Manipulating spin-state conversion to promote asymmetric d–p orbital hybridization for high-efficiency nitrate electroreduction to ammonia

Abstract

The electrochemical nitrate reduction reaction (eNO₃⁻RR) presents a sustainable solution for water pollutant management and green ammonia (NH₃) synthesis. However, hindered by the spin-forbidden barrier, the sluggish hydrogenation kinetics of the key intermediate *NO severely limits the production of NH₃. Here, we reported for the first time the realization of a controllable transition of the inner Co spin-state from a low spin to a high spin in CuCo₂O₄ through the Mn doping-driven oxygen vacancy strategy (Mn–CuCo₂O_4−x). The elevated Co spin-state enhanced Co 3d (d_xz/d_yz/d_z²)–*NO 2p asymmetrical orbital hybridization, facilitating *NO intermediate adsorption and the subsequent hydrogenation. Thanks to the Cu–Co synergistic effect enhanced via spin-state modulation, the Mn–CuCo₂O_4−x/graphene oxide aerogels (GAs) exhibited an attractive NH₃ yield rate of 2.14 mg h⁻¹ cm⁻² with a dramatic NH₃ faradaic efficiency of 98.37% at an environmentally relevant NO₃⁻ level (10 mM NO₃⁻–N), far superior to that of Co₃O₄/GAs, CuCo₂O₄/GAs and as-reported catalysts. Moreover, the strong interfacial interaction between GAs and Mn–CuCo₂O_4−x suppresses structural reconstruction of Mn–CuCo₂O_4−x, endowing the hybrid with robust stability. Herein, we confirm that spin-state modulation can enhance the Cu–Co synergistic effect and reveal a universal strategy to optimize intermediate adsorption/conversion through the spin-state, opening up a new avenue for deep purification of water pollutants based on spin optimization and providing general principles for the rational design of catalytic materials.

---

1. Introduction

Increasingly serious problems of nitrogen cycle disruption and environmental contamination pose significant threats to the global ecosystem and public health.^1–3 Nitrate (NO₃⁻), a predominant contaminant in industrial effluents and agricultural runoff, is identified as a critical contributor to eutrophication, drinking water pollution and ecosystem disruption.^4,5 Concurrently, ammonia (NH₃) serves as the core raw material for agricultural fertilizers and clean hydrogen energy carriers. The conventional NH₃ production method depends on the energy-consuming Haber–Bosch process, generating a considerable carbon footprint.^6,7 The electrochemical nitrate reduction reaction (eNO₃⁻RR) can convert NO₃⁻ into value-added NH₃ under mild conditions, providing a sustainable solution to the above dual challenges.^8–10 Despite its promise, the practical implementation of the eNO₃⁻RR is fundamentally throttled by the kinetically sluggish hydrogenation of a critical intermediate (*NO), where a spin-forbidden barrier imposes severe limitations on reaction rates and faradaic efficiency (FE).

The hydrogenation of *NO constitutes the rate-determining step (RDS) in the eNO₃⁻RR, as its spin-polarized triplet ground state inherently resists forming singlet *HNO intermediates—a classic manifestation of spin selection rules. The Cu–Co bimetallic system in the eNO₃⁻RR partially mitigates this kinetic barrier through functional decoupling, in which the Cu site preferentially adsorbs NO₃⁻ and drives the initial deoxygenation reaction and the Co site is responsible for the subsequent *NO hydrogenation conversion.^11–13 However, current design paradigms predominantly focus on macro/meso-structure engineering (e.g., alloying, defect modulation, and reconstruction).^14–16 These approaches fail to deeply modulate the essential electronic structure determinants. As a core feature of the d-orbital electron arrangement of transition metals, the spin-state can significantly change the orbital adsorption strength and hybridization mode of metal intermediates, thereby impacting the reaction process.¹⁷ Spinel-type CuCo₂O₄ is an ideal platform for resolving the Co spin-state effect due to its unique structural tunability. Co³⁺ occupies the octahedral site in CuCo₂O₄, and its spin-state is significantly affected by the local coordination environment and lattice stresses.¹⁸ Some studies have shown that by introducing hetero-atoms for doping in Co-based materials to induce lattice distortions, the ligand field strength of Co³⁺ can be effectively tuned, consequently enabling controllable switching of the spin-state.^19–21

The spin-state of Co sites critically determines the d-orbital energy level configuration, where the HS state exhibits stronger ligand field splitting effects and higher occupancy in the e_g orbitals.^22,23 This electronic configuration enhances the adsorption and activation of reaction intermediates through optimized orbital interactions, thereby facilitating the RDS in catalytic processes.²⁴ Notably, compelling theoretical and experimental validation for such spin-dependent electronic modulation effects exists in analogous spin-sensitive electrochemical processes, such as the oxygen reduction and evolution reactions, yet this potent strategy remains conspicuously unexplored and underutilized within the eNO₃⁻RR framework.^25,26 Consequently, it is imperative to elucidate whether HS Co sites can indeed enhance *NO adsorption and accelerate its hydrogenation kinetics through changing the orbital hybridization pattern with the *NO intermediate and to unravel the interplay between the Co spin-state and the established Cu–Co synergistic effect. These interesting validations will unlock unprecedented activity and selectivity for next generation NO₃⁻ reduction electrocatalysts.

Here, we utilized CuCo₂O₄ as a model catalytic system and proposed a Mn doping-driven oxygen vacancy strategy to precisely modulate the spin-state of Co. Experimental studies demonstrate that the HS state Co can greatly optimize the adsorption of *NO and solve the problem of slow hydrogenation kinetics in the RDS (*NO → *HNO) via enhancing the asymmetric hybridization of the Co 3d (d_xz/d_yz/d_z²)–*NO 2p orbitals. This spin state-optimized bimetallic synergistic effect endowed Mn–CuCo₂O_4−x/GAs with excellent catalytic performance (Faradaic Efficiency (FE): 98.37%, yield rate: 2.14 mg h⁻¹ cm⁻² at −0.6 V vs. RHE, and 10 mM NO₃⁻–N), exceeding that of Co₃O₄/GAs, CuCo₂O₄/GAs and as-reported catalysts. This work provides a theoretical paradigm for extending the spin-engineering strategy to bimetallic synergistic catalytic systems, while also inaugurating a new dimension of electronic structure modulation for designing efficient NO₃⁻ conversion catalysts.

2. Experimental section

2.1 Chemicals

All chemical reagents are depicted in Text S1.

2.2 Catalyst preparation

2.2.1 Preparation of GAs. Graphene aerogels (GAs) were synthesized by a hydrothermal combined calcination method. A standard synthesis protocol involved homogenizing 60 mg carboxymethyl cellulose (CMC) with 30 mL graphene oxide (GO) (2 mg mL⁻¹) under ultrasonic processing for 60 min. This precursor solution underwent hydrothermal treatment at 180 °C for 12 h. The synthesized hydrogels were purified using deionized water, freeze-dried, and then thermally treated at 900 °C for 2 h under N₂ flow to generate GAs. Finally, the GAs were functionalized with 0.2 mL aminopropyltriethoxysilane (APTES) to impart a positive surface charge.

2.2.2 Preparation of Mn–CuCo₂O_4−x yolk–shell spheres. A blend of Cu(NO₃)₂·3H₂O (0.1 mmol), Co(NO₃)₂·6H₂O (0.2 mmol), a certain amount of Mn(NO₃)₂·6H₂O, and 8 mL glycerol was introduced into 40 mL isopropanol and stirred to obtain homogeneous solution. This solution was moved into a 100 mL solvothermal reaction vessel lined with Teflon and maintained at a temperature of 180 °C over 6 h. The resulting precursor from the solvothermal reaction was subjected to a dual centrifugal rinse using deionized water and ethanol before undergoing a 12 h drying phase at 80 °C. The Mn–CuCo₂O_4−x yolk–shell spheres were then generated by placing the dried precursor in a tube furnace set at 400 °C and sustained for 1 h under O₂ conditions. Co₃O₄ and CuCo₂O₄ were prepared similarly to Mn–CuCo₂O_4−x except that the corresponding metal salts were not added.

2.2.3 Preparation of Mn–CuCo₂O_4−x/GAs. The Mn–CuCo₂O_4−x/GAs composite was self-assembled from GAs and CuCo₂O₄. More specifically, an initial step involves dispersing 0.1 g Mn–CuCo₂O_4−x yolk–shell spheres in an isopropanol solution loaded with poly(sodium 4-styrenesulfonate) (PSS) (concentration: 0.2 mL PSS and 20 mL isopropanol), yielding PSS-treated Mn–CuCo₂O_4−x yolk–shell spheres. These spheres were subsequently cleaned twice by centrifugation using alcohol and deionized water. The cleaned product was then dispersed in deionized water and gently added to a GA suspension (60 mL, 0.5 mg mL⁻¹) and stirred for 180 min to achieve a dispersion of Mn–CuCo₂O_4−x/GAs composite material. The Mn–CuCo₂O_4−x/GAs composite material was isolated via vacuum filtration and sequentially purified with ethanol-aqueous solution under ambient conditions. Co₃O₄/GAs and CuCo₂O₄/GAs were prepared by the same procedure as that of Mn–CuCo₂O_4−x/GAs.

2.2.4 The preparation of the working electrode. 10 mg of the catalysts were re-dispersed into 470 μL of ethanol and 470 μL of deionized water, followed by adding 60 μL of Nafion solution to make a homogeneous solution. Subsequently, it was ultrasonicated for about 1 h to enable the good dispersion of catalyst inks. Afterwards, 100 μL of catalyst inks were dropped onto the carbon paper with an area of 1 cm² (1 cm × 1 cm).

2.3 Characterization

More details on the characterization methods are provided in the SI (Text S2).

2.4 Electrochemical measurements

Electrochemical tests were performed using a three-electrode system connected to a CHI 760E electrochemical workstation in a typical H-cell. The H-cell was separated using a Nafion 117 membrane (Dupont) that was pretreated following reported procedures. The catalysts were used as the working electrode, while an Ag/AgCl electrode and platinum mesh were used as the reference and counter electrodes, respectively. The electrolytes were Ar-saturated 0.1 M Na₂SO₄ containing 10 mM NO₃⁻–N. All potentials were calibrated to the reversible hydrogen electrode (RHE) scale using the Nernst equation (E_RHE = E_Ag/AgCl + 0.197 V + 0.059 V × pH). The current density was normalized to the geometric electrode area (∼1 cm²). The linear sweep voltammetry (LSV) test was carried out in a three-electrode system at scanning rates of 10 mV s⁻¹. Electrochemical impedance spectroscopy (EIS) measurements were conducted in a frequency range from 10⁵ to 0.01 Hz with 5 mV amplitude. Cyclic voltammetry (CV) curves were obtained in the non-faradaic region with different scan rates (20, 40, 60, 80, and 100 mV s⁻¹).

2.5 Analytical methods

Detailed information of nitrogen species concentration determination and the calculation of the eNO₃⁻RR parameters are provided in Text S3 and S4, respectively. Detailed steps of in situ Fourier transform infrared spectroscopy (FT-IR), online differential electrochemical mass spectrometry (DEMS), and electron spin resonance (ESR) measurements are described in Text S5–S7, respectively. Density functional theory (DFT) calculations are presented in Text S8.

3. Results and discussion

3.1 Catalyst characterization

The preparation process of the Mn–CuCo₂O_4−x/GAs composite framework is shown in Fig. 1a. First, Mn–CuCo₂O_4−x yolk–shell nanospheres were prepared via solvothermal combined high-temperature calcination. Subsequently, Mn–CuCo₂O_4−x/GAs was synthesized via a charge neutralization process. A thorough examination of the microstructure of Mn–CuCo₂O_4−x was conducted using scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Mn–CuCo₂O_4−x possessed a well-defined yolk–shell architecture, as clearly observed in Fig. 1b and c, characterized by an average particle diameter of approximately 400 nm and distinct surface roughness. Selected area electron diffraction (SAED) analysis confirmed the polycrystalline characteristics of Mn–CuCo₂O_4−x (Fig. 1d). Elemental mapping via energy-dispersive spectroscopy (EDS) confirmed homogeneous spatial distribution of Mn, Cu, Co, and O constituents throughout the yolk–shell spheres (Fig. 1e–i). Inductively coupled plasma mass spectrometry (ICP-MS) quantified Mn content at 9.36 wt% (Table S1). The combined results from EDS and ICP-MS analysis conclusively verified the successful incorporation of Mn within the composite structure. The Mn–CuCo glycerate precursors underwent structural evolution into yolk–shell architectures via the Ostwald maturation mechanism, as also supported via comparative TEM analyses of Co₃O₄ and CuCo₂O₄ phases (Fig. S1–S4).²⁷ To impart Mn–CuCo₂O_4−x with negative electrical properties, we utilized PSS for its modification via electrostatic self-assembly. During the PSS modification process, the sulfonic acid group (SO₃⁻) of PSS was adsorbed on the surface of Mn–CuCo₂O_4−x through electrostatic interaction, forming a stable negatively charged layer (Fig. S5).²⁸ Simultaneously, the amino group (–NH₂) of the APTES molecule underwent a protonation reaction with protons in solution to form positively charged ammonium ions (–NH₃⁺), and –NH₃⁺ was stably anchored to the surface of the GAs via covalent bonds, thus conferring a positively charged character.²⁹ Then, the oppositely charged Mn–CuCo₂O_4−x and GAs were mixed to form the Mn–CuCo₂O_4−x/GAs composites in situ through a charge neutralization process. As illustrated in Fig. 1j and k, Mn–CuCo₂O_4−x was embedded into the 3D porous skeleton of GAs to form a stable Mn–CuCo₂O_4−x/GAs composite framework. The introduction of GAs promoted rapid electron transport while effectively maintaining the stability of the structure, as will be detailed in the subsequent stability characterization section. High-resolution TEM (HRTEM) images demonstrated clear lattice streaks and GA layers, and the measured crystal plane spacing of 0.245 nm corresponded to the (311) crystal plane of CuCo₂O₄, confirming that the spinel structural features were maintained after Mn doping (Fig. 1l). TEM images of Co₃O₄/GAs and CuCo₂O₄/GAs composites suggested that Co₃O₄ and CuCo₂O₄ were also tightly coupled to the GA scaffolds, thus confirming the general applicability of the charge neutralization strategy in constructing metal oxide/GAs composites (Fig. S6 and S7).

Fig. 1 (a) The synthesis schematic of Mn–CuCo₂O_4−x/GAs. (b) SEM, (c) TEM, (d) SAED, and (e–i) elemental mapping of Mn–CuCo₂O_4−x. (j) SEM, (k) TEM, and (l) HRTEM images of Mn–CuCo₂O_4−x/GAs (inset: HRTEM images of Mn–CuCo₂O_4−x).

The successful synthesis of the CuCo₂O₄ spinel-type framework was confirmed by the X-ray diffraction (XRD) analysis showing consistent crystallographic diffraction peaks at (220), (311), (400), (422), (440), and (533) with the standard card (PDF#78-2177) (Fig. 2a).³⁰ After Mn doping, the diffraction peaks all exhibited a systematic shift to the low angle direction, among which the displacement of the (311) crystal plane was the most significant, originating from the lattice expansion triggered by the difference in ionic radii between Mn and Co. A weakening of the diffraction peak intensity and a slight broadening of the half height width were also observed, suggesting that the doping process may lead to a decrease in crystallinity. The Raman analysis of CuCo₂O₄ exhibited characteristic peaks at about 475, 521, 627, and 684 cm⁻¹ attributed to the E_2g, F_2g, and A_1g vibrational modes of the CoO₆ octahedron (Fig. 2b).³¹ Since there is a spectral overlap between the F_2g vibrational mode of the CuO₄ tetrahedron (190 cm⁻¹) and the F_2g vibrational mode of the tetrahedral CoO₄ tetrahedron (195 cm⁻¹), the characteristic signals of the CuO₄ tetrahedron were not effectively recognized.^32,33 The Raman peak of A_1g in CuCo₂O₄ was shifted toward the low wave number direction, which originated from the lattice substitution effect of the CoO₆ octahedron by the MnO₆ octahedron with a larger radius. The larger ionic radius can induce local lattice distortion, leading to an increase in metal–oxygen bond length and structural stress, which is consistent with the theoretical model of cation substitution-induced bond length rearrangement.³⁴ Meanwhile, the characteristic vibrational modes of the D and G bands of the graphitic carbon layer of GAs were detected in Raman spectra, confirming the effective coupling of CuCo₂O₄ to GAs.

Fig. 2 (a) XRD and (b) Raman spectra of Co₃O₄/GAs, CuCo₂O₄/GAs, and Mn–CuCo₂O_4−x/GAs. (c) High-resolution XPS images of Mn 2p of Mn–CuCo₂O_4−x/GAs. (d) Mn K-edge XANES spectra of Mn–CuCo₂O_4−x/GAs and the standard reference materials, and (e) their corresponding FT-EXAFS spectra. (f) Mn K-edge WT-EXAFS contour plots for Mn–CuCo₂O_4−x/GAs, Mn foil, MnO, Mn₂O₃ and MnO₂. (g) Co K-edge XANES spectra of CuCo₂O₄/GAs and Mn–CuCo₂O_4−x/GAs and the standard reference materials, and (h) their corresponding FT-EXAFS spectra. (i) EPR images of CuCo₂O₄/GAs and Mn–CuCo₂O_4−x/GAs; (j) the influence of Ov on the electron occupancy of Co 3d orbitals. (k) Magnetic hysteresis loops of CuCo₂O₄/GAs and Mn–CuCo₂O_4−x/GAs.

The chemical valence and coordination environments of the elements were characterized using X-ray photoelectron spectroscopy (XPS) analysis. The XPS full spectra revealed characteristic peaks associated with Mn–CuCo₂O_4−x and GAs, indicating the successful construction of CuCo₂O₄ and GA composites (Fig. S8). The high-resolution XPS spectra of the Cu 2p orbital presented characteristic double peaks at 933.9 eV (2p_3/2) and 954.0 eV (2p_1/2), confirming that the Cu element exists in the +2-valence state in CuCo₂O₄ and Mn–CuCo₂O_4−x (Fig. S9).³² Remarkably, the doping of Mn did not change the chemical state of Cu since the binding energy positions and ratios of the two orbitals were not changed. The high-resolution XPS spectra of the Mn 2p orbital were divided into Mn 2p_3/2 and Mn 2p_1/2, and the predominant oxidation state of Mn in the samples was Mn⁴⁺ (Fig. 2c).³⁵ The Co 2p orbital of CuCo₂O₄ exhibited double state features at 779.7/781.5 eV (2p_3/2) and 796.6/794.7 eV (2p_1/2), confirming that Co²⁺ and Co³⁺ coexist in the material, and Co³⁺ occupies the main component (Fig. S10).^32,36 Notably, the decrease in the Co³⁺/Co²⁺ ratio in Mn–CuCo₂O_4−x confirmed the preferential occupation of CoO₆ octahedral sites by Mn⁴⁺. The O 1s orbitals presented peaks at 529.6, 531.1, and 532.6 eV, corresponding to metal–oxygen (M–O), Ov, and H₂O, respectively, and the Ov concentration of Mn–CuCo₂O_4−x was higher than that of CuCo₂O₄ (Fig. S11).³⁷ X-ray absorption spectroscopy (XAS) elucidated the electronic state and local coordination of metal elements. Based on the linear combination fitting of the absorption edge positions (Fig. 2d and S12a), the oxidation state of Mn in Mn–CuCo₂O_4−x was found to be +3.8, which is consistent with the XPS results. Extended X-ray absorption fine structure (EXAFS) analysis of the Mn K-edge (Fig. 2e and S13) revealed no evidence of the Mn–Mn bond, confirming the atomic dispersion of Mn and absence of clusters. The wavelet transform (WT) spectroscopy in R-space (Fig. 2f) of Mn foil, MnO, Mn₂O₃, MnO₂, and Mn–CuCo₂O_4−x demonstrated spectral similarity between Mn–CuCo₂O_4−x and octahedrally coordinated MnO₂. These results confirm an octahedral coordination environment for Mn, supporting its primary substitution at octahedral Co sites. The normalized Co K-edge X-ray absorption near-edge structure (XANES) spectra (Fig. 2g and S12b) of CuCo₂O₄ was located between Co₃O₄ and Co₂O₃, but closer to Co₂O₃, indicating that Co was predominantly +3 valence in CuCo₂O₄. In contrast, the Co K-edge XANES spectra of Mn–CuCo₂O_4−x were shifted towards lower energy, suggesting that the substitution of Co³⁺ by Mn⁴⁺ leads to an increase in the Co²⁺ ratio. Co K-edge EXAFS analysis (Fig. 2g) revealed a significant reduction in oscillation amplitude for Mn–CuCo₂O_4−x compared to pristine CuCo₂O₄, indicative of increased local structural disorder around Co atoms induced by the incorporation of high-valent Mn⁴⁺ ions. Both the Co K-edge EXAFS spectra of CuCo₂O₄ and Mn–CuCo₂O_4−x and their corresponding R-space WT spectra (Fig. S14 and S15) show no evidence of Co–Co bonding contributions. Moreover, the EXAFS fitting result (Table S2) of the first-shell peak for CuCo₂O₄ and Mn–CuCo₂O_4−x revealed that the introduction of Ov reduces the average coordination number of Co.

Prior research has indicated that the addition of metal doping can create lattice distortions and reduce the strength of the metal–oxygen bond, thereby promoting the creation of oxygen vacancies (Ov).³⁸ Consequently, we analyzed the variations in the Ov concentration in the samples using electron paramagnetic resonance (EPR) tests. Compared with CuCo₂O₄/GAs, Mn–CuCo₂O_4−x/GAs exhibited significantly enhanced signal intensity in EPR spectra, confirming the elevated Ov concentration (Fig. 2i). Since Ov are positively charged defects, they would release free electrons into the surroundings to maintain electrical neutrality. These electrons can be trapped by neighboring Co³⁺ to form transient Co²⁺, enhancing the spin-state of Co³⁺. In the pristine CuCo₂O₄, Co³⁺ is usually in a LS state (t_2g⁶e_g⁰) with a high octahedral coordination field splitting energy, and the electrons preferentially populate the t_2g orbitals in the lower energy levels (Fig. 2j). The introduction of Ov induces localized lattice distortion. This structural alteration further modifies the coordination environment of transition metal ions, causing their coordination field splitting energy to fall below the critical threshold. At this point, Co³⁺ transitions from the LS state to the HS state (t_2g⁴e_g²), and the number of unpaired electrons increases to four. To validate the proposed introduction of Ov altering the Co³⁺ spin-state, we measured sample magnetism via vibrating sample magnetometry. As displayed in Fig. 2k, original CuCo₂O₄ exhibited a weak saturation magnetization intensity whereas the saturation magnetization intensity was significantly increased by the introduction of Ov. This phenomenon indicated that Ov may induce the transition of Co³⁺ from the LS to the HS state through modulation of the localized crystal field and electron occupation, leading to a larger number of spin electrons. In summary, Ov can significantly enhance the number of spin-polarized electrons via triggering the HS state transition of Co³⁺, promoting a more effective “spin promoter” for the orbital interaction between the catalyst and the *NO intermediate.

3.2 Electrocatalytic performance for the NO₃⁻RR

The electrocatalytic activities of Co₃O₄/GAs, CuCo₂O₄/GAs, and Mn–CuCo₂O_4−x/GAs were assessed in an H-type cell. NO₃⁻, NO₂⁻, and NH₄⁺ concentrations during the eNO₃⁻RR were quantified via UV-Vis spectrophotometry (Fig. S16–S18). The linear sweep voltammetry (LSV) curves revealed enhanced current densities for Co₃O₄/GAs, CuCo₂O₄/GAs, and Mn–CuCo₂O_4−x/GAs versus NO₃⁻-free systems, confirming their intrinsic NO₃⁻ reduction activity (Fig. 3a). Notably, CuCo₂O₄/GAs exhibited higher current density than Co₃O₄/GAs, implicating synergistic advantages of Cu–Co bimetallic sites in the eNO₃⁻RR. Additionally, Mn–CuCo₂O_4−x/GAs possessed higher current density and more positive starting reduction potential compared with CuCo₂O₄/GAs, suggesting that Mn doping further can enhance the catalytic activity of the eNO₃⁻RR. Electrochemical impedance spectrum (EIS)-based analysis of electron transport kinetics showed that Mn–CuCo₂O_4−x/GAs possessed the smallest capacitive arc radius in the Nyquist plot (Fig. 3b). Moreover, the R_ct values showed a dramatic and conclusive trend: Co₃O₄/GAs (66.17 Ω) > CuCo₂O₄/GAs (26.69 Ω) > Mn–CuCo₂O_4−x/GAs (15.64 Ω), which was in excellent agreement with the observed catalytic activity (Table S3). This decrease in R_ct unequivocally demonstrated that the charge transfer kinetics for the eNO₃⁻RR was significantly enhanced upon introducing Cu, indicating that the bimetallic synergy drastically facilitated the electron transfer process. Most impressively, Mn–CuCo₂O_4−x/GAs achieved the lowest R_ct value, which provided direct electrochemical evidence that the Mn doping and the induced Ov created a more favorable electronic structure, further accelerating the rate-determining electron transfer steps in the eNO₃⁻RR. Electrochemical active surface area (ECSA) was quantified by electrochemical double-layer capacitance (C_dl) measurements from cyclic voltammetry (CV) in non-faradaic windows (Fig. S19). A higher C_dl value was obtained for Mn–CuCo₂O_4−x/GAs (5.20 mF cm⁻²) versus Co₃O₄/GAs (1.62 mF cm⁻²) and CuCo₂O₄/GAs (2.22 mF cm⁻²), confirming that Cu–Co bimetallic synergism together with Mn doping can increase the active site density, thus enhancing the intrinsic activity of the eNO₃⁻RR. Based on the LSV results, −0.4 to −0.8 V vs. RHE was identified as the characteristic potential window. The chronoamperometry quantified the FE and NH₃–N yield rates of catalysts (Fig. S20). The trend of volcanic distribution of FE for each catalyst confirmed the existence of a critical starting overpotential for the eNO₃⁻RR (Fig. 3c). Excessively negative potentials attenuated FE due to the competing hydrogen evolution reaction (HER). Mn–CuCo₂O_4−x/GAs exhibited higher FE than both counterparts across tested potentials, achieving a maximum of 98.37% at −0.6 V vs. RHE. Mn–CuCo₂O_4−x/GAs presented the highest NH₃–N yield rate at all potentials; especially at the voltage corresponding to the optimal FE, the NH₃–N yield rate was 2.14 mg h⁻¹ cm⁻² (Fig. 3d). Additionally, the selectivity of the catalysts was further evaluated based on the yield rate and FE of the intermediate NO₂⁻ (Fig. S21). As the applied voltage increased, both the NO₂⁻ yield rate and corresponding FE of the three catalysts exhibited a gradual decrease. Among them, Mn–CuCo₂O_4−x/GAs maintained the lowest values for both metrics, indicating its greater ability to promote the further conversion of the intermediate NO₂⁻. This highlights the outstanding selectivity of this catalyst in the NO₃⁻ reduction process.

Fig. 3 (a) LSV curves, (b) EIS curves, (c) FE, and (d) NH₃–N yield rates of Co₃O₄/GAs, CuCo₂O₄/GAs, and Mn–CuCo₂O_4−x/GAs. (e) ¹H NMR spectra for ¹⁴NO₃⁻ and ¹⁵NO₃⁻ electrolyte after 1 h of electrolysis. (f) Comparison of eNO₃⁻RR performance of Mn–CuCo₂O_4−x/GAs with that of reported electrocatalysts. (g) In situ Raman spectra of Mn–CuCo₂O_4−x/GAs at different voltages. (h) The cycling stability tests of Mn–CuCo₂O_4−x/GAs.

The properties of the nitrogen source in NH₃–N were verified through a blank control. The results revealed almost negligible ammonia production in the blank control, clearly confirming that the nitrogen in the product originated from NO₃⁻ rather than from environmental pollutants (Fig. S22). ¹H-nuclear magnetic resonance (¹H-NMR) was used to monitor the dynamics of the nitrogen conversion pathways. When ¹⁵NO₃⁻ and ¹⁴NO₃⁻ were employed as nitrogen sources, the corresponding ¹⁵NH₄⁺ and ¹⁴NH₄⁺ showed characteristic double-peak and triple-peak splitting patterns, providing direct evidence for the directional conversion of NO₃⁻ to NH₃ (Fig. 3e). Although Mn–CuCo₂O_4−x/GAs had excellent catalytic activity, the performance comparison between this material and similar advanced catalysts still needs to be clarified. To this end, we systematically summarized the recent literature for the low-concentration NO₃⁻ system (Fig. 3e and Table S4). Both the FE and NH₃–N yield rate of Mn–CuCo₂O_4−x/GAs were significantly better than those of other reported systems, fully demonstrating the leading position of this catalyst in the eNO₃⁻RR. The structural evolution of the Mn–CuCo₂O_4−x/GAs electrocatalyst during the eNO₃⁻RR was dynamically tracked using in situ Raman spectroscopy. In the absence of GAs, the characteristic Raman peak corresponding to the A_1g vibrational mode of Mn–CuCo₂O_4−x exhibited a rapid decrease in intensity as the applied potential shifted negatively, indicating insufficient structural stability under operating conditions (Fig. S23). In contrast, after the introduction of GAs, the same characteristic Raman mode remained clearly detectable even at highly negative potentials up to −0.7 V vs. RHE, demonstrating significantly enhanced structural integrity (Fig. 3g). This pronounced contrast underscores the critical role of GAs in stabilizing the active material through strong interfacial interactions, which effectively anchor Mn–CuCo₂O_4−x, prevent agglomeration or dissolution, and thus maintain the structural and catalytic durability of the Mn–CuCo₂O_4−x/GAs composite throughout the reaction. The FE and NH₃–N yield rate of the system exhibited only minor fluctuations during the continuous electrolysis test for up to 20 h, demonstrating excellent cycling stability (Fig. 3h). The catalysts before and after the reaction were systematically characterized by XRD and TEM, and neither the crystal structure nor the microscopic morphology significantly changed, further verifying the excellent structural stability of Mn–CuCo₂O_4−x/GAs at the microscopic level (Fig. S24 and S25).

3.3 Reaction pathways and mechanisms

Time-resolved electrochemical mass spectrometry (DEMS) identified key eNO₃⁻RR intermediates on CuCo₂O₄/GAs and Mn–CuCo₂O_4−x/GAs (Fig. 4a, b and S26). m/z signals at 15, 16, 17, 30, and 31 revealed *NH, *NH₂, *NO, HNO, and NH₃ intermediate generation. Notably, the characteristic signal peak intensity of the relevant intermediates on Mn–CuCo₂O_4−x/GAs was significantly higher than that on CuCo₂O₄/GAs, suggesting that the introduction of Ov can significantly enhance the generation capacity of the key intermediates. Operando Fourier transform infrared spectroscopy (FT-IR) tracked real-time evolution of surface intermediates during the reaction (Fig. 4c and S27). The vibrational peaks at 1650 cm⁻¹ (δH–O–H) and 3300 cm⁻¹ (νH–O–H) corresponded to interfacial water adsorption and cleavage.^39,40 The progressive intensification of vibrational signatures at 1460 cm⁻¹ and 3200 cm⁻¹ can be unambiguously assigned to the stretching vibrations of N–H bonds in NH₃ molecules.^41,42 Based on the key intermediate species and their dynamic evolution patterns monitored online in this study, combined with the mechanistic models reported in the literature, we systematically deduced and constructed the reaction pathway dominated by the stepwise hydrogenation of *NO (Fig. S28). The dynamic evolutionary behavior of active hydrogen species (*H) during the eNO₃⁻RR process was systematically investigated by the in situ EPR technique (Fig. 4d). The 5,5-dimethyl-1-pyrroline-N-oxide (DMPO)–*H spin adduct signal intensity for Mn–CuCo₂O_4−x/GAs doubled that of CuCo₂O₄/GAs in 0.1 M Na₂SO₄ electrolyte lacking NO₃⁻, confirming that Mn doping effectively promoted hydrolytic dissociation to generate *H. Intriguingly, upon introducing 10 mM NO₃⁻ into the electrolyte, the characteristic signal intensity corresponding to the *H on Mn–CuCo₂O_4−x/GAs exhibited significant attenuation during the electrocatalytic process. This apparent inhibitory effect may stem from the rapid depletion caused by *H and NO₃⁻ reactions. These findings demonstrate that NO₃⁻ reduction on Mn–CuCo₂O_4−x/GAs proceeded via an *H-mediated indirect pathway and the Ov promoted the hydrogenation process of nitrogen-containing intermediates. To further evaluate the interfacial H₂O dissociation capability of the prepared catalysts, kinetic isotope effect (KIE) tests were conducted by replacing H₂O with D₂O. The KIE value was calculated by comparing the current densities measured in H₂O and D₂O, thereby revealing the dynamic process of *H transfer during H₂O dissociation (Fig. S29). The results indicated that Mn–CuCo₂O_4−x/GAs exhibits an average KIE value of 1.24, significantly lower than that of other reference electrocatalysts, demonstrating that the introduction of Ov effectively promotes interfacial H₂O dissociation. Concurrently, it also demonstrated that Ov possesses a significant interface water regulation behavior. Gibbs free energy change (ΔG) plots for CuCo₂O₄ and Mn–CuCo₂O_4−x in the eNO₃⁻RR are shown in Fig. 4e. Both CuCo₂O₄ and Mn–CuCo₂O₄ synthesize NH₃via the 8-electron reduction pathway (NO₃⁻ → *NO₃ → *HNO₃ → *NO₂ → *NO → *HNO → *NH → *NH₂ → *NH₃ → NH₃) (Fig. S30 and S31). The ΔG profile identified *NO hydrogenation to *HNO as the rate-determining step (RDS), exhibiting the highest energy barrier. Apparently, Ov lowered the RDS energy barrier from 1.39 eV to 0.41 eV, thereby enhancing eNO₃⁻RR kinetics. Promoting the dissociation of H₂O to produce *H and inhibiting the coupling of *H to produce H₂ are essential to enhance the performance of the eNO₃⁻RR. As illustrated in Fig. 4f, S32 and S33, the H₂O adsorption energy of Mn–CuCo₂O_4−x (−0.41 eV) was significantly negatively shifted compared with that of CuCo₂O₄ (−0.19 eV), indicating that its activation of H₂O was largely enhanced. The H₂O dissociation energy barrier was reduced from 0.32 eV to 0.14 eV, effectively promoting the generation of *H active species. Meanwhile, the ΔG of H₂ generation on the surface of Mn–CuCo₂O_4−x (ΔG = 0.38 eV) showed a significant positive shift compared with that of CuCo₂O₄ (ΔG = 0.07 eV), suggesting that Ov significantly inhibited the generation of H₂ by the coupling of *H. In conclusion, Ov can lower the hydrogenation barrier of *NO → *HNO, accelerating the H₂O dissociation and inhibiting the HER side reaction, thereby enhancing selective NO₃⁻-to-NH₃ conversion.

Fig. 4 (a and b) Online DEMS spectra of CuCo₂O_4−x/GAs and Mn–CuCo₂O_4−x/GAs. (c) In situ FT-IR spectra of Mn–CuCo₂O_4−x/GAs. (d) EPR spectra of CuCo₂O_4−x/GAs and Mn–CuCo₂O_4−x/GAs in electrolytes without/with NO₃⁻. Gibbs free energy profiles for (e) eNO₃⁻RR and (f) HER on CuCo₂O₄ and Mn–CuCo₂O_4−x. (g) DOS of *NO and *HNO species. (h) Differential charge density of critical intermediates adsorbed on CuCo₂O₄ and Mn–CuCo₂O_4−x interfaces. (i) PDOS analyses of *NO intermediates adsorbed on CuCo₂O₄ and Mn–CuCo₂O_4−x. (j) Schematic illustration of orbital hybridization of Co and *NO in different spin states. (k) Spin charge density of CuCo₂O₄ and Mn–CuCo₂O_4−x.

After identifying *NO → *HNO as the RDS, density functional theory (DFT) deciphered orbital interaction mechanisms between CuCo₂O₄/Mn–CuCo₂O_4−x and reaction intermediates. Density of states analysis of *NO and *HNO intermediates revealed pronounced spin asymmetry in *NO across spin-up/down channels, contrasting with spin symmetry in *HNO (Fig. 4g). This shift in spin configuration revealed a spin rearrangement phenomenon occurring during the reaction, which essentially stems from the pairing process of unpaired electrons. Differential charge density analysis revealed that *NO adsorbed on the surface of Mn–CuCo₂O_4−x (0.28|e|) had a higher electron transfer number compared with CuCo₂O₄ (0.16|e|), suggesting that Ov intensified the interfacial electron transfer process and the bonding tendency to the *NO intermediates was stronger (Fig. 4h).⁴³ Meanwhile, enhanced *NO adsorption was observed on Mn–CuCo₂O_4−x (−1.96 eV) relative to CuCo₂O₄ (−1.32 eV). Thus, the spin-state transition promoted *NO adsorption on catalysts, corroborated by projected density of states (PDOS) analysis. Fig. 4i reveals a significant overlap region between the hybridization of the Co-3d orbitals in Mn–CuCo₂O_4−x and the N-2p orbitals of *NO and the hybridization energy intervals skewed towards the Fermi energy level, suggesting the existence of a stronger orbital interaction in Mn–CuCo₂O_4−x and *NO.⁴³ Furthermore, Mn–CuCo₂O_4−x enhanced *NO hydrogenation by stabilizing critical *HNO intermediates, which can be illustrated by the fact that *HNO adsorbed on Mn–CuCo₂O_4−x exhibited a larger electron transfer number and a more negative adsorption energy than CuCo₂O_4−x.⁴⁴ Based on the previous analysis of the orbital configuration of the Co high spin-state (t_2g⁴e_g²), its d_xz/d_yz/d_z² orbitals can form directional hybridization with the σ/π* antibonding orbitals of *NO, whereas the d_xy and d_x²−y² orbitals are excluded from the effective coupling regime due to symmetry mismatch.^43,45 In this hybridization, the d_z² orbitals of Co can accept electron occupation from the σ orbitals while the d_xz/d_yz orbitals feed electrons back to the π* orbitals. This Co 3d (d_xz/d_yz/d_z²)–*NO 2p asymmetrical orbital hybridization enhanced the adsorption of *NO intermediates and weakened the N–O bond, thereby facilitating *NO hydrogenation to *HNO (Fig. 4j). The unique catalytic properties of CuCo₂O₄ and Mn–CuCo₂O_4−x can also be illustrated by the magnetic moment of Co orbitals. The magnetic moment of Co increased from 1.356μ_B to 1.661μ_B after the introduction of Ov (Fig. 4k). The elevated magnetic moment is directly associated with the elevated occupancy of the HS state of Co, and the increase in the number of its unpaired electrons facilitates electron exchange between the active site and *NO, accelerating the kinetics of *NO hydrogenation. To summarize, by combining experimental characterization with DFT calculations, we revealed that the elevation of the internal Co spin state in the Cu–Co bimetallic active sites can further enhance the Cu–Co synergistic catalysis through the role of orbital hybridization.

4. Conclusions

In summary, the spin-state of Co³⁺ in spinel CuCo₂O₄ was modulated by an Mn doping-driven Ov strategy, and the controllable transition from the LS state to the HS state was successfully realized, significantly enhancing the Cu–Co bimetallic synergistic catalysis. Experiments and DFT revealed that the HS state Co³⁺ can optimize the adsorption of the key intermediate *NO and hydrogenation step via enhancing the asymmetric hybridization of the Co 3d (d_xz/d_yz/d_z²)–*NO 2p orbitals, thus accelerating the eNO₃⁻RR kinetic process. Mn–CuCo₂O_4−x/GAs demonstrated superior electrocatalytic performance, achieving an NH₃ yield rate of 2.14 mg h⁻¹ cm⁻² and a faradaic efficiency of 98.37% at −0.6 V vs. RHE, outperforming both Co₃O₄/GAs and CuCo₂O₄/GAs, as well as previously reported catalysts. This work proposes the first theoretical correlation between spin-state modulation and bimetallic synergistic catalysis, offering a new avenue for the design of high-performance catalytic systems for the deep purification of water pollutants and resource recovery.

Author contributions

Ke Wang and Tong Zhao: data curation and writing – original draft. Hou Wang and Shiyu Zhang: investigation. Rupeng Wang: formal analysis. Meng Wang: writing – review & editing. Zixiang He: supervision. Shih-Hsin Ho: funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Data availability

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d5sc06823a.

Acknowledgements

This work received research funding from the National Natural Science Foundation of China (No. 52070057), State Key Laboratory of Urban–Rural Water Resources and Environment (Harbin Institute of Technology) (No. 2025DX06), and State Key Laboratory of Water Pollution Control and Green Resource Recycling Foundation (No. PCRRF25010).

References

1. J. G. Chen, R. M. Crooks, L. C. Seefeldt, K. L. Bren, R. M. Bullock, M. Y. Darensbourg, P. L. Holland, B. Hoffman, M. J. Janik, A. K. Jones, M. G. Kanatzidis, P. King, K. M. Lancaster, S. V. Lymar, P. Pfromm, W. F. Schneider and R. R. Schrock, Science, 2018, 360, eaar6611.
2. Y. Wang, C. Wang, M. Li, Y. Yu and B. Zhang, Chem. Soc. Rev., 2021, 50, 6720–6733.
3. P. H. van Langevelde, I. Katsounaros and M. T. M. Koper, Joule, 2021, 5, 290–294.
4. Y. Xue, Q. Yu, Q. Ma, Y. Chen, C. Zhang, W. Teng, J. Fan and W.-x. Zhang, Environ. Sci. Technol., 2022, 56, 14797–14807.
5. W. Duan, Y. Chen, H. Ma, J.-F. Lee, Y.-J. Lin and C. Feng, Environ. Sci. Technol., 2023, 57, 3893–3904.
6. S. Han, H. Li, T. Li, F. Chen, R. Yang, Y. Yu and B. Zhang, Nat. Catal., 2023, 6, 402–414.
7. F.-Y. Chen, Z.-Y. Wu, S. Gupta, D. J. Rivera, S. V. Lambeets, S. Pecaut, J. Y. T. Kim, P. Zhu, Y. Z. Finfrock, D. M. Meira, G. King, G. Gao, W. Xu, D. A. Cullen, H. Zhou, Y. Han, D. E. Perea, C. L. Muhich and H. Wang, Nat. Nanotechnol., 2022, 17, 759–767.
8. K. Fan, W. Xie, J. Li, Y. Sun, P. Xu, Y. Tang, Z. Li and M. Shao, Nat. Commun., 2022, 13, 7958.
9. K. Wang, R. Mao, R. Liu, J. Zhang, H. Zhao, W. Ran and X. Zhao, Nat. Mater., 2023, 1, 1068–1078.
10. S. Zhang, J. Wu, M. Zheng, X. Jin, Z. Shen, Z. Li, Y. Wang, Q. Wang, X. Wang, H. Wei, J. Zhang, P. Wang, S. Zhang, L. Yu, L. Dong, Q. Zhu, H. Zhang and J. Lu, Nat. Commun., 2023, 14, 3634.
11. G.-F. Chen, Y. Yuan, H. Jiang, S.-Y. Ren, L.-X. Ding, L. Ma, T. Wu, J. Lu and H. Wang, Nat. Energy, 2020, 5, 605–613.
12. H. Liu, X. Lang, C. Zhu, J. Timoshenko, M. Rüscher, L. Bai, N. Guijarro, H. Yin, Y. Peng, J. Li, Z. Liu, W. Wang, B. R. Cuenya and J. Luo, Angew. Chem., Int. Ed., 2022, 61, e202202556.
13. W. He, J. Zhang, S. Dieckhöfer, S. Varhade, A. C. Brix, A. Lielpetere, S. Seisel, J. R. C. Junqueira and W. Schuhmann, Nat. Commun., 2022, 13, 1129.
14. J.-Y. Fang, Q.-Z. Zheng, Y.-Y. Lou, K.-M. Zhao, S.-N. Hu, G. Li, O. Akdim, X.-Y. Huang and S.-G. Sun, Nat. Commun., 2022, 13, 7899.
15. Z. Ren, K. Shi, Z. Meng, M. D. Willis and X. Feng, ACS Energy Lett., 2024, 9, 3849–3858.
16. R. Yan, H. Yin, X. Zuo, W. Peng, X. Zhu, L. Shi, J. Hou, D. Wang, F. Ye, J. Li, B. Mao and C. Hu, Appl. Catal., B, 2025, 361, 124609.
17. Z. Gu, Y. Zhang, X. Wei, Z. Duan, Q. Gong and K. Luo, Adv. Mater., 2023, 35, 2303107.
18. Y. Zhou, R. Duan, H. Li, M. Zhao, C. Ding and C. Li, ACS Catal., 2023, 13, 10846–10854.
19. C. Zhou, C.-Z. Yuan, F. Jing, C.-H. Li, H. Zhao, Y. Sun and W. Yuan, J. Energy Chem., 2025, 106, 142–150.
20. Y. Xie, Y. Feng, S. Zhu, Y. Yu, H. Bao, Q. Liu, F. Luo and Z. Yang, Adv. Mater., 2025, 37, 2414801.
21. M. Duan, C. Huang, G. Zhang, H. Shi, P. Zhang, L. Li, T. Xu, Z. Zhao, Z. Fu, J. Han, Y. Xu and X. Ding, Angew. Chem., Int. Ed., 2024, 63, e202318924.
22. L. Zhao, C. Xin, C. Yu, Y. Xing, Z. Wei, H. Zhang, T. Fei, S. Liu, H. Zhang and T. Zhang, InfoMat, 2025, 7, e12634.
23. M. Xu, H. Zhou, X. Lv, Y. Fang, X. Tu, F. Wang, Q. Han, X. Wang and G. Zheng, Adv. Mater., 2025, 2505286.
24. J. Qi, Q. Bai, X. Bai, H. Gu, S. Lu, S. Chen, Q. Li, X. Yang, J. Wang and L. Wang, Adv. Sci., 2025, 2503665.
25. S. Zhang, W. Zhao, J. Liu, Z. Tao, Y. Zhang, S. Zhao, Z. Zhang and M. Du, Adv. Sci., 2024, 11, 2407301.
26. J. Ding, Z. Wei, F. Li, J. Zhang, Q. Zhang, J. Zhou, W. Wang, Y. Liu, Z. Zhang, X. Su, R. Yang, W. Liu, C. Su, H. B. Yang, Y. Huang, Y. Zhai and B. Liu, Nat. Commun., 2023, 14, 6550.
27. J. Qiao, C. Lu, L. Kong, J. Zhang, Q. Lin, H. Huang, C. Li, W. He, M. Zhou and Z. Sun, Adv. Funct. Mater., 2024, 34, 2409794.
28. H. Xue, J. Wang, H. Cheng, H. Zhang, X. Li, J. Sun, X. Wang, L. Lin, Y. Zhang, X. Liao and Y. He, Appl. Catal., B, 2024, 353, 124087.
29. Z. Huang, Y. Liang, Z. Wu, Y. Kong, M. Bai, M. Li, B. Hong, T. Huang, S. Huang, H. Chen and S. Zhang, Adv. Mater., 2025, 37, 2410318.
30. Y.-X. Zhang, Y.-D. Li, A.-K. Du, Y. Wu and J.-B. Zeng, J. Mater. Sci. Technol., 2024, 173, 114–120.
31. J. Geltmeyer, G. Vancoillie, I. Steyaert, B. Breyne, G. Cousins, K. Lava, R. Hoogenboom, K. De Buysser and K. De Clerck, Adv. Funct. Mater., 2016, 26, 5987–5996.
32. K. Liu, X. Gao, C.-X. Liu, R. Shi, E. C. M. Tse, F. Liu and Y. Chen, Adv. Energy Mater., 2024, 14, 2304065.
33. Y. Chen, Q. Li, R. Su, Y. Gao, N. An, Y. Rong, X. Xu, D. Ma, Y. Wang and B. Gao, Water Res., 2025, 123765.
34. H. Lin, J. Wei, Y. Guo, Y. Li, X. Lu, C. Zhou, S. Liu and Y.-y. Li, Adv. Funct. Mater., 2024, 34, 2409696.
35. X. Guo, W. Cai, Y. Zhuo and S. Liu, Appl. Catal., B, 2025, 373, 125354.
36. L. Wu, Q. Wu, Y. Han, D. Zhang, R. Zhang, N. Song, X. Wu, J. Zeng, P. Yuan, J. Chen, A. Du, K. Huang and X. Yao, Adv. Mater., 2024, 36, 2401857.
37. K. Wang, T. Zhao, N.-Q. Ren and S.-H. Ho, Water Res., 2024, 265, 122304.
38. Q. Deng, H. Li, K. Pei, L. W. Wong, X. Zheng, C. S. Tsang, H. Chen, W. Shen, T. H. Ly, J. Zhao and Q. Fu, ACS Nano, 2024, 18, 33718–33728.
39. M. Xie, G. Zhu, H. Yang, B. Liu, M. Li, C. Qi, L. Wang, W. Jiang, P. Qiu and W. Luo, Adv. Energy Mater., 2024, 14, 2401717.
40. J. Li, H. Li, K. Fan, J. Y. Lee, W. Xie and M. Shao, Chem Catal., 2023, 3, 100638.
41. Y. Huang, C. He, C. Cheng, S. Han, M. He, Y. Wang, N. Meng, B. Zhang, Q. Lu and Y. Yu, Nat. Commun., 2023, 14, 7368.
42. J. Ni, J. Yan, F. Li, H. Qi, Q. Xu, C. Su, L. Sun, H. Sun, J. Ding and B. Liu, Adv. Energy Mater., 2024, 14, 2400065.
43. J. Liang, S. Deng, Z. Li, M. Zhou, S. Wang, Y. Su, S. Yang and H. Li, Adv. Mater., 2025, 37, 2418828.
44. J. Dai, Y. Tong, L. Zhao, Z. Hu, C.-T. Chen, C.-Y. Kuo, G. Zhan, J. Wang, X. Zou, Q. Zheng, W. Hou, R. Wang, K. Wang, R. Zhao, X.-K. Gu, Y. Yao and L. Zhang, Nat. Commun., 2024, 15, 88.
45. Y. Sun, S. Sun, H. Yang, S. Xi, J. Gracia and Z. J. Xu, Adv. Mater., 2020, 32, 2003297.

---

Footnote

† These authors contributed equally to this work.

---