Constructing high-capacity and ultra-stable biphasic layered cathodes via reductive coupling mechanism for sodium-ion batteries

Abstract

Stimulated anionic redox reactions are commonly utilized to design layered oxide cathodes for high-energy sodium-ion batteries (SIBs). Unfortunately, excessive oxygen redox often causes irreversible lattice oxygen loss and cation migration, leading to rapid capacity and voltage fading, as well as slow reaction kinetics. Here, we design and synthesize a P2/O3 biphasic cathode material, Na_0.8Ni_0.3Mn_0.5Cu_0.1Ti_0.1O₂ (NNMCT), which can activate the redox of oxygen at significantly lower voltage while leveraging the high stability of the P2 phase and the high capacity of the O3 phase. Furthermore, we elucidate the reductive coupling mechanism (RCM) of the uncommon electron transfer from oxygen to Ni/Mn ions within the material, which enhances the reversibility of the anion redox reaction, and induces the formation of strong covalent Ni/Mn-(O-O) bonds that effectively inhibit excessive oxygen oxidation and avoid rapid capacity decay. As a result, NNMCT cathode materials exhibit excellent reversible capacity (134.0 mAh g^–1 at 10 mA g^–1) and superior long-cycle stability (85.2% capacity retention after 500 cycles at 100 mA g^–1). The intrinsic functional mechanism of RCM is fully elucidated through a series of in situ/ex situ characterizations as well as theoretical calculations. These findings underscore the advantages of biphasic/multiphasic cathodes for high-energy SIBs.

Graphical abstract

A P2/O3 mixed-phase Na_0.8Ni_0.3Mn_0.5Cu_0.1Ti_0.1O₂ (NNMCT) cathode material with Cu/Ti co-doping is synthesized to inspire reductive coupling mechanism (RCM) to stabilize anionic redox reactions by forming strong covalent Ni/Mn−(O−O). NNMCT delivers a reversible capacity of 134.0 mAh g⁻¹ and exhibits a high cyclic stability for 85.2% capacity retention after 500 cycles at 100 mA g⁻¹.

1. Introduction

In the era of carbon neutrality, new grid-scale energy storage systems have received widespread attention [[1], [2], [3], [4]]. Due to the abundant global reserves of sodium and their low cost, sodium-ion batteries (SIBs) show great potential as a viable alternative to lithium-ion batteries (LIBs) [[5], [6], [7], [8], [9]]. However, has the higher redox potential and larger mass of Na⁺ compared to Li⁺ lead to a reduced energy density, which poses challenges for the practical application of SIBs [[10], [11], [12], [13], [14]]. Therefore, there is an urgent necessity to develop high-energy cathode materials to promote the commercial adoption of SIB technology.

Among the diverse cathode materials, transition metal (TM) layered oxides are at the forefront of SIB cathode research due to their high theoretical capacity, elevated voltage, fast electron conduction, and improved multiplicity properties [[15], [16], [17]]. Based on the positioning of Na and the unique number of oxide layers, Na_xTMO₂ is mainly classified into O3-type and P2-type oxides. P2-type oxides feature a wide Na transport channel, allowing Na⁺ to diffuse directly between adjacent trigonal prism sites, whereas O3-type oxides accommodate a higher Na content. As a result, P2-type oxides typically exhibit superior rate capability, while O3-type oxides provide greater capacity [18,19]. The strategy of designing and synthesizing P2/O3 hybrid phases aims to combine the advantages of both types [[20], [21], [22]]. However, achieving high capacity solely through the valorization of transition metal elements in layered oxide structures is challenging and often necessitates the stimulation of anionic redox reactions to further increase the material capacity. Moreover, this process is often accompanied by drastic phase transitions and lattice slip in both P2 and O3 phases, leading to irreversible lattice oxygen delocalization, as well as migration and dissolution of transition metal elements, which subsequently causes rapid capacity decay, greatly limiting the practical application of anionic redox cathode materials [23,24]. In recent years, Tarascon, Doublet, and their colleagues were the first to identify an anomaly in the transfer of electrons from oxygen to TM cations (i.e., TM reduction) during the electrochemical oxidation of Li-rich layered oxide cathodes. This phenomenon is known as the reduction coupling mechanism (RCM) [[25], [26], [27], [28]]. However, there has been limited research on such mechanism in sodium layered oxides [29,30]; in particular, the effect of RCM on the reversibility and kinetics of the oxygen redox reaction in sodium oxide cathodes remains poorly understood. Consequently, it is highly desirable to stimulate a cost-effective reductive coupling mechanism between 3d TM ions and oxygen to create robust covalent TM−O interactions, thereby enhancing the redox stability of oxygen.

Herein, we successfully designed and synthesized P2/O3 mixed-phase Na_0.8Ni_0.3Mn_0.5Cu_0.1Ti_0.1O₂ (NNMCT) cathode materials (Fig. 1a), revealing the RCM involving Ni/Mn ions and O. The doping of Cu and Ti aims to construct cathode materials with a nanoscale mixed-phase structure. Specifically, the introduction of Cu modifies the electronic structure of the surrounding ions, thereby enhancing the intrinsic electronic conductivity of the materials, while the introduction of Ti leverages strong covalent Ti−O bonds to stabilize the layered structure. The obtained P2/O3 NNNMCT cathode material can stimulate anionic redox at lower voltages, while at higher voltages, it relies on the RCM to establish a Ni/Mn−(O−O) covalent structure. This structure effectively inhibits excessive oxygen oxidation and largely avoids irreversible lattice oxygen loss and rapid capacity decay caused by the migratory dissolution of transition metal elements. As a result, the NNMCT cathode material merges the remarkable capacity of the O3 phase (134.0 mAh g⁻¹ at 10 mA g⁻¹ and 85.0 mAh g⁻¹ at 2000 mA g⁻¹) with the excellent cycling stability of the P2 phase (85.2% capacity retention after 500 cycles at 100 mA g⁻¹ and 80.8% capacity retention after 1000 cycles at 500 mA g⁻¹) within the voltage window of 2-4.15 V. The complete phase transition of NNMCT during Na⁺ extraction/insertion is demonstrated by in situ X-ray diffraction (XRD). Furthermore, ex situ X-ray absorption spectroscopy (XAS) and X-ray photoelectron spectroscopy (XPS) analyses elucidate the charge transfer mechanism of the cathode during the charging/discharging. By calculating the partial density of states (pDOS) and the crystal orbital Hamiltonian population (COHP), we reveal the evolution of the electronic structure upon desodiation and achieve a comprehensive understanding of the intrinsic mechanism underlying the oxygen stabilization reaction of the RCM.

Fig. 1.

(a) Schematic illustration of the synthetic P2/O3 biphasic structure. (b) XRD pattern and Rietveld refinement of NNMCT. (c) The calculated formation energies of samples. (d) SEM image, (e) TEM image, (f) HRTEM image, (g) HADDF-STEM image, and (h) TEM-EDS mapping images of the NNMCT sample.

2. Materials and methods

2.1. Materials synthesis

All materials were synthesized via a typical solid-state reaction. In a typical procedure, stoichiometric amounts of Na₂CO₃, NiO, MnO₂, CuO, and TiO₂ were added to a ball mill tank with an adequate amount of ethanol. An additional 5 mol% of Na₂CO₃ was included to compensate for sodium loss at high temperatures. After 12 h of ball milling at 450 rpm, the precursors were dried at 80°C, and subsequently calcined at 1,000 °C for 15 h with a heating rate of 5 °C min^-1.

2.2. Materials characterization

The elemental ratios of all samples were determined by ICP-OES (iCAP 7400). The crystal structures of the samples were characterized by X-ray diffraction (XRD) (Rigaku, TTR- Ⅲ) using Cu-Kα radiation. The morphologies and microstructures of the samples were tested by scanning electron microscopy (SEM) (GeminiSEM 450) and transmission electron microscope (TEM) (JEM-2100F, JEOL, Japan). XAS of Ni K-edge, Mn K-edge and Cu K-edge were collected at BL14W beamline in the Shanghai Synchrotron Radiation Facility (SSRF), while the XAS of Ni L-edge, Mn L-edge and Cu L-edge were tested by the soft X-ray magnetic circular dichroism end station (XMCD) of National Synchrotron Radiation Laboratory (NSRL). The acquired EXAFS data was processed by Demeter software package. EXAFS was k³-weighted and Fourier transformed in the k range of 0-6 Å^-1. The Wavelet transformed (WT) of EXAFS was performed via HAMA software. The surface chemistry of the samples was characterized using X-ray photoelectron spectroscopy (XPS) testing.

2.3. Electrochemical measurements

The cathodes were prepared by mixing the active materials (70 wt%) with Super P (20 wt%) and polytetrafluoroethylene (PVDF, 10 wt%), and then dispersed on carbon-coated aluminum (Al) foil. Then the film was dried overnight in a vacuum oven at 120°C for 12 h. The loading mass of NNMCT on the electrodes was 2-3 mg cm⁻². The electrochemical performance was assessed using CR2032 coin half-cells mounted in a glove box filled with O₂ and H₂O concentrations less than 0.1 ppm. Na was used as the counter electrode and a glass microfiber filter (GF/D, Whatman) was used as the separator. The electrolyte used was 1 M NaClO₄ in propylene carbonate (PC) with a 5% fluoroethylene carbonate (FEC) additive. All batteries underwent a preliminary activation process before electrochemical performance testing, which involved a pre-cycling process at a current density of 10 mA g⁻¹ for 5 cycles. The cycling and rate performances were conducted on the Neware battery cycler (BTS BTS-610, Shenzhen, China). The CHI660D workstation was used for the cyclic voltammetry (CV) measurements and electrochemical impedance spectroscopy (EIS) tests.

2.4. Computational details

All calculations in this study were performed using density functional theory (DFT) methods within the Vienna Ab-initio Simulation Package (VASP) [31,32]. The projector augmented wave (PAW) [33,34] method was utilized to model electron-ion interactions, with a plane-wave energy cutoff of 500 eV. Exchange-correlation interactions were treated using the generalized gradient approximation (GGA) as formulated by Perdew, Burke, and Ernzerhof (PBE) [35]. The convergence criterion for interatomic forces during optimization was set to 0.02 eV/Å, while the total energy convergence criterion for wavefunction self-consistency was set to 10^-5 eV. Considering that these systems in our study contain transition metal atoms (such as Mn, Ni, Cu, Ti), which possess d or f electrons and are thus strongly correlated, the GGA functional is insufficient for accurate calculations. Therefore, the inclusion of an on-site Coulomb interaction term (U) is necessary, and LDAU = .TRUE. was applied in the calculations. The U values for Mn, Ni, Cu, and Ti were set to 3.9, 3, 4, and 4, respectively. In the DOS calculations, NEDOS was set to 3,000, while for the COHP calculations, NBANDS was set to 1100, with post-processing carried out using the LOBSTER tool [[36], [37], [38], [39], [40]].

3. Result and discussion

3.1. Structure analysis

All cathode materials were successfully synthesized via solid-state reactions, as detailed in the Supporting Information. Their stoichiometries were determined by inductively coupled plasma-atomic emission spectroscopy (ICP-AES), with results presented in Table S1. The elemental content of Na_0.8Ni_0.4Mn_0.6O₂ (NNM), Na_0.8Ni_0.3Mn_0.6Cu_0.1O₂ (NNMC), Na_0.8Ni_0.3Mn_0.6Ti_0.1O₂ (NNMCT) and Na_0.8Ni_0.3Mn_0.5Cu_0.1Ti_0.1O₂ (NNMCT) are as designed. X-ray diffraction (XRD) patterns and corresponding Rietveld refinements in Fig. 1b and Fig. S1 indicate that the NNMCT and NNMC crystallize in a mixed hexagonal P2-type layered structure and rhombohedral O3-type layered structure (space group: R-3m), while the NNM and NNMT crystallize in the hexagonal P2-type layered structure (space group: P6₃/mmc). Their refined results are provided in Table S2–S5. The proportions of P2-type and O3-type phases in NNMCT are approximately 31.3% and 68.7%, respectively, which are comparable to those in NNMC (39.6% P2-type phase and 70.36% O3-type phase). This biphasic layered structure suggests that doping with elements of low cationic potential can induce a transition from the P2-type phase to the O3-type phase (e.g., replacing Ni with Cu and Mn with Ti), in samples with equivalent sodium content. As shown in Fig. 1c, Cu/Ti doping can reduce the formation energy of the material, suggesting a more stable structure, with NNMCT exhibiting the highest stability, which implies superior electrochemical cycling stability. The scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images of NNMCT, shown in Fig. 1d and Fig. 1e, exhibit lamellar morphologies ranging from 2 to 5 μm in size. The morphology of NNMCT is similar to that of the other samples (Fig. S2-S7). Additionally, the microstructure of NNMCT was further surveyed through high-resolution TEM (HRTEM). As demonstrated in Fig. 1f, the P2-type and O3-type phases coexist at the nanoscale within the bulk phase structure. The interplanar spacings of 2.48 Å and 2.21 Å correspond to the (100) and (012) planes of the P2 structure, respectively, while the interplanar spacing of 2.54 Å corresponds to the (101) planes of the O3 structure. In addition, the scanning transmission electron microscopy (STEM) image of a high-angle annular dark field (HAADF) shows clear interplanar spacings of 5.55 Å for the (002) plane of P2 structure and 5.41 Å for the (003) plane of O3 structure (Fig. 1g). Thus, the P2/O3 phase in the NNMCT forms a symbiotic structure at the nanoscale, rather than merely developing as isolated individual particles. The energy dispersive X-ray spectroscopy (EDS) mappings (Fig. 1h) verify the successful doping of Cu and Ti elements, as well as the uniform distribution of all elements.

3.2. Electrochemical performance

To assess the effect of doping with different elements, we assessed the electrochemical performance of four cathode materials in conjunction with Na metal anode using coin cells, as shown in Fig. 2. The galvanostatic charge/discharge curves of all samples were initially performed within a voltage window of 2.0-4.15 V at 10 mA g⁻¹, as depicted in Fig. 2a. The initial specific charge capacities of NNM, NNMC, NNMT and NNMCT are 79.3, 105.3, 92.2 and 120.4 mAh g⁻¹, respectively; while their corresponding initial Coulombic efficiencies (ICE) are 103.9%, 118.0%, 99.4% and 111.3%. When the ICE reaches 110% or more, significantly higher than 100%, it means that more sodium ions are replenished from the negative (Na) side to the positive side during the first discharge process, which is something we do not want to get because in a full cell, the negative (hard carbon) side is not able to supply the excess sodium ions, which will result in a decrease in the capacity of the material in the full cell. Notably, NNMCT exhibits the highest specific capacity, which is nearly 50% higher than that of the original NNM. The NNM shows a characteristic charge/discharge curve typical of P2-type layered oxides, featuring two voltage plateaus at 2.7/2.6 V and 3.6/3.5 V, indicative of the Na⁺/vacancy ordering rearrangement. Such rearrangement inevitably results in a reduction of the Na⁺ diffusion coefficient and a diminished dimensionality of ionic transport. The interplay of these phenomena contributes to kinetic retardation and rapid capacity decay during charging and discharging. In contrast, the charge/discharge curves of NNMCT are smoother, which implies the elimination of the ordered rearrangement of the Na⁺/vacancy. This is consistent with the CV results in Fig. 2b. The mitigation of this effect is also illustrated by the weak intensity of the two superlattice peaks that represent the ordered rearrangement of Na⁺/vacancy between 25° and 30° in the XRD patterns (Fig. S8). Particularly, the Cu-doped sample exhibits a more pronounced elimination effect, which accelerates the sodium ion transport kinetics significantly. Moreover, within the voltage range of 4-4.15 V, a small redox peak emerges in the CV curve of the NNMCT compared to the NNM, indicating the onset of anionic redox reactions in the NNMCT. The elimination of the ordered rearrangement of Na⁺/vacancy, coupled with the activation of anionic redox reactions at lower voltages (< 4.15 V), contributes to the surprisingly high reversible capacity, quality multiplicity performance, and excellent cycling stability of NNMCT. Furthermore, the NNMCT cathode delivers admirable discharge capacity of 85.0 mAh g^–1 even at 2000 mA g⁻¹ compared to other samples (Fig. 2c), and maintains 85.2% capacity after 500 cycles at 100 mA g⁻¹ (Fig. 2d) and 80.8% capacity after 1000 cycles at 500 mA g⁻¹ (Fig. 2e). All cathodes can be cycled stably at a current density of 10 mA g⁻¹, with NNMCT having a higher capacity along with more than 90% capacity retention after 100 cycles (Fig. S9). The charge/discharge curves of NNMCT at various current densities are shown in Fig. S10, NNMCT exhibits excellent specific capacity compared to other samples at different current densities. On the other hand, the electrochemical tests of NNM and NNMCT were carried out within a wider voltage window of 2.0-4.3 V to evaluate their high-voltage performance, as depicted in Fig. 2f. The NNMCT delivers a high reversible capacity of 159.1 mAh g⁻¹ at 10 mA g⁻¹ (515 Wh kg⁻¹, Fig. S11), corresponding to the insertion of 0.634 mol Na⁺ per mol of NNMCT. This capacity is slightly lower than the 168.2 mAh g⁻¹ (equaling to insertion of 0.670 mol Na⁺ per mol of NNM). This difference can be attributed to the presence of more electrochemically inert elements in NNMCT, resulting in a slightly lower capacity in the deep discharge state compared to NNM. In addition, the initial CV curve of NNM exhibits significantly stronger oxidation peaks at high voltage than those of NNMCT in the voltage window of 2-4.3 V (Fig. 2g). This indicates that the degree of the oxygen redox reaction is suppressed at high voltage in NNMCT, which is conducive to a more stable cycling of the material. Intense oxygen redox reactions can lead to the irreversible collapse of the laminated structure. As a result, the NNMCT achieves a capacity retention of 84.1% over 200 cycles at a current density of 100 mA g⁻¹ (Fig. 2h). Comparing the CV curves of NNM and NNMCT at different sweep speeds (Fig. S12) reveals that the high-voltage redox peaks in NNMCT remain distinctly observable even at high sweep speeds. In contrast, the redox peaks in NNM appear to be significantly weakened in intensity, implying a rapid decay of the capacity for the NNM electrode when cycled within a large voltage window (2-4.3 V). Therefore, Cu doping can eliminate the ordered rearrangement of Na⁺/vacancy to a certain extent, improve the diffusion coefficient of Na⁺, stimulate the redox reaction of oxygen at lower voltage, and increase the reversible capacity; whereas Ti doping plays a role in stabilizing the structure of the material, which is manifested in the inhibition of the violent redox reaction of oxygen at high voltage, improving the cyclic stability of the material. The synergistic effects of these two dopants lead to a more pronounced enhancement in material performance.

Fig. 2.

(a) Galvanostatic charge/discharge profiles of NNM, NNMC, NNMT and NNMCT electrodes at 10 mA g⁻¹ in the voltage range of 2.0-4.15 V. (b) CV curves of NNM and NNMCT electrodes at 0.1 mV s⁻¹ in the voltage range of 2.0-4.15 V. (c) Rate capabilities of NNM, NNMC, NNMT and NNMCT electrodes. (d) Cycling performance of NNM, NNMC, NNMT and NNMCT electrodes at 100 mA g⁻¹ in the voltage range of 2.0-4.15 V. (e) Cycling performance of NNM and NNMCT electrodes at 500 mA g⁻¹ in the voltage range of 2.0-4.15 V. (f) Galvanostatic charge/discharge profiles of NNM and NNMCT electrodes at 10 mA g⁻¹ in the voltage range of 2.0-4.3 V. (g) CV curves of NNM and NNMCT electrodes at 0.1 mV s⁻¹ in the voltage range of 2.0-4.3 V. (h) Cycling performance of NNM and NNMCT electrodes at 100 mA g⁻¹ in the voltage range of 2.0-4.3 V.

The CV curves of NNM and NNMCT at different sweep speeds are shown in Fig. S13. At high sweep speeds, NNMCT still maintains a distinct redox peak in the high voltage interval, while the corresponding redox peak intensity of NNM decreases significantly, indicating its poor reversibility of anionic redox. This corresponds well to the superior cycling stability of NNMCT at high voltage, while the capacity of NNM decays rapidly. Further, the morphology of NNMCT and NNM after two hundred cycles was shown in Fig. S14-S15, and the NNMCT still maintains the original lamellar structure after cycling, while the NNM particles show very obvious cracks and obvious structural degradation, which implies that the capacity decline of NNM is highly related to the structural degradation. After two hundred cycles, the impedance of NNMCT was significantly smaller than that of the other samples, corroborating the role played by the elimination of the ordered rearrangement of Na⁺/vacancy in accelerating the diffusion kinetics of sodium ions (Fig. S16). In addition, according to the fitting results of the variable temperature impedance test (Fig. S17-S18), the E_a of the NNMCT electrodes (24.26 kJ mol⁻¹) was considerably lower than that of the NNM electrodes (43.12 kJ mol⁻¹). Additional analyses were performed using the Gamma Electrostatic Intermittent Titration Technique (GITT). The results of the GITT test (Fig. S19) showed that the sodium ion diffusion coefficient of the NNMCT electrodes was significantly higher than that of the NNM electrodes throughout the charging and discharging process. Together, these findings indicate that the ion migration kinetics of the NNMCT electrode is significantly faster than that of the NNM electrode. This enhanced kinetic behavior is attributed to the optimized structural features of the NNMCT material, which promotes more efficient sodium ion transport and improves electrochemical performance.

3.3. Charge compensation mechanism

To gain a deeper understanding of the electrochemical reaction processes of the samples during the charging/discharging, we employed XAS and XPS for a comprehensive investigation of the charge compensation mechanism. Initially, we examined the potential valence changes of Ni and Mn elements in the NNMCT sample by ex situ XAS measurements. Fig. 3a-c shows the normalized X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) spectra at the Ni K-edge. Higher energy in the XANES spectra corresponds to higher elemental valence states, while the EXAFS spectra of typical layered oxides provide local structural information, with the first peak indicating the TM−O bond distance and the second peak reflecting the TM−TM distance. During the initial state of charging, as the voltage increased to 4.0 V, the Ni K-edge XANES spectrum exhibited a shift toward higher energy, accompanied by a significant contraction of the first Ni−O coordination shell, suggesting the oxidation of Ni ions during this process. Intriguingly, when fully charged to 4.3 V, the Ni K-edge XANES spectrum unexpectedly shifts towards lower energy, implying a paradoxical decrease in the valence state of Ni. This phenomenon is attributed to the redox reaction of oxygen in a deeply desodiation state, where O²⁻ is transformed into (O₂)ⁿ⁻, resulting in the partial transfer of electrons from oxygen to nickel ions and the formation of Ni−(O−O) bonds with strong covalent interactions. Previous studies have identified this phenomenon as a reductive coupling mechanism (RCM), which has thus far been observed only in elements such as Fe and Cu. Upon discharging to 4.15 V, such Ni−(O−O) bonding disappears as the electrons return to oxygen, resulting in partial oxidation of the nickel ion and an increase in the Ni−O bond distance. It is worth noting that the Ni K-edge XANES spectrum at this stage remains at a lower energy compared to that at 4.0 V, indicating the persistence of some Ni−(O−O) bonds within the structure. This implies that the material has already undergone the redox reaction of oxygen, accompanied by an electron transfer process, in the voltage window of 4-4.15 V. This observation aligns with the unexpectedly high-capacity exertion of the NNMCT discussed previously. Furthermore, the strong covalent nature of the Ni−(O−O) bonding can effectively avoid oxygen escape due to anion redox during the deep desodiumization state, thus maintaining excellent structural stability at high voltages. A similar transformation occurs in the second Ni−TM coordination layer, attributed to the edge-sharing configuration of the octahedral structure. Additionally, ex situ sXAS in the total electron yield (TEY) mode was utilized to monitor the valence variations of TMs in NNMCT during charge/discharge (Fig. S20–22). The variation in the L₃/L₂ ratio of Ni L-edge, which responds to the valence change of Ni during these processes, further elucidates the RCM (Fig. S20). Specifically, the L₃/L₂ ratio of Ni L-edge increases with an elevation in the Ni valence state and decreases when the state is reduced. Upon discharging to 2.0 V, both XANES and EXAFS spectra at Ni K-edge return to their initial states, confirming the reversibility of the electrochemical reaction process.

Fig. 3.

(a) Ex situ XANES, (b) enlarged plots and (c) Fourier-transformed EXAFS spectra at Ni K-edge of NNMCT electrode. (d) Ex situ XANES, (e) enlarged plots and (f) Fourier-transformed EXAFS spectra at Mn K-edge of NNMCT electrode. Contour plots of the (g) Ni and (h) Mn K-edge WT-EXAFS spectra at different voltage states.

Similarly, as shown in Fig. 3d-e, we observed analogous features in the XANES and EXAFS spectra at the Mn K-edge, indicating that manganese is also involved in the center of redox activity in its initial charge state prior to reaching 4.0 V. Subsequently, at 4.15 V, the redox reaction of oxygen starts to be stimulated, resulting in the formation of Mn−(O−O) bonds, akin to those of Ni−(O−O), via the RCM. Furthermore, the variations in the intensity of the characteristic peaks representing the different valence states of the Mn in the L-edge sXAS spectrum corroborate the occurrence of Mn²⁺/Mn³⁺ and Mn³⁺/Mn⁴⁺ redox reactions throughout the charging/discharging processes (Fig. S21). In order to visualize the variation of bond lengths, the wavelet-transform (WT) EXAFS spectra at Ni and Mn K-edge were also recorded (Fig. 3g, h), revealing two scattering peaks at around 1.6 and 2.6 Å, which correspond to the TM-O and TM-TM bonds, respectively. However, the XANES spectra at Cu K-edge exhibit a brief shift toward higher energy in the initial charging state, remaining relatively stable until discharged to 2.0 V (Fig. S23). This behavior aligns with the weak redox peak located at 3.90 V/3.87 V in the CV curves of NNMCT, indicating minimal involvement of copper in charge compensation. In addition, the peak shapes of the ex situ sXAS spectra at Ti L-edge (Fig. S22) as well as the ex situ Ti 2p XPS spectra (Fig. S24) show that the pristine tetravalent Ti ions, which do not participate electrochemically during the charging process, serve solely as a stabilizing framework for the laminar structure.

To investigate the role of oxygen redox in the charge compensation process, we collected ex situ O 1s XPS spectra (Fig. S25). Normally, the deconvoluted peaks located at 532–534 eV correspond to surface oxygen species, while the deconvoluted peak centered at 529.8 eV is attributed to lattice oxygen. Notably, when charged to 4.15 V, a new peak appears at 530.8 eV, which is assigned to the (O₂)ⁿ⁻ species, indicating the oxidation of lattice oxygen. This peak remains present in the fully charged state but disappears during discharge. This observation is further corroborated by the ex situ O K-edge sXAS spectra of NNMCT (Fig. S25), where the appearance of a saddle peak between the e_2g peak and the t_g peak suggests the generation of the O−O dimer, indicative of oxygen oxidation. Upon discharging to 4.0 V, the spectrum returns to its original shape, signifying oxygen reduction.

3.4. Structure evolution and air stability

The structural evolution of NNMCT was monitored using in situ XRD during the initial Na⁺ de-intercalation and intercalation processes, as shown in Fig. 4a and Fig. S26. Given that the O3 phase occupies nearly 70% of mass fraction in NNMCT and possesses a higher Na content compared to the P2 phase, it is thermodynamically favorable for Na in the bulk O3 phase to be preferentially removed. During the charging process, the (003) peak of the O3 phase, located near 16.3°, initially shifts to a lower angle as Na⁺ is extracted. This shift is attributed to the spatial enlargement of the c-axis, which occurs due to increased electrostatic repulsive forces between neighboring oxygen layers. This process is accompanied by a structural transition from the O3 phase to the P3 phase, completing at 3.6 V. In contrast, the (002) peak of the P2 phase at 15.9° maintains during the early stages, only beginning to shift to a higher angle once the O3 phase has completely transformed to the P3 phase, indicating a decrease in the c-axis. At this juncture, we propose that the P2 phase transitions into a new P'2 phase with smaller layer spacing. This may be due to the fact that after the transformation of the O3 phase, which occupies most of the bulk phase, into the P3 phase, the volume expansion it exhibited along the c-axis compresses the P2 phase, which is in between the neighboring P3 phases, to a certain extent, resulting in the shift of the (002) peak (This transition may be attributed to the volume expansion of the P3 phase along the c-axis after the transformation of the predominantly bulk O3 phase, which exerts compressive forces on the intervening P2 phase, thereby facilitating the observed shift of the (002) peak. This phenomenon may be inevitable due to the nanoscale distribution of the pristine O3 and P2 phases within NNMCT. In situ XRD test of the NNM exhibits a pure P2 phase, as shown in Fig. S27, and its (002) peak shifts to a smaller angle during the charging process, which is in accordance with the conventional theory. This observation suggests that the anomalous shift of the (002) peak of the P2 phase is characteristic of the specific NNMCT mixed-phase structure. It may play a significant role in maintaining the structural stability in a deeply denatured state, which correlates with the excellent electrochemical stability exhibited in the P2/O3 biphasic NNMCT. Simultaneously, throughout the charging process, the (101), (012), and (104) peaks of the O3 phase, as well as the (012) peak of the P2 phase, keep moving toward higher angles, indicating that the oxidation of TM ions induces the contraction in the ab plane. An increase in the valence of the reactive TM cation leads to an increase in the TM−O bonding energy and a decrease in the ionic radius, resulting in a contraction of the a/b axis. During the initial discharge process, as Na⁺ gradually intercalates into the bulk phase, the diffraction peaks shift in the opposite direction, implying that the P3 and P'2 phases are gradually transformed back to the original O3 and P2 phases. Notably, some diffraction peaks associated with the P3 phase persist in the XRD pattern even in the fully discharged state, indicating that the structural transformation is not completely reversible. This partially irreversible structural transformation may be beneficial for subsequent cycles. Given that the P3 phase has a prismatic Na occupation site similar to that of the P2 phase, it is generally accepted that the P3 phase offers a superior Na⁺ diffusion coefficient compared to the O3 phase, which corresponds to improved multiplicity performance. Thus, throughout the entire charge/discharge process, the phase transformation can be summarized as follows: P2+O3→P2+O3+P3→P'2+P3→P2+O3+P3. TEM images and XRD pattern of the NNMT cathode after the first cycle are shown in Fig. S28–S29 to proof that the NNMCT after cycling remains a P2/O3 mixed-phase structure.

Fig. 4.

(a) Contour map of in situ XRD patterns and corresponding phase transition for NNMCT electrode during the first cycle at 30 mA g⁻¹ between 2.0 and 4.3 V. Comparison of (b) XRD patterns, (c) galvanostatic charge/discharge profiles and (d) cycling performance of the NNMCT and NNMCT after 30 days of exposure to air in the voltage range of 2.0-4.15 V. Comparison of (e) XRD patterns, (f) galvanostatic charge/discharge profiles and (g) cycling performance of NNMCT and water-washed NNMCT in the voltage range of 2.0-4.15 V.

It is well established that the air stability of the majority of O3-phase layered oxides for SIBs is of concern; their reliance on stringent environmental conditions, such as low humidity and low oxygen conditions levels, often leads to increased costs associated with their preparation and storage [41,42]. Given that the primary component of NNMCT is the O3 phase, we conducted further tests to assess the air stability of NNMCT in order to evaluate its potential for practical applications. Firstly, we analyzed the XRD pattern of NNMCT powders that had been exposed to air for 30 days and compared it with that of the fresh sample (Fig. 4b). The diffraction peaks of the treated NNMCT, especially the O3-(003) and P2-(002) peaks, exhibit no significant attenuation in intensity or angular shift, indicating that the lamellar structure of the material remains stable. In addition, no diffraction peaks corresponding to impurities such as hydrates and NaHCO₃, which would indicate reactions with H₂O, O₂, and CO₂ in the air, are found in the XRD pattern. Further, we investigated the electrochemical properties of the air-exposed sample (Fig. 4c). The NNMCT exposed to air for 30 days exhibits no significant degradation in its initial specific capacity during cycling at a current density of 10 mA g⁻¹, a finding that sharply contrasts with the behavior of other common O3 phase materials. Moreover, we assessed its long-term cycling stability, as shown in Fig. 4d. After three hundred stable cycles at a current density of 100 mA g⁻¹, the capacity decay rate is minimal when compared with that of the fresh sample, indicating that NNMCT possesses superior resistance to air corrosion. Subsequently, the water-washed NNMCT sample was characterized by XRD after drying to assess the moisture resistance under more extreme environmental conditions (Fig. 4e). The resulting XRD pattern reveals that no diffraction peaks corresponding to the products of the acid degradation reactions with air components are present in the water-washed NNMCT, and distinct layered diffraction peaks indicative of the intact mixed-phase structure of the material are observed. As depicted in Fig. 4f, the initial cycling capacity of the water-washed NNMCT is 132.1 mAh g⁻¹ at 10 mA g⁻¹, which is comparable to that of the fresh sample (134.0 mAh g⁻¹). Similarly, the capacity degradation observed after three hundred cycles of electrochemical cycling for the water-washed NNMCT is relatively minor (Fig. 4g). SEM images of NNMCT exposed to air for 30 days and water-washed NNMCT are shown in Fig. S30–S31. The morphology of NNMCT did not change much after different treatments. Collectively, these findings demonstrate that air-stable NNMCT has impressive resistance to both acid and oxidative degradation, with minimal capacity decrease during electrochemical cycling remaining within acceptable limits. This suggests that NNMCT can be largely liberated from the harsh conditions associated with synthesis, storage, and transportation, thereby reducing costs for practical applications.

3.5. Theoretical calculations

The density of states (DOS) for the NNM, NNMC, NNMT and NNMCT were analyzed using DFT-based calculations. The computational models are shown in Fig. S32–S35. The results of DOS for NNM and NNMT are presented in Fig. 5a-b, revealing that both materials exhibit band gaps. In contrast, the DOS of NNMCT does not display a significant band gap, a phenomenon attributed to the introduction of the Cu element, which alters the electronic structure and consequently improves the material's conductivity. This conclusion is further supported by Fig. S36, which shows that the DOS of NNMC also lacks an open band gap. Meanwhile, the density of states of NNMCT near the Fermi energy level is significantly larger than that of NNM and NNMT, which implies that the conductivity of NNMCT is better and the migration of electrons is more likely to occur. To elucidate the critical role of the RCM in stabilizing the oxygen reaction, we performed partial density of states (pDOS) and the crystal orbital Hamiltonian population (COHP) calculations based on DFT for a series of Na_xNi₁₅Mn₂₅Cu₂Ti₂O₁₀₀ (x = 40, 25, 10) supercells. These three structures correspond to the initial charging state, the anion redox excited state, and the deep sodium deficiency state, respectively. As shown in Fig. 5d, at x = 40, the valence band near the Fermi energy level is predominantly occupied by the Ni 3d orbitals, indicating the dominance of Ni redox in charge compensation during the initial charging period. At x = 25, the valence band near the Fermi level is mainly composed of O 2p states, suggesting the involvement of oxygen redox reactions in charge compensation at this stage. Notably, the Ni/Mn 3d band around the Fermi level decreases, further indicating the oxidation of Ni/Mn during the transition from x = 40 to x = 25. At x = 10, a significant increase in the O 2p states (hole density) above the Fermi level is observed, coinciding with an increase in the Ni/Mn 3d states near the Fermi energy level. This suggests that the charge is transferred from the oxygen to the Ni/Mn ions during the deep desodiation (i.e., RCM). In addition, information on Ni−O, Mn−O and O−O bonding was also investigated by COHP calculations (Fig. 5e–g). Positive COHP values correspond to the antibonding state, while negative values indicate the bonding state [43]. The Ni/Mn−O COHP curves demonstrate substantial antibonding participation, reflecting a strong Ni/Mn−O interaction throughout the desodiation process, which is conducive to stabilizing the oxygen oxidation reaction [44,45]. The integrated COHP (ICOHP) is subsequently employed to evaluate the bonding strength of the Ni/Mn−O and O−O bonds (Fig. 5h–i). During the transition from x = 25 to x = 10, the −ICOHP value for Ni/Mn−O bonds increases, mainly due to Ni/Mn oxidation. As for the O−O bonds, the relatively low −ICOHP values indicate minimal bonding between neighboring oxygen ions. Following further Na⁺ extraction, the −ICOHP values for both Ni/Mn−O and O−O bonds increase, suggesting the formation of a stable Ni/Mn−(O−O) configuration through a reductive coupling mechanism, wherein the oxygen ions are covalently bonded to the Ni/Mn ions. Previous experimental results indicate that this Ni/Mn−(O−O) interaction can effectively improve oxygen stability.

Fig. 5.

The density of states for (a) NNM, (b) NNMT and (c) NNMCT with the corresponding pDOS spectra. (d) pDOS of the Na_xNi₁₅Mn₂₅Cu₂Ti₂O₁₀₀ (x = 40, 25, 10) supercells. The COHP profiles of (e) Ni–O, (f) Mn–O and (g) O–O bands. (h) and (i) The corresponding −ICOHP values of Na_xNi₁₅Mn₂₅Cu₂Ti₂O₁₀₀ (x = 40, 25, 10).

4. Conclusion

In conclusion, the reductive coupling mechanism facilitating electron transfer from oxygen ions to active transition metal ions has been elucidated, enhancing the reversibility of the anionic redox reaction in a novel biphasic P2/O3-Na_0.8Ni_0.3Mn_0.5Cu_0.1Ti_0.1O₂ cathode in a deeply desodiumated state. The mixed-phase strategy combined with Cu and Ti double doping enables the NNMCT to stimulate the oxygen redox reaction at lower voltage and stabilize the process at high voltage. This results in excellent reversible capacity and high multiplicity capability (134.0 mAh g⁻¹ at 10 mA g⁻¹ and 85.0 mAh g⁻¹ at 2000 mA g⁻¹ within the voltage range of 2-4.15 V). Furthermore, the material exhibits remarkable cycling stability (85.2% capacity retention after 500 cycles at 100 mA g⁻¹ and 80.8% capacity retention after 1,000 cycles at 500 mA g⁻¹), effectively combining the advantages of both the P2 and O3 phases. In situ XRD reveals a complete phase transformation of NNMCT during Na⁺ extraction/insertion, while the presence of the P3 phase promotes the material’s kinetics. Systematic ex situ XANES, XAS, and XPS characterizations provide insights into the charge transfer mechanism during charging and discharging, and DFT calculations elucidate the intrinsic chemical mechanism of RCM in stabilizing anion redox. More excitingly, NNMCT has superb resistance to both acid and oxidative degradation, showing promising applications in air stabilization. This study highlights the synergistic effect of anionic and cationic redox reactions, offering a novel approach to designing high-capacity and long-life cathode for SIBs.

Declaration of competing interest

The authors declare that they have no conflicts of interest in this work.

Biographies

Shaolong Zhu obtained his B.S. degree from University of Science and Technology of China (USTC) in 2022 and now he is pursuing the Ph.D. degree in the Department of Materials Science and Engineering, USTC. His research focuses on the design of electrode materials for Na-ion batteries.

Yan Yu is a professor of material science at the University of Science and Technology of China (USTC). She received her Ph.D. in material science at USTC in 2006. Her current research interests mainly include the design of novel nanomaterials for clean energy, especially for batteries and the fundamental science of energy storage systems.