Non-linear spin correlation of intermediates in enhanced electrochemical nitrate reduction under magnetic fields†

Dongsheng Shao‡ ^ab, Qian Wu‡ ^b, Yuwei Zhang‡ ^b, Xiyang Cai ^b, Chencheng Dai ^b, Siyuan Zhu ^b, Fanxu Meng ^b, Pengfei Song ^b, Xiaoning Li ^b, Xiaoming Ren ^a, Tianze Wu *^b and Zhichuan J. Xu *^bc
aState Key Laboratory of Materials-Oriented Chemical Engineering and College of Chemistry and Molecular Engineering, Nanjing Tech University, Nanjing 211816, China
^bSchool of Materials Science and Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore. E-mail: tianze.wu@ntu.edu.sg; xuzc@ntu.edu.sg
^cThe Centre of Advanced Catalysis Science and Technology, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore

First published on 24th June 2025

---

---

Broader context Spin effects in electrocatalysis have shown great potential for overcoming kinetic and energetic limitations in a range of reactions. Notably, studies have demonstrated that spin-related phenomena can enhance the performance of ferromagnetic catalysts, particularly in oxygen electrocatalysis, where activity improvements are observed under applied magnetic fields. However, NH3 synthesis, especially via electrochemical nitrate reduction, involves a much more complicated elementary reaction than oxygen electrocatalysis, which prevents spin effects from being revealed in a straightforward way. In this study, beyond the observed enhancement of NO3−RR using ferromagnetic catalysts under magnetic fields, the improvements are associated with the net spin of the intermediates. The improvements are correlated with the net spins of reaction intermediates. Due to the diverse spin configurations of these intermediates, the relationship between catalytic enhancement and the degree of ferromagnetic ordering becomes non-linear and more intricate.

Introduction

Electrocatalysis plays a pivotal role in the sustainable use of renewable energy.^1–6 The development of efficient electrocatalysts depends on a deep understanding of the relationship between their electrocatalytic activity and electronic structure. Structure–activity relationships have been established based on relevant descriptors such as the d-band center, e_g occupancy, O 2p band energy, metal–oxygen covalency, valence state of the active site, charge-transfer energy and the number of d electrons.^3,7–11 Typically, electrocatalytic activity is regulated according to these descriptors to achieve optimal adsorption/desorption energetics of intermediates.^7,8 However, considering only the energetics of intermediates can maintain the limitations in the reaction kinetics. For instance, in oxygen electrocatalysis, *OH, *O, and *OOH intermediates can scale adsorption energies, introducing an energy barrier that is difficult to overcome.^12,13 In addition, electrochemical reactions can be limited by significant kinetic barriers, primarily caused by the high activation energy required for the conversion of intermediates.

Spin manipulation for catalysts potentially introduces an additional degree of freedom to overcome the scaling relationships for intermediates and the activation energy among intermediates with different spin-multiplicity.^14–16 The crucial roles of spin have been demonstrated in the reaction kinetics of oxygen electrocatalysis.^14–19 For example, in some DFT studies, the binding energy of reaction intermediates is regulated by open–shell quantum spin exchange interactions (QSEI), leading to higher catalytic activity with ferromagnetic (FM) coupling compared to that with antiferromagnetic (AFM) coupling.^20,21 This potentially bypasses the energy scaling between key intermediates and could lead to spin-related reaction pathways with more favourable energetics and kinetics.

Approaches to control electron spins include the manipulation of chirality and magnetic ordering of catalysts.^{15–19,22–25} Chirality-induced spin selectivity (CISS) effect describes the preferential transmission of electrons with a specific spin orientation through chiral molecules, effectively enabling them to function as spin filters. Zhang et al. synthesized chiral nanostructured Ag films and observed a pronounced improvement in faradaic efficiency (FE) for C₂⁺ products compared with racemic and achiral ones in the electrochemical CO₂ reduction reaction.²⁴ Other than molecular strategies, spin regulation can also be assisted by magnetic ordering of catalysts under an external magnetic field. Pioneering studies on the OER revealed that magnetizing ferromagnetic (FM) catalysts enhances current density under magnetic fields.^17–19

To realize effective spin regulation in catalysis, it is essential to establish the relationship between spin polarization and catalytic performance. Magnetic field-regulated oxygen evolution reaction (OER) has been investigated as a model system for exploring possible spin-related effects in catalysis (Fig. 1(a) and (b)).^{14,15,17–19} In FM catalysts, magnetization eliminates the domain wall region to evolve a single-domain state, which promotes the spin polarization of catalysts (Fig. 1(c)).¹⁹ In these cases, the OER enhancement showed a linear relationship with the degree of spin polarization. This linear relationship is based on a straightforward association between the O–O coupling efficiency and the spin polarization of intermediate oxygen radicals on the catalysts, which is typically rate-determining in OER (Fig. 1(b)).¹⁸ However, complicated spin correlations will provide more spin-related insights into the reaction kinetics of complex catalytic systems to extend the study of spin electrocatalysis. Recent progress for Haber–Bosch ammonia synthesis has demonstrated that spin-mediated promotion activates traditionally unreactive magnetic materials such as cobalt, suggesting an interesting role of spin in NH₃ synthesis.²⁶ As one of the methods for the electrochemical synthesis of NH₃, NO₃⁻RR involves eight electron transfer steps (NO₃⁻ + 6H₂O + 8e⁻ → NH₃ + 9OH⁻),^27–30 featuring complicated reaction intermediates including *NO₃, *NO₂, *NO, *NHO, *NH₂O, *NH₂OH, *NH and *NH₃.^31–35 In the pathway for NO₃⁻RR, the net spins in these intermediates frequently change (Fig. 1(d)),^36–39 which calls for the spin regulation in relevant elementary steps.

In this study, we showcase the complicated spin correlations in NO₃⁻RR by demonstrating activity enhancement under magnetic fields. Model catalysts, including Cu, Fe₃O₄, and CuFe₂O₄, were examined for NO₃⁻RR under magnetic field conditions. While the NH₃ yield rates and FE of Cu foil in the NO₃⁻RR process show negligible changes upon the application of a magnetic field, the magnetic catalysts Fe₃O₄ and CuFe₂O₄ exhibit significant NO₃⁻RR enhancements. The NH₃ yield rate and FE enhancement for CuFe₂O₄ reached 93.2% and 30.3%, respectively, at −1.1 V vs. RHE. Notably, an activity enhancement for NO₂⁻ production achieves more than one order of magnitude. The activity enhancement shows linear and non-linear correlations for NO₂⁻ and NH₃ production, respectively. Such a difference originates from the varied response of the elemental electron transfer step to magnetization according to the magnetic nature of key intermediates, as revealed by DFT simulation. This study presents essential fundamentals for NO₃⁻RR electrocatalyst optimization by spin manipulation and maximizing the practical potential of employing a magnetic field to elevate the intrinsic NO₃⁻RR activity.

Results and discussions

Spin-promoted NO₃⁻RR under a magnetic field condition

The magnetic CuFe₂O₄ (spinel oxide) was prepared by a conventional sol–gel method as previously reported.⁴⁰ The powder X-ray diffraction (XRD) pattern confirms the crystal structure. The pattern is consistent with that of the standard tetragonal spinel (I41/amd), indicating the high phase purity of the catalyst (Fig. S1, ESI†). Field emission scanning electron microscopy (FESEM) images of CuFe₂O₄ catalyst at different scales and the distribution of Cu, Fe, and O elements are shown in Fig. S2 and S3 (ESI†). The prepared sample nanoparticles have irregular shapes with a size of approximately 100 nm. Elemental mapping images and atom percentage analysis revealed that Cu, Fe, and O were homogeneously distributed. The magnetic properties of the employed catalysts, Cu foil, Fe₃O₄, and CuFe₂O₄, were investigated using a vibrating sample magnetometer (VSM) at room temperature. The MH curves for the employed catalysts were tested in the magnetic field range from −14 to +14 kOe, and are shown in Fig. 2(a). Cu foil exhibited typical diamagnetic behavior, whereas Fe₃O₄ and CuFe₂O₄ exhibited S-shaped hysteresis loops with a coercive field, indicating ferromagnetic behavior. The coercive field, which measures a material's resistance to demagnetization, was small for Fe₃O₄ at 107.8 Oe but significantly larger for CuFe₂O₄ at 1519.7 Oe.

The NO₃⁻RR performance of the Cu foil, Fe₃O₄, and CuFe₂O₄ was investigated in a 0.5 M Na₂SO₄ electrolyte with 0.1 M NaNO₃ at room temperature in an optimized setup (Fig. S4, see Methods for details). As shown in linear sweep voltammetry (LSV) curves of the catalysts in Fig. 2(b), the current density (j) of all catalysts shows an obvious increment after adding 0.1 M NaNO₃. Furthermore, the NO₃⁻RR performance of the ferromagnetic Fe₃O₄ and CuFe₂O₄ is significantly enhanced under an external magnetic field of 2500 Oe, while the changes in the nonmagnetic catalyst Cu foil are negligible. Once the magnetic field was removed, the improved NO₃⁻RR activity of ferromagnetic Fe₃O₄ promptly reverted nearly to its unmagnetized state, whereas the activity of CuFe₂O₄ after removing the magnetic field still exceeds the one before magnetization. This is attributed to the higher coercivity of CuFe₂O₄ than that of Fe₃O₄, which contributes to the maintenance of the spin polarization in materials after magnetization.

Chronoamperometry (CA) measurements were carried out at a constant potential of −1.1 V vs. RHE for 0.5 h, after which the NH₃ concentration was measured by ultraviolet-visible (UV-Vis) spectrophotometry (Fig. S5, ESI;† see Methods for details). As shown in Fig. 2(c), the NH₃ yield rate follows the order Cu foil < Fe₃O₄ < CuFe₂O₄ without an applied magnetic field. When an external magnetic field of 2500 Oe was applied, the NH₃ yield rate and FE of the nonmagnetic Cu foil catalyst showed negligible improvement at −1.1 V vs. RHE. In contrast, for Fe₃O₄, the NH₃ yield rate increased from 1.9 ± 0.1 to 6.0 ± 0.2 mg h⁻¹ cm⁻² (a 3.2-fold increase), and the FE improved from 38.3 ± 3.7% to 87.6 ± 3.4% (a 2.3-fold increase) under a 2500 Oe magnetic field. The corresponding NH₃ partial current density increased from −23.3 to −74.7 mA cm⁻² (Fig. 2(d)). Similarly, for CuFe₂O₄, the NH₃ yield rate increased from 3.1 ± 0.1 to 6.0 ± 0.2 mg h⁻¹ cm⁻² (a 1.9-fold increase), and FE increased from 65.6 ± 4.8% to 85.5 ± 2.9% (a 1.3-fold increase), with the NH₃ partial current density rising from −38.9 to −75.3 mA cm⁻² (Fig. 2(d)). To exclude the effects of magnetic fields on the mass transport, such as Lorentzian movement,¹⁷ we measured the NH₃ produced by Cu foil with and without a magnetic field under different potentials (Fig. S6, ESI†) and found a negligible improvement in NH₃ production. This evidence excludes the mass transport effects as a major contributor to the significant NO₃⁻RR enhancement under the magnetic field. To exclude the influence of the substrate electrode, we tested the NO₃⁻RR performance of a bare carbon paper electrode, which showed negligible nitrate reduction activity to NH₃ with and without an external magnetic field (Fig. S7, ESI†). In addition, we excluded the influence of magnetic fields on the electrochemical surface area (ECSA) of catalysts (Fig. S8 and S9, ESI†). However, the ECSA-normalized NO₃⁻RR activity still exhibited activity enhancement under the magnetic field (Fig. S10, ESI†).

CA measurements were conducted for Fe₃O₄ and CuFe₂O₄ at different potentials with and without an external magnetic field of 2500 Oe. The NH₃ yield rate and FE are shown as a function of applied potentials in Fig. 3(a) and Fig. S6 (ESI†). The significant activity enhancement for CuFe₂O₄ is found at potentials of −1.1 V and −0.9 V vs. RHE, while as the potential decreased to −0.7 V vs. RHE, the performance improvement is weakened and becomes minimal at −0.5 V and −0.3 V vs. RHE. The FE for NH₃ in NO₃⁻RR system is typically below 100%, primarily due to competing side reactions and minor NO₃⁻RR byproducts, including NO₂⁻ and potentially trace nitrogen-containing species. We quantify both NH₃ and NO₂⁻ faradaic efficiencies with and without an external magnetic field across a range of potentials to clarify the origin of the FE deficit (Fig. S11, ESI†).

During NO₃⁻RR, the effects of the magnetic field demonstrate profound scientific implications during the studies of elemental reactions (e.g., NO₃⁻→ NO₂⁻, NO₂⁻ → NH₃). Nitrite ion (NO₂⁻) is an important intermediate by-product during NO₃⁻RR, and its yield rate and selectivity under magnetic fields are screened along with the applied potentials (Fig. 3(b) and Fig. S12, ESI†). Under an applied magnetic field of 2500 Oe, the yield rate of NO₂⁻ increased significantly at nearly all potentials. Maximal enhancement was found at −0.7 V vs. RHE, with the yield rate of NO₂⁻ increased from 0.26 to 7.13 mg h⁻¹ cm⁻², which represents a remarkable enhancement by more than one order of magnitude. Similarly, the NH₃ and NO₂⁻ partial current density for CuFe₂O₄ increased significantly under a magnetic field of 2500 Oe (Fig. 3(c)).

NO₂⁻ is an intermediate byproduct that is further converted to NH₃. We performed NO₂⁻ reduction reaction (NO₂⁻RR) on CuFe₂O₄ with and without a magnetic field. As shown in Fig. S13 (ESI†), CuFe₂O₄ exhibited the highest performance for NO₂⁻RR at −1.1 V vs. RHE. After a 2500 Oe magnetic field was applied, the NH₃ yield rate increased from 3.35 mg h⁻¹ cm⁻² to 6.99 mg h⁻¹ cm⁻², and FE increased from 51.7% to 79.0% respectively, demonstrating a significant promotion due to the magnetic field.

Furthermore, we analyzed NH₃ produced from ¹⁵N isotope-labeled NO₃⁻ using ¹H nuclear magnetic resonance (NMR) analysis (Fig. 3(d)). Only two peaks of ¹⁵NH₄⁺ appeared in the ¹H NMR spectra, and their peak intensity approximately doubled under the magnetic field, confirming the promoted NH₃ production directly sourced from NO₃⁻. The durability of CuFe₂O₄ in NO₃⁻RR was evaluated through 14 consecutive CA measurements, with the magnetic field (2500 Oe) repeatedly applied or removed. The NH₃ yield rate and FE were analyzed 1 h after each session (Fig. 3(e)). Please note that after each session, the electrolyte was completely collected and refilled with a new electrolyte. We did not find notable activity differences due to the replacement of electrolytes (Fig. S14, ESI†). As shown in Fig. 3(e), the NH₃ yield rate and FE in each segment show negligible difference, indicating the high repeatability of NO₃⁻RR improvement by applying magnetic fields and its durable electrochemical performance. After the CA tests, the high-resolution transmission electron microscope (HRTEM, Fig. S15, ESI†), FESEM images (Fig. S16a and b, ESI†), and elemental mapping images (Fig. S16c–f, ESI†) showed that the structure and morphology of the CuFe₂O₄ catalyst remained nearly identical to that before the electrochemical test, indicating the material stability of CuFe₂O₄ during NO₃⁻RR.

Catalyst chemistry after NO₃⁻RR

The catalyst chemistry during NO₃⁻RR is studied by post-electrochemistry analyses and operando characterizations. The powder XRD pattern of CuFe₂O₄ showed negligible change after 1 h of CA tests in 0.5 M Na₂SO₄ + 0.1 M NaNO₃ (Fig. S17, ESI†). The operando Raman tests were performed under different applied potentials to track the structural change at a smaller scale. As shown in Fig. S18 (ESI†), CuFe₂O₄ displays several Raman active modes of E_g (446 cm⁻¹), B_1g (491 cm⁻¹ and 571⁻¹), and A_1g (706 cm⁻¹). A_1g is identified as symmetric stretching of metal–oxygen bonds in tetrahedral sites. The other E_g and B_g modes are assigned to metal ions in octahedral sites.^41,42 In addition, the vibrational bands of 980 cm⁻¹ and 1047 cm⁻¹ can be assigned to adsorbed SO₄²⁻ and NO₃⁻, respectively. As the applied potential shifts from −0.1 to −1.1 V vs. RHE, almost no change in the vibrational modes was observed. HRTEM before and after NO₃⁻RR proved that the CuFe₂O₄ crystal structure was well maintained (Fig. S19, ESI†).

As shown in Fig. S20 and S21 (ESI†), X-ray photoelectron spectrum (XPS), and Auger electron spectroscopy (AES) tests were performed on CuFe₂O₄ after electrochemical measurements. In Cu 2p XPS spectra, the peak located at 934.4 eV in CuFe₂O₄ is assigned to Cu²⁺, while an additional small peak at 932.5 eV is assigned to Cu⁰ after nitrate electrochemical reduction.^43,44 The Auger peaks of Cu⁰ and Cu²⁺ were observed at 568 and 568.9 eV,⁴⁵ respectively, and a small peak for Cu⁰ emerges after NO₃⁻RR. The peaks at 711.0 and 724.4 eV in the Fe 2p XPS spectra can be attributed to Fe³⁺ in the octahedral sites, and the peaks at 713.6 and 726.9 eV can be attributed to Fe³⁺ in the tetrahedral sites. The O 1s spectra have three main peaks at 530.5, 532.1, and 535.3 eV, which can be attributed to bulk lattice oxygen (O_lattice), surface oxygen (O_surface), and adsorbed oxygen (O_absorb). These peaks remained unchanged in location before and after the NO₃⁻RR.

The valence state and the density of partially/empty filled electronic states of CuFe₂O₄ were further revealed by soft X-ray absorption spectroscopy (SXAS) at the Cu 2p, Fe 2p, and O 1s edges (Fig. S22, ESI†) before and after NO₃⁻RR. The dipole transitions of Cu 2p3/2 and 2p1/2 electrons into empty d-states are illustrated in the Cu L-edge XAS spectrum.^46,47 In Cu L-edge XAS spectra, the peaks at 930.6 and 950.6 eV in CuFe₂O₄ are assigned to Cu²⁺, corresponding to the transitions of Cu 2p3/2 and Cu 2p1/2 (spin–orbital coupled states) to the 3d⁹ unoccupied orbital, respectively. A small peak at 939.9 eV appears after NO₃⁻RR, which can be attributed to Cu⁰. This finding is consistent with the XPS and AES results, confirming that a small amount of Cu²⁺ was reduced to Cu⁰. In the Fe L-edge XAS spectra, four intense peaks at 708/709.4 eV and 721.2/722.8 eV are assigned to the Fe 2p3/2 and Fe 2p1/2, respectively.⁴⁶ Compared to the Fe L-edge XAS spectra of CuFe₂O₄ before the NO₃⁻RR, the peaks remained unchanged after the NO₃⁻RR. Furthermore, the O K-edge XAS spectra of CuFe₂O₄ had five peaks, which were resolved at 530, 531, 537.4, 540.6, and 546.5 eV, respectively. The peaks at 530 and 531 eV are assigned to O 1s and O p orbitals hybridized with metal 3d orbitals, and the other three peaks are assigned to O 1s and O p orbitals hybridized with metal 4s and 4p orbitals.⁴⁶ Slight changes in peak intensity after NO₃⁻RR suggest a small amount of Cu²⁺ was reduced, altering the metal-oxygen hybridization. Overall, CuFe₂O₄ is the main catalytic motif for the spin-sensitive NO₃⁻RR process. Although minor Cu⁰ segregation was identified on the CuFe₂O₄ surface, the Cu foil experiment excluded the contribution of Cu⁰ to the enhanced NO₃⁻RR under magnetic fields. Our subsequent theoretical study was conducted using CuFe₂O₄ as the simulation model.

Insights into spin-polarized NO₃⁻RR

To investigate the reaction mechanism of NO₃⁻RR on CuFe₂O₄, in situ ATR-FTIR tests were performed to detect the key absorbed intermediates and products during NO₃⁻RR (Fig. 4(a) and Fig. S23, ESI†). In the potential-dependent and time-dependent in situ ATR-FTIR spectra, a series of characteristic peaks associated with NO₃⁻RR intermediates gradually intensified as the applied potential became more negative and the reaction progressed. Notably, the bands at 1456 cm⁻¹ and 1228 cm⁻¹ correspond to the N–H bending vibration of *NH₃ species^48–50 and the N–O stretching vibration of *NO₂, respectively.^49,51 Their gradual intensification indicates an enhancement in the reaction, which is consistent with the results of the electrochemical testing. In addition, several weaker but distinguishable peaks emerged, reflecting the presence of multiple intermediates: the NO scaling vibration of *NHO (1397 cm⁻¹),⁴⁸ the antisymmetric N–H bending of *NH₂ (1538 cm⁻¹),^48,49 the N–O stretching vibration of *NO₃⁻ (1338 cm⁻¹),^48,49 the N–O bending vibration in *NO (1611 cm⁻¹),^49,50,52 and the N–O stretching of hydroxylamine *NH₂OH (1153 cm⁻¹).^50,52 The gradual enhancement of these characteristic peaks indicates an intensifying reaction that is consistent with electrochemical testing observations. Based on the in situ ATR-FTIR results, the NO₃⁻RR could proceed through the * + NO₃⁻ → *NO₃ → *NO₂ → *NO → *NHO → *NHOH or *NH₂O → *NH₂OH → *NH₂ → *NH₃ pathways. Either with *NHOH or *NH₂O intermediate will decide for alternative pathways, which were both studied by spin-polarized density functional theory to identify the favorable pathway. A low-index (110) surface was used to model the adsorption of the intermediates on the spinel CuFe₂O₄ surface, and a 1 × 2 supercell was constructed. As shown in Fig. S24 (ESI†), *NO₃ is preferably adsorbed on Fe atoms rather than Cu atoms; thus, Fe could serve as the main site for subsequent hydrogenation and deoxygenation processes. The free energy diagram of all reaction intermediates on CuFe₂O₄ is shown in Fig. S25 (ESI†), and the corresponding intermediate configurations are summarized in Fig. 4(b). As displayed, the NO₃⁻ adsorption step, * + NO₃⁻ → *NO₃ + e⁻, shows a downhill Gibbs free energy change of −0.653 eV. Subsequently, *NO₃ undergoes two deoxygenation steps to become *NO with a downhill free energy change of −2.651 eV and −1.975 eV, respectively. Therefore, the two steps of NO₃⁻ deoxygenated to *NO on CuFe₂O₄ are thermodynamically favorable. The potential limiting step (PLS) is found in the first hydrogenation step from *NO to *NHO with a limiting potential of 0.474 eV. The hydrogenation of *NHO could result in either *NH₂O or *NHOH, whereas the formation of *NHOH is energetically more favorable. In addition, the free energy diagram of NO₂⁻RR on CuFe₂O₄ is shown in Fig. S26 (ESI†); the only difference from NO₃⁻RR is the adsorption of NO₂⁻ in the first reaction step. The diagram indicates that the free energy changes from *NO₂ to *NH₃ in NO₂⁻RR are consistent with those observed in NO₃⁻RR, with the PLS also being the conversion of *NO to *NHO.

To explain the improvement by magnetization, we employ DFT studies to reveal the spin-related interactions among reaction intermediates and identify the spin effects on the PLS. The d-band density of states (DOSs) for the Fe sites of CuFe₂O₄ were calculated, and the results are shown in Fig. 4(c). The Fe 3d spin-up and spin-down states around the Fermi level (E_F) are asymmetric about the x-axis, indicating that the unpaired electrons belong to the 3d orbitals. Fe was calculated using a spin population of 3.526μ_B. Similarly, according to the DOSs of all NO₃⁻RR intermediates (Fig. S27, ESI†), significant spin density is identified for the atoms (N or O) directly associated with Fe. The corresponding electronic interactions within these reaction intermediates are crucial for leading spin-polarized pathways.

Subsequently, we studied the DOS (Fig. 4(c) and Fig. S28, ESI†) of the Fe of CuFe₂O₄ before and after the adsorption of the intermediates, revealing changes in the electronic structure of the orbitals. After the adsorption of *NO₃, new occupied electronic states appear in the Fe d_xy orbital near the Fermi level, whereas the occupied states of the Fe d_x²−y² and d_z² orbitals disappear. Notably, the spin direction of the filled electrons in the d_xy orbital after adsorption is consistent with that in the Fe d_x²−y² and d_z² orbitals before adsorption. Such association by spin direction is schematically illustrated in Fig. 4(d) to indicate the role of ferromagnetism in reaction kinetics. When a ferromagnetic substrate is magnetized, spin polarization in the materials facilitates spin filtering. During electrochemical reduction, the electrons passing through the electrode are aligned by the magnetized substrate to the same spin direction. The spin electron transfer, mediated by Fe d_x²−y² and d_z² orbitals, induces significant spin-polarized electrons in the Fe d_xy orbital, which results in the π interactions between Fe d_xy and N or O p orbitals of adsorbed nitrogenous species to be spin-correlated. In the absence of spin filtering by a magnetized substrate or using inherently non-ferromagnetic catalysts, significantly disordered spins can dominate Fe d_x²−y² and d_z², leading to electron pairing in the d_xy orbital without spin interactions with the adsorbed species. Therefore, in terms of orbital-level spin interactions, the ferromagnetic spin ordering of electrocatalysts could be a prerequisite for the spin polarization of intermediates during NO₃⁻RR.

The correlation between activity enhancement and the degree of spin polarization in electrocatalysts is important for understanding the performance improvement mechanism. In a past study, a linear correlation for OER enhancement as a function of spin polarization degree was indicated by the portion of disappeared domain walls after NiFe thin films were magnetized.²⁹ In this study, we achieved varying degrees of spin polarization in CuFe₂O₄ by gradually increasing the magnetic field from zero to magnetic saturation (2500 Oe according to the hysteresis loop of CuFe₂O₄ in Fig. S29, ESI†). The inset in Fig. 5(a) shows the increment in current density corresponding to the magnetic field strength, ranging from 0 to 2500 Oe. A product analysis was conducted after a series of CA measurements (−1.1 V vs. RHE, 0.5 h) under elevating magnetic fields. Tables S2 and S3 (ESI†) summarize the yield rate and FE for both NH₃ and NO₂⁻ under varying magnetic fields, respectively. Fig. 5(a) plots the yield rate increment for NH₃ and NO₂⁻ as a function of the magnetic field strength. For NH₃, the yield rate increment exhibits a nearly linear correlation with the magnetic field strength up to 1000 Oe. However, once magnetic field strength exceeded 1000 Oe, the effect on NH₃ yield rate increment plateaus, disrupting the linear relationship with spin polarization. In contrast, the yield rate increment for NO₂⁻ maintains a linear correlation with the magnetic field strength throughout the field range. The dependence of the reaction kinetics on spin polarization shows a striking difference between the production of NO₂⁻ and NH₃, implying that spin has distinct roles in their respective reaction pathways.

The magnetic moments of Fe and adsorbed intermediates were calculated along with the NO₃⁻RR pathway and are summarized in Fig. 5(b). These calculations reveal dynamic variations in the magnetic moments during deoxygenation and hydrogenation processes. Upon the formation of *NO₃, both the Fe atoms and the adsorbed intermediates exhibit increasing magnetic moments. Significant magnetic moments were identified for the *NO and *NHO intermediates. The transition to these intermediates requires substantial spin polarization, making the relevant electrochemical step highly sensitive to the magnetization of the electrocatalyst. As electrons pass through the magnetized catalyst, they are more likely to be filtered into the same spin direction, enhancing the spin polarization in the reaction.

The step of *NO → *NHO is spin-sensitive and coincides with PLS. Tuning it by improving the spin-polarization of the catalyst has a high likelihood of promoting the reaction kinetics. Upon the formation of *NHO, Fe reaches its maximal magnetic moment (3.5μ_B) in the whole pathway. The electron transfer in the PLS was screened through Bader charge analyses of Fe, substrate, and adsorbed intermediates along with the reaction pathway, as shown in Fig. 5(c). The formation step of *NHO involves the largest charge transfer (0.606e) among all steps, with Fe atoms experiencing the most significant charge increment during this transition. Notably, the charging of Fe sites is accompanied by the discharging of the substrate. This interaction not only highlights a spin-polarized electron transfer mediated by Fe and indicates a plausible source of spin electrons from the ferromagnetic substrate as a spin filter.

The major spin polarization process in the PLS of *NO → *NHO, mediated by Fe in the ferromagnetic catalyst, should maximize the spin-related improvement in NH₃ production. However, when the step of *NO → *NHO is significantly optimized by magnetization, it may no longer be the limiting step for NH₃ production. The steps of *NO → *NHO and *NHOH → *NH₂OH show a close energetic barrier of 0.474 and 0.423 eV, respectively. However, according to the analysis of magnetic moments (Fig. 5(b)), the step of *NHOH → *NH₂OH is much less spin polarized compared to the step of *NO → *NHO. Upon increasing the strength of the magnetic field, more optimization by magnetization is expected for the step of *NO → *NHO compared to *NHOH → *NH₂OH (Fig. 5(d)). When the improvement reaches a critical point, the limiting step may be shifted to the *NHOH → *NH₂OH. Afterward, the NH₃ production rate could be less sensitive to the increasing spin polarization, where a critical turning point makes the non-linear correlation between NH₃ yield rate increment and magnetic field strength. In contrast, NO₂⁻ production is determined only by the step of *NO₃ → *NO₂, which maintains a linear correlation with the following.

Conclusion

We demonstrated the critical role of spin polarization in altering the limiting step and optimizing the efficiency of NO₃⁻RR under magnetic fields. The Cu substitution in Fe₃O₄ contributes to a higher intrinsic NO₃⁻RR activity and elevated magnetic coercivity. The highest NO₃⁻RR activity has been achieved on magnetic CuFe₂O₄ after magnetization. Theoretical studies have revealed a spin correlation specific to Fe 3d orbitals near the Fermi level before and after the NO₃⁻ adsorption on the Fe site of CuFe₂O₄, which facilitates the spin-related π interactions with reaction intermediates and leads to significant magnetic moments. The highest magnetic moments of the intermediates were identified for the step of *NO → *NHO, which is the primary limiting step. However, as it is optimized by magnetization, the limiting step is shifted to the step of *NHOH → *NH₂OH, which is less spin polarized, for which the reaction kinetics is less responsive to the magnetic field. Therefore, a non-linear correlation to spin polarization of CuFe₂O₄ has been found for NH₃ production. In contrast, NO₃⁻RR to produce NO₂⁻ shows linear correlation to spin polarization due to unchangeable PLS.

The linear and non-linear correlations highlight the importance of specifying the spin effects to intermediates with varied magnetic nature. From a practical perspective, linear correlations offer predictable scalability, making it easier to improve catalytic activity with increasing magnetic field strength. However, non-linear correlation indicates that simply increasing spin polarization may not be applicable for optimizing every elemental step, which makes spin engineering for catalysts complicated but scientifically interesting. This study revealed the complicated enhancement of NO₃⁻RR activity under magnetic field conditions. It further provided new fundamental insights into the role of spin polarization in each elemental step of NO₃⁻RR. This work inspires interest in investigating more reactions that contain complicated intermediates and steps that can be tuned through a spin effect.

Methods

Reagents and materials used

All chemicals used in this study were of analytical grade and were sourced from commercial suppliers. No further purification was conducted before their use.

Preparation of CuFe₂O₄

CuFe₂O₄ spinel oxides were prepared using a conventional sol–gel method, as described elsewhere.⁴⁰ Iron nitrate nonahydrate (Fe(NO₃)₃·9H₂O) and Copper acetate monohydrate (Cu(CH₃COO)₂·H₂O) were mixed in a 2:1 molar ratio and then dissolved in 60 mL of deionized water. Citric acid, ethylene diamine tetraacetic acid and ammonium hydroxide were added to the solution, which was stirred at 90 °C for about 2 h until a homogeneous gel was formed. The gel was dried in air at 200 °C for 12 h to obtain black oxide powders. These powders were ground and calcined at 800 °C in air for 6 h to obtain CuFe₂O₄ spinel oxides. This synthesis procedure was repeated multiple times and showed good reproducibility.

Material characterizations

Powder X-ray diffraction (XRD) data were recorded using a Bruker D8 diffractometer under Cu-Kα radiation source (λ = 1.5418 Å) at a scanning rate of 2° min⁻¹ at ambient temperature. The scanning electron microscopy images and energy-dispersive X-ray spectroscopy (EDS) elemental mapping were performed using a scanning electron microscope (SEM, JEOL JSM7600F). High-resolution transmission electron microscopy (HRTEM) images were obtained on a JEOL JEM-2100 plus microscope at an accelerating voltage of 200 KV. X-ray photoelectron spectroscopy (XPS) equipped with a standard Al Kα X-ray source (PHI-5400 machine) operated at 250 W was employed for surface analysis. Soft X-ray absorption spectroscopy (XAS) at the Cu L-edge, Fe L-edge, and O K-edge was conducted at the Australian Synchrotron, part of ANSTO. The field dependence of magnetization data at room temperature was collected from −14 to +14 kOe using vibrating sample magnetometry (VSM, Lake Shore 7400). Ultraviolet-visible (UV-Vis) absorbance spectra were recorded on a Shimadzu UV-2700 spectrophotometer. In situ ATR-FTIR was conducted using a Bruker Vertex 80 instrument with an electrochemical VeeMax III apparatus from PIKE. The operando Raman tests were performed using an i-Raman® Plus 785 H Raman spectrometer with a laser wavelength of 785 nm. Nuclear magnetic resonance (NMR) spectra were performed on a Bruker AV 400-MHz NMR spectrometer.

Electrocatalytic NO₃⁻RR measurements

The electrochemical experiments were performed in an H-type cell using a three-electrode setup, consisting of a carbon paper (10 × 10 mm) working electrode (WE) with an effective electrode area of 1.0 cm² (one-side loading), a platinum foil (15 × 15 mm) counter electrode (CE), and a saturated calomel reference electrode (RE). A BioLogic SP 150 potentiostat workstation was used to record the electrochemical response. The WE was prepared by the ink-cast method, using a mixture of catalysts and acetylene black carbon in ethyl alcohol with a mass ratio of 5:1, then Nafion was added as a binder. The ink was ultrasonicated in an ice bath for 60 min to ensure homogeneity. Finally, a certain volume of ink (51 μL) was dropped onto carbon paper with a loading mass of 255 μg cm⁻². All potentials were measured against the SCE without iR correction versus the RHE reference scale by E (vs. RHE) = E (vs. SCE) + 0.0591 × pH + 0.241. For the electrochemical NO₃⁻ reduction reaction, a 25 mL solution containing 0.5 M Na₂SO₄ and 0.1 M NaNO₃ was used as the electrolyte.² The electrolyte was distributed to the anode and cathode compartments of the H-cell, separated by a Nafion 117 membrane, with each compartment containing 25 mL of electrolyte. Before the measurement, the electrolyte was purged with argon for 20–25 minutes. The linear sweep voltammetry (LSV) was performed in high-purity argon-saturated 0.5 M Na₂SO₄ with or without 0.1 M NaNO₃ at a scan rate of 10 mV s⁻¹. Cyclic voltammetry (CV) tests for the double-layer capacitance (C_dl) were conducted in 0.5 M Na₂SO₄ solution in the non-faradaic potential range at scan rates of 10, 20, 40, 60, 80, and 100 mV s⁻¹. The ECSA of the working electrodes was calculated by ECSA = C_dl/C_s. Chronoamperometry (CA) tests were carried out at different applied potentials, with high-purity argon continuously supplied to the cathode compartment throughout the experiments. The current density presented in this work is normalized to the geometric surface area.

Determination of ammonia content

The generated ammonia concentration was quantified using UV-Vis spectrophotometry and a modified indophenol blue method. The measured electrolyte volume was diluted to 1 mL to ensure accurate measurement within the detectable range. Subsequently, 1 mL of a 1 M NaOH solution containing 5 wt% sodium citrate and 5 wt% salicylic acid was added to the diluted sample. Next, 0.5 mL of 0.05 M NaClO solution and 0.1 mL of 1.0 wt% sodium nitroferricyanide solution. Next, the resulting indophenol blue solution's absorption spectrum was then recorded using a UV-Vis spectrophotometer (UV-2700) at 655 nm. Finally, the ammonia concentration was calculated based on the absorbance data from the standard ammonium chloride calibration curve.

Calculation of ammonia yield rate and FE

NH3 yield rate (μg h−1 cm−2) = (cNH3 × V)/(t × S)

FENH3 = (n × F × cNH3 × V)/(M × Q) × 100%

where c_NH3 (μg mL⁻¹) is the measured ammonia concentration, V (mL) is the electrolyte volume in the cathodic compartment, t (h) is the electrochemical reduction time, S is the geometric surface area of WE, n is the number of electrons transferred (n = 8), F is the Faraday constant (96485 C mol⁻¹), M is the relative molecular mass of NH₃ and Q (C) is the total charge of applied electricity.

Computational methods

Spin-polarized density functional theory (DFT) calculations were carried out using the Vienna ab initio simulation package (VASP).⁵³ The projector-augmented plane-wave method was applied to describe the ion-electron interaction⁵⁴ and the generalized gradient approximation within the Perdew–Burke–Ernzerhof functional was used to describe the exchange-correlation interaction.⁵⁵ A low-index (110) surface was used to model the adsorption of the intermediates on the spinel CuFe₂O₄ surface, and a 1 × 2 supercell was constructed. The cut-off energy was set as 500 eV, and the systems were optimized until the energy and force were less than 10⁻⁵eV and 0.01eV Å⁻¹. GGA+U calculations were performed using the model proposed by Dudarev et al., with U_eff (=Coulomb U − exchange J) values of 4 eV and 5 eV for Fe and Cu, respectively. The Monkhorst–Pack k-point grids were set as 3 × 2 × 1. During the structural optimization, the top two layers and the intermediates were allowed to fully relax while the bottom layers were kept fixed. A vacuum layer larger than 20 Å in the z-direction is employed to prevent the interaction between slabs. Grimme's D3 method is used to consider the van der Waals (vdW) interactions.⁵⁶

The overall reaction pathway can be described by the following sequence of elementary steps:

* + NO3− → *NO3 + e−

*NO3 + 2H+ + 2e− → *NO2 + H2O

*NO2 + 2H+ + 2e− → *NO + H2O

*NO + H+ + e− → *NHO

*NHO + H+ + e− → *NHOH

*NHOH + H+ + e− → *NH2OH

*NH2OH + H+ + e− → *NH2 + H2O

*NH2 + H+ + e− → *NH3

where * denotes the catalytically active site.

The corresponding Gibbs free energy change for each step is calculated using the following equation:

ΔG = ΔE + ΔEZPE − TΔS

where ΔE is the total energy difference before and after the intermediate adsorption, ΔE_ZPE is the difference in zero-point energy, and ΔS is the entropy change. The zero-point energies and entropies of both free molecules and adsorbed species were obtained from vibrational frequency calculations. To describe the charged NO₃⁻ species as a reference, neutral HNO₃ in the gas phase was used as a reference, and the energy of NO₃⁻ was derived via a thermodynamic cycle, avoiding challenges in modeling charged systems using periodic DFT.⁵⁷

Author contributions

D. S., T. W., and Z. X. conceived the idea and designed the experiment. D. S. was responsible for catalyst design. D. S. and X. C. performed electrochemical measurements and all authors contributed to the data analysis. Q. W. conducted the DFT studies. C. D. conducted the in situ ATR-FTIR and operando Raman tests. S. Z. conducted the magnetic hysteresis loop. F. M. conducted the isotope-labeling test and analysis. Y. Z. and P. S. performed the HRTEM characterizations. X. L., S. Z., and T. W. performed the soft XAS measurement and analyses. Y. Z., D. S., and T. W. prepared the figures. D. S., T. W., Y. Z., and Z. X. prepared the manuscript with contributions from all authors.

Conflicts of interest

There are no conflicts to declare.

Data availability

All data supporting the findings of this study are available within the article and its ESI† file or from the authors upon reasonable request.

Acknowledgements

This work was financially supported by the Singapore Ministry of Education Tier 2 Grant (MOE-T2EP10220-0001). C. Dai acknowledges the support by the National Research Foundation, Prime Minister's Office, Singapore, under its Campus for Research Excellence and Technological Enterprise (CREATE) program. The authors thank for the XAS measurements from the Soft X-ray Spectroscopy beamline, the Australian Synchrotron (AS, Australia), part of ANSTO.

References

1. J. Huang, B. Hu, J. Meng, T. Meng, W. Liu, Y. Guan, L. Jin and X. Zhang, Energy Environ. Sci., 2024, 17, 1007–1045 RSC.
2. N. T. Suen, S. F. Hung, Q. Quan, N. Zhang, Y. J. Xu and H. M. Chen, Chem. Soc. Rev., 2017, 46, 337–365 RSC.
3. S. Jin, J. M. Kevin, A. G. Hubert, B. G. John and S.-H. Yang, Science, 2011, 334, 1383–1385 CrossRef PubMed.
4. M. G. Kibria, J. P. Edwards, C. M. Gabardo, C. T. Dinh, A. Seifitokaldani, D. Sinton and E. H. Sargent, Adv. Mater., 2019, 31, e1807166 CrossRef PubMed.
5. G. Zhang, B. Li, Y. Shi, Q. Zhou, W.-J. Fu, G. Zhou, J. Ma, S. Yin, W. Yuan, S. Miao, Q. Ji, J. Qu and H. Liu, Nat. Sustain., 2024, 7, 1251–1263 CrossRef.
6. F.-Y. Chen, A. Elgazzar, S. Pecaut, C. Qiu, Y. Feng, S. Ashokkumar, Z. Yu, C. Sellers, S. Hao, P. Zhu and H. Wang, Nat. Catal., 2024, 7, 1032–1043 CrossRef CAS.
7. J. K. Norskov, F. Abild-Pedersen, F. Studt and T. Bligaard, Proc. Natl. Acad. Sci. U. S. A., 2011, 108, 937–943 CrossRef CAS PubMed.
8. F. Calle-Vallejo, N. G. Inoglu, H.-Y. Su, J. I. Martínez, I. C. Man, M. T. M. Koper, J. R. Kitchin and J. Rossmeisl, Chem. Sci., 2013, 4, 1245–1249 RSC.
9. A. Grimaud, K. J. May, C. E. Carlton, Y. L. Lee, M. Risch, W. T. Hong, J. Zhou and Y. Shao-Horn, Nat. Commun., 2013, 4, 2439 CrossRef PubMed.
10. J. Suntivich, W. T. Hong, Y.-L. Lee, J. M. Rondinelli, W. Yang, J. B. Goodenough, B. Dabrowski, J. W. Freeland and Y. Shao-Horn, J. Phys. Chem. C, 2014, 118, 1856–1863 CrossRef CAS.
11. W. T. Hong, R. E. Welsch and Y. Shao-Horn, J. Phys. Chem. C, 2015, 120, 78–86 CrossRef.
12. M. T. M. Koper, J. Electroanal. Chem., 2011, 660, 254–260 CrossRef CAS.
13. M. T. M. Koper, Chem. Sci., 2013, 4, 2710–2723 RSC.
14. J. Gracia, Phys. Chem. Chem. Phys., 2017, 19, 20451–20456 RSC.
15. J. Gracia, R. Sharpe and J. Munarriz, J. Catal., 2018, 361, 331–338 CrossRef CAS.
16. J. Munarriz, V. Polo and J. Gracia, Chemphyschem, 2018, 19, 2843–2847 CrossRef CAS PubMed.
17. X. Ren, T. Wu, Y. Sun, Y. Li, G. Xian, X. Liu, C. Shen, J. Gracia, H. J. Gao, H. Yang and Z. J. Xu, Nat. Commun., 2021, 12, 2608 CrossRef CAS PubMed.
18. T. Wu, X. Ren, Y. Sun, S. Sun, G. Xian, G. G. Scherer, A. C. Fisher, D. Mandler, J. W. Ager, A. Grimaud, J. Wang, C. Shen, H. Yang, J. Gracia, H. J. Gao and Z. J. Xu, Nat. Commun., 2021, 12, 3634 CrossRef CAS PubMed.
19. X. Ren, T. Wu, Z. Gong, L. Pan, J. Meng, H. Yang, F. B. Dagbjartsdottir, A. Fisher, H. J. Gao and Z. J. Xu, Nat. Commun., 2023, 14, 2482 CrossRef CAS PubMed.
20. J. Gracia, Phys. Chem. Chem. Phys., 2017, 19, 20451–20456 RSC.
21. J. Gracia, J. Phys. Chem. C, 2019, 123, 9967–9972 CrossRef CAS.
22. W. Mtangi, V. Kiran, C. Fontanesi and R. Naaman, J. Phys. Chem. Lett., 2015, 6, 4916–4922 CrossRef CAS PubMed.
23. Z. Chen, X. Li, H. Ma, Y. Zhang, J. Peng, T. Ma, Z. Cheng, J. Gracia, Y. Sun and Z. J. Xu, Natl. Sci. Rev., 2024, 11, nwae314 CrossRef PubMed.
24. W. Zhang, J. Ai, T. Ouyang, L. Yu, A. Liu, L. Han, Y. Duan, C. Tian, C. Chu, Y. Ma, S. Che and Y. Fang, J. Am. Chem. Soc., 2024, 146, 28214–28221 CAS.
25. 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 CrossRef PubMed.
26. K. Zhang, A. Cao, L. H. Wandall, J. Vernieres, J. Kibsgaard, J. K. Nørskov and I. Chorkendorff, Science, 2024, 383, 1357–1363 CrossRef CAS PubMed.
27. Y. Wang, A. Xu, Z. Wang, L. Huang, J. Li, F. Li, J. Wicks, M. Luo, D. H. Nam, C. S. Tan, Y. Ding, J. Wu, Y. Lum, C. T. Dinh, D. Sinton, G. Zheng and E. H. Sargent, J. Am. Chem. Soc., 2020, 142, 5702–5708 CrossRef CAS PubMed.
28. Q. Hu, Q. Huo, S. Qi, X. Deng, J. Zhuang, J. Yu, X. Li, W. Zhou, M. Lv, X. Chen, X. Wang, C. Feng, H. Yang and C. He, Adv. Mater., 2023, 36, 2311375 CrossRef PubMed.
29. K. Liu, H. Li, M. Xie, P. Wang, Z. Jin, Y. Liu, M. Zhou, P. Li and G. Yu, J. Am. Chem. Soc., 2024, 146, 7779–7790 CrossRef CAS PubMed.
30. S. Han, H. Li, T. Li, F. Chen, R. Yang, Y. Yu and B. Zhang, Nat. Catal., 2023, 6, 402–414 CrossRef CAS.
31. Y. Wang, C. Wang, M. Li, Y. Yu and B. Zhang, Chem. Soc. Rev., 2021, 50, 6720–6733 RSC.
32. J. Crawford, H. Yin, A. Du and A. P. O'Mullane, Angew. Chem., Int. Ed., 2022, 61, e202201604 CrossRef CAS PubMed.
33. J. Cai, Y. Wei, A. Cao, J. Huang, Z. Jiang, S. Lu and S.-Q. Zang, Appl. Catal., B, 2022, 316, 121683 CrossRef CAS.
34. Q. Gao, H. S. Pillai, Y. Huang, S. Liu, Q. Mu, X. Han, Z. Yan, H. Zhou, Q. He, H. Xin and H. Zhu, Nat. Commun., 2022, 13, 2338 CrossRef CAS PubMed.
35. 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 CrossRef CAS PubMed.
36. J. Wang, K. Zhao, Y. Yao, F. Xue, F. Lu, W. Yan, F. Yuan and X. Wang, Nat. Commun., 2025, 16, 1129 CrossRef CAS PubMed.
37. X. He, F. Yang, Q. Hu, J. Yin and J. Zhang, Crystals, 2023, 13, 1640 CrossRef CAS.
38. W. Chen, X. Yang, Z. Chen, Z. Ou, J. Hu, Y. Xu, Y. Li, X. Ren, S. Ye, J. Qiu, J. Liu and Q. Zhang, Adv. Funct. Mater., 2023, 33, 2300512 CrossRef CAS.
39. C. Liu, X.-C. Zhou, G. Li, J. Su, L. Tang, Q. Liu, X. Han, S. Lv, Z. Mu, Y. Sun, S. Yuan, F. Gao, J.-L. Zuo, S. Li and M. Ding, Sci. Adv., 2025, 11, eadq3554 CrossRef CAS PubMed.
40. T. Wu, S. Sun, J. Song, S. Xi, Y. Du, B. Chen, W. A. Sasangka, H. Liao, C. L. Gan, G. G. Scherer, L. Zeng, H. Wang, H. Li, A. Grimaud and Z. J. Xu, Nat. Catal., 2019, 2, 763–772 CrossRef CAS.
41. K. Verma, A. Kumar and D. Varshney, Curr. Appl. Phys., 2013, 13, 467–473 CrossRef.
42. R. Zhang, M. Liu, L. Lu, S.-B. Mi, C.-L. Jia and H. Wang, RSC Adv., 2016, 6, 100108 RSC.
43. V. N. Popok, S. M. Novikov, Y. Y. Lebedinskij, A. M. Markeev, A. A. Andreev, I. N. Trunkin, A. V. Arsenin and V. S. Volkov, Plasmonics, 2020, 16, 333–340 CrossRef.
44. M. Swadzba-Kwasny, L. Chancelier, S. Ng, H. G. Manyar, C. Hardacre and P. Nockemann, Dalton Trans., 2012, 41, 219–227 RSC.
45. Y. Wang, W. Zhou, R. Jia, Y. Yu and B. Zhang, Angew. Chem., Int. Ed., 2020, 59, 5350–5354 CrossRef CAS PubMed.
46. M. Nasir, N. Patra, D. K. Shukla, D. Bhattacharya, S. Kumar, D. M. Phase, S. N. Jha, S. Biring, P. M. Shirage and S. Sen, RSC Adv., 2016, 6, 103571 RSC.
47. J. Yang, T. Regier, J. J. Dynes, J. Wang, J. Shi, D. Peak, Y. Zhao, T. Hu, Y. Chen and J. S. Tse, Anal. Chem., 2011, 83, 7856–7862 CrossRef CAS PubMed.
48. Y. Li, C. Wang, L. Yang, W. Ge, J. Shen, Y. Zhu and C. Li, Adv. Energy Mater., 2023, 14, 2303863 CrossRef.
49. M. A. Larrubia, G. Ramis and G. Busca, Appl. Catal., B, 2001, 30, 101–110 CrossRef CAS.
50. V. Rosca, G. L. Beltramo and M. T. M. Koper, J. Electroanal. Chem., 2004, 566, 53–62 CrossRef CAS.
51. Z. Huang, B. Yang, Y. Zhou, W. Luo, G. Chen, M. Liu, X. Liu, R. Ma and N. Zhang, ACS Nano, 2023, 17, 25091–25100 CrossRef CAS PubMed.
52. Q. Hu, K. Yang, O. Peng, M. Li, L. Ma, S. Huang, Y. Du, Z. X. Xu, Q. Wang, Z. Chen, M. Yang and K. P. Loh, J. Am. Chem. Soc., 2024, 146, 668–676 CrossRef CAS PubMed.
53. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186 CrossRef CAS PubMed.
54. P. E. Blochl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979 CrossRef PubMed.
55. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775 CrossRef CAS.
56. S. Grimme, J. Comput. Chem., 2006, 27, 1787–1799 CrossRef CAS PubMed.
57. S. Guo, K. Heck, S. Kasiraju, H. Qian, Z. Zhao, L. C. Grabow, J. T. Miller and M. S. Wong, ACS Catal., 2017, 8, 503–515 CrossRef.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ee02132d

‡ These authors contributed equally to this work.

This journal is © The Royal Society of Chemistry 2025