1 Introduction
Sodium-ion batteries (SIBs) have attracted significant attention in recent years due to the abundance and widespread availability of sodium resources, coupled with high safety and low cost. These advantages position SIBs as strong candidates for large-scale energy storage systems [
1,
2]. As a critical component of SIBs, various cathode materials have been extensively explored over the past two decades, including Prussian blue analogs, polyanionic compounds, layered oxides, and others [
3,
4]. Inspired by the successful application of Li
xTMO
2 in lithium-ion batteries [
5,
6], layered oxides (Na
xTMO
2, where TM=Ni, Fe, Mn, Co, etc.) have been widely investigated as promising cathodes for SIBs due to their facile synthesis, competitive specific capacities, and favorable Na-ion conductivity [
7].
In Na
xTMO
2 structures, different stacking sequences of transition metal and oxygen atoms give rise to two primary configurations: P2-type and O3-type [
8]. In the P2 structure, sodium ions are positioned at the edges and centers of triangular prisms formed by NaO
6 units. The letter “P” stands for “prism,” while the numeral “2” indicates that the oxygen stacking sequence has a periodicity of two, described as “ABBAABBA…”. In contrast, sodium ions occupy the centers of NaO
6 octahedra in the O3 structure, where “O” denotes “octahedron” and the numeral “3” signifies a periodicity of three with an “ABCABC…” stacking sequence. These distinct stacking arrangements lead to different ion diffusion pathways and energy barriers, significantly influencing sodium-ion kinetics in cathode materials.
Recent studies have revealed that P2-type oxides with specific compositions can enhance energy density through lattice oxygen redox (LOR), due to their high operating potentials and additional capacity contributions. This provides a pathway to overcome the energy density limitations of traditional layered oxide cathodes [
9–
12]. LOR can be triggered by the presence of unique “Li–O–Na” or “Mg–O–Na” configurations along the
c-axis in P2-type oxide crystal structures, as reported for P2-Na
0.6Li
0.2Mn
0.8O
2 [
13] and P2-Na
2/3[Mg
0.28Mn
0.72]O
2 [
14]. However, LOR at high voltages leads to the generation of O 2p states with uncoordinated oxygen and electron-deficient, resulting in irreversible lattice oxygen release and structural degradation, often accompanied by phase transitions from P2 to O2/OP4 [
13,
15].
These complicated phase transitions not only induce significant volume change of unit cell and lattice stress, but also severely degrade cycling stability and rate capability [
16,
17]. The phase transition from P2 to O2 and P2 to OP4 involves the conversion of NaO
6 prism into NaO
6 octahedron, which increases the Na-ion diffusion barrier and results in poor Na-ion kinetics in O-type stacking [
18]. Therefore, suppressing such phase transitions is a viable strategy to enhance structural stability and improve Na-ion diffusion kinetics in cathodes materials.
Elemental doping is a common strategy to inhibit the phase transitions, where the dopants serve as “pillar” in the Na layers or “rivets” in the transition metal layers. For example, Mg doping has been reported to act as a pillar in the Na layers, stabilizing the P2 structure throughout charge/discharge process and significantly improving rate capability and cycling stability [
15]. Previous work has demonstrated that doping with Sb, due to its high ionic potential, acts as a “rivet” in the transition metal layers, constraining phase transitions and promoting the formation of the Z phase, thereby enhancing energy density and Na-ion kinetics [
19]. Similarly, Nb doping has been found to suppress phase transitions in P2-NaLiMnO
2 by reinforcing the structure through highly covalent Nb–O bonds [
20]. These findings highlight the importance of strong covalent interactions in improving structural stability and Na-ion diffusion. However, the relationship between covalent bond strength and structural stability remains insufficiently understood.
In this context, a P2-type Na2/3Li1/6Al1/6Mn2/3O2 cathode (denoted as LAM) was designed by incorporating flexible Al–O bonds with relatively weaker covalency. This approach successfully suppressed the formation of O-type stacking during LOR and facilitated the formation of the beneficial Z phase. X-ray absorption spectroscopy (XAS) characterizations and density functional theory (DFT) calculations revealed that the flexible AlO6 octahedrons function as buffer units within the crystal structure modulating Al–O bond strength to alleviate local structural stress. The phase transitions mechanism was further elucidated using in situ X-ray diffraction (XRD), while charge compensation behavior was confirmed through XAS. Benefiting from the Z phase, the LAM cathode exhibited improved Na-ion kinetics and superior rate capability compared to the undoped P2-Na2/3Li2/9Mn7/9O2 (denoted as LM). At a high current density of 1 A/g, the LAM cathode delivered a discharge capacity of 86 mAh/g, demonstrating excellent potential for fast-charging energy storage applications.
2 Experimental sections
2.1 Materials synthesis
To synthesize Na2/3Li1/6Al1/6Mn2/3O2, stoichiometric amounts of Al2O3, Mn2O3, LiOH·H2O, and Na2CO3 were thoroughly mixed and ground in an agate mortar. The resulting powder was pressed into pellets with a diameter of 10 mm. These pellets were then placed in a corundum porcelain boat and sintered in a tube furnace. The sintering process involved heating the sample to 450 °C over 90 min and holding at this temperature for 6 h. Subsequently, the temperature was raised to 900 °C within 90 min and maintained for 12 h under an O2 atmosphere. After sintering, the sample was cooled to room temperature and transferred to a glovebox under an Ar atmosphere.
Na2/3Li2/9Mn7/9O2 was synthesized using the same procedure, with stoichiometric amounts of Mn2O3, LiOH·H2O, and Na2CO3 as starting materials.
2.2 Electrochemical measurements
The working electrode was prepared using 70% (mass fraction) of the synthesized P2-type oxides as the active material, 20% (mass fraction) Super P as the conductive additive, and (10%, mass fraction) polyvinylidene fluoride (PVDF, Sigma-Aldrich) as the binder. To form the slurry, these components were thoroughly mixed and homogeneously dispersed in N-methyl-2-pyrrolidone (NMP, 99.5%, Sigma-Aldrich). The resulting slurry was uniformly coated onto an aluminum foil and subsequently dried at 70 °C under vacuum overnight. The dried electrode film, along with the aluminum foil, was then punched into circular slices with a diameter of 10 mm for use as the working electrode.
The electrochemical performance of the prepared samples was evaluated using CR2032 coin-type cells. The assembled cells consisted of the fabricated working electrode, a sodium metal as counter electrode, a GF/D (Whatman) separator, and 1 mol/L NaClO4 electrolyte dissolved in a solution of propylene carbonate and ethyl carbonate (1:1 by volume) with 5% (volume fraction) fluoroethylene carbonate (FEC) as an additive. Galvanostatic charge/discharge (GCD) tests were performed on a Neware CT-4008Tn battery test system within a voltage range of 1.5–4.5 V voltage at 25 °C. Differential capacity (dQ/dV) profiles were derived from the GCD data using the same system. Glvanostatic intermittent titration technique (GITT) measurements were also performed on the Neware CT-4008Tn at a current density of 20 mA/g within same voltage range. The GITT protocol involved a 30-min current pulse followed by a 2-h relaxation period. Electrochemical impedance spectra (EIS) was conducted using the Gamry electrochemical workstation with a 10 mV amplitude over a frequency range from 0.01 to 105 Hz. Cyclic voltammetry (CV) tests were performed at various scan rates within the 1.5–4.5 V range using the same workstation.
2.3 Materials characterizations
A Rigaku Mini Flex 600 X-ray diffractometer was utilized to characterize powder XRD patterns using Cu-Kα radiation (
λ = 1.54 Å) at 40 kV and 15 mA. The obtained data were refined using the GSAS II software based on the Rietveld refinement method [
21].
In situ XRD measurements were conducted using a Bruker D8 Advance diffractometer with Cu-Kα radiation (
λ = 1.54 Å), scanning over a range of 10° to 70°. The
in situ electrochemical cell was tested on a Neware CT-4008Tn battery test system, operated at a current density of 30 mA/g within a voltage window of 1.5–4.5 V. Beryllium windows were incorporated into the
in situ cell to enable X-ray transmission.
Scanning electron microscopy (SEM) image was obtained using a field emission scanning electron microscope (JSM-7800F). High-resolution transmission electron microscopy (HRTEM) images and energy-dispersive X-ray (EDX) spectroscopy were obtained using a Talos F200X G2 microscope. X-ray photoelectron spectroscopies (XPS) measurements were performed using an AXIS Ultra DLD energy spectrometer.
XAS characterizations were conducted at beamline 7-BM (QAS) of the National Synchrotron Light Source II (NSLS-II), Brookhaven National Laboratory. Soft XAS measurements were performed at beamline BL02B02 of the Shanghai Synchrotron Radiation Facility (SSRF). XAS data were processed and normalized using the Athena software package [
22], and wavelet transformed EXAFS results were analyzed using the HAMA Fortran Version software package.
2.4 Theoretical calculations
First-principles calculations were performed based on DFT using the Vienna
Ab-initio Simulation Package (VASP) [
23]. The projector augmented wave (PAW) method was utilized to describe the interactions between ions and valence electrons [
24,
25]. The Perdew-Burke-Ernzerhof (PBE) functional was employed to account for exchange-correlation effects [
26,
27]. To accurately describe strong correlation effects, the Hubbard model of DFT + U was applied, with U values adopted from Refs. [
28,
29]. The plane-wave energy cutoff was set to 400 eV. The convergence criteria were set to 0.02 eV/Å for atomic forces and 1×10
−5 eV for total energy.
For geometry optimizations, a k-point mesh density of 0.4 Å
−1 was adopted, while denser meshes were utilized for density of states (DOS) calculations. Van der Waals interactions were considered using the DFT-D3 method developed by Grimme [
30]. The climbing image nudged elastic band (CI-NEB) method was utilized to evaluate sodium ion migration barriers along different diffusion pathways [
31,
32], as implemented in VASP. Crystal orbital Hamilton population (COHP) analyses were performed using the Lobster program [
33].
3 Results and discussion
3.1 Structural design and analysis
The LAM cathode material was synthesized via a conventional solid-state sintering process conducted in an oxygen atmosphere. The structural properties of the obtained LAM were characterized using powder XRD (Fig. 1(a)) and further analyzed through Rietveld refinement (Tables S1 and S2, Supplementary Material). The results confirm that the synthesized sample exhibits a typical layered P2 structure with P63/mmc symmetry. The atomic site occupancies of Al, Li, and Mn within the crystal structure are equivalent, indicating their presence in the transition metal layers. Additionally, inductively coupled plasma (ICP, Table S3) analysis revealed a Li:Al molar ratio of approximately 1:1, consistent with the designed stoichiometric composition.
To determine the oxidation state of Mn, XAS was performed on the pristine LAM, with the results displayed in Figs. 1(b) and S1. The corresponding energies at half intensity for LAM and the standards were determined to be 6552.6 eV (LAM), 6553 eV (MnO2 standard), and 6551.8 eV (LiMn2O4 standard). Based on the Mn valence states of the two standards, the Mn valence in LAM was calculated to be approximately +3.83, suggesting a mixture of 83% Mn(IV) and 17% Mn(III). The presence of a small amount of Mn(III) can be attributed to Mn reduction occurring on the particle surface. Therefore, the majority of the capacity contribution arises from oxygen redox activity.
XPS was performed to investigate elemental composition and valence states. According the full-spectrum analysis of LAM (Fig. S2(a)), Na, Mn, O, C, Li, and Al elements were detected in the LAM. The oxygen spectrum was deconvoluted into two peaks (Fig. S2(b)): the peak at 531.1 eV corresponds to metal carbonates, and the peak at 529.0 eV is attributed to metal oxides. The Mn 2p XPS spectrum (Fig. S2(c)) shows the characteristic Mn(IV) 2p1/2 and 2p3/2 peaks, confirming a Mn valence state of approximately +4, consistent with the XAS results. Al 2p was also detected (Fig. S2(d)), confirming the presence of Al(III) in the sample.
The morphology of the LAM particles was characterized by SEM and HRTEM (Figs. S3 and S4), revealing hexagonal-shaped particles with diameters of 0.6–1.0 μm and a thickness of around 200 nm. The interlayer spacing was measured to be 2.4 Å, corresponding to the (100) lattice plane of the P63/mmc, consistent with XRD refinement results. Moreover, EDX elemental mappings (Figs. S4(d)–S4(h)) confirmed the presence and homogenous distribution of Na, Al, Mn, and O elements. These characterization results validate that the obtained LAM corresponds to the designed Al-substituted P2-type sodium layered oxide structure with substitution in the transition metal layers.
Based on the above findings, the crystal structure of the synthesized LAM cathode is illustrated in Fig. 1(c). The P2-type layered structure consists of alternating Na layers and transition metal layers. The transition metal layers comprise edge-sharing MnO6, AlO6, and LiO6 octahedrons, while the Na layers feature NaO6 trigonal prisms. The oxygen stacking sequence connecting the transition metal and Na layers follows an “ABBAABBA” pattern, a typical feature of the P2 configuration. Along the c-axis of the LAM unit cell, a “Na−O-Li” configuration is observed (Fig. 1(d)), which contributes to lattice oxygen redox (LOR) activity due to the ionic nature of the Li–O bond. Within the transition metal layers, the local “Al–O(2)–Li” configuration is also identified, aligning with designed compositional strategy. The edge-sharing geometry between AlO6 and LiO6 octahedra leads to mutual structural interactions, influencing the local bonding environment and potentially modulating redox behavior and phase stability.
3.2 Electrochemical performance
The electrochemical performance of the synthesized LAM cathode was evaluated using GCD tests. To highlight the improvements achieved by Al substitution, an undoped P2-Na2/3Li2/9Mn7/9O2 cathode (denoted as LM) was used as a reference. The LM cathode was prepared under identical conditions to those of LAM (see Section 2 for details). XRD patterns of LM and corresponding rietveld refinement results confirm the same P63/mmc symmetry as LAM (Fig. S5 and Tables S4 and S5).
As shown in Fig. 2(a), during charge at a current density of 100 mA/g, both LAM and LM exhibit a distinct plateau in their GCD curves characteristic of LOR activity. The charge capacity of LAM reaches 118 mAh/g, corresponding to the extraction of 0.380 Na ions per formula unit. For LM, the first cycle charge capacity is 111 mAh/g, equivalent to the extraction of 0.379 Na ions. The Coulombic efficiency of LAM exceeds 100%, indicating that the amount of Na+ inserted into the crystal during the sodiation surpasses the amount extracted during the initial desodiation. The nearly identical Na+ extraction quantities for both samples suggest comparable Na+ extraction capabilities, likely due to their similar structural symmetry and initial Na+ content.
During discharge, while the GCD curves of LAM and LM share similar shapes, featuring a short plateau above 4.0 V and a long slope below 3.0 V, an inflection point at 3.0 V is clearly observed for LAM, whereas LM displays voltage hysteresis. This inflection point marks the boundary between two chemical reactions, corresponding to the transition point of redox peaks in the dQ/dV profile (Fig. 2(b)). The region before at 3.0 V in Fig. 2(a) is attributed to oxygen redox, while the process beyond this point corresponds to Mn redox. A distinct boundary at 3.0 V is also evident in the redox peaks in Fig. 2(b), related to the inflection point, signifying different electrochemical processes occurring during discharge. Moreover, the gap in the discontinuous dQ/dV profiles between charge and discharge reflect polarization effects in the two electrochemical processes. The observation suggests that Al substitution effectively mitigates voltage hysteresis in LM, resulting in a higher discharge capacity for LAM (172 mAh/g) compared to LM (149 mAh/g). This voltage hysteresis during discharge is also evident in the dQ/dV plots (Figs. 2(b) and 2(c)), indicating performance degradation in the redox processes occurring below 3.0 V. For LAM, the oxidation peak at 4.25 V during charging, as shown in Fig. 2(b), is attributed to oxygen oxidation. The corresponding reduction processes are reflected by the peaks at 4.0 V. The differences in these redox potentials arise from voltage hysteresis and the degradation in the oxygen electronic structure. Additionally, the reduction peak below 2.0 V is ascribed to Mn redox activity. LM exhibits similar redox behavior to those of LAM due to the comparable GCD curve profiles Fig. 2(c).
Figures 2(d), S6 and S7 compare the rate capabilities of LAM and LM. LAM demonstrates superior rate performance across a wide range of current densities from 100 to 1000 mA/g. Even at a high current density of 1 A/g (approximately 10 C), LAM maintains a discharge capacity of 85 mAh/g, whereas LM delivers only 32 mAh/g. The capacity recovery observed when the current density is reduced from 1000 to 100 mA/g reflects both the reversibility and structural robustness of the LAM cathode under high rate cycling conditions. To further demonstrate this reversibility, GCD curves at various rates are compared in Fig. S8. After cycling at different current densities, the LAM cathode retains a discharge capacity of 122 mAh/g at the 32nd cycle, corresponding to 85% capacity retention compared to its discharge capacity of 144 mAh/g at the 5th cycle. Analysis of the GCD curves from the 5th and 32nd cycles reveals voltage hysteresis and polarization effects, which contributes to the observed capacity loss.
These findings underscore the faster Na+ kinetics and enhanced fast-charging capability of LAM. Figure 2(e) depicts the cycling stability of LAM at 100 mA/g, showing a retained discharge capacity of 97 mAh/g after extended cycling, equivalent to 57% of the initial capacity. The GCD curves and dQ/dV profiles above 4.0 V are attributed to oxygen redox processes. As shown in Fig. S8(a), the reversible capacity in this voltage region gradually decreases with cycling, accompanied by significant voltage hysteresis during discharge. Conversely, Mn redox activity, associated with voltages below 2.5 V, also shows capacity decline over repeated cycles. Irreversible capacity loss is further evidenced by the dQ/dV profiles in Fig. S8(b), where the redox peaks in the 4.0–4.5 V and 2.5–1.5 V ranges progressively diminish from the 2nd to the 50th cycle, indicating irreversible redox processes. Additionally, the peak shifts and pronounced voltage hysteresis reflect degradation of the oxygen electronic structure.
The cyclic stability of LAM was also assessed at a high current density of 1 A/g within the voltage range of 1.5–4.5 V, as shown in Fig. 2(f). The initial specific capacity increases during the first cycle, likely due to the activation of μm-scale particles. At this high rate, the initial discharge capacity reaches 88 mAh/g, demonstrating excellent Na-ion diffusion kinetics and fast-charging performance. After 100 cycles, the discharge capacity stabilizes at 55 mAh/g, representing 62.5% retention of the initial capacity. The observed capacity degradation during high-rate cycling may be ascribed to irreversible degradation of the oxygen electronic structure and particle pulverization. While LAM exhibits excellent rate capability, capacity decay during cycling can be attributed to irreversible LOR processes, which will be further analyzed through variations in electronic structures in the subsequent sections.
3.3 Phase transition mechanisms
The phase transitions and structural evolution of LAM were investigated using
in situ XRD. The noticeable (002) peak at 15.8° shifts to a lower angle before 4.2 V and then returns to its original position after 4.2 V (Figs. 3(a) and S9). This shift corresponds to an increase in the
c lattice parameter during the initial charge process up to 4.2 V, followed by a decrease beyond this voltage (Fig. 3(b)). The initial increase of the
c parameter can be attributed to the enhanced repulsive forces between adjacent transition metal (TM) layers due to Na
+ extraction. Notably, the overall volume change before the phase transformation is minimal, resulting in only an approximately 1.2% volume reduction (Fig. S10). Beyond 4.2 V, the intensity of the (002) peak sharply diminishes, although a weak signal still remains. This behavior is indicative of the formation of the Z phase, which arises from interlayer slipping [
19].
During discharge, the initial reduction in the c lattice parameter is driven by the formation of Na–O bonds within the Z phase. As more Na ions are inserted into the crystal structure, the Z phase gradually transforms back into the P2 phase, resulting in an increase in interlayer spacing. Subsequently, during further sodiation, the interlayer spacing decreases again due to the formation of Na–O bonds within the P2 phase. By the end of the first full discharge cycle, the Na content exceeds that of the pristine state, and the additional Na–O bonds contribute to a reduced c parameter compared to the pristine state. A more significant volume change is observed during discharge, with about a 5% variation between the fully charged and fully discharged states. Throughout the second cycle, the material retains a single P2 phase, with crystal parameters varying reversibly.
Figure 3(c) schematically illustrates the structural evolution from P2 to Z phase, where regular NaO
6 prisms distort into oblique prisms, while the oxygen stacking sequence remains as “ABBAABBA.” Furthermore, the mechanism of Z phase formation is extendable to other transition metal-doped systems. For example, in a recent study on the Mg and Cu co-doped P2-Na
43/60Li
1/20Mg
7/60Cu
1/6Mn
2/3O
2 cathode [
34], the Z phase was detected without significant O-type structure formation, indicating effective suppression of oxygen layer gliding even under highly desodiated conditions. This supports the notion that the Z phase forms via slight and reversible gliding of oxygen layers. Notably, the interlayer distance in the Z phase closely resembles that of the P2 structure, and the local Na environment retains the NaO
6 prismatic configuration, consistent with the observations in this work.
To assess Na
+ diffusion kinetics during bulk redox processes in LAM, GITT measurements were conducted over the first cycle (Fig. 3(d)). The Na
+ diffusivity during discharge ranges from 10
−9 to 10
−11 cm
2/s, comparable to the previously reported Sb-substituted P2-Na
2/3Li
1/4Sb
1/12Mn
2/3O
2 cathode material. During charging, capacity is primarily driven by oxygen redox activity. However, diffusivity varies across different states of charge (SOCs). A sharp decrease in diffusivity is observed between 4.3 and 4.5 V, attributed to significant structural distortion [
11]. In discharge, diffusivity increases below 2.5 V, where oxygen redox dominates and a phase transition from Z phase to P2 phase occurs, suggesting that the prismatic local structure is more conducive to enhancing Na
+ diffusion kinetics. As Na
+ ions further inserting into the crystal, the diffusivity gradually decreases from 2.5 to 1.5 V. In this region, Mn redox activity contributes to charge compensation, enabling additional Na
+ ions to occupy sites within the crystal. The decline in diffusivity can be ascribed to steric hindrance caused by the increasing Na
+ content.
CV tests at various scan rates were also performed to investigate the redox kinetics for LAM. Five distinct peaks (Fig. S11(a)) correspond to Mn oxidation (peak 1), O oxidation (peak 2), O reduction (peak 3), and Mn reductions (peaks 4 and 5). To investigate the diffusion coefficient of Na+ through CV tests, peak currents were extracted from the curves and fitted with the various scan rates (shown in Fig. S11(b)). According to the linearity of Ip and ν1/2, the coefficients of Na+ at the five peaks can be calculated as 4.7×10−10 (peak 1), 4.4×10−11 (peak 2), 2.1×10−11 (peak 3), 1.5×10−10 (peak 4), 1.7×10−10 (peak 5) cm2/s, which are consistent with the results of the GITT tests.
In situ EIS measurements were also performed during the first cycle to monitor variations in charge transfer resistance. Nyquist plots (Fig. S12) show that the semicircle corresponds to charge transfer resistance and the low-frequency straight line reflects diffusion processes. Charge transfer resistance initially increases then decreases during charging, likely due to cathode electrolyte interphase (CEI) formation. Additionally, the slope of the low-frequency line gradually decreases during charging, indicating an increase in diffusion resistance, in agreement with GITT data. During discharge, charge transfer resistance remains relatively stable, suggesting formation of a robust CEI, while a slight decrease in slope corresponds to diffusivity trends observed the GITT. Collectively, these results demonstrate excellent Na-ion kinetics in the LAM cathode, highlighting its promising performance in sodium-ion battery applications.
Notably, the flexibility of Al–O bonds helps mitigate structural distortions during the sodiation, ensuring the stability of the P2 phase and maintaining efficient Na+ diffusion pathways. These findings suggest that suppression of O-type stacking formation can be achieved not only through strong covalent Sb–O bonds but also via weaker but flexible Al–O interactions.
3.4 Charge compensation mechanisms
XAS experiments were performed on the LAM cathode at various states of charge (SOCs) to investigate the charge compensation mechanisms during the initial two cycles. Variations in Mn valence at different SOCs were quantified using XANES spectra collected at the Mn K-edge of the LAM cathode (Fig. 4(a)). At the fully charged state of 4.5 V, the Mn K-edge energy closely matches that of the pristine state, indicating that the Mn valence remains essentially unchanged during the initial charge. This suggests that Mn does not contribute to the capacity in the first charge process, and that the charge capacity originates exclusively from oxygen redox activity. During the first discharge, however, the Mn K-edge shifts from 6552.6 to 6552.2 eV, indicating the reduction of Mn to an average valence of approximately +3.7, corresponding to a mixed Mn(III)/Mn(IV) state. In the subsequent second charge process, the Mn K-edge energy returns to 6552.6 eV, indicating a reversible redox behavior consistent with the first discharge. These XAS results thus confirm that oxygen redox dominates capacity during the initial charge, while both Mn and O contribute to capacity in subsequent charge/discharge cycles.
Soft XAS measurements at the O K-edge were conducted to probe variations in the oxygen electronic structure and provide direct evidence of oxygen redox activity. The pre-edge features of the O K-edge are mainly influenced by the hybridization and energy splitting of transition metal (TM) 3d orbitals and oxygen 2p orbitals within the octahedral crystal field. Since Mn does not contribute to charge compensation during the first charge, observed changes in the pre-edge region are primarily attributed to variations in the oxygen electronic structure. During the initial charge, a noticeable increase in the intensities of the
t2g and
eg peaks is observed (Fig. 4(b)), indicating the formation of empty states above the Fermi level [
35]. This confirms forming oxygen holes during the initial charge, indicating electron loss from oxygen and participation in charge compensation. Upon discharge, the intensities of the
t2g and
eg peaks decrease, reflecting a reversible electron transfer consistent with the Mn valence variations observed in the hard XAS.
Furthermore, Mn EXAFS data were extracted from the hard XAS spectra to investigate the local structural evolution at different SOCs (Fig. S13). The Fourier-transformed EXAFS spectra reveal negligible changes in Mn−O and Mn–TM coordination shells, consistent with the limited variation in Mn valence and electronic structure. The minor presence of Mn(III) can also help mitigate structural distortion since larger amounts of Mn(III) are prone to inducing significant Jahn-Teller distortions. However, wavelet-transformed EXAFS analyses (Figs. 4(c)–4(f)) reveal subtle variations at different SOCs: during the first charge, the Mn–O first coordination shell contracts slightly (highlighted by the dotted area), likely due to electron loss from oxygen and minor distortions of the MnO6 octahedra. In the subsequent discharge process, the Mn–O shell expands back to its pristine state, indicating a reversible process. In the second charge cycle, however, these subtle variations disappear, suggesting that the MnO6 octahedra remain largely intact, consistent with in situ XRD findings, show a stable single P2 phase throughout the entire second cycle.
3.5 Electronic and local structural variations
The local and electronic structural variations of LAM were further investigated through DFT calculations. Figure S14 displays the structural models of the LAM crystal at pristine (Na0.67), desodiated (Na0.29), and sodiated (Na0.90) states, respectively. The corresponding partial density of states (pDOS) for different state of charges (SOCs) are shown in Fig. S15, simulating the electronic structural evolution during the first cycle. In the pristine state (Fig. S15(a)), a narrow energy gap is observed, indicating good electronic conductivity, which is crucial for the excellent rate capability of LAM. Notably, the occupied O 2p states near the Fermi level lie above the Mn 3d states, suggesting that oxygen is more prone to losing electrons compared to Mn, thereby exhibiting the potential involvement in LOR.
According to the Zaanen–Sawatzky–Allen theory, Mn 3d orbitals hybridize with O 2p orbitals within the octahedral crystal field, forming Mn−O bonding and (Mn−O)* antibonding orbitals. Typically, in transition metal oxides, the (Mn−O)* antibonding orbital is partially filled, resulting in the empty upper Hubbard band (UHB) and filled lower Hubbard band (LHB) due to Mott-Hubbard splitting. The on-site Coulomb interaction term
U, defined as the energy difference between the UHB and LHB, quantifies the electron repulsion within the d orbitals [
36]. Additionally, the charge transfer energy Δ, defined as the energy difference between O 2p and (Mn−O)* states, reflects the electronegativity of oxygen. Reversible oxygen redox in transition metal oxides occurs only when Δ ≈
U/2.
In pristine LAM, the (Mn−O)* and O 2p states overlap at the Fermi level, resulting in collective distortions that stabilize the degenerate electronic state upon Na
+ extraction, which creates partially filled states. This distortion induces interactions between (Mn−O)* and O 2p by lowering the symmetry of the MnO
6 structure, forming Mn(3d)–O
2(
σ) covalent bonds [
37]. However, when Δ <
U/2, O 2p states near the Fermi level exhibit minimal overlap with Mn 3d, leading to the formation of unstable oxygen holes and irreversible oxygen redox.
To clarify the relative positions of the Mn 3d and O 2p states, schematic band structures for different SOCs were constructed (Figs. 5(a)–5(c)) based on the pDOS results. For pristine LAM, the UHB, LHB, and O 2p band centers are calculated as 2.15, −2.08, and −1.52 eV, respectively. This indicates that Δ is slightly less than U/2, deviating from the ideal reversible LOR condition (Δ ≈ U/2) and implying a limited ability to stabilize oxygen, which may contribute to the observed limited cycling stability of the LAM cathode. After desodiation, the O 2p center shifts down to −1.88 eV due oxygen hole formation, and it returns to −1.27 eV upon sodiation, consistent with charge compensation behaviors observed in XAS analyses.
To investigate the impact of Al-substitution in the cathode, variations in covalency within the “Al−O(2)−Li” local structure were analyzed using crystal orbital Hamilton population (COHP) calculations (Fig. S16). The integrated −COHP values serve as a measure of covalency, representing the cumulative effect of both negative and positive contributions below the Fermi level. The integrated −COHP values for Li−O are 0.09, 0.095, and 0.10 for the pristine, desodiated, and sodiated states, respectively (Fig. 5(d)), indicating predominantly ionic Li−O bonding that remains nearly constant throughout the charge/discharge process.
In contrast, the integrated −COHP values for Al−O are significantly higher at 0.646, 0.640, and 0.675 in the pristine, desodiated, and sodiated states, respectively. These values lie between those of Mg−O/Li−O and Sb−O bonds, reflecting moderate covalency and indicating limited orientation and flexibility of the Al−O bonds. While the Al−O bond covalency remains essentially unchanged during desodiation, it increases upon sodiation. Based on the charge compensation analysis, the Mn−O bond elongates due to the reduction of Mn, leading to an increase in the occupied states of the Al−O hybridized orbitals. This suggests that flexible Al−O bonds accommodate asymmetric changes in the occupied states.
Structurally, the contraction of Al−O bonds and distortion of the AlO
6 octahedron mitigate structural evolution, stabilizing Na
+ diffusion channels and maintaining steady Na
+ diffusion within the crystal lattice. Furthermore, Na
+ migration simulations, as displayed in Fig. 5(e) reveal a maximum energy barrier of 0.47 eV, comparable to previously reported Sb-substituted cathodes [
19].
The mechanisms by which Al−O bonds stabilize the local structure is described in Fig. 5(f), where MnO6 and AlO6 octahedron share a common edge, causing oxygen’s electronic structure to be influenced by both Mn and Al. During discharge, as Mn−O bonds elongate due to Mn reduction, increased Al–O covalency compensates for asymmetrical changes in the local structure. This enhanced Al–O covalency helps stabilize the local structure, ensuring its stability throughout the process.
Thus, the suppression of O-type stacking during Z-phase formation facilitates Na+ diffusion, while the flexibility of Al–O bonds mitigates structural distortions through bond contraction. These effects collectively ensure stable Na+ diffusion within the crystal lattice and contribute to the superior rate capability of the material. Although the cycling stability of LAM remains limited, this Al-substitution strategy and the accompanying electronic structure analysis may inspire further exploration of more effective approaches to simultaneously enhance both kinetics and cycling stability.
4 Conclusions
In summary, Z-phase generation was successfully achieved and fast Na+ kinetics was enabled through Al substitution in a P2-type Na2/3Li1/6Al1/6Mn2/3O2 cathode. The phase transition and charge compensation mechanisms were systematically investigated. The improvements in the cathode can be summarized in two key aspects:
(1) Suppression of O-type stacking: In situ XRD analysis reveals that Z-phase formation occurs during the first charge process instead of O-type stacking. This Z-phase formation exhibits a lower energy barrier, facilitating Na+ diffusion.
(2) Flexible Al–O covalent bonds: DFT calculations and XAS characterizations indicate that the flexible nature of Al–O bonds accommodates asymmetric variations in occupied states during the sodiation process. This flexibility mitigates local structural distortions through Al–O bond contraction, thereby stabilizing the crystal structure.
These structural enhancements maintain stable Na+ diffusion channels, enabling the LAM cathode to achieve excellent rate capability (86 mAh/g at 1 A/g) and demonstrate significant potential for fast-charging applications. The detailed analysis of electronic structural variations provides valuable insights to guide the development of more effective strategies for further enhancing the electrochemical performance of P2-type cathodes.
PDF
(3810KB)
