Thermal processing to modulate surface chemistry and bulk charge distribution in nickel-rich layered lithium positive electrodes

Abstract

The broader application of nickel-rich layered oxides as positive electrode materials for lithium-ion batteries has been hindered by their high manufacturing cost and inferior cycling stability. Thermal processing, which is integral to electrode materials manufacturing and fundamental in materials science, has not been fully utilized to design advanced positive electrode materials. Herein, we demonstrate the capability of using quenching heat treatment to regulate Li distribution and modulate electronic structure near particle surface. The resulting materials exhibit less parasitic reactions with the electrolyte and an improved charge distribution homogeneity in secondary particles, leading to more stable cycling performance at high voltages (4.5 V vs Li/Li⁺). Our synchrotron X-ray analyses reveal the underlying interplay between surface structure and bulk charge distribution in positive electrode materials particles. While strategies used to stabilize positive electrode materials through compositional control, surface modification, and electrolyte engineering have become mature, thermal processing can be advantageous to further improve positive electrode materials manufacturing.

Nickel-rich Li-ion positive electrodes face challenges such as high cost and poor cycling stability. Here, authors show that quenching heat treatment can lead to more stable performance at high voltages, with synchrotron analyses revealing the roles of surface chemistry and bulk charge distribution.

Introduction

Nickel-rich LiNi_xMn_yCo_zO₂ (x ≥ 0.8, NMC) layered positive electrode materials with high specific capacity (≥200 mAh/g) hold great potential for high-energy lithium batteries, the most significant power devices for long-range electric vehicles. However, the cycling stability of nickel-rich positive electrode material is still unsatisfactory at present, especially at high voltages. The severe structural and chemo-mechanical degradation results in rapid capacity decay of these positive electrode materials during cycling, and the capacity decay is accelerated with an increase of Ni content in materials¹. The surface stability plays a vital role in nickel-rich positive electrode materials, because their performance deterioration can initiate at the surface due to parasitic reactions with the electrolyte, transition metal dissolution, and surface reconstruction^2–4. Moreover, the surface stability of NMCs is related to their bulk chemo-mechanical stability during cycling. Due to the mutual modulation between surface chemistry and bulk microstructure, a higher surface structural stability can alleviate anisotropic stress formation in NMC polycrystalline particles^5,6. To mitigate surface instability of NMCs, researchers have built functional layers on positive particles or prepared NMCs with an elemental concentration gradient and core-shell structure^7–10. A surface coating layer such as Al₂O₃ can enhance the cycling stability of NMCs by suppressing undesired side reactions with the electrolyte and lowering interfacial charge transfer impendence¹¹. The concentration gradient NMCs exhibit improved long-term cycle life through enhancing structural stability with a Mn-rich surface layer. However, the effectiveness of these strategies depends on the capability to precisely control the synthesis conditions^12,13.

Considering that the capacity delivery and electrochemical kinetics of positive electrode materials are associated with reversible Li ion (de)intercalation throughout entire battery particles, the structural and chemical evolution in the bulk of NMC particles should be considered when investigating degradation mechanisms upon cycling. Researchers have introduced heteroatoms into the NMC lattice, which may mitigate undesired bulk degradation^14–18. However, due to the crystallographic anisotropy, it is unlikely to make zero-strain layered positive electrode materials although from time to time one may claim minimal volume changes upon cycling. Due to their high Young’s moduli, a seemingly insignificant strain can still lead to high mechanical stress. Meanwhile, it is non-trivial to precisely place heteroatoms in the desired site of NMC lattice. Another hardly discussed drawback of doping chemistry, especially those with multiple dopants, is the vulnerable raw materials supply chain. Adding more elements means a more complex raw materials supply chain. It is within such context that controlling materials microstructures without excessively using dopants has become an attractive strategy for long-term manufacturing reliability^6,19–21.

As a key step in positive electrode material manufacturing, the calcination process can drive mass transport and reaction kinetics to redistribute constituting elements to become either homogeneously or heterogeneously distributed in NMC secondary particles²². In a typical NMC calcination, the ramping and cooling rates are controlled at a few or tens of degrees per minute. Quenching can produce materials far from equilibrium. Quenching from the peak of the calcination temperature to low temperature has been shown to increase the reversible capacity of lithium titanate negative electrodes²³. Some studies adopted a similar quenching treatment for Ni-rich positive electrode materials and found that Ni-rich positive electrodes could deliver enhanced electrochemical performance after the quenching treatment^24,25. It has remained largely unexplored regarding how quenching can change the mutual modulation between the surface and bulk of layered positive electrode particles and thus impact battery performance. If successful, such a thermal treatment can reduce the overall calcination duration and become an effective strategy to tailor battery performance.

Herein, we report quenching as a strategy to simultaneously mitigate surface structural degradation and address the bulk charge distribution heterogeneity issue. We manipulate elemental distribution in NMC polycrystalline particles by regulating the cooling rate from the peak of calcination temperature. We observe substantial lithium diffusion from the bulk to the surface of NMC polycrystalline particles, creating a surface Ni reduction layer to decrease the surface reactivity. Subsequent X-ray spectroscopic and imaging analyses show the improved surface stability allows the redox reactions to proceed more homogeneously throughout polycrystalline particles and results in more favorable capacity retention cycled to high voltages. Our work provides insights into regulating the mutual modulation between surface chemistry and bulk charging reactions within polycrystalline layered positive electrode particles.

Results

Li ion redistribution near particle surface upon quenching

The conventional thermal protocol for synthesizing layered positive electrode material is to calcine the precursor at high temperatures followed by natural cooling to room temperature. In this work, we followed the same heating procedure as the conventional positive electrode material synthesis but quenched the positive electrode material instead of slow cooling after the high-temperature holding. The quenching process can manipulate the diffusion of light elements in particles such as Li ions in the present context, leading to distinct structural evolution and formation of Li-containing species at the surface of NMC particles. The as-prepared LiNi_0.8Mn_0.1Co_0.1O₂ positive electrode materials by natural cooling and quenching are denoted as NMC-S and NMC-Q, respectively. We performed X-ray photoelectron spectroscopy (XPS) and soft X-ray absorption spectroscopy (XAS) to study the surface evolution for NMC-Q and NMC-S powder samples. In Fig. 1a, Li 1 s XPS spectra are deconvoluted into Li₂CO₃ peak with a higher binding energy and Li in the lattice of NMC positive electrode with a lower binding energy^17,25. The surface of NMC-Q has a higher Li₂CO₃ content than that of NMC-S. Accordingly, the Li assigned to NMC lattice in NMC-Q is lower than that in NMC-S. Similarly, we found that the surface O assigned to lithium-containing surface species (e.g., carbonate) increases after the quenching treatment (Fig. 1b), where the spectra were deconvoluted to differentiate the surface and lattice O species²⁶. Next, we performed soft XAS to further investigate the impact of quenching on the surface chemical properties of NMC. Soft XAS is sensitive to the oxidation states and electronic configurations of transition metals at the surface²⁷. Herein, we applied two detection modes, i.e., total electron yield (TEY) and fluorescence yield (FY), to probe the surface and subsurface information at ~10 nm and ~50 nm depths, respectively^26,27. The peak intensity ratio between the high energy shoulder (L_3high) and low energy shoulder (L_3low) of L₃-edge is positively correlated to the Ni oxidation state^2,16,17. At a ~ 10 nm probing depth by the TEY mode, the quenched sample NMC-Q shows a lower L_3high/L_3low ratio and thereby a lower Ni oxidation state than NMC-S on the particle surface (Fig. 1c). The Ni oxidation state remains virtually the same for NMC-S and NMC-Q at the ~50 nm probing depth based on the FY mode, showing a high similarity of Ni oxidation state in the bulk between NMC-Q and NMC-S samples. One can conclude that quenching has a negligible effect on the bulk Ni oxidation state of NMC materials. To further validate our conclusion, we performed scanning transmission electron microscopy-electron energy loss spectroscopy (STEM-EELS) measurements on NMC-Q and NMC-S (Supplementary Fig. 1). From the O K-edge SETM-EELS spectra in Supplementary Fig. 1b, e, NMC-Q and NMC-S show apparent pre-edge peak marked by ‘*’. This pre-edge peak is associated with the transition from O1s to O2p-TM3d hybridized state and is sensitive to the transition metal (TM)’s oxidation state^2,26,28. For NMC-Q, the pre-edge peak of O K-edge is absence at the subsurface region but is distinct and sharp in the bulk of particles in Supplementary Fig. 1b. For NMC-S, the pre-edge peak of O K-edge can be seen at both subsurface and bulk regions in Supplementary Fig. 1e. Similarly, from Supplementary Fig. 1c and f, we can see a peak shift of Ni L-edge for NMC-Q, whereas insignificant peak shift for NMC-S from the subsurface to the bulk. Thereby we can confirm that NMC-Q shows a thicker Ni reduction surface layer than NMC-S. Combining soft XAS and STEM-EELS results, it can be concluded that (1) NMC-Q shows a lower Ni oxidation state than NMC-S on the surface; and (2) NMC-Q and NMC-S display a high similarity of Ni oxidation state in the bulk, showing the negligible effect of quenching on the Ni oxidation state in the bulk. The XAS pre-edge of O K-edge spectra can be attributed to TM3d-O2p hybridization and Li₂CO₃ (~534 eV) (Fig. 1d). The O K-edge spectra in the TEY mode show a higher Li₂CO₃ intensity in NMC-Q than that in NMC-S, which is consistent with the XPS results above. In the FY mode, O K-edge shows a smaller Li₂CO₃ intensity difference between NMC-Q and NMC-S, demonstrating that Li₂CO₃ is a surface species and that the impact of quenching on the materials properties remains at the top tens of nanometers. Furthermore, the TM3d-O2p hybridization peaks are weaker for NMC-Q in both TEY and FY modes, suggesting that surface TMs are reduced. From the bright field-scanning transmission electron microscopy (BF-STEM) images, we can see the thickness of the Li₂CO₃ formation on the quenched (NMC-Q) sample (Supplementary Fig. 2a) is 3–5 nm, whereas it is only 1–2 nm for the NMC-S sample (Supplementary Fig. 2b), confirming more Li₂CO₃ existence on the surface of NMC-Q than that on NMC-S. The complementary results from XPS, STEM-EELS and soft XAS show the distinct surface chemistry of the NMC-Q sample, i.e., lower TM oxidation states and more Li₂CO₃ at the particle surface than NMC-S. The rapid cooling forces Li ions to diffuse out of the lattice, resulting in the formation of surface carbonate and surface TM reduction, as schematically shown in Fig. 1e. It should be noted that Ni-rich positive electrode materials are vulnerable in the ambient environment^17,29,30, so we transferred both powders and electrodes into an argon-filled glovebox immediately after preparation. All the materials and electrodes underwent similar transferring processes, so the differences observed between NMC-Q and NMC-S are attributed to different thermal processing.

Fig. 1. The surface structural and chemical properties of pristine NMC-S and NMC-Q powders to show distinct Li distribution upon different thermal processing.

XPS spectra of a Li 1 s and b O 1 s for NMC-S and NMC-Q pristine powders. Li 1 s spectra are deconvoluted into surface Li₂CO₃ and Li ions in the NMC lattice, and O 1 s spectra are deconvoluted into surface Li₂CO₃ and lattice oxygen. Soft XAS c Ni L-edge and d O K-edge spectra in the TEY and FY modes for NMC-S and NMC-Q powders. e Schematic illustration of surface structure evolution of NMC particles during the quenching process.

Intact bulk properties upon quenching

Although Li ion intercalation and deintercalation initiate at the particle surface, the overall electrochemical performance involves the bulk properties of positive electrode particles. Therefore, probing the bulk properties is equally important. Synchrotron X-ray diffraction (XRD) patterns show both NMC-S and NMC-Q have a hexagonal α-NaFeO₂ layered structure with a $\mathrm{R}\overline{3}\mathrm{m}$ space group (Supplementary Fig. 3). The peak position and shape of NMC-Q are consistent with NMC-S, revealing a similar crystal structure and crystallinity between quenched and slow cooled samples. We further performed neutron diffraction (ND) and Rietveld refinements for NMC-Q and NMC-S (Fig. 2a, b, Supplementary Fig. 4 and Supplementary Table 1). Ni²⁺/Li⁺ cation mixing and lattice parameters are similar for NMC-Q and NMC-S, suggesting minimal changes of bulk crystal structure are induced by quenching.

Fig. 2. The bulk structural and chemical properties of NMC-S and NMC-Q pristine powder materials.

Rietveld refinement of structure of a NMC-S and b NMC-Q using neutron powder diffraction data. The R_wp for NMC-Q and NMC-S is 5.54% and 4.64%, respectively. Experimental data are shown in black dots, calculated results in red curve and difference curves in blue. Bragg reflections are in purple marks. Hard XAS Ni K-edge c XANES and d EXAFS spectra of NMC-S and NMC-Q. e Cross-sectional 3D TXM images and f Ni K-edge energy distribution histogram in the whole particle of NMC-S and NMC-Q at the pristine state.

To probe the transition metal electronic structure and local chemical environment in bulk, we performed hard XAS that includes X-ray absorption near edge structure (XANES) and extended X-ray absorption fine structure (EXAFS). Ni K-edge XANES spectra show well-aligned shape and nearly no shift of the pre-edge and edge positions for NMC-Q and NMC-S, indicating similar bulk Ni oxidation state in the two samples (Fig. 2c). In addition, Ni K-edge EXAFS for NMC-Q and NMC-S almost overlaps with each other, which means the interatomic distances of Ni-O and Ni-TM and the atomic environment of Ni are nearly identical for the two samples (Fig. 2d). Mn and Co XANES and EXAFS results (Supplementary Fig. 5) further confirm that different cooling protocols barely lead to the bulk chemical environment variance. We further examine the morphology of the two samples through scanning electron microscopy (SEM) at various magnification (Supplementary Fig. 6). Both NMC-Q and NMC-S possess a spherical polycrystalline particle morphology around 10 μm in diameter. No crack formation is observed in either sample, showing a minimal effect of cooling protocol on the particle morphology.

Charge heterogeneity is strongly correlated with the redox behavior within positive electrode particles, which can lead to internal polarization and performance degradation upon cycling^19,31,32. To complement the surface-sensitive soft XAS and ensemble-averaged hard XAS, we further performed the full field transmission X-ray microscopy (TXM) at the single particle level for NMC-S and NMC-Q at the pristine state. TXM is a powerful technique that can non-invasively reconstruct the bulk morphology of particles with a spatial resolution down to tens of nanometers and provide chemical information, such as oxidation states. From the cross-section of three-dimensional (3D) TXM images for NMC-S and NMC-Q particles in Fig. 2e, we observe similar Ni K-edge energy level and overall uniform distribution, indicating similar Ni oxidation state and distribution inside the particles at pristine state. We further plot the energy distribution histogram in the whole particle (Fig. 2f), which shows similar average Ni K-edge absorption energy at around 8352.4 eV for both samples. Both TXM and hard XAS results show similar bulk Ni oxidation state in bulk NMC-S and NMC-Q. To resolve the spatial details of energy distribution, we did depth-dependent analyses (Supplementary Fig. 7). Shell 1 is the outmost shell at the particle surface with a thickness of 20 nm (Supplementary Fig. 7a). Each shell with increasing numbers is 20 nm away from its outer shell. We plot the energy distribution within Shell 1 for NMC-Q and NMC-S (Supplementary Fig. 7b). The histogram of NMC-Q shifts to lower energy compared to that of NMC-S, which indicates the Ni oxidation state of NMC-Q is overall lower than that of NMC-S at the top 20 nm surface. These results are consistent with the trend revealed by XPS, STEM-EELS and soft XAS that NMC-Q has more surface reduced Ni. More Li₂CO₃ on the surface of NMC-Q originates from the Li diffusion from the bulk of NMC crystal structure to the surface, which is driven by the rapid decrease of temperature on the surface during quenching. The rapid temperature decrease reduces the solubility of Li ion in the lattice. The Li diffusion from the bulk to the surface can also be regarded as Li loss, which happens together with O loss. Then, Li₂O forms on the surface of NMC-Q, which is later changed to Li₂CO₃ by reacting with the CO₂ in the air. Ultimately, NMC-Q shows a reconstructed surface layer with a lower Ni oxidation state and more Li₂CO₃ formation on the particle surface. Based on our understanding of the materials and their characterization, we determine that the particle surface has been reconstructed by the rapid quenching process.

Quenching enhances cycling stability of layered positive electrode at high voltages

Considering the capacity decay impedes the future application of polycrystalline NMC positive electrode, we examine the electrochemical cycling performance of NMC-S and NMC-Q positive electrodes (Fig. 3). NMC-Q and NMC-S are paired with Li metal negative electrode and cycled within the voltage window of 2.8–4.5 V at 22 °C. The initial specific discharge capacity of NMC-Q and NMC-S at a specific current of 100 mA/g is 195 mAh/g and 198 mAh/g, respectively (Fig. 3a, b and Supplementary Fig. 8). NMC-Q shows a slightly lower initial capacity than NMC-S at 100 mA/g. It could be attributed to the Li diffusion from the bulk to the surface during quenching, shown in the XPS, STEM-EELS and soft XAS results, leading to some Li inventory loss for initial cycling. After 200 cycles, the capacity retention of NMC-Q is much higher than that of NMC-S, showing enhanced cycling stability of NMC-Q. When increasing the charging rate to 200 mA/g, the peak specific discharge capacity for NMC-Q and NMC-S is 181 mAh/g and 176 mAh/g, respectively. The capacity retention of NMC-Q after 300 cycles at 200 mA/g is 90%, which is much higher than 72% of NMC-S under the same conditions (Fig. 3c). It is noted that the retention is calculated by dividing the final value by the peak value through the cycling history.

Fig. 3. Electrochemical cycling performance of NMC-Q and NMC-S positive electrodes measured in Li metal cells.

The charge/discharge profiles of a NMC-Q, b NMC-S for 1st, 2nd, 20th, 50th, 100th and 200th cycle at a specific current of 100 mA/g with a voltage range of 2.8–4.5 V at 22 °C. c The discharge specific capacity of the NMC-Q and NMC-S at a specific current of 200 mA/g with a voltage range of 2.8–4.5 V at 22 °C, where error bars are created based on the standard deviation of the data from three repeated cells.

The cells show increasing specific capacity during the first several cycles in Fig. 3c, which is associated with the increasing Li ion transport kinetics due to factors such as improved electrolyte wetting. The increase is more obvious for the NMC-Q because it takes more cycles to overcome the initial kinetic barrier created by the surface reduced layer³³. When cycled at a higher specific current (200 mA/g in Fig. 3c), the kinetic barrier plays a larger role than when cycled at a lower specific current (100 mA/g in Fig. 3a, b). We further compare the rate capability and phase transition during cycling of NMC-Q and NMC-S. Comparing to NMC-S, NMC-Q shows improved rate capability, especially at high current rate. For instance, the NMC-Q maintains a reversible specific capacity of 164 mAh/g at 400 mA/g, while the NMC-S delivers a specific capacity of only 136 mAh/g in Supplementary Fig. 9a. The cycling stability of NMC-Q is also better than that of NMC-S at 100 mA/g (Supplementary Fig. 9b). From the dQ/dV curves at 100 mA/g in Supplementary Fig. 9c, d, it can be observed that the curve of NMC-Q exhibits less significant changes than that of NMC-S, contributing to better cycle life for NMC-Q. Supplementary Fig. 10 shows that the charge transfer resistances of NMC-Q after 1 cycle and 50 cycles are both lower than those of NMC-S.

Improved interfacial stability of quenched layered positive electrode upon battery cycling

To investigate the origin of the enhanced cycling stability and the evolution of the surface chemical and structural properties in NMC-Q, we performed soft XAS for NMC-S and NMC-Q electrodes at the pristine state and after 200 cycles at a specific current of 100 mA/g with a voltage range of 2.8–4.5 V. The peak intensity ratio Li_3high/L_3low of NMC-Q shows minor change after 200 cycles, whereas the NMC-S experiences profound Ni reduction at the surface based on both TEY and FY modes (Fig. 4a, b). These results show that NMC-Q demonstrates improved interfacial structural stability. We further quantify the L₃ edge peak intensity ratio L_3high/L_3low for the two samples at the pristine state and after 200 cycles (Fig. 4c). L_3high/L_3low of NMC-Q slightly decreases from 0.64 to 0.51 in the TEY mode while maintaining at 0.80 in the FY mode after 200 cycles (Fig. 4c). In contrast, the value for NMC-S has a sharp decrease from 0.69 to 0.25 in the TEY mode and from 0.80 to 0.54 in the FY mode after 200 cycles, which is attributed to more severe surface side reactions upon cycling. We further performed TXM on NMC-Q and NMC-S after 100 cycles at 200 mA/g and plot the Ni K-edge energy distribution for Shell 1 (Fig. 4d). The histogram pattern of NMC-Q is sharp, and the energy distributes in a relatively small range, indicating a more uniform Ni oxidation state distribution at the top surface of NMC-Q (Fig. 4d). In contrast, the energy distribution peaks of NMC-S are broader and exhibit a bimodal feature. The low energy peak indicates a large portion of Ni at the top surface of NMC-S is reduced (Fig. 4d). Ni is the primary charging compensating element; thus, the oxidation or reduction of Ni can be utilized as an indicator for delithiation or lithiation. 3D Ni K-edge TXM shows that after extensive cycles the charge distribution at the surface of NMC-S is more heterogeneous than that of NMC-Q (Fig. 4e). The results of soft XAS and TXM collectively show that the surface changes in NMC-Q during quenching promotes charge homogeneity at the positive electrode surface and improves interfacial stability between positive electrode and the electrolyte during battery cycling.

Fig. 4. The surface chemical and structural evolution of NMC-S and NMC-Q electrodes after cycling.

Soft XAS Ni L-edge in (a) TEY, and (b) FY mode of the NMC-S and NMC-Q electrodes at the pristine state and after 200 cycles at a specific current of 100 mA/g with a voltage range of 2.8–4.5 V at 22 °C. c The Ni L_3high/L_3low values of NMC-S and NMC-Q electrodes at the pristine state and after 200 cycles in the TEY and FY modes at a specific current of 100 mA/g with a voltage range of 2.8–4.5 V at 22 °C. d The first shell (20 nm away from particle surface) Ni K-edge energy distribution and e 3D TXM images of NMC-S and NMC-Q particles after 100 cycles at a specific current of 200 mA/g with a voltage range of 2.8–4.5 V at 22 °C.

Homogeneous charge distribution in quenched particles after long-term cycling

We further study the evolution of bulk structure upon cycling of NMC-S and NMC-Q using hard XAS and TXM. Ni K-edge XANES (Fig. 5a) and EXAFS (Fig. 5b) results show minimal difference between NMC-S and NMC-Q after 100 cycles at 200 mA/g with a voltage range of 2.8–4.5 V, which shows that Ni oxidation state, Ni-O and Ni-TM interatomic distances, and the local environment of Ni are similar in two samples at the ensemble-averaged scale. Co K-edge and Mn K-edge XANES and EXAFS spectra in Supplementary Fig. 11 show similar peak intensity and shape without noticeable peak shift for NMC-S and NMC-Q, further confirming similar overall bulk Co and Mn oxidation states. However, the particle scale analyses from 3D TXM show large differences between NMC-S and NMC-Q (Fig. 5c, d). The 3D TXM cross-sectional tomographic images of NMC-S show higher charge distribution heterogeneity, whereas the charge distribution within NMC-Q particle is relatively homogeneous (Fig. 5c, d). Ni K-edge energy distribution further confirms that NMC-Q has a more uniform charge and oxidation state distribution than NMC-S due to a narrower peak in Fig. 5e. We further plot the average valence gradient vector size as a function of vector angle θ following the method developed in our previous study^19,34. The overall smaller vector size of NMC-Q than NMC-S confirms that the charge heterogeneity of NMC-Q is smaller in Fig. 5f, which is beneficial to the structure and cycling stability improvement of NMC positive electrodes upon extensive cycling.

Fig. 5. The bulk chemical and structural evolution of NMC-S and NMC-Q after cycling.

Hard XAS a XANES and b EXAFS results of Ni K-edge after 100 cycles at a specific current of 200 mA/g with a voltage range of 2.8–4.5 V in the discharged state at 22 °C. 3D rendering of the tomographic data over the c NMC-S and d NMC-Q particles with the perspective views of a few virtual slices displayed in the center. e Ni K-edge energy distribution in the whole particle of NMC-S and NMC-Q after 100 cycles at 200 mA/g with a voltage range of 2.8–4.5 V at 22 °C. f The average valence gradient vector size as a function of θ where 0°−180° were divided to 36 intervals.

Discussion

Over the past decade, substantial research activities have been steered towards understanding the relative importance of the surface and bulk properties of layered positive electrodes in determining battery performance. Due to the presence of chemo-mechanical degradation upon long-term cycling, the surface and bulk properties collectively impact battery performance. In particular, when there is microcrack formation, the bulk of positive electrode particles can be exposed to the electrolyte and essentially becomes the surface. It is within such context that researchers have concluded the mutual modulation between surface and bulk processes in these materials⁵. In general, there is a great need to better control the surface properties to improve bulk redox reactions. In summary, we engineered the surface of NMC secondary particles using the non-equilibrium quenching method. This method can be applied in addition to other materials stabilization strategies. Our characterization results show that quenching can improve the battery performance of NMC materials in multiple aspects. Specifically, we presented that NMC-Q can deliver an initial discharge specific capacity of 195 mAh/g and 181 mAh/g at 100 mA/g and 200 mA/g, respectively. Upon 300 cycles at 200 mA/g, the capacity retention of NMC-Q can be up to 90%, higher than 72% of NMC-S. The enhanced long-term cycling stability of NMC-Q is dependent on the less parasitic reactions with the electrolyte and higher charge distribution homogeneity upon cycling. Our study also reveals the underlying interplay between positive electrode surface structure and bulk charge distribution in polycrystalline secondary particles. The more homogeneous charge distribution in NMC positive electrode is closely related to their improved interfacial stability. Based on our understanding, a homogeneous charge distribution may eliminate local high stress regions in the particles, thus promoting higher chemo-mechanical stability and higher utilization of redox active Ni. While we observe a positive impact of quenching experiments on cycling the NMC positive electrode, we have to note that the Li₂CO₃ residue on the NMC surface could influence the full cell performance when paring NMC with a graphite negative electrode. Properly regulating the decomposition path of Li₂CO₃ in the full cell would be needed.

As the community looks for ways to reduce energy consumption and CO₂ emission in CAM manufacturing, we believe new thermal processing methods such as the one presented here could be a promising manufacturing path forward. Applying the quenching technique to real production of battery materials could potentially save time and energy for manufacturing. When the thermal processing strategy is integrated with other strategies such as doping and coating, higher quality positive electrode materials are on the horizon. Several emerging battery chemistries, especially sodium ion battery positive electrode materials, need thermal quenching to synthesize materials with desired phases for good battery performance³⁵. Therefore, some efforts in designing thermal treatment apparatus at the manufacturing line are recommended to garner the manufacturing benefit of quenching. Certainly, when implementing special thermal processing technologies in CAM manufacturing, one would also need to evaluate the capital cost of the manufacturing facility.

Methods

Material synthesis

The precursor of Ni_0.8Mn_0.1Co_0.1(OH)₂ was obtained from Shuangdeng Group Co, Ltd, which was synthesized by a co-precipitation method. Then, the precursor was thoroughly ground with LiOH (dehydrated from LiOH·H₂O, Sigma Aldrich. Purity 99.9%) for one hour by a mortar with a pestle. 5 mol% excess LiOH was added to compensate for the Li loss during high-temperature calcination. The obtained powder after mixing the Ni_0.8Mn_0.1Co_0.1(OH)₂ precursor and LiOH was calcined at 460 °C for two hours, followed by 750 °C for six hours with a temperature ramping rate of 3 °C/min. Subsequently, two protocols of cooling were performed to obtain the LiNi_0.8Mn_0.1Co_0.1O₂ positive electrode materials: (1) slow cooling under constant oxygen flow at a rate of 2.0 L/min (denoted as NMC-S); (2) quenching in the air (denoted as NMC-Q). The approximate cooling rate for NMC-S and NMC-Q is approximately 3 °C/min and 140 °C/min, respectively, by the calculation from total cooling time to room temperature. The air in the quenching process has about 38% relative humidity.

Materials characterizations

Synchrotron X-ray diffraction (XRD) was performed at the beamline 11-3 of Stanford Synchrotron Radiation Lightsource (SSRL). The calibrated wavelength of the applied X-ray is 1.5406Å. Experiments involving neutron diffraction (ND) was carried out on the Nanoscale-Ordered Materials Diffractometer (NOMAD) beamline, specific BL-1B, at the Spallation Neutron Source (SNS) in Oak Ridge National Laboratory. For ND measurements, the power sample between 0.2 and 0.3 g was placed into a quartz tube with a diameter of 3 mm, and each session of data acquisition lasted roughly one hour. The structure refinements against neutron diffraction patterns were conducted using the software GSAS EXPGUI to identify the crystal structure and Li/Ni cation mixing in NMC-S and NMC-Q samples³⁶. To study the effect of quenching on morphology of NMC-S and NMC-Q, scanning electron microscopy (SEM) was conducted using FEI Quanta 600 FEG. Hard XAS measurement, including XANES and EXAFS, was carried out at the beamline 20-ID of the Advanced Photon Source (APS) at Argonne National Laboratory using a Si (111) monochromator. The calibration of the absorption energy was referenced against the initial inflection points from the spectra of Ni, Mn, and Co, which were captured in tandem with the sample data. For powder samples, about 50 mg powder were sealed by the Kapton tape in an Ar-filled glove box (H₂O < 0.5 ppm, O₂ < 0.5 ppm) for hard XAS measurements. For positive electrode samples, they were collected by tearing down the cycled cells, then rinsing with dimethyl carbonate and drying in the glove box. All powder and electrode samples were sealed in Mylar aluminum bags for transfer to measurements at room temperature. X-ray photoelectron spectroscopy (XPS) measurements were performed on the PHI VersaProbe III with a monochromatic Al K-alpha X-ray source (1486.6 eV). The samples were transferred into the chamber using an airtight inert transfer vessel for XPS measurements. Soft XAS was performed using total electron yield (TEY) and fluorescence yield (FY) mode under ultrahigh vacuum (10⁻⁹Torr) to study the oxidation state of Ni, Co and Mn elements in the bulk and on the surface of NMC particles. The data of soft XAS was collected at the Stanford Synchrotron Radiation Lightsource (SSRL) beamline 10-1, SLAC National Accelerator Laboratory. The high resolution (∼10¹¹ph s⁻¹ at 0.2 eV in a 1 mm² beam spot) was provided using a ring current of 350 mA and a 1000 L mm⁻¹ spherical grating monochromator with 20 μm entrance and exit slits. The samples were mounted on an aluminum sample holder in an Ar-filled glove box and well-sealed for transfer. The STEM and EELS measurements were performed at an accelerating voltage of 200 kV using an instrument of Hitachi HD2700C The background of EELS spectra was removed using DigitalMicrograph software (Gatan). The comprehensive analysis of charge and oxidation state distribution was conducted using TXM at the beamline 18-1D of the National Synchrotron Light Source II (NSLS-II) located at Brookhaven National Laboratory. For the reconstruction and analysis of the tomographic data, we employed a python-based tool named TXM-Sandbox^37,38. Additionally, a commercial visualization software, Avizo, was utilized for enhanced graphical representation. In order to map the charge distribution with NMC-S and NMC-Q particles, we determined the 2D nanodomain valence gradient vectors through the aggregation of two partial valence gradient vectors. The orientation of the vectors indicates the regional pattern of Ni valence states, while the vectors’ strength corresponds to the gradient of the local valence states. Our previous publication detailed the concept of 3D nanodomain valence gradient vectors, which are hereby referenced for clarity^19,34.

Electrochemical measurements

The NMC-S and NMC-Q electrodes were prepared using a same procedure. To prepare the composite electrode slurry, poly-vinylidene difluoride (PVDF, Sigma-Aldrich, ≥99.5%, molecular weight of 534,000) as the binder was first dissolved into N-methyl-2-pyrrolidone (Sigma-Aldrich, ≥99%), then positive electrode powder and acetylene black (Fisher Chemical, ≥99.9%) were added into the solution and thoroughly mixed, following the ratio of active material: PVDF: acetylene carbon = 90 wt %: 5 wt %: 5 wt %. The slurry was subsequently cast on the carbon-coated aluminum foil by a doctor blade and dried at 120 °C for 12 h in a vacuum oven. The carbon-coated aluminum foil was purchased from MTI Co. with the purity >99.9%, the total thickness of 16 μm (aluminum foil of 15 μm and carbon coating of 1 μm) and the width of 180 mm (carbon coated width of 150 mm). The dry composite electrode was punched into disks of a 10 mm diameter and a thickness of about 70 μm without the current collector by a precision disc cutter (MTI Co., MSK-T10). The areal capacity loading of active materials is ∼1.4 mAh/cm² and the mass loading is ~6 mg/cm². The obtained composite electrodes were used as the positive electrode and Li metal disks were used as the negative electrode with Whatman glass fiber (1827-047934-AH) as the separator to assemble the CR2032 coin cells in an Ar-filled glovebox (MBraun, H₂O < 0.5 ppm, O₂ < 0.5 ppm). The Li metal disks with a diameter of 15.6 mm and a thickness of 450 μm were purchased from China Energy Lithium Co., Ltd. without other treatments when used for negative electrodes. The electrolyte was 1 M LiPF₆ in ethylene carbonate (EC) and ethyl methyl carbonate (EMC) with 2 wt. % vinylene carbonate (VC), which was obtained by manually mixing LP57 electrolyte (Gotion Inc., battery grade) in a formula of 1 M LiPF₆-EC/EMC (with a weight ratio of 3/7) and VC (Sigma Aldrich, anhydrous, ≥99%, inhibitor-free) in the glove box. The obtained electrolyte was placed into a glass bottle and then stored in the glove box at room temperature no more than 2 weeks before use. The material of pipette’s tip for electrolyte transfer is polypropylene. The electrolyte volume in each cell is about 100 μL. All coin cells were cycled using a LAND battery testing system (Wuhan Land Company) from 2.8 to 4.5 V vs. Li/Li⁺ under ambient condition at 22 °C. For galvanostatic charge/discharge experiments, cells were cycled at the specific current of 100 mA/g and 200 mA/g with a voltage range of 2.8–4.5 V at 22 °C. The electrochemical impedance spectroscopy (EIS) was measured by applying an AC voltage of 5 mV over the frequency range of 0.01 Hz to 100 kHz with 74 data points in total using a potentiostat (Biologic, SP-150). The cells for EIS measurements were fully discharged to 2.8 V at a specific current of 100 mA/g after 1 cycle and 50 cycles and placed for 24 h to get an equilibrium state.

Author contributions

F.L. conceived the project. F.L. and H.S. designed the experiments. H.S. performed the synthesis, electrochemical measurements and data analysis. R.G. prepared samples for the TEM measurements and performed data analysis. S.H. performed TEM measurements. A.H. performed SEM measurements. J.L. performed neutron diffraction measurements. C.S. assisted with the hard XAS measurements. S.S. and D.N. collected soft XAS data. X.X. performed synchrotron TXM measurements. Z.Y. and Y.Z. assisted with the TXM measurement and analysis and participated in scientific discussions. H.S., Z.Y. and F.L. wrote the manuscript with feedback and assistance from all coauthors.

Peer review

Peer review information

Nature Communications thanks Xiangming He, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-56075-7.