Distinguishing bulk redox from near-surface degradation in lithium nickel oxide cathodes

Abstract

Ni-rich layered oxide cathodes can deliver higher energy density batteries, but uncertainties remain over their charge compensation mechanisms and the degradation processes that limit cycle life. Trapped molecular O₂ has been identified within LiNiO₂ at high states of charge, as seen for Li-rich cathodes where excess capacity is associated with reversible oxygen redox. Here we show that bulk redox in LiNiO₂ occurs by Ni–O rehybridization, lowering the electron density on O sites, but importantly without the involvement of molecular O₂. Instead, trapped O₂ is related to degradation at surfaces in contact with the electrolyte, and is accompanied by Ni reduction. O₂ is removed on discharge, but excess Ni²⁺ persists forming a reduced surface layer, associated with impeded Li transport. This implicates the instability of delithiated LiNiO₂ in contact with the electrolyte in surface degradation through O₂ formation and Ni reduction, highlighting the importance of surface stabilisation strategies in suppressing LiNiO₂ degradation.

---

Broader context

Increasing the capacity of Li-ion batteries requires cathodes which can reversibly deintercalate more lithium without leading to structural instability and severe capacity fade. To this end, Ni-rich layered cathodes are under development for next-generation batteries, with LiNiO₂ the archetypal system for investigating their charging mechanisms. However, the role played by different redox centres in LiNiO₂ is still debated, and the connections with structural instabilities and associated degradation are not yet fully established. Recent reports have suggested the involvement of molecular O₂ in the bulk redox process at high states of charge, with direct experimental detection of O₂ based on techniques that probe 100–200 nm into the surface of the few μm-sized cathode particles. Here, we combine a broad suite of X-ray spectroscopies with varying information depths (10 nm to 10 μm) to separate the bulk redox from surface degradation. We reveal that trapped O₂ formation in LiNiO₂ is primarily associated with degradation at surfaces in contact with the electrolyte, rather than contributing to the bulk redox process. Interpretation of experimental spectra using theoretical calculations shows that bulk charge compensation proceeds by Ni–O rehybridization. These findings highlight the importance of using bulk sensitive techniques to understand redox, and suggests design strategies for stabilising high energy density Ni-rich cathodes.

Introduction

Layered transition metal (TM) oxides, LiTMO₂ (TM = Co, Ni, Mn, etc.), are the cathode materials of choice for commercial high-energy density Li-ion batteries, reversibly intercalating Li over thousands of cycles.^1,2 Ni-rich stoichiometries are increasingly favoured to increase capacity, and lower Co content, which is expensive and has ethical concerns around its mining.³ In the traditional picture of charge compensation, Li⁺ removal is compensated by an increase in the formal oxidation state of the redox-active TM centres via a single electron transfer. However, it is well-established that there are accompanying changes in TM–O bond covalency meaning both TM and O sites are involved.^4–9

The archetypal Ni-rich cathode material, LiNiO₂, undergoes several first-order structural phase transitions upon delithiation, with significant degradation observed at high potentials that has been associated with severe lattice changes,^10,11 structural degradation,^12,13 gas evolution,^14,15 and parasitic reactions with the electrolyte.^16,17 However, the causality and connections between these different modes of degradation are not yet fully established. Oxygen loss from the cathode surface and the associated formation of a reduced surface layer have been widely observed for Ni-rich cathode materials particularly above the H2–H3 transition,^15,18–20 and are found to depend critically on the upper cut off voltage (UCV) and electrolyte formulation.^17,21–23 Recently, studies of Ni-rich cathodes at high potentials (≥4.3 V during charge) have also shown the emergence of a distinct signature in O K-edge resonant inelastic X-ray scattering (RIXS) spectra at an excitation energy of ∼531.5 eV.^24–26 For Li-rich materials that show excess capacity beyond TM cation redox, this signature is typically taken as evidence of the formation of molecular O₂ trapped in pores throughout the cathode bulk, as a result of charge compensation by non-bonding O orbitals.^27,28 However, the TM vacancies necessary to accommodate this are not expected in LiNiO₂, with experimental samples typically containing excess Ni. Furthermore, LiNiO₂ does not exhibit excess capacity that might be associated with molecular O₂ redox. Nevertheless, bulk sensitive Ni K-edge X-ray absorption Near Edge Structure (XANES) measurements of LiNiO₂ have indicated a plateauing of the main edge half-height position at similarly high states of charge, which has been taken as evidence of the formal Ni oxidation state no longer changing in the bulk, and thus a change in the redox mechanism.²⁹

Although O K-edge RIXS and the related fluorescence yield X-ray absorption spectroscopy (FY-XAS) are widely referred to as bulk-sensitive (∼200 nm information depth), in the context of Li-ion cathode materials where typical secondary particle diameters are >5 μm, these methods probe <10% of the particle volume nearest to the surface. Attribution of bulk molecular O₂ redox based on these methods alone is therefore ambiguous. Similar concerns have been raised around identifying oxygen redox with hard X-ray photoelectron spectroscopy (HAXPES), where typical probing depths are tens of nm.³⁰ Solid-state ¹⁷O magic-angle-spinning nuclear magnetic resonance (NMR) spectroscopy provides an alternative bulk-averaged approach to estimate the amount of O₂ present in the lattice.³¹ However, it does not resolve the spatial distribution of O₂ molecules, nor has it been reported for LiNiO₂ to date. Whereas, online electrochemical mass spectrometry (OEMS) can quantify the gas release associated with O-loss from the cathode surface, it does not probe molecular O₂ that remains trapped within the cathode.^15,20,32 The extent to which oxygen redox is involved in charge compensation in the LiNiO₂ bulk thus remains unclear, motivating approaches that can provide comparable information with surface- and bulk-sensitivity.

Here we combine complementary core-loss spectroscopies to obtain a depth-resolved (10 nm to >10 μm) account of the redox processes in LiNiO₂ and distinguish reversible bulk redox processes from near-surface degradation. X-ray Raman Spectroscopy (XRS, >10 μm information depth) reveals that in the bulk of LiNiO₂ secondary particles there is a continuous change in both the Ni L_3,2-edge and O K-edge spectra with state of charge (SoC) up to 4.8 V. This is consistent with charge compensation proceeding by rehybridization between the Ni and O centres, lowering the electron density on O sites but with Ni–O coordination still preserved. Features of trapped molecular O₂ appear at potentials of ≳4.2 V in O K-edge FY-XAS, accompanied by increased Ni²⁺ contributions in the Ni L_3,2-edge. Importantly, these changes are less pronounced in bulk-averaged XRS measurements indicating that formation of molecular O₂ is a predominantly surface process. Total Electron Yield (TEY)-XAS measurements (∼10 nm information depth) confirm that a densified NiO-like layer forms in direct contact with electrolyte, whilst FY-XAS measurements are consistent with an extended cation mixing layer in which Ni²⁺ ions have migrated to occupy Li sites. Scanning transmission electron microscopy–electron energy loss spectroscopy (STEM–EELS) further confirms this picture of a reduced surface layer (RSL) that extends ∼200 nm into the surface for LiNiO₂ which has been cycled to 4.8 V vs. Li/Li⁺. This understanding emphasises the importance of strategies to stabilise the interfaces of Ni-rich cathode materials in contact with electrolyte (e.g. cathode coatings/gradients, electrolyte formulation), rather than bulk stabilisation approaches (e.g. pillaring) that might sacrifice capacity.

Results and discussion

Trends in bulk chemical state

The charge–discharge profile for the 2nd cycle of the composite polycrystalline LiNiO₂ electrode is shown in Fig. 1(a), together with an inset showing a scanning electron micrograph (SEM) of the pristine LiNiO₂ active material (see ESI,† Fig. S1 for further characterisation). The ∼5 μm diameter spheroidal LiNiO₂ secondary particles are composed of sub-μm primary particles. The voltage profiles show distinct plateaus associated with the first-order structural phase transitions of LiNiO₂ on delithiation, apparent as maxima in the dQ/dV plots (Fig. 1(b)), at potentials consistent with prior literature.^15,33,34 Powder X-ray diffraction (XRD) of the pristine material (see ESI,† Fig. S1c) closely resembles the calculated pattern for LiNiO₂ with the R3̄m space group. This hexagonal H1 phase transitions to the monoclinic M phase at ∼3.67 V, then to the H2 phase at ∼4.0 V, followed by the H3 phase at ∼4.2 V. The voltage profile shows noticeable hysteresis above ∼4.3 V, with the voltage rapidly dropping from 4.8 V to ∼4.2 V on discharge. However, the capacity reached at 4.8 V is 256 mA h g⁻¹ which compares with a maximum theoretical capacity of 264 mA h g⁻¹, based on the pristine material having ∼4% Ni excess as determined by inductively coupled plasma-optical emission spectroscopy (ICP-OES).²⁶ This provides an initial indication that the full capacity of the electrode can be accounted for by formal Ni redox alone, without obvious excess capacity associated with molecular O₂ redox.

Fig. 1 Bulk-sensitive probing of LiNiO₂ redox processes. (a) 2nd cycle charge–discharge profile of LiNiO₂ electrode cycled at a rate of C/20 between 3.0 and 4.8 V vs. Li/Li⁺. Inset: Scanning electron microscope (SEM) image of pristine LiNiO₂ particles. (b) Corresponding differential capacity plots (dQ/dV). (c), Normalised Ni K-edge XANES spectra (transmission mode) of LiNiO₂ at different SoC. (d) Plot of the energy shift in normalised Ni K-edge whiteline and 50% edge height positions relative to pristine LiNiO₂. (e) and (f) XRS (∼10 μm information depth) of the O K-edge and Ni L₃-edge core-loss spectra for LiNiO₂ electrodes at different SoC during the 2nd charge cycle. Experimental XRS data is marked as black dots and represented with smooth solid trace lines. Charge transfer multiplet (CTM) calculations of formally Ni²⁺ (green), Ni³⁺ (purple), and Ni⁴⁺ (pink) environments. See ESI,† Fig. S2 for fitted XRS Ni L_3,2-edges.

Fig. 1(c) shows normalised transmission Ni K-edge XANES spectra for the LiNiO₂ electrodes at different SoC (x in Li_xNiO₂) during the 2nd charge cycle. As expected, the Ni K-edge shifts to higher energies as the formal Ni oxidation state increases, with the removal of valence electrons leaving the Ni nucleus less-shielded such that it has a higher effective charge, and the core-level becomes more strongly bound. Both the energy of the fractional (normalised) edge height and the position of the whiteline (intensity maximum) are routinely used as indirect measures of average oxidation state.^35,36 A continuous shift to higher energy in both the edge half-height and whiteline is observed up to 4.2 V, x = 0.22 (Fig. 1(d)). The two trends diverge with further delithiation, with the whiteline monotonically shifting to higher energy up to the furthest measured extent of delithiation (4.8 V, x = 0.03), while the half-height position plateaus with little variation between x = 0.10 and x = 0.03. The plateau of half-height position has previously been taken as an indication that Ni is no longer involved in the redox mechanism at high SoC,^24,25 however the continuing shift in whiteline position would suggest otherwise. Indeed, the edge-position is known to be sensitive to other factors including bond length and ligand covalency.³⁷

To resolve this ambiguity without introducing surface sensitivity as a confounding factor, bulk-sensitive XRS was performed to collect O K-edge (Fig. 1(e)) and Ni L₃-edge (Fig. 1(f)) spectra at the same SoC as the XANES. XRS probes lower-energy O 1s → 2p and Ni 2p → 3d transitions using hard X-rays (10 keV), achieving an information depth of ∼10 μm which is similar to Ni K-edge XANES. In Fig. 1(e), pristine LiNiO₂ exhibits a prominent asymmetric O K pre-edge feature centred at 528.8 eV associated with transitions from O 1s → O 2p-Ni 3d hybridised states, and main edge features above 535.0 eV associated with transitions from O 1s → O 2p-Ni 4s,p hybridised states. On delithiation, the pre-edge peak is seen to continuously increase in relative intensity, whilst losing its asymmetry and shifting by 0.4 eV to a higher peak energy of 529.2 eV. There is also an accompanying shift in the main edge half-height position from ∼536.0 eV for pristine LiNiO₂ up to 539.5 eV at 4.8 V, and the shape of the main edge changes indicating a change in the O2p and Ni4s,p orbital hybridisation. Importantly, across the potentials probed, the feature arising at ∼531.5 eV associated with the formation of molecular O₂ is not strongly pronounced.^24–26

The corresponding Ni L₃-edge XRS (Fig. 1(f)) for pristine LiNiO₂ shows a broad line shape composed of three main features at 853.6 eV, 854.9 eV, and 856.1 eV. There remains debate over the ground state of LiNiO₂ (see Supplementary Note 1, ESI†) and a variety of models based on alternating layers of NiO₆ octahedra and Li have been proposed. The simplest model, in which all NiO₆ octahedra are equivalent with a formal oxidation state of Ni³⁺, is compatible with XRD data but not with measurements using more local probes.^38,39 As a result, more complex models involving time or spatially varying distortions of the octahedra have been proposed. These include structures with Jahn–Teller (J–T) distortions, where two different Ni–O bond lengths are present and the formal oxidation state remains Ni³⁺,⁴⁰ and spin disproportionated structures, where Ni²⁺ (S = 1), Ni³⁺ (S = ½), and Ni⁴⁺ (S = 0) octahedra coexist and interconvert dynamically at room temperature.^41,42 Recent temperature-dependent XAS and X-ray magnetic circular dichroism (XMCD) shows strong evidence for such disproportionation in LiNiO₂,⁴¹ which is consistent with other correlated nickelate compounds, including AgNiO₂, which show disproportionation and strong covalency between frontier O 2p and Ni 3d states.^43–47

Charge transfer multiplet (CTM) calculated L₃-edges for the three Ni environments with formal oxidation states of +2, +3 and +4 are overlaid on the pristine LiNiO₂ spectra in Fig. 1(f), corresponding to the three main features at 853.6 eV, 854.9 eV, and 856.1 eV seen in Ni L₃-edge XRS. Simulation parameters have been optimised based on experimental data (Supplementary Note 3, ESI†). Each simulated spectra can be thought of as a superposition of metal–ligand hole configurations,⁴⁷ with the formally Ni²⁺, Ni³⁺, and Ni⁴⁺ octahedra having ground-state configurations of 0.80|3d⁸〉 + 0.19|3d⁹L̲〉 + 0.01|3d¹⁰L̲²〉, 0.25|3d⁷〉 + 0.58|3d⁸L̲〉 + 0.16|3d⁹L̲²〉 + 0.01|3d¹⁰L̲³〉, 0.04|3d⁶〉 + 0.33|3d⁷L̲〉 + 0.48|3d⁸L̲²〉 + 0.14|3d⁹L̲³〉 + 0.01|3d¹⁰L̲⁴〉 respectively. In CTM calculations, increasing ligand hole contributions indicate an increasing degree of Ni–O covalency for higher formal oxidation states. Linear combinations of the simulated spectra match closely to the Ni L₃-edge spectra from XRS, FY-XAS and TEY-XAS at all SoC (ESI,† Fig. S2–S5), indicating that the simulated spectra for the Ni²⁺, Ni³⁺ and Ni⁴⁺ environments are suitable descriptions despite the small changes in octahedral environment expected for different phases.

On cycling to higher potentials, the XRS shows a continuous growth in the intensity of the Ni⁴⁺ feature (see Fig. 2(d)), initially at the expense of Ni²⁺ up to 3.9 V, x = 0.55, and then Ni³⁺ up to 4.8 V, x = 0.03. This evolution of Ni species upon delithiation matches that expected from disproportionation.⁴¹ At 4.8 V, the spectrum closely matches Ni L₃-edge simulations of Ni⁴⁺ (ESI,† Fig. S6) with 4–5% Ni²⁺. This is consistent with the excess Ni detected with ICP-OES occupying Li sites, as similarly sized Ni²⁺, and thus preventing all sites reaching Ni⁴⁺.^26,48 The bulk sensitivity of XRS suppresses contributions from surface layers which are otherwise seen even for inverse partial fluorescence yield (IPFY) measurements (ESI,† Fig. S7), including for reference Ni⁴⁺ compounds.^49,50 Importantly this shows that charge compensation in the LiNiO₂ bulk proceeds predominantly through Ni–O rehybridization across the whole cycling range, lowering the electron density on O sites, but without a significant role for molecular O₂ redox. This contrasts with several reports of oxygen redox in this potential range for LiNiO₂ and Ni-rich layered cathode materials, based on detection of the molecular O₂ feature with less bulk-sensitive O K-edge RIXS.^24–26,51

Fig. 2 Near-surface probing of LiNiO₂ redox processes. (a) and (b) FY-XAS (∼200 nm information depth) of the O K-edge and Ni L₃-edge, and (c) TEY-XAS (∼10 nm information depth) of the Ni L₃-edge for LiNiO₂ at different SoC. (d) Relative intensities of Ni²⁺, Ni³⁺, and Ni⁴⁺ components based on fitting CTM calculated spectra to XRS, FY-XAS, and TEY-XAS spectra (see fitting results in ESI,† Fig. S2–S5).

Near-surface degradation

To further investigate the origins of molecular O₂ reported at high SoC, the same core levels were measured using soft XAS in FY mode (Fig. 2(a) and (b)). The spectra for pristine LiNiO₂ closely resemble those obtained with XRS, however an additional feature is apparent at 532.3 eV in the O K-edge, and the Ni²⁺ feature in the Ni L₃-edge is more intense. These same features are seen for NiO (see ESI,† Fig. S8 for O K-edge), and correspond to a NiO-like RSL,⁵² whose contribution is not detected in the more bulk-sensitive XRS. On delithiation, the XAS data show similar trends to the XRS until 4.1 V, x = 0.33, with the Ni L₃-edge showing the Ni⁴⁺ feature increasing at the expense of Ni²⁺ and then Ni³⁺, and some growth in the O K pre-edge. At higher SoC there are significant deviations between FY-XAS and XRS spectra. Most notably a feature at ∼531.5 eV is seen to emerge in the O K-edge, which although initially weak at 4.2 V, x = 0.22, shows significant intensity at 4.8 V, x = 0.03 (see integrated peak areas in ESI,† Fig. S9). This feature corresponds to the same absorption energy as molecular O₂, whose vibrational structure has been detected in LiNiO₂ and other conventional Ni-rich layered oxides in several recent reports.^25,26 Whereas this molecular O₂ signature and the Ni²⁺ feature grow in FY-XAS, the O K pre-edge peak and the Ni⁴⁺ feature in the Ni L₃-edge are supressed in FY-XAS compared to the XRS. This suggests a near-surface molecular O₂ redox process associated with RSL growth, in which Ni is reduced toward Ni²⁺ and molecular O₂ forms i.e., NiO₂ → NiO_2−x + ½xO₂. Similar trends are observed with the more surface-sensitive TEY-XAS (fits shown in ESI,† Fig. S4) consistent with RSL formation proceeding from electrolyte-exposed surfaces.^21,52 Importantly, the bulk-sensitive XRS (Fig. 1(e) and (f)) does not detect such Ni reduction or O₂ formation, even after charging to 4.8 V, highlighting the key connection between the formation of trapped molecular O₂ and the increase in Ni²⁺ species close to the cathode surface.

We now investigate the reversibility of this near-surface molecular O₂ redox process and how its extent changes with upper cutoff voltage (UCV). Fig. 3(a) shows that after discharging from a UCV of 4.8 V to 4.0 V, the molecular O₂ feature at ∼531.5 eV disappears from the O K-edge, but a prominent RSL feature at 532.6 eV remains. On discharge to 3.0 V, the RSL feature further grows in intensity relative to the pre-edge feature, with accompanying increases in the Ni²⁺ feature for the Ni L₃-edge spectra (Fig. 3(b) and (c)). This is even more prominent in the surface-sensitive TEY-XAS (Fig. 3(c)), indicating the RSL is more densified near to the surface. Comparison to an electrode where the UCV is 4.2 V confirms that the extent of RSL formation is much greater for the UCV of 4.8 V, consistent with previous studies where significant RSL formation occurs at SoC above the H2-H3 transition in Ni-rich cathodes.^20,21,52 Longer-term cycling (150 cycles, ESI,† Fig. S10) further shows that the UCV of 4.8V leads to greater voltage hysteresis and charge transfer impedance reflecting this more extensive RSL formation.

Fig. 3 Discharge behaviour of LiNiO₂. (a), (b) FY-XAS (∼200 nm information depth) of the O K-edge and Ni L₃-edge, and (c) TEY-XAS (∼10 nm information depth) of the Ni L₃-edge for LiNiO₂ cycled to a UCV of 4.2 V before being discharged to (i) 4.0 V and (ii) 3.0 V, with parallel samples cycled to a higher UCV of 4.8 V and then back to (iii) 4.0 V and (iv) 3.0 V.

Although comparison of TEY and FY mode XAS confirms the RSL is found predominantly near the sample surface, it provides only limited insight into the depth over which it is distributed. To spatially resolve the extent of the RSL at high SoCs, STEM-EELS was performed for LiNiO₂ charged to 4.8 V (Fig. 4). Depth-resolved Ni L₃-spectra show a decreasing proportion of the lower energy (more reduced) component (peak A1) on moving towards the bulk of the particle, stabilising at ∼200 nm from surface, consistent with more Ni²⁺ species at the surface and more Ni⁴⁺ in the bulk. Similarly, the O K-edge shows a higher pre-edge intensity (peak B1) towards the bulk of the particle correlating with higher Ni oxidation state and Ni–O covalency. This extended RSL region where the Ni oxidation state is seen to vary over ∼200 nm is attributable to a cation mixing layer in which Ni²⁺ ions have migrated to occupy Li sites, and is consistent with the differences seen between TEY-XAS, FY-XAS and XRS observations. Notably, a similar extent of RSL formation is not observed at intergranular cracks away from the LiNiO₂ surface, presumably as electrolyte does not fully penetrate these cracks for the low cycle numbers considered here. This indicates a key role of the electrolyte in promoting RSL formation, with electrolyte infiltration into internal cracks likely proceeding over multiple cycles.

Fig. 4 (a) Cross-sectional scanning electron microscopy (secondary electron detection) of LiNiO₂ particle from an electrode charged to 4.8 V. (b) Selected STEM-EELS scan area of 1.5 μm from surface to bulk (left to right) of the particle. (c) and (d) Fitted peak ratios of depth-resolved Ni L₃- and O K-edge EELS spectra using a simplified two peak fit in each case (see ESI,† Fig. S11). Insets: Examples of EELS spectra.

Bulk electronic and geometric structural evolution of LiNiO₂

Having shown that bulk redox in LiNiO₂ occurs by Ni–O rehybridization, we now consider further the associated changes in electronic and geometric structure. Fig. 5(a) shows high-energy-resolution fluorescence detection (HERFD-)XANES Ni K-edge spectra of pristine LiNiO₂ and after cycling to 4.8 V. Notably the main edge half-height position is shifted ∼2.1 eV higher compared to LiNiO₂, a more distinct change than seen in the transmission mode measurements of Fig. 1(c) (∼1.5 eV), as a result of the fine-structure features along the rising edge now being better resolved. Fig. 5(c) compares the similarly bulk-sensitive O K-edge XRS spectra of the same samples. Since we anticipate that differing Ni–O bond lengths yield distinct signatures in the O K pre-edge in any model of the material, we chose the zigzag J–T P2₁/c structure for LiNiO₂ spectral calculations. More complex time-varying LiNiO₂ model structures are computationally prohibitive for spectral calculations and although the P2₁/c structure has a formal oxidation state of Ni³⁺, this will predominantly affect the Ni 3d states, which only weakly influence the main features of the Ni and O K-edge spectra. Density functional theory (DFT) calculated Ni and O K-edge spectra for LiNiO₂ and NiO₂ (Fig. 5(b) and (d)), reproduce the features of the experimental spectra extremely well, showing the same pre-edge peaks, number of fine structure features, and similar trends in intensity and linewidths across the whole spectral range. The relative energy shifts are also captured well, giving confidence in the sufficiency of the chosen structure models (P2₁/c for LiNiO₂, and R3̄m for NiO₂).

Fig. 5 Electronic and geometric structural changes of LiNiO₂ upon delithiation. (a) Experimental HERFD-XANES and b, core–hole calculated Ni K-edge spectra of pristine and charged LiNiO₂. (c) Experimental XRS with smooth trace lines and (d) core–hole calculated O K-edge spectra of pristine and charged LiNiO₂. (e) Ground-state partial and total density of states for LiNiO₂ (top) and NiO₂ (bottom). Fermi energies are set to zero. (f) Ni–Ni and Ni–O distances determined from the Fourier-transformed EXAFS spectra (details in ESI,† Table S1 and Fig. S12, S13). Note that the short/long Ni–O lengths of pristine (hexagons), 3.0 V and 3.8 V (crosses) LiNiO₂ are related to the disproportionated model applied for EXAFS fitting. Bond lengths for the geometry optimised structures from DFT calculations used in (b), (d), (e) are shown as triangles in f.

The origin of the spectral features can be understood by comparison to ground-state partial density-of-states (pDOS) shown in Fig. 5(e), and consideration of the allowed spectroscopic transitions. The first unoccupied states in both LiNiO₂ and NiO₂ lie just above 0 eV, showing mixed O 2p and Ni 3d orbital character and giving rise to the pre-edge peaks in the experimental Ni (∼8335 eV) and O (∼529 eV) K-edges. A sizable gap separates the next set of unoccupied states which give rise to the main edges in the Ni (≳8340 eV) and O (≳535 eV) K-edges, and have Ni 4s,p character, with some Li 2s contribution also seen in this region for LiNiO₂. This gap widens by ∼2.9 eV from LiNiO₂ to NiO₂ which can be related to a decrease in average Ni–O bond length associated with the change in geometric structure.^53,54 We note that the DFT calculated Ni K-edge spectra show weaker pre-edge features than experiment, attributable to quadrupolar transitions not being considered in the calculations.⁵⁵

A clear splitting of the calculated O K pre-edge peak in Fig. 5(d) for LiNiO₂ resembles the asymmetric pre-edge in the XRS experimental data. The O K pre-edge becomes far more intense in the 4.8 V sample and the peak splitting seen in the calculated spectrum of LiNiO₂ is lost. This corresponds with the change from D_4h site symmetry for the J–T distorted Ni³⁺ octahedra used in the LiNiO₂ calculation, where d orbital splitting arises from the elongation of two Ni–O bonds, to O_h site symmetry for the Ni⁴⁺ octahedra of NiO₂, where this d orbital splitting is lost. The growth in intensity of the O K pre-edge feature is also consistent with the CTM calculations, where the increased ligand hole contributions for the Ni⁴⁺ octahedra indicate an increasing degree of Ni–O covalency on delithiation. The increase of O K pre-edge intensity by a factor of ∼2 on full delithiation (see ESI,† Fig. S9) corresponds closely to the factor of ∼1.8 obtained based on the proportions of Ni species fitted to the Ni L_3,2-edge XRS spectra (Fig. 2(d)) and their respective electron configurations. Further evidence for increased Ni–O covalency is apparent from the emergence of more distinct fine-structure features (∼8347 eV and 8351 eV) in the Ni K-edge, attributable to ligand-to-metal charge transfer shakedown transitions,⁵⁶ as well as satellite peaks in the Ni L_3,2-edge that are most clearly seen in FY-XAS measurements (see ESI,† Fig. S14b) and are well-reproduced in the CTM calculated Ni⁴⁺ spectrum. In addition, Bader charge analysis⁵⁷ based on the ground-state DFT calculations shows the ionic charge of the Ni only modestly changes from +1.41 to +1.56 e⁻ between LiNiO₂ and NiO₂, whilst a more significant change from −1.15 to −0.78 e⁻ is seen for the O charges.

Fig. 5(f) shows the nearest Ni–O and Ni–Ni distances extracted by fitting to EXAFS spectra for LiNiO₂ at different SoC. Since fitting with the single Ni–O bond length model showed significantly higher Debye–Waller factors at low SoC, and disproportionation is expected to persist up to 3.9 V based on Fig. 2(d), a model with two Ni–O bond lengths (ratio of short:long Ni–O bond based on the disproportionated model and associated XRS fittings) was instead used to fit pristine, 3.0 V and 3.8 V LiNiO₂ (see ESI,† Fig. S15 and Table S2). The Ni–O and Ni–Ni bond distances obtained show good agreement with both the J–T P2₁/c LiNiO₂ and the disproportionated structure, however, the short:long bond ratios of the disproportioned model show lower Debye–Waller factors, supporting assignment of this structure.

Similar trends in weighted average Ni–O bond lengths are seen to operando neutron diffraction measurements,³⁴ with Ni–O bond length gradually shrinking in line with the change in structure, increased oxidation state and increased covalency at high SoC. Notably, above the H2–H3 transition (x ≤ 0.22) a modest increase in the Ni–O bond length is observed. This has been associated with a loss of the stabilising effect of Li–O covalency at high SoC, leading to Ni–O bond elongation alongside the sudden c-lattice collapse related to the H2–H3 transition, and increased charge transfer from the O to Ni sites.^11,58,59 This changing covalency, seen as continuous spectral changes in Fig. 1(c) and (d), can account for the plateauing in half-height position of the Ni K main-edge at high SoC in transmission XANES (Fig. 1(d)). As well as highlighting the limitations of applying a single metric to assess changes in oxidation state, the limited sensitivity of the Ni K-edge fractional-edge height reflects that it arises from transitions to Ni 4s,p states, in contrast to the O K- and Ni L_3,2-edges which probe transitions to O 2p-Ni 3d hybridised states.

Conclusion

In summary, bulk sensitive XRS measurements reveal that in the bulk of LiNiO₂, charge compensation occurs by Ni–O rehybridization without the involvement of molecular O₂. From an initially disproportionated structure where formally Ni²⁺, Ni³⁺, and Ni⁴⁺ octahedra coexist, the Ni⁴⁺ features of the Ni L_3,2-edge continuously grow on delithiation, initially at the expense of Ni²⁺ and subsequently Ni³⁺ features (see Fig. 6). There is a concomitant increase in O K pre-edge intensity, consistent with significant lowering of electron density on the O sites at potentials where O loss is expected.^9,50 However, significant signatures of molecular O₂ formation are not detected throughout the bulk suggesting its formation remains kinetically hindered.⁶⁰

Fig. 6 Schematic representation of LiNiO₂ bulk charge compensation mechanism and surface degradation processes probed by different X-ray spectroscopy techniques. The differences in LiNiO₂ delithiation occurring at the surface and in the bulk, including the accommodation of trapped O₂ in pores formed near the surface by Ni²⁺ species migrating to the Li layer. Continuous oxidation of Ni in bulk LiNiO₂ at high SoC is distinguished from the near surface degradation.

FY-XAS measurements reveal evidence of molecular O₂ formation in the outer ∼200 nm of the cathode surface, with the growth in intensity of the Ni⁴⁺ feature plateauing above 4.2 V, and features of trapped molecular O₂ emerging alongside increased Ni²⁺ contributions. This is consistent with molecular O₂ trapped in voids formed by Ni²⁺ entering the Li layers (see Fig. 6). STEM-EELS reveals a RSL that extends ∼200 nm into the LiNiO₂ surface following cycling to 4.8 V, showing a gradual change in oxidation state across its thickness. The absence of this extended RSL at internal surfaces of the secondary cathode particles, e.g. interparticle cracks, suggests its formation is driven by contact with the electrolyte.

The trapped molecular O₂ feature disappears on discharging to 4.0 V, but a significantly increased near-surface Ni²⁺ contribution is retained. Although our results do not fully exclude some reversible molecular O₂ redox, online electrochemical mass spectroscopy studies have reported O₂ evolution occurring on discharge.¹⁵ We therefore suggest that this may arise from the release of trapped O₂ associated with structural changes, including abrupt c-lattice expansion and particle cracking (ESI,† Fig. S16).

Our findings highlight the importance of combining bulk- and surface-sensitive techniques to fully confirm the extent to which molecular O₂ redox processes in cathode materials are bulk phenomena contributing to reversible charge compensation, rather than involved in surface degradation as revealed here for LiNiO₂. The understanding developed of the surface instability of LiNiO₂ associated with rehybridisation at high SoC, emphasises the importance of strategies such as cathode coatings, composition gradients, and electrolyte formulation to stabilise Ni-rich cathode surfaces in contact with electrolyte, rather than bulk stabilisation approaches (e.g. pillaring) that might unduly sacrifice capacity. This study thus provides a solid basis for future exploration of molecular O₂ formation and Ni–O rehybridisation in Ni-rich cathodes in different electrolyte environments, and for further investigations to separate bulk redox and near-surface degradation processes in a broad range of cathode materials.

Experimental

Sample preparation

Commercial grade LiNiO₂ powder was obtained from BASF, without any deliberate doping or coating added. This was characterised by SEM (Zeiss Merlin, 2 kV, Inlens detector), XRD (Rigaku Miniflex, Cu Kα source), and X-ray Photoelectron Spectroscopy (XPS, PHI Versaprobe III, Al Kα source), see ESI,† Fig. S1. Free-standing electrodes were prepared by calendaring a mixture of 80 wt% LiNiO₂ powder, 10 wt% conductive acetylene black and 10 wt% polytetrafluoroethylene (PTFE) binder. Electrochemical tests of the LiNiO₂ cathodes were performed in 2032 coin cells (316 stainless steel, Cambridge Energy Solutions) using Li metal disks as negative electrodes and borosilicate glass fibre separators (borosilicate, GF/A, Whatman) soaked with 120 μL of LP57 electrolyte (1 M LiPF6 in 3 : 7 of EC : EMC). The assembled cells were charged up to 4.2 V at C/20 (calculated based on a theoretical capacity of LiNiO₂ of 275 mA h g⁻¹), held for 30 min and then cycled back to 3.0 V at the same rate. This was immediately followed by a second charge at C/20 (ESI,† Fig. S1), with a voltage hold for 10 hours at the desired endpoint. The SoC of delithiated Li_xNiO₂ is calculated based on the charge–discharge capacity curve presented in Fig. 1(a), starting from x = 0.96 for pristine and 3.0 V LiNiO₂ (based on the ∼4% Ni excess). All potentials mentioned in this work are referenced to Li/Li⁺. For ex situ measurements, cathodes were recovered from the cycled coin cells by disassembly in a glove box under Ar atmosphere (O₂ < 1 ppm, H₂O < 1 ppm). Recovered cathodes were washed in dimethyl carbonate (anhydrous, ≥99%, Sigma Aldrich) solvent and dried before heat-sealing in aluminised mylar (Fresherpack, 130 μm) pouches for XANES, EXAFS, and XRS, or transferring in a vacuum suitcase for XAS, XPS.

Electron microscopy

A Thermo Scientific Helios G4 CXe Plasma FIB (PFIB) was used to prepare the STEM lamellae. For lift-out, a thin Pt layer was deposited onto the surface region of interest (ROI), trenches were patterned around the ROI to make 4 μm-thick lamellae. A W needle was then used to lift each lamella and place them on a FIB lift-out grid (Cu, Agar Scientific). Each lamellae was then thinned down to around 50 nm thick for STEM-EELS and polished with a low dose, low energy beam (<0.3 nA, 5 keV) to minimise ion beam damage. Inert transfer between PFIB and an Ar glovebox was achieved using a Gatan iLoad system.

Spatially resolved EELS of the lamellae was performed using a JEOL ARM200F equipped with cold field emission gun operated at 200 keV and spherical aberration probe corrector. Dual EELS was acquired using a Gatan GIF Quantum 965 ER with energy resolution of around 1 eV at 0.25 eV per channel dispersion. Inert transfer between glovebox and STEM was achieved using a JEOL double-tilt vacuum transfer holder.

Given LiNiO₂ is less stable when highly delithiated, radiation damage should be considered when evaluating the oxidation state of Ni in EELS. The energised Xe ion beam in PFIB and electron beam in STEM can both induce reduction of LiNiO₂ and NiO₂ towards NiO.⁶¹ The absolute A2/A1, and B2/B1 ratios seen in the LiNiO₂ bulk reflect some degree of ion/electron beam induced reduction. Nevertheless, equal acquisition time and constant electron beam current during the EELS scans ensure a consistent radiation dose (3 × 10³ e⁻ Å⁻²) such that the trends in EELS spectra and the spatial variations seen near their surface are still representative.

X-ray spectroscopy

Transmission Ni K-edge XANES and EXAFS spectra were collected with a laboratory-based easyXAFS300+ spectrometer (easyXAFS, WA, US). X-rays are generated with a liquid-cooled Ag anode X-ray tube, before monochromation by a Si (551) spherically bent crystal analyser. A helium-filled box with polyimide windows is placed in the beam path for better X-ray transparency while a steel plate with a 9 × 3 mm slot is placed after each sample to lower the background. The transmitted intensity is measured with a silicon drift detector (KETEK, Munich, Germany) placed behind the sample. Each acquisition was performed over 45 min and 30 scans were collected for each sample to obtain good statistics. NiO reference spectra were also collected for each batch of measurements for energy calibration. Data pre-processing was performed with the EasyXANES package to convert the measured intensity into linear attenuation coefficient, μ. Data reduction and analysis were performed using the Demeter package (version: 0.9.26).

In situ HERFD-XANES was performed at Diamond Light Source's beamline I20 with aluminised mylar (Fresherpack, 130 μm) pouch cells containing free-standing LiNiO₂ cathodes, with Li metal disks as negative electrodes and a Celgard 2325 separator soaked with 80 μl of LP57 electrolyte (1 M LiPF6 in 3 : 7 of EC : EMC). The cells were held at the desired potentials, and measured through the cathode side of the pouches using an X-ray emission spectrometer equipped with three Si(444) analyser crystals.⁶² The spectrometer was set to the maximum of the Ni Kβ_1,3 line (8266 eV), and the incident energy was scanned using the four-bounce Si(111) monochromator. The spectrometer was calibrated using a Ni foil, measuring the Kβ line with the incident energy tuned +300 eV from the Ni K-edge.

TEY- and FY-XAS measurements were performed at ES-2 of beamline B07-B at Diamond Light Source, with the exit slits set to 50 μm in the dispersive direction, yielding a flux of between 1 × 10¹¹ (O K-edge) and 2 × 10¹¹ (Ni L_3,2-edge) photons s⁻¹. All samples were measured with the incident beam normal to the electrode surface, yielding a beam footprint of 150 × 100 μm. FY-XAS measurements were acquired using an Al coated Si photodiode directed at the sample with its surface normal at ∼45° to incident beam direction. Simultaneous TEY-XAS measurements were obtained using a SR570 low-noise current amplifier (Stanford Research Systems) to collect the current between the sample plate and an isolated steel washer in front of the sample biased to +90 V. Separate IPFY-XAS measurements of the Ni L_3,2-edge were acquired using a Vortex silicon drift detector (Hitachi) at the I10 beamline at Diamond Light Source, with FY and TEY mode measurements simultaneously acquired. All spectra are divided by the drain current measured from the last X-ray mirror, to correct for variations in incident photon flux. The photon energy scale is calibrated using a NiO sample.⁶³ O K-edge spectra are background-subtracted using a straight line fitted to the pre-edge region, followed by intensity normalization to the post-edge region at 550 eV. Ni L_3,2-edge spectra are normalized to the intensity at 867 eV after removal of a linear background.

XRS measurements were performed at the European Synchrotron Radiation Facility at the ID20 beamline.⁶⁴ X-rays are generated from three U26 revolver undulators, before being collimated, and then monochromated by a liquid–nitrogen cooled double-crystal Si(111) pre-monochromator. The beam from a second Si(311) channel-cut post-monochromator is focussed onto a ∼20 × 20 μm² spot at the sample position by a mirror in Kirkpatrick–Baez geometry. The sample surface was positioned at a grazing angle of ∼1° relative to the incident beam direction, to maximise the illuminated area and the sample was scanned over a region of ∼10 mm during the 4-hour measurement to minimise beam-induced changes. Inelastically scattered photons were recorded using 72 spherically bent Si(660) crystal analysers with energy loss events in the vicinity of both O K-edge and Ni L_3,2-edge. O K- and Ni L_3,2-edges were recorded at momentum transfers of q = 6.9 ± 0.5, and all data extraction and treatment were performed as described in ref. 65.

Charge-transfer multiplet calculations

Ni L_3,2-edge multiplet simulations were performed at the ligand field level of theory using the many-body code, Quanty.⁶⁶ This was implemented using the same single-cluster NiO₆ Hamiltonian as Green et al.,⁴⁷ where Ni 2p, Ni 3d ligand shells are explicitly included (see Supplementary Note S3, ESI†). Parameters used in Ni²⁺ calculation (eV): Δ = 5.5, 10D_q = 0.71, V_eg = 2.627, V_t2g = 1.524. Parameters used in Ni³⁺ calculation (eV): Δ = −0.5, 10D_q = 0.93 with Jahn–Teller splitting of Δ_eg = 0.15 and Δ_t2g = 0.10 where Δ_eg is the difference between the x² − y² and 3x² − r² onsite energies and Δ_t2g is the difference between the xy and xz/yz onsite energies, V_{3x² − r²} = 2.43, V_{x² − y²} = 3.33, V_xz/yz = 1.41, V_xy = 1.93. Parameters used in Ni⁴⁺ calculation (eV): Δ = −6.5, 10D_q = 0.78, V_eg = 3.456, V_t2g = 2.004.

DFT spectral calculations

Density functional theory (DFT) calculations were carried out using the plane wave pseudopotential code CASTEP⁶⁷ and the Perdew–Burke–Ernzerhof (PBE) form of the generalized gradient approximation functional,⁶⁸ with the addition of the G06 semi-empirical dispersion correction⁶⁹ to better account for van der Waals forces. The zig-zag J–T P2₁/c structure for LiNiO₂, and the R3̄m structure for NiO₂ were used for pristine LiNiO₂ and fully delithiated materials respectively.³³ Each structure was initially geometry optimised using appropriate plane wave cut-off energies (900 eV) and k-points (0.03 Å⁻¹k-point spacing) determined via convergence of the total energy. The geometry of the system was considered optimized when the maximum forces on the ions were below 0.001 eV Å⁻¹ for NiO₂ and 0.01 eV Å⁻¹ for LiNiO₂ consistent with other studies.^33,70 Calculations of the pDOS and core–hole spectra were subsequently performed. The energy scale of the ground-state pDOS assumes the material is an insulator and sets the Fermi energy, E_f, to zero. Since core orbitals are not treated explicitly in the pseudopotential method, a unique pseudopotential is generated for an excited atom possessing a core–hole. For O and Ni K-edges, a core–hole is placed on the O 1s or Ni 1s orbitals respectively. A supercell is generated to prevent interactions between neighbouring core–holes. For spectral calculations, the plane wave energy cut-off, k-point sampling and cell size were increased until no visible effect on the spectrum was seen. Spectral calculations were handled using the OptaDOS programme.⁷¹ Lorentzian broadening was performed using full widths at half maximum of 0.14 and 0.8 eV for the O and Ni K-edges respectively, which should reflect the lifetimes of radiative and non-radiative transitions.^72,73 The Gaussian component was then adjusted as a free parameter to match the experimental data, but remained fixed for the same edges to allow for comparison. The Lorentzian component is given energy dependence to account for the energy dependence of the lifetime. This was done by summing the set width with a factor that varies linearly with energy as implemented in Optados. The calculated spectra were rigidly shifted to align with the first absorption peaks of the experimental data to allow better comparison. In cases where the system under investigation possessed more than one inequivalent excitation site, separate spectra were generated, energy aligned⁷⁴ and combined before rigidly shifting.

Author contributions

LA, JENS and RSW conceived the study. LA performed electrode preparation and electrochemical testing, with assistance from RAH. MWF performed SEM. LA, JENS, CJS and RSW performed the XRS. LA, AP, EB, PND, MWF, LAHJ, CMEP and RSW performed the soft XAS with support from DCG, PF and PB. LA, JENS, AP, CJS, and RSW performed XAS and XRS analysis. LA and PC performed the lab-based XANES and EXAFS, with analysis performed by PC. LA, JENS, EB, PND and RSW performed HERFD-XANES with support from SH. JENS, NR and RJN performed DFT calculations. RZ and PDN performed PFIB, STEM-EELS and related analysis. GH and RJG performed CTM calculations. LA, JENS, PC and RSW wrote the paper with contributions from all authors.

Data availibility

The data supporting this article have been included as part of the ESI.† The corresponding data sets are available from the ORA repository, https://doi.org/10.5287/ora-yxpnqgero.

Conflicts of interest

The authors declare that there are no conflicts of interest.

Acknowledgements

The authors acknowledge funding from the Faraday Institution (Faraday.ac.uk; EP/S003053/1, FIRG001, FIRG007, FIRG008, FIRG016, FIRG024) and the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (EXISTAR, grant agreement No. 950598) and under the Marie Sklodowska-Curie Actions (ISOBEL, grant agreement No. 101032281). We acknowledge support from the Engineering and Physical Science Research Council (EPSRC) through grants EP/K040375/1, EP/L022907/1, EP/T001038/1, and EP/R010145/1 (Henry Royce Institute). R. J. G. acknowledges funding from the Natural Sciences and Engineering Research Council of Canada (NSERC). R. A. H. acknowledges funding from the Royal Academy of Engineering under the Research Fellowship scheme. R. S. W. acknowledges a CAMS-UK Fellowship through the Analytical Chemistry Trust Fund and a UKRI Future Leaders Fellowship (MR/V024558/1). We acknowledge access to the David Cockayne Centre for Electron Microscopy. We thank Diamond Light Source (DLS) for beamtime on beamlines B07B, I10, and I20 under proposals SI33283, MM33062, and SP32010. We acknowledge the European Synchrotron Radiation Facility (ESRF) for provision of synchrotron radiation facilities under proposal MA-5753 and thank Blanka Detlefs for assistance and support in using beamline ID20.

References

1. E. Wikner, E. Björklund, J. Fridner, D. Brandell and T. Thiringer, How the utilised SOC window in commercial Li-ion pouch cells influence battery ageing, J. Power Sources Adv., 2021, 8, 100054.
2. J. E. Harlow, et al., A Wide Range of Testing Results on an Excellent Lithium-Ion Cell Chemistry to be used as Benchmarks for New Battery Technologies, J. Electrochem. Soc., 2019, 166, A3031–A3044.
3. “This Is What We Die For”: Human Rights Abuses in the Democratic Republic of the Congo Power the Global Trade in Cobalt, Amnesty International, 2016, Index: AFR 62/3183/2016.
4. P. Kuiper, G. Kruizinga, J. Ghijsen, G. A. Sawatzky and H. Verweij, Character of Holes in Li_xNi_1−xO and Their Magnetic Behavior, Phys. Rev. Lett., 1989, 62, 1214.
5. T. Mizokawa, et al., Role of oxygen holes in Li_xCoO₂ revealed by soft X-ray spectroscopy, Phys. Rev. Lett., 2013, 111, 1–5.
6. J. M. Tarascon, et al., In Situ Structural and Electrochemical Study of Ni_1−xCo_xO₂ Metastable Oxides Prepared by Soft Chemistry, J. Solid State Chem., 1999, 147, 410–420.
7. K. Kleiner, et al., On the Origin of Reversible and Irreversible Reactions in LiNi_xCo(1 − x)/2 Mn(1 − x)/2O₂, J. Electrochem. Soc., 2021, 168, 120533.
8. Z. Wu, et al., Unveiling the Evolution of LiCoO₂ beyond 4.6 V, ACS Energy Lett., 2023, 8, 4806–4817.
9. A. R. Genreith-Schriever, et al., Oxygen hole formation controls stability in LiNiO₂ cathodes, Joule, 2023, 7, 1623–1640.
10. H. Li, N. Zhang, J. Li and J. R. Dahn, Updating the Structure and Electrochemistry of Li_xNiO₂ for 0 ≤ x ≤ 1, J. Electrochem. Soc., 2018, 165, A2985–A2993.
11. S. Lee, L. Su, A. Mesnier, Z. Cui and A. Manthiram, Cracking vs. surface reactivity in high-nickel cathodes for lithium-ion batteries, Joule, 2023, 7, 2430–2444.
12. S. Ahmed, et al., Visualization of Light Elements using 4D STEM: The Layered-to-Rock Salt Phase Transition in LiNiO₂ Cathode Material, Adv. Energy Mater., 2020, 10, 2001026.
13. C. S. Yoon, D. W. Jun, S. T. Myung and Y. K. Sun, Structural Stability of LiNiO₂ Cycled above 4.2 v, ACS Energy Lett., 2017, 2, 1150–1155.
14. J. K. Papp, et al., A comparison of high voltage outgassing of LiCoO₂, LiNiO₂, and Li₂MnO₃ layered Li-ion cathode materials, Electrochim. Acta, 2021, 368, 137505.
15. L. de Biasi, et al., Phase Transformation Behavior and Stability of LiNiO₂ Cathode Material for Li-Ion Batteries Obtained from In Situ Gas Analysis and Operando X-Ray Diffraction, ChemSusChem, 2019, 12, 2240–2250.
16. D. Aurbach, et al., The Study of Surface Phenomena Related to Electrochemical Lithium Intercalation into Li_xMO_y Host Materials (M = Ni, Mn), J. Electrochem. Soc., 2000, 147, 1322.
17. R. Pan, E. Jo, Z. Cui and A. Manthiram, Degradation Pathways of Cobalt-Free LiNiO₂ Cathode in Lithium Batteries, Adv. Funct. Mater., 2023, 33, 1–11.
18. F. Lin, et al., Surface reconstruction and chemical evolution of stoichiometric layered cathode materials for lithium-ion batteries, Nat. Commun., 2014, 5, 3529.
19. R. Jung, M. Metzger, F. Maglia, C. Stinner and H. A. Gasteiger, Oxygen Release and Its Effect on the Cycling Stability of LiNi_xMn_yCo_zO₂ (NMC) Cathode Materials for Li-Ion Batteries, J. Electrochem. Soc., 2017, 164, A1361–A1377.
20. D. Streich, et al., Operando Monitoring of Early Ni-mediated Surface Reconstruction in Layered Lithiated Ni–Co–Mn Oxides, J. Phys. Chem. C, 2017, 121, 13481–13486.
21. W. M. Dose, et al., Electrolyte Reactivity at the Charged Ni-Rich Cathode Interface and Degradation in Li-Ion Batteries, ACS Appl. Mater. Interfaces, 2022, 14, 13206–13222.
22. C. M. E. Phelan, et al., Role of Salt Concentration in Stabilizing Charged Ni-Rich Cathode Interfaces in Li-Ion Batteries, Chem. Mater., 2024, 36, 3334–3344.
23. F. Kong, et al., Kinetic Stability of Bulk LiNiO₂ and Surface Degradation by Oxygen Evolution in LiNiO₂-Based Cathode Materials, Adv. Energy Mater., 2019, 9, 1–12.
24. N. Li, et al., Unraveling the Cationic and Anionic Redox Reactions in a Conventional Layered Oxide Cathode, ACS Energy Lett., 2019, 4, 2836–2842.
25. A. S. Menon, et al., Oxygen-Redox Activity in Non-Lithium-Excess Tungsten-Doped LiNiO₂ Cathode, PRX Energy, 2023, 2, 1.
26. M. Juelsholt, et al., Does trapped O₂ form in the bulk of LiNiO₂ during charging?, Energy Environ. Sci., 2024, 17, 2530.
27. M. Zhang, et al., Pushing the limit of 3d transition metal-based layered oxides that use both cation and anion redox for energy storage, Nat. Rev. Mater., 2022, 7, 522–540.
28. R. A. House, et al., The role of O₂ in O-redox cathodes for Li-ion batteries, Nat. Energy, 2021, 6, 781–789,  DOI:10.1038/s41560-021-00780-2 Preprint at.
29. N. Li, et al., Unraveling the Cationic and Anionic Redox Reactions in a Conventional Layered Oxide Cathode, ACS Energy Lett., 2019, 4, 2836–2842.
30. Z. W. Lebens-Higgins, et al., How Bulk Sensitive is Hard X-ray Photoelectron Spectroscopy: Accounting for the Cathode-Electrolyte Interface when Addressing Oxygen Redox, J. Phys. Chem. Lett., 2020, 11, 2106–2112.
31. R. A. House, et al., First-cycle voltage hysteresis in Li-rich 3d cathodes associated with molecular O₂ trapped in the bulk, Nat. Energy, 2020, 5, 777–785.
32. R. Jung, M. Metzger, F. Maglia, C. Stinner and H. A. Gasteiger, Chemical versus electrochemical electrolyte oxidation on NMC111, NMC622, NMC811, LNMO, and conductive carbon, J. Phys. Chem. Lett., 2017, 8, 4820–4825.
33. M. Mock, M. Bianchini, F. Fauth, K. Albe and S. Sicolo, Atomistic understanding of the LiNiO₂–NiO₂ phase diagram from experimentally guided lattice models, J. Mater. Chem. A, 2021, 9, 14928–14940.
34. P. H. Chien, et al., New Insights into Structural Evolution of LiNiO₂ Revealed by Operando Neutron Diffraction, Batteries Supercaps, 2021, 4, 1701–1707.
35. R. J. Woolley, B. N. Illy, M. P. Ryan and S. J. Skinner, In situ determination of the nickel oxidation state in La₂NiO_4+δ and La₄Ni₃O_10−δ using X-ray absorption near-edge structure, J Mater Chem, 2011, 21, 18592–18596.
36. W. E. O’Grady, K. I. Pandya, K. E. Swider and D. A. Corrigan, In Situ X-Ray Absorption Near-Edge Structure Evidence for Quadrivalent Nickel in Nickel Battery Electrodes, J. Electrochem. Soc., 1996, 143, 1613–1617.
37. M. L. Baker, et al., K- and L-edge X-ray absorption spectroscopy (XAS) and resonant inelastic X-ray scattering (RIXS) determination of differential orbital covalency (DOC) of transition metal sites, Coord. Chem. Rev., 2017, 345, 182–208.
38. A. Rougier, C. Delmas and A. V. Chadwick, Non-cooperative Jahn-Teller effect in LiNiO₂: An EXAFS study, Solid State Commun., 1995, 94, 123–127.
39. J. H. Chung, et al., Local structure of LiNiO₂ studied by neutron diffraction, Phys. Rev. B: Condens. Matter Mater. Phys., 2005, 71, 8–11.
40. S. Sicolo, M. Mock, M. Bianchini and K. Albe, And Yet It Moves: LiNiO₂, a Dynamic Jahn-Teller System, Chem. Mater., 2020, 32, 10096–10103.
41. A. D. Poletayev, R. J. Green, J. E. N. Swallow, L. An, L. Jones, G. Harris, P. Bencok, R. Sutarto, J. P. Cottom, B. J. Morgan, R. A. House, R. S. Weatherup and M. S. Islam, Temperature-Dependent Dynamic Disproportionation in LiNiO₂, arXiv, 2024, preprint, arXiv:2211.09047 DOI:10.48550/arXiv.2211.09047.
42. K. Foyevtsova, I. Elfimov, J. Rottler and G. A. Sawatzky, LiNiO₂ as a high-entropy charge- and bond-disproportionated glass, Phys. Rev. B, 2019, 100, 1–7.
43. T. Mizokawa, D. Khomskii and G. Sawatzky, Spin and charge ordering in self-doped Mott insulators, Phys. Rev. B: Condens. Matter Mater. Phys., 2000, 61, 11263–11266.
44. V. Bisogni, et al., Ground-state oxygen holes and the metal-insulator transition in the negative charge-transfer rare-earth nickelates, Nat. Commun., 2016, 7, 1–8.
45. S. Johnston, A. Mukherjee, I. Elfimov, M. Berciu and G. A. Sawatzky, Charge disproportionation without charge transfer in the rare-earth-element nickelates as a possible mechanism for the metal-insulator transition, Phys. Rev. Lett., 2014, 112, 1–5.
46. E. Wawrzyńska, et al., Charge disproportionation and collinear magnetic order in the frustrated triangular antiferromagnet AgNiO₂, Phys. Rev. B: Condens. Matter Mater. Phys., 2008, 77, 094439.
47. R. J. Green, M. W. Haverkort and G. A. Sawatzky, Bond disproportionation and dynamical charge fluctuations in the perovskite rare-earth nickelates, Phys. Rev. B, 2016, 94, 1–5.
48. J. Zheng, et al., Ni/Li Disordering in Layered Transition Metal Oxide: Electrochemical Impact, Origin, and Control, Acc. Chem. Res., 2019, 52, 2201–2209.
49. L. Jin, et al., Hidden Hydroxides in KOH-Grown BaNiO₃ Crystals: A Potential Link to Their Catalytic Behavior, Chem. Mater., 2023, 35, 9434–9443.
50. H. Huang, et al., Unusual double ligand holes as catalytic active sites in LiNiO₂, Nat. Commun., 2023, 14, 1–14.
51. Z. W. Lebens-Higgins, et al., Revisiting the charge compensation mechanisms in LiNi_0.8Co_0.2-: YAlyO₂ systems, Mater. Horiz., 2019, 6, 2112–2123.
52. E. Björklund, et al., Cycle-Induced Interfacial Degradation and Transition-Metal Cross-Over in LiNi_0.8Mn_0.1Co_0.1O₂-Graphite Cells, Chem. Mater., 2022, 34, 2034–2048.
53. A. H. De Vries, L. Hozoi and R. Broer, Origin of the chemical shift in X-ray absorption near-edge spectroscopy at the Mn K-edge in manganese oxide compounds, Int. J. Quantum Chem., 2002, 91, 57–61.
54. A. Miglio, C. P. Heinrich, W. Tremel, G. Hautier and W. G. Zeier, Local Bonding Influence on the Band Edge and Band Gap Formation in Quaternary Chalcopyrites, Adv. Sci., 2017, 4, 1700080.
55. F. M. F. De Groot, et al., 1s2p Resonant inelastic X-ray scattering of iron oxides, J. Phys. Chem. B, 2005, 109, 20751–20762.
56. S. DeBeer, et al., X-ray absorption edge and EXAFS studies of the blue copper site in stellacyanin: Effects of axial amide coordination, J. Phys. Chem. B, 2000, 104, 10814–10819.
57. R. F. W. Bader, Atoms in Molecules, Acc. Chem. Res., 1985, 18, 9–15.
58. C. Delmas, et al., On the behavior of the Li_xNiO₂ system: An electrochemical and structural overview, J. Power Sources, 1997, 68, 120–125.
59. A. O. Kondrakov, et al., Charge-transfer-induced lattice collapse in Ni-rich NCM cathode materials during delithiation, J. Phys. Chem. C, 2017, 121, 24381–24388.
60. F. Kong, et al., Kinetic Stability of Bulk LiNiO₂ and Surface Degradation by Oxygen Evolution in LiNiO₂-Based Cathode Materials, Adv. Energy Mater., 2019, 9, 1–12.
61. Y. Koyama, T. Mizoguchi, H. Ikeno and I. Tanaka, Electronic structure of lithium nickel oxides by electron energy loss spectroscopy, J. Phys. Chem. B, 2005, 109, 10749–10755.
62. S. Diaz-Moreno, et al., The Spectroscopy Village at Diamond Light Source, J. Synchrotron Radiat., 2018, 25, 998–1009.
63. G. Van Der Laan, J. Zaanen, G. A. Sawatzky, R. Karnatak and J. M. Esteva, Comparison of x-ray absorption with x-ray photoemission of nickel dihalides and NiO, Phys. Rev. B: Condens. Matter Mater. Phys., 1986, 33, 4253–4263.
64. S. Huotari, et al., A large-solid-angle X-ray Raman scattering spectrometer at ID20 of the European Synchrotron Radiation Facility, J. Synchrotron Radiat., 2017, 24, 521–530.
65. C. J. Sahle, et al., Planning, performing and analyzing X-ray Raman scattering experiments, J. Synchrotron Radiat., 2015, 22, 400–409.
66. M. W. Haverkort, Quanty for core level spectroscopy - Excitons, resonances and band excitations in time and frequency domain, J. Phys.: Conf. Ser., 2016, 712, 012001.
67. S. J. Clark, et al., First principles methods using CASTEP, Z. Kristallogr., 2005, 220, 567–570.
68. J. P. Perdew, K. Burke and M. Ernzerhof, Generalized gradient approximation made simple, Phys. Rev. Lett., 1996, 77, 3865–3868.
69. S. Grimme, Semiempirical GGA-type density functional constructed with a long-range dispersion correction, J. Comput. Chem., 2006, 27, 1787–1799.
70. H. Chen, C. L. Freeman and J. H. Harding, Charge disproportionation and Jahn–Teller distortion in LiNiO₂ and NaNiO₂: A density functional theory study, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 84, 85108.
71. A. J. Morris, R. J. Nicholls, C. J. Pickard and J. R. Yates, OptaDOS: A tool for obtaining density of states, core-level and optical spectra from electronic structure codes, Comput. Phys. Commun., 2014, 185, 1477–1485.
72. S. T. Perkins, et al., Tables and Graphs of Atomic Subshell and Relaxation Data Derived from the LLNL Evaluated Atomic Data Library (EADL), Z = 1–100. UCRL-50400-30, 1991.
73. J. L. Campbell and T. Papp, Widths of the atomic K-N7 levels, At. Data Nucl. Data Tables, 2001, 77, 1–56.
74. D. A. Eustace, et al., First-principles calculation of spectral features, chemical shift and absolute threshold of ELNES and XANES using a plane wave pseudopotential method, J. Phys.: Condens. Matter, 2009, 21, 6.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ee02398f

---