, two *H species react with the adsorbed carbonyl group of HMF on the amide-functionalized carbon layer, producing DHMF via a L-H pathway. Experimental analyzes reveal that lattice nitrogen in Cu3N strongly promotes water dissociation, accelerating *H generation, while surface amide groups regulate the local interfacial microenvironment through hydrogen bonding and electrostatic interactions. This cooperative modulation enhances HMF adsorption and stabilized *H species, collectively demonstrating that the multi-nitrogen coordination environment plays a pivotal role in enabling high ECH activity. The kinetics of HER and HMF ECH were evaluated using Tafel analysis. As shown in Fig. 5a, the Tafel slopes for HER are 146, 170, and 151 mV dec−1 for Ami-Cu3N/CF, Cu3N/CF, and Cu/CF, respectively, indicating that the Volmer step is the rate-determining step (RDS) for all catalysts64. Upon addition of HMF, the Tafel behavior varies significantly among the samples. For Cu/CF, the slope increases from 151 to 192 mV dec−1, suggesting a less favorable HMF ECH pathway compared to HER. In contrast, Ami-Cu3N/CF and Cu3N/CF exhibit similar Tafel slopes under HMF ECH (118 and 154 mV dec−1), implying that the Volmer step remains the RDS. Isotopic experiments using H2O/D2O electrolytes (Figs. 5b and S60–S62) are conducted to further probe the RDS. For HER, Cu/CF displays a significant kinetic isotope effect (KIE) with a 127 mV cathodic shift at 20 mA cm−2, whereas Ami-Cu3N/CF and Cu3N/CF show smaller shifts (92 and 52 mV), confirming that Cu3N facilitates water dissociation and promotes the Volmer step. During HMF ECH, the KIE values for Ami-Cu3N/CF and Cu3N/CF (119 and 72 mV) closely match those in HER, supporting a L-H mechanism involving adsorbed *H species. In contrast, Cu/CF exhibits a smaller KIE decrease (114 mV) during HMF ECH, implying a shift of the RDS toward a CPET pathway. Reaction-order analysis with respect to HMF concentration further distinguished these mechanisms. For Ami-Cu3N/CF, a negative reaction order (Supplementary Fig. 60) indicates competitive adsorption between HMF and H2O, consistent with the L-H mechanism. Cu3N/CF maintains a negative reaction order up to 100 mM, but the slope approaches 0.56 beyond this concentration, suggesting limited *H availability and a transition toward Eley-Rideal (E-R) behavior. Cu/CF, however, shows a positive order of 0.95, consistent with a predominant E-R pathway. Notably, the concentration-dependent rate analysis in Supplementary Fig. 63 was performed under matched-potential potentiostatic conditions (−0.35 VRHE) for all three catalysts in an H-cell, so that the observed differences in apparent reaction order reflect catalyst-dependent concentration responses under the same electrochemical driving force. Product analysis (Supplementary Fig. 64) reveals that C−C coupling is promoted on Cu/CF and emerges on Cu3N/CF at high HMF concentrations, but is effectively suppressed on Ami-Cu3N/CF, indicating that amide functionalization inhibits radical-mediated coupling and stabilizes the selective L-H hydrogenation pathway65–67. Kinetic analysis of competing HER and HMF hydrogenation pathways. a Tafel plots of Ami-Cu3N/CF (red), Cu3N/CF (blue), and Cu/CF (yellow) in 0.5 M PBS (dashed lines) and 0.5 M PBS containing 20 mM HMF (solid lines). b Comparison of polarization curves measured on Ami-Cu3N/CF, Cu3N/CF, and Cu/CF in H2O (pH 7.0) and D2O (pD 7.0) electrolytes, with and without 20 mM HMF. c, d Nyquist plots and corresponding Bode phase plots of Ami-Cu3N/CF recorded in 0.5 M PBS at different applied potentials. e Equivalent circuit model used for EIS fitting and schematic illustration of the hydrogen evolution mechanism. R1 represents the electron-transfer resistance at the electrode surface, R2 corresponds to intermediate (*H) accumulation and interfacial charge storage, and R3 denotes the charge-transfer resistance associated with surface hydrogenation reactions. f, g Comparison of Nyquist and Bode plots for HER and HMF ECH processes over Ami-Cu3N/CF (red), Cu3N/CF (blue), and Cu/CF (yellow) at −0.3 VRHE in 0.5 M PBS with and without 20 mM HMF. h Potential-dependent evolution of equivalent resistances (R2 and R3) for Ami-Cu3N/CF in the potential range from −0.1 to −0.5 VRHE, with and without 20 mM HMF. All potentials are referenced to RHE. Tafel plots and polarization data were derived from LSV measurements with 95% iR compensation, whereas EIS measurements were conducted without iR correction unless otherwise specified. The uncompensated solution Rs was in the range of 2.5–4.8 Ω cm2. Source data for a–h are provided as a Source Data file. Operando electrochemical impedance spectroscopy (EIS) was employed to interrogate the hydrogen supply and consumption dynamics during HMF ECH (Fig. 5c–h, Supplementary Figs. 66–72, Supplementary Tables 10–12)68,69. Relevant reaction steps are summarized in Scheme S2. It should be noted that the assignment of equivalent circuit elements to specific elementary steps (e.g., Volmer or Heyrovsky) is subject to inherent model dependence. Therefore, the circuit model presented here is interpreted as a semi-quantitative framework to deconvolute the impedance features and track the kinetic trends across different potentials, rather than as an absolute mechanistic determination. Under HER conditions, the Bode plots exhibit two distinct relaxation features: a low-frequency response attributed to the relaxation of surface coverage (*H formation/Volmer step) and a mid-frequency feature associated with interfacial charge transfer (Heyrovsky step). To quantify the coupled dynamics, we extracted the characteristic relaxation time constant (τ) from the fitted CPE parameters, serving as a kinetic fingerprint for the coverage-dependent Volmer-to-L-H sequence (Supplementary Fig. 70). Notably, upon HMF introduction, τ is selectively and strongly modulated within the low-overpotential window (−0.10 to −0.30 VRHE) for Ami-Cu3N/CF and Cu3N/CF. This behavior signals a regime where *H generation and its consumption via HMF hydrogenation are tightly coupled, specifically, the *H-mediated L-H step becomes kinetically rate-determining. Consistent with this τ signature, HMF addition predominantly reshapes the medium-to-low frequency impedance response for the nitride-based catalysts (Fig. 5f, g). The fitted resistance parameters indicate that while the *H formation pathway remains robust, the interfacial reduction resistance (Rct) increases in the presence of HMF at low overpotentials (−0.10 to −0.45 VRHE). This effectively implies that the surface *H reservoir is diverted from the HER pathway toward HMF hydrogenation, leading to a suppression of parasitic HER, in line with the enhanced FEDHMF observed (Fig. 3d). In stark contrast, Cu/CF exhibits a negligible τ response and a decrease in resistance upon HMF addition, pointing to a distinct ECH trajectory less reliant on a stable *H reservoir, consistent with a concerted CPET-type mechanism. The significantly lower polarization resistance for Ami-Cu3N/CF relative to its counterparts further corroborates its enhanced capability for facilitating rapid interfacial electron delivery and efficient *H utilization. It follows that HMF reduction proceeds via divergent mechanisms: whereas Ami-Cu3N/CF and Cu3N/CF facilitate HMF hydrogenation through an L-H pathway, the reaction on Cu/CF follows a distinct direct electroreduction mechanism involving CPET, yielding a radical intermediate that may either be hydrogenated to DHMF or participate in C−C coupling (Supplementary Fig. 73). PBS was selected not merely for bulk pH control, but for its enhanced capability in modulating interfacial proton dynamics (Supplementary Figs. 74–78). While control experiments in borate buffer (BBS) resulted in a loss of HMF-response and sluggish kinetics, PBS enables rapid interfacial charge transfer. XPS analysis confirms that phosphate species adsorb at the interface without acting as sacrificial reactants, thereby functioning as a stable proton-management scaffold that sustains the *H coverage required for the coupled L-H mechanism. Mechanistic studies of HMF ECH to DHMF Guided by the post-reaction structural characterizations and the experimentally identified self-limited reconstruction behavior, we constructed a static Ami-Cu/Cu3N interfacial model to represent the inferred steady-state active motif, in which amide groups are selectively ligated at the Cu–Cu3N boundary. This model was used to analyze the electronic structure and reaction energetics of the reconstructed interface, rather than to simulate the dynamic reconstruction process itself. Using this representative motif (Supplementary Figs. 79–84 and Supplementary Data 1), we deciphered the cooperative mechanism by which the heterojunction architecture and ligand functionalization synergistically minimize the energy barriers for HMF ECH. The adsorption energies of HMF on Ami-Cu/Cu3N, Ami-Cu3N, Ami-Cu, Cu/Cu3N, Cu3N and Cu were compared. As shown in Fig. 6a, the introduction of lattice-N (Cu3N) alone improves adsorption relative to metallic Cu (−1.81 eV vs. −1.78 eV) by creating polar Cu–Nlatt motifs that enhance basal affinity for the polar HMF substrate. Building on this, amide-functionalized catalysts (Ami-Cu/Cu3N, Ami-Cu3N, Ami-Cu) exhibited significantly more negative adsorption energies (ΔEad = −2.35, −2.11, and −1.83 eV, respectively). This hierarchy confirms a dual-site synergy, that is, lattice-N constructs a receptive active platform, while amide groups function as geometric anchors to maximize substrate capture. Electron localization function analysis revealed localized electrons at unsaturated Cu sites, with Ami-Cu/Cu3N showing the strongest electronic localization (Figs. 6b and S85). This is complemented by significant electron redistribution across the amide-functionalized heterojunction (Fig. 6c), creating a highly active electronic landscape. Further electronic analysis (Supplementary Figs. 86 and 87) reveals that the amide functionalities induce a distinct polarization of the HMF carbonyl bond. Corroborated by Bader charge transfer data, this polarization arises from amide-mediated hydrogen bonding, which geometrically stabilizes the adsorption and electronically pre-activates the substrate for efficient hydrogenation. Theoretical insights into interfacial electronic structure and reaction energetics. a Adsorption energy of HMF at Ami-Cu/Cu3N, Ami-Cu3N, Ami-Cu interface and Cu/Cu3N, Cu3N, Cu surface. b Electron localization function patterns of Ami-Cu, Ami-Cu3N and Ami-Cu/Cu3N. c Electron density difference plots of the Ami-Cu, Ami-Cu3N and Ami-Cu/Cu3N interface, respectively. d Gibbs free energy diagrams for L-H pathway of HMF hydrogenation on Ami-Cu/Cu3N, Ami-Cu3N and Ami-Cu. e pDOS plots of Cu/Cu3N, Cu3N, Cu, Ami-Cu/Cu3N, Ami-Cu3N, and Ami-Cu. f Rigid energy band diagrams illustrating the interaction between Cu 3d orbitals and the −C=O highest occupied molecular orbital (HOMO) are presented. The diagrams correspond, respectively, to conditions in which unpaired electrons are absent (left), present to a lesser extent (center), and present in greater abundance (right) within the Cu 3d orbitals. Source data for a, d and e are provided as a Source Data file. Two distinct electron transfer pathways were identified for aldehyde group hydrogenation: the L-H and CPET mechanisms. For the L-H pathway, Ami-Cu/Cu3N exhibited the lowest activation energy (−2.11 eV) for the first hydrogenation step at the oxygen site (Figs. 6d and S88), initiating ECH via formation of an alkyl radical. In the second hydrogenation step, Ami-Cu/Cu3N showed a reaction barrier of 0.69 eV for DHMF* formation, lower than that of Cu/Cu3N (0.76 eV) and significantly lower than Cu3N (1.04 eV) and Cu (1.10 eV). This demonstrates that nitrogen incorporation, through both Cu3N and amide functionalization, synergistically reduced the thermodynamic energy barrier for keto radical hydrogenation. The highest barrier in the second step constitutes the potential-determining step (PDS) for all catalysts. To examine the sensitivity of the reaction thermodynamics to electrode polarization, an auxiliary charged-slab analysis was performed by varying the total electron number in the simulation cell. The methodological details are provided in the Supplementary Information. Cathodic polarization induces a general stabilization of the reaction intermediates (Supplementary Table 14). Crucially, this energetic shift is uniform across the Ami-Cu/Cu3N, Ami-Cu3N, and Ami-Cu models. Consequently, neither the relative stability order of the intermediates nor the identity of PDS is altered by the applied potential (Supplementary Fig. 89). This confirms that the mechanistic conclusions and the activity trends established herein are robust against potential-dependent field effects. Due to keto radical coupling, the L-H mechanism requires a higher barrier for BHH formation, making it more unfavorable on Ami-Cu/Cu3N, consistent with their enhanced ECH efficiency toward DHMF. In contrast, for the CPET pathway (Supplementary Figs. 90 and 91), Cu exhibited lower energy barriers for direct DHMF formation, suggesting Cu prefers CPET, a finding corroborated by experimental results. The influence of amide functionalization on electronic structure was further probed via density of states (DOS) calculations. As shown in Fig. 6e, the Cu/Cu3N heterostructure exhibits a higher density of Cu 3d states near the Fermi level compared with Cu3N and metallic Cu, indicating enhanced electrical conductivity that facilitates charge transfer during electrocatalysis. Upon amide functionalization (Ami-Cu/Cu3N, Ami-Cu3N, and Ami-Cu), the d-band centers (εd) of all catalysts shift upward toward the Fermi level relative to their unmodified counterparts, suggesting strengthened interactions between the catalysts and adsorbates. Physically, this upshift originates from the amide-induced inductive effect (Supplementary Fig. 87), which creates an electron-deficient environment at the Cu sites. Upon HMF adsorption, the d-band center shifts to lower energy due to orbital stabilization. The Ami-Cu/Cu3N system retains a relatively higher d-band center compared to the control samples, indicating enhanced electronic coupling between adsorbed *HMF and the active sites. Mechanistically, the adsorption of HMF on Cu/Cu3N induces electron transfer and orbital mixing, pushing the anti-bonding states above the Fermi level and increasing the population of unpaired electrons (holes) in the d-band (Fig. 6f). Furthermore, amide functionalization enhances HMF adsorption on the catalyst surface, further optimizing the electronic state of the Ami-Cu/Cu3N system. This modulation facilitates the depopulation of Cu-adsorbate anti-bonding orbitals (labeled as “Less-filling” in Fig. 6f), thereby strengthening intermediate adsorption and stabilizing the catalytic interface. Ami-Cu/Cu3N also possesses a decisive electronic advantage for hydrogen activation and supply. Lattice-N sites serve as the primary reservoir for *H generation, and they are electronically tuned by the neighboring amide groups. Although Amide-N does not act as an independent hydrogen-activation site, it exerts a long-range electronic modulation that upshifts the interfacial Cu d-states. This strengthens the Cu−H interaction at adjacent Cu–Nlatt motifs, facilitating water activation. Consequently, the Gibbs free energy for H adsorption (ΔG*H) is optimized to 1.34 eV on Ami-Cu/Cu3N, enhanced to both pristine Cu3N (1.67 eV) and Cu (1.76 eV) (Supplementary Fig. 92). Collectively, these findings elucidate a sophisticated dual-nitrogen synergy. The lattice-N skeleton establishes a fundamental polar thermodynamic platform for basal reactivity, which is distinctively potentiated by amide-N functionalities. Acting as a bifunctional modulator, the amide group orchestrates a concerted mechanism, simultaneously locking the HMF substrate via geometric hydrogen bonding and fine-tuning the electronic structure to optimize *H supply, thereby minimizing the energetic span for efficient ECH. Discussion In summary, a heterogeneous catalyst, denoted as Ami-Cu3N/CF, featuring surface amide functionalization on a Cu3N substrate, was successfully developed. This catalyst achieves both high FE and ECH activity under elevated HMF concentrations. The synergistic interaction between copper nitride and the amide groups enhances the efficient generation of *H and targeted HMF adsorption, thereby finely tuning the local microenvironment during the ECH process. Notably, the Ami-Cu3N/CF catalyst accomplishes HMF electroreduction with a FE exceeding 98% at an operating potential of merely −26 mV, and even at an HMF concentration of 1000 mM, the FE for DHMF remains above 90%. Moreover, the dual-electrode flow electrolytic cell lowers the cell voltage and energy consumption, thereby improving the economic feasibility of biomass upgrading and demonstrating significant practical potential. This work provides a foundational understanding and a design blueprint for cooperative active sites, paving the way for advanced electrocatalytic systems in renewable chemical synthesis. Methods Materials Except noted, all chemicals were purchased and used without further purification. Deionized water was used throughout the experiments. Cu foam substrates were purchased from Jinghong New Energy Technology Co., Ltd (Zhengzhou). Sodium hydroxide (NaOH) was purchased from Tianjin Zhiyuan Chemical Reagent Co. Ltd. Ammonium persulfate ((NH4)2S2O8, AR, 98.5%) was purchased from Rhawn Co. Ltd. Potassium hydroxide (KOH) was purchased from Tianjin Guangfu technology development Co. Ltd. Hydrochloric acid (HCl) was purchased from Sinopharm reagent Co. Ltd. Dipotassium hydrogen phosphate (K2HPO4), Potassium dihydrogen phosphate (KH2PO4), Benzene-1,3,5-Benzenetricarboxylic acid (C9H6O6, BTC), 5-hydroxymethylfurfural (HMF, 99%), 2,5-furandicarboxylic acid (FDCA, 99%), 5-hydroxymethyl-2-furancarboxylic acid (HMFCA, 99%), 2,5-diformylfuran (DFF, 99%), 5-formyl-2-furancarboxylic acid (FFCA, 99%), benzyl alcohol (BA, 99%), Benzaldehyde (BAH, 99%), Benzoic acid (BZA, 99%), furfural (FF, 99.5%), 2- furoic acid (FA, 99%), 4-methoxybenzyl alcohol (MBA, 98%), 4-methoxybenzyldehyde (MBH, 98%) and 2,5-dihydroxymethylfuran (DHMF, 98%) were purchased from Aladdin Chemical Co. Ltd. Electrode preparation Pretreatment of the Cu foam (CF) The purchased Cu foam (CF) was firstly cut into 3 × 4 cm2. The CF was sonicated in acetone for 15 min to remove impurities on the surface. After washing with water, the CF was transferred into the 0.5 M HCl for 1 min to remove the oxide layers on the surface, rinsed with distilled water for several times to remove any additional HCl. Synthesis of Cu(OH)2/CF Cu(OH)2 nanowire arrays were grown on Cu foam by surface chemical oxidation in an alkaline medium. The Cu(OH)2/CF were fabricated by immersing the thoroughly cleaned Cu foam into a solution containing 0.12 M NaOH and 0.01 M ammonium persulfate ((NH4)2S2O8) at 5 °C temperature for 30 min without disturbance. The chemically oxidized Cu(OH)2/CF were gently rinsed with DI water and then dried at room temperature. Synthesis of Cu/CF The contrast Cu/CF was obtained by heating the as-formed Cu(OH)2/CF at 400 °C for 3 h in H2 atmosphere at a ramping rate of 2 °C min−1. Synthesis of Cu3N/CF The contrast Cu3N/CF was obtained by heating the as-formed Cu(OH)2/CF at 325 °C for 2 h in NH3 atmosphere at a ramping rate of 5 °C min−1. Synthesis of BTC-Cu/CF The octahedral BTC-Cu particles, used as MOF precursors, were synthesized according to previous research with some modifications. Briefly, a 3 × 4 cm2 Cu(OH)2/CF was placed in 500 mL of methanol solution containing 8.6 g C6H3(COOH)3 and kept at room temperature for 12 h. The resulting BTC-Cu/CF washed several times with methanol and finally dried in an oven at 60 °C. Synthesis of Ami-Cu3N/CF The contrast Ami-Cu3N/CF was obtained by heating the as-formed BTC-Cu/CF at 325 °C for 2 h in NH3 atmosphere at a ramping rate of 5 °C min−1. The catalyst loading on CF was determined by gravimetric analysis before and after catalyst growth, yielding an average metal-containing catalyst (Supplementary Table 2). Electrode preparation The as-prepared Cu-based materials were directly used as self-supported electrodes without additional binders. The geometric area of the working electrode was precisely defined (typically 1 cm2 unless otherwise specified). Catalyst loading was determined based on the mass difference before and after synthesis. Characterizations The morphology and phase composition of as prepared catalysts were studied by SEM (Hitachi S-4800 at an accelerating voltage of 5 kV), TEM (JEM-2100 at an accelerating voltage of 200 kV), and XRD (Bruke D8 diffractometer using Cu Kα (λ = 1.5406 Å) radiation with accelerating voltage 40 kV). The chemical state was characterized by XPS, performed on VG ESCALAB MK II using a Mg Kα (1253.6 eV) achromatic X-ray radiation. Center of gravity calculation for the valence band spectra is conducted as ∫N(ε)εdε∫N(ε)dε in the range −5~10 eV, where N(ε) is the density of states. The BET equation was used to calculate the specific surface area. The electron paramagnetic resonance (EPR) tests of radicals were carried out on the Bruker EMX plus model spectrometer. The Cu K-edge X-ray absorption spectroscopy was collected in the transmission mode using inhouse laboratory-based X-ray absorption spectrometer (Super XAFS). All XAFS data were analyzed using the program Demeter70. For all samples, the EXAFS oscillations were extracted from the normalized XAFS spectra by subtracting the atomic background using a cubic spline fit to k3-weighted data, where k is the photoelectron wave number. The χ(k) functions were then Fourier transformed into R-space. The Fourier-transform window was in the k range 3–11 Å−1. Cu, CuO, Cu2O and Cu3N single crystals were measured simultaneously for energy calibration of Cu K-edge. Soft XAS were collected at the Taiwan Light Source (TLS) beamline BL-11A of the National Synchrotron Radiation Research Center (NSRRC). The N-K spectra were obtained using the total electron yield (TEY) mode. Scanning Kelvin probe (SKP) measurement (SKP5050 system, Scotland) was performed in ambient atmosphere, and a gold electrode was used as the reference electrode. Thermogravimetric analysis (TGA) was performed using a Pyris Diamond thermogravimetric/differential thermal analyzer (TG/DTA) at a constant dry inert atmosphere flow rate of 10 °C min−1. In-situ Fourier transform infrared (FTIR) spectroscopy measurement utilized a hemicylindrical CaF2, and the spectra were collected by Thermo Scientific Nicolet iS50 FTIR. FTIR optical accessory was produced by Shanghai Linglu Co. Ltd. Quasi-in-situ XRD, XPS and FTIR were conducted to investigate the structural and chemical evolution of catalysts after HMF ECH. In detail, the electrode after electrolysis for different HMF ECH time was immediately removed from the cell, rinsed with deionized water or ethanol to remove residual electrolyte and adsorbed species, and dried under inert gas flow prior to characterization. The total air exposure time was minimized to suppress additional oxidation or reconstruction. While slight surface oxidation during transfer cannot be completely excluded, the dominant features observed by quasi-in-situ XRD and XPS are considered representative of the reconstructed catalyst state established under reaction conditions. The content of Cu leaching after HMF ECH was determined by an inductively coupled plasma-atomic emission spectrometer (ICP-AES). Electrochemical measurements The electrochemical performances of all the samples were performed with a three-electrode system using a CHI electrochemical workstation (Model 760E). All reported potentials were quoted with respect to the reversible hydrogen electrode (RHE) through ERHE = EAg/AgCl + 0.059 × pH + 0.197 VRHE. All the tests for electrochemical reduction of HMF were conducted under magnetic stirring of 500 rpm. All measurements were conducted at ~25 ± 2 °C. Cyclic voltammetry (CV) was conducted at a scan rate of 100 mV s−1 to activate the catalysts. All LSVs requiring iR compensation were performed using the potentiostat’s positive-feedback mode with a compensation fraction of 95%, which was selected to maximize ohmic loss compensation while avoiding potential oscillation. The compensated potential was calculated as: where Emeas. is the raw (uncompensated) potential, Rs is the series resistance, and the compensation fraction was fixed at f = 0.95. To ensure consistency across different electrolytes and electrodes, Rs was measured individually for each system prior to the corresponding electrochemical tests using the high-frequency intercept of the electrochemical impedance spectroscopy (EIS) (frequency range: 105 to 10−2 Hz, with an amplitude of 10 mV). The Rs values are reported in Supplementary Tables 10–12 and explicitly indicated in the corresponding figure and table captions. In contrast, no real-time iR compensation was applied during any CV measurements, chronoamperometry (CA) and chronopotentiometry (CP) experiments to avoid potential feedback-induced instabilities under prolonged high-current operation. Unless otherwise noted, all CA/CP potentials discussed in the main text refer to raw, uncompensated values, with the corresponding raw potential-time traces provided in the Supporting Information. In-situ EIS measurements were performed over a frequency range from 105 to 10−2 Hz with AC amplitude of 10 mV with an AUTOLAB electrochemical workstation at room temperature. All the experiments were carried out at room temperature. The stability and durability were explored by ten continuous electrolysis cycles at a constant current density of 200 mA cm−2. During the test, another 20 mM HMF was subsequently added to the system after the previous run was completed. Electrolyte preparation Phosphate buffer solution (PBS, 0.5 M) was prepared freshly using analytical-grade reagents and deionized water prior to each experiment. The electrolyte was used immediately after preparation to ensure reproducibility. Flow cell configuration A custom-built flow electrolyzer was used for high-current-density electrolysis. The cell consisted of two compartments separated by a Nafion 212 membrane (thickness ~50 μm), which was pretreated by sequential boiling in H2O2, deionized water, and H2SO4. The catholyte (0.5 M PBS containing HMF) and anolyte were circulated using peristaltic pumps. The working electrode (Ami-Cu3N/CF) was placed on the cathode side, while a Ni-based electrode was used as the anode. Product analysis The liquid products were quantified by high-performance liquid chromatography (HPLC; Scion LC6000) using calibration curves constructed from authentic standards. In addition to DHMF, the HPLC method was able to resolve and quantify the C–C coupling product BHH. BHH was clearly observed on Cu/CF and became detectable on Cu3N/CF at elevated HMF concentrations, while no quantifiable BHH signal was detected on Ami-Cu3N/CF within the current HPLC detection window. Accordingly, BHH was included in the product distribution and Faradaic efficiency calculations where applicable. For electrooxidation product analysis, the HPLC system was equipped with a UV-vis detector (265 nm) and a 4.6 mm × 250 mm Shim-pack GWS C18 column (5 μm). During chronoamperometry at 1.432 V (three-electrode configuration), 10 μL of electrolyte was withdrawn, acidified with 100 μL of 0.1 M HCl, diluted with 390 μL of water, and analyzed at 30 °C. Methanol and 5 × 10−3 M ammonium formate aqueous solution were used as mobile phases A and B (A:B = 3:7) at a flow rate of 1 mL min−1. The identification and quantification of HMF and electrocatalytic reduced products were conducted by comparing them to authentic samples, with the exception of BHH. As there was no commercially available sample for BHHs, the calibration was estimated to be twice that of DHMF, as stated in previous reports37. The present HPLC-based analysis captures major liquid-phase products but may not fully exclude trace higher-molecular-weight or poorly detectable species outside the current analytical window. Product identification and quantification were performed using calibration curves obtained from standard solutions of commercially available reactants, intermediates, and final products. Under the reported conditions, the total molar (carbon) balance based on HMF and all detected liquid products determined by HPLC typically closes to >98%. The theoretical total charge for HMF electrooxidation was calculated as: The HMF conversion, FDCA yield, and FE were calculated using the following equations, respectively where F is the Faraday constant (96,485 C mol −1). For reductive product analysis, HPLC measurements were performed at 230 nm using a binary gradient of water and acetonitrile (CH3CN) at a flow rate of 0.6 mL min−1 and a column temperature of 45 °C. DHMF and HMF eluted at approximately 6.0 and 7.1 min, respectively. The passed charge was calculated as: where 2 is the number of electron transfers for hydrogenation of HMF to DHMF, 96,485 C mol−1 is the Faraday constant. The FE value and product selectivity of electrocatalytic reductive were calculated with the following equations: where n is the number of electron transfers for each product formation, n = 2 for DHMF, F is the Faraday constant (96,485 C mol−1). Liquid-phase products were analyzed by HPLC using calibrated standards, yielding a molar closure of ≥98%. In addition to DHMF, the C−C coupling product BHH was detected and quantified. Specifically, BHH was prominently observed on Cu/CF and became increasingly detectable on Cu3N/CF at elevated HMF concentrations, whereas it remained below the detection limit on Ami-Cu3N/CF. Accordingly, BHH was explicitly included in the product distribution and FE analysis wherever detected. For conditions where no additional quantifiable liquid-phase reduction products were observed, the remaining charge was conservatively attributed to H2 evolution. where selectivity is normalized to the converted HMF to avoid potential bias from undetected minor products and to ensure consistency with the closed molar balance. The average productivity of DHMF was accorded to de Luna’s report the area corresponds to the geometric area of electrodes. SKP measurement Scanning Kelvin probe (SKP) measurement (SKP5050 system, Scotland) was performed in ambient atmosphere, and a gold electrode was used as the reference electrode. The work function (Φ) was calculated by the following formula where ΦAu is the work function of Au, ΦAu = −239.75 eV, η = 5.1 eV. EQCM measurements Quartz crystal microbalance (QCM) operates based on the inverse piezoelectric effect of a quartz crystal and the change of the resonant frequency due to the change in the mass of the specimen. EQCM measurements were taken by CHI400B series. The CHI400B series EQCM contains a potentiostat/galvanostat (CHI440B), an external box with oscillator circuitry, and the EQCM cell. The EQCM cell consists of three round Teflon pieces. The total height is 37 mm with a diameter of 35 mm. The top piece is the cell top to hold the reference and counter electrodes. There are also two 2 mm holes for manual purging. The centerpiece is the cell body for the solution. The bottom piece is for mounting purposes. Four screws are used to tighten the bottom piece and center piece together. Do not over-tighten the screws to avoid damage of the screw threads. The quartz crystal is located between the center and bottom pieces. The seal is through two O-rings that are pressed together by the four screws mentioned above. The diameter of the quartz crystal is 13.7 mm. The gold electrode diameter is 5.11 mm. The QCM experiment relies on the calculation of mass change (∆m) at the quartz crystal using measured frequency change (∆f) (Sauerbrey equation): where f0 is the resonant frequency of the fundamental mode of the crystal, A the area of the gold disk coated onto the crystal (0.205 cm2 for given crystal), ρ the density of the crystal and µ the shear modulus of quartz. The density of the crystal is 2.648 g cm−3 and the shear modulus m is 2.947 × 1011 g cm−1 s−2. For 8 MHz crystal, the mass change is 0.14 ng for 0.1 Hz frequency change. In this work, all deposited film layers are regarded as thin and rigid films. EQCM measurements were carried out using a three-electrode system with an Au quartz crystal resonator as the working electrode, a platinum wire as the counter electrode and Ag/AgCl as the reference electrode. The geometric area of the Au quartz crystal resonator with a standard frequency of f0 = 8.000 M Hz. is 0.205 cm2. The electrode was cleaned in an aqueous solution containing H2O2 and H2SO4 (0.1 M) in the ratio of 1:3 followed by thorough rinsing with deionized water. After drying with nitrogen, the electrode was then immersed in the electrochemical cell. Theoretical calculations All the DFT calculations were performed by the projector-augmented-wave method implemented in the Vienna Ab-initio Simulation Package (VASP). The generalized gradient approximation of Perdew-Burke-Ernzerhof was applied to describe the exchange-correlation potential. The energy cutoff for the plane wave was set as 400 eV. The conjugate gradient algorithm was employed to predict the lowest energy position based on the energy and strain changes at each step. The convergence thresholds for optimization were 1 × 10−5 eV in energy and 0.01 eV Å−1 in force. Monkhorst–Pack mesh was used for k-space samplings with set as 3 × 3 × 3 k-points sampled for the optimization of all structures and 3 × 3 × 1 k-points sampled for the computation of electronic structure. The water solvation effect was also considered by using VASPsol. Transition states and energy barriers were calculated using the climbing-image nudged elastic band (Cl-NEB) method. DFT calculations The climbing-image nudged elastic band (Cl-NEB) is a method for finding saddle points and minimum energy paths between known reactants and products. The method operates by optimizing a series of intermediate images along the reaction pathway. Each image finds the lowest energy possible while maintaining equal spacing to neighboring images. This constrained optimization is done by adding spring forces along the band between images and by projecting out the component of the force due to the potential perpendicular to the band. Construction of DFT models To elucidate the reaction mechanism at the atomic scale, the catalyst models were constructed to represent the experimentally inferred steady-state surface motifs identified by post-reaction characterizations (XRD, HRTEM, and XPS), rather than to simulate the dynamic electrochemical reconstruction process itself. Specifically, the heterostructure model was built by interfacing a Cu(111) slab with a Cu3N(100) slab to capture the local Cu/Cu3N boundary formed after the self-limited reconstruction observed experimentally. Pristine Cu3N and Cu were modeled using stoichiometric Cu3N(100) and Cu(111) surfaces, respectively, corresponding to the primary crystallographic facets identified by XRD. Based on the experimental evidence that amide groups act as persistent surface ligands under reaction conditions, amide molecules were placed selectively at the Cu–Cu3N phase boundary to represent the local anchored configuration inferred for the reconstructed interface. This interfacial model was therefore used as a static, experimentally constrained descriptor of the post-reconstruction active motif for analyzing electronic modulation, adsorption behavior, and reaction energetics. A vacuum layer of 15 Å was applied to avoid periodic interactions, and the bottom atomic layers were fixed while the interface and adsorbates were fully relaxed. Physical basis and implementation of the charged-slab approach To evaluate the sensitivity of reaction thermodynamics to electrode polarization, we performed auxiliary calculations using the charged-slab approach. The physical basis and computational details are as follows. The electrode potential (U) in an electrochemical cell is fundamentally linked to the electronic chemical potential (μe) of the interface. In our DFT model, we modulate μe by explicitly varying the total number of electrons (Nelect) in the simulation cell. This process shifts the Fermi level (EF) and the work function (Ф) of the slab, effectively mimicking the charging of the electrode surface under bias: U ∝ Ф/e + constant. Calculations were performed by adjusting the NELECT parameter (e.g., NELECT = NELECT0 + 0.5 e or +1.0 e). To maintain the requirements of periodic boundary conditions and avoid electrostatic divergence, a uniform compensating background charge (Jellium background) was automatically introduced by VASP. This background charge serves as a first-order approximation for the counter-ions in the electric double layer (EDL), ensuring that the perturbation arises solely from the modification of the surface electronic state. For each adsorbed intermediate i, the potential-induced energy shift (ΔEcharge, i) was calculated relative to the neutral surface (q = 0): where Ei(q) is the total energy of intermediate i on the charged slab and Ei(0) is the energy on the neutral surface. These intermediate-specific shifts account for the Stark effect and electronic reorganization at the polarized interface, providing a semi-quantitative probe of the potential-determining step (PDS) sensitivity (Supplementary Fig. 89 and Table 14). We emphasize that this charged slab approach does not rigorously correspond to a constant-potential electrochemical model, nor does it explicitly account for solvent or electric double-layer effects. In addition, as is well known for charged slab calculations under periodic boundary conditions, the total energies may exhibit slow convergence with respect to vacuum thickness due to interactions with the compensating background charge. Therefore, the present analysis is intended as a semi-quantitative sensitivity probe, aimed at assessing whether moderate surface charging can qualitatively alter the relative stability of reaction intermediates and the identity of the potential-determining step, rather than providing quantitatively accurate potential-dependent energetics. Graphics and visualization Graphical schematics in Figs. 1a, 3a and Supplementary Figs. 10 and 32 were created using Microsoft PowerPoint (Microsoft Office), while the process flow diagram was generated using Cinema 4D (Maxon Computer GmbH). All graphics were produced by the authors without the use of third-party copyrighted materials requiring additional licensing. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Transparent Peer Review file Source data Source Data Acknowledgements We gratefully acknowledge the support of this research by the National Key Research and Development Program of China (2023YFA1507204, H.F.), the National Natural Science Foundation of China (22539002, H.F. and 22179035, Z.R.), Shanghai Science and Technology Innovation Action Plan (22142200300, J.W.) and the Photon Science Center for Carbon Neutrality (J.W.). H.F. discloses support for the research of this work from the National Key Research and Development Program of China (Grant No. 2023YFA1507204) and the National Natural Science Foundation of China (Grant No. 22539002). Z.R. discloses support for the research of this work from the National Natural Science Foundation of China (Grant No. 22179035). J.W. discloses support for the research of this work from the Shanghai Science and Technology Innovation Action Plan (Grant No. 22142200300) and the Photon Science Center for Carbon Neutrality. J.H., J.Z., B.L., C.D., Z.C., H.W., G.L., S.D., and Z.J.X. declare no relevant funding. Author contributions J.H. designed and performed the experiments and wrote the original manuscript. J.Z. and J.W. conducted the synchrotron radiation experiments. B.L. performed the theoretical calculations. C.D. contributed to data interpretation and provided critical feedback on the manuscript. Z.C. provided guidance on data analysis. H.W., G.L., and S.D. assisted with experimental measurements and data collection. Z.R. supervised the experimental design and manuscript revision. Z.J.X. and H.F. provided critical review and editorial input. All authors discussed the results and contributed to the final manuscript. Peer review Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Data availability All data generated or analyzed during this study are included in this Article (and its Supplementary Information). Source data are provided with this paper. The data that support the findings of this study are available in the paper and its Supplementary Information or from the corresponding author upon request. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Zhiyu Ren, Email: zyren@hlju.edu.cn. Zhichuan J. Xu, Email: xuzc@ntu.edu.sg Honggang Fu, Email: fuhg@hlju.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-72690-4. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Transparent Peer Review file Source Data Data Availability Statement All data generated or analyzed during this study are included in this Article (and its Supplementary Information). Source data are provided with this paper. The data that support the findings of this study are available in the paper and its Supplementary Information or from the corresponding author upon request. Source data are provided with this paper.