Raman spectroscopy complemented with XRD and TEM for studying structural evolution in initial cycles of LiNi_1/3Mn_1/3Co_1/3O₂ cathode material

Abstract

In this study, we present a detailed investigation of the structural evolution of NMC111 (LiNi_1/3Mn_1/3Co_1/3O₂) cathode material during cycling over an extended potential window, using a combination of in situ and ex situ Raman spectroscopy, ex situ X-ray diffraction (XRD), and high-resolution transmission electron microscopy (HR-TEM). The in situ Raman spectroscopy enabled real-time monitoring of structural changes under operating conditions, while ex situ Raman provided a more detailed post-mortem analysis. We revealed an energy-dependent Raman response that offered further insights into the electronic band structure and phase transitions within the material. We also observed a significant surface layer reconstruction at high potentials, where the NMC111 layered structure transitions into a cubic phase. This surface layer transformation is reversible during the initial stages of cycling but contributes to irreversible degradation with extended cycling, especially at elevated voltages. Ex situ XRD was employed to study the bulk structural evolution of NMC111, confirming that conventional XRD effectively captures large-scale structural changes during cycling, including significant volume variations. Additionally, TEM imaging revealed stress fringes in relithiated grains, indicating cycling-induced stress accumulation from unit cell changes and revealing particle cracking mechanism due to repeated volume fluctuations. This complete analysis provided a complementary view of both surface and bulk modifications, illustrating the advantages of integrating multiple characterization techniques. It offers valuable insights into the mechanisms behind the capacity fading, cracking of the material, and performance loss observed in lithium-ion batteries with NMC cathodes. This comprehensive understanding is essential for improving the design and performance of such batteries, particularly for high-voltage operation.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s11671-025-04319-2.

Introduction

In situ, operando and ex situ techniques are used to study the structural stability, dynamic properties, chemical changes, and morphological evolution of materials and devices. In situ methods involve measurements taken while the sample is subjected to changing conditions, allowing for real-time observation of processes as they occur, often under conditions that approximate normal operation. Operando methods are a specific type of in situ study where measurements are conducted directly on a functioning device, providing insights into its behavior under actual working conditions. In contrast, ex situ methods involve analyzing a sample that has been extracted from the system or a device after being paused at a specific state, enabling detailed characterization without ongoing external influences. All of these methods are increasingly used to study materials and processes in a wide range of scientific fields, and their application has become particularly popular in the energy storage field in recent years [1–6].

The majority of structural and morphological studies of Li-ion battery materials focus on powders measured directly after synthesis. To understand the structural changes occurring within the grains of a given electrode material during oxidation/reduction processes, one can study cells during operation using in situ/operando techniques or under partial-lithiation conditions using ex situ techniques [1, 7]. These studies offer a deeper understanding of the relationships between the crystallographic structure of the analyzed compound, the mechanisms of oxidation/reduction reactions, and its chemical and physical properties. Improving the electrochemical performance of lithium-ion batteries requires investigating mechanisms underlying irreversible structural phase transitions and morphological changes, such as those induced by mechanical stresses and grain cracking, cation site mixing in the crystal structure, oxygen release from the lattice, or electrolyte decomposition [8].

It is important to note that in situ and operando spectroscopic and diffraction measurements, such as Raman and XRD, are typically conducted using specialized electrochemical cells with transparent windows to allow probe access. While these cells are developed to mimic the electrochemical environment of commercial Li-ion batteries, the differences in cell architecture, specific electrolyte volume or composition, and the presence of inert windows can somewhat influence the electrochemical performance or reaction kinetics compared to standard commercial cells. However, there is the general agreement in the field that the fundamental structural and vibrational changes, phase transformations, and mechanistic insights revealed by these in situ/operando methods are representative and critically important for understanding real battery behavior. The insights gained from these techniques have been essential in understanding complex processes, such as phase transitions in electrode materials, SEI and/or CEI formations, and electrolyte decomposition, which are directly relevant to optimizing the performance and longevity of practical Li-ion batteries. Further details on the relevance and limitations of in situ/operando techniques in relation to commercial battery performance can be found in recent reviews such as [4] or [9] and [10].

Ex situ measurements are a foundational approach to understanding processes occurring during the operation of electrochemical cells and often guide more specialized experiments, they come with significant limitations [11]. A major drawback is the loss of dynamic information due to the thermodynamic relaxation of oxides during sample preparation. Additionally, there is a risk of contamination and surface modification when transferring samples between analytical techniques, necessitating meticulous sample preparation. The control over the state of charge during each measurement is limited, especially when stopping the cycle at different potentials, which is critical during high-voltage charging and deep discharging of cells. Using different electrodes for each measurement requires highly controlled electrode preparation to maintain consistent composition and electrochemical properties. Furthermore, because lithium-ion cell components are highly sensitive to air and moisture, ex situ results may not fully represent the processes occurring inside the cell [7].

Obtaining information about a cell under actual operating conditions is crucial for advancing lithium-ion battery technology. In situ, and especially operando, measurements can provide valuable insights into electrode materials without disassembling test cells, aiding in the study of correlations between structural and electrochemical properties. The main advantages of operando methods include: (I) monitoring processes under real conditions (directly in the device for operando studies), (II) the ability to track full charge/discharge cycles with high time resolution, (III) monitoring cells at different stages of formation and throughout their life cycles, (IV) detecting even the smallest structural changes and pinpointing the exact potentials at which they occur relative to lithium or cell voltage (particularly at low current densities), (V) tracking side reactions, and (VI) assessing the qualitative and quantitative reversibility of electrochemical processes.

Despite the value of in situ methods for analyzing and monitoring lithium-ion cells, their implementation presents specific challenges. Common issues include difficulties in monitoring processes over small areas of a component (especially with optical techniques), which may not represent the overall state of the battery; [1, 6] variability in conditions during a single measurement (due to a lack of equilibrium while data is collected); challenges in positioning analytical probes within measurement devices to replicate real operating conditions; and limited access to specialized techniques such as synchrotron radiation, which often require expensive equipment and accessories [4].

Many aspects of chemical changes within electrodes, electrolytes, and electrode/electrolyte interactions during cycling can be better understood using commonly employed techniques such as Raman spectroscopy [1] and X-ray diffraction (XRD) [12]. Vibrations within a crystal are influenced by its symmetry, chemical bonds, composition, structural order, and the presence of defects and distortions [7]. Raman spectroscopy is a highly sensitive vibrational technique capable of detecting changes in short-range order, local bonding environments, and molecular species, particularly those involving transition metal-oxygen vibrations, electrolyte degradation products, and SEI/CEI components [13]. This makes it suitable for identifying phase transitions, cation migration, local structural distortions, or oxygen loss in electrode materials, which often manifest as changes in peak position, intensity, or width. The detecting volume in Raman spectroscopy is primarily determined by the laser spot size and the penetration depth of the laser light. Typical laser spot diameters are in the several micrometer range (1–10 μm), while the probing depth can range from approximately 100 nm to 10 μm, depending on the laser wavelength and the material’s optical properties. Consequently, the Raman detection volume has approximately a few micrometers in diameter and a depth of up to 10 μm, which makes it a highly localized probe, sensitive to near-surface phenomena. In contrast, XRD is primary sensitive to bulk structural properties. It probes the long-range structural order of materials [14] and provides quantitative information on structural changes such as changes of unit cell parameters of crystalline phases, phase fraction evolution, and bulk structural transformations, including lattice contraction/expansion, phase segregation, and changes in crystal structure symmetry during electrochemical cycling. In typical reflection-mode XRD, the detection volume is determined by the incident X-ray beam size and the penetration depth. The X-ray beam size can vary from millimeters to centimeters for conventional laboratory setups, while the penetration depth is typically tens of micrometers, depending on the X-ray energy and the material’s density and composition. Therefore, the detection volume is often relatively large, with dimensions in the millimeter to centimeter range laterally and tens of microns in depth for common laboratory instruments. Therefore, Raman spectroscopy complements XRD analysis. Additionally, Raman spectroscopy is particularly useful for examining amorphous phases or electrode materials with low crystallinity, where XRD may not be effective. Both methods are non-destructive and non-invasive, allowing for the same sample to be analyzed using both techniques.

As detailed in the following paragraphs, previous studies using in situ Raman and XRD techniques have provided critical insights into the structural evolution of various Ni-rich and Li-rich layered oxide cathodes during cycling. These include phase transitions, changes in local bonding environments, and oxygen loss associated with surface reconstruction. These prior findings establish an important foundation for the following research in the field and beyond. A comparison of the new findings in high-voltage windows presented in this work with existing literature aims to underscore the consistency and novel aspects of the structural and electronic evolution observed in NMC111.

In situ techniques are often used to analyze structural and morphological changes in layered transition metal oxides during lithium-ion cell operation. For instance, Lin et al. [15] presented reconstructions of the NMC111 surface, where structural changes were directly associated with variations in the oxidation states of transition metals and atomic arrangements, as shown through synchrotron analysis using X-ray absorption spectroscopy (XAS) and annular dark-field scanning transmission electron microscopy (ADF-STEM). After cycling or exposure to the electrolyte, NMC particle surfaces exhibited progressive reconstruction—transitioning from the layered structure to the “rock salt” structure—along with the formation of a complex cathode electrolyte interphase (CEI). This reconstruction, involving a shift of transition metals to lower oxidation states, occurred primarily along the direction of lithium-ion transport and was highly anisotropic [15].

In the work by Ghanty et al. [16] in situ XRD and Raman spectroscopy were used to study lithium-ion extraction/insertion in the oxide matrix of NMC811 electrodes. These electrodes underwent reversible structural transformations from their initial hexagonal H1 phase to the H2 phase and a coexisting H1 + H2 phase, distinguished by different lithium cation concentrations and changes in lattice constants a and c. The results from both techniques were consistent [16]. In contrast, Lanz et al. [6]. observed that the Raman spectrum of Li-enriched material with a Li_1.2Mn_0.54Ni_0.12Co_0.12O₂ structure remained unchanged when charged up to 4.2 V. However, at a potential of 4.25 V, a new broad peak emerged due to Ni³⁺–O bonding around 542 cm⁻¹, which sharpened in the 4.35–4.47 V range. Other spectral changes linked to lithium-ion removal included the disappearance of the peak at 445 cm⁻¹ and its shift to 490 cm⁻¹. These changes, along with a change in the slope of the charge profile beyond 4.2 V, suggested lithium-ion extraction from the oxide matrix. During this process, transition metal ions moved into vacant lithium sites to maintain charge balance. Additionally, between 4.25 and 4.5 V, a significant increase in the Raman background signal was attributed to Li₂MnO₃ activation, indicating local structural modifications and oxygen release, as reported in previous studies. During the discharge process of Li₁.₂Mn₀.₅₄Ni₀.₁₃Co₀.₁₃O₂, the Raman spectrum reverted to its initial state, with a reduction in the background intensity.

Daemi et al. [17] reported structural transformations in NMC due to volume changes in the unit cell during cycling. When charged up to 4.7 V, the unit cell contracted. The overall decrease in specific capacity observed in various cells was attributed to a combination of factors such as poor cell sealing, electrolyte degradation, and cathode degradation. However, no direct correlation between unit cell volume changes and particle degradation was observed when individual particles were analyzed. These structural changes were attributed to the highly non-uniform internal structure of the particles, as further revealed through focused ion beam-scanning electron microscopy (FIB-SEM), which showed extensive cracking caused by high-voltage cycling.

From these studies, it is evident that in situ techniques, combined with additional ex situ analysis, are highly effective for describing the electrochemical processes in lithium-ion cells. The application of in situ Raman spectroscopy, which can detect subtle structural changes, provides numerous measurement opportunities. When combined with other, less sensitive but more volume-representative structural techniques such as in situ or ex situ XRD, as well as surface analysis methods like ex situ SEM or XPS, a comprehensive picture of changes in electrode materials during lithium-ion cell operation can be achieved. Skillful use of these techniques, alongside the continued development of in situ and ex situ procedures, is critical for advancing energy storage technologies.

In this work, we present a complementary study using in situ and ex situ Raman spectroscopy, combined with ex situ XRD and high-resolution transmission electron microscopy (HR-TEM), to investigate the structural evolution of NMC111 (LiNi_1/3Mn_1/3Co_1/3O₂) cathode material. By combining Raman and XRD techniques, supplemented with additional TEM analysis, our study offers complementary insights into both localized surface and bulk structural changes during cell operation, providing a more complete picture of the electrochemical processes occurring within the NMC cathode. Our results reveal significant surface reconstruction at high potentials (up to 4.7 V), where the layered structure of NMC111 transitions into a cubic phase. These changes were reversible during the initial cycles but may contribute to degradation mechanisms during prolonged cycling, especially under high-voltage conditions. We compared the in situ and ex situ Raman results, highlighting the advantages and limitations of each approach. Another novel insights from this work is the ability to track changes in electronic conductivity of NMC111 using Raman band intensity modulation. This provides a real-time probe of conductivity loss or recovery during electrochemical reaction. Furthermore, we explored how different laser excitation energies influence the Raman response of the NMC111 structure, finding that distinct wavelengths accentuate specific vibrational modes, allowing for a more detailed analysis of structural changes. Additionally, our ex situ XRD analysis confirmed that this method is highly effective for studying bulk structural changes in NMC111, complementing the surface-sensitive insights from Raman spectroscopy and advanced TEM analysis. Our near-surface HR-TEM analysis confirms the onset of surface degradation, lattice disorder, defects, phase transitions above 4.3 V, and stress generation that correlate strongly with Raman and XRD findings. Together, these techniques provide a comprehensive understanding of phase transitions in NMC111 and present a new pathway for correlating electronic and structural evolution and degradation in layered oxide cathodes under cycling conditions, offering insights into its stability and performance in lithium-ion batteries.

Experimental section

Sample preparation for in situ and ex situ measurements

The preparation of cells for these measurements differed from that of standard laboratory cells due to the specific requirements of each method: the need for a measurement window in the in situ method, and the need to obtain a sufficient amount of material after cycling in the battery for further analysis in ex situ methods.

Both methods required the preparation of the electrode paste. A total of 40 mg of NMC111 commercial material (Sigma Aldrich, 98% purity) was wet milled with 4 mg of a carbon mixture (PTFE (Sigma Aldrich, powder) with acetylene black (MSE, powder), 1:1 wt ratio) dissolved in 2 ml of ethanol (Sigma Aldrich ≥ 99.5%, 200 proof). The resulting paste was then pressed into a stainless-steel mesh and dried in a vacuum dryer to remove any residual water and solvent. Using a steel mesh instead of an aluminum collector enabled a 3D electrode design and facilitated in situ observation using Raman microscopy, as the electrode material filled the openings of the mesh.

To prepare sufficient samples for ex situ measurements, the same steel mesh with the pressed active material (AM) was used with identical material proportions (total mass of approximately 44 mg: 91% AM and 9% PTFE/AB).

Electrochemical tests

Chronopotentiometric tests for both in situ and ex situ studies were performed over a wide potential range (2–4.7 V) to observe structural changes during high-voltage charging and deep discharging for the first two cycles. For in situ measurements, a potentiostat/galvanostat (Autolab) was used, and the prepared electrode on the steel mesh was assembled in an in situ split test cell, with metallic lithium as the reference electrode. A standard electrolyte of 1 M LiPF6 in 1:1 vol. EC/DMC (Sigma Aldrich), soaked in a glass fiber membrane (Whatman), was used to separate the in situ cell electrodes.

For ex situ samples, the preparation involved the use of Swagelok-type three-electrode cells, with the same electrolyte and a Celgard^® membrane as a separator. In this setup, two lithium electrodes served as the reference and counter electrodes. Chronopotentiometric measurements were conducted using a multi-channel battery tester (Sollich Atlas 1361). Each sample was charged and discharged at a constant current of 0.1 C. After reaching the desired potential, each sample was held at a constant potential for two hours to stabilize the electrochemical potential. The cells were then transferred to a glovebox filled with argon, where the electrode potential relative to the metallic lithium reference was checked and recorded before disassembling. The positive electrode containing the tested material was gently removed, rinsed with dimethyl carbonate to remove electrolyte residues, and placed on a glass slide to dry for 48 h in the glovebox atmosphere (H2O/O2 content below 5 ppm). The prepared electrodes were then ready for ex situ measurements using Raman, X-ray diffraction, and TEM.

Additionally, the cyclability tests were performed within standard (2.0–4.3 V) and extended (2.0–4.7 V) potential windows at 0.2 C-rate to compare the specific capacity and capacity retention beyond initial cycles with standard and high-voltage conditions.

In situ and ex situ Raman studies

Raman spectroscopy measurements were performed using a Renishaw inVia confocal Raman microscope. For ex situ measurements, a Nd laser (532 nm, maximum power 50 mW) with a 2400 l mm⁻¹ grating, and a HeNe laser (633 nm, maximum power 17 mW) with a 1800 l mm⁻¹ grating, were used. The measurements were taken using x50 long-distance objectives, with further analysis conducted using OriginLab software.

In situ Raman studies were conducted using the inVia Raman microscope (Renishaw) equipped with a 633 nm laser source and a 1200 l mm⁻¹ grating, measured in a dedicated in situ split test cell with a quartz window (EQ-STC-RAMAN, MTI Corporation). A potentiostat/galvanostat was used to control the potential of the electrode containing the NMC111 relative to Li⁺/Li⁰ in real-time. The cell was charged at a constant current of 0.1 C up to 4.7 V and discharged at the same rate until the potential of the working electrode reached 2 V. Each cycle measurement took approximately 20 h, with open circuit voltage (OCV) applied between cycles. During charging and discharging, Raman spectra were recorded in real-time at 10-minute intervals (averaged from ten accumulations) to monitor changes at different states of charge. The setup for in situ Raman studies is shown in Figure S1. For all in situ measurements, 100% laser power was used to compensate for the longer penetration path and signal reduction, minimizing sample decomposition effects.

Ex situ Raman spectra were recorded from ten spots (10 s each, with 3 accumulations per spot) from each sample and averaged. Two excitation sources (532 nm and 633 nm) were used to assess the dependence of the Raman response on excitation energy. For all ex situ measurements, the laser power was reduced to 1% to prevent sample decomposition.

Ex situ X-ray diffraction

X-ray diffraction (XRD) measurements of the powder material and electrodes were carried out in classical Bragg-Brentano geometry using a Bruker D8 Advance diffractometer. A Cu X-ray source with Kα radiation (λ = 1.5419 Å) was employed, operating at 40 mA and 40 kV. Data were collected over a 2θ range of 10° to 80° with a step size of 0.02° s⁻¹ and a recording time of 1 s per step. The XRD data analysis was performed using Match!, FullProf, and OriginLab software. The instrumental broadening (FWHM_inst.) was ca. 0.063 degree. The estimation of crystal size was derived from five diffraction peaks, with 2θ ranging from 30 to 50 degrees.

Ex situ transmission electron microscopy studies

The electrode material was gently scratched off the aluminum current collector using a sharp scalpel directly onto holey carbon TEM grids without using any solvent dispersion step. This approach minimized mechanical damage and helped preserve the surface features of the particles. Imaging was performed on surface and near-surface regions of the NMC particles, which are accessible to the electron beam and most relevant to electrochemical reactions at the electrode–electrolyte interface. We also ensured that the observed features were not artifacts of mechanical disruption by selectively imaging intact particles and avoiding heavily fragmented regions. The microstructure of NMC111 electrode samples was examined using transmission electron microscopy (TEM) with a FEI Tecnai F20 operated at 200 kV. TEM-based techniques, including high-resolution transmission electron microscopy (HRTEM) and selected area electron diffraction (SAED), were used to analyze the OCV, charged, and discharged samples. Microscopic data analysis was performed using DigitalMicrograph^® (GATAN) software.

Results

Before conducting in situ and ex situ electrode measurements, the NMC111 powder was characterized using XRD and SEM methods (Fig. 1). The hkl indices correspond to the hexagonal space group of the NMC structure. Rietveld refinement was used to determine the unit cell parameters, yielding a and c parameters of 2.857 Å and 14.213 Å, respectively, consistent with the typical parameters observed for NMC111’s layered structure. Additionally, a (003)/(104) intensity ratio of 1.21 confirmed low cation mixing in the material. The separation of the (018)-(110) doublet (0.76) further verified good hexagonal ordering. Using the Scherrer equation [18], the average crystallite size was estimated to be approximately 33 ± 4 nm. SEM analysis revealed that the spherical particles of the powder ranged from 8 to 50 μm in diameter, with a microstructure indicating grain sizes between 20 and 40 nm, aligning well with the XRD data.

Fig. 1.

A XRD pattern and B and C SEM images of the NMC111 powder

In situ Raman measurements on NMC111 electrodes

This study focuses on the initial cycles of NMC111 material. Raman spectroscopy was employed to collect time-dependent spectra of individual particles during the first two cycles, aiming to investigate the dynamics of the Li_xMO₂ system (where M is manganese, nickel, and cobalt) as a function of lithium content (x denotes the Li fraction). The aim was to correlate the spectroscopic response of the NMC111 structure during the first two cycles with results from ex situ studies.

Figure 2 shows the in situ potential-dependent Raman mapping of the NMC111 material during the first cycle at a 0.1 C-rate, alongside selected spectra. The spectra shown in Fig. 2A were normalized for better visualization of structural changes, while background subtraction was applied to those in the map.

Fig. 2.

In situ Raman study of the first cycle of the NMC111 electrode: A Raman spectra from OCV through charge (up to 4.3 V) and discharge (down to 2.3 V); B corresponding Raman map of the 1st cycle process. Excitation source: 633 nm

Group theory predicts two Raman-active vibrational modes for the lithiated hexagonal NMC structure with space group: the E_g mode, involving the opposite movement of oxygen atoms along adjacent oxide layers, and the A_1g mode, originating from the symmetric movement of oxygen atoms along the c-axis, typically showing a stretching character [19–22]. For single transition metal oxide structure such as LiCoO₂ with all three ion types at their confined oxidation states, the Raman spectra presents two Raman bands. The situation becomes more complex when additional transition metal ions are introduced to the crystal structure. The presence of at least nine distinguishable bands for the layered NMC111 oxide powder suggests additional vibrational modes beyond the theoretically expected E_g and A_1g modes for the symmetry system (Figure S2) [6]. Deconvoluting these modes, particularly the E_g/A_1g pair, is challenging due to the presence of additional bands above 620 cm⁻¹ and highly overlapping bands below 580 cm⁻¹. These extra modes arise from distinct lower-symmetry local environments, related to variations in length and strength of the (Ni/Co/Mn)-O bonds [23]. The complexity arises from the unique vibrational behavior of the oxide framework in the material, with different vibrational modes of various transition metals contributing to the Raman spectra, potentially influenced by their different oxidation states. Calculations of the band structure of NMC materials would greatly enhance our understanding of the structural changes occurring during cycling. However, the exact origin of the phonons in NMC structures remains unresolved [1].

As is known, the structure of the material changes at different battery states of charge due to the progression of the electrochemical reaction. Starting from the open circuit potential to a potential of ~ 3.8 V vs. Li⁺/Li, we observed an increase in the band intensity around 600 cm⁻¹, clearly visible in in situ Raman mapping. This indicates an initial decrease in electron conductivity within the material at this potential [24, 25]. It is noteworthy that a first-order Mott transition in LiCoO₂ (LCO) was theoretically predicted [26]. A first-order Mott transition is a phase transition between a Mott insulator and a metal where the transition is sudden. Marianetti et al. [26] used density functional theory (DFT) to study LCO structure and demonstrated that for dilute Li-vacancy concentrations, a vacancy binds a hole and forms an impurity band yielding a Mott insulator. Consequently, it was determined that LCO should exhibit an insulating state for Li fractions greater than 0.95 and a metallic state for fractions less than 0.75. As experimentally demonstrated by E. Flores et al. [1], upon the initial delithiation, the intensity rapidly drops to zero once the lithium content falls below x = 0.92, showing that the phase with metallic properties appears sooner than theoretically predicted [27, 28]. Since the metallic properties of the material reduce the optical skin depth due to strong interaction between free electrons and incident light, the Raman intensity drops as a result of this skin effect [1, 29]. Given that, it is not surprising that once two other transition metals (Ni and Mn) are incorporated into the crystal structure, the electronic bands become even more complex [30]. From a potential of 3.8 V to approximately 4.6 V, we observed a progressive decrease in the intensity of the bands, indicating an increase in electron conductivity within the grain, due to the narrowing of the energy gap in the partially deintercalated NMC111 [1, 31]. This electronic conductivity increase is associated with the presence of mixed Ni³⁺/Ni⁴⁺ valence states resulting from delithiation, which leads to hole formation in the narrow (Ni⁴⁺/Ni³⁺) band [24]. Around 4.7 V, a sudden increase of the intensity of the bands was observed, corresponding to a second decrease in electron conductivity as the material becomes high-voltage charged. This is related to the near-complete deintercalation of lithium from the structure. As will be demonstrated in the following diffraction data, this phenomenon is associated with a phase transition from a hexagonal to a cubic phase and thus another change in electronic structure of the delithiated material. Given the established properties of cubic NiO and MnO, which are known for their insulating characteristics [32, 33], and of Co₃O₄, which is known for its semiconducting properties [34], the observed increase in Raman intensity in fully delithiated NMC111 is not unexpected.

During the first discharge below 4.5 V, there is a gradual decrease in the intensity of the bands, again indicating the transition from an insulator/semiconductor to metallic phase and thus an increase in electronic conductivity at the discharge plateau, down to 3.6 V. However, the magnitude of this change is smaller than that observed during the first charging process. From 3.6 to 2 V, the ~ 600 cm⁻¹ band once again increases in intensity, indicating a decrease in electronic conductivity related to the metal-insulator transition. Similar conclusions were presented by Qui et al. [35] based on their EIS measurements performed during the first charging of NMC111 (for the OCP potential range up to 4.4 V), although they did not study the range above 4.4 V. This highlights the remarkable usefulness of in situ Raman measurements for studying electrode materials in lithium-ion batteries, not only in term of the crystallographic changes but also for tracking changes in electronic properties during cell operation.

Continuing the structural analysis, new bands appear between 400 and 600 cm⁻¹ from a potential of 3.8 V upwards, clearly visible in the corresponding in situ Raman spectra. Above 3.8 V, we observe a decrease in the ~ 600 cm⁻¹ band and an increase in the 468–580 cm⁻¹ range. Maximum intensities at ~ 472, ~535, and ~ 576 cm⁻¹ are observed at the end of the first charging (4.7 V), consistent with findings previously described in the literature by Flores et al., whose charging potential range was OCV − 4.3 V [1, 6]. In addition to previous data, charging the cell above 4.3 V, causes an increase in the bands above ~ 600 cm⁻¹, due to significant structural changes on the crystal surface of the NMC111 material and the formation of an additional crystalline phase, such as a cubic transition metal oxide [36]. As will be discussed further, the diffraction data confirms that assumption. When the cell is discharged to 3.6 V vs. Li⁺/Li⁰, the bands at ~ 472 cm⁻¹ and ~ 535 cm⁻¹ disappear, while the band intensity at ~ 600 cm⁻¹ increases. Further discharge restores a structure close to the original one, indicating that the reactions occurring during the first cycle are nearly reversible. At deep discharge (~ 2 V), broadening of the bands is observed (Figure S3), particularly evident in the in situ Raman map, which is related to the worsening crystal quality due to the intercalation process, as will be discuss further.

Ex situ Raman measurements on NMC111 electrodes

The charge/discharge profile of NMC111 cycled at 0.1 C-rate, along with the points representing the ex situ collected samples, is shown in Fig. 3. The first and second cycle curves appeared similar, with the first charge exhibiting additional plateaus at 4.06 V and 4.44 V (Figure S4), likely due to CEI formation and associated phase transitions. The chemical nature and evolution of the CEI on NMC111 are known to be voltage-dependent, particularly with standard electrolytes such as 1 M LiPF₆ in EC/DMC (1:1 v/v), which was used in this study. At around 4.06 V vs. Li⁺/Li, early stages of electrolyte decomposition starts, marked by hydrogen evolution from residual moisture and a slow, linear increase in CO₂ concentration from carbonate solvent oxidation [37]. These processes contribute to the initial nucleation of the CEI. As the potential increases towards 4.44 V, electrolyte oxidation accelerates, leading to the formation of a complex CEI composed of both inorganic species like LiF and Li₂CO₃, and organic components such as polycarbonates and lithium alkyl carbonates (ROCO₂Li), derived from the decomposition of EC, DMC, and LiPF₆ [38].

Fig. 3.

Charge/discharge profile of NMC111 cycled at 0.1 C-rate with ex situ sample points

While more Ni-rich NMCs exhibit significant lattice oxygen release around 4.4 V, NMC111 shows relatively less severe degradation and lower gas evolution at 4.44 V, with its primary lattice oxygen release occurring at higher potentials (~ 4.7 V) [39]. The formation of the CEI is critical for preventing direct contact between the cathode and electrolyte, thereby limiting parasitic side reactions, impedance rise, and gas evolution, which are interconnected degradation pathways affecting overall cell performance. Additionally, the first charge was higher than the second, and the discharge values showed some irreversibility, with 110 mV and 20 mV differences between the primary oxidation and reduction peaks in the first and second dQ/dt curves, respectively. This suggests a different charging behavior during the first cycle.

Figure S4 compares in situ and ex situ Raman measurements, revealing very similar changes and strong consistency between the two methods. This consistency indicates that ex situ studies can also be used reliably, offering several advantages, such as longer measurement times, enhanced spectral intensity, and more accurate analyses across multiple material grains, which leads to better measurement statistics. The lower intensity observed in the in situ measurements during the 2nd cycle is likely due to CEI formation and reduced laser penetration into the crystal surface.

The ex situ spectra were obtained by averaging Raman measurements from ten different grains per sample. Compared to the first cycle, the second cycle shows a faster reduction of the characteristic peaks of NMC111 at a Raman shift of ~ 600 cm⁻¹ and an increase in the intensity of bands at ~ 470 cm⁻¹, ~ 535 cm⁻¹, and 580 cm⁻¹. Notably, samples recorded ex situ exhibit different dynamics due to the specific sample preparation method: the constant current– constant voltage method. The 2-hour constant voltage hold at each potential stabilizes the reaction equilibrium, resulting in a slightly different state of charge or discharge compared to chronopotentiometric cycling. This extended time allows the entire sample to equilibrate at the set potential before being disassembled at its equilibrium state. It also suggests that localized overpotential effects may occur in in situ measured particles, altering their response. During the second cycle, the ~ 600 cm⁻¹ band does not clearly disappear above 4.3 V, indicating that the electrochemical reaction is not fully reversible. This trend was observed for both measurement methods, implying that not all Li ions can be deintercalated from the structure, leading to capacity loss after the first cycle, as is commonly observed in electrochemical measurements (Fig. 3). Another factor affecting the first cycle is the tendency of Li⁺ ions to prefer four-fold coordination with oxygen atoms in the cubic structure, compared to six-fold coordination in the quasi-layered structure.

Ex situ measurements were also performed using two laser sources: 633 nm and 532 nm (Fig. 4, S5 and S6). The Raman spectra clearly depend on the excitation energy used. The band intensity varies depending on the laser source, similar to observations in the LiMn₂O₄ spinel structure [40]. This energy dependence suggests changes in the electronic band structure of the NMC111 electrode influenced by the excitation energy. Specifically, variations in the observed band gap imply that changes in the Li content within the NMC material directly affect its electronic properties. As the Li content varies, shifts in the band gap occur, indicating a distinct electronic band structure at different lithium concentrations. This demonstrates that the distribution and amount of lithium within the NMC111 structure significantly influence its electronic characteristics, consistent with intensity changes observed in the in situ Raman study and previous EIS measurements [35]. These findings underscore the complex relationship between lithium distribution and the electronic structure of NMC, which has implications for structural changes in subsequent cycles and overall performance in lithium-ion batteries.

Fig. 4.

Ex situ Raman spectra of the NMC111 electrode recorded at data points along the first two cycles starting from OCV through high-voltage (4.7 V) state following with deep discharging (2 V). Spectra recorded using two different excitation sources: 633 nm and 532 nm for comparison

Additionally, we compared OCV samples with the 1st and 2nd discharged electrodes using these two laser sources (Figure S7). The trends were similar, with some Raman bands appearing more intense when using the 532 nm laser, further indicating the presence of multiple Raman bands in the NMC spectra. After the first deep discharge (2 V), the NMC111 sample nearly returns to the same state as the OCV sample. Samples stopped at 3 V show slightly incomplete lithiation, particularly in the second cycle, reflected in variations in band intensities.

Ex situ XRD measurements on NMC111 electrodes

The samples were further examined using ex situ XRD (Fig. 5). All diffractograms were calibrated using the 2θ reflection positions of metallic iron, originating from the steel grid (current collector). The diffractograms were analyzed using the Rietveld method.

Fig. 5.

Ex situ XRD study of the 1st and 2nd cycle of the NMC111 electrode: A X-ray diffraction pattern progression and B unit cell parameters evolution

During the charging process, as lithium ions are deintercalated from the NMC material, shifts occur in the diffraction peaks, corresponding to changes in the distance between lattice planes in the NMC structure (Fig. 5A). These shifts, mirrored in the changes of the a and c unit cell parameters (Fig. 5B) originate from complex atomic rearrangements occurring within the layered α-NaFeO₂ structure (R-3 m space group). These changes are driven by an interplay of lithium deintercalation from 3a octahedral sites, charge compensation through the oxidation of transition metal (TM) ions (primarily Ni²⁺ and Co³⁺ to higher valence states) at 3b octahedral sites, and the resulting changes in ionic radii and electrostatic interactions between atomic layers. Specifically, some families of planes, such as the (00l) and (018), shift to lower 2θ angles during the middle of the charge, indicating an expansion of the c-axis attributed to the increased electrostatic repulsion between the negatively charged O-TM-O slabs upon lithium removal. The removal of positively charged Li⁺ ions from the inter-slab space significantly reduces the electrostatic shielding that these ions previously provided between the adjacent, negatively charged oxygen layers (O-TM-O slabs) [41, 42]. Furthermore, due to the nature of low-spin Ni³⁺ and Co⁴⁺ ions as Jahn-Teller ions, it is likely that upon charging, and thus TM oxidation, the ongoing deformation from the octahedral symmetry of CoO₆ and /or NiO₆ octahedra occurs to lower the energy [43]. The Jahn-Teller effects during Li deintercalation from NMC111 were also theoretically predicted using extensive Coulomb energy analysis and DFT calculations [30]. With reduced shielding, the inherent coulombic repulsion between the negatively charged oxygen sheets intensifies, effectively pushing them further apart, thus, leading to c-axis expansion during early delithiation.

However, at the high-voltage charge, these reflections shift to higher 2θ values, signaling a significant contraction of the c-axis, driven by the high concentration of lithium vacancies and a dominant electrostatic attraction between the increasingly oxidized TM layers. Specifically, at high potentials, the continued removal of Li⁺ ions and strong, charge neutrality-induced, oxidation of TM ions (e.g., Ni²⁺ to Ni⁴⁺) results in a significant increase in the effective positive charge density within the TM layers. This leads to a stronger electrostatic attraction between the now highly charged O-TM-O slabs, causing them to pull closer together and resulting in a significant contraction of the c-axis [44]. Furthermore, at these high voltages, lattice oxygen anions (O²⁻) can begin to participate in the charge compensation mechanism, a process known as anionic redox [45]. The interplay of cationic and anionic redox can influence the overall structural response, potentially contributing to the c-axis contraction by altering the inter-slab interactions or inducing structural instabilities. This c-axis contraction at high delithiation is a well-known phenomenon and is often associated with structural degradation, increased internal stress, and potential phase transitions (e.g., the H2→H3 transition) in NMC materials, particularly Ni-rich variants [44].

In contrast, the (10l) family and (110) plane exhibit the opposite trend, initially shifting to higher 2θ values, indicating that the a parameter contracts (~ 1.4%) during the early stage of charging. This contraction is largely driven by the oxidation of TM ions, as the cathode must maintain overall charge neutrality during lithium deintercalation. Charge compensation is primarily achieved through redox reactions involving the TM ions [46, 47]. For NMC111, nickel (Ni²⁺) and cobalt (Co³⁺) are the main electrochemically active species. During charging, they oxidize to higher valence states, such as Ni³⁺, Ni⁴⁺, and Co⁴⁺ [30, 48, 49]. This oxidation of Ni and Co to species with smaller ionic radii (e.g., Ni²⁺ to Ni³⁺/Ni⁴⁺) leads to a concomitant contraction or shortening of the average TM − O bond lengths within the TM − O₆ octahedra [23]. This shortening of the intra-layer bonds directly causes the contraction of the a lattice parameter, which defines the dimensions within these TM-O layers. At high-voltage charge, the a-axis relaxes, but does not fully return to its initial value. At such high states of charge, the material may undergo subtle structural rearrangements to accommodate the extreme state of delithiation and the high internal strain that accumulates [50]. This structural adjustment is the material’s attempt to relieve the accumulated strain from extensive lithium removal and strong electrostatic interactions within the lattice. Factors contributing to this relaxation include oxygen redox, which becomes a significant charge compensation mechanism at high delithiation, and TM migration (e.g., Ni ions moving from TM layers to Li layers), which introduces structural disorder and can induce anisotropic stress, influencing the lattice parameters and potentially contributing to a-axis relaxation by altering the local environment and bonding within the TM layers [51].

Notably, a broadening of peaks is observed above 4.3 V in both cycles, indicating phase transitions and the coexistence of multiple phases at high charge states. This effect is reversible in those initial two cycles. Previous studies have shown that changes in the oxidation state of transition metals during lithiation/delithiation can induce the formation of additional structures (e.g., cubic phases) at the surface of various NMC compositions [15, 52]. The formation of high-spin Mn³⁺ ions, potentially originating from Li₂Mn₂O₄-LiMn₂O₄ spinel structures, has been described by Mohanty et al., [53] although a spinel structure was not observed when within a narrower potential range (2.4–4.2 V). Our Raman study and subsequent TEM examinations are consistent with these findings.

There are minor differences in the changes to the a and c unit cell parameters between the 1st and 2nd cycles. During the first charge, the c parameter increases by about 0.15 Å at 3.8 V, then further by 0.1 Å at 4 V, expanding the interplanar distances in that direction. This expansion persists up to 4.5 V before contracting significantly at high-voltage charge as lithium is heavily deintercalated. In the second cycle, the expansion at 3.8 V is only 0.1 Å, and shrinkage begins earlier and at 4.5 V is already ~ 0.075 Å. The a parameter shows an initial decrease of ~ 0.03 Å at 3.8 V, with a further 0.01 Å reduction upon charging to 4.3 V, followed by a small increase at 4.7 V. In the second charge, changes to the a parameter occur slower, similar to the c parameter. During discharge, similar trends are observed for both cycles, indicating structural consistency across them. This correspondence between the dQ/dt discharge curves reflects similar structural changes in the NMC111 during both cycles.

During the first charge, the unit cell volume contracts by 1.6%, while during discharge, it expands by only 0.25% compared to the clean sample. The unit cell parameters revert to their initial values after a deep discharge to 2 V, confirming that intercalation changes can be reversible. The minimal expansion of a and c parameter of ca. 0.003 Å and 0.001 Å, respectively, was observed. However, discharging to 3 V, as is often done in NMC research, may not be sufficient to fully restore the original NMC111 structure, suggesting that deep discharges are recommended during formation cycles. When the cell is fully discharged to 2 V in the second cycle, the a and c parameters are again close to their original values, although with a slightly higher c unit cell parameter (by ca. 0.02 Å). After 10 cycles, the unit cell parameters align closely with the values after the second cycle of operation (~ 0.001 Å and ~ 0.02 Å). This shows that the crystalline structure of the NMC111 material will rather slowly degrade over cycling within the extended potential window (2–4.7 V). While the main oxidation and reduction peak differences increase slightly to about 40 mV, the capacity retention remains at 87% after ten cycles (Figure S5). Additionally, the difference between the second discharge at 3 V and 2 V is minimal, suggesting that deep discharge after formation cycles may not be strictly necessary for restoring unit cell parameter values. The excess electrolyte in the Swagelok cell likely minimizes the effects of electrolyte decomposition due to the high operating potential.

Ex situ TEM measurements on NMC111 electrodes

We further investigated selected samples using HR-TEM imaging (Figs. 6 and 7, S8–S11), focusing on OCV, 1st and 2nd highly charged, and 2nd deeply discharged samples. Figure S8 shows all the samples with marked d-spacing values. OCV and discharged samples exhibit hexagonal (101) or (102) plane distances, whereas charged samples reveal d-spacings indicative of a cubic-like structure.

Fig. 6.

HR-TEM images showing NMC111 samples at 1st charge up to 4.7 V. The analysis revealed the change from hexagonal into cubic-like phase at high charge. The calculations of angles between to the (111) and (200) planes of the cubic phase and their d-spacing are provided. The enlargement of the image with assigned planes showed that atomic ordering does not follow the hexagonal structure rules

Fig. 7.

A–C SAED patterns of the electrodes at OCV, delithiated, and lithiated states. The calculations of angles between to the planes and their d-spacing are provided showing the initial reversibility of hexagonal to cubic phase transition at the near surface of the grain. D Structural model of the NMC111 delithiation. E and F TEM images of the lithiated and delithiated electrodes revealing stress fringes across the grain bulk after lithiation and planar defects (streaks parallel to the (111) planes of the cubic phase) after high-voltage charging

Detailed analysis of the charged samples (Fig. 6) showed that atomic ordering does not follow the hexagonal structure rules. The angles between the planes were approximately 55° and 70°, consistent with a cubic structure [54]. The d-spacings correspond to the (111) and (200) planes of the cubic phase, similar to Ni₆MnO₈ [55, 56] or NiO [54], which crystallize in the space group. This suggests that at high potentials, transition metal ions in the NMC111 structure reorganize at the grain surface. Additionally, it is worth noting that the d-spacing of the (111) plane varies at the surface of the delithiated NMC111 grains (Figure S8) due to numerous dislocations. The (111) plane spacing ranged from 4.85 Å to as much as 5.5 Å, depending on the location, indicating highly defective delithiated NMC surface at the high potential state. We also observed that the charged grains contained elongated voids, which are clearly visible in Figure S9. These enlarged planar defects are also spotted in the Fig. 6 where the contrast is low.

The SAED analysis confirmed the structural transformation of NMC111 during cycling, showing a transition from hexagonal to cubic and back to hexagonal structure within the extended potential window (Fig. 7). The highly lithiated phases displayed the six-fold symmetry along the [001] direction, with 60° angles between the (110), (-210), and (1–20) planes. In contrast, the highly charged particles exhibited a clear cubic structure, with approximately 54° between the (200) and (111) planes, and around 70° between planes within the (111) family. This cubic structure suggests a reorganization of transition metal (TM) ions at the surface of the NMC grains after delithiation, with some TMs partially occupying lithium vacancies, resulting in new symmetry. Additionally, planar defects were observed in the SAED image of the delithiated structure, evident from the well-defined streaks in the electron diffraction patterns (Fig. 7B and F, and S10) [57]. These streaks are parallel to the (111) planes in the cubic phase, which correspond to the (003) planes in the lithiated structure. This indicates that the planar defects form along the [110]_hex direction, which aligns with the lithium out-diffusion path from the NMC structure (Fig. 7D). As the Li layers become vacant, TM ions appear to migrate into these vacancies, creating regions with increased spacing, visible as planar defects. This phenomenon aligns with previously reported oxygen release from the NMC111 structure at high potentials vs. Li/Li⁺ [58]. As the TM ions migrate from their initial layers, oxygen anions may be reduced and released from the NMC grain’s planar voids and surfaces. The atomic reorganization, such as TM hopping into Li-vacant layers, might not be the only contributing factor; TM dissolution at high deintercalation levels could also occur when the material operates in a high potential window [59].

The TEM images further revealed that the lithiated grains experience stress buildup after re-lithiation (Fig. 7E and S11). Compared to the OCV state, the lithiated samples showed stress fringes across the grain bulk, characterized by alternating bands of dark and bright contrast. This is likely a result of changes in unit cell parameters that occur during delithiation and re-lithiation processes. It also suggests that if grain volume changes occur in each cycle, eventual particle cracking is inevitable. Given the composition of NMC particles, which consist of smaller grains interconnected through grain boundaries, the cyclic grain shrinkage and expansion can disrupt the grain boundary structure over time. Consequently, the observed accumulation of stress resulting from cycling supports the grain boundary cracking mechanism rather than the grain bulk cracking. This is because within the particle it is the amorphous grain boundary structure that is most prone to fracturing during the electrochemical process due to its disorderly nature.

Discussion

Our studies using combined in situ and ex situ Raman, XRD and TEM techniques align well with findings presented by Lin et al. [15] They observed the surface reconstruction after 1 cycle (2.0–4.7 V) of the LiNi_0.4Mn_0.4Co_0.18Ti_0.02O₂ structure, transitioning from an layered structure to an rock-salt structure, as revealed through synchrotron XAS and ADF-STEM analysis. Similarly to our study, they also observed orientation-dependent surface changes at 4.7 V along the lithium diffusion channels. Analysis of NMC532 material after 50 cycles at 4.8 V showed that structural changes were mainly localized on the surface, while the bulk region remained rhombohedral [52]. The authors observed the formation of a spinel phase as a continuous layer extending 12–15 nm from the surface, with a narrow top-surface region (~ 2–3 nm) containing a cubic phase (NiO-like rock-salt structure). Another study on Li_1.44Ni_0.2Co_0.2Mn_0.6O₂ material revealed that after cycling, some grains showed a thick layer of rock-salt structure () about 15 nm thick between the bulk phase and a surface amorphous CEI layer [60]. Other grains exhibited mixed phases, including spinel-like and / phases. The authors suggested that transition metal cation migration occurred continuously during cycling, resulting in the formation of dense spinel-like and NiO phases.

TM migration into Li vacancies was also observed by Boulineau et al. [61], who applied high-resolution HAADF-STEM imaging and EELS mapping on cycled Li_1.2Mn_0.61Ni_0.18Mg_0.01O₂ material. They concluded that during the first charge, the extraction of lithium ions led to oxygen loss from the surface to maintain charge balance, consistent with the work of Jung et al. [58] This oxygen loss caused some TM cations on the surface to become coordinated by only five oxygen atoms, making them unstable. To stabilize the structure, the cations migrate to vacant octahedral sites left by the extracted lithium ions, initiating the formation of a spinel phase. This study also observed particle shrinkage, which was similarly detected in NMC111 using CT-XRD, allowing spatially resolved characterization of local heterogeneities in the unit cell volume [17]. The authors observed expansion along the c-axis and shrinkage along the a-axis due to extended cycling, with larger shrinkage observed in the crystal structure of electrodes cycled to 4.7 V.

The findings discussed above are closely related to and consistent with our study on NMC111 material. Figure S12 presents the specific capacity and capacity retention over cycling for the NMC111 material tested within standard (2.0–4.3 V) and extended (2.0–4.7 V) potential windows at 0.2 C-rate. The sample cycled within the standard window delivered specific discharge capacities of 159 mAh g⁻¹ and 130 mAh g⁻¹ during the first and fiftieth cycles, respectively, corresponding to 82% capacity retention after 50 cycles. In contrast, the sample cycled between 2.0 and 4.7 V exhibited higher initial capacity of 206 mAh g⁻¹, although experience capacity drop to 140 mAh g⁻¹ in the 50th cycle, resulting in a greatly reduced capacity retention of 68%. Its initial capacity drop from the 1st charged state was 40 mhA g⁻¹ (16%) versus 13 mhA g⁻¹ (7%) for 4.3 V. Expanding the upper cutoff voltage increased the initial specific capacity, but it significantly accelerated the degradation of the electrode. Thus the reduced cycling stability observed at higher voltages (above 4.3 V) is attributed to progressing structural degradation of the active material as discussed above (phase transformation from hexagonal to cubic and increased mechanical stress due to lattice size changes).

Conclusions

In this work, we present a study using in situ and ex situ Raman spectroscopy, along with ex situ XRD and HR-TEM, to investigate the structural evolution of NMC111 cathode material over the extended potential window cycling. Our results show significant surface reconstruction at high potentials, where the layered structure transitions into a cubic phase. In situ Raman spectroscopy revealed a sudden increase in band intensity around 4.7 V during the first charge, likely due to decreased electron conductivity associated with high levels of lithium deintercalation. Additionally, charging above 4.3 V led to the increased bands above ~ 600 cm⁻¹, indicating structural changes and the possible formation of the cubic phase. At deep discharge (~ 2 V), band broadening was observed, possibly due to worsening crystal quality, as further confirmed by XRD and TEM. In situ and ex situ findings were consistent, highlighting the benefits of ex situ studies, such as longer measurement times, better statistics, and the ability to perform additional complementary studies. We also observed excitation energy dependence in the Raman signal of NMC111, with band intensity varying based on the excitation source. This suggests changes in the electronic band structure of NMC111.

During charging, lithium-ion deintercalation caused specific shifts in diffraction peaks observed in ex situ XRD studies, reflecting changes in the NMC lattice over cycling. There were evident changes of the a and c unit cell parameters, with the c parameter initially expanding and then shrinking at high-voltage charge, while the a parameter exhibited the opposite behavior. TEM and SAED confirmed structural changes from hexagonal to cubic and back, with the formation of planar defects along the [110]_hex direction of the lithium out-diffusion pathway. These defects likely result from TM ion migration into lithium vacancies, creating spaced regions visible as planar defects. Grain stress built-up and volume changes seen in TEM and XRD during delithiation and re-lithiation suggest that grain boundry cracking is likely unavoidable over time, especially during the cell operation at extended potential windows. These findings, using a combination of in situ and ex situ Raman spectroscopy, XRD, and TEM, improve our understanding of the performance of lithium-ion batteries based on NMC materials, especially in applications requiring high-voltage operation.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

CEI

Cathode electrolyte interphase

CT-XRD

Computed tomography X-ray diffraction

EELS

Electron energy-loss spectroscopy

EIS

Electrochemical impedance spectroscopy

FIB-SEM

Focused ion beam scanning electron microscopy

HR-TEM

High resolution transmission electron microscopy

HAADF-STEM

High-angle annular dark-field scanning transmission electron microscopy

NMC

Lithium nickel manganese cobalt oxide

OCV

Open circuit voltage

SAED

Selected area electron diffraction

SEI

Solid electrolyte interphase

TM

Transition metal

TEM

Transmission electron microscopy

XAS

X-ray absorption spectroscopy

XPS

X-ray photoelectron spectroscopy

XRD

X-ray diffraction

Author contributions

D.A.B. conceived the original idea, designed the study, and secured funding, contributing significantly to result interpretation. M.B. performed and analyzed electrochemical data, prepared samples for ex situ analysis, and conducted in situ Raman experiments, with D.A.B. performing the analysis. D.A.B. and M.B. carried out ex situ XRD experiments. D.A.B. performed ex situ Raman investigation. J.B.J. performed TEM experiments, contributed to data analysis and interpretation, and provided additional support for the in situ Raman experiments. D.A.B. conducted comprehensive structural and morphological analysis of all in situ and ex situ data and developed crystallographic models. D.A.B. wrote the manuscript and prepared visual materials, while J.B.J. revised the manuscript. D.A.B. and A.C. supervised the study. All authors approved the final version of the manuscript.

Funding

This work was supported by Homing program of the Foundation for Polish Science (POIR.04.04.00-00-5EC3/18 − 00).

Data availability

The authors declare that the data supporting the findings of this study are available within the paper and its Supplementary Information files. Should any raw data files be needed in another format they are available from the corresponding author upon reasonable request.

Declarations

Ethical consent to participate, and consent to publish declarations

Not applicable.

Competing interests

The authors declare no competing interests.