Main

Lithium-ion insertion and extraction compounds based on layered oxide frameworks are widely used as cathode materials in high-energy-density Li-ion batteries1,2,3,4,5,6,7,8,9. Owing to the ionic radius of the redox species and the Coulomb repulsion of oxygen, the anisotropic lattice strain and stress caused by Li-ion insertion and extraction result in the accumulation of lattice strain. Moreover, the Fermi energy level of the transition metal (TM) ions in the high-nickel oxide cathode material continuously decreases during delithiation, which causes the energy bands of the TM 3d orbitals t2g and O 2p to overlap, leading to the participation of oxygen in charge compensation10,11. During deep charge compensation, lattice O2− loses electrons and forms highly reactive oxygen species12,13. The reactive oxygen species are compounded and released as oxygen gases; after reacting with the electrolyte, the gases are released in the form of CO2 and CO, causing thermal runaway of the battery14,15. In addition, lattice oxygen vacancies result from the release of oxygen diffusing into the bulk and accumulating on specific crystal planes, resulting in cracks and the formation of new surfaces16,17,18,19. At the same time, oxygen vacancies also accelerate the migration of TM ions, which, together with strain accumulation, produce an irreversible phase transition from layered phase to spinel phase to rock-salt phase, decreasing the battery performance20,21,22. This is widely considered to be the primary reason for the deterioration of high-capacity layered Ni-rich cathode materials. Therefore, the high stability of oxygen anions in layered oxides during Li-ion insertion/extraction is crucial for developing high-capacity cathode materials. Despite numerous efforts in material physics and synthetic chemistry to develop high-capacity layered oxide cathode materials, enhancing the stability of the oxygen anions in layered high-nickel oxides remains a challenge.

Previous studies have primarily focused on three aspects to enhance the stability of oxides: bulk doping to improve the structural stability of materials, surface coating to suppress surface side reactions, and the preparation of gradient materials and single crystals to enhance stress–strain stability23,24,25,26,27. Among these, doping strategies have become universal in numerous studies for their ability to modulate the microstructural morphology. For example, doping high valent Ta5+ or W6+ cations into the bulk of high-Ni cathode can form radially distributed particles with grain refinement, showing great ability to mitigate lattice strain within long cycles28,29. However, while the lattice strain is an inherent feature highly related to electronic structure and oxygen evolution, the functioning mechanism of doping strategies remains inadequate in the local structural scale. Therefore, there is an urgent need for doping strategies to simultaneously tune the particle morphology and local atomic structures. Furthermore, deep understanding of effective doping methods is required to suppress thermodynamically irreversible phase transitions and enhance the materials stability. Ordered structures possess low thermodynamic energy and good structural stability30,31. Tellurium has a 5d unoccupied orbital, suitable ionic radius and good lattice matching with layered high-nickel oxide32.

Here we show that constructing Ni6Te ordered structures exhibiting highly thermodynamically stable states within the TM layers can significantly enhance the stability of lattice oxygen. Moreover, the high valent Te6+ cations can potentially refine the particle morphology which can eliminate stress–strain effect. We have prepared layered ultrahigh-nickel oxide LiNi0.94Co0.05Te0.01O2 with a refined particle morphology and Ni6Te ordered structure (NC95T). The combined enhancement in particle size and atomic arrangement leads to a multi-effect improvement in lattice strain and oxygen stability. The resulting structure demonstrates an increased metal–oxygen binding energy and interlayer spacing (Supplementary Table 1), mitigating the lattice strain at high potential and preventing cooperative lattice distortion. Furthermore, the significantly lower oxygen band centre after Li removal suggests that the Ni6Te ordered structure effectively enhances the lattice oxygen stability, preventing cumulative stress–strain and irreversible phase transitions. The as-prepared NC95T cathode exhibits an ultrahigh specific capacity and excellent cycling stability.

Results

Materials characterization

The X-ray diffraction (XRD) patterns of NC95T showed a structure identical to that of layered α-NaFeO2 in the R\(\bar{3}\) m space group; however, interestingly, weak reflection peaks appeared in the 2θ range of 20–25° (Supplementary Fig. 1). The results indicate the existence of ordered distribution33 in NC95T, which is similar to the ordered structure of LiM6 in Li2MnO3 (ref. 24) and different from that of the pristine LiNi0.95Co0.05O2 (NC95) (Fig. 1a,b). Using density functional theory (DFT) computation, the ordered arranged Te cations manifest lower total energy than the randomly distributed model, as is shown in Supplementary Fig. 2. This indicates that formation of the Te–Ni–Ni–Te superstructure is more thermodynamically favourable. X-ray photoelectron spectroscopy (XPS) and Te L-III edge34 measurements confirmed that Te existed as Te6+, and the XPS signals of Te at 576.2 eV (3d 5/2) at different etching depths revealed that tellurium was uniformly distributed in the bulk (Supplementary Fig. 3), which was consistent with the energy-dispersive spectroscopy mapping results (Supplementary Fig. 4). Three-dimensional (3D) atom probe tomography (APT) also shows a uniform spatial distribution of Li, Ni, Co, Te and O elements in the bulk (Fig. 1c). The measured element count ratios are in good agreement with the theoretical expectations of the material design (Supplementary Fig. 5). It is worth noting that TeNi (average valence: +4) peaks were also detected in the mass spectrum (Supplementary Fig. 5); this is also evidence of strong interaction due to Te/Ni proximity in the TM layer. While 3D APT is not precise at atomic scale to determine the distributing pattern of specific atoms, it is more visual in high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images. The observed images along [001] direction show clusters of connected NiTe6 hexagons that contain one brighter cation in the centre, indicating the superstructure formation in the lattice (Fig. 1d and Supplementary Fig. 6). In contrast, the pristine NC95 only shows uniform cation intensity (Supplementary Fig. 7). The HAADF-STEM results in the TM layer further show an ordered cation arrangement of ‘two bright and one brighter’ of NC95T (Fig. 1e), corresponding to Te–Ni–Ni–Te atoms arrangement (Fig. 1f,g), which is due to the difference in cation contrast and confirms the ordered structure observed using XRD (Supplementary Fig. 1). The statistical quantity of the tellurium concentration is around 1.14 atomic percent (at%) according to three observed areas, which is further verified via Rietveld refinements of NC95T where approximately 1.2 at% Te substituted at the Ni site (Supplementary Fig. 8). Introducing Te with a high valence state into the TM layer increases the Ni2+ content in the material (Supplementary Fig. 9), resulting in a slightly higher degree of Li/Ni disorder (~1.9% in Supplementary Table 2) in NC95T than in NC95 (1.73%), as determined by neutron diffraction refinement (Supplementary Fig. 10). Minor variations in the Li/Ni mixing ratio appear to have a negligible impact on the electrochemical performance35. The refinement also highlights an increase in c-spacing (Supplementary Table 1), attributed to the generation of Ni2+ ions, which have larger ionic radii than Ni3+, for charge compensation by adding Te6+. In addition, using transmission Kikuchi diffraction (TKD) to detect the entire particle morphology shows that the introduction of high-valent tellurium clearly refines the grains (Supplementary Fig. 11), which can effectively improve the stress–strain property during long cycling. But unlike tungsten and tantalum elements36, the primary grains do not have an obvious preferred orientation and preferential growth37. These results confirm the successful synthesis of LiNi0.94Co0.05Te0.01O2 with a refined particle morphology and a stable Te–Ni–Ni–Te ordered structure.

Fig. 1: Crystal structure analysis.
figure 1

a,b, XRD pattern with Rietveld refinement of NC95 (a) and NC95T (b). Y_Obs. is experimental data, Y_Cal. is software fitted data, Y_Differ. is the difference value between Y_Obs. and Y_Cal, and Rw is weighted graphical residual variance factor. c, 3D atomic probe image. Scale bar, 20 nm. d, HAADF-STEM of NC95T along [001] direction. The hexagonal boxes mark the featured Ni6Te honeycombs which connect each other. e, Spherical aberration-corrected HAADF-STEM of NC95T along [110] direction in uncharged state. Colour rendered by DigitalMicrograph software; scale bar, 5 nm. f,g, Corresponding intensity profiles of selected regions 1 (f) and 2 (g) in e.

Source data

Full size image

Electrochemical performance

To highlight oxygen stability, a cut-off charge voltage of 4.6 V was used. NC95T displays a high capacity of 239 milliampere-hours (mAh) per gram at 0.1 C, slightly higher than that of NC95, and excellent cycle stability of 94.5% after 200 cycles at 0.5 C (Fig. 2a,c and Supplementary Table 3). In contrast, NC95 exhibited a retention rate of 59.2%. As shown in Fig. 2b, the width of the H2–H3 phase transition23 peaks (~4.2 V) at half-maximum of NC95T was around 2.2 times that of NC95. The broadening of the phase transition reaction potential maximizes the high-nickel cathode’s capacity and helps prevent structural damage caused by stress concentration23. After 200 cycles, NC95 exhibited a considerable voltage decay of 0.2568 V, along with an average voltage polarization four times higher than that of the initial cycle (Fig. 2d). In contrast, NC95T showed negligible voltage decay and an average voltage polarization of only 0.1585 V. The C-rate cycling in Fig. 2e shows an improved ordered structure for fast charging performance. After undergoing high-C-rate cycling, the NC95T sample demonstrated good capacity recovery, whereas NC95 exhibited a visible decline in capacity. Even at a high temperature of 55 °C, NC95T also shows excellent cycling stability (87% in Fig. 2f), which is considerably higher than that of NC95 (33%).

Fig. 2: Electrochemical performance.
figure 2

a, Initial charge/discharge curves of NC95 and NC95T. b, dQ/dV curves of NC95 and NC95T. c,d, Cycling performance (c) and average voltage (d) of NC95 and NC95T at 0.5 C current density in the voltage range of 2.7–4.6 V. e, C-rates performance. f, Cycling performance at 55 °C. Data are presented as mean values ± s.e.m. and the sample size is 3.

Source data

Full size image

Additionally, when the cut-off charge voltage is limited to 4.4 V, NC95T exhibited an excellent retention rate of nearly 99% after 100 cycles at 0.5 C. When lithium dendrite-free Li4Ti5O12 is used as the negative electrode, the retention rate is still greater than 83% after 1,000 cycles at 10 C (Supplementary Fig. 12), indicating the superior structural stability. The as-prepared pouch cell (Li anode) shows a high monomeric energy density of 545 watt-hours (Wh) per kilogram (Supplementary Fig. 13 and Supplementary Table 5) and the energy density of lithium-ion batteries prepared by combining silicon anodes reaches an unprecedented high value of about 404 Wh kg−1. Compared with the 24% retention rate of NC95 after 300 cycles, the NC95T sample still has an excellent energy retention rate (91.2%) of 300 cycles at the rate of 1 C. The mechanism of cyclic stability needs to be further revealed.

Structural stability during cycling

The structural evolution of NC95 and NC95T was studied using in situ XRD, spherical aberration electron microscopy and nickel K-edge absorption spectra (Ni K-edge XAFS). As lithium removal, NC95 and NC95T exhibit a typical H1–M–H2–H3 phase transition process (Fig. 3a and Supplementary Figs. 14–17), consistent with previously reported results38. During the initial charging process, diffraction peaks of NC95 demonstrate pronounced biphasic behaviour, but this diminishes in subsequent processes, which is consistent with Chapman’s work39. However, the as-prepared NC95T displays solid-solution behaviour during the initial cycle, which may indicate the high surface stability in ambient air, confirmed by XPS measurement (Supplementary Fig. 16). Furthermore, during the H1–H2 phase transition process, the peak intensity and angle changes of NC95T are smoother compared to NC95 (Supplementary Fig. 17), indicating the excellent stability of its bulk structure. Beyond 4.2 V, both NC95 and NC95T exhibit an H2–H3 phase transition, with the disappearance of the pillar effect in the lithium layers leading to lattice contraction. At the end of the initial cycle, under the same delithiation state, the maximum deviation in 2θ angle for the (003) peak of the NC95 sample is 1.31°. By contrast, the maximum deviation in 2θ angle for the (003) peak in NC95T is 0.98°, reducing by 25% compared to the NC95 sample, indicating superior preservation of the layered structure during lithium extraction. Notably, the superlattice diffraction peak does not disappear in the fully charged state with 0.5 C (Supplementary Fig. 18) and tellurium did not segregate after long cycles (Supplementary Fig. 19), which further explains the beneficial effect of a stable ordered structure on structural stability.

Fig. 3: Structure stabilization mechanism.
figure 3

a, In situ XRD of NC95T of the first two cycles. b, Corresponding calculated lattice strain. c, Scanning electron microscopy cross-sectional view after 100 cycles. Scale bar, 10 nm. d–f, HADDF-STEM of NC95T after 100 cycles in the voltage range of 2.7–4.6 V (d,e) and localized strain distribution analysed by GPA (f). Scale bars, 1 μm (d) and 5 nm (e). g, Ni K-edge before and after the cycle. μ(E) is all weights between 0 and 1. h,i, EELS O K-edge (h) and Ni L-edge (i) after 100 cycles from the surface (0 nm) to the bulk area (15 nm).

Source data

Full size image

As shown in Fig. 3b, the lattice strain of the materials during charging and discharging was determined using the Williamson–Hall equation38. During the charging process, the NC95 and NC95T strains initially increased and then decreased, which is consistent with previous literature6. NC95 exhibits greater strain variations during the charging process. In particular, before 4.0 V, the coexistence of the H1 and M phases in the NC95 materials results in asynchrony between the strain and lattice changes, potentially leading to local stress concentration and slow kinetics40. In the H2–H3 phase transition stage, NC95T exhibited small strain fluctuations of ~0.05%, which were about three times lower than NC95. The large strain changes during the charge–discharge processes may lead to lattice oxygen loss, TM migration and lattice mismatch. The small strain fluctuations of NC95T in the highly delithiated state contribute to its enhanced structural stability. This smaller lattice change is also shown in the ex situ XRD of the highly charged samples (Supplementary Fig. 20 and Table 4). Therefore, the migration trend of TMs in long-range structures can be qualitatively characterized by calculating the lattice mismatch ratio along the c-direction. As illustrated in Supplementary Fig. 21, the lattice mismatch ratio of NC95T consistently remained below zero during both charging and discharging, suggesting a weak inclination toward the NiO rock-salt phase formation35. This transformation is further discernible in the Ni K-edge absorption spectrum, as illustrated in Supplementary Figs. 22 and 23. After 100 cycles, the Ni K-edge of the NC95 specimen shifted markedly toward the lower-energy side and a conspicuous Ni2+–O bond length (indicated by a dashed line) also displayed in the first coordination shell of extended X-ray absorption fine structure (EXAFS). In comparison to NC95T, NC95 exhibits a substantial decrease in peak intensity in the second coordination shell after long-term cycling, indicating an increase in local structural disorder degree41. This is also consistent with the XRD result (Supplementary Fig. 24) after long-term cycling. These results indicate that the Te–Ni–Ni–Te ordered structure effectively alleviates the lattice strain at high potentials, thereby preventing particle pulverization (Fig. 3c), the formation of the rock-salt phase (Fig. 3d,e) and aggravation of side effects during long-term cycling (Supplementary Figs. 25–29).

Furthermore, at the atomic scale, as shown in Supplementary Fig. 30, the original NC95 sample had numerous point (rock-salt phase) line defects (lattice cracks) on the primary particles, with fragmentation and amorphization between the primary particles after a long cycle. The geometrical phase analysis (GPA) mapping also shows that the particle strain changes obviously. In contrast, the GPA (Fig. 3f) indicates that NC95T had almost zero defects and zero strain, with observable changes at the particle edges. In addition, the electron energy loss spectrometer (EELS) lattice O K-edge and Ni L-edge after cycling (Fig. 3h,i) showed that the NC95T sample exhibits only a NiO rock-salt phase at the 2 nm scale; the thickness of the rock-salt phase of the NC95 sample was more than 30 nm (Supplementary Fig. 31), indicating that NC95T has excellent structural stability during long cycling, which is consistent with the Ni K-edge absorption fine structure (XAFS) results (Fig. 3g). Macroscopic and microscopic analyses revealed that materials with a TM layer ordered structure could effectively alleviate lattice strain during cycling and avoid numerous side reactions caused by oxygen loss.

Stability of lattice oxygen framework

The K-edge XAFS results obtained during the charging process offer intriguing insights into the redox behaviour of the sample41. The Ni K-edge in NC95 shifts toward the high-energy side before 4.2 V and subsequently to the low-energy side on charging to 4.6 V owing to the oxygen release (Fig. 4a). In situ differential electrochemical mass spectrometry (DEMS) measurements showed CO2 gas release above 4.3 V (Supplementary Fig. 32), resulting from the catalytic decomposition of the electrolyte by the highly active species12. In contrast, the K-edge of Ni in NC95T shifts monotonically toward the high-energy side on charging to 4.6 V (Fig. 4b). The DEMS measurements did not detect any gas release, indicating that the lattice oxygen activity may be passivated. Moreover, the peak intensity change observed in the radial distribution function for NC95T differs from that of NC95, presenting a monotonic change trend and indicating that the local structure of NC95T remains ordered throughout the charging process (Fig. 4c,d). However, further research is still needed to better understand the properties and behaviour of highly reactive lattice oxygen.

Fig. 4: Electronic structure evolutions during charge and discharge processes.
figure 4

a,b, Ni K-edge XANES of charging state of NC95 (a) and NC95T (b). c,d, Corresponding Fourier transform magnitudes of Ni K-edge EXAFS spectrum during the initial charge processes of NC95 (c) and NC95T (d). χ(R) is the result of transforming χ(k), the oscillatory function in wavevector k-space, into real space R. e,f, RIXS for O K-edge of NC95 (e) and NC95T (f) with excitation energy at 531 eV. g–j, O K-edge XAFS during the charging process of NC95 (g,i) and NC95T (h,j).

Source data

Full size image

In the total electron yield mode of O K-edge XAFS (Fig. 4g,h), the absorption peaks at 528.9 eV (t2g) and 532 eV (eg) correspond to the transitions from the O 1s orbitals to the empty TM d orbitals mixed with O 2p orbitals42, which provides essential information about the distribution of surface oxygen hole states and effective charges on the oxygen atoms. The peak located at 534 eV mainly stems from Li2CO3, and the nearly invisible peaks in NC95T indicate minor surface byproducts during the charging process, consistent with the XPS result. As shown in Fig. 4g,i, the continuously increasing t2g unoccupied state before 4.4 V, relatively decreasing eg state and decreasing Ni L-edge peak intensity ratio indicate that the TMs are the only redox species near the surface. Therefore, the release of CO2 gas detected by the DEMS (Supplementary Fig. 32a) during charging is mainly related to Ni4+-catalysed electrolyte decomposition. When charged to 4.6 V, a sharp increase in the eg state indicates the participation of anions in charge compensation. The combined effect of the high catalytic activity of Ni4+ and highly active oxygen exacerbates the deterioration of battery performance. Conversely, for NC95T, the continuously increasing eg unoccupied state before 4.4 V and the almost constant t2g state and Ni L-edge intensity ratios indicate that anion charge compensation occurred on the surface of the NC95T material (Supplementary Fig. 33). The DEMS result (Supplementary Fig. 32b) shows that no gas is released during the charging process. Subsequently, after charging to 4.6 V, the valence state of Ni on the surface was reduced by the reductive coupling between TM–O. Low-valence Ni with weak catalytic activity is beneficial for lowering interfacial side reactions and improving cathode–electrolyte interphase stability.

Neither NC95 nor NC95T shows obvious O redox characteristic signals near the 523.7 eV emission energy in resonant inelastic X-ray scattering (RIXS) spectra before 4.4 V (Supplementary Figs. 34–36), indicating that there was no reversible O redox reaction in the bulk43. As shown in Fig. 4e, compared with 3.8 V, the peak shape at 4.6 V changed significantly, but no apparent inversion peak of the O redox was observed, reflecting the lack of substantial reversible O redox in NC95 owing to lattice oxygen loss, which resembles the behaviour of Li2MnO3 materials reported earlier44. Compared to NC95, the peak shape and area of the NC95T at 4.6 V are similar to those at 3.8 V (Fig. 4f), and the apparent d–d excitation signal at 2.4 eV from the resonance elastic peak indicates an increase in the ionicity of TM–O in the bulk. Although the peak signal of O redox was enhanced when charged to 5.0 V, the peak area (~524 eV) of the NC95T decreased by about 10% compared with NC95 (Supplementary Table 6). Combining the Ni K-edge, total electron yield and RIXS results, the NC95T ultrahigh-nickel material evidently exhibits excellent lattice oxygen stability.

Discussion

The grain refinement phenomenon in NC95T cathode materials can eliminate the internal strain and avoid severe particle pulverization. Moreover, the smaller size of the particles can shorten the Li diffusion channel, making it better for releasing more capacity within varied rates45. In the representative works related to microstructure modulation, the primary particles can grow along (003) surface with grain refinement and arrange in a specific direction, which become the two key reasons in the microstructural change for better stability45,46,47,48,49. Under this scenario, the consistent expansion and contraction of the primary particles can inhibit the electrolyte corrosion and alleviate the stress–strain effect along the c axis. Nonetheless, NC95T shows grain refinement without particle arrangement and preferential growth along a specific direction, as revealed by the TKD results. Because of the lack in these two primary factors, the role of particle morphology in the electrochemical performance should be limited in our system.

A representative work29 has interpreted this kind of phenomenon well. In this study they synthesized a Ti-doped high-Ni cathode in which an obvious grain refinement shows in the particle morphology. Despite the promoted cycling stability compared to the pristine sample, the absence of primary particles with radial arrangement from the inside out leads to limited promotion and still bears the problems of particle cracks, retaining 76.5% of initial capacity after 1,000 cycles at 1 C in the range of 3.0–4.2 V in the pouch cells. However, the counterparts of Ta- and Mo-doped cathodes with radial arrangement of primary particles can show superior cycling performance and better strain property, with 95% retention after 3,000 cycles at 1 C in the range of 3.0–4.2 V in pouch cells. This demonstrates that radial arrangement of primary particles is the key for enhancing performance in the aspect of morphology. Therefore, in our work, with the absence of the radial arrangement of primary particles, the superior cycling performance and good maintenance of the long-cycled structure indicate an additional functional mechanism promoting the battery performance besides the grain refinement effect.

To further reveal the exceptional stability of lattice structure and oxygen during delithiation, the electronic structure was investigated using DFT. Careful examination of the electronic structure (Fig. 5a) shows the presence of localized O 2p states near the Fermi level owing to orbital hybridization between O 2p and Ni 3d. Compared with NC95, NC95T with Ni–Te order structure has more occupied states below the Fermi level, which could be attributed to the effective modulation of the local TM electron structure through the superlattice structure, resulting in decreased unoccupied energy level state by ~0.08 eV, showing a smaller bandgap and higher electronic conductivity.

Fig. 5: Stability of oxygen.
figure 5

a, The density of states of oxygen. b, Bader charge on oxygen and average Bader charge difference between pristine and Ni–Te ordered sample. c, Oxygen vacancy formation energy (pristine means the NC95 and Ni-Te ordered represents the NC95T). d, Oxygen p-band centre evolution during delithiation process. e, Band structure diagram.

Source data

Full size image

Figure 5a clearly illustrates the changes in the oxygen electronic structure, especially in the fully delithiated state, where localized O 2p states near the oxygen Fermi level emerge in the ordered Ni–Te structure. This occurrence prevents a pronounced tendency for lattice oxygen to lose electrons. The discreteness of Bader charges (Fig. 5b) indicates the partial passivation of oxygen activity. The non-uniform variation of lattice oxygen in the framework structure effectively avoids cooperative lattice distortions. Additionally, the oxygen vacancy formation energy (Fig. 5c) suggests the outstanding stability of lattice oxygen in the ordered Ni–Te structure.

The p-band centre showed the oxidation trend of the lattice oxygen during lithium removal (Fig. 5d,e). Compared with the pristine structure, the ordered structure has a lower p-band centre (Li1), mainly because of the modulation of the electronic structure of the TM layer. During lithium removal, charge compensation is primarily performed by the Ni3+ eg orbitals; therefore, the p-band centre of oxygen does not change obviously (from Li1/2 to Li1/12). However, in the fully delithiated state (Li0), the p-band centre of the ordered structure was significantly further from the Fermi level than that of the original sample. Owing to the reduction of the p-band centre, there will be no apparent overlap of the Ni4+–O energy bands; therefore, the material will not experience anion charge compensation and lattice oxygen loss. Furthermore, the evolution of the Te L-III edge X-ray absorption near-edge structure (XANES) spectra in the local structure shows the invariance of the Te valence state and reversibility of the local coordination environment during charge–discharge cycles, which might relate to the symmetry change in the Te–O bond length along the c-direction in the Ni–Te ordered structure calculated by DFT (Supplementary Fig. 37). Symmetric stretching of the local structure could also alleviate cooperative lattice distortion during lithium removal and avoid stress concentration to some extent. These results suggest that our strategy of modulating the electronic structure of TM layers and ligands through an ordered structure, relieving high-potential lattice strain and delaying the oxygen reaction potential significantly improves the stability of lattice oxygen in ultrahigh-nickel materials.

To summarize, we report a multi-effect strategy to modulate lattice strain and promote anion stability in layered Te-doped NC95 cathode materials. The as-prepared material exhibits excellent electrochemical performance, with an initial capacity of up to 239 mAh g−1 at a current rate of 0.1 C and a cut-off voltage of 4.6 V while maintaining an impressive 94.5% capacity retention over 200 cycles. The resulting pouch cell using silicon carbon as anode exhibits a high energy density of 404 Wh kg−1. Superior performance of lattice strain mitigation and oxygen stabilization during cycling have been attributed to the particle refinement and formation of a new Te–Ni–Ni–Te ordered structure uniformly distributed within the TM layers. This superstructure can serve as a regulator for O 2p energy level with low thermodynamic energy, effectively maintaining stability of the layered oxides during charge–discharge cycles. This doping strategy of using the Ni6Te ordered structure and simultaneously tuning the particle morphology can further help to achieve highly stable layered oxide crystal structures, with potential applications in advanced high-energy secondary battery systems.

Methods

Preparation of cathode

LiNi0.94Co0.05Te0.01O2 (NC95T) cathode material was synthesized via coprecipitation to prepare the precursor, followed by a high-temperature solid-phase method to obtain the target product. NiSO4·6H2O, CoSO4·7H2O and H6TeO6 were dissolved in deionized water to form a 2 mol l−1 salt solution with a stoichiometric ratio. A 4 mol l−1 solution of NaOH was weighed and prepared for the precipitant and 1 mol l−1 aqueous ammonia was added as the chelating agent. The two solutions were added separately to a reaction kettle with a constant pH of 11.0 at a stirring speed of 900 rpm and a temperature of 55 °C. After 30 h of reaction and 10 h of aging, the obtained precursor was washed several times with deionized water and dried in a 110 °C oven for 12 h to obtain Ni0.94Co0.05Te0.01(OH)2 precursor. The precursor was mixed thoroughly with LiOH·H2O in a stoichiometric ratio of 1:1.02 and placed in a tube furnace, which was heated at a rate of 2 °C min−1 under an oxygen atmosphere of 8 l min−1 to 450 °C and held for 4 h. Subsequently, the reaction mixture was heated at a rate of 1 °C min−1 to 740 °C and held for 12 h to obtain the target product. The reference sample LiNi0.95Co0.05O2 was prepared using the similar synthetic process.

To prepare the cathode for the 2032-type coin cell, the first step was to mix the cathode material completely, using Super P and polyvinylidene difluoride (PVDF) in a mass ratio of 8:1:1. Next, an appropriate amount of N-methyl-2-pyrrolidone was added to the mixture to make a conductive paste, which was then applied to an aluminium foil and dried overnight in a blast oven at 110 °C. Finally, the dried electrode was cut into a disc of diameter 14 mm and the active material loading was maintained at ~3 mg cm−2.

Battery assembly and testing

After preparing the cathode, the coin cell was assembled in a glove box with low water and oxygen content (<0.1 ppm). The anode was lithium metal and a Celgard2320 microporous membrane was used as the separator. The electrolyte used was model LB-365 (Duoduo Cop. 1.2 M LiPF6 in ethylene carbonate:ethyl methyl carbonate (EC:EMC) = 3:7 vol%). The constant current charge–discharge test was carried out on a Neware tester, with a current density of 1 C = 200 mA g−1, at test temperatures of 30 and 55 °C on a NEWARE BTS-4000 battery test system.

Lithium metal batteries

The lithium batteries were fabricated using NC95T as the cathode at a loading density of 25 mg cm−2 (active materials:Super P:PVDF = 94:3:3 in weight) and a lithium strip with a thickness of 20 μm coated onto a 4 μm copper foil as anode. The lithium metal battery consisted of 21 layers of cathodes and 22 layers of anodes, with a total electrolyte (1.2 M LiPF6 in EC:EMC = 3:7 vol% + 2%VC) injection amount of 10.6 g. Initially, they were charged at a current rate of 0.1 C, with a voltage cut-off set at 3.8 V. Subsequently, a charging current rate of 0.3 C was applied, with a voltage cut-off set at 4.3 V, completing the battery activation process. During the initial charge–discharge cycle, the voltage window was set between 2.5 V and 4.45 V, with a current density of 0.05 C. For the subsequent cycling process, the voltage window was adjusted to between 2.5 V and 4.3 V, and the discharge rate was set at 0.2 C.

Lithium-ion batteries

In the case of the lithium-ion battery, anode was prepared using a 6 μm copper foil with a silicon-carbon (1,600 mAh g−1) loading density of 4.09 mg cm−2. The cathode was fabricated using a 10 μm aluminium foil with a loading density of 25 mg cm−2. A total of 15 cathodes and 16 anodes were stacked to assemble the lithium-ion battery, with a total electrolyte (1 M LiPF6 in DMC:EC:EMC = 1:1:1 vol% with 5%FEC,1%VC) injection amount of 10 g. The prepared battery cells were aged for 48 hours. Following the aging process, the cells were placed into a conditioning apparatus for high-temperature conditioning. Initially, they were charged at a current rate of 0.1 C, with a voltage cut-off set at 3.8 V. Subsequently, a charging current rate of 0.3 C was applied, with a voltage cut-off set at 4.3 V, completing the battery activation process. For the initial charge–discharge cycle of the Li-ion battery, the voltage window was set between 2.7 V and 4.45 V, with a current density of 0.05 C. During the cycling process, the voltage window was adjusted to between 2.8 V and 4.2 V (versus silicon), and the charge–discharge rate was set at 1 C.

Ex situ structure characterization

Crystal structure information was obtained by XRD measurements using a D8-Advance tester (Cu Kα radiation source (α = 1.5406 Å) at 40 kV and 40 mA) with an integration time of 5 s in the 2θ range of 10–120°. For neutron powder diffraction, approximately 3 ml of the powdered sample was placed in a quartz tube and Neutron powder diffraction data were collected between 7.5° and 152° with a scan step of 0.07° (wavelength λ = 1.88919 Å) by the high-resolution power diffractometer located at the China Advanced Research Reactor, China Institute of Atomic Energy. The obtained data were then refined using GSAS II software. For TKD, a Thermo Fisher Helios 5 focused ion beam/scanning electron microscope workstation was used to produce thin foils. It was found that the high-quality polished surface of the thin foils is of great importance for achieving TKD maps with a high indexing rate. TKD analyses were performed using a HKL Channel 5 electron backscatter diffraction system with a Nordlys II detector, mounted on a field emission gun scanning electron microscope. An accelerating voltage of 30 kV in high current mode and an aperture size of 120 µm were used to maximize the signal.

3D APT

Three-dimensional APT samples are prepared using the focused ion beam (FEI Helios G4 UC) for single-particle lift-out and annular milling method on prefabricated silicon microtip arrays. Laser-assisted APT analysis was conducted using a CAMECA LEAP 5000 XR atom probe tomography system with a 355 nm ultraviolet laser, 20 pJ laser pulse energy, 40 K specimen temperature and evaporation rate maintained at 0.005 atoms per pulse. APT data were reconstructed and analysed using IVAS v.3.6.6 software.

In situ testing

In situ XRD testing was conducted by uniformly dispersing the active material, Super P and PVDF in N-methyl-2-pyrrolidone in an 8:1:1 ratio, forming a slurry with a certain viscosity. The cathode slurry was dripped onto the centre position of the in situ cell with a Be window and, after it had dried, it was assembled into the in situ cell inside a glove box. Cycling was carried out at a charge–discharge rate of 0.2 C in the voltage range of 2.7–4.6 V for the first two cycles. Bruker XRD data were obtained using a D8-Advance diffractometer (Bruker, Germany) with Cu Kα radiation (λ = 1.5406 Å) operated at 40 kV and 40 mA. Data was collected in the 2θ range of 15–70°, with a step size of 0.02°, and each data point took approximately 15 min to collect. DEMS testing was conducted to detect gas production during the charging process. Approximately 10 mg of the cathode sample was placed in an in situ cell and Celgard membranes and lithium foil were sequentially added as the separator and anode. Then, 40 μl of the electrolyte (LB-365, Duoduo Company 1.2 M LiPF6 in EC:EMC = 3:7 vol%) was introduced. The assembled in situ cell was connected to 0.2 MPa of argon gas, serving as the carrier gas. Volatile products generated by the electrochemical reaction were transported through a hydrophobic membrane interface into the vacuum system of the mass spectrometer. The test was conducted at a current density of 5 mA g−1 (0.05 C) within the voltage range of 2.7–4.4 V.

Electron microscopy characterization

Transmission electron microscopy samples were prepared by cutting the particle into thin sections smaller than 100 nm using a focused ion beam (ZEISS Crossbeam 540) and HAADF-STEM testing was performed using an aberration-corrected electron microscope. EELS testing and analysis were conducted using a Gatan GIF 965 equipped with a FEI Titan Themis Z. The cross-sectional morphology and elemental energy-dispersive spectroscopy mapping distribution of samples after extended cycling were characterized using scanning electron microscopy (Merlin Compact). The cross-sections of the sample were obtained by sectioning with the Leica EM UC6 Ultramicrotome.

X-ray absorption spectroscopy

Coin cells were charged at a rate of 0.1 C to various states of charge. The coin cells were carefully disassembled to obtain the cathode after cycling in the glove box. This process should be conducted to prevent the occurrence of battery short circuits. The obtained electrodes were cleaned with dimethyl carbonate to remove any residual electrolyte, resulting in the target material for absorbance spectrum testing. XAFS were acquired in transmission mode at beamline 1W1B using a Si (111) double-crystal monochromator at the Beijing Synchrotron Radiation Facility (BSRF). Fourier transformation was carried out using Athena software. Te L-edge absorption spectra were collected in FY mode at 4B7A at BSRF. O K-edge XANES data were obtained in total electron yield mode at the XMCD experimental station at the National Synchrotron Radiation Laboratory. RIXS data for O K-edge were collected at the BL20U beamline of the Shanghai Synchrotron Radiation Facility. The distinctive features of the O redox reaction were extracted at an excitation energy of 531 eV. Mapping of resonant inelastic X-ray scattering was performed by collecting emission spectrum signals within the excitation energy range of 527–532 eV at 0.2 eV intervals.

Theoretical calculations

DFT calculations were conducted via the generalized gradient approximation with the inclusion of an on-site Coulomb interaction method within the Vienna Ab initio Simulation Package code for two different compositions, namely Li36Ni36O72 and Li36Ni34Te2O72. The projector-augmented-wave method represented the core electron states and the Perdew–Burke–Ernzerhof exchange-correlation functional. A plane-wave basis set with a cut-off energy of 600 eV was used for the calculations and a 3 × 3 × 2 Monkhorst–Pack k-point mesh was applied for structure optimization. The total energies converged to within 10−5 eV per formula unit and the final forces on all atoms were less than 0.05 eV Å−1.

The oxygen vacancy formation energy, \(\Delta{{E}_{({\rm{O}}_2\,{\rm{release}})}}\), was evaluated using equation (1).

$$\Delta {{{E}}}_{({\rm{O}}_2{\rm{release}})}=2\left({{{E}}}_{({\rm{X}}-{\rm{O}})}+1/2\,{{{E}}}_{({\rm{O}}_2)}-{{{E}}}_{({\rm{X}})}\right)$$
(1)

where E(X−O) and E(X) are the total energies of the oxygen-deficient and pristine structures, respectively. The chemical potential of oxygen was determined to be 9.426 eV per molecule, as reported previously.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.