Inorganic–organic hybrid cobalt spinel oxides for catalyzing the oxygen evolution reaction

Abstract

Fully triggering the deep-seated potential of traditional nanomaterials, such as the classic spinel family, is of paramount importance in the field of materials science, which is yet believed to heavily depend on advanced conceptual designs and synthetic strategies. Herein, a type of inorganic–organic hybrid spinel oxide is designed using a π-conjugated azobenzene single-tooth coordination method to overcome their stubborn problems of moderate activity and phase instability in electrocatalytic reactions. Taking spinel Co₃O₄ nanocubes as a pre-catalyst, after subtle etching of the cube surfaces, some oxygen atoms in the tetrahedral Co–O coordination field are replaced and selectively linked to weakly polar azo-extended π-conjugated units (π*–N=N–π*) via electrophilic carboxyl groups. The π-conjugation structure in Co₃O₄ suppresses the covalency competition between the tetrahedral and octahedral Co–O coordination fields, successfully preventing the phase transition during the electrocatalytic process and improving the electrocatalytic activity and durability. This study not only expands the spinel family but also provides useful guidelines for developing advanced functional materials.

Unlocking the potential of traditional nanomaterials like spinel oxides is crucial for improving the catalytic oxygen evolution reaction. Here, the authors report an inorganic–organic hybrid spinel oxide that enhances both catalytic activity and structural stability through a coordination method.

Introduction

Spinel oxide (AB₂O₄) is an important category of inorganic materials^1–3. AB₂O₄ contains two types of metal cations, i.e., tetrahedrally coordinated A²⁺ (M_Td) and octahedrally coordinated B³⁺ (M_Oh)^4–6. Owing to their different coordination structures, tetrahedral A²⁺ and octahedral B³⁺ have different electronic structures. Therefore, the properties of spinel oxides can be regulated by adjusting the ratio of tetrahedral to octahedral coordinations or the composition of metal elements. Owing to this characteristic, spinel oxides have important application value as effective catalysts in various catalytic reactions, including water electrolysis, anodic reactions in lithium-ion batteries, and carbon reactions in the chemical industry (C₁)^7–9. For example, in electrocatalytic oxygen evolution reactions (OERs), cobalt-based spinel oxides (Co₃O₄) exhibit notable electrocatalytic activity, making them among the most promising alternatives to noble metal-based catalysts¹.

However, the covalency competition between Co_Td and Co_Oh in cobalt spinel oxides can cause the material to become extremely unstable during catalytic reactions^10,11. Therefore, when catalyzing the alkaline OER, Co₃O₄ invariably transitions into hydroxyl oxides, losing the characteristics and advantages of spinel oxides^10,12. More seriously, neither metal cation doping nor elemental substitution can effectively suppress the covalency competition between the tetrahedral or octahedral Co–O fields formed by strongly polar oxygen coordinations, which may be why it was almost impossible to avoid the phase transition of spinel oxides during the OER in previous works^13–16. Compared to oxygen, sulfur, and halogen anions, the polarity of organic compounds is weak and easily adjustable^17–19. As a result, replacing the oxygen anion of Co₃O₄ with weakly polar or nonpolar organic groups can effectively tune the crystal field properties of metal cations. Encouragingly, the widespread development of strategies for organic ligand substitution in the synthesis of metal‒organic frameworks (MOFs) and perovskite materials has opened up possibilities for designing more efficient catalysts^20–22. However, a major challenge lies in the lack of stable linking units for the selected ligands. This often leads to weak binding between the ligand and the metal cation, which compromises the stability of catalytic materials under operational conditions. Drawing from these insights and challenges, it is clear that developing a compound that transcends traditional inorganic spinels is crucial for addressing the stability issue, yet this remains a challenge.

In this work, we propose a concept of inorganic–organic hybrid spinels. Using spinel oxide Co₃O₄ nanocubes as a pre-catalyst, a carboxyl single-tooth coordination method is used to replace some oxygen anions in the tetrahedron field of Co₃O₄ with organic groups, forming an inorganic–organic hybrid spinel called (R–COO)_xσ-Co₃O_4−x. At the inorganic end of the hybrid, the spatial symmetry is broken by replacing some strongly polar oxygen anions in the tetrahedron of Co₃O₄, which suppresses the covalency competition between the tetrahedral (Co_Td)–O₄ and octahedral (Co_oh)–O₆ fields to stabilize the spinel structure. At the organic end of the hybrid, azo-extended π-conjugated units (π*–N = N–π*) enhance the stability of the whole molecular system. The effective combination of the inorganic and organic ends results in a unique asymmetric structural unit, which not only regulates the electronic properties of cation sites in the metal coordination crystal fields but also endows the biphenyl group with good hydrophilicity. Consequently, the inorganic–organic hybrid spinel oxide exhibits comparable OER catalytic activity and structural stability in alkaline media. Our proposal not only expands the family of spinels but also provides a reference for developing advanced functional materials.

Results and discussion

Design and synthesis of the hybrid compound

The benzene ring with its unique π–π structure has stimulated many research studies on organic and supramolecular syntheses^23–25. According to the literature, the stability of a molecular system can be enhanced by expanding the π-conjugated structure and the azo group (–N=N–) exhibits reversible redox activity^26–28. Inspired by these facts, herein, nitrobenzoic acid was used as the initial reactant. After a series of reduction and coupling reactions, azobenzene-4,4ʹ-dicarboxylic acid was ultimately synthesized (named azobenzene groups). As shown in Fig. 1. I, nitrobenzoic acid first undergoes a reduction reaction with carbohydrates, where the nitro group (–NO₂) is reduced to the metastable nitroso group (–N=O) and the amino group (–NH₂). Upon continuous heating, –N=O and –NH₂ undergo dehydration to form a nitroso compound, which participates in the intermolecular coupling reaction to form the azobenzene backbone structure. This process involves intramolecular hydrogen bonding and π–π stacking interactions, which are responsible for the subsequent coupling of the organic and inorganic components. The formed azobenzene structure further undergoes substitution reactions in an acidic environment, affording the final product carboxylic acid-azobenzene. Particularly, the introduction of the carboxylic acid group enhances the electrophilicity of azobenzene. The SEM image of the resulting azobenzene molecules is shown in Supplementary Fig. 1. Spinel oxide Co₃O₄ nanocubes were synthesized using a high-temperature liquid-phase method, as shown in Fig. 1.II. To break the stable spatially symmetrical structure of inorganic spinels, some tetrahedral bonds (Co_Td–O) are cleaved to form a defective tetrahedron via the strong reduction solution method. Finally, as shown in Fig. 1.III, an inorganic–organic hybrid spinel (R–COO)_xσ-Co₃O_4−x is successfully synthesized using the carboxylate single-tooth coordination method; (R–COO)_xσ-Co₃O_4−x effectively coordinates carboxylate and azobenzene molecules with defective tetrahedrons.

Fig. 1. Synthetic routes for inorganic–organic hybrid spinels.

I Organic synthesis route, II Inorganic synthesis route, and III Inorganic–organic hybrid spinel synthesis route.

First, high-resolution transmission electron microscopy (HRTEM) was used to reveal the morphology of the samples. The TEM images in Supplementary Fig. 2 show that the synthesized spinel Co₃O₄ has a cubic shape with a size of ~10 nm. Moreover, the selected area electron diffraction (SAED) pattern in Supplementary Fig. 3 demonstrates the successful synthesis of the Co₃O₄ spinel structure. To create defective tetrahedrons in spinel Co₃O₄ (σ-Co₃O_4−x), spinel Co₃O₄ nanocubes are etched using NaBH₄, and the product maintains the cube shape even after etching, as shown in Fig. 2a. Interestingly, at the 10 nm scale (Fig. 2b), it is possible to clearly observe pores with a size of ~2 nm on the surface of the etched spinel Co₃O₄ nanocubes. After hydrothermal treatment (the detailed procedure is provided in the Experimental Section), tetrahedron-defective spinel Co₃O₄ with a carboxyl single-tooth coordination [named (R–COO)_xσ-Co₃O_4−x] is obtained. As shown in Fig. 2d and e, the cube size of (R–COO)_xσ-Co₃O_4−x remains ~10 nm and pores with a size of ~2 nm still exist on the surface of the cube, but the overall cubic structure remains intact. To further prove the connection between the azobenzene groups and spinels, energy-dispersive X-ray spectroscopy (EDS) line scanning and point scanning were used to characterize the designated pore locations (Supplementary Figs. 4 and 5). As shown in the HRTEM image in Fig. 2c, the Co and O contents in σ-Co₃O_4−x decrease inside the pores, proving that the cube surface is successfully etched. Additionally, the EDS scan-line spectrum of σ-Co₃O_4−x reveals almost negligible contents of C and N. In contrast, as shown in Fig. 2f, the contents of C and N inside the pores in (R–COO)_xσ-Co₃O_4−x notably increase, which may be attributed to the binding of azobenzene groups with oxygen in the pores. The elemental distribution of (R–COO)_xσ-Co₃O_4−x was obtained via elemental mapping analysis (Fig. 2g), which revealed that the overall distributions of Co and O were relatively uniform and decreased inside the pores. Moreover, when the elemental mapping images of C and N are superimposed (Supplementary Fig. 6), it clearly demonstrates the distribution of organic molecules inside the pores. This means that the organic groups are “locked” inside the pores of the σ-Co₃O_4−x nanocubes, allowing them to bind firmly with the spinels. These morphological characterization and elemental analyses confirm the successful synthesis of the inorganic–organic hybrid structure.

Fig. 2. Morphological structure characterizations.

a TEM images of σ-Co₃O_4−x, b HRTEM images of σ-Co₃O_4−x, c EDS line-scan profiles of σ-Co₃O_4−x, d TEM images of (R–COO)_xσ-Co₃O_4−x, e HRTEM images of (R–COO)_xσ-Co₃O_4−x, f EDS line-scan profiles of (R–COO)_xσ-Co₃O_4−x, and g Elemental mapping images of (R–COO)_xσ-Co₃O_4−x. Source data are provided as a Source Data file.

Structural identification

Advanced spectroscopy techniques can provide useful information about the atomic and molecular structures of chemical substances^29–31. To identify the molecular structure of the synthesized hybrid, Raman spectroscopy was used to reveal the cation distribution using the molecular vibration dynamics of the spinel phase. As shown in Fig. 3a, five characteristic peaks are observed in the Raman spectrum of pristine Co₃O₄, which are labeled A_1g, F_2g (1), F_2g (2), F_2g (3), and E_g. The peaks at 531 and 457 cm⁻¹ are observed in the Raman spectra of Co₃O₄, σ-Co₃O_4−x, and (R–COO)_xσ-Co₃O_4−x and can be attributed to the F_2g (CoO₄) and E_g modes, respectively. Among them, the A_1g band originates from the asymmetric stretching and bending vibrations of the M–O bond at the octahedral position, the F_2g (1) band is related to the M–O stretching bond at the tetrahedral position, the F_2g (2) band is caused by the asymmetric stretching of the M–O bond, and the E_g band is caused by the asymmetric bending of the M–O bond^32–34. The changes in the positions and intensities of the vibration bands reflect the subtle differences in the crystal structure of the hybrid. The F_2g (1) peak in the spinel phase of the azobenzene organic molecule moves toward a low wavenumber of 181 cm⁻¹ and becomes weaker and wider in the Raman spectrum of (R–COO)_xσ-Co₃O_4−x, indicating that the central metal of CoO₄ is occupied by organic molecules and its symmetry is broken^32,35,36. In contrast, the A_1g peak shifts to a higher wavenumber, which may be caused by the tension generated in the (R–COO)_xσ-Co₃O_4−x crystal. Additionally, the difference (Δλ) in the low-wavenumber region observed in the Raman spectrum of (R–COO)_xσ-Co₃O_4−x demonstrates the interaction between the d orbitals of the spinel and the π bonds of azobenzene^32,35,36. Furthermore, two distinct peaks are observed in the high-wavenumber region (800−1000 cm⁻¹) in the Raman spectrum of (R–COO)_xσ-Co₃O_4−x: the peak at 846 cm⁻¹ is attributed to the stretching vibration of the M–OR bond and the peak at 926 cm⁻¹ represents ring vibration, which indicates the coordination between the tetrahedral metal Co and oxygen in the carboxylate ester.

Fig. 3. Atomic and electronic structure characterizations.

a Raman spectra of (R–COO)_xσ-Co₃O_4−x, (R–COO)_xCo₃O₄, σ-Co₃O_4−x, and Co₃O₄. b FTIR spectra of (R–COO)_xσ-Co₃O_4−x, (R–COO)_xCo₃O₄, σ-Co₃O_4−x, and Co₃O₄. c Structural diagram of the inorganic–organic hybrid spinel. d ¹H NMR spectrum of azobenzene-4,4ʹ-dicarboxylic acid. e O K-edge normalized XANES spectra of azobenzene-4,4ʹ-dicarboxylic acid, (R–COO)_xσ-Co₃O_4−x, σ-Co₃O_4−x, and Co₃O₄. f Co L-edge normalized XANES spectra of (R–COO)_xσ-Co₃O_4−x, σ-Co₃O_4−x, and Co₃O₄. g C and N K-edge normalized XANES spectra of azobenzene-4,4ʹ-dicarboxylic acid and (R–COO)_xσ-Co₃O_4−x. Source data are provided as a Source Data file.

Fourier transform infrared (FTIR) spectroscopy was performed to obtain information on the molecular bonding of the spinel samples. As shown in Fig. 3b, in the FTIR spectrum of spinel Co₃O₄, the characteristic peaks at 580 and 667 cm⁻¹ are attributed to Co–O_Td and Co–O_Oh, respectively³⁷. Furthermore, in the FTIR spectrum of the azobenzene group, the characteristic peaks of νCOOH (1290 and 1425 cm⁻¹) are observed in the high-wavenumber region, which correspond to the two carboxyls in azobenzene group^38–41. In the FTIR spectrum of (R–COO)_xσ-Co₃O_4−x, the two characteristic peaks of νCOOH merge into one peak at 1378 cm⁻¹, which may be due to the metal–ester bond (C = O–O–Co) formed by the coordination between the proton-deprived carboxyl group and positive metal center⁴². Meanwhile, the characteristic peak of aromatic amino νC–N at 1250 cm⁻¹ is still observed in the FTIR spectra of the (R–COO)_xσ-Co₃O_4−x and azobenzene groups. These results indicate that a stable inorganic–organic unit structure (R–COO-Co_Td) is formed, and the intuitive structural diagram of the hybrid is shown in Fig. 3c. Additionally, the characteristic peaks of νCOO (1378 cm⁻¹) and the azobenzene group are absent from the FTIR spectrum of the (R–COO)_xCo₃O₄, which indicates that the mixing of the azobenzene group and Co₃O₄ cannot form a hybrid structure under the same solvothermal conditions. The Co 2p XPS spectra of pristine Co₃O₄ and (R–COO)_xCo₃O₄ are shown in Supplementary Fig. 12. It can be observed that the oxidation state of Co in the two samples does not change evidently, suggesting that the organic molecules do not bond with Co₃O₄ in (R–COO)_xCo₃O₄. Furthermore, nuclear magnetic resonance (NMR) spectroscopy was used to clarify the structural formula of azobenzene-4,4ʹ-dicarboxylic acid. As shown in Fig. 3d, the 1H NMR spectrum of azobenzene group shows quadruple peaks at around 8 PPM, each representing hydrogen in four different environments of azobenzene, which indicates the presence of a symmetrical benzene ring structure at the center of the organic molecule. In contrast, the isolated peak at around 2.3 ppm in the ¹H NMR spectrum of azobenzene groups represents the hydrogen of the carboxylic acid group. Moreover, the ¹H NMR spectrum of the solution exhibits a distinct characteristic peak at 4.8 ppm, which differs from the azobenzene group and is likely attributed to the presence of hydrogen (H) in the aqueous solution. These results confirm the successful synthesis of the azobenzene structure and indicate that, during the reaction process, the azobenzene remains firmly integrated within the hybrid spinel without detaching.

To determine the ligand structure and coordination environment, X-ray absorption near-edge structure (XANES) spectrum was measured. As shown in Fig. 3e, in the O K-edge XANES spectrum of Co₃O₄, the characteristic peak at ~534 eV is attributed to the electronic transitions from O 1s to 2p hybrid orbitals (O 2p hybridized with M 3 d), which belongs to inorganic oxygen (O_M)^43,44. In the O K-edge XANES spectrum of azobenzene group, the peak at ~535 eV arises due to electronic transitions between O 1s and 2p orbitals, corresponding to the organic oxygen in the azobenzene group (O_–N=N–)^43,44. Interestingly, the O K-edge XANES spectrum of (R–COO)_xσ-Co₃O_4−x shows a wide peak that encompasses inorganic and organic oxygen species (O_M and O_–N=N–), indicating that organic molecules have formed a strong chemical coordination bond with the inorganic spinel. The Co L-edge XANES spectra of (R–COO)_xσ-Co₃O_4−x are also shown in Fig. 3f, which exhibit two independent peaks corresponding to the L₃-edge (~535 eV) and L₂-edge (~801 eV). The shoulder peak on the L₃-edge represents the electron transition from Co 2p_3/2 to 3d_5/2, and the intensity is positively correlated with the tetrahedral Co (Co_Td) content in the spinels⁴³. Therefore, an obvious enhancement in the shoulder peak intensity in the XANES spectrum of (R–COO)_xσ-Co₃O_4−x implies that more Co_Td cations are exposed in (R–COO)_xσ-Co₃O_4−x to bind with azobenzene group. The normalized XANES spectra of the C K-edge and N K-edge for (R–COO)_xσ-Co₃O_4−x are shown in Fig. 3g. The characteristic peaks at approximately 285.9, 286.7, and 294.2 eV in the C K-edge spectra of the samples correspond to π*(C=C), π*(C–N–C), and σ*(C–C), respectively⁴⁵. π*(C=C) represents the electronic transition from C 1s to 2pπ* antibonding orbitals in the C=C hybrid bond, and the π*(C=C) peak intensity in the XANES spectrum of (R–COO)_xσ-Co₃O_4−x is evidently lower than that of the azobenzene group⁴⁶. Meanwhile, the decrease in the peak intensity of π*(C–N–C) in the XANES spectrum of (R–COO)_xσ-Co₃O_4−x is due to the weakened electronic transition from the C 1s to 2pπ* antibonding orbitals of N^47,48. These results demonstrated that the structure becomes more disordered after azobenzene is attached to σ-Co₃O_4−x. Furthermore, the σ*(C–C) peak in the XANES spectrum of (R–COO)_xσ-Co₃O_4−x shifts to lower energy, indicating that electron transfer occurs between C and Co, which corresponds to the C=O–O–Co metal–ester bond in (R–COO)_xσ-Co₃O_4−x⁴². The N K-edge XANES spectra of (R–COO)_xσ-Co₃O_4−x and azobenzene group show differences at 391 eV, possibly due to the electronic excitation from N 1s to oxygen 2p orbitals. The intensity of the p–π*(C–N–O) peak at ~399.6 eV in the XANES spectrum of (R–COO)_xσ-Co₃O_4−x is higher than that of the azobenzene group, which is probably due to the electron excitation from N 1s to O 2p_x and 2p_y orbitals (π orbitals). These results indicate that the azo double bond (–N=N–) in (R–COO)_xσ-Co₃O_4−x may have cleaved and hybridized with oxygen atoms inside the lattice, which breaks the original covalency competition within M_Oh–O–M_Td and ensures the sharing of electrons from polar groups. The peaks at 400.9, 404.1, and 410.7 eV in the XANES spectrum of (R–COO)_xσ-Co₃O_4−x correspond to π* (C=N–C), π* (N–C₃), σ* (N–C), respectively⁴⁵. Among these, the intensity of the σ* (N–C) peak is higher than that of the azobenzene group, indicating a higher electronic transition strength within the hybrid structure, which endows the azo group (–N=N–) with comparable reversible redox activity²⁸. These spectroscopic results indicate the successful synthesis of the inorganic–organic hybrid spinel at the molecular and atomic levels.

To verify the crystallinity and phase of the samples, X-ray diffraction (XRD) was performed. As shown in Fig. 4a, the XRD patterns show that the overall structure of the spinel remains intact after organic molecule hybridization. Furthermore, in the XRD pattern of (R–COO)_xσ-Co₃O_4−x, the diffraction peaks at 19.0°, 31.3°, 36.9°, 44.8°, 55.7°, 59.4°, and 65.3° can be indexed to the (111), (220), (311), (400), (422), (511), and (440) planes of cubic spinel Co₃O₄, respectively. To clarify the local coordination structure of (R–COO)_xσ-Co₃O_4−x, X-ray absorption fine structure (XAFS) spectroscopy at the Co K-edge was performed. In Fig. 4b, the characteristic peak at ~1.5 Å represents the single scattering path of the Co–O bond in the first shell, and the characteristic peaks at approximately 2.4 and 3.0 Å are attributed to the scattering paths from the Co central atom at the octahedral and tetrahedral sites to the nearest adjacent metal cations, respectively. These results confirm that the crystal structure of (R–COO)_xσ-Co₃O_4−x still retains the structural characteristics of the spinel crystal, corroborating the Raman and FTIR results. The k³-weighted extended XAFS (EXAFS) spectral fitting curves (Supplementary Fig. 7) and corresponding structural parameters (Supplementary Table 1) show that the peak of (R–COO)_xσ-Co₃O_4−x in the first shell is derived from Co–O coordination (R=1.95 Å). More importantly, the Co–O bond length in (R–COO)_xσ-Co₃O_4−x decreases by ~0.03 Å compared with that in σ-Co₃O_4−x, which is likely due to Co coordinating with the oxygen in the organic molecules (O_-N=N-). Furthermore, the fitting results show that the coordination of the second shell mainly arises from M_Oh–O–M_Oh and M_Td–O–M_Td, with coordination numbers of 3.8 and 8.6, respectively. Compared with that in σ-Co₃O_4−x, the disorder of M_Td–O–M_Td in (R–COO)_xσ-Co₃O_4−x increases, indicating that the introduction of azobenzene molecules reduces the symmetry of the metal coordination environment. These results verify that the azobenzene groups are successfully connected to the tetrahedral metal centers of the spinel. As shown in the Co K-edge XANES spectra in Fig. 4c, the white line of the Co K-edge XANES corresponds primarily to the transition from the occupied 2p state to the empty 3d state. Furthermore, the position of the white line as a function of the oxidation state of the metal is shown in Fig. 4d. Notably, the metal Co center of (R–COO)_xσ-Co₃O_4−x has a higher oxidation state (2.75) than that of σ-Co₃O_4−x (2.68), indicating increased hybridization between the Co 2p orbitals and p–π conjugated orbitals of azobenzene molecular groups^26,27,49. Additionally, the chemical composition and valence states were studied via X-ray photoelectron spectroscopy (XPS). A detailed analysis of the XPS spectra for C, N, and O (Supplementary Figs. 8‒10) reveals the variation in the non-metallic species present in different samples. The high-resolution Co 2p XPS spectra of (R–COO)_xσ-Co₃O_4−x, σ-Co₃O_4−x, and Co₃O₄ in Fig. 4e show the characteristic peaks of Co 2p_3/2 and Co 2p_1/2. Compared to pristine Co₃O₄, the Co 2p_3/2 orbital of σ-Co₃O_4−x shifts to a lower binding energy, indicating a reduction in the oxidation state of Co after oxygen etching. In contrast, the binding energy of the Co 2p orbitals in the XPS spectrum of (R–COO)_xσ-Co₃O_4−x increases, implying that interactions with strongly polar groups ((R–COO)_x) increase the oxidation state of Co in the σ-Co₃O_4−x, which is conducive to optimizing the adsorption strength of the oxygen-containing intermediates. The increase of electron state density of Co sites is conducive to optimize the adsorption strength of the oxygen-containing intermediates and further enhancing the reaction kinetics rate.

Fig. 4. Atomic and electronic structure characterizations.

a XRD patterns of (R–COO)_xσ-Co₃O_4−x, σ-Co₃O_4−x, and Co₃O₄. b Fourier transforms of the Co K-edge EXAFS oscillations of (R–COO)_xσ-Co₃O_4−x, σ-Co₃O_4−x, Co(OH)₂, CoOOH, and the Co foil. c Co K-edge normalized XANES spectra of (R–COO)_xσ-Co₃O_4−x, σ-Co₃O_4−x, Co(OH)₂, CoOOH, and the Co foil. d White-line positions of (R–COO)_xσ-Co₃O_4−x, σ-Co₃O_4−x, Co(OH)₂, CoOOH, and the Co foil depending on the oxidation state. e XPS spectra of (R–COO)_xσ-Co₃O_4−x, σ-Co₃O_4−x, and Co₃O₄. Source data are provided as a Source Data file.

Electrocatalytic application demonstration

Electrochemical properties are an important index to evaluate the practicability of materials⁵⁰. Herein, using the electrocatalytic OER as a model, the OER performance of (R–COO)_xσ-Co₃O_4−x in a 1.0 M KOH solution was evaluated using σ-Co₃O_4−x, Co₃O₄ and commercial RuO₂ as control samples. The picture of the electrochemical testing setup is shown in Supplementary Fig. 28. As shown in Fig. 5a and Supplementary Fig. 13b, linear sweep voltammetry (LSV) curves show that (R–COO)_xσ-Co₃O_4−x exhibits the highest activity, with overpotentials of 230 and 330 mV at the current densities of 10 and 100 mA cm⁻², respectively, which is better than those of Co₃O₄ (370 and 580 mV), σ-Co₃O_4−x (350 and 520 mV), and RuO₂ (340 and 490 mV; Fig. 5a and b). To evaluate the OER kinetics, the Tafel slopes of the samples were analyzed based on the Supplementary Fig. 13a. (R–COO)_xσ-Co₃O_4−x has the smallest Tafel slope (60.9 mV dec⁻¹), exhibiting enhanced reaction kinetics (Fig. 5c). Furthermore, (R–COO)_xσ-Co₃O_4−x exhibits comparable OER overpotentials and kinetic rates compared to previously reported spinel OER electrocatalysts (Supplementary Table 2). Furthermore, electrochemical impedance spectroscopy (EIS) shows that (R–COO)_xσ-Co₃O_4−x has an obviously minimum charge transfer resistance (R_ct; Fig. 5d, Supplementary Fig. 14 and Supplementary Table 3), indicating its rapid electron transfer capability during the OER process. The comparable reaction kinetics of (R–COO)_xσ-Co₃O_4−x is closely related to the introduction of the π-conjugated units of the azo group (–N=N–). The intrinsic electrochemical activity is an important index to determine the performance of the catalyst, which was further investigated via the electrochemical active surface area (ECSA) and electrochemical double-layer capacitance (C_dl) analysis (Supplementary Fig. 15). As shown in Fig. 5e, (R–COO)_xσ-Co₃O_4−x shows the highest C_dl (5.86 mF cm⁻²) among all reference samples, indicating that (R–COO)_xσ-Co₃O_4−x had the highest number of active sites. Furthermore, the mass activity (MA) and turnover frequency (TOF) of (R–COO)_xσ-Co₃O_4−x at an overpotential of 330 mV are 678 A g_metal⁻¹ and 710.5 h⁻¹, respectively, which are tens of times those of control samples (Fig. 5f).

Fig. 5. Electrochemical performance measurements.

a LSV curves with a 90% iR drop correction (scan rate = 5 mV s⁻¹, solution resistance = 2.0 ± 0.5 Ω.). b overpotentials at 10 and 100 mA cm⁻², and c Tafel slopes of (R–COO)_xσ-Co₃O_4−x, σ-Co₃O_4−x, RuO₂, and Co₃O₄. d EIS measurement of different samples. e C_dl values of different samples at 0.19 V vs. RHE. f MAs and TOFs of (R–COO)_xσ-Co₃O_4−x, σ-Co₃O_4−x, RuO₂, and Co₃O₄ at overpotentials of 1.56 V vs. RHE. g Function of the temperature and activation energy of (R–COO)_xσ-Co₃O_4−x, based on the Arrhenius equation. h Stability curves of (R–COO)_xσ-Co₃O_4−x. All measurements were performed in a 1 M KOH solution (pH=13.8 ± 0.05). The error bars are the standard deviations of three individual calculation. Source data are provided as a Source Data file.

To explore the influence of the organic hybrid on the OER kinetics of spinels, LSV was performed at different temperatures. Results show that the activity of the hybrid gradually increases with an increase in the solution temperature (Supplementary Fig. 16a and b) and the hybrid exhibits the highest catalytic activity at 80 °C (210 mV at the current density of 10 mA cm⁻²). According to the transition state theory, this increase in the catalytic activity may be due to the increase in the reaction rate constant with an increase in the reaction temperature, which facilitates the reaction progress⁵¹. To understand the mechanism underlying the OER kinetics of the hybrid catalyst, the activation energy of the sample was analyzed. The activation energy of the reaction depends only on the inherent properties of the catalyst and is independent of its surface area and other external factors. We applied Arrhenius’ deformation formula in electrocatalytic reactions i_k = Ae^−Eaw/RT, where i_k is the kinetic current, A is the electrode area, R is the universal gas constant, T is the reaction temperature, and E_aw is the activation energy at a specific overpotential (η) (details provided in the Experimental Section). Figure 5g shows that the activation energy of (R–COO)_xσ-Co₃O_4−x is 21.5 kJ mol⁻¹ whereas that of σ-Co₃O_4−x is 33.3 kJ mol⁻¹ (Supplementary Fig. 16c). Usually, polar molecules or groups form hydrogen bonds in aqueous systems, which increases the hydrophilicity of OH* intermediates and hinders OER kinetics. Introducing a weakly polar group such as R–COO can disrupt these hydrogen bonds and reduce the interactions between polar molecules, making the *OH intermediate relatively hydrophobic and more thermodynamically favorable for the OER⁵². In addition, the durability of the catalyst is an important index to evaluate its application prospects. To evaluate the catalytic stability of (R–COO)_xσ-Co₃O_4−x, chronoamperometry measurements were performed at specific potential (Fig. 5h). The performance of (R–COO)_xσ-Co₃O_4−x shows almost no degradation after 100 h of continuous operation at 20 mA cm⁻². Chronoamperometry measurements at 20 mA cm⁻² was also conducted on pristine Co₃O₄ for comparison. As a result, the Co₃O₄ catalyst degraded over a 5-hour period (Supplementary Fig. 18). The ICP‒OES results shown in Supplementary Fig. 17 indicate that only a trace amount of Co was leached during stability testing. The chronoamperometry test at elevated current densities (50 mA cm⁻²) and higher temperatures (60 °C) was also conducted. The results show that (R-COO)_xσ-Co₃O_4−x maintains basic stable performance after 50 hours and that the lattice does not undergo reconstruction after the reaction (Supplementary Figs. 19, 20).

Structural stability verification

Due to the structural coordination characteristics of spinel metal oxides, they undergo reconstructive phase changes during catalytic reactions. To investigate the structural stability characteristics of hybrid spinels, various structural and morphological characterizations of hybrid spinels were performed at different stages of the reaction process. The Raman spectroscopy results are shown in Fig. 6a, which shows that the characteristic peak of the spinel remains intact after 20 min of the catalytic reaction and that the peak intensity fluctuates minimally after 30 min of the reaction. The Raman results of (R–COO)_xσ-Co₃O_4−x after the catalytic reaction show that the intensity of the characteristic peak of the spinel decreases but the crystal structure of the spinel is preserved without phase transition. Importantly, in the high-wavenumber region (800–1000 cm⁻¹), the characteristic peaks corresponding to organic molecules are still observed, indicating that organic molecules are still coordinated with the metal active center. Moreover, we conducted Raman spectroscopy on Co₃O₄ both before and after the OER to further investigate its structural stability. As shown in Supplementary Fig. 25, five characteristic peaks belonging to Co₃O₄ (before the OER) were maintained. After the reaction, the characteristic peak of spinel disappeared and formed a cobalt hydroxide mixed phase, indicating phase transformation.

Fig. 6. Morphological and structural characterizations at different reaction stages.

a Raman spectrum of (R–COO)_xσ-Co₃O_4−x (before reaction, 20 min, 30 min, after reaction). b FTIR spectra of(R–COO)_xσ-Co₃O_4−x (before reaction, 20 min, 30 min, after reaction). c Fourier transforms of the Co K-edge EXAFS oscillations of (R–COO)_xσ-Co₃O_4−x (before reaction, 20 min, 30 min, after reaction). d Co–Co bond length of (R–COO)_xσ-Co₃O_4−x at different reaction stages. e O K-edge normalized XANES spectra of (R–COO)_xσ-Co₃O_4−x (before reaction, 20 min, 30 min, after reaction), σ-Co₃O_4−x, and Co₃O₄. f TEM images of (R–COO)_xσ-Co₃O_4−x (before reaction, 20 min, after reaction). Source data are provided as a Source Data file.

These results show that the overall crystal structure of the hybrid spinel does not change during the reaction due to the π-conjugated structure of azo groups (–N=N–). FTIR spectroscopy (Fig. 6b) shows that after 20 min of the catalytic reaction, the ester bond peak (νCOO) at 1378 cm⁻¹ is still present and the metal–oxygen bond of spinel is also present. After 30 min of the reaction, the intensities of the νCOO and Co–O peaks remain relatively stable. Furthermore, after the reaction, the peaks of νCOO and spinel-related Co–O bonds are still can be observed in the FTIR spectrum of (R–COO)_xσ-Co₃O_4−x, indicating that the organic molecule is still coordinated with the metal active center. The atomic structural characteristics of the hybrid spinel at different reaction stages were obtained at the atomic level via XAFS spectroscopy at the Co K-edge. As shown in Fig. 6c, the characteristic peaks remain unchanged, indicating that the spinel structure remains stable during the reaction. According to the fitting results in Supplementary Figs. 21, 22 and Supplementary Table 1, the coordination of the tetrahedra changes slightly during the reaction process, and the overall coordination relation shows good stability (Fig. 6d).

XRD was used to further confirm the stability of the crystal structure. Supplementary Fig. 24 shows that after the reaction, the crystal structure of the (R–COO)_xσ-Co₃O_4−x does not obviously change. The XRD characterization of the Co₃O₄ sample after the reaction confirmed that the structure had undergone reconstruction. The valence state of Co in (R-COO)_xσ-Co₃O_4-x after the reaction was evaluated by XPS. As shown in Supplementary Fig. 11, compared with those of the (R–COO)_xσ-Co₃O_4−x sample before the OER, the Co 2p characteristic peaks of (R–COO)_xσ-Co₃O_4−x (after the OER) exhibit only a minor shift, indicating that the oxidation state of Co essentially remains stable before and after the reaction. Figure 6e shows the changes in the O K-edge XANES spectra of the hybrid. As previously mentioned, the peak at ~534.4 eV represents inorganic oxygen (O_M) in the Co₃O₄ spinel and that at ~535.6 eV represents organic oxygen (O_–N=N–) in azobenzene groups. As the reaction proceeds, the intensity of the inorganic oxygen peak slightly decreases after 20 min. Subsequently, in the (R–COO)_xσ-Co₃O_4−x, the intensity of the inorganic oxygen peak stabilizes after the reaction whereas that of the organic oxygen peak slightly decreases. In particular, the electron transition from the organic oxygen (O_–N=N–) O 1s orbital to the adjacent metal (Co_Td) Co 2p orbital implies the strong coordination bond formed between organic oxygen and the metal tetrahedron (Co_Td). The TEM images of the (R–COO)_xσ-Co₃O_4−x sample show that its morphology does not undergo notable changes at different reaction stages (Fig. 6f). HRTEM images of (R–COO)_xσ-Co₃O_4−x before and after the reaction exhibits no evidence of lattice reconstruction (Supplementary Fig. 23). The slight changes in the lattice spacing are likely due to alterations in the crystal facets caused by the adsorption/desorption of oxygen-related intermediates. In addition, the characteristic signals of azobenzene molecule groups are not observed in the NMR spectrum of the electrolyte (Fig. 3d) after a long reaction⁵³. These characterization results show the high activity and structural stability of (R–COO)_xσ-Co₃O_4−x during continuous operation. In the hybrid spinel, the azo-extended π-conjugated unit (π*–N=N–π*) of the azobenzene group replaces some of the oxygen anions in the tetrahedral sites, breaking the covalency competition between tetrahedral and octahedral sites. Consequently, the enhanced metal–organic oxygen coordination and optimized local electronic distribution ensure the structural stability of materials during catalytic reactions. Furthermore, the reversible redox activity of the azo group (–N=N–) can resist the dissolution of the hybrid during oxidation reactions, further enhancing the overall stability of the material in electrochemical reactions.

Theoretical investigation of the OER mechanism

To further investigate the active site and oxygen evolution mechanism of (R–COO)_xσ-Co₃O_4−x, density functional theory (DFT) calculations were conducted (Supplementary data 1). First of all, the two configurations (vertical and tilted) of the organic molecules relative to the spinel tetrahedra were simulated. The results show that the electronic energy of the vertical structure [(R–COO)_vertical-Co_Td] is −9807.42 Ha, whereas the electronic energy of the tilted system [(R–COO)_tilt-Co_Td] is −9807.53 Ha. This result probably indicates that the (R–COO)_tilt-Co_Td structure is more stable for electrocatalytic reactions. More importantly, the electrostatic potential of (R–COO)_xσ-Co₃O_4−x with the (R–COO)_tilt-Co_Td structure was calculated to investigate the role of the organic molecules. As shown in Fig. 7a, the tetrahedral Co (Co_Td) sites without the linkage of organic ligands are located in a blue mapping region, indicating a lower electrostatic potential. In contrast, the Co_Td sites linked to organic molecules are found in a weak red mapping region, representing a higher electrostatic potential. This probably suggests that the (R–COO)_tilt-Co_Td unit has a dense electron field around the region of higher electrostatic potential, making it more receptive to electrons. It is known that H₂O molecules have low electrostatic potential and are more prone to electron loss⁵⁴. Therefore, the difference in the potential energy field facilitates the adsorption of H₂O molecules onto the (R–COO)_tilt-Co_Td unit, thereby accelerating the evolution of reaction intermediates. Furthermore, the differential charge density was calculated to clarify the impact of the relative position of organic molecules on the spinel structure. As shown in Fig. 7b and Supplementary Fig. 26, a strong energy overlap near the (R–COO)_vertical-Co_Td junction indicates a strong electronic interaction between Co_Td and the organic molecule, weakening the coupling with oxygen-related intermediates during catalysis (Supplementary Table S4). In contrast, an energy gap around the (R–COO)_tilt-Co_Td junction enhances electron transfer with oxygen intermediates (O*, OH*, OOH*), accelerating the catalytic process. Thus, a more curved spatial structure promotes the H₂O oxidation reaction. Figure 7c shows the structure model of H₂O adsorption on the (R–COO)_tilt-Co_Td-based organic hybrid spinel (top view). DFT was used to simulate the five fundamental steps of the OER mechanism, comprising four electrochemical steps (Supplementary Fig. 27) and one non-electrochemical O₂ desorption step, as shown in Fig. 7d. The Gibbs free energy (ΔG) distribution for each step of the OER process on (R–COO)_xσ-Co₃O_4−x was calculated at both U = 0 V and U = 1.23 eV. The results show that the rate-determining step (RDS) for (R–COO)_xσ-Co₃O_4−x is step III (O* → OOH*) with a ΔG of 1.90 eV at U = 0 V. Combining the theoretical and experimental results, it is evident that organic molecule units, which serve as electron aggregation centers, enhance the OER performance of (R–COO)_xσ-Co₃O_4−x. The (R–COO)_tilt-Co_Td) structure disrupts the covalent competition within the tetrahedron-oxygen-octahedron (M_Td‒O‒M_Oh), thereby improving structural stability during the electrocatalytic process.

Fig. 7. DFT investigation of oxygen evolution process on (R–COO)_xσ-Co₃O_4−x.

a Calculated electrostatic potential maps and b Differential charge density of (R–COO)_xσ-Co₃O_4−x. c Top view of (R–COO)_xσ-Co₃O_4−x adsorbing water molecules. d Free-energy diagram of (R–COO)_xσ-Co₃O_4−x at U = 0 eV and U = 1.23 eV. Source data are provided as a Source Data file.

In summary, we report a π-conjugated azobenzene single-tooth coordination method for the development of an inorganic–organic hybrid spinel oxide (named (R–COO)_xσ-Co₃O_4−x). The Raman, FTIR, and XAFS spectroscopic results demonstrate the successful connection of extended π-conjugated azobenzene (π*–N=N–π*) to the tetrahedral Co sites of spinel Co₃O₄ using well-designed vacancy-induced oxygen replacements. XPS and XANES results further confirm that the breaking of the spatial symmetry structure and the weakening of the M_Oh–O–M_Td covalency competition between tetrahedral and octahedral coordination fields in inorganic–organic hybrids, stabilizing the spinel phase structure during electrocatalytic reactions. In the electrochemical OER (which was used as a model), the as-obtained (R–COO)_xσ-Co₃O_4−x catalysts exhibit MA and TOF 10–20 times relative to pristine Co₃O₄ and 100 h structural stability during long-term operation. Going beyond electrocatalysis, the synthetic strategy proposed herein for expanding the spinel family will arouse widespread interest in the fields of renewable energy conversion and storage, practical catalysis, and large-scale chemical industries.

Methods

Materials

Co(CH₃COO)₂·4H₂O (≥98%), NaBH₄ (≥99.5%), C₇H₅NO₄ (≥99%), C₆H₁₂O₆ (≥98%), NaOH (≥99%), KOH (≥99%), 25% NH₃·H₂O (≥98%), C₂H₅OH (≥99%), CH₃OH (≥99%), CH₃COCH₃ (≥99.5%), CH₃C(O)N(CH₃)₂ (≥99%), and 10% HCl (≥98%) were purchased from Sinopharm Chemical Reagent Co., Ltd. Commercial RuO₂ was purchased from Aladdin Biochemical Technology Co., Ltd. All chemicals were used directly without further purification. Deionized (DI) water (18.2 MΩ cm resistivity) was used in all the experiments.

Synthesis of inorganic substances

Synthesis of nanocubic spinels (Co₃O₄)

Nanocubic spinels (Co₃O₄) were synthesized via the high-pressure liquid‒phase method. First, 0.5 g of Co(CH₃COO)₂·4H₂O was dissolved in 25 mL of a mixed solution (23 mL of ethanol and 2 mL of deionized (DI) water), followed by the addition of 2.5 mL of 25% ammonia under intense agitation. Second, the mixture was stirred in air for ~10 min to form a uniform slurry. Next, the suspension was transferred to a 48.0 mL autoclave and reacted at 150 °C for 3 h. After the reaction, the autoclave was cooled to room temperature. The black solid product was centrifuged with DI water, dispersed in DI water, washed with ethyl alcohol, and finally vacuum dried at 60 °C for 12 h. The dried product was collected for characterization.

Synthesis of nanocubic spinels with pores (σ-Co₃O_4−x)

To obtain nanocubic spinels with pores (σ-Co₃O_4−x), 60 mg of the Co₃O₄ (10 nm) sample was immersed in 10 mL of a 0.5 M NaBH₄ solution for 6 h. The samples were washed three times with ethanol to remove residual impurities and dried in a vacuum oven at 60 °C for 12 h. After drying, the product was collected for characterization.

Synthesis of organic substances

The azobenzene-4,4ʹ-dicarboxylic acid–based synthesis method was used with some modifications. First, C₇H₅NO₄ (11 g, 66 mmol) and NaOH (40 g, 100 mmol) were dissolved in 150 mL of water. Second, the mixed solution was transferred to a 500 mL flask, and the temperature of the mixed solution was controlled via a 50 °C water bath, with care taken to maintain the water level above the liquid level of the solvent in the flask. Then, the C₆H₁₂O₆ solution (90 g in 100 mL of water) was added dropwise to the mixed solution over 60 min. The mixture was air-bubbled under continuous stirring overnight, followed by the addition of 50 mL of methanol. The precipitates were collected via filtration, washed with methanol, and dissolved in hot water. Finally, the solution was acidified with a 10% HCl solution, and the product was collected via filtration and further characterized.

Synthesis of inorganic–organic hybrid spinels ((R–COO)_xσ-Co₃O_4−x)

(R–COO)_xσ-Co₃O_4−x was synthesized via a single-tooth coordination reaction, where σ-Co₃O_4−x and carboxylate were used under solvothermal conditions at 120 °C. The prepared σ-Co₃O_4−x precursor, azobenzene-4,4ʹ-dicarboxylic acid, dimethyl formamide (DMF; 16 mL), water (1 mL), and ethanol (1 mL) were packed into Pyrex vials. Next, the mixture was heated at 120 °C for 12 h. After cooling to room temperature, the (R–COO)_xσ-Co₃O_4−x was collected via centrifugation, washed with DMF (twice) and acetone (twice), vacuum-dried at 100 °C, and collected for characterization. The organic molecule loading on (R-COO)_xσ-Co₃O_4-x is around 5.04 wt%.

Morphology and structure characterizations

The X-ray diffraction of the samples were measured using a Philips Xʹ Pert Pro Super X-ray diffractometer with Cu Kα radiation (λ = 1.54178 Å). TEM, HRTEM, and EDS were performed using a JEM-F200 microscope at an acceleration voltage of 200 kV. XPS spectra were recorded using a Thermo ESCALAB250Xi spectrometer with a monochromatized Al Ka excitation source (hv = 1486.6 eV) and a pass energy of 30 eV. The values of the binding energies were calibrated using the C 1 s peak of contaminant carbon at 284.80 eV. Raman measurements were performed using a laser confocal Raman microscope (LabRAM HR Evolution). The Fourier transform infrared spectra of the samples were measured using the Nicolet 8700 FTIR spectrometer. The ¹H NMR spectra were measured using the AVANCE III 400 superconducting Fourier NMR spectrometer. The XANES were measured at the MCD-A and MCD-B beamlines (Soochow Beamline for Energy Materials) at the National Synchrotron Radiation Laboratory (NSRL, China), which were measured in the total electron yield mode in a vacuum chamber (< 5 × 10⁻⁸ Pa).

Electrochemical measurements

The OER performance assessments were carried out through a conventional three-electrode electrochemical setup employing a CHI760E electrochemical workstation (CH Instruments). A clean carbon cloth (CC) with dimensions of 1 cm × 1 cm served as the working electrode, paired with a carbon rod counter electrode and an Hg/HgO reference electrode. The measurements were conducted in 1 M KOH electrolyte. The preparation method of the electrolyte includes the following steps: the solute uses potassium hydroxide (KOH) with a purity of ≥99%, and the solvent is deionized water. The weighing operation is performed first to calculate the mass of the required base based on the target concentration. The mixing operation is then performed, and the alkali is slowly added to the water. The mixture was then stirred and cooled, and the stirring was continued until it was completely dissolved. Finally, the mixture was sealed and stored in a cool, dry place, maintaining a storage temperature between 4 and 25 °C. Multiple pH tests were performed on the configured electrolyte, and the average pH was 13.8 ± 0.05. In this work, the potentials shown are all referenced to the reversible hydrogen electrode (RHE), which was corrected experimentally by using Pt/C as the working electrode potential in a H₂-saturated solution. The carbon paper was thoroughly soaked and cleaned in ethanol, acetone, and 10% hydrochloric acid. The mixture was then vacuum dried at 60 °C for 2 h to avoid oxidation. The catalyst solutions were prepared by mixing 5.0 mg of the catalyst in a solution containing 250 μL ethanol, 730 μL DI water, and 20 μL 5 wt.% Nafion solution, which were sonicated to form homogeneous inks. Using a micropipette gun, 10 μL slurry was uniformly added to the pretreatment substrate surface each time. The mixture was allowed to stand at room temperature for 10 minutes. This process was repeated to a target load of 0.3 mg cm^-2. Loading capacity = (slurry mass × catalyst mass fraction)/base area. The optimal electrochemical experimental device was used for testing, and then the solution resistance was measured. After several measurements to obtain the average value, the resistance value is 2.0 ± 0.5 Ω. All final potentials are converted to the RHE using the following equation: E (vs. RHE) = E (vs. Hg/HgO) + 0.098 V + 0.059 × pH. LSV curves were obtained at a rate of 5 mV s⁻¹ with 90% iR correction. Furthermore, the potential was determined using the following equation: E_Corrected = E_Uncorrected – iR_S. Electrochemical stability tests are carried out using chronoamperometry at a constant voltage of 1.50 V vs RHE, and two constant temperatures of 25 and 60 °C are maintained by an electrothermal sensor. The electrochemical surface area (ECSA) was evaluated through cyclic voltammetry (CV) measurements conducted within a non-Faradaic potential range. The ECSA values were derived from the double-layer capacitance (C_dl) using ECSA = C_dl/C_s, where C_s represents the specific capacitance of the material. For KOH electrolyte, the standard value of C_s is 0.04 mF cm⁻². Mass activity (MA (A g_Co⁻¹)) was calculated using the equation MA = (j × A) ÷ m, where j denotes the current density at 1.56 V vs. RHE, A is the geometric electrode area, and m is the mass of the loaded active metal. Turnover frequency (TOF (s⁻¹)) was determined via TOF = (j × A) ÷ (4 × F × M), where F is the Faraday constant (96,485 C mol⁻¹) and M corresponds to the molar quantity of the active metal on the electrode⁵⁰.

Calculation and derivation of the Arrhenius formula for electrocatalytic reactions

The variation in current density with electrolyte temperature (25, 30, 40, 50, 60, 70, and 80 °C) at 1.44 V was determined using the Arrhenius relationship. The Arrhenius equation, expressed as k = A·exp(−Eₐ/(RT)), relates the reaction rate constant (k) to the pre-exponential factor (A), activation energy (Eₐ), universal gas constant (R), and temperature (T). For electrocatalytic processes, the rate constant k also depends on the applied overpotential (η) through the equation k = k₀·exp(−αFη/(RT)), where k₀ denotes the standard rate constant, F is the Faraday constant (96,485 C·mol⁻¹), and α represents the charge transfer coefficient. The kinetic current (i_k) is governed by i_k = knFAC, with n as the number of electrons transferred, A as the electrode area, and C as the bulk analyte concentration. Under the assumption of negligible mass transport limitations and constant parameters in the expression i_k = knFAC, the Arrhenius formalism simplifies to i_k = A·exp(−Eₐ/(RT)), where Eₐ corresponds to the activation energy at a fixed overpotentia⁵⁵.

XAFS measurements

The XAFS experiments for Co K-edge were conducted at the BL11B beamline of the Shanghai Synchrotron Radiation Facility (SSRF) in China. The SSRF storage ring operated at an energy of 3.5 GeV and a maximum beam current of 300 mA. Si (111) double-crystal monochromator was employed to monochromatize the X-ray beam extracted from the bending magnet. Notably, data acquisition utilized transmission mode for the Co foil control sample, and fluorescence excitation mode for the experimental samples³¹.

XAFS data analysis

The EXAFS data analysis for determining the local coordination structure of Co in the samples was conducted through ATHENA and ARTEMIS modules within the IFEFFIT software package⁴⁸. The k³-weighted χ(k) data (2.5‒12.5 Å⁻¹ of k-range) were Fourier-transformed to R-space under the Hanning windows of dk = 1.0 Å⁻¹. The fitted R-range was 1.0‒3.5 Å. Accordingly, the number of independent points (N_ipt) was evaluated using the equation: N_ipt = 2Δk × ΔR/π = 2 × (12.5−2.5) × (3.5 − 1.0)/π = 15. The amplitude reduction factor of S₀² was obtained by fitting the Co foil data, and the value was 0.73 herein. S₀² was treated as a fixed parameter in the fitting procedure for (R–COO)_xσ-Co₃O_4−x and σ-Co₃O_4−x. Further, the coordination numbers, bond lengths (R), Debye–Waller factors (σ²), and energy shifts (ΔE₀) were treated as adjustable parameters (N_para) according to the fitting criterion, and N_para = 10 <N_ipt under this condition.

Computational methods

The spin unrestricted density functional theory methods were performed as implemented in the DMol³ code of the Materials Studio package⁵⁶. The Perdew–Burke–Ernzerh functional within the generalized gradient approximation was employed to describe the electron exchange–correlation potential, and a double numerical plus polarization basis set was selected⁵⁷. The convergence tolerances for energy change, maximum force, and maximum displacement were set to 2.0 × 10⁻⁴ Ha, 0.004 Ha Å⁻¹, and 0.005 Å, respectively. The calculated model consists of a 3 ×2 × 1 supercell, and a vacuum layer of 20 Å in the z direction was used to avoid unnecessary interactions between periodic images.

Author contributions

Q.H.L., W.R.C. and S.W.B. conceived the project. S.W.B. carried out the experiments. C.M.W. carried out the infrared spectroscopy measurements. H.J.W. carried out the TEM measurement. X.C. carried out the DFT calculations. S.W.B., W.L.Z., X.X.Z., W.R.C., and Q.H.L. analyzed the experimental data. The manuscript was written by Q.H.L., S.W.B., W.R.C. and W.L.Z. with contributions from all authors.

Peer review

Peer review information

Nature Communications thanks Ali Morsali, Jian-Gan Wang and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data are available in the main text or the supplementary materials. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-57799-2.