High‐Nickel Heterostructured Cathodes with Local Stoichiometry Control for High‐Voltage Operation

The growing demand for lithium‐ion batteries to power electric vehicles and other energy‐dense devices continues to fuel the need for cathodes of increasingly higher nickel in cathodes. The relentless pursuit of high Ni content, however, raises concerns on compromising cell lifetime and safety, especially under high‐voltage operation. Alternative to the traditional design of uniform or core–shell composition, we report a rational control of local stoichiometry in high‐Ni cathodes, enabling their high thermal and cycling stabilities—up to 258 °C at the fully charged state and 91.4% capacity retention for 100 cycles between 2.7 and 4.4 V. Multimodal synchrotron X‐ray characterization unveils the heterostructure of secondary particles, featuring a high‐Ni core (LiNi0.90Mn0.05Co0.05O2) covered by a thin Ni‐gradient layer that remains stable over prolonged cycling due to suppressed oxygen release and structural deterioration. This work underlines, the intricate interplay between local stoichiometry and redox reactions in stabilizing high‐Ni cathodes for high‐voltage operation while ensuring safety.


Introduction
Over the past three decades since the commercialization of lithium-ion batteries, the layered transition metal (TM) oxide as represented by LiCoO 2 (LCO) has been the dominant cathode on the market. [1,2]LiNiO 2 attracted immediate attention in the initial search for alternatives to LCO because of its isostructural characteristics, low cost, and even higher practical capacity (>200 mAh g À1 ). [3,4]However, stoichiometric LiNiO 2 is hard to be synthesized, and its thermal instability at high states of charge (SoCs) has been a safety concern.One strategy to mitigate these issues is the substitution of Ni by Mn and Co to form the solid solution LiNi x (MnCo) 1-x O 2 (NMC; x > 0.5). [5,6]To minimize surface areas exposed directly to electrolyte while increasing packing density, spherical secondary particles comprising primary particles are mostly fabricated via a coprecipitation process (for producing hydroxide precursors) followed by hightemperature calcination.As Ni concentration is increased to achieve higher capacity, fast capacity fading becomes an issue, [6][7][8] especially when operating to high voltages (>4.2 V), [9][10][11] as it causes structural transformation from layered to spinel or NiO-type rock-salt phases. [12,13]Another big concern is the poor thermal stability associated with high-Ni-layered oxides, caused by severe oxygen release from the charged electrodes wherein the lattice oxygen framework becomes highly unstable. [6,14,15][18][19][20] One appealing strategy is to construct a heterogeneous structure, such as core-shell or concentration gradient, [21][22][23] where the high-Ni core delivers high capacity and a relatively lower Ni containing The growing demand for lithium-ion batteries to power electric vehicles and other energy-dense devices continues to fuel the need for cathodes of increasingly higher nickel in cathodes.The relentless pursuit of high Ni content, however, raises concerns on compromising cell lifetime and safety, especially under highvoltage operation.Alternative to the traditional design of uniform or core-shell composition, we report a rational control of local stoichiometry in high-Ni cathodes, enabling their high thermal and cycling stabilities-up to 258 °C at the fully charged state and 91.4% capacity retention for 100 cycles between 2.7 and 4.4 V. Multimodal synchrotron X-ray characterization unveils the heterostructure of secondary particles, featuring a high-Ni core (LiNi 0.90 Mn 0.05 Co 0.05 O 2 ) covered by a thin Ni-gradient layer that remains stable over prolonged cycling due to suppressed oxygen release and structural deterioration.This work underlines, the intricate interplay between local stoichiometry and redox reactions in stabilizing high-Ni cathodes for high-voltage operation while ensuring safety.
surface layer improves the cycling and thermal stabilities.However, intense debates persist regarding their fundamental impediments to practical use and about the mechanisms for high-voltage and safe operation in Li-ion cells. [24,25]It remains unclear why and how the heterostructured high-Ni NMC works at high voltages, while the homogeneous ones mostly fail.[28] In this work, we report a rational design of the heterostructured high-Ni-layered oxides for high-voltage operation, to meet the multifaceted performance requirements, including high energy density as well as high cycling and thermal stabilities (Scheme 1a).Specifically, local stoichiometry-controlled (LSC) secondary particles, consisting of an ultrahigh-Ni core with the stoichiometry close to LiNi 0.90 Mn 0.05 Co 0.05 O 2 (NMC9055) covered by a Ni-gradient layer, are produced through a scalable continuous coprecipitation process followed by post-annealing (as detailed in Experimental Section and (Figure S1, Supporting Information)).The precursors for LSC-NMC9055 were synthesized in a continuous mode using two continuous stirring tank reactors (CSTRs), different from the previous work in fabricating core-shell-structured NMC811 (core-shell NMC811) via batch mode using a single tank. [21]The local stoichiometry is intentionally controlled via thermally driven elemental interdiffusion during the post-annealing process, turning the core-shell structured secondary particles into particles consisting of an ultrahigh-Ni core (LiNi 0.90 Mn 0.05 Co 0.05 O 2 ; NMC9055) covered by a Ni-gradient layer.The obtained LSC-NMC9055 exhibit superior electrochemical performance and thermal stability NMC9055, compared to the traditional compositionally homogeneous NMC9055 (homo-NMC9055) and core-shell NMC811. [21]To understand the role of LSC in tuning the cycling and thermal stabilities, multimodal X-Ray characterization using operando transmission X-Ray microscopy (TXM), X-Ray absorption near-edge structure spectroscopy (XANES), X-Ray absorption spectroscopy (XAS), and X-Ray diffraction (XRD) is carried out (as illustrated in Scheme 1b and further detailed in Note S1, Supporting Information).The electrochemical redox and the associated structural evolution are probed at different length scales, both in bulk electrodes and locally in individual particles, shedding light on how the local redox reaction proceeds during cycling and on its interplay with local stoichiometry in the heterostructured LSC-NMC9055 during cycling.

Heterostructure
A continuous coprecipitation process was adapted to synthesize hydroxide precursors for controlling the local stoichiometry across secondary particles (Figure S1, Supporting Information).During the process, high-Ni particles (with stoichiometry close to NMC9055) were first formed in the core region and then coated with a low-Ni layer (with the stoichiometry close to LiNi 1/3 Mn 1/3 Co 1/3 O 2 , NMC333), leading to coreshell-structured secondary particles.As shown in Figure S2 (Supporting Information), the maps of elemental distribution, obtained by scanning transmission electron microscopy energy dispersive X-Ray spectroscopy (STEM-EDX), were obtained from as-synthesized precursors, and showed highly concentrated Ni in the core and low Ni concentration near the surface (within %200 nm).Through post-annealing, heterostructured NMC9055 (LSC-NMC9055) layered oxides were obtained.Detailed chemical and structural analysis of LSC-NMC9055, in comparison to that of homo-NMC9055 and core-shell811, were obtained from both pristine and cycled samples, for correlation with their electrochemical and thermal stability properties.
Figure 1a shows the 2D valence map of the Ni 2þ (red) and Ni 3þ (green) within a single secondary particle of LSC-NMC9055, obtained by TXM-XANES maps fitted using PyXAS, [29] an in-house-developed XANES image analysis package.The Ni 2þ /Ni 3þ distribution is heterogeneous, with Ni 2þ Scheme 1. Rational design of high-Ni NMC cathodes with local stoichiometry control for high-voltage and safe operation.a) Schematic illustration of making heterostructured particles via the local stoichiometrycontrolled (LSC) route, alternative to the traditional way of employing uniform or core-shell composition, to meet the multifaceted performance requirements (e.g., high energy density, high cycling, and thermal stabilities).b) Multimodal synchrotron X-Ray characterization for disentangling the interplay between local stoichiometry and electrochemical redox in stabilizing high-Ni cathodes.The unique tool of transmission X-Ray microscopy X-Ray absorption near-edge structure spectroscopy (TXM-XANES) with nanometer-scale resolution is employed to probe local electrochemical redox across individual particles (bottom), which is complemented by X-Ray diffraction (XRD) and X-Ray absorption spectroscopy (XAS) to investigate the electrochemical redox and the associated structural evolution in the bulk (top).
(red) mainly in the near-surface region and Ni 3þ (green) in the core, as also observed from different cuts across the particle (Figure S3, Supporting Information).Such a heterogeneous distribution of Ni 2þ /Ni 3þ across the core and surface regions is largely attributed to the original compositions in the precursor, namely, NMC9055 (with Ni oxidized to %3þ during sintering) in the core and NMC333 (with Ni valence remaining as 2þ) in the surface layer, respectively.However, no obvious core-and-shell structure as in the original precursors, but rather a gradual change in the Ni 2þ /Ni 3þ distribution, was seen in the near-surface region, as also indicated by the relative concentration projection (Figure 1a, bottom).The concentration of Ni 2þ drops from 50% near the surface to %20% in the core region; the concentration of Ni 3þ is equivalent to that of Ni 2þ near the surface but quickly increases to 80% in the majority of the core.
No sharp boundaries, but instead a gradual change of the Ni 3þ /Ni 2þ concentration ratio in the final products, implies that elemental interdiffusion occurred during the high-temperature sintering. [30]The Ni distribution along the radial direction across the LSC-NMC9055 particles was retrieved through reverse Radon transform using the filtered-back-projection method (with the assumption of a spherical particle shape; see the detailed discussion in Note S2, Supporting Information).As shown in Figure 1c, a high concentration of Ni 3þ (up to 90%) is distributed across the majority of the core region in the particle, and is covered with a thin Ni-gradient layer, which is restricted to the nearsurface region with a depth of %1.2 μm (or 15% of the particle radius).In the LSC-NMC9055, a low level of Li/Ni cation mixing (%4.6%) due to Li/O loss was induced by high-temperature sintering (Figure 1d).

Electrochemical Performance
In coin cells against Li metal, the presence of the Ni-gradient layer NMC9055 led to different electrochemical behaviors of the LSC-NMC9055 compared to that of the homo-NMC9055, first shown by the voltage profiles (Figure 2a).The plateau at around 4.2 V, clearly seen in homo-NMC9055, was less visible in LSC-NMC9055.The more striking change is the high-voltage cycling stability gained in LSC-NMC9055 (up to 4.4 V), which is significantly improved in this respect compared to the homo-NMC9055, as shown in Figure 1b.Homo-NMC9055 experienced a serious capacity fade, at a rate of C/2, from the initial 197-125 mAh g À1 after 100 cycles.In contrast, LSC-NMC9055 retained a capacity of about 180 mAh g À1 , meaning about 91.4% capacity retention after 100 cycles.The cyclability of LSC-NMC9055 is comparable to that of NMC333, even though the Ni-gradient layer is restricted to the near-surface region of LSC-NMC9055, indicating the effectiveness of local stoichiometry for stabilizing the high-Ni cathodes for high-voltage operation.The cycling stability in NMC333 is due to the absence of redox within the voltage range (between 2.7 and 4.4 V), as shown in Figure 2c.In contrast, the homo-NMC9055 exhibits a strong redox peak at 4.2 V (corresponding to the plateau in Figure 1a arising from the H2 to H3 phase transformation), which fades away quickly with cycling (within about 20 cycles), indicating structural instability (Figure 2e).The redox peak at 4.2 V is also seen in LSC-NMC9055, but with a much reduced amplitude, and is retained over cycling even after 100 cycles (Figure 2d), indicating the crucial role of the Ni-gradient layer in maintaining structural stability by suppressing the H2-to-H3 phase transformation.
The long cycling performance of the LSC-NMC9055 was tested in the pouch-type full cells (against graphite) at voltages between 3.0 and 4.3 V using a C/3 charge and C/2 discharge rates.Electrochemical tests were also made to the baseline NMC811 and core-shell NMC811 for comparison, as detailed in Experimental Section.The main results are given in Figure 3. Being consistent with half-cell tests (Figure 2), LSC-NMC9055 exhibited exceptional cycling stability, maintaining 80% capacity by 700 cycles, in contrast to 76% retention in the core-shell NMC811% and 54% capacity by 500 cycles in the baseline NMC811 (Figure 3a).Correspondingly, negligible voltage decay, only 0.04 V by 700 cycles, was observed LSC-NMC9055, in contrast to 0.08 V in core-shell NMC811 by 700 cycles and 0.22 V drop by 500 cycles in the baseline NMC811 (Figure 3b), indicating the important role of LSC in stabilizing the layered structure both in the bulk and at the particle surface.This is further corroborated by the high Coulombic efficiency (nearly 100%) by the slow increase of the discharge area-specific impedance (ASI) at 50% depths of discharge (DOD), as given in

Retention of the Heterostructure After Long Cycling
To check the durability of the heterostructure with cycling, 3D TXM tomography of the Ni concentration was performed on the cycled LSC-NMC9055 by scanning X-Ray energies across the Ni K-edge.The reconstruction with volume rendering and a 2D cross section of the Ni distribution is plotted in Figure 4a,b.The local stoichiometry with the low-Ni surface region and high-Ni core region persisted after 120 cycles, which is better viewed by examining the 2D Ni-distribution in cross-section maps.Moreover, the close-packing morphology of secondary particles is maintained after cycling, without obvious cracks being observed; this behavior is crucial to the long cycling stability by excluding electrolyte penetration. [31,32]he maintenance of the particle morphology was further correlated to the local stoichiometry in the cycled LSC-NMC9055 through examination by 2D valence mapping (Figure 4c-e).As in the pristine samples, Ni 2þ (red) is mainly distributed in the near-surface region and Ni 3þ (green) is distributed in the core region; this pattern is further confirmed by the intensity profiles in Figure 4f.In comparison to the valence distribution in the pristine sample, there is only %5% change in Ni 2þ /Ni 3þ concentration after cycling (Figure 4g), indicating the negligibly small stoichiometry change even after 120 cycles.The persistent Ni-gradient layer should have played an important role in maintaining the heterostructure, likely through alleviating lattice mismatch over the long cycling. [33,34]

Local Electrochemical Redox During Cycling
To understand the role of local stoichiometry in LSC-NMC9055, local electrochemical redox within individual particles was studied using operando 2D TXM-XANES.The main results are given in Figure 5.During the charging process, 2D XANES maps were collected at different SoCs from multiple secondary particles in a large area (%30 Â 30 μm; Figure 5a,b).Data from one particle that was well separated from other particles are given in Figure 5c, with false colors (red for Ni 2þ , green for Ni 3þ , and blue for Ni 4þ , respectively).At the early stage of charging, the outer layer in red (mostly Ni 2þ ) was maintained overall, but it began to shrink at around t = 5.0 h (at %3.9 V), owing to oxidation to Ni 3þ (turning to green).During charging to voltages above 3.9 V, Ni 4þ (blue) appears in the core region and the Ni 2þ (red) in the surface layer starts to vanish because of the oxidization of Ni to the higher valence states, both at the surface region and in the bulk.Upon further charging, more and more of the core region turns to blue (Ni 4þ ) at the expense of the green region (Ni 3þ ).By the end of charging (t = 14.2 h), only a very thin surface region remains green (%3þ).Similar behavior is seen from the examination of another set of particles (see Figure S5 and Movie S2, Supporting Information).The maintained low valence state of Ni (%3þ) near the particle surface is due to the presence of Ni-gradient layer (Figure 2), as further discussed later through quantitative spectral data analysis.Such inhomogeneous valence distribution across secondary particles was not observed in the secondary particles with homogeneous composition, [35] suggesting the phenomenon may not be due to the preferential Li depletion across secondary particles.
To quantify the Ni valence change by 2D XANES mapping and its dependence on the local stoichiometry across the surface and core regions, XANES spectra were selected from different regions of interest (ROIs), both from the surface (S1) and the core (C1) regions, as shown in Figure 5d,e.The XANES K-edge of Ni in the surface region shifts continuously to higher energies until the end of charging (Figure 5d).In the core region, the K-edge of Ni shifts gradually to higher energies, reaching its highest energy position at t = 9.0 h, and it remains at almost the same energy position as the charging is continued (Figure 5e).A similar analysis was also done for other ROIs, which showed similar behavior (Figure S6, Supporting Information).
Quantitative analysis of the local redox was performed through spectral fitting (with examples provided in Figure S7, Supporting Information), with the main results provided in Figure 5f.In the initial state, the valences of Ni were about 2.40þ and 2.95þ in the surface and core regions, respectively, and they increased in both regions upon charging.The Ni valence in the core reached its limit at 4.2 V and remained almost constant up to the end of charging (even after 1 h of holding at 4.4 V).The slight increase (by 0.5%) is mostly attributed to the sampling of the surface layer (wherein Ni is continuously oxidized).The valence change of Ni  c) Zoom-in view of the valence maps within a single selected particle as a function of charging time (as labeled).Red, green, and blue pixels represent Ni 2þ , Ni 3þ , and Ni 4þ , respectively (see also Movie S1, Supporting Information).d,e) XANES spectra of the Ni K-edge from the surface region (S1 in Figure 5c) and core region (C1 in Figure 5c), respectively.f ) Quantitative analysis of valence evolution of Ni in the selected core and surface regions during charging.For better statistics, the valence of Ni was calculated by averaging the fitting values from a number of spectra obtained in the core and surface regions (as marked in Figure S6, Supporting Information).The valence of Ni for the whole particle was also obtained by fitting the average XANES spectra obtained from all pixels of the particle.The error bar represents the standard deviation.g) Proportion from the surface (orange) and core (purple) regions participating in Ni oxidation during charging, derived from Figure 5f by calculating the surface/core valence weight over the whole particle.in the surface region followed a different trend, showing a rapid increase at voltages below 3.9 V, followed by a slow but continuous increase until the end of charging.During the whole charging process, the valence of the whole particle stayed between that of core and surface, with a value close to 4þ at the end of charging owing to the dominant volume proportion of the core region.The evolution of region proportion during charging (Figure 5g) was evaluated by calculating the weight of the valence in the surface/core over the whole particle.It is observed that the surface region accounted for most of the early-stage charge compensation (t < 5.0 h), with the weight ranging from 27% to 57%.Once the voltages increased above 3.8 V (t > 5.0 h), the core region dominated the charge compensation and reached 98% weight at the end of charge.The time variation of the regional proportion indicates the interaction between surface and core regions, which has a vital influence on the stability of high-Ni cathode materials, as discussed later.

Charge Compensation During Cycling
To complement the 2D TXM-XANES results on local redox, operando Ni K-edge XAS measurements were performed on LSC-NMC9055 and homo-NMC9055 electrodes to investigate the charge compensation during cycling (up to 4.4 V).The main results, including both XANES and Fourier transformation of the extended X-Ray-absorption fine structure (EXAFS), are provided in Figure 6.As shown in Figure 6a,b, the peak positions of Ni K-edge XANES spectra from both LSC-NMC9055 and homo-NMC9055 gradually shifted to high energies during the early charging.As charging increased above 4.2 V, the peak positions remained almost the same, as Ni has already reached its full oxidation state, which is consistent with local studies as shown in Figure 5.The trend is verified by quantitative analysis through linear combination fitting (Figure 6c) and further by the evolution of the local environment in the EXAFS spectra (Figure 6d).Specifically, the first coordination of homo-NMC9055 shifted from 1.942 to 1.891 Å (ΔR = 2.5%) at the beginning of charging, which corresponds to the continuous shrinking of the Ni-O bond (Figure 6e).A similar trend was observed in the second coordination of homo-NMC9055, where the R value shifted from 2.874 to 2.837 Å (ΔR = 1.3%), indicating shrinkage of the Ni-TM (TM representing Ni, Mn, and Co) bond.This shrinkage of both the Ni-O bond and Ni-TM bond demonstrates the decrease in the ionic radius of Ni because of oxidation.6][37][38] However, further charging above 4.2 V causes little change to the first and second coordination, with only a 0.00075 Å (ΔR = 0.039%) change in the Ni-O bond length and a 0.006 Å (ΔR = 0.216%) change in the Ni-TM bond length, supporting the claim that Ni has almost reached its full oxidation state.There is approximately 20% charge capacity delivery in LSC-NMC9055 or homo-NMC9055 above 4.2 V, but Ni has been almost fully oxidized and cannot contribute to such a large capacity.
To identify the contribution to the excess capacity beyond 4.2 V charging, valence change in the Co and Mn was also examined in homo-NMC9055 during the first charge (as given in Figure S8, Supporting Information), indicating that neither of Co or Mn was oxidized above 4.2 V (see detailed discussed in Note S3, Supporting Information).This finding is consistent with other reports that, at voltages >4.2 V, oxygen rather than the TM ions (Ni, Co, and Mn) is the electron donor for the charge compensation in high-Ni NMC systems. [10,11,39,40]s is well established, cationic redox with delithiation reaction below 4.2 V does not cause capacity decay in high-Ni cathodes, [41,42] whereas severe capacity fading due to abrupt non-monotonic changes of the lattice framework and surface reconstruction has been reported to occur at high voltages because of the associated oxygen release. [7,8,35,39]Although lattice oxygen was involved in charge compensation in both LSC-NMC9055 and homo-NMC9055 during high-voltage charging, Ni redox in LSC-NMC9055 was more reversible than that in homo-NMC9055 (as shown in Figure 2 and 3); this reversibility should be correlated to the asynchronous redox process between surface and core regions of LSC-NMC9055 (Figure 5g).The delayed reaction of the surface region originates from LSC.To further verify that statement, the redox process of NMC333 (which has the similar composition as the surface end-member in LSC-NMC9055) was investigated by operando XAS measurement (Figure S9-S12, Supporting Information), and is discussed in Note S4 (Supporting Information), indicating that TMs can provide charge compensation even at charge voltages beyond 4.2 V, therefore causing a fundamentally different redox process in LSC-NMC9055 versus homo-NMC9055.
Operando XRD measurement was also carried out on LSC-NMC9055 during charging in the voltage range from 2.7 to 4.4 V (Figure S13, Supporting Information).The refinement results show an expansion of lattice parameter c from 14.159 to 14.291 Å, and then a slight collapse from 14.291 to 13.840 Å (Δc = 3.16%).The lattice collapse Δc of LSC-NMC9055 is about 2/3 of the reported values of homogeneous high-Ni NMC, [43,44] therefore making the secondary particles robust during cycling.

Thermal Stability
The thermal stability of electrodes in the fully charged state (to 4.4 V) was evaluated with respect to exothermic reactions using differential scanning calorimetry (DSC).During heating at a rate of 5 °C min À1 , the heat flow of the electrodes was measured as a function of temperature (Figure 1a).In contrast to the measured high-temperature, low-intensity exothermic peak in NMC333, homo-NMC9055 exhibited a sharp exothermic peak at 219 °C (with a total released heat of 2250 J g À1 ).Strikingly, in LSC-NMC9055, a much broader exothermic peak was observed, with the main peak at 258 °C and a shoulder at around 275 °C, indicating a much-enhanced thermal stability in the presence of the Ni-gradient layer.
The much-improved thermal and high-voltage cycling stabilities in LSC-NMC9055 compared to homo-NMC9055 (Figure 2) are attributed to the higher structural stability enabled by the Ni-gradient layer.In addition, the particle architecture, i.e., the close-packing primary particles in LSC-NMC9055 (as shown by the cross-section high-angle annular dark-field [HAADF]  image, Figure 6b) may also play an important role, particularly in suppressing oxygen release under the thermal treatment and electrochemical cycling conditions, as discussed later.

Discussion
High Ni loading in layered oxides is desired for high specific capacities but the high Ni loading creates structural-instability-related issues during high-voltage operation and consequently compromises the lifetime and safety of the cells. [6,45,46]In the present work, through LSC to construct a Ni-gradient layer, both high cycling stability and thermal stability were obtained in LSC-NMC9055, giving rise to the exceptional performance (Figure 2, 3, and 7a).The high structural and chemical stabilities were evidenced by the retention of the Ni-gradient layer and the particle integrity in 2D/3D TXM elemental/valence maps.Further multimodal operando X-Ray characterization, using XAS combined with 2D-XANES, unveiled the direct correlation between the bulk and local electrochemical redox.Specifically, results from XAS studies of the homo-NMC9055 and NMC333 (similar to the core and surface compositions, respectively, of the hetero-NMC9055) are consistent with those from the TXM-XANES studies, indicating the interplay between electrochemical redox and local stoichiometry.47] From these measurements, we propose a mechanism that enables structure stabilization in LSC-NMC9055, as illustrated in Figure 7c.The covalency of the Ni-O bond, predicted by theoretical calculations in high-Ni NMC, may apply to the homo-NMC9055: namely, the significant overlap between Ni and oxygen bands facilitates oxygen redox at high voltages (above 4.2 V). [48] Also, the low Co content in the high-Ni NMCs brings about the instability of Ni in the TM layer because of the opposing magnetic moments.For magnetic frustration relief, Ni prefers to migrate to the Li site and induce a structural transition. [49]After charging to a high voltage (above 4.2 V), oxygen redox may occur and the unstable oxidized O species on the particle surface tend to form oxygen gas or react with electrolyte (as illustrated in Figure 7c). [7]As a consequence, the resulting oxygen vacancies promote Ni migration from the TM layer to the lithium layer and eventually lead to a phase transition from layered spinel to rock salt. [40,50]The high concentration of oxygen vacancies on the surface layer may induce the injection of oxygen vacancies into the bulk, which further facilitates Ni migration and deteriorates the bulk structure. [51]This behavior may explain the fast capacity decay in the homo-NMC9055 (Figure 1).Moreover, the oxygen vacancies also contribute to poor thermal stability of high-Ni NMC, [52] which is also demonstrated in the homo-NMC9055 (Figure 1c, top).
Oxygen redox may also be involved in LSC-NMC9055 at high voltages, but only in the core region.Although anionic redox activity has been demonstrated in NMC333 (the similar composition as the surface region), it only occurs at high voltages (%4.6 V), well above the upper-limit voltage (4.4 V), and shows negligible effects at 4.4 V and lower voltages. [47]As a result, the oxygen framework is well preserved in the surface region of LSC-NMC9055 (as illustrated in Figure 7c, bottom).Owing to the high kinetic barriers for dimer formation and low diffusivity as predicted by theoretical calculations, [53] oxygen can be confined in the core.Recent studies have further confirmed that the structural reconstruction and oxygen release only occur in the near-surface region of close-packed secondary particles, rather than the core region, provided that there is no direct contact with liquid electrolyte. [31,32]In other words, oxygen release can be kinetically prohibited by a stable surface-oxygen framework and close packing of primary particles within the secondary particles.The local stoichiometry, with relatively higher Mn/Ni ratio in the surface region, has a positive effect on stabilizing the oxygen framework and reduces the irreversible side reactions with electrolyte, owing to the lower reaction activity of Mn 4þ relative to Ni 4þ . [46]Normally, stronger TM-O ionicity corresponds to a larger bandgap, so inducing formation of ions with higher ionicity such as Mn 4þ can effectively hinder oxygen redox and improve the stability of the oxygen framework. [54,55]Therefore, continuous oxidation of Ni and/or Co without involving oxygen redox in the low-Ni surface region of LSC-NMC9055 is crucial to the cycling and thermal stability.In addition, the LSC also suppresses the notorious lattice collapse, which is considered as the leading cause of cracking and pulverization of high-Ni NMC secondary particles. [56]Further evidences were obtained by operando XRD measurements, showing much reduced lattice change in the LSC-NMC9055, with Δc = 3.16% during charging in the voltage range from 2.7 to 4.4 V (Figure S13, Supporting Information), much lower than the reported Δc values of homogeneous high-Ni NMC. [43,44]

Conclusion
In this work, we report a rational design of LSC high-Ni cathodes (up to 90% Ni), with high cycling and thermal stability enabled by the heterostructure in secondary particles.Specifically, the LSC-NMC9055, with a Ni-gradient layer covering an utrahigh-Ni core, delivered capacities as high as 200 mAh g À1 with 91.4% retention for 100 cycles at voltages up to 4.4 V. Exceptional cycling performance of 80% retention for 700 cycles was obtained at voltages between 3.0 and 4.3 V against graphite in the full cells.Evidence for structural stabilization of LSC-NMC9055 was obtained by DSC, showing its high thermal stability at up to 258 °C, and by operando multimodal X-Ray characterization of the interplay between local stoichiometry and electrochemical redox.Unlike homo-NMC9055, in which both Ni and lattice oxygen were involved in charge compensation across the secondary particles, the high-Ni core and low-Ni surface regions in LSC-NMC9055 went through distinct redox routes.As a consequence, a stable inactive lattice-oxygen framework can remain as evidenced by the retention of Ni-gradient layer and the particle integrity even after elongated cycling, indicating the crucial role of the Ni-gradient layer in alleviating oxygen release and structure deterioration.We believe, with the demonstrated practical solution to the long-standing issues preventing high-voltage operation of high-Ni cathodes, this work constitutes an important step toward the deployment of these cathodes in next-generation batteries.was fed from the storage tank into the first CSTR system.Simultaneously, 10 mol L À1 NaOH solution (aq.) for pH adjustment (molar ratio of sodium hydroxide to TM%1.5) and 3.8 mol L À1 NH 4 OH solution (aq.) for chelating purposes (molar ratio of ammonium hydroxide to TM = 1.0) were fed into the first CSTR system, along with a measured amount of deionized water.In this first CSTR system, the precursor with the core structure grows to the desired particle size through a coprecipitation process, and the core particles are separated via a hydro-cyclone for transfer to the second CSTR system.Sequentially, to establish the surface composition, a high-Mn aqueous solution, composed of NiSO 4 •6H 2 O, CoSO 4 •7H 2 O, and MnSO 4 •5H 2 O (molar ratio of Ni:Co:Mn = 33:33:33), was fed from the storage tank into the second CSTR system.Simultaneously, 10 mol L À1 NaOH solution (aq.) for pH adjustment and 3.8 mol L À1 NH 4 OH solution (aq.) for chelating purposes were fed into the second CSTR system, along with a measured amount of deionized water.Thereafter, the core precursor, [Ni 0.90 Co 0.05 Mn 0.05 ](OH) 2 , was transferred from the first CSTR to the second CSTR containing the surface precursor, [Ni 1/3 Co 1/3 Mn 1/3 ](OH) 2 .The precursor with the heterogeneous structure, LSC-NMC9055, was formed and grown to the desired particle size through a coprecipitation process, and the formed heterogeneous particles were separated via a hydro-cyclone to be collected.Then, the LSC-NMC9055 precursor was filtered, washed, and dried for 20 h at 100 °C.Next, the dried precursor was mixed with LiOH•H 2 O and calcined at 800 °C for 20 h under oxygen flow to produce the heterogeneously structured LSC-NMC9055 cathode material.

Experimental Section
For electrochemical tests, the LSC-NMC9055 powder, Super P carbon, and 8 wt% polyvinylidene fluoride (PVDF) binder solution dissolved in N-methyl-2-pyrrolidone were mixed at a 90:5:5 weight ratio.The spread slurry was dried in an oven at 80 °C for 3 h, and 2032-type coin-cell tests were performed with pure lithium metal as the opposite (anode) electrode.The electrolyte solution was Gen2, 1.2 M LiPF 6 in a mixture of ethylene carbonate and ethyl methyl carbonate in a volume ratio of 3:7, and the separator was Celgard 2325.The half coin cells were assembled in an Ar-filled glove box and tested with a Maccor (Series 4000) apparatus for cycle performance, with current density at a rate of 0.1 C (20 mA g À1 ) and a voltage window between 2.7 and 4.4 V at a temperature of 30 and 50 °C.For comparison, homo-NMC9055 and NMC333 were also tested with the same procedure.
Electrode Fabrication and Operando Measurements: The 2032-type coin cell, with holes on both sides, was specially designed to perform operando measurements.The holes, kept small (%3 mm) to ensure a small cell impedance, were sealed by Kapton tape to exclude oxygen and moisture.The operando hard X-Ray chemical mapping was performed with full-field TXM at the 18-ID (FXI) beamline, National Synchrotron Light Source II, Brookhaven National Laboratory.To acquire a high-quality 2D image, the dispersed electrodes for operando measurements were made of 25 wt% active material, 30 wt% carbon black, and 45 wt% PVDF binder.Carbon papers (%100 μm thickness) were used as current collectors for the electrodes.The cells were fixed in a custom-built holder mounted on a motorized X, Y, Z, and R stage.The energy of the incident X-Ray was scanned from 8200 to 8500 eV to cover the Ni K-edge XANES region.The energy step was set to 1 eV in the range from 8320 to 8400 eV to improve data quality and 10 eV in the other ranges to reduce acquisition time.A low current density (0.1 C, 1 C = 200 mA g À1 ) was applied in all operando measurements.
The single-layer pouch cells and correlated baseline NMC811, core-shell NMC811, core-gradient NMC811, and capacity-matched graphite electrodes were fabricated by the cell analysis, modeling, and prototyping (CAMP) facility at Argonne National Laboratory in a climate-controlled dry room with a dew point less than À42 °C (<100 ppm moisture).The following processes are described in greater detail in prior work. [57]The single-side electrodes were coated using a roll-to-roll reverse comma coater and calendered by a hydraulic-driven roll press.The cathodes were coated on 20 μm aluminum foil and the anode on 10 μm copper foil.The electrodes were then punched to the dimensions pertaining to a single layer pouch cell (xx3450) format, with the cathode and anode dimensions being 31.3mm wide and 45.0 mm high (14.1 cm 2 ) and 32.4 mm wide and 46.0 mm high (14.9cm 2 ), respectfully.Then All all electrodes were dried overnight under vacuum at 120 °C The assembled cells used Celgard 2320 as the separator, were dried at 60 °C overnight under vacuum, and then were filled with 1.2 M LiPF6 in EC:EMC (3:7 by wt%) electrolyte targeting a electrolyte volume to pore volume of %4.4Â.
The same electrode/cell fabrication of the LSC-NMC9055 was used for operando XRD, which was performed at beamline 28-ID-2, National Synchrotron Light Source II.Data were collected with a fixed wavelength of 0.1885 Å.During charging, a 2D X-Ray detector was deployed to collect the XRD patterns.The focused spot size was 0.5 mm (horizontal) Â 0.5 mm (vertical).
Operando XAS experiments were carried out at beamline 7-BM, National Synchrotron Light Source II.Energy calibration of each spectrum was performed by aligning the first-derivative maximum of the reference K-edge XANES spectra of TMs (i.e., Ni: 8333 eV), which were collected simultaneously from the metal foils by the reference detector.
Samples for cross-section transmission electron microscopy (TEM) of LSC-NMC9055 were prepared using a focused-ion beam (FIB).The TEM lamellae were produced using the standard in situ lift-out procedure on the Helios 600 Nanolab dual-beam FIB with final thinning performed at 5 keV.The thickness of the prepared TEM samples was approximately 50 nm.
Thermal Measurements: DSC analysis was performed on the fully charged electrodes (to 4.4 V) made of LSC-NMC9055, homo-NMC9055, and NMC333, using an STA 449 F3 Jupiter (NETZSCH, Germany).The measurements were done in the presence of electrolyte solution within high-pressure sample crucibles of chrome-nickel steel.The ratio of the cathode material to electrolyte solution was kept the same in measuring the three different samples.The heat flows of the cathode materials were measured as a function of temperature during heating at 5 °C min À1 , from room temperature up to 380 °C.From the experiments, the onset temperature, heat flow, and total heat generation as a function of mass were obtained.
Data Analysis: The Ni valence-states map was analyzed through a linear least-squares fitting of each pixel of the 2D XANES spectrum collected at the 18-ID beamline, given the reference spectra of Ni with different oxidation states provided.An in-house package (PyXAS) was used to perform the fitting process.Program ImageJ was used to perform valence mapping for the selected lines. [58]he Athena software package [59] was used to perform linear combination fitting for Ni K-edge XANES.The fitting range is from À20 to 30 eV deviation from the energy of edge E 0 .When fitting the Ni K-edge spectra, commercial NiO and LiNiO were chosen as references for Ni 2þ and Ni 3þ , respectively.To get a reference for Ni 4þ , the spectrum with the highest energy shift among all pixels was chosen.The EXAFS data were preprocessed by Athena software and then fitted by Artemis software.The normalized k 3 -weighted EXAFS spectra were Fourier-transformed in k space with integration limits of 3.2-10.5Å À1 for the Ni, Co, and Mn data.The least-square fits were carried out in R space between 1.0 and 3.0 Å.To simplify the fitting algorithm, the LiNiO 2 (space group: R-3m) structural model was used to perform EXAFS fittings.The Li content in the models was adjusted according to the capacity analysis.Because of the coupling effect of the distortion factor σ 2 and the coordination number, we refined only the distortion factor σ 2 and fixed the coordination number for the sake of clear analysis.
Rietveld refinements of XRD patterns were carried out using the FullProf Suite. [60]LiNi 0.8 Co 0.1 Mn 0.1 O 2 (SG: R-3m) was used as the structural model.The refining parameters included background coefficients, overall atomic displacement parameter, peak shape parameters, lattice parameters, positions of atoms, and occupancy of Li, Ni, Co, and Mn.A constrained condition was set to make the sum of Li and Ni occupancy at 3a sites or 3b sites be equal to 1.

Figure 1 .
Figure 1.Heterostructure properties of local stoichiometry-controlled LiNi 0.90 Mn 0.05 Co 0.05 O 2 (LSC-NMC9055).a) Local Ni valence distribution using false-colored 2D valence mapping, with the red color for Ni 2þ and green for Ni 3þ .Inset: relative concentration ratio of Ni 2þ /Ni 3þ across a single secondary particle (from the region indicated by the white dashed lines; see also Figure S3, Supporting Information for more analysis).b) Schematic illustration of transmission X-Ray microscopy X-Ray absorption near-edge structure spectroscopy (TXM-XANES) mapping for determining Ni-concentration distribution by projection, using the hypothetical core-shell model (top) and gradient model (bottom).See details in Note S2, Supporting Information.c) Projection of the 2D Ni concentration determined by TXM-XANES measurement and fitted radial Ni concentration, showing the gradient Ni-concentration layer in the near-surface region (marked with orange rectangles).Details are provided in Note S2 and Figure S4, Supporting Information.d) Synchrotron X-Ray diffraction (XRD) patterns of LSC-NMC9055, in comparison to the patterns calculated by Rietveld refinement.In the plots, Bragg positions associated with rhombohedral symmetry (space group: R-3m) are indicated by green bars.Black circles and red and blue lines are used for the experimental pattern, the calculated pattern (from the Rietveld refinement), and their difference, respectively.

Figure 2 .
Figure 2. Electrochemical properties of the LSC-NMC9055 in half-cells (against Li metal).a) Voltage profiles for the first charge and discharge and b) cycling performance of LSC-NMC9055 (orange) between 2.7 and 4.4 V at a rate of C/10 (1 C = 200 mAh g À1 ), in comparison to that of homo-NMC9055 (black) and NMC333 (blue), between 2.7 and 4.4 V at a rate of C/2.c-e) Differential capacity (dQ/dV ) plots as a function of cycles from the 1st up to the 100th cycle (measured at every 20 cycles), obtained from NMC333, LSC-NMC9055, and homo-NMC9055, respectively.Inset: zoom-in view of the redox peak at around 4.2 V, showing the retention of the peak after 100 cycles in LSC-NMC9055 in contrast to the fast decay in homo-NMC9055.

Figure
Figure 3c,d, respectively.The inferior behaviors were observed in the baseline NMC811 and core-shell NMC811.

Figure 3 .
Figure 3. Electrochemical performance of the LSC-NMC9055 in pouch-type full cells (against graphite): a) discharge capacity, b) discharge voltages, c) Coulombic efficiency, and d) discharge area-specific impedance (ASI) at 50% depths of discharge (DOD), measured from LSC-NMC9055 (green circles) between 3.0 and 4.3 V at a rate of C/3 charge, C/2 discharge.Data taken under the same conditions from the baseline NMC811 (blue circles) and core-shell NMC811 (red circles) were provided for comparison to NMC9055.More details on the electrochemical tests were provided in Experimental Section.

Figure 4 .
Figure 4. Retention of the heterostructure in LSC-NMC9055 after long cycling.a,b) The 3D and 2D maps of local elemental distribution in the pristine and cycled LSC-NMC9055, respectively, via X-Ray tomography.The volume rendering and 2D cross section of the Ni distribution shows the maintained morphology and local stoichiometry of LSC-NMC9055 after 120 cycles (scale bar: 8 μm).c-e) False-colored 2D Ni valence distribution in LSC-NMC9055 after 120 cycles, shown by the Ni 2þ (red), Ni 3þ (green), and the composite maps, respectively (scale bar: 8 μm).f ) Relative concentration ratio of Ni 2þ /Ni 3þ across a single secondary particle of LSC-NMC9055 after 120 cycles (from the region indicated by the white dashed box).g) Comparison of relative concentration ratio of Ni 2þ /Ni 3þ between pristine sample (dashed line) and the sample after 120 cycles (solid lines).

Figure 5 .
Figure 5. Tracking of local redox in LSC-NMC9055 during the 1st charge by operando 2D TXM-XANES.a) Valence distribution of Ni from multiple particles of the pristine sample in a large area.b) Voltage profile recorded from the LSC-NMC9055 electrode during the charge process, up to 4.4 V (at 0.1 C).c) Zoom-in view of the valence maps within a single selected particle as a function of charging time (as labeled).Red, green, and blue pixels represent Ni 2þ , Ni 3þ , and Ni 4þ , respectively (see also Movie S1, Supporting Information).d,e) XANES spectra of the Ni K-edge from the surface region (S1 in Figure5c) and core region (C1 in Figure5c), respectively.f ) Quantitative analysis of valence evolution of Ni in the selected core and surface regions during charging.For better statistics, the valence of Ni was calculated by averaging the fitting values from a number of spectra obtained in the core and surface regions (as marked in FigureS6, Supporting Information).The valence of Ni for the whole particle was also obtained by fitting the average XANES spectra obtained from all pixels of the particle.The error bar represents the standard deviation.g) Proportion from the surface (orange) and core (purple) regions participating in Ni oxidation during charging, derived from Figure5fby calculating the surface/core valence weight over the whole particle.

Figure 6 .
Figure 6.Charge compensation in the bulk electrodes during charging/discharging visualized by operando X-Ray absorption spectroscopy (XAS).a) Voltage profiles recorded from LSC-NMC9055 and homo-NMC9055 during operando XAS measurement as the electrodes were cycled in the voltage range of 3.0-4.4V, at a rate of 0.1 C. b) From the bottom up, Ni K-edge XANES spectra taken from the two electrodes at consecutively later times during the 1st cycle.Black dashed lines represent XANES at the fully charged state.c) Valence change of Ni during the 1st cycle, calculated by linear combination fitting, showing much higher reversibility of redox in LSC-NMC9055 (red) compared to homo-NMC9055 (black).d) k 3 -weighted Fourier-transformed Ni K-edge XAS fine structure (EXAFS) spectra (circles) and fitting data (lines) for homo-NMC9055 during the 1st charge.e) Variation of Ni-O and Ni-TM bond lengths for homo-NMC9055, showing a rapid decrease at the early stage (below 4.2 V, labeled by black dashed line), and little change at voltages above 4.2 V.

Figure 7 .
Figure 7. Mechanisms for thermal and high-voltage cycling stabilities of LSC-NMC9055.a) Thermal release profiles of the electrodes charged to 4.4 V, produced by differential scanning calorimetry (DSC; scanning at a rate of 5 °C min À1 ), showing greatly improved thermal stability of LSC-NMC9055 (orange) compared to homo-NMC9055 (black).The DSC profile from NMC333 (blue) is also provided for reference.b) High-angle annular dark-field (HAADF) image with the zoom-in view (inset) showing the close packing of particles.c) Electrochemical redox process in homo-NMC9055 (top) in comparison to that in LSC-NMC9055(bottom). Owing to the high Ni content in homo-NMC9055, lattice oxygen gets involved in charge compensation at voltages above 4.2 V (in addition to Ni), leading to oxygen release and cation migration in the near-surface region.In contrast, in LSC-NMC9055, lattice oxygen is only involved in the core region when charged to voltages above 4.2 V, and in the presence of a Ni-gradient layer (with low Ni concentration), oxygen release is highly suppressed.