Spatial engineering and d-orbital coupling in axial dual-atom sites for bifunctional oxygen catalysis Authors: Xinru Yan, Xiaoliang Yuan, Ning Liu, Bin Liao, Zhanhao Liang, Wencai Liu, Yihui Huang, Longwen Yan, Qing Zheng, Shufen Chen, Xi Xie, Xuchun Gui, Hong Bin Yang, Jiling Li, Dingshan Yu, Zhiping Zeng, Guowei Yang Published: 2025 ABSTRACT AbstractHeterogeneous dual-atom catalysts (DACs) offer great potential to enhance electrocatalytic reaction kinetics and provide diverse active sites. However, achieving precise tuning of metal atom coordination in DACs remains a significant challenge. Here, the axial dual atom (ADA) embedded within a covalent organic framework and N-doped graphene, features an axial intermetallic distance tuned by alcohol solvent treatment, resulting in efficient bifunctional electrocatalysis of oxygen reduction reaction (ORR) and oxygen evolution reaction (OER). In situ X-ray absorption near-edge spectroscopy (XANES) and Raman spectroscopy reveal that Fe acts as the primary active center in the axially coordinated FeCo-ADA, with Co providing synergistic effects. The in-depth theoretical analysis elucidates that the axial Fe-Co orbital coupling results in optimized orbital energy levels, higher Fe oxidation state, weakened oxygen intermediate binding strength, and reduced reaction energy barrier. The zinc-air battery (ZAB) equipped with FeCo-ADA achieves a high peak power density of 464.5 mW cm-2 and exhibits long rechargeability of 3710 hours at 10 mA cm-2. Our findings pave an avenue for the rational design in axial DACs to enhance the electrocatalytic performance for energy conversion.Subject terms: Electrocatalysis, Fuel cells Introduction Oxygen redox reaction plays a pivotal role in promising sustainable energy conversion technologies such as metal-air batteries, fuel cells, and electrocatalytic water splitting1–3. However, the complex multielectron reaction process, involving strong O-O bond (498 kJ mol−1) rupture and formation, result in sluggish kinetics for both the oxygen reduction reaction (ORR) and the oxygen evolution reaction (OER), highlighting the urgency for efficient bifunctional catalysts to facilitate these reactions4–9. Despite platinum-, ruthenium-, and iridium-based electrocatalysts being the current benchmark for ORR or OER applications, their high cost, scarcity, and instability restrict widespread use10,11. Extensive research is focused on exploring cost-efficient non-precious metal catalysts12. Recently, metal-N-C single-atom catalysts with precise coordination configurations have emerged as a promising alternative to precious metal catalysts, owing to the utilization of metal atoms and adjustable active sites13–18. The FeN4 site is regarded as optimal for efficient ORR activity, attributed to the high atomic utilization enabled by the dispersed Fe-N-C configuration19–27. When evaluated at identical loadings, the Fe–N–C catalyst delivers ORR performance comparable to the Pt/C28. For example, a hierarchically porous Fe-SAC with Fe-N4 sites demonstrates enhanced ORR activity, due to the long-range electron modulation that facilitates the desorption of the *OH intermediate29. However, due to the inherent limitation of linear relationship, SACs possess singular active centers that can primarily accommodate only one single oxygen-containing ORR/OER intermediate (*O, *OH, or *OOH), thereby constraining their bifunctional oxygen catalysis. The development of robustly stable and efficient bifunctional catalysts remains an urgent challenge. Dual-atom catalysts (DACs), featuring diverse active sites, tunable spatial configuration, and variable surrounding coordination species and number, hold promising for overcoming the inherent activity limits of SACs30–37. For example, FeCo-N3O3@C with its strongly coupled FeN3 and CoO3 species optimizes the desorption-adsorption of oxygen intermediates, displaying robust and efficient bifunctional oxygen catalytic activity38. Beyond planar DACs in two-dimensional configurations, axial-stacked DACs with three-dimensional confinement effectively reduce the out-of-plane mobility of metal atoms. This structural arrangement promotes strong electronic coupling and synergistic interaction, boosting structural stability39–42. The heterogeneous characteristics of the axial coordination configuration give rise to distinct catalytic sites and reaction pathways, thereby influencing the overall catalytic performance. However, the precise tuning of axial dual-atom sites remains a huge challenge, especially in terms of adjusting the types of substrates and controlling the intermetallic distance. Herein, we designed and synthesized an axial DACs, integrating a covalent organic framework (COF) with N-doped graphene, which exhibits efficient electrocatalytic activity for oxygen reduction reaction (ORR) and oxygen evolution reaction (OER). A solvent-based approach enables adjustable axial dual-atom spacing, allowing for the tailed axial coordination environment. In situ Raman spectroscopy and X-ray absorption near-edge spectroscopy (XANES) reveal that Fe functions as the primary active center in the axially coordinated FeCo-ADA catalyst, with Co providing synergistic effects. The zinc-air battery (ZAB) equipped with FeCo-ADA achieves a high power density (464.5 mW cm−-2) and a long-term rechargeability (3710 hours at 10 mA cm−2). The in-depth theoretical analysis elucidates that the axial Fe-Co orbital coupling results in optimized orbital energy levels, a higher iron oxidation state, weakened oxygen intermediate binding, and reduced reaction energy barrier. Our findings pave an avenue for the rational design in axial DACs to enhance their electrocatalytic performance for energy conversion. Result and discussion Synthesis and characterization of the FeCo-ADA catalyst Axial tuning DACs can improve site accessibility over planar configurations via reduced metal mobility and enhanced orbital coupling (Fig. 1a). FeCo-ADA with an axial-stacked configuration was designed using COFBTC and N-doped graphene (NG) as two-dimensional substrates (Supplementary Fig. 1). To evaluate the dynamic stability of FeCo-ADA, the ab initio molecular dynamics (AIMD) simulations were carried out at 800 K (Fig. 1b). The negligible structural deformation and energy fluctuation after 9 ps indicate its thermodynamic stability. The incorporation of Fe and Co leads to a redistribution of electrons, as evidenced by the symmetry-breaking charge distribution around the Fe site in FeCo-ADA, which exhibits a higher concentration of positive charges (Fig. 1c). In contrast, the Co site exhibits minimal electronic rearrangement (Supplementary Figs. 2–4), attributable to compensatory charge redistribution between Co and its coordinating N ligands upon axial Fe-Co bonding. Moreover, the projected density of states (PDOS) analysis of the five d-orbitals (dxy, dyz, dxz, dz², and dx²-y²) for both Fe and Co in FeCo-ADA shows strong d-d orbital coupling between heteroatoms (Fig. 1d and Supplementary Fig. 5). Specifically, the spin polarization of Fe 3 d orbitals is significantly strengthened upon Co coordination. FeCo-ADA possesses a higher Fe magnetic moment (2.00 μB) relative to Fe-COF (1.03 μB) and a notably enhanced spin polarization near the Fermi level, reaching 88% at −1.5 eV compared to Fe-COF (59%). This enhancement amplifies the hybridization between the O2 π* antibonding orbitals and the Fe d-orbitals, thereby lowering the O-O bond dissociation energy barrier and accelerating the four-electron ORR process43,44. Furthermore, the increased spin polarization in FeCo-ADA induces a splitting between the spin-up and spin-down orbital energy levels, facilitating preferential electron transfer through specific spin channels on the catalyst surface. Total DOS (TDOS) analysis reveals that FeCo-ADA achieves enhanced electron transfer efficiency and electrical conductivity relative to the monomeric Fe-COF and Co-NG (Supplementary Fig. 6). The introduction of an axial Co atom modifies the Fe d-orbital energies, particularly downshifting the dxy and dx²-y² levels (Fig. 1e, f and Supplementary Figs. 7,8). This orbital modification weakens the binding of OH, reducing the desorption energy barrier for the ORR and consequently enhancing catalytic performance. Electronic structure analysis and schematic diagram of orbital hybridization. a Geometric structures of planar dual atom and axial dual atom catalysts. b The variations of energy during the 9 ps AIMD simulation. Charge density differences of FeCo-ADA (c) and Fe-COF (d). e PDOS for Fe-3d and Co-3d orbitals in FeCo-ADA. The energy level of dxy (f) and dx²-y² (g) in FeCo-ADA. The shade of color refers to the probability densities of spin-up/spin-down electrons. Based on theoretical design, axially coordinated DACs were synthesized by combining microwave-synthesized COFBTC-based SACs with pyrolysis-treated metal-NG (Supplementary Fig. 9). Using a solvent-based approach, axial dual-atom spacing was further adjusted to tailor the coordination environment. Methanol, ethylene glycol, and 1,4-butanediol serve as solvents during the fabrication of FeCo-MeOH, FeCo-ADA, and FeCo-BDO, respectively. Fourier-transform infrared spectroscopy (FTIR) of the Fe-COF and Co-NG mixed with different alcohols reveals distinct Co-O and Fe-O peaks at 507 and 635 cm⁻¹, respectively (Supplementary Fig. 10). These peaks reflect solvent-mediated control of coordination chemistry during synthesis. After argon-purged calcination, the samples display a high density of bright spots (Fig. 2a, c and Supplementary Figs. 11). Interestingly, the statistical analysis of marked metal pairs indicates that the atomic distances for the majority of DACs are 0.299 nm and 0.312 nm for FeCo-ADA and FeCo-BDO (Fig. 2b, d). In contrast, FeCo-MeOH exhibits a disordered single atom distribution (Supplementary Figs. 12). To assess the impact of extended alcohol chains on axial distance, FeCo-POD was synthesized with 1,5-pentanediol. AC-HAADF-STEM coupled with statistical analysis determined an intermetallic distance of 0.684 nm (Supplementary Fig. 13), suggesting the structural impact of the alcohol chain extension. The transmission electron microscopy (TEM) and field emission scanning electron microscopy (FESEM) confirmed the preservation of graphene-layered structure in FeCo-ADA, Fe-COF, and Co-NG (Supplementary Figs. 14-16). Dark-field scanning transmission electron microscopy (STEM) and energy-dispersive X-ray spectroscopy (EDS) mapping of FeCo-ADA revealed a uniform distribution of Fe and Co atoms (Fig. 2e). Inductively coupled plasma atomic emission spectrometry (ICP-AES) Fe and Co contents of 1.67 and 1.05 wt%, respectively (Supplementary Table 1). Atomic force microscopy (AFM) analysis indicate that the thickness of FeCo-ADA (3.037 nm) closely matches the combined thickness of Fe-COF (2.056 nm) and Co-NG (1.079 nm) (Fig. 2f, Supplementary Figs. 17-19). X-ray diffraction (XRD) of FeCo-ADA shows peaks at 5.8° and 27.4° (Fig. 2g and Supplementary Fig. 20), corresponding to the (100) plane and π-π stacking of COFBTC45, respectively. The 27.4° peak indicates an interlayer spacing of approximately 0.3 nm in FeCo-ADA, consistent with the 0.299 nm lattice spacing observed by HRTEM. N2 adsorption-desorption analysis reveals that FeCo-ADA possesses a large surface area of 290.50 m2 g−1, surpassing that of Fe-COF (109.52 m2 g−1) and Co-NG (217.52 m2 g−1) (Supplementary Fig. 21). Synthesis and characterization of FeCo-ADA. Aberration-corrected HAADF-STEM images of a FeCo-ADA, c FeCo-BDO. The distance distribution between neighboring atoms in b, FeCo-ADA, d FeCo-BDO. The insets in panel depict the HAADF-STEM image and the proposed structural model of a typical neighboring atom pair. e HAADF-STEM image and corresponding EDS mapping images. f AFM image of FeCo-ADA. g XRD patterns of FeCo-ADA. The inset in panel depicts the intensity profiles of the two atoms labeled in (a). The black, blue, yellow, and purple balls represent carbon, nitrogen, iron, and cobalt atoms, respectively. h Fe K-edge XANES spectra. i FT-EXAFS spectra. j Experimental Fe K-edge XANES spectra and the corresponding theoretical spectra of FeCo-ADA. k FT-EXAFS fitting curves at the R space of FeCo-ADA. The black, blue, yellow, and purple balls represent carbon, nitrogen, iron, and cobalt atoms, respectively. X-ray photoelectron spectroscopy (XPS) was conducted to investigate the elemental chemical states of FeCo-ADA (Supplementary Figs. 22-24). The overall survey spectra reveal distinct peaks for Co (779.9 eV), Fe (709.8 eV), C (284.6 eV), and N (398.4 eV). Compared to Fe-COF and Co-NG, the Fe and Co 2p peaks in FeCo-ADA exhibit positive shifts, suggesting an increase in valence state at the Fe and Co sites in FeCo-ADA (Supplementary Fig. 25). X-ray absorption spectroscopy (XAS) was further carried out to explore the chemical states and coordination environments of the metallic elements in FeCo-ADA. The X-ray absorption near-edge structure (XANES) at the Fe K-edge of FeCo-ADA and Fe-COF lies between the Fe foil and FeO, suggesting an oxidation state of Fe between 0 and +2 (Fig. 2h). Fourier-transform (FT) analysis of the EXAFS spectra exhibited a peak at 1.44 Å for FeCo-ADA, attributed to Fe-N scattering paths, and an apparent peak at 2.52 Å, indicating Fe-Co coordination (Fig. 2i). The simulated Fe K-edge XANES spectrum matched experimental data for FeCo-ADA, showing agreement and accurately replicating the characteristic features (Fig. 2j). The FT-EXAFS spectrum can be satisfactorily fitted by a model comprising both Fe-N and Fe-Co scattering paths. Notably, the fit reveals a Fe-Co bond length (2.99 Å) with a coordination number of 1.37, alongside a primary Fe-N shell. In addition, the presence of Fe-Co coordination supports the formation of an axial Fe-Co dual-atom center (Fig. 2k and Supplementary Table 2). In comparison, FeCo-BDO exhibited a distinct peak at 2.54 Å. Its structure features Fe-N and Fe-Co coordination numbers of 4.23 and 1.36, with bond lengths of 1.92 and 3.00 Å, respectively, indicative of longer Fe-Co coordination (Supplementary Table 3). The Co K-edge XANES spectra reveal that the absorption edges of Co in FeCo-ADA and Co-NG are intermediate between those of Co foil and CoO, indicating that Co oxidation states range between 0 and +2. FT-EXAFS of the Co K-edge for FeCo-ADA revealed peaks at 1.41 Å and 2.41 Å, assignable to Co-N and Fe-Co interactions, respectively. A notable peak at 2.43 Å in FeCo-BDO confirms longer Fe-Co bond (Supplementary Figs. 26,27). Simulated Co K-edge XANES data for FeCo-ADA also showed strong agreement with the actual sample structure (Supplementary Fig. 28a). Consistent with the FT-EXAFS result (Supplementary Fig. 28b), the Wavelet Transform (WT) contour plot for the Fe K-edge of FeCo-ADA exhibited two peaks at 4.8 and 7.5 Å−1, corresponding to Fe-N and Fe-Co coordination, respectively. The Co K-edge WT-EXAFS similarly indicated Co-N and Co-Fe coordination at 4.8 and 7.2 Å−1 (Supplementary Fig. 29). Accordingly, the Fe-Co scattering path and its fitting parameters provide direct evidence for an axial coordination mode in FeCo-ADA, characterized by the formation of an Fe-Co axial bond. Electrochemical performance The electrochemical performance of catalysts in ORR and OER were systematically investigated using cyclic voltammetry (CV) and linear sweep voltammetry (LSV) in a three-electrode system. Compared to SACs, FeCo-ADA demonstrates the largest electrochemically active surface area (ECSA), with a capacitance (Cdl) value of 7.6 mF/cm2 (Supplementary Fig. 30). In an O2-saturated 0.1 M KOH, FeCo-ADA displays a high ORR activity, achieving a half-wave potential (E1/2) of 0.92 V. This ORR activity outperforms mixed-Fe-COF+Co-NG (0.83 V), FeCo-MeOH (0.91 V), FeCo-BDO (0.88 V), Co-NG (0.82 V), Fe-COF (0.60 V), and Pt/C (0.85 V) (Fig. 3a and Supplementary Figs. 31,32). FeCo-ADA processes the highest catalytic activity of 28.58 mA cm−2 at 0.85 V and the lowest Tafel slope of 42.7 mV dec−1 (Fig. 3b, c). These values surpass those of Pt/C (5.74 mA cm−2 and 112.1 mV dec−1), confirming enhanced kinetic rate of FeCo-ADA for ORR. LSV curves at various scan rates and the corresponding Koutecky-Levich (K-L) plots indicate a 4-electron ORR pathway for FeCo-ADA (Supplementary Fig. 33). Over the potential range from 0.3 to 0.8 V (vs. RHE), the ring current of the FeCo-ADA, corresponding to H2O2 oxidation on the disk electrode is low and the average n value is more than 3.82 (Supplementary Figs. 34,35). This suggests that FeCo-ADA primarily promotes the 4e− reaction pathway, in agreement with previous literature46. The current density of FeCo-ADA remained virtually unchanged after 40000 s of continuous operation at 0.65 V (vs. RHE), with a retention of 98.2%, outperforming that of Co-NG (66.8%), Fe-COF (58.5%), and Pt/C (66.8%) (Fig. 3d). This reflects the long-term stability of FeCo-ADA. Following the United States Department of Energy (DoE) protocol, FeCo-ADA exhibited long-term stability, with only an 8 mV negative shift in its E1/2 after 10,000 cycles, outperforming the Fe-COF and Co-NC (Supplementary Figs. 36,37). Additionally, it demonstrated enhanced methanol tolerance in alkaline media compared to Pt/C (Fig. 3e). In-situ Raman spectroscopy was conducted in the potential range from 1.0 to 0.3 V (vs. RHE, Supplementary Fig. 38). Initially, without the application of an external potential, no characteristic Raman peaks were detected in the range of 200 to 800 cm−1. However, upon applying a potential of 1.0 V (vs. RHE), a distinct peak at 699 cm-1 emerged, corresponding to the Fe-O bond47. When the potential gradually decreases to lower than 0.7 V (vs. RHE), four weak Raman peaks appear at 290, 435, 510, and 611 cm−1 (Fig. 3f and Supplementary Fig. 39a), corresponding to Fe-OOH, Fe-OH, Co-OH, and Co-OOH, respectively48–50. The OER performance of FeCo-ADA was further envaulted. The overpotential at 10 mA cm−2 (EJ10) was 310 mV for FeCo-ADA, surpassing Fe-COF+Co-NG Mixed (320 mV), FeCo-MeOH (430 mV), FeCo-BDO (390 mV), Co-NG (390 mV), Fe-COF (450 mV), and commercial RuO2 (330 mV) (Fig. 3g and Supplementary Fig. 40). FeCo-ADA exhibited the lowest Tafel slope among the samples (Supplementary Fig. 41), suggesting its favorable OER kinetics. FeCo-ADA demonstrated a high stability over an 40000 s duration (Fig. 3h). In situ Raman spectroscopy was employed to gain further insights into the OER process. At an applied potential of 1.2 V (vs. RHE), a characteristic peak at 699 cm−1 was observed (Fig. 3i and Supplementary Fig. 39b), attributed to Fe-O47. This peak intensified with increasing potential, indicating the formation of higher oxidation states. As the potential increased, four distinct peaks emerged at 407, 476, 550, and 611 cm−1, corresponding to Fe-OH, Co-OOH, Co-O, and Co-OOH, respectively47,48,50,51. These finding indicate that Fe acts as the initial active site during ORR and OER, while Co exerts a synergistic catalytic effect. The bifunctional oxygen electrocatalytic activity of FeCo-ADA was further assessed in Supplementary Table 4. Post-electrochemical characterization via STEM image and XPS of FeCo-ADA demonstrated no discernible changes in surface morphology or chemical structure (Supplementary Figs. 42,43), confirming its enhanced stability. Bifunctional oxygen electrocatalysis performance. a Linear sweep voltammograms for the ORR in 0.1 M KOH solution. b Tafel plots. c Comparison of kinetic current density (Jk) @ 0.85 V and E1/2. d Chronoamperometry measurement under 0.65 V (vs. RHE). e Methanol tolerance test. g LSV plots for the OER in 1.0 M KOH solution. h Chronoamperometry measurement under 1.45 V (vs. RHE). In-situ Raman spectra for the ORR (f) and OER (i). Operando XAS study To elucidate the redox transformations in FeCo-ADA, in situ X-ray absorption spectroscopy (XAS) were conducted in a three-electrode configuration (Supplementary Fig. 44). The electronic structure and coordination environment of Fe and Co sites were monitored under different applied potentials in a 0.1 M KOH electrolyte. When the applied potential was increased to 0.75 V (vs. RHE) from the open-circuit potential (OCP), the Fe absorption edge shifted to higher energy, resulting in an increased oxidation state, which is attributed to the adsorption of oxygen species (Fig. 4a, b). Subsequently, a noticeable decrease in the Fe K-edge energy was observed from 0.75 to 0.35 V (vs. RHE), indicating the oxidation state of Fe gradually decreases. Higher potentials facilitate electron transfer to the Fe site, leading to enhanced oxygen species desorption and subsequent promotion of the ORR52. The Co K-edge XANES spectrum and its first-derivative XANES curve shift towards lower energy direction, when the potential increases from 0.75 V to 0.35 V (vs. RHE) (Fig. 4c, d). This suggests that Co atoms play a significant role as co-catalytic sites during the ORR process, which is in accordance with the in-situ Raman spectroscopy results. The dynamic electronic structure of Fe and Co for OER was also clarified by the in-situ XAS spectra. The absorption edge of Fe-XANES shifts to higher energies from 1.25 V to 1.65 V (vs. RHE), indicative of an increased oxidation state due to the adsorbed *OH under electrochemical conditions (Fig. 4e, f). The Co absorption edge moves in the positive direction (Fig. 4g, h), following a similar trend to the Fe sites. As the OER potential increases, the Co K-edge spectrum shifts towards higher energy, indicating an increase in the oxidation valence state of Co. This finding is consistent with the in-situ Raman spectroscopy analysis. The enhanced oxidation state of Fe and Co during OER facilitates the binding of oxygen species, thereby boosting the OER efficiency53. In summary, in-situ XAFS analysis, combined with in-situ Raman spectroscopy, elucidated that both Fe and Co sites serve as active centers for ORR and OER (Fig. 4i). In-situ X-ray absorption near-edge structure spectroscopy and the reaction pathways. In-situ XANS spectra of Fe (a) and Co (c) K-edge during ORR. The trend of the oxidation states of Fe (b) and Co (d) at different ORR potentials. In situ XANS spectra of Fe (e) and Co (g) K-edge during OER. The trend of the oxidation states of Fe (f) and Co (h) at different OER potentials. i Schematic of the reaction pathways on FeCo-ADA. Density functional theory analysis The underlying mechanisms of FeCo-ADA, including the charge transfer, catalyst-adsorbate bonding strength, and orbital hybridization, were conducted to elucidate the origin of its enhanced ORR and OER performance. O2 adsorption energies (ΔE) at the interlayer Fe-Co and outer Fe sites in FeCo-ADA were constructed (Supplementary Fig. 45). It reveals a significantly lower value for the interlayer site (−0.74 eV) than the outer Fe site (−0.51 eV). Indicating that the initial O2 adsorption predominantly occurs at the interlayer site. Since the rate-determining step (RDS) in FeCo-ADA involves *OH desorption, the orbital coupling between Fe and *OH was further analyzed. Fig. 5a, b illustrates a schematic comparison of Fe-OH orbital coupling in FeCo-ADA and Fe-COF. Strong orbital coupling (Fe-dz² and O-pz, Fe-dxz and O-px) is evident. In FeCo-ADA, the head-to-head overlap between the Fe-dz² and O-pz orbitals forms a σ bond, while their out-of-phase interaction generates an antibonding σ* orbital. Compared to Fe-COF, the downshifted energy level of the Fe-dz² orbital in FeCo-ADA increases electron occupancy in the antibonding σ* orbital, thereby weakening the σ bond. The side-by-side overlap of the Fe-dxz and O-px orbitals forms a weaker π bond. The downshifted Fe-dxz orbital in FeCo-ADA elevates electron occupancy in the antibonding π* orbital, further weakening the π bond. Consequently, the weakened σ and π bonding reduce *OH adsorption strength, lowering the energy barrier of the RDS. The charge density differences analysis reveals enhanced electron transfer from FeCo-ADA (0.464 |e | ) to adsorbed *OH compared to Fe-COF (0.402 |e | ) and Co-NG (0.451 |e | ) (Supplementary Fig. 46). This increased electron transfer promotes the electron-accepting desorption of *OH during ORR (*OH + e⁻ → * + OH⁻). PDOS analysis displays a downshifted d-band center for Fe in FeCo-ADA (−1.52 eV) compared to Fe-COF (−0.92 eV), indicating increased antibonding orbital electron occupancy (Fig. 5c and Supplementary Fig. 47). This weakens the Fe-OH bond in FeCo-ADA, facilitating *OH intermediates desorption. The integrated crystal orbital Hamiltonian population (ICOHP) results also displays that a significantly reduced Fe-O bonding strength in FeCo-ADA (−1.35 eV) compared to Fe-COF (−2.55 eV), indicating facilitated *OH desorption and optimized RDS energy barrier (Supplementary Fig. 48). The projected crystal orbital Hamiltonian population (pCOHP) analysis of FeCo-ADA reveals the electronic coupling between the Fe-dz² and O-pz orbitals, as well as the Fe-dxz and O-px orbitals (Supplementary Fig. 49). Structural optimization of the adsorbed *OH configurations further confirms the weakened Fe-O interaction in FeCo-ADA, with a longer Fe-O bond length (2.16 Å) than that in Fe-COF (1.79 Å). The ORR and OER free energy diagrams for FeCo-ADA under adsorbate evolution mechanism (AEM) are presented (Fig. 5d, e). The RDS for ORR of FeCo-ADA is *OH desorption step (*OH → * + OH-), with a lower overpotential of 0.34 eV compared to Fe-COF (0.55 eV) and Co-NG (0.48 eV). Free energy calculations indicate the oxide pathway mechanism (OPM, Supplementary Fig. 50) is energetically unfavorable, due to its high ORR overpotential (1.04 eV). Thus, FeCo-ADA follows the AEM pathway during ORR, corroborated by in-situ Raman spectroscopy identifying the intermediates *OH, *O, and *OOH. The RDS for OER via the AEM pathway in FeCo-ADA is *O → *OOH, with an overpotential of 0.56 eV, lower than those of Fe-COF (0.96 eV) and Co-NG (0.62 eV). These findings demonstrate the high catalytic activity of FeCo-ADA in both ORR and OER. Additionally, the calculations of the outer-layer Fe and Co reaction pathway free energies exhibit high overpotentials, supporting the previously proposed inner-layer pathway (Supplementary Figs. 51, 52). Orbital interactions between the adsorbed *OH and Fe site along with the reaction free energy diagrams. Schematic illustration of orbital interactions between adsorbed O (px) and Fe 3 d orbital (dz2, dxz) in FeCo-ADA (a) and Fe-COF (b). c, PDOS for Fe 3 d orbitals in FeCo-ADA and Fe-COF. The reaction free energy for ORR (d), and OER (e). Rechargeable zinc-air batteries performance Given the enhanced bifunctional catalytic activity of FeCo-ADA, a rechargeable ZAB equipped with FeCo-ADA was assembled. FeCo-ADA served as the cathode, while zinc foil acted as the anode, with the electrolyte comprising 6.0 M KOH and 0.2 M Zn(CH₃COO)₂ (Fig. 6a). In comparison to the ZAB utilizing Pt/C + RuO2, which exhibit an open-circuit voltage of 1.454 V and a specific capacity of 676 mAh g−1, the FeCo-ADA-based ZAB demonstrates a higher open-circuit voltage (1.537 V, Supplementary Fig. 53) and an enhanced specific capacity (750 mAh g−1, Fig. 6b). Under a series of discharge current densities (1.0, 5.0, 10, 20, 40 mA cm−2), the ZAB equipped with FeCo-ADA exhibited higher discharge voltage plateaus (Fig. 6c), indicative of high-performance ORR kinetics and rate capability. Notably, the FeCo-ADA-based ZAB achieves the high peak power density of 464.5 mW cm−2, outperforming the Pt/C-based ZAB (104.9 mW cm−2) (Fig. 6d). To evaluate the durability and reversibility of FeCo-ADA, a constant current cycling test was conducted. The FeCo-ADA-based battery exhibits long-term stability, maintaining an operation for over 3710 h at 10 mA cm−2 (Fig. 6e), 285 h at 20 mA cm−2 (Supplementary Fig. 54), and 235 h at 50 mA cm−2 (Supplementary Fig. 55). These performances are competitive with those of previously reported zinc-air batteries. (Supplementary Table 5). In addition to liquid-state ZABs, a flexible solid-state ZAB has been developed, utilizing FeCo-ADA as the cathode, zinc foil as the anode, and polyacrylamide (PAM) hydrogel as the electrolyte. The FeCo-ADA-based solid-state ZAB displays a peak power density of 170.3 mW cm−2 and a high open-circuit voltage of 1.582 V, outperforming the benchmark Pt/C + RuO2 solid-state ZAB (Fig. 6f and Supplementary Fig. 56). The solid-state ZAB with FeCo-ADA also exhibits long rechargeability (over 70 h at 10 mA cm−2), along with larger discharge voltage plateaus across a range of current densities from 1 to 10 mA cm−2 (Fig. 6g and Supplementary Fig. 57). Interestingly, the solid-state ZAB with FeCo-ADA demonstrates high cycle stability, with virtually unchanged voltage across a range of operational angles from 0 to 180 at 2 mA cm−2 (Supplementary Fig. 58). The FeCo-ADA based solid-state ZAB effectively powers a sensor, an LED, and a mobile phone even under bend conditions (Fig. 6h, Supplementary Figs. 59,60). The solid-state ZAB based on FeCo-ADA exhibits effective flexibility and stability, highlighting its significant potential for practical applications in wearable electronics. Performance of the rechargeable ZAB using FeCo-ADA as the air cathode. a Schematic illustration of the liquid-state ZAB. b Specific capacity of the battery assembled with catalyst. c Galvanostatic discharge plots of ZABs. d The power densities of the ZABs. e Galvanostatic charge-discharge curves of the ZABs at 10 mA cm−2 (inset: Magnified view of charge-discharge cycles). f Discharge polarization and corresponding power densities of solid-state ZABs. g Galvanostatic charge-discharge curves of the solid-state ZABs at 10 mA cm−2. h A sensor powered by a flexible ZAB. Discussion In summary, the axially coordinated FeCo-ADA with high-performance bifunctional electrocatalytic activity was successfully designed and synthesized by integrating COF and NG. In-situ XANE and Raman spectroscopy unravel that Fe functions as the primary active center in FeCo-ADA, with Co providing synergistic effects. The ZAB using FeCo-ADA delivers a high peak power density (464.5 mW cm−2) and long-term rechargeability (3710 hours at 10 mA cm−2). The theoretical investigations elucidates that the axial Fe-Co orbital coupling results in optimized orbital energy levels, higher Fe oxidation state, weakened oxygen intermediate binding, and boosted bifunctional activity. This study offers a strategy for the rational design and fabrication of axial DACs for efficient energy conversion. Methods Synthesis of Fe-COF Benzene-1,2,4,5-tetracarbonitrile (272.7 mg), anhydrous iron (III) chloride (202.7 mg), and 1,8-diazabicyclo[5.4.0]undec-7-ene (1.24 mL) were mixed in ethylene glycol (30 mL) and N,N-dimethyl acetamide (30 mL)27. Next, the resulted solution was reacted in a microwave for 45 minutes to form a suspension. Then, the suspension was washed using 1 M HCl, followed by drying overnight. Finally, the dried product was grinded to obtain fine powder. Synthesis of Co-NG Cobalt(II) acetate (50 mg), melamine (12 g) and L-alanine (2 g), and were uniformly mixed by ball-milling for 35 min. Subsequently, ethanol (10 mL) and hydrochloric acid (2 mL) was added. The mixture was further ball-milled. The product was dried at 60 °C overnight. The resultant powder was pyrolyzed under a flowing Ar atmosphere. Synthesis of Fe-COF + Co-NG mixed Separately, Fe-COF (20 mL) and Co-NG (20 mL) were sonicated in 5 mL ethylene glycol for 1 h. The suspensions were then combined, sonicated again for 1 h, and stirred overnight. Then the product was dried in a vacuum drying oven at 60 °C for 24 hours. Synthesis of FeCo-ADA, FeCo-BDO, FeCo-MeOH Methanol, ethylene glycol, and 1,4-butanediol served as solvents for fabricating FeCo-MeOH, FeCo-ADA, and FeCo-BDO, respectively. For each sample, Fe-COF (20 mL) and Co-NG (20 mL) were individually sonicated in 5 mL of the respective solvent for 1 h. The suspensions were then combined, sonicated for an additional hour, and stirred overnight. The resulting powders were pyrolyzed in a tubular furnace under flowing argon. The temperature was ramped from room temperature to 500 °C at 5 °C min−¹, held for 1 h, and then cooled to room temperature. Electrochemical measurements A standard pre-treatment protocol was employed to ensure a contamination-free and stable electrode-electrolyte interface. This involved purging the electrolyte with N2 for 30 minutes, followed by electrochemical conditioning via cyclic voltammetry. Then, the electrolyte was saturated with O2 ( ≥ 30 min purging) to establish the desired reaction environment before commencing the main electrochemical characterization. The subsequent cycling was conducted in O2-saturated electrolyte. The oxygen reduction reaction (ORR) was carried out utilizing a CHI-760E workstation, connected to a Pine Instrument Company rotating apparatus, in a conventional three-electrode setup in a 0.1 M KOH solution. A 5 mm diameter rotating disc electrode (RDE) and a 5.61 mm diameter rotating ring disc electrode (RRDE), supplied by Pine Research Instrumentation, were used for the measurement. A graphite rod was used as the counter electrode and a Hg/HgO electrode as the reference. The potentials were normalized to the reversible hydrogen electrode (RHE) using the Nernst equation: ERHE = EHg/HgO + 0.059 × pH + 0.098 V. In order to prepare the electrode, a solution was created by combining 5 mg of the catalyst with 485 μL of isopropanol, 485 μL of distilled water, and 30 μL of a 5% Nafion solution. This solution was then subjected to sonication in order to ensure that it was homogeneous. 13.74 µL of the aforementioned ink of the catalyst was applied to the surface of the RED. The electrolyte was saturated with pure oxygen for at least 30 minutes, and a continuous flow of oxygen was maintained throughout the testing period. The durability of the catalyst was evaluated through an iR-corrected i-t curve, obtained via linear sweep voltammetry (LSV) at a scan rate of 10 mV s−1. The double layer capacitance (Cdl) was calculated from CV result within the non-Faradaic potential window, using scanning rates that varied from 10 to 50 mV s−1. The electrochemically active surface area (ECSA) was determined based on the following equation: where Cdl is the double layer capacitance and CS is the specific capacitance of the catalyst on a flat surface (60 μF cm−2). The calculation of the double layer capacitance Cdl relies on the following equations. where Q is the charge and φ is the potential. To calculate the selectivity of catalytic during the ORR process, the LSV curve is recorded at varying speeds from 400 to 2025 rpm on the RDE. The number of electron transfers, n, is then determined following the Koutecky-Levich (K-L) equation. where J is the measure current density, JL is the limiting diffusion current density, and Jk is the kinetic current density. ω is the angular speed of the rotating disc (2πN, where N is the rotational rate), n is the number of electron transfers, F is Faraday constant (96485 C mol−1), C0 is the volume concentration of oxygen in the electrolyte (1.2×10−6 mol cm−3), D0 is the diffusion coefficient of O2 (1.9×10−5 cm2 s−1), and ν is the kinematic viscosity of the electrolyte (0.01 cm2 s−1). Electron transfer number, n, and the yield of H2O2 were calculated according to the following equation: where Id is the disc current, Ir is the ring current and N is the Pt ring collection current efficiency (N = 0.4). XAFS measurements The BL14W1 beamline of the Shanghai Synchrotron Radiation Facility (SSRF) and the commercial Laboratory-Based XAFS spectrometer (RapidXAFS 2 M, Anhui Absorption Spectroscopy Analysis Instrument Co., Ltd.) were employed to conduct X-ray absorption fine structure (XAFS) analyses of the Fe and Co. Post-monochromatization, the X-rays traversed the sample and were collected using a high-energy-resolution silicon drift detector (SDD) to obtain the X-ray intensity. The XAFS measurements were performed in transmission mode, with the absorption edge (Eo) position calibrated against standard metal foil samples of Fe and Co. The entire dataset was acquired within a unified temporal frame. Assembly of Liquid Zn-air batteries A liquid zinc-air battery was constructed with a thick zine foil as the anode, and FON-TM/NC-coated carbon-paper as the cathode. The electrolyte consisted of a 6.0 M KOH solution mixed with 0.2 M Zn(AC)2. AL and 2001A multichannel tester were employed for conducting tests on constant current charging and discharging cycles. Assembly of flexible solid-state Zn-air batteries The acrylamide (3 g) was dissolved in 6 M KOH (30 mL), followed by the addition of N, N-methylenebisacrylamide (100 mg) and sodium citrate (2.5 g) under vigorous stirring at room temperature. After complete dissolution, 3 ml of 5 wt % K2S2O8 solution was added to the mixture. The mixture was then polymerized at 60 °C for 8 h to yield the PAM-SC gel electrolyte. Following the polymerization process, the PAM-SC organohydrogel was immersed in a dimethyl sulfoxide (DMSO) solution. The flexible solid-state ZAB, which utilized PAM-SC as the electrolyte, underwent charge/discharge cycling stability tested at various temperatures. Each cycle consisted of 5 min charging at a specific current density, followed by 5 min discharging. Assembly of flexible self-powered sensing system The flexible self-powered sensing system was constructed by integrating a flexible zinc-air battery with an external piezoresistive sensor. The sensor, powered by the battery, was interfaced with a host computer via a USB data cable for real-time signal acquisition. To characterize the system’s response, a calibrated pressure stimulus was applied to the sensor using a metal letter, converting the mechanical input into an electrical signal that was subsequently processed and displayed on the computer. DFT calculations All the spin-polarized density functional theory (DFT) calculations were implemented in the Vienna Ab-initio Simulation Package (VASP) based on projected-augmented wave (PAW) method54. The electron exchange and correlation energy were described by the Perdew-Burke-Ernzerhof (PBE) functional of the generalized gradient approximation (GGA)55,56. The FeCo-ADA model was constructed based on experimental characterization, with systematic evaluation of a comprehensive set of potential configurations57,58. The atomic coordinates for all optimized structures are provided in Supplementary Data 1. A dispersion-corrected DFT (DFT-D3) framework, which has been proven to provide an adequate description of long-range van der Waals interactions in previous studies, was adopted59. Since the GGA fails to properly describe the strongly correlated interactions in transition metals, the DFT + U method, which accounts for on-site coulomb (U) and exchange (J) interactions, was employed to correct the localized 3 d electrons in fourth-period transition metals60. The correlation energy values (U-J) for the 3 d orbitals of Fe and Co were set to 3.29 eV and 3.42 eV, respectively61. The initial magnetic moments assigned to Fe and Co in FeCo-ADA were set to 4 and 4 μB, respectively. The cutoff energy was set to 450 eV and convergence criteria of energy and force were set to 10−5 eV and 0.01 eV/Å, respectively. A Г-centered Brillouin zone sampling strategy with 2 × 2 × 1 and 4 × 4 × 1 grids was adopted for structural relaxation and electronic structure calculations, respectively. A Vacuum space of 15 Å along the direction vertical to surface of the 2D slab is used to avoid the interactions between the adjacent layers. To inspect the thermodynamic stability, ab-initio molecular dynamics (AIMD) simulations are executed at 800 K with a time step of 3 fs for a total of 9 ps. The crystal orbital Hamilton populations (COHP) implementing LOBSTER code is used to analyze the Fe-O and Co-O bonds’ bonding nature for the adsorbed systems62. The calculation of the OER and ORR process is based on the computational hydrogen electrode (CHE) model proposed by Nørskov63. The free energies of reactants, intermediates, and products were calculated using the following formula: G = E + ZPE - TS, Where E represent total energy of species, ZPE, T and S is zero point energy correction term, temperature (298.15 K) and entropy. The ORR proceeds via two distinct mechanisms: the Adsorbate Evolution Mechanism (AEM) and the Oxide Pathway Mechanism (OPM)64. In alkaline conditions, the Gibbs free energy of ORR following the AEM pathway can be calculated through the following four-electron transfer steps: The Gibbs free energy of ORR following the OPM pathway can be calculated through the following four-electron transfer steps: Although DFT calculations offer valuable insights, the inherent limitations should be considered. The static nature of the computational models may not fully account for the dynamic evolution of the catalyst surface and the influence of solvation under realistic reaction conditions. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Transparent Peer Review file Source data Source data Acknowledgements This work was supported by the National Key R&D Program of China (2024YFB4607600) and National Natural Science Foundation of China (NSFC, Grant No. 52303359), Basic and Applied Basic Research Foundation of Guangdong Province (2023B1515040027, Young Top Talents Program 2021QN02C174), Open program of State Key Laboratory of Optoelectronic Materials and Technologies (OEMT-2024-KF-06), and Basic and Applied Basic Research Project of Guangzhou (2025A04J2379). We acknowledge the Hefei advanced computing center. Author contributions X.Y. (Xinru Yan), P.Z. and G.Y. conceived and designed the project. P.Z. directed and supervised the research. X.Y. (Xinru Yan) were responsible for the synthesis, characterizations and electrochemical measures. X.Y. (Xinru Yan), N.L., B.L., W.L., Z.L., X.G. and H.B.Y. performed all the XAFS experiments and operando XAFS experiments and the in-situ Raman experiments. X.Y. (Xiaoliang Yuan) and J.L. carried out the theoretical calculations. X.Y. (Xinru Yan), Y.H., Q.Z., S. C. and D.Y. participated in SEM and TEM characterization. N.L. carried the BET tests. L.Y. and X.X. contributed to the flexible ZAB. All authors discussed the results. Peer review Peer review information Nature Communications thanks Xijun Wang, Shenlong Zhao, 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 from the corresponding author upon request. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Xinru Yan, Xiaoliang Yuan. Contributor Information Dingshan Yu, Email: yudings@mail.sysu.edu.cn. Zhiping Zeng, Email: zzhip8@mail.sysu.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-025-66362-y. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Transparent Peer Review file Source data Data Availability Statement All data are available from the corresponding author upon request. Source data are provided with this paper.