Comprehensive analysis of degradation mechanisms in 18650 Li-Ion cells under prolonged cycling conditions

Research Papers Comprehensive analysis of degradation mechanisms in 18650 Li-Ion cells under prolonged cycling conditions Pavel Blaˇ zek a,c , Ondˇ rej Klvaˇ c b,e , Martin ˇ Sedina b , Ondˇ rej ˇ Cech b , Mark´ eta Tkadlecov´ a a , Zuzana Stravov´ a a , Tom´ aˇ s Kazda b , Tom´ aˇ s Zikmund a,* , Robert H. Schmitt c,d , Jozef Kaiser a a Central European Institute of Technology, Brno University of Technology, Brno, Czech Republic b Department of Electrical and Electronic Technology, Faculty of Electrical Engineering and Communication, Brno University of Technology, Czech Republic c WZL | RWTH Aachen University, Aachen, Germany d Fraunhofer Institute for Production Technology IPT, Germany e Thermo Fisher Scientic, Vlastimila Pecha 12, Brno 627 00, Czech Republic ARTICLE INFO Keywords: Lithium-ion battery degradation 18650 cylindrical cell Electrode delamination Aging ABSTRACT Li-ion batteries are essential for applications like electromobility and stationary energy storage, where long-term performance and safety are critical. While previous studies have explored degradation mechanisms in Li-ion cells, detailed insights into structural changes during extended cycling remain limited. This study aims to investigate the degradation mechanisms in a Samsung 18650 cylindrical Li-ion cell over 800 cycles (10 % to 90 % State of Charge) to understand performance fade and structural changes. Utilizing periodic micro-CT scans, we observed signicant geometric alterations in the electrode stack, including delamination and bending towards the cell axis, correlating with capacity loss. Post-mortem analyses using broad ion beam (BIB), scanning electron microscopy (SEM), energy-dispersive spectroscopy (EDS), and synchrotron CT conrmed voids in the cathode active material, copper deposition on the anode surface, and elevated levels of phosphorus and uorine, likely from electrolyte decomposition and SEI layer formation. These insights shed light on the structural changes and failure modes in cylindrical Li-ion cells during prolonged cycling and highlight the need for design improvements and optimized manufacturing processes to enhance mechanical robustness and reduce defects, thereby extending battery cycle life and safety. 1. Introduction Lithium-ion batteries (LIBs) have become a key technology for energy storage due to their high energy density, efciency, and versatility, providing a sufciently long lifetime. They play a pivotal role in various technologies, from small portable electronic devices to energy storage systems and electric vehicles [1–3]. One of the most common types of LIB cells is 18650-format, which is a cylindrical cell of 18 mm diameter and 65 mm length. It is used as a standard choice in numerous applications like power tools, electric bicycles and electric cars [4]. Battery aging manifests itself through a multitude of complex symptoms, which reect changes in micro and macroscopic level. It leads to a decrease in capacity, an increase in internal resistance, and physical changes such as swelling and delamination of electrode layers. It can lead not only to deterioration of performance but also damage battery structure, affect the safety of battery usage, and even culminate in complete battery failure [5–15]. The issue of battery aging is a very broad topic; further described are those processes that have been the subject of research in this paper. During cycling, the electrodes undergo volumetric changes manifested at the level of whole electrodes as well as individual grains of electroactive material [16–18]. This is mainly due to electrochemical processes involving lithium-ion intercalation and also lattice structural changes [19]. Temperature changes induced by high C-rate also play an essential role [20]. These effects in the cylindrical cells lead to the deformation of the electrodes and their buckling towards the center of the battery. This can result in delamination of the electroactive material and, in extreme cases, disruption of the separator. This phenomenon can be partially counteracted by the use of a central pin - a metal tube that lls the space in the center of the cylindrical cell, which, however, increases the weight [16]. Volume changes are also a cause of particle cracking [15,21,22]. It * Corresponding author. E-mail address: [email protected] (T. Zikmund). Contents lists available at ScienceDirect Journal of Energy Storage journal homepage: www.elsevier.com/locate/est https://doi.org/10.1016/j.est.2025.117436 Received 11 February 2025; Received in revised form 3 May 2025; Accepted 12 June 2025 Journal of Energy Storage 130 (2025) 117436 Available online 18 June 2025 2352-152X/© 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). can lead to the material losing contact with the surrounding matrix, making it inactive and decreasing the capacity. Moreover, during cracking, the existing SEI layer is disrupted and starts to grow on the newly formed surfaces. This consumes lithium ions, resulting in further capacity loss. An additional risk lies in the possibility of lithium dendrite formation manifesting itself mainly at higher loads and lower temperatures [10,23,24]. In the application of very high currents and operation of the cell outside the voltage operating window, dissolution of the current collectors may also occur. The cases of anodic oxidation of the copper collector and deposition of Cu ions on the cathode surface increasing its volume have been described [25]. Also, copper can form an internal short circuit through the formation of Cu dendrites. Rarely has the deposition of copper on the surface of the anode or individual grains been described. The reduction of copper dissolved in the electrolyte by reaction with the SEI layer compounds is assumed [26,27]. Evidently, the processes contributing to battery degradation and aging are complex, and their deeper understanding is crucial for the further development of LIBs. X-ray-based techniques are suitable for their investigation due to their non-destructive character and can also be used for in-situ analysis, where a scan is performed after a certain number of electrochemical cycles. This includes X-ray radiography [28] as well as advanced CT methods [29,30]. It is possible to achieve resolution from micrometers when analyzing whole cells down to submicrometer resolution of small pieces of electrode materials and detect even mild structural changes [31,32]. With scanning electron microscopy (SEM), even higher resolution can be achieved to observe individual grains, their cracking, and the growth of the SEI layer [11,33–38]. In combination with electron dispersive spectroscopy (EDS), it is also possible to determine the elemental composition and its changes after electrochemical cycling [39,40]. These analyses are usually performed only on the surface of the electrodes and if it is a cross-section, again it is a cross-section through only one side of one electrode, which cannot fully describe the process taking place in the whole battery [41]. At the same time, electrochemical data give an idea of the change in capacity and other properties of the battery, but do not provide accurate information about the degradation process, both in terms of deformation of the internal cell structure and changes at the microscopic level in the battery electrodes [42]. Since the cell must be disassembled before analysis, it is advisable to discharge it for safety reasons, to avoid short-circuiting during further handling, and to minimize contact with air. Argon gloveboxes are typically used for this purpose. The separated electrodes are then examined from the surface or cross-section. The materials are often brittle and can be damaged by mechanical stress. For this reason, focused ion beam (FIB) also allowing FIB/SEM 3D reconstruction [43–45] or broad ion beam (BIB) [8,46–49] techniques are often used for sample preparation, which sputter the material without mechanical force and allow the samples to be observed in their native state [7,8]. This study introduces a comprehensive, multi-scale, and multi-modal approach to characterizing degradation in 18650-type LIBs. We combine micro-CT, sub-micron CT, SEM, and EDS with electrochemical performance data across 800 charge-discharge cycles to create a full structural and chemical prole of battery aging. Notably, we apply virtual unrolling techniques to micro-CT data—enabling quantitative tracking of electrode deformation in cylindrical geometry, a technique rarely used in battery studies. CT scans were conducted every 200 cycles to observe geometric evolution in parallel with electrochemical metrics, including capacity, hysteresis, and impedance via Electrochemical Impedance Spectroscopy (EIS). By correlating large-scale structural deformation with localized chemical changes, this work provides a uniquely integrated view of LIB aging. It not only highlights degradation mechanisms at multiple length scales but also demonstrates how advanced CT methods can serve as predictive tools for battery health assessment. Our methodology offers a replicable blueprint for future studies aiming to bridge the gap between imaging and performance in real-world battery systems. 2. Materials and methods 2.1. Samples We investigated commercially available LIB Samsung INR18650-29E (manufacturer stated capacity 2860 mAh) based LiNi x Mn y Co 1-x-y O 2 (NMC), the most used cathode materials in LIBs. We used two accumulators from the same manufacturing batch. The rst one, fresh cell, was used for destructive analysis before cycling, and the other one was used for long-term cycling and degradation study. After the cycling was nished, we used it for destructive analysis of the effects of aging by submicron CT, SEM and EDS. Visualization of analyzed cell components and inner structure is shown in. 2.2. Aging and electrical analysis The test cell was characterized by a baseline test using two cycles in the full manufacturer-dened voltage range of 2.5–4.2 V at 0.1C/0.1C, 0.2C/0.2C, 0.2C/0.5C and 0.2C/1C (charge/discharge). EIS was performed at 100 % SoC during 0.1/0.1C cycle in a frequency range from 1 MHz to 30 mHz, with an amplitude of 10 mV, followed by CT scanning. The CCCV method was used with the limiting current set to 0.02C. The initial test was followed by long-term cycling at 1C using the CCCV method with a limit current of 0.02C in the range of 10 % to 90 % SoC for 200 cycles, followed by a baseline test and subsequently by CT scanning. This process was repeated until 800 cycles were reached. After the last baseline test and CT analysis, the cell was disassembled and analyzed by SEM. The electrochemical measurements were performed with a BioLogic battery cycler BCS-815 with an EIS module. 2.3. Micro CT analysis The whole LIB was analyzed by micro CT in the fresh state, after 200, 400, 600, and 800 cycles. For these measurements, we used a Thermo Scientic HeliScan microCT system equipped with a at panel detector with a resolution of 3072 ×3072 px 2 and pixel size 139 um and 160 kV microfocus tube. The whole battery was analyzed by one scan with a space-lling helical trajectory, which allowed the use of higher geometrical magnication and increased the signal-to-noise ratio while avoiding cone-beam artifacts. The X-ray tube was set to 150 kV, and the beam was ltered with 0.2 mm of stainless steel and 0.5 mm thick Sn foils to reduce beam-hardening. To obtain sufcient signal, the exposure time was set to 0.65 s, and ve radiographs were averaged in each of the 5400 projections per scan. The data were reconstructed with the iterative algorithm in software provided by the manufacturer. An algorithm for sample drift correction and the self-calibration algorithm [50] were used to correct for geometrical errors. The reconstructed volume size was approximately 22 ×22 ×73 mm 3 with (8 μ m) 3 voxel size. The virtual unrolling technique was applied to all the whole-cell datasets in Thermo Scientic Avizo. We used the procedure described in [51]. The cathode was segmented using the multithresholding method, and morphological operations such as opening/closing were applied. We analyzed the distance from the core on the segmented cathode and compared results between datasets. An unrolled distance map then allows for the assessment of cathode position and deformities in the entire cell volume. 2.4. Sample preparation for SEM, EDS, and submicron CT Before disassembly, batteries were deeply discharged to 1 V with a current of 0.05C. Inside the Ar-lled glovebox, the metal case was cut about 1 mm below the edge of the positive pin using a manual tube P. Blaˇ zek et al. Journal of Energy Storage 130 (2025) 117436 2 cutter. The CT inspection revealed that there is a free space inside with minimal risk of electrode damage. The positive pin was then removed using pliers, and the metal case was unrolled to approximately half the cell’s height. The electrolyte was dried by leaving the cell under a vacuum in the glovebox antechamber for about 1 h. Pieces of the individual electrodes were cut from the unrolled section using scissors. Further manipulation took place outside the glovebox. The cell was cut across on the negative pin side with a hacksaw about 1 cm above the pin edge. The metal case in this area serves as a mechanical xation and prevents structure shifting during cross-section preparation through the entire structure. This specimen was then mechanically ground by 30, 15, 8, and 5 μ m grain size SiC grinder papers with isopropanol. The Broad Ion Beam (BIB) polisher model 1061 SEM Mill (Fischione Instruments) was used for further sample preparation. The individual electrodes were prepared in cross-section mode, and the whole structure in planar mode using Cryo cooling. SEM imaging and EDS analysis were performed on a Scios 2 scanning electron microscope (Thermo Fischer Scientic). These samples were obtained from the upper half of the cell (near the positive pin) approximately in the middle of the winding (area between the center of the cell and its edge). For nanoCT, pieces of approximately 0.5 ×2 mm were cut from individual electrodes with a razor blade from similar region. The samples were inserted into a Kapton tube and xed with molten ethylene carbonate. For SEM/CT correlative analysis, the sample was rst polished using BIB, then SEM analysis was performed, and nally placed in the Kapton tube. 2.5. Submicron CT We used the unpolished part of the cell and unrolled the electrodes. A piece of it was cut to dimensions approximately 0.5 ×2 mm. This piece was then xed on a special holder. It was inserted in a Kapton tube and poured with ethylene carbonate to reduce oxidation. The sample was then imaged using nanoCT Rigaku nano3DX with Mo target, 50 kV tube voltage, 35 s exposure time, 800 projections, and voxel size of 0.54 μ m. 3. Results and discussion Continuous measurements revealed capacity decrease during cell cycling (Fig. 2 a). We also observed that after interrupting the cycling for analysis with CT, regeneration occurred, and capacity was increased. This effect was most signicant after 200 and 400 cycles. The most considerable capacity reduction of 20.8 % occurred after the rst 200 cycles. After that, the capacity regenerated to 90.8 % of the original value and, during another 200 cycles, decreased by a similar value as in the rst 200 cycles. This effect repeated also between 400 and 600 cycles. A notable capacity reduction of 11.3 % occurred between 600 and 800 cycles as well. In total, capacity dropped by 30.5 % after 800 cycles (see Table 1). The discharge characteristics at different C-rates before and after cycling are shown in Fig. 2 b. At all C-rates after 800 cycles of cycling, there was a decrease in the achieved capacity and, at the same time, a decrease in the discharge plateau. The capacity at 0.1C load has been reduced from 2865 mAh, which is equivalent to the capacity declared by the manufacturer, to 2442 mAh (14.8 % capacity drop). At 0.2C load, the capacity dropped from 2807 to 2344 mAh (16.5 % capacity drop). At loads of 0.5C and 1C, the capacity dropped by 19.2 % and 22.1 %, respectively. It is evident that cycling did not only lead to a decrease in capacity but also to a decrease in load ability at higher C-rates. The change in capacity during the baseline tests is shown in Table 2. When comparing the capacities of the characterization cycles, a significant change can be seen after 200 and 800 cycles, where after 200 cycles, there was a signicant change in capacity but stability at higher loads was maintained. Thus, at a load of 1C, there was a 4.2 % decrease before cycling and a 4.5 % decrease after cycling compared to a current of 0.1C. Another signicant change in this parameter occurred after 800 cycles. The hysteresis changes at different C-rates during cycling, as shown in Fig. 2 c. It is evident that the hysteresis increases with increasing Crate and is highest at 1C. After 200 cycles, there was an increase in hysteresis at all C-rates. Its value then remained stable during the following cycling until the last 200 cycles, when the value increased again signicantly for all C-rates. This increase of hysteresis is in correlation with the signicant capacity drop during the last 200 cycles of cycling and corresponds to the large capacity drop at a current load of 1C that was observed at the end of cycling. The EIS analysis data including the equivalent circuit is shown in Fig. 1 d. The equivalent circuit used to interpretation of the measured EIS data includes serial resistance Rs featuring ohm internal resistance, Rct represent charge transfer resistance of the charge transfer process, Q is constant phase element represents internal capacitance and Zw is Warburk impedance which represents diffusion and mass transport. This circuit was chosen because the curve forms one semicircle as in publications of Liu et al. [52] and Santoni et al. [53]. The EIS analysis revealed that the Rct at the beginning of cycling was double (12.3 mΩ) that after 200 cycles (5.6 mΩ), which is visible in Nyquist plots in Fig. 2 d. This is due to the changes in the structure at the beginning of cycling, which lead to improved contact with the electrolyte. Liu et al. give a similar description of this change [54]. The Rct value was almost identical between 200 and 400 cycles (5.5 mΩ), corresponding to the observed capacity regeneration and nearly identical change in hysteresis. Subsequently, Rct increases with cycling, and after 800 cycles, its value approaches the original values before cycling (9.4 mΩ). The dV/dQ analysis before cycling and after a different number of cycles (200, 400, 600, and 800) is visualized in Fig. 3. The gure shows Table 1 Capacity and capacity retention in different stages of cycling. Cycle number Capacity/mAh Capacity retention/% 1 2165 – 200 1714 20.8 % 400 1668 23.0 % 600 1717 20.7 % 800 1505 30.5 % Table 2 Degradation under different loads during long-term cycling. Capacity/mAh Capacity drop/% Cycle number 0.1C 0.2C 0.5C 1C 0.2C vs 0.1C 0.5C vs 0.1C 1C vs 0.1C 0 2865 2807 2742 2744 98.0 % 95.7 % 95.8 % 200 2660 2591 2591 2541 97.4 % 97.4 % 95.5 % 400 2583 2555 2503 2490 98.9 % 96.9 % 96.4 % 600 2478 2459 2400 2363 99.2 % 96.9 % 95.4 % 800 2442 2344 2215 2139 96.0 % 90.7 % 87.6 % Fig. 1. A 3D render of the top part of an 18650 cell with a section showcasing its internal structure. A detailed view highlights the components of the electrode stack in a tomographic cross-section. P. Blaˇ zek et al. Journal of Energy Storage 130 (2025) 117436 3 the activity of the contained materials at the anode and at the cathode and their gradual degradation. An anodic peak at 3.42 V and cathodic at 3.38 V is related to lithiation of graphite anode [55]. After 200 cycles, both peaks shifted to a higher voltage, and peaks were less evident. During the cycling, peaks become more evident and, at the same time, change position close to 3.5 V. The magnitude of the anode-related peak after 200, 400, and 600 cycles was similar; however, after 800 cycles, the peak magnitude decreased, indicating a decrease in the activity of the anode material. This drop might be associated with a higher drop in capacity at a high C-rate after 800 cycles. Another very signicant anodic peak can be observed at a potential of about 3.65 V. This peak is related to the transition from a hexagonal to a monoclinic lattice of the NMC532 cathode [56]. Its activity gradually decreases, with the most signicant change can be observed after the rst 200 cycles and then after the last 200 cycles of cycling. The anodic peak around 4.1 V is then associated with high Ni cathode materials such as NMC811 or NCA [57] and represents the H2 to H3 phase transition. The small anodic peak located around 3.8 V represents the transition from the M to the H2 phase. The peak around 4.1 V was relatively stable for the rst 400 cycles, but it dropped signicantly after 600 and 800 cycles. This decrease may then be related to the capacity drop during the subsequent cycling, when the partial regeneration of the battery capacity no longer occurred when cycling started again. Prolonged cycling of the cell caused damage to the inner geometry of the cell. The electrodes are deformed, and the electroactive material is cracked and delaminated from the current collector. These deformations are clearly visible in tomographic cross-sections (Fig. 4) as well as an SEM image of the cross-section through the whole structure (Fig. 5). The deformation is rst visible after 200 cycles on the rst three windings of the electrode stack and is getting more pronounced with cycling. After 800 cycles, they reach up to the seventh winding of electrodes. SEM image (Fig. 5) shows, in addition to electrode waving, cracks, and loss of contact with the current collector, also loss of contact of the anode with the separator and, therefore, with the opposite cathode. The virtual unrolling technique allowed us to visualize the geometry Fig. 2. a) Capacity decreases due to cycling, b) Discharging proles at different C-rate before and after cycling, c) Changes of hysteresis at different c-rates during cycling, d) EIS analysis in different stages of long-term cycling. Fig. 3. dQ/dV analysis for INR18650-29E cell before cycling and after 200, 400, 600 and 800 cycles. P. Blaˇ zek et al. Journal of Energy Storage 130 (2025) 117436 4 of the cathode across the entire volume at different cycling stages. Electrode deformations appear as vertical stripes in the distance-fromcore map. Stripes visible in the fresh dataset (Fig. 6 a) that persist in data from the cycled cell indicate that the electrode deviates from an ideal spiral. Deformities arising in the inner winding due to cycling are also evident in the unrolled data as additional stripes in the Fig. 4. Tomographic cross-sections, detail of the deformation’s development through cycling. Top row: Axial view, bottom row: axial view. Red arrows point to the highest deformation present in the innermost winding, which is the same position as show in Fig. 6. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.) Fig. 5. Detail of cross-section through the whole battery structure. Electrodes are bending towards the central pin, arrows point to loss of contact with separator and current collector. Fig. 6. Visualization of 10 innermost windings of the unrolled cathode as an unrolled distance from the cell core color map. Red arrows point to the highest deformation present in the innermost winding, which is the same position as shown in Fig. 4. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.) Fig. 7. Length of cathode after unrolling. P. Blaˇ zek et al. Journal of Energy Storage 130 (2025) 117436 5 corresponding sections. These deformities extend through the entire height of the electrode spiral and become more pronounced with an increasing number of cycles. A limitation of this unrolled map is the presence of artifacts in the form of sharp borders between horizontal sections. These artifacts stem from the unrolling algorithm in Avizo, which required dividing the data into sections to reduce computational demands and introduced interpolation errors that likely struggled to handle deviations from the ideal spiral. Due to these artifacts, this data cannot be used for quantitative measurements, such as comparing distance from the core between datasets. However, the unrolled data show a cathode length increase of 3.5 mm (0.5 %) over 800 cycles (see Fig. 7), likely resulting from increased bending and deformations. The data shows that the most prominent change occurred again after the rst 200 cycles and after the last 200 cycles, which correlates with the results of electrochemical measurements. The change of hysteresis occurred at a similar time; at the last 200 cycles, there was a signicant decrease in Fig. 8. Cross-section SEM image of fresh battery a) Overview of the cross-section b, c) details of the manufacturing defects. Table 3 Elemental composition of regions 1–4, as shown in Fig. 9 a). Atomic % Element Region 1 (NCA) Region 2 (NMC) Region 3 (NMC) Region 4 (NCA) C 2.9 1.7 1.6 5.7 O 59.1 59.1 56.0 55.6 Mn 0.7 11.2 13.0 0.2 Co 3.0 7.6 7.5 3.2 Ni 33.2 20.1 21.4 34.3 Al 0.9 0.3 0.4 0.8 P 0.2 0.0 0.1 0.2 Fig. 9. Results of EDS analysis of fresh cathode a) Regions of analysis, the elemental composition of each region is shown in Table 3. Figures show the weight percentage of elements from emission line Kb) Mn c) Co d) Ni. P. Blaˇ zek et al. Journal of Energy Storage 130 (2025) 117436 6 capacity at high C-rates. Cross-section images of the fresh cells obtained both by CT (Fig. 4) and SEM (Figs. 5 and 8) show the manufacturing defects in the structure. These are unlled areas within the electroactive layer structure and bends of the current collector. These areas correspond to the ‘voids’ in the CT model. The cause is probably due to the inhomogeneous composition of the cathode paste, where a bubble forms on one side of the current collector. Subsequent pressing leads to its denting and damage of the cathode material particles. Details of these areas are visible in SEM images shown in Fig. 8. Analysis of three pieces of the cathode by synchrotron CT (see Supplemental Material A.1) shows the presence of approximately 9 defects / mm2 of the cathode with a diameter of (72 ±19) μ m and volume (26,000 ±19,000) μ m 3 . There is a high variance in defects shape and size, but no dependency on the position in the battery or inuence due to cycling was discovered. EDS elemental analysis of fresh cell cathode prepared individually revealed that the electroactive material comprises two types of materials with different elemental abundances. As can be seen in the EDS analysis results shown in Table 3 and Fig. 9 below one of the cathode materials is NMC532 (LiNi 0.5 Mn 0.3 Co 0.2 O 2 ) and the other cathode material is most likely NCA (LiNi 0.88 Co 0.1 Al 0.02 O 2 ). In Fig. 9, the materials can be distinguished based on the presence of manganese in the given area - if manganese is present, it is NMC; if manganese is absent, it is NCA. This is consistent with the ndings from electrochemical testing of the battery, where a peak associated with the NCA cathode was observed at dQ/dV spectra. Piece of cathode after 800 cycles was analyzed with microCT and SEM. Same region at the surface of the cathode piece is visualized from both techniques in Fig. 10. The analysis of the cycled cathode crosssection by EDS showed a 3 % decrease in Ni, a 2 % decrease in Mn, a 1 % decrease in Co for NMC grains, and a 4 % decrease in Ni in NCA grains (Table 4) compared with EDS results of pristine cathode (Table 3). On the other hand, the oxygen fraction increased by approximately the same amount. Thus, it is probably not a dissolution of transition metal oxides. It can be conrmed by the fact that they were not present on the anode. The particles appear to be more cracked and fragmented after Fig. 10. Analysis of a piece of cathode after 800 cycles. a) cross-section from submicron CT, b) 3D render of the CT, c) Image of cathode section on from SEM, d) Detail of SEM image (position shown by green arrows) with elemental composition from EDS. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.) Table 4 Elemental composition of regions (shown in Fig. 10 d). Atomic % Element Region 1 (NCA) Region 2 (NCA) Region 3 (NMC) Region 4 (impurity) C 2.6 2.6 2.2 8.6 O 63.2 62.9 64.2 64.7 Al 0.6 0.5 0.3 19.7 Mn 0.6 0.6 10.1 1.3 Co 3.1 3.2 6.6 0.9 Ni 30 30.2 16.6 2.9 S 0 0 0 1.9 Fig. 11. SEM images of cathode cross-section; fresh cell (a); cycled cell (b). P. Blaˇ zek et al. Journal of Energy Storage 130 (2025) 117436 7 cycling, especially the NMC particles, compared with fresh cell (Fig. 11). The closer analysis showed the local presence of barium sulfate at some grain edges for both, pristine and cycled cell. Also, pieces of aluminum are present between grains (impurity in region 4 Fig. 9d). These areas have also increased amounts of sulfur and reach about 2 %. In the anode case, a typical layered structure of graphite grains was observed. The analysis also showed minor amounts of uorine and phosphorus present at the edges of the grains and in the space between them. The origin may be a binder, lithium salt, or a formed SEI layer. The comparison of the anode of the fresh cell and the cycled cell (Fig. 12) shows that the gaps between the graphene layers are more expanded in the cycled cell. At the same time, there are pieces of copper between the grains. Copper is also found in a larger amount on the surface of individual grains as part of the SEI layer (Fig. 13). Higher concentrations of copper were additionally detected in a cross-section through the entire cell structure on the surface of the anode as a light stripe (Fig. 14). This was conrmed by EDS analysis showing a decreasing copper concentration from the surface towards the center of the anode (Fig. 15). The surface of the anode of a fresh cell analyzed in the same way does not show an increased copper concentration. 4. Conclusions This study presents a detailed analysis of degradation in Samsung 18650 cylindrical cells during long-term cycling. While overall capacity fade aligned with datasheet expectations and EIS results, CT imaging revealed signicant internal mechanical degradation—specically delamination and inward bending of electrode layers despite the central pin. Notably, deformation extended to the eight innermost windings after 800 cycles, highlighting the need for improved mechanical design, such as a thicker central pin. A novel nding includes the identication of voids in the cathode structure, with BIB/SEM/EDS analysis linking these regions to increased NMC grain cracking. These insights suggest optimization of the slurry mixture and coating process could mitigate such defects. The primary changes due to aging include the expansion of graphite grains and fracturing of NMC grains around defects. In the cycled material, elevated P and F levels typical of SEI formation and unexpectedly high Cu concentrations (up to 20 %) on the anode and graphite grains. Since sample preparation was ruled out, cycling-related Cu deposition is likely—a phenomenon not fully explained and warranting further study. Literature [26,27] suggests this could involve interactions between Cu ions and SEI layer components. Different electrode composition and optimized cycling protocols may prevent these issues. Additionally, electrode deformation from current collector expansion led to cracking and delamination, increasing hysteresis and reducing capacity due to particle detachment. These changes may compromise performance at high C-rates. Our ndings support predictive modeling for battery health and underscore how design and cycling protocol optimization can mitigate failure modes and extend battery life. Fig. 12. SEM images of anode cross-section; fresh cell (a); cycled cell (b). Fig. 13. EDS mapping of cycled anode cross-section. Fig. 14. SEM images of whole structure cross-section, study of increased amounts of copper on the anode surface; fresh cell (a); cycled cell (b). P. Blaˇ zek et al. Journal of Energy Storage 130 (2025) 117436 8 CRediT authorship contribution statement Pavel Blaˇ zek: Writing – review & editing, Writing – original draft, Methodology, Investigation, Conceptualization. Ondˇ rej Klvaˇ c: Writing – review & editing, Writing – original draft, Methodology, Investigation, Conceptualization. Martin ˇ Sedina: Methodology, Investigation. Ondˇ rej ˇ Cech: Supervision, Methodology, Investigation. Mark´ eta Tkadlecov´ a: Methodology, Investigation. Zuzana Stravov´ a: Investigation, Formal analysis. Tom´ aˇ s Kazda: Writing – review & editing, Validation, Supervision, Data curation, Conceptualization. Tom´ aˇ s Zikmund: Validation, Supervision, Resources, Project administration, Conceptualization. Robert H. Schmitt: Supervision, Resources. Jozef Kaiser: Supervision, Resources, Project administration, Funding acquisition. Declaration of competing interest The authors declare the following nancial interests/personal relationships which may be considered as potential competing interests: Jozef Kaiser reports a relationship with CactuX s.r.o. that includes: board membership and equity or stocks. Tomas Zikmund reports a relationship with CactuX s.r.o. that includes: board membership and equity or stocks. Zuzana Stubianova reports a relationship with CactuX s.r.o that includes: employment. Marketa Tkadlecova reports a relationship with CactuX s.r.o that includes: employment. Pavel Blazek reports a relationship with Baker Hughes Digital Solutions GmbH that includes: employment. If there are other authors, they declare that they have no known competing nancial interests or personal relationships that could have appeared to inuence the work reported in this paper. Acknowledgments We acknowledge CzechNanoLab Research Infrastructure supported by MEYS CR (LM2023051). We acknowledge Elettra Sincrotrone Trieste for providing access to its synchrotron radiation facilities and for nancial support under the IUS internal project, and we thank Lucia Mancini for assistance in using beamline SYRMEP. This work was supported by the project The Energy Conversion and Storage funded as project No 02.01.01/00/22_008/0004617 by Programme Johannes Amos Comenius call Excellent Research. This work was developed in cooperation with Thermo Fisher Scientic Brno. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.est.2025.117436. Data availability Dataset is available at DOI 10.5281/zenodo.14833100 References [1] B. Scrosati, J. Garche, Lithium batteries: status, prospects and future, J. Power Sources 195 (2010) 2419–2430, https://doi.org/10.1016/j.jpowsour.2009.11.048. [2] R. Schmuch, R. Wagner, G. H¨ orpel, T. Placke, M. Winter, Performance and cost of materials for lithium-based rechargeable automotive batteries, Nat. Energy 3 (2018) 267–278, https://doi.org/10.1038/s41560-018-0107-2. [3] D. Deng, Li-ion batteries: basics, progress, and challenges, Energy Sci. Eng. 3 (2015) 385–418. [4] V. Muenzel, A.F. Hollenkamp, A.I. Bhatt, J. de Hoog, M. Brazil, D.A. Thomas, I. Mareels, A comparative testing study of commercial 18650-format lithium-ion battery cells, J. Electrochem. Soc. 162 (2015) A1592, https://doi.org/10.1149/ 2.0721508jes. [5] M. Pin, J. Choi, J.H. Chang, A.S. Schenk, J. Han, S. Wacławek, Y. Kim, J.Y. Cheong, In situ X-ray based analysis of anode materials for lithium-ion batteries: Current status and future implications, Energy Storage Mater. 73 (2024) 103798, https:// doi.org/10.1016/j.ensm.2024.103798. [6] L. Willenberg, P. Dechent, G. Fuchs, M. Teuber, M. Eckert, M. Graff, N. Kürten, D. U. Sauer, E. Figgemeier, The development of jelly roll deformation in 18650 lithium-ion batteries at low state of charge, J. Electrochem. Soc. 167 (2020) 120502, https://doi.org/10.1149/1945-7111/aba96d. [7] T. Waldmann, A. Iturrondobeitia, M. Kasper, N. Ghanbari, F. Aguesse, E. Bekaert, L. Daniel, S. Genies, I.J. Gordon, M.W. L¨ oble, E.D. Vito, M. Wohlfahrt-Mehrens, Review—post-mortem analysis of aged lithium-ion batteries: disassembly methodology and physico-chemical analysis techniques, J. Electrochem. Soc. 163 (2016) A2149, https://doi.org/10.1149/2.1211609jes. [8] Kovachev, Schr¨ ottner, Gstrein, Aiello, Hanzu, Wilkening, Foitzik, Wellm, Sinz, Ellersdorfer, Analytical dissection of an automotive Li-ion pouch cell, Batteries 5 (2019) 67, https://doi.org/10.3390/batteries5040067. [9] S.J. An, J. Li, C. Daniel, D. Mohanty, S. Nagpure, D.L. Wood, The state of understanding of the lithium-ion-battery graphite solid electrolyte interphase (SEI) and its relationship to formation cycling, Carbon 105 (2016) 52–76, https://doi. org/10.1016/j.carbon.2016.04.008. [10] T. Waldmann, M. Wilka, M. Kasper, M. Fleischhammer, M. Wohlfahrt-Mehrens, Temperature dependent ageing mechanisms in Lithium-ion batteries – a postmortem study, J. Power Sources 262 (2014) 129–135, https://doi.org/10.1016/j. jpowsour.2014.03.112. [11] J. Cheng, X. Li, Z. Wang, H. Guo, Mechanism for capacity fading of 18650 cylindrical lithium ion batteries, Trans. Nonferrous Met. Soc. China 27 (2017) 1602–1607, https://doi.org/10.1016/S1003-6326(17)60182-1. [12] M. Ecker, N. Nieto, S. K¨ abitz, J. Schmalstieg, H. Blanke, A. Warnecke, D.U. Sauer, Calendar and cycle life study of Li(NiMnCo)O2-based 18650 lithium-ion batteries, J. Power Sources 248 (2014) 839–851, https://doi.org/10.1016/j. jpowsour.2013.09.143. [13] K. Dai, Z. Wang, G. Ai, H. Zhao, W. Yuan, X. Song, V. Battaglia, C. Sun, K. Wu, G. Liu, The transformation of graphite electrode materials in lithium-ion batteries after cycling, J. Power Sources 298 (2015) 349–354, https://doi.org/10.1016/j. jpowsour.2015.08.055. [14] G. Kwak, J. Park, J. Lee, S. Kim, I. Jung, Effects of anode active materials to the storage-capacity fading on commercial lithium-ion batteries, J. Power Sources 174 (2007) 484–492, https://doi.org/10.1016/j.jpowsour.2007.06.169. [15] H.-J. Kim, T. Krishna, K. Zeb, V. Rajangam, C.V.V.M. Gopi, S. Sambasivam, K.V. G. Raghavendra, I.M. Obaidat, A comprehensive review of Li-ion battery materials and their recycling techniques, Electronics 9 (2020) 1161, https://doi.org/ 10.3390/electronics9071161. [16] T. Waldmann, S. Gorse, T. Samtleben, G. Schneider, V. Knoblauch, M. WohlfahrtMehrens, A mechanical aging mechanism in lithium-ion batteries, J. Electrochem. Soc. 161 (2014) A1742–A1747, https://doi.org/10.1149/2.1001410jes. [17] J.H. Lee, H.M. Lee, S. Ahn, Battery dimensional changes occurring during charge/ discharge cycles—thin rectangular lithium ion and polymer cells, J. Power Sources 119–121 (2003) 833–837, https://doi.org/10.1016/S0378-7753(03)00281-7. [18] M. Ko, S. Chae, J. Cho, Challenges in accommodating volume change of Si anodes for Li-ion batteries, ChemElectroChem 2 (2015) 1645–1651, https://doi.org/ 10.1002/celc.201500254. [19] N. Zhang, H. Tang, Dissecting anode swelling in commercial lithium-ion batteries, J. Power Sources 218 (2012) 52–55, https://doi.org/10.1016/j. jpowsour.2012.06.071. [20] S. Yun, C. Kim, Inuences of thermal volume changes in LiMn2O4 particles on the performance of a Li-ion battery, J. Mech. Sci. Technol. 32 (2018) 3233–3239, https://doi.org/10.1007/s12206-018-0625-7. [21] C. Lin, A. Tang, H. Mu, W. Wang, C. Wang, Aging mechanisms of electrode materials in lithium-ion batteries for electric vehicles, J. Chem. 2015 (2015) 1–11, https://doi.org/10.1155/2015/104673. [22] A.M. Boyce, E. Martínez-Pa˜ neda, A. Wade, Y.S. Zhang, J.J. Bailey, T.M.M. Heenan, D.J.L. Brett, P.R. Shearing, Cracking predictions of lithium-ion battery electrodes by X-ray computed tomography and modelling, J. Power Sources 526 (2022) 231119, https://doi.org/10.1016/j.jpowsour.2022.231119. [23] M. Broussely, Ph. Biensan, F. Bonhomme, Ph. Blanchard, S. Herreyre, K. Nechev, R. J. Staniewicz, Main aging mechanisms in Li ion batteries, J. Power Sources 146 (2005) 90–96, https://doi.org/10.1016/j.jpowsour.2005.03.172. Fig. 15. The concentration of Cu on line proles marked in Fig. 14 in the direction towards copper current collector. Detailed results of elemetnal analysis are in Supplemental material A.2. P. Blaˇ zek et al. Journal of Energy Storage 130 (2025) 117436 9