Breaking scaling relations via Fe/Ni diatomic catalysts towards highly efficient electrocatalysts for rechargeable Na-air batteries

Abstract

Exploiting bifunctional and highly efficient electrocatalysts that can facilitate the reversible formation and decomposition of discharge products is crucial for rechargeable Na-air batteries (SABs). However, restricted by scaling relations, air cathodes based on single metal systems usually exhibit inferior activity and fail to achieve a stable and long cycle life of Na-air batteries. Dual single-atom catalysts (DACs) with two metal sites can circumvent the scaling relations and endow the single-atom catalysts with superior activity. Herein, we present a hollow carbon microsphere loaded with Ni and Fe single atoms (Ni-HCMs-Fe) as a demonstration of DAC air cathodes for SABs. Notably, Na-air batteries with Ni-HCMs-Fe as an air cathode can achieve a low overpotential gap of 530 mV, a high specific capacity of 5382.9 mAh g⁻¹, and an ultralong cycle life of over 450 cycles (1800 hours) with NaO₂ as the main discharge product. Due to the different adsorption energies between oxygenated intermediates and Ni/Fe sites, the Ni-HCMs-Fe catalysts can break through the scaling relations and display optimized binding ability towards intermediates, thus boosting the reversible formation and decomposition of NaO₂ in SABs. This work pioneers the use of DACs in SABs, paving the way for their practical applications.

---

Fangxi Xie

Fangxi Xie is currently an Associate Professor in the School of Chemical Engineering and Technology at Sun Yat-sen University. He obtained his PhD from the University of Adelaide under the supervision of Prof. Shizhang Qiao, and subsequently carried out postdoctoral research at both the University of Adelaide and the University of Melbourne. His research focuses on the design and synthesis of electrode materials for rechargeable batteries, as well as the development of operando and in situ characterization techniques for battery systems.

Introduction

Rechargeable Na-air batteries (SABs) are deemed promising candidates for energy conversion and storage systems by virtue of their high energy density (1602 or 1105 Wh kg⁻¹ based on the discharge products of Na₂O₂ or NaO₂, respectively) as well as the low cost and abundance of Na element.^1,2 Na-air batteries operate based on conversion reactions, which is different from the intercalation mechanism in Na-ion batteries. The overall electrochemical reactions for O₂ at the air cathode during discharge and charge are the oxygen reduction reaction (ORR) and oxygen evolution reaction (OER),respectively.^3,4 Considering that the OER and ORR are bidirectional reactions, an adsorption energy that is either too high or too low for oxygen-containing reaction intermediates can hinder the reversibility of SABs.⁵ Therefore, optimizing the adsorption energy of these intermediates is crucial for developing high-performance SABs. For a given site where reaction intermediates are adsorbed, the adsorption energies of specific intermediates have fixed values.^6,7 However, these values may not be optimal for both the OER and ORR. A representative example is the ORR, where achieving optimal catalytic performance requires strong adsorption of OOH* while maintaining weak adsorption of OH*.⁸ However, since both species bind to active sites through oxygen atoms, balancing these opposing adsorption strengths is inherently challenging. This limitation arises from the scaling relationship phenomenon, where the adsorption energies of related intermediates change in tandem at a given active site. As illustrated in Fig. 1a, this linear scaling relation restricts the ability to achieve optimized adsorption energies for reaction intermediates, ultimately limiting the development of high-performance SABs. Consequently, most reported SAB cathodes with single metal or metal oxide catalysts fail to break the linear scaling relationship, resulting in SABs with insufficient achievable capacities and limited cycle life (Fig. 1b).^9–18

Fig. 1 (a) Scaling relationship between ΔG_*OH and ΔG_*OOH for M@NC and rutile MO₂(M: Pt, Pd, Fe, Co, Ni, Cu…).⁸ FeNi DACs can break the linear relationship between ΔG_*OH and ΔG_*OOH.²⁴ (b) Discharge–charge overpotential and cyclic performance of Na-air batteries with different cathodes. Data obtained from ref. 9–18.

Atomically dispersed metal catalysts, known as single atom catalysts (SACs), are endowed with outstanding activity and selectivity due to their distinctive HOMO–LUMO gap, discrete energy level distribution and unique electronic structure.^19–22 This promising class of catalysts demonstrates superior catalytic performance across various applications. However, since the isolated atomic sites in SACs are based on a specific metal, SACs with a single-metal site exhibit a fixed adsorption mode.⁶ This mode is closely linked to the adsorption energy of multistep reaction intermediates. Consequently, SACs with a single metal are unable to independently control the adsorption of one adsorbate relative to another. Adsorption behaviours are highly correlated among different adsorbates, a phenomenon known as the adsorbate linear scaling relationship, as mentioned above.⁷ As a result, the performance of this promising class of catalysts is strictly constrained by the linear scaling relationship, leading to the limited reported application of SACs in SABs. A promising strategy to break the linear scaling relationship is to introduce another metal to form diatomic catalysts (DACs).²³ By introducing a second metal as a “dual site”, the two intermediates can be selectively adsorbed on different sites. This enables the independent adjustment of the interactions between the catalyst and each intermediate, effectively breaking the linear scaling relationship and endowing the catalysts with bifunctional ORR/OER activity (Fig. S1, SI).^8,24–26 For example, Han et al. demonstrated that introducing Co atoms into isolated Ni atom sites can overcome the unfavorable OH* adsorption on Ni sites, achieving the optimized adsorption of OH* and thereby efficiently promoting the ORR/OER process.²⁵ Similarly, Zhou et al. demonstrated that the charge transfer between Co and Fe can effectively reshape the electronic structure of the metal active sites and regulate the adsorption energy of reaction intermediates, greatly increasing the ORR/OER bifunctional activity.²⁶ Ma et al. reported a case where the different adsorption energies of Fe and Ni sites to OOH* and OH* can substantially improve the electrocatalytic performance for bidirectional reactions.²⁷ Therefore, selective combination of one metal with higher ORR activity and another metal with higher OER activity forming a DAC is an effective strategy to break the scaling relation, providing a promising class of catalysts for the development of high-performance SABs. However, even though such a promising strategy has been applied in Zn-air batteries, it remains scarcely applied in SABs.

Herein, we demonstrate that introducing additional metal sites to SACs to form DACs is an effective strategy for developing high-performance SABs. In this study, we utilize Ni and Fe, which have different adsorption energies for oxygen-containing reaction intermediates, to break the linear scaling relationship by fabricating a diatomic catalyst (Fig. 1a).²⁴ This catalyst consists of two types of single atoms (Ni and Fe) anchored on hollow carbon microspheres (Ni-HCMs-Fe). The synergistic effect of Fe and Ni species can endow the Ni-HCMs-Fe catalyst with optimized binding ability towards intermediates, which can effectively break the stubborn adsorbate scaling relationship and achieve superior ORR and OER activities. Benefitting from these advantages, Na-air batteries assembled with the Ni-HCMs-Fe cathode can achieve a high specific capacity of 5382.9 mAh g⁻¹, a low charge overpotential of 530 mV, and long cycling stability over 450 cycles with NaO₂ as the main discharge product. These results demonstrate the superior performance of Ni-HCMs-Fe cathodes over their carbon-based and monometallic counterparts. This work showcases the promising potential of diatomic catalysts in Na-air batteries, paving the way for their broader implementation.

Results and discussion

Fig. 2a illustrates the detailed synthesis procedure of Ni-HCMs-Fe: (1) melamine cyanuric acid complex (MCA) was prepared by mixing melamine and cyanuric acid in dimethyl sulfoxide (DMSO) solution.²⁸ Scanning electron microscopy (SEM) images (Fig. S2, SI) show that the MCA complex microsphere composed of numerous 3D plates presents a flower-like appearance with a uniform size of 1.5–2.0 μm. The X-ray diffraction (XRD) pattern (Fig. S3, SI) can be perfectly indexed to the pure MCA. Since pure MCA will be decomposed into gas containing C, N, and O and totally evaporate above 400 °C, it can serve as a sacrificial template and nitrogen source in the formation of Ni-HCMs-Fe.²⁹ (2) A Ni²⁺ chelating polydopamine (PDA-Ni) nanolayer can be uniformly coated on the surface of MCA to form core–shell MCA@PDA-Ni. (3) An Fe³⁺ chelating polydopamine (PDA-Fe) nanolayer can be subsequently coated on the surface of MCA@PDA-Ni to obtain MCA@PDA-Ni@PDA-Fe. The morphologies (Fig. S4, SI) and XRD patterns (Fig. S5, SI) of MCA@PDA-Ni and MCA@PDA-Ni@PDA-Fe are almost identical to those of pristine MCA. Such nearly identical morphology demonstrates the uniform PDA coating on MCA, which can be attributed to the exceptional chelating ability with diverse metal ions and strong adhesion to substrates with the complex morphology of PDA. Besides, the PDA nanolayer skeleton can be converted into N/O co-doped carbon nanosheets during the high-temperature pyrolysis. The N/O atoms on the carbon nanosheets coordinated with Ni or Fe atoms can efficiently prevent the aggregation of Ni or Fe.²⁵ Ni-HCMs-Fe with isolated Ni and Fe atoms anchored on the carbon microspheres can be obtained after pyrolyzing MCA@PDA-Ni@PDA-Fe precursors at 800 °C.

Fig. 2 (a) Synthetic route of Ni-HCMs-Fe. (b) SEM, (c) aberration-corrected HAADF STEM, and (d) elemental mapping images of C, N, O, Ni and Fe (scale bar is 500 nm) of Ni-HCMs-Fe. Aberration-corrected HAADF STEM images of (e) Ni-HCMs and (f) HCMs-Fe. (g) XRD patterns of Ni-HCMs, HCMs-Fe, and Ni-HCMs-Fe.

The morphology and structure of Ni-HCMs-Fe were characterized by SEM and transmission electron microscopy (TEM). As shown in Fig. 2b, Ni-HCMs-Fe can retain its original flower-like structure assembled by nanosheets. Due to the complete decomposition of MCA microspheres and transformation of PDA nanolayers, the plate-like subunits of MCA@PDA can be transformed into ultrathin nanosheets. The Ni-HCMs-Fe sample, composed of numerous ultrathin nanosheets, presents a nest-like appearance with a diameter ranging from 1.0 to 1.5 μm. During the thermal pyrolysis, the decomposition of MCA reduces the overall size of microspheres. Meanwhile, the rapid decomposition of the dense MCA core without the PDA coating produces an obvious hollow interior, forming a unique carbon hollow structure assembled from nanosheets. With the assistance of electronegative N and O atoms, the Ni and Fe atoms can be uniformly dispersed on the carbon framework. TEM images in Fig. S6 (SI) verify that the hollow structure of each microsphere is composed of numerous ultrathin nanosheets. An aberration-corrected high-angle annular dark-field scanning transmission electron microscope (HAADF-STEM) image (Fig. 2c) clearly shows a large number of bright dots (Ni or Fe SAs, as marked by the purple circle) are atomically dispersed on the carbon nanosheets, indicating the successful synthesis of Ni and Fe SAs. Elemental mapping images (Fig. 2d) show the uniform distribution of C, N, O, Ni, and Fe elements throughout the microsphere. The loading mass of Ni and Fe in Ni-HCMs-Fe is about 2.4 and 2.5 wt%, respectively, determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES) (Table S1, SI). The Ni-HCMs (Fig. S7, SI), HCMs-Fe (Fig. S8, SI), and HCMs (Fig. S9, SI) samples were synthesized as control groups. The HAADF-STEM results certify that the Ni (blue circles, Fig. 2e) and Fe species (red circles, Fig. 2f) are atomically dispersed on the surface of the carbon microspheres. No diffraction peaks related to metal phases (pure metal, alloy, and oxide) are detected in their XRD patterns (Fig. 2g and S10, SI). The above control experiments strongly prove that atomically dispersed Ni and Fe sites anchored on the microspheres can be finally obtained. Raman spectra of HCMs, Fe-HCMs, HCMs-Ni, and Ni-HCMs-Fe in Fig. S11 (SI) show two characteristic peaks at 1350 cm⁻¹ (D-band: defective or disordered carbon) and 1570 cm⁻¹ (G-band: graphitic carbon).³⁰ The I_D/I_G values of Ni-HCMs (1.02), HCMs-Fe (1.04), and Ni-HCMs-Fe (1.05) are slightly higher than that of HCMs (1.00), suggesting that Fe or Ni doping can increase the disorder degree of the carbon matrix and generate more defective sites in the carbon skeleton.¹¹ Nitrogen adsorption–desorption isotherms (Fig. S12a, SI) were measured to analyze the porous structure and properties of these catalysts. All of them exhibit type-IV isotherm characteristics, indicating the existence of rich microporous and mesoporous structures. Pore size distributions (Fig. S12b, SI) indicate that the pore size of these catalysts is broadly distributed in the range of 2–50 nm, exhibiting a hierarchical porous structure.

To further disclose the chemical environment of Ni and Fe atoms in Ni-HCMs-Fe, Ni and Fe K-edge X-ray absorption spectroscopy (XAS) of Ni-HCMs-Fe was performed with Ni foil, NiO, Fe foil, FeO, and Fe₂O₃ as references. The X-ray absorption near edge structure (XANES) spectra of the Ni K-edge (Fig. 3a) demonstrate that the edge position of Ni is between Ni and NiO, indicating that the mean valence state of Ni is close to +2.³¹ The k³-weighted Fourier-transform (FT) of the extended X-ray absorption fine structure (EXAFS) of Ni shows a main peak near 1.42 Å (Fig. 3b), which corresponds to the Ni–N/O bonds.³² No peaks associated with Ni–Ni paths are detected at about 2.2 Å and larger bond lengths, suggesting that Ni atoms are atomically dispersed in Ni-HCMs-Fe. The coordination configuration of Ni with N/O is further calculated by quantitative EXAFS fitting (Fig. 3c). The appropriate fitting results show that the coordination number of the central Ni atom bonding to the adjacent N and O atoms is approximately 4 and 1, respectively (Table S2, SI). The above analysis reveals that the coordination configuration of Ni sites is Ni–N₄O. Similarly, the XANES spectra of the Fe K-edge (Fig. 3g) show an adsorption energy of Ni-HCMs-Fe located between FeO and Fe₂O₃, indicating the mean valence state of Fe is between +2 and +3.³¹ The FT-EXAFS spectrum of Fe exhibits a main peak near 1.45 Å (Fig. 3h), which can be attributed to the Fe–N/O bonds.³² No Fe–Fe peaks are detected at about 2.2 Å or longer bond lengths, demonstrating that Fe atoms are atomically dispersed in Ni-HCMs-Fe. The fitting results confirm that the coordination configuration of Fe sites is Fe–N₄O (Fig. 3i and Table S3, SI). Wavelet-transform (WT) plots of Ni and Fe are focused at about k = 4.60 Å⁻¹ and r = 1.45 Å (Fig. 3d–f, j–l and S13, SI). No substantial Fe foil (Fe–Fe) and Ni foil (Ni–Ni) signals can be observed, further confirming their atomically isolated nature.^33,34 X-ray photoelectron spectroscopy (XPS) was conducted to analyse the surface composition and elemental state of the catalysts. High-resolution N 1s spectra of Ni-HCMs-Fe, Ni-HCMs, and HCMs-Fe (Fig. S14, SI) can be deconvoluted into pyridinic-N (398.3 eV), M–N (399.6 eV), pyrrolic-N (400.3 eV), graphitic-N (400.9 eV), and oxidized-N (403.2 eV). The O 1s spectrum (Fig. S15, SI) shows three bands, corresponding to M–O (530.8 eV), O–H (531.8 eV), and O=C–O (533.5 eV), respectively. Notably, compared with pure HCMs, the additional peaks at 399.6 eV in the N 1s spectrum and 530.8 eV in the O 1s spectrum indicate the presence of metal-nitrogen bonds (Fe–N and Ni–N) and metal–oxygen bonds (Fe–O and Ni–O) in the Ni-HCMs-Fe.³⁵ Two peaks of Ni (2p_3/2: 855.0 eV and 2p_1/2: 871.9 eV) (Fig. S16, SI) and Fe (2p_3/2: eV, 2p_1/2: 723.8 eV) (Fig. S17, SI) verify the dominant oxidation states of Ni and Fe single atoms, which is caused by N/O-coordination.³⁶

Fig. 3 (a) Ni K-edge XANES spectra, (b) FT-EXAFS spectra, and (c) EXAFS fitting curves of Ni in Ni-HCMs-Fe. WT of the Ni K-edge of (d) Ni foil, (e) NiO, and (f) Ni-HCMs-Fe. (g) Fe K-edge XANES spectra, (h) FT-EXAFS spectra, and (i) EXAFS fitting curves of Fe in Ni-HCMs-Fe. WT of the Fe K-edge of (j) Fe foil, (k) Fe₂O₃, and (l) Ni-HCMs-Fe.

To compare the electrochemical performance of DACs and SACs, air cathodes composed of HCMs, Ni-HCMs, HCMs-Fe, and Ni-HCMs-Fe were fabricated onto nickel foam and tested as air cathodes in Na-air batteries. Differential capacity (dQ/dV) analysis of the initial discharge–charge profiles was conducted to evaluate the electrochemical catalytic activity of different cathodes (Fig. 4a). Their relevant cathodic/anodic peak positions are listed in Fig. 4b. The HCMs and HCMs-Fe cathodes exhibit a single cathodic peak at about 2.10 and 2.27 V, respectively. Meanwhile, the Ni-HCMs and Ni-HCMs-Fe cathodes display pairs of cathodic/anodic peaks at 2.17/2.77 V and 2.20/2.73 V, respectively. The presence of redox peaks during the cathodic and anodic processes indicates the formation and decomposition of discharge products, respectively. The absence of an anodic peak in the dQ/dV curves of the HCMs and HCMs-Fe cathodes indicates they only exhibit ORR activity in Na-air batteries. In contrast, the Ni-HCMs and Ni-HCMs-Fe cathodes display both ORR and OER activities. The onset potential gap between the cathodic and anodic peaks of the Ni-HCMs-Fe cathode is just 530 mV, suggesting its excellent electrochemical kinetic performance in Na-air batteries. The initial discharge–charge curves of SABs with different cathodes were investigated within the voltage range of 1.8–3.8 V at 100 mA g⁻¹ at room temperature (Fig. 4c and d). A Na-air battery with the Ni-HCMs-Fe cathode can deliver a high specific discharge capacity of 5382.9 mAh g⁻¹ and a charge capacity of 4259.3 mAh g⁻¹. Both the discharge and charge capacities of Ni-HCMs-Fe far exceed those of HCMs (1574.5/931.0 mAh g⁻¹), Ni-HCMs (2531.2/1735.8 mAh g⁻¹), and HCMs-Fe (4178.5/1770.8 mAh g⁻¹). The significantly increased discharge and charge capacities of the Ni-HCMs-Fe cathode indicate its superior ORR and OER activities over its monometallic-based and carbon-based analogs. Additionally, to evaluate the durability and stability of air cathodes and prevent the overgrowth of discharge products and parasitic reactions, the cycling performance of SABs was tested with a fixed capacity of 500 mAh g⁻¹ at a current density of 250 mA g⁻¹. Notably, the Ni-HCMs-Fe cathode can steadily operate over 450 cycles (about 1800 h) without obvious performance degradation (Fig. 4e). In contrast, Na-air batteries with HCMs, Ni-HCMs, and HCMs-Fe cathodes can only run for 25 cycles (about 100 h), 40 cycles (about 160 h), and 30 cycles (about 120 h), respectively, before the discharge–charge potential rapidly decreases. The significant decline in the discharge–charge plateaus of the Ni-HCMs-Fe cathodes especially after 700 hours (Fig. S18, SI) might be associated with the morphology of discharge products on the cathode.³⁷ Considering that the Ni foam may affect the performance of batteries due to the ORR/OER activity of Ni, control tests using pure Ni foam as the cathode and carbon paper as the substrate have been conducted (Fig. S19, SI). The initial discharge–charge profile of pure Ni foam as an air cathode in Fig. S19a shows that the pure Ni foam contributes limited capacity to the discharge/charge process. Additionally, both the discharge/charge capacity (Fig. S19b) and cyclic performance (Fig. S19c) of the air cathodes loaded on carbon paper are inferior to those of the air cathodes loaded on Ni foam. The excellent performance of air cathodes loaded on Ni foam might be attributed to the porous structure of Ni foam. Moreover, the energy efficiency (terminal discharge voltage divided by terminal charge voltage) of the Ni-HCMs-Fe cathode remained at over 55% after 450 cycles (Fig. 4f), far exceeding that of HCMs (40% at 25 cycles), Ni-HCMs (48% at 40 cycles), and HCMs-Fe (42% at 30 cycles) as air cathodes. In addition, regular fluctuations can be observed in the discharge–charge plateaus of these cathodes between cycles 60 and 300. These fluctuations in battery performance are attributed to diurnal temperature variations between morning and night. The increased specific capacity, extended cycle life and superior energy efficiency of the Ni-HCMs-Fe cathode over its carbon-based and monometallic analogs indicate that the Ni-HCMs-Fe cathode can exhibit excellent performance as an air cathode for Na-air batteries.

Fig. 4 (a) dQ/dV (capacity increment) curves of the initial discharge–charge profile and (b) cathodic/anodic peaks of HCMs, Ni-HCMs, HCMs-Fe, and Ni-HCMs-Fe cathodes in the voltage range of 1.8–3.8 V at a current density of 100 mA g⁻¹. (c) Initial discharge–charge curves and (d) initial discharge/charge capacities of different cathodes at the current density of 100 mA g⁻¹. (e) Cyclic performance and (f) energy efficiency vs. cycle number of SABs with different cathodes at the current density of 250 mA g⁻¹ with a fixed capacity of 500 mAh g⁻¹.

To reveal the origin of the superior performance of the Ni-HCMs-Fe cathode over its carbon-based and monometallic analogs, ex situ SEM, XRD, Raman, and Fourier transform infrared spectroscopy (FTIR) were performed to examine the detailed structure and composition of discharge products and the air cathodes after the first recharge process to measure their reversible formation and decomposition. SEM images show evident differences in the surface coverage of discharged products on these different cathodes. Specifically, the HCMs cathode exhibits only a few particles and film-like layers on its surface (Fig. S20a, SI), suggesting its limited ORR activity. In contrast, the surfaces of Ni-HCMs (Fig. S20b, SI), HCMs-Fe (Fig. S20c, SI), and Ni-HCMs-Fe (Fig. S20d, SI) cathodes are densely covered with particles and film-like layers. The incremental coverage demonstrated enhanced ORR activity for metal single atom sites on the catalysts, as evidenced by the presence of a substantial amount of discharge products. After recharging, the discharge products on these cathodes can be decomposed (Fig. S20e–h, SI). XRD analysis reveals that the discharge products of these cathodes are crystalline Na₂O₂ and NaO₂ (Fig. S21, SI). However, the discharge products show a weak intensity of the characteristic Na₂O₂ and NaO₂ peaks. Considering the complex diffraction signals of these cathodes, the detailed compositions of the discharge products were further examined by Raman, FTIR, and XPS spectroscopies. Raman spectra show the formation and decomposition of NaO₂ (1140 cm⁻¹) during the discharging and charging process (Fig. S22 and S23, SI). However, the discharge products of HCMs and HCMs-Fe cathodes exhibit a weak intensity of the characteristic NaO₂ peak compared to that of Ni-HCMs and Ni-HCMs-Fe cathodes. No Raman peak related to the Na₂O₂ phase (740 cm⁻¹) was detected, probably due to its weak Raman signal. Raman and IR spectroscopies are complementary analytical techniques, and as a result, a compound with a weak Raman signal may show strong absorbance in the IR.³⁸ Accordingly, FTIR spectra of these discharged cathodes were also recorded. Compared with the FTIR spectra of reference NaClO₄, Na₂CO₃, and Na₂O₂ (Fig. S24, S I) and the pristine cathodes (Fig. S25, S I), the FTIR spectra of these discharged cathodes (Fig. S26, SI) present two characteristic peaks of Na₂O₂ at 1440 and 1600 cm⁻¹, providing further evidence for the formation of Na₂O₂ during the discharging process.⁵ After the recharging process, the peak of Na₂O₂ disappeared, suggesting the decomposition of Na₂O₂ (Fig. S27, SI). Overall, the Raman spectra confirm the presence of NaO₂, while the FTIR spectra reveal the existence of Na₂O₂. Based on the above results, it can be concluded that the discharge product in these cathodes is the mixture of Na₂O₂ and NaO₂. Afterwards, XPS analysis was performed to examine the detailed composition of discharge products. Specifically, the O 1s spectra of these discharged cathodes display three peaks at 530.9 eV (Na₂O₂), 532.9 eV (NaO₂), and 536.0 eV (Na auger), respectively (Fig. S28, SI).³⁹ Notably, the percentage of NaO₂ (Fig. 5a), calculated from the O 1s spectra in Fig. S28, follows the order: Ni-HCMs-Fe (62.0%) > Ni-HCMs (56.3%) > HCMs-Fe (39.4%) > HCMs (34.8%). Conversely, the percentage of Na₂O₂ follows the order: HCMs (65.2%) > HCMs-Fe (60.6%) > Ni-HCMs (43.7%) > Ni-HCMs-Fe (38.0%). The highest percentage of NaO₂ in the Ni-HCMs-Fe cathode suggests that Ni-HCMs-Fe can more effectively promote the formation of NaO₂ than its carbon-based and monometallic analogs. Additionally, the high percentage of NaO₂ demonstrates that the discharge products on the Ni-HCMs-Fe cathode are mainly kinetically favourable NaO₂, resulting in a lower charge overpotential and ultra-long cycle life of the Ni-HCMs-Fe cathode. Consequently, the Ni-HCMs-Fe cathode with dual atom catalytic sites exhibits substantially improved performance compared to other reported air cathodes (Fig. 5b). Our results demonstrate that constructing DACs as air cathodes is a promising strategy to develop high-performance SABs.

Fig. 5 (a) Percentage of NaO₂ and Na₂O₂ calculated on O 1s spectra of different discharged cathodes. (b) Discharge–charge overpotential and cyclic performance of Na-air batteries with different cathodes. Data obtained from (ref. 4, 9–18, 40–42). (c) Binding energy of O₂ and NaO₂ on Ni–N₄O, Fe–N₄O and Fe–Ni sites. Free energy diagram of the formation and decomposition of NaO₂ with the most stable structure of intermediates on (d) Ni–N₄O, (e) Fe–N₄O and (f) Fe–Ni surfaces.

Therefore, to gain theoretical insight into the detailed redox mechanisms and explain the enhanced electrochemical performance of the Ni-HCMs-Fe cathode, density functional theory (DFT) calculations were first employed to estimate the binding energy of O₂ and NaO₂ on Ni–N₄O, Fe–N₄O and Fe–Ni sites, as shown in Fig. 5c. From the perspective of O₂ adsorption, DFT results reveal that the binding energy of O₂ on the Fe–Ni sites is −0.79 eV, much lower than that of Ni–N₄O (−0.12 eV) and Fe–N₄O (−0.21 eV). This suggests that the Ni–Fe sites exhibit a stronger adsorption towards O₂ than Ni–N₄O and Fe–N₄O sites. From the perspective of NaO₂ adsorption, DFT results reveal the binding energy of NaO₂ on the Fe–Ni sites is −1.68 eV, much lower than that of Ni–N₄O (−0.91 eV) and Fe–N₄O sites (−1.20 eV). This indicates that the Ni–Fe sites exhibit a stronger adsorption of NaO₂ than Ni–N₄O and Fe–N₄O sites. The ORR process in SABs is complicated, which can be split into four parts:¹⁴.

(a) 4(Na⁺ + e⁻) + 2O₂ → 3(Na⁺ + e⁻) + O₂ + NaO₂*

(b) 3(Na⁺ + e⁻) + O₂ + NaO₂* → 2(Na⁺ + e⁻) + 2NaO₂*

(c) 2(Na⁺ + e⁻) + O₂ + 2NaO₂* → (Na⁺ + e⁻)+ 3NaO₂*

(d) (Na⁺ + e⁻) + O₂ +3NaO₂* → 4NaO₂

The pristine configuration of the Ni–Fe sites, along with the four configurations representing the four-step growth/desorption mechanism of NaO₂ and its intermediates adsorbed on the Ni–Fe sites, are presented in Fig. S29 (SI). The inverse reactions represent the OER process and * indicates the intermediates are adsorbed on the surface of sites. Fig. 5d–f shows the calculated free energy for the ORR/OER on Ni–N₄O, Fe–N₄O and Ni–Fe sites at equilibrium potential (U_eq), discharge potential (U_dis) and charge potential (U_ch). The calculated overpotential decreases in the order of Ni–N₄O (2.55 V) > Fe–N₄O (1.85 V) > Ni–Fe (1.54 V), indicating that Ni–Fe sites have better catalytic ability than Ni–N₄O and Fe–N₄O sites. Compared to SACs with single-metal catalytic sites (e.g., HCMs-Fe with Fe–N₄O sites, Ni-HCMs with Ni–N₄O sites) that are constrained by the scaling relationship, DACs (Ni-HCMs-Fe with Ni–N₄O and Fe–N₄O sites) can simultaneously achieve an optimized adsorption mode for O₂, NaO₂ and its intermediates. Consequently, by leveraging the OER-favorable Ni–N₄O sites and the ORR-favorable Fe–N₄O sites, the Ni-HCMs-Fe cathode can break the linear scaling relationships, thereby exhibiting excellent performance in SABs.

Experimental

Materials

Melamine (99%), cyanuric acid (98%), ammonia (AR), and DMSO (GC) were bought from Macklin. Ni(NO₃)₂·6H₂O (99%), Fe(NO₃)₃·9H₂O (AR), NaClO₄ (AR), N-methylpyrrolidine (NMP) (AR), and 1,2-dimethoxyethane (DME) (AR) were purchased from Aladdin. Dopamine hydrochloride (98%) was obtained from Acmec. Polyvinyl pyrrolidone (M_w = 40 k) and poly(vinylidene fluoride) (PVDF) (M_w = 400 k) were bought from Sigma-Aldrich. Na (99.5%) was acquired from DAMAO. Super P carbon (K90) was received from Rhawn. Ni foam (210 × 297 × 1.0 mm³) was obtained from N-buliv.

Synthesis of MCA

MCA microspheres were synthesized based on a previous report.²⁸ In a typical synthesis, 3.0 g of melamine and 3.06 g of cyanuric acid were dissolved in 120 mL and 60 mL DMSO, respectively. After complete dissolution, the solutions were mixed together and stirred for 10 min to give a white precipitate. Subsequently, the solid product was collected by centrifugation and washed with ethanol three times. The resulting sample was dried at 65 °C.

Synthesis of Ni-HCMs-Fe

1 g of MCA, 24 mg of Ni(NO₃)₂·6H₂O, 240 mg of dopamine hydrochloride, and 400 μL ammonia were sequentially added into 160 mL deionized (DI) water and stirred for 24 h.²⁹ The solid product was collected by centrifugation and washed with deionized (DI) water to obtain MCA@PDA-Ni. Then, the MCA@PDA-Ni sample, 24 mg of Fe(NO₃)₃·9H₂O, 240 mg of dopamine hydrochloride, and 400 μL ammonia were sequentially added into 160 mL DI water and stirred for 24 h. The solid product was collected by centrifugation and washed with DI water to obtain MCA@PDA-Ni@PDA-Fe. The MCA@PDA-Ni@PDA-Fe sample was calcined at 800 °C for 2 h in a N₂ atmosphere with a heating rate of 2 °C min⁻¹. The final carbonized product was denoted as Ni-HCMs-Fe.

Synthesis of HCMs

1 g of MCA, 240 mg of dopamine hydrochloride, and 400 μL ammonia were sequentially added into 160 mL DI water and stirred for 24 h. After being centrifuged and washed with DI water three times, the MCA@PDA powders were prepared. Then the MCA@PDA sample was carbonized at 800 °C as above and denoted as HCMs.

Synthesis of Ni-HCMs

The synthetic procedure is similar to that of Ni-HCMs-Fe. The obtained MCA@PDA-Ni, 240 mg of dopamine hydrochloride, and 400 μL ammonia were sequentially added into 160 mL DI water and stirred for 24 h. After being centrifuged and washed with DI water three times, the MCA@PDA-Ni@PDA powders were prepared. Then the powders were pyrolyzed at 800 °C.

Synthesis of HCMs-Fe

The synthetic procedure is similar to that of Ni-HCMs-Fe. The obtained MCA@PDA, 24 mg of Fe(NO₃)₃·9H₂O, 240 mg of dopamine hydrochloride, and 400 μL ammonia were sequentially added into 160 mL DI water and stirred for 24 h. After being centrifuged and washed with DI water three times, the MCA@PDA@PDA-Fe powders were prepared. Then the sample was carbonized at 800 °C. The obtained sample was denoted as HCMs-Fe.

Electrochemical characterization

HCMs, Ni-HCMs, HCMs-Fe, and Ni-HCMs-Fe cathodes were prepared by mixing 80 wt% active materials, 10 wt% Super P carbon and 10 wt% PVDF (NMP solvent) binder. The slurry was then cast onto Ni foam (1 × 1 cm²), followed by drying at 120 °C under vacuum for 12 h. A conventional two-electrode battery was constructed in a dry air-filled glovebox with the samples on Ni foam as the cathode and one sheet of high-purity sodium foil as the anode.⁴³ The electrolyte was composed of 1 M NaClO₄ in 1,2-dimethoxyethane (DME). Rate performance and cyclic tests were carried out on a LANHE CT3001A battery testing system.

Characterization

The morphology, structure, and chemical composition of samples were characterized using a scanning electron microscope (SEM, ZEISS Gemini SEM500) and transmission electron microscope (FEI Tecnai G2 F30). The phase structure of samples was tested using a powder X-ray diffractometer (Rigaku Ultima IV XRD with Cu Kα radiation (λ = 1.54056 Å)). An X-ray photoelectron spectrometer (XPS, Thermo Fisher/ESCALAB QXi) was used to analyze the structural information of the sample. X-ray absorption spectroscopy spectra were collected in the Australian Synchrotron. Raman spectra were recorded using an Invia reflex Raman microscopy system (RENISHAW). Fourier transform infrared spectroscopy (FTIR) was carried out on a Nicolet iS50 RaptIR (Thermo Scientific). The composition of samples was analyzed by ICP (PerkinElmer Optima 8300). Ex situ SEM, XRD, XPS, Raman and FTIR measurements were collected from air electrodes after an initial discharge and charge process. The cells were disassembled in an Ar-filled glovebox and the air electrodes were rinsed with DME several times to remove the residual sodium salts. To avoid any exposure of the air electrodes to oxygen and water, they were rapidly transferred into the chambers for characterization.³⁷

Theoretical calculations

Spin-polarized electronic structure calculations were performed using the plane-wave basis set approach as implemented in the Vienna ab initio simulation package (VASP). The projector augmented wave (PAW) method was used to represent the ion-core electron interactions. The valence electrons were represented with a plane wave basis set with an energy cutoff of 450 eV. Electronic exchange and correlation were described with the Perdew–Burke–Ernzerhof (PBE) functional. The DFT-D3 method was used to treat the van der Waals interaction. A 2 × 2 × 1 Monkhorst–Pack scheme was used to generate the k-point grid for the modeled surfaces. The convergence criteria for the self-consistent electronic structure and geometry were set to 10⁻⁵ eV and 0.05 eV Å⁻¹, respectively.^44,45

Conclusions

In summary, we reported a pioneering demonstration of utilizing dual single-atom catalysts (DACs) with atomically dispersed Ni and Fe species as effective air cathodes for non-aqueous Na-air batteries. The unique hierarchical hollow and porous structures of the catalysts as well as highly exposed Ni and Fe sites contributed to the outstanding performance of Na-air batteries. Specifically, the Ni-HCMs-Fe cathode assembled SABs delivered a low overpotential of about 530 mV, a high specific capacity of 5382.9 mAh g⁻¹, and an ultra-long cycle stability over 450 cycles (1800 h). We conducted various ex situ spectroscopy techniques and DFT calculations to unveil the origins of the superior performance of Ni-HCMs-Fe over its carbon-based and monometallic counterparts. These findings reveal that the Fe and Ni species on the catalysts can optimize the interaction of O₂, NaO₂ and its intermediates on the active centers and thus break the scaling relationship between the adsorption energy of reaction intermediates, which leads to the significantly increased discharge capacity and extended cycle life with reduced overpotential of Na-air batteries. This work provides a novel insight into the rational design of air cathodes for non-aqueous Na-air batteries, paving the way for their practical application.

Author contributions

Wenwen Yin: writing – original draft, conceptualization, data curation, methodology, validation. Jiawei Ma: data curation, validation. Yanyan Li: data curation, validation. Qing Cheng: writing – review & editing. Bernt Johannessen: data curation, writing – review & editing, resources. Fangxi Xie: writing – original draft, writing – review & editing, resources, funding acquisition. Mingmei Wu: writing – review & editing, resources, funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the SI. Additional data will be made available on request.

Supporting information: scaling relationship between ΔG_*OH and ΔG_*OOH of M@NC and rutile MO₂ (M: Pt, Pd, Fe, Co, Ni, Cu…), SEM/TEM images and XRD/Raman/N₂ adsorption–desorption isothems/XPS spectra of catalysts, WT-EXAFS of Fe K-edge of FeO, electrochemical performance of SABs with different cathodes, SEM/XRD/Raman/FTIR/XPS spectra of discharged/charged cathodes, growth/desorption mechanism of NaO₂ on Ni-HCMs-Fe catalysts and tables of Ni or Fe contents of catalysts tested by ICP. See DOI: https://doi.org/10.1039/d5ta03806e.

Acknowledgements

This work was financially supported by the China Postdoctoral Science Foundation (Grant No. 2021M703669), the Project of National Natural Science Foundation of China (No. 22405297), the Talent Recruitment Project of Guangdong (No. 2023QN10C330), the joint Project of National Natural Science Foundation of China (NSFC) and Guangdong Province (No. U1801251), and the Science and Technology Planning Project of Guangzhou City for International Cooperation Program (No. 201704030020). Part of this research was undertaken on the MEX-1 and XAS beamlines at the Australian Synchrotron, part of ANSTO. B.J. acknowledges an Honorary Professorial Fellowship at the University of Wollongong.

Notes and references

1. B. Xu, D. Zhang, C. Peng, F. Liang, H. Zhao, B. Yang, D. Xue and Y. Lei, Adv. Energy Mater., 2023, 13, 2302325.
2. X. He, Y. X. Ni, Y. X. Li, H. X. Sun, Y. Lu, H. X. Li, Z. H. Yan, K. Zhang and J. Chen, Adv. Mater., 2022, 34, 2106565.
3. X. T. Lin, Q. Sun, J. T. Kim, X. F. Li, J. J. Zhang and X. L. Sun, Nano Energy, 2023, 112, 10846.
4. W. W. Yin, X. Zhi, Y. Y. Li, J. W. Ma, B. Johannessen, F. X. Xie and M. M. Wu, Small, 2025, 24, 2500109.
5. Q. Sun, J. Liu, X. Li, B. Wang, H. Yadegari, A. Lushington, M. N. Banis, Y. Zhao, W. Xiao, N. Chen, J. Wang, T. Sham and X. Sun, Adv. Funct. Mater., 2017, 27, 1606662.
6. A. Khorshidi, J. Violet, J. Hashemi and A. A. Peterson, Nat. Catal., 2018, 1, 263.
7. Z. Seh, J. Kibsgaard, C. F. Dickens, I. Chorkendorff, J. K. Nørskov and T. F. Jaramillo, Science, 2017, 355, 146.
8. C. Fang, J. Zhou, L. Zhang, W. Wan, Y. Ding and X. Sun, Nat. Commun., 2023, 14, 4449.
9. F. Wu, Y. Xing, J. Lai, X. Zhang, Y. Ye, J. Qian, L. Li and R. Chen, Adv. Funct. Mater., 2017, 27, 1700632.
10. B. Sun, K. Kretschmer, X. Xie, P. Munroe, Z. Peng and G. Wang, Adv. Mater., 2017, 29, 1606816.
11. J. Ma, N. Li, Q. Zhang, X. Zhang, J. Wang, K. Li, X. Hao and J. Yan, Energy Environ. Sci., 2018, 11, 2833.
12. X. Bi, R. Wang, Y. Yuan, D. Zhang, T. Zhang, L. Ma, T. Wu, R. Shahbazian-Yassar, K. Amine and J. Lu, Nano Lett., 2020, 20, 4681.
13. X. Wang, X. Zhang, M. Tang, Q. Shen and J. Ma, J. Alloys Compd., 2021, 882, 160702.
14. Z. Zheng, J. Jiang, H. Guo, C. Li, K. Konstantinov, Q. Gu and J. Wang, Nano Energy, 2021, 81, 105529.
15. Q. Sun, H. Yadegari, M. Banis, J. Liu, B. Xiao, B. Wang, S. Lawes, X. Li, R. Li and X. Sun, Nano Energy, 2015, 12, 698.
16. S. Zhang, Z. Wen, K. Rui, C. Shen, Y. Lu and J. Yang, J. Mater. Chem. A, 2015, 3, 2568.
17. L. Ma, D. Zhang, Y. Lei, Y. Yuan, T. Wu, J. Lu and K. Amine, ACS Energy Lett., 2018, 3, 276.
18. X. Lin, J. Wang, X. Gao, S. Wang, Q. Sun, J. Luo, C. Zhao, Y. Zhao, X. Yang, C. Wang, R. Li and X. Sun, Chem. Mater., 2020, 32, 3018.
19. H. Lee, D. Park, J. Park, Y. Kim and J. Kang, Adv. Energy Mater., 2023, 13, 2300041.
20. F. X. Xie, X. L. Cui, X. Zhi, D. Z. Yao, B. Johannessen, T. Lin, J. N. Tang, T. B. F. Woodfield, L. Gu and S. Z. Qiao, Nat. Synth., 2023, 2, 129.
21. P. Zhou, Q. Zhang, Z. Xu, Q. Shang, L. Wang, Y. Chao, Y. Li, H. Chen, F. Lv, Q. Zhang, L. Gu and S. Guo, Adv. Mater., 2020, 32, 1904249.
22. Y. Zhang, L. Jiao, W. Yang, C. Xie and H. Jiang, Angew. Chem., Int. Ed., 2021, 60, 7607.
23. W. Zhang, Y. Chao, W. Zhang, J. Zhou, F. Lv, K. Wang, F. Lin, H. Luo, J. Li, M. Tong, E. Wang and S. Guo, Adv. Mater., 2021, 33, 2102576.
24. Y. Zhou, W. Yang, W. Utetiwabo, Y. Lian, X. Yin, L. Zhou, P. Yu, R. Chen and S. Sun, J. Phys. Chem. Lett., 2020, 11, 1404.
25. X. Han, X. Ling, D. Yu, D. Xie, L. Li, S. Peng, C. Zhong, N. Zhao, Y. Deng and W. Hu, Adv. Mater., 2019, 31, 1905622.
26. X. Zhou, J. Gao, Y. Hu, Z. Jin, K. Hu, K. M. Reddy, Q. Yuan, X. Lin and H. Qiu, Nano Lett., 2022, 22, 3392.
27. J. Chen, H. Li, C. Fan, Q. Meng, Y. Tang, X. Qiu, G. Fu and T. Ma, Adv. Mater., 2020, 32, 2003134.
28. Y. Jun, E. Lee, X. Wang, W. Hong, G. Stucky and A. Thomas, Adv. Funct. Mater., 2013, 23, 3661.
29. F. Ma, Z. Liu, G. Zhang, H. Fan, Y. Du, L. Zhen and C. Xu, Small, 2023, 19, 2207991.
30. M. Zhang, M. Shoaib, H. Fei, T. Wang, J. Zhong, L. Fan, L. Wang, H. Luo, S. Tan, Y. Wang, J. Zhu, J. Hu and B. Lu, Adv. Energy Mater., 2019, 9, 1901663.
31. S. Liu, C. Li, M. Zachman, Y. Zeng, H. Yu, B. Li, M. Wang, J. Braaten, J. Liu, H. Meyer, M. Lucero, A. Kropf, E. Alp, Q. Gong, Q. Shi, Z. Feng, H. Xu, G. Wang, D. Myers, J. Xie, D. A. Cullen, S. Litster and G. Wu, Nat. Energy, 2022, 7, 652.
32. X. Xie, L. Peng, H. Yang, G. I. N. Waterhouse, L. Shang and T. Zhang, Adv. Mater., 2021, 33, 2101038.
33. H. Funke, A. C. Scheinost and M. Chukalina, Phys. Rev., 2005, B71, 094110.
34. H. Funke, M. Chukalina and A. C. Scheinost, J. Synchrotron Radiat., 2007, 14, 426.
35. T. Zhang, X. Han, H. Liu, M. Biset-Peiro, J. Li, X. Zhang, P. Tang, B. Yang, L. Zheng, J. Morante and J. Arbiol, Adv. Funct. Mater., 2022, 32, 2111446.
36. X. Liu, F. Zhao, L. Jiao, T. Fang, Z. Zhao, X. Xiao, D. Li, K. Yi, R. Wang and X. Jia, Small, 2023, 19, 2300289.
37. W. W. Yin and Z. W. Fu, Chem. Commun., 2017, 53, 1522.
38. H. Yadegari, Y. Li, M. Banis, X. Li, B. Wang, Q. Sun, R. Li, T. Sham, X. Cui and X. L. Sun, Energy Environ. Sci., 2014, 7, 3747.
39. Z. P. Jovanov, L. Lutz, J. G. Lozano, C. Holc, X. W. Gao, A. Grimaud, J. M. Tarascon, Y. H. Chen, L. R. Johnson and P. G. Bruce, Batteries Supercaps, 2022, 5, e202200055.
40. C. G. Jiang, H. Zhang, P. Li, X. Y. Zhan, Z. J. Liu, L. Wang, B. H. Mao, Q. T. Li, Z. Y. Wen, Z. Q. Peng, S. L. Chen and Z. Liu, Adv. Funct. Mater., 2022, 32, 2202518.
41. J. Chen, H. Y. Zhang, F. J. Yu and Y. H. Chen, ACS Appl. Mater. Interfaces, 2024, 16, 26226.
42. X. F. Lei, M. L. He, P. L. Wu, C. Ma, X. Liu, C. Y. Yao, W. F. Cui, Q. X. Zhang, C. C. Li, H. Q. Li and X. Z. Liu, Small Methods, 2025, 9, 2401930.
43. W. W. Yin, J. L. Yue, M. H. Cao, W. Liu, J. J. Ding, F. Ding, L. Sang and Z. W. Fu, J. Mater. Chem. A, 2015, 3, 19027.
44. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865.
45. S. Grimme, J. Antony, S. Ehrlich and H. Krieg, J. Chem. Phys., 2010, 132, 154104.

---