Geometrical Inhomogeneities as Cause of Mechanical Failure in Commercial 18650 Lithium Ion Cells

The mechanical degradation of 18650 lithium ion cells is studied by X-ray computed tomography and is correlated to the electrochemical performance. A method for the geometrical analysis of electrodes by computer tomography is developed and applied to charged and discharged cells. As shown in earlier studies, the geometry of the jelly roll is inhomogeneous leading to mechanical stress during charge/discharge cycles. This effect leads to significant deformations of the jelly roll, which can be analyzed by computed tomography. The detailed analysis reveals that expansion of the anode takes place as expected during charging, but the degree of expansion depends on the position within the battery cell: the largest expansion during charging was found within the area of strongest deformations, whereas other areas without any expansion of the jelly roll were also observed. It is reasoned that the observed inhomogeneous expansion/contraction contribute significantly to cell degradation. The strong expansion within the deformed areas leads to sharp bending of the electrodes resulting in delamination of active layers. On the other hand, the absence of anode expansion reflected by a lack of increase in thickness when charging may indicate pore clogging assuming that the additional volume of graphite with intercalated lithium has to be accommodated within the pore structure. © The Author(s) 2019. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.0551914jes]

Lithium ion battery technology has become a game changer for mobile applications: with the rise of high-energy and high-power battery cells with long lifetime, disruptive innovations like smartphones and full electric vehicles have become reality. Therefore, lithium ion technology, which itself is a multi-billion Euro market, is a key enabler for entire industries addressing even larger markets worldwide. Within a few years, automotive applications have become a dominant application market and development driver of lithium ion battery technology. 1 A major argument for the overwhelming success of the lithium ion technology is its versatility and the ability to custom-design cell performance with respect to power, energy, and lifetime according to the requirements of the application while ensuring a sufficient level of safety and acceptable cost. Performance indicators such as energy and power as well as required level of safety are determined by the technical requirements of the device. However, sufficient lifetime is more crucial for the profitable operation because in many cases an exchange of the battery is too costly, too inconvenient or simply impossible.
Therefore, degradation and reliability are key parameters not only when potential new battery materials are assessed, but a large effort is also still underway to study established lithium ion battery cell chemistries. Cell models describing ageing behavior based on thermal and electric properties have been developed. [2][3][4][5][6] However, despite an improved understanding of underlying mechanisms, the large number of variations in cell form factors, cell balancing, electrolyte additives and manufacturing quality in combination with a wide range of application conditions make it extremely difficult to predict accurately lifetime and reliability ab-initio for a given cell.
The scientific literature dealing with lithium ion batteries considers extensively degradation of materials, electrodes and cells. [7][8][9][10] Most commonly, potential degradation mechanisms are already identified when new materials with improved energy or power density are presented. Nevertheless, these investigations can only give hints under which circumstances these processes are relevant in commercial full cells, because often either only half cell results are presented or the electrode design does not reflect what is later used in commercial cells. Therefore, detailed investigations of degradation in commercially available cells are mandatory for identifying relevant degra-dation mechanisms amongst the multitude of potential degradation paths. It is well established that the loss of lithium inventory due to the reaction of lithium at the anode is one of the major mechanisms for capacity fade. 11,12 Also, the growth of deposition layers on the anode has been identified as failure mode leading to increased internal resistance: this can be either solid electrolyte interphase growth evenly on all particles throughout the electrode or the deposition of a relatively thick layer of electrolyte degradation products on top of the anode, eventually leading to pore clogging resulting in a severe impedance growth and rapid capacity loss. 11 Several other mechanisms have been postulated and described in the literature depending on the chemistry of the active and passive materials as well as other design parameters. 8 Beside electrochemical and chemical side reactions, mechanical processes also contribute to ageing. Diffusion-induced stresses can build up in particles by composition inhomogeneity during mass transfer and may result in faster degradation 13 due to pulverization of the particles. Lithiation causes elastic, inelastic and plastic volume changes of the active layer, 14 which results in in-plane stress buildup or permanent deformation. The range of stress measured by bendingbeam technique [15][16][17] are in the GPa range for Si electrodes, which is also high enough to deform the current collectors, especially on the anode side. 18 These volume effects have been investigated at electrode and material level. [19][20][21][22][23][24][25][26] In contrast to chemistry-related degradation mechanisms and mechanical effects on particle level, macroscopic effects of the geometry of the jelly roll and mechanical and form factors related failure has been considered to a much lesser degree. 27,28 Such mechanisms seem to be of less interest for academic researchers since they seem to pose rather an engineering than a material research challenge. Nevertheless, during the last years a number of studies have been published showing how mechanical issues caused by extensive charge-discharge cycling can play a significant role in degradation and need to be considered in the overall understanding of reliability of lithium ion battery cells. 2,29 Moreover, uncertainties in cell geometry and the resulting impact on mechanical behavior are a significant challenge for engineering of battery packs, since the dimensions of cells, and hence module, can change over the lifetime and mechanical stresses in the components will rise.
In previous work we have shown, using micro X-ray computer tomographic (CT) imaging and post-mortem analysis that charge-discharge cycling even at moderate rates can cause deformations of the jelly roll causing rapid capacity fade. 30 In the current publication we show volume variations as a function of position along the jelly roll, which are interpreted as the underlying reason for deformations leading to cell degradation and failure due to delamination of electrode layers.

Experimental
Battery cell testing.-All data presented in this contribution refers to the high energy 18650 cell type labelled UR18650E manufactured by Sanyo, which has been used in an extensive empirical battery degradation study and for developing ageing models. 3 The gravimetric energy density is 165 Wh/kg given by the nominal capacity of 2.05 Ah at a voltage range between 2.5 and 4.2 V and a nominal voltage of 3.6 V.
The cell contains a graphite anode and a Li(Ni 0.33 Mn 0.33 Co 0.33 )O 2 cathode. In the current investigation two cells from a previous study 31 were examined: a) cell S000b was new and cycled only for conditioning for computed tomography analysis; b) cell S073 was cycled at 1 C rate at 33°C from 5 to 15% SoC (1684 equivalent full cycles amounting to an overall testing time of 377 days, and 1.6 Ah remaining capacity at end of cycling). The cell capacity was defined as the charge capacity (CC from 2.5 V to 4.2V/CV till the current decreased to less than 0.02C). The actual SoC was determined by comparing open circuit voltage (OCV) with the OCV/capacity curve and was set applying a CC/CV charge step until the required SoC was reached. During testing the cells were charged and discharged Ah-based and the mean SoC was adjusted during check-up, 10%, via CC/CV charging. The number of equivalent full cycles was calculated by dividing the overall charge provided by the cell over all cycles by the nominal battery capacity.
For CT evaluation, the cell was charged after finalizing the ageing experiments with a time delay of 4 years. Despite the significant time delay this cell was chosen, because it has been part of a large and comprehensive testing matrix including an in-depth understanding of degradation by modeling, which enables a solid interpretation of the data. The cell had been stored at 5°C and it seems reasonable to assume that the dimensions of the cell and electrodes were not altered during storage when no thermal nor electrochemical stress is applied. A potential impact of calendaric ageing on geometry might be the slow decomposition of electrolyte leading to gas evolution, which might impact the jelly roll geometry by hosting gas bubbles. Nevertheless, CT, electrochemical and post-mortem evaluation did not give an experimental indication of such a failure mode.

Micro X-ray computed tomography (CT).-A
Nanotom S X-ray computed tomography system (GE Sensing & Inspection Technologies, phoenix X-ray, Wunstorf, Germany) was used. For further details on the tomography system and the evaluation software see. 30 An accelerating voltage in the range of 130 to 140 kV was selected in combination with a tungsten target. A voxel size of 32 μm was used for imaging the complete cell (in the following referred to as 'low resolution') and a voxel size of 5 μm for 'high resolution' imaging. 2000 projections were imaged for each dataset for low resolution and 3700 for high resolution using the full pixel resolution of the detector. Integration time for an image was 750 ms and 2 or 3 images were averaged for one projection. For high resolution, the size of the detector was virtually enlarged by a factor of two by moving the detector horizontally by approx. the width of the detector (for each projection two images were subsequently stitched together automatically).

Methods for quantitative evaluation of CT data.-A prerequisite
for assessing structural changes based on comparison of different CT scans of the same sample is the determination of corresponding locations of the structures of interest in the different CT scans. In this study, this means that corresponding locations of structures of interest have to be identified in CT scans of the same cell in charged and discharged state.
Identification of slices for quantitative analysis.-An evaluation method for this purpose was developed and is illustrated on two corresponding cross-sections in Figure 1: inhomogeneities in the cathode layers (see e.g. white circles in bottom images in Figure 1) were used as markers. Several slices with such markers were evaluated with the purpose of determining spatial shifts between scans. Out of these, slices at five different z-positions (z-axis parallel to the axis of rotational symmetry of the cylinder casing of the cell) were selected for quantitative analysis. As deformations were the key target of this investigation, slices were chosen, where a marker was located close to the area of deformation (see also Figure 1) and therefore it can be assumed that not only the marker is visible in the identified corresponding layers, but also corresponding sections of the deformations. Further the z-positions of the selected slices were chosen such that they are approximately evenly distributed over the considered z-height (i.e. the height range covered by the high resolution CT scan).
In each CT scan, slices (perpendicular to the z-axis) are numbered in ascending order with ascending z-coordinate. The position of the cell in the holder can be slightly different from one measurement to the next and the volume to be imaged is selected manually. As a consequence, an offset between slices can be expected when comparing different scans (even of an identical sample). The evaluation showed that the offset in slice number between the two different CT scans (of the cell in as received discharged state and charged state) was identical for all slices (namely 14 slices, e.g. 1426-1412 for the example slice shown in Figure 1). Consequently, the distance between the investigated five slices in vertical direction was identical in discharged and charged state and it is reasonable to assume that no deformation in z-direction occurred in this part of the jelly roll.
Procedure for measuring the thickness of the jelly roll.-Deformations of the jelly roll were observed in the aged cell S073 between the Al current collector and the centre of the cell. In-plane dimensions of the layers cannot be easily determined, therefore a procedure was developed and applied to determine the thickness of the deformed jelly roll in this area (see also Figure 2) and a comparable thickness in a non-deformed part of the jelly roll. The following procedure was applied to CT datasets of the discharged and charged cell: 1. Perform measurement on CT data of discharged cell a. Determine centre of cell cross-section: fit a circle to casing in selected slice and determine centre of the circle, which is also considered the centre of the jelly roll b. Determine diameter of free area in centre of jelly roll: construct a circle with same centre with a radius that equals the minimum distance of centre to first Cu layer; ignore Cu layers without active material (see circle in Figure 2). Use the point 1 (marked in Figure 2) where the circle touches the inner surface of the first Cu layer (with active material) for further construction. c. Determine direction of largest deformation: define straight line through centre and point 1. d. On this line, select the Cu layer, which is closest to the current collector: on inner surface of that Cu layer, define point 2 as the point at the surface on the side toward the centre of the cell (marked in Figure 2). e. Measure thickness of deformed area between current collector and centre: determine the distance between point 1 to point 2 (which corresponds to 8 times a distance from Cu layer to Cu layer and is assumed to be representative for the deformation in the deformed area) f. Measure thickness in non-deformed area: construct a second line, which is rotated by 120°relative to the first line (180°was not considered suitable as deformations at 0°might have an impact on the opposite side). Determine where this second line crosses the inner sides of Cu layers which were used to define point 1 and point 2. Determine the distance between those two points (which should be representative of non-deformed area, not shown in Figure 2). c. Determine direction corresponding to measurements on discharged cell: Use this point and centre of circle to construct a line and continue as described for the discharged cell (1d to 1h) Figure 3).-As complementary measurement, the overall thickness of a Cu layer and two adjacent anode layers (i.e. the thickness of the double coated anode) was determined over the deformed areas in discharged state and after charging. In order to measure thicknesses for layers that show a high expansion as compared to average layer thickness, the following procedure was applied:

Procedure for measuring layer thicknesses (see
1. Layer thickness measurement on discharged cell a. In the cross-section of the discharged cell, the line that exhibits the highest thickness increase of anode layers was chosen and the angle between this line and the line going through point 1 (see 1.b of the previous procedure) was measured as the reference angle (e.g. 13°in Figure 3). This line crosses the Cu layers between centre of the cell and Al current collector. b. For each of the Cu layers, measurements of the overall thickness anode/Cu/anode layer through the point where this line crosses the Cu layer are performed; please note that the thickness measurement is carried out perpendicular to the layers (see Figure 3, top right) 2. Layer thickness measurement on charged cell a. A line is constructed using point 1 and the reference angle determined in 1a. b. Measurements of thicknesses were performed as described in 1b.

Results
Electrochemical performance.-The investigated aged cell (S073) had been part of a larger ageing study 3 and was selected for detailed investigations in this work for a number of reasons. Firstly in the tests the cell showed a discontinuous decrease of capacity during cyclic ageing in combination with a corresponding discontinuous rise of the internal resistance ( Figure 4). Moreover, mechanical issues connected to lithium intercalation are in the center of interest for this work and learning from literature that charge/discharge cycling between 5 and 15% state of charge is connected to a significant change in volume of the graphite anode leading (phase 1L 32 ) to potentially detectable geometric variations, a cell with such a electrochemical history had been selected for in-depth analysis. 30 In the previous investigation, no obvious explanation for the discontinuities in discharge capacity could be identified, but it is clearly related to the varying internal resistances observed in parallel ( Figure 4). Our CT study as outlined below, may give hints that resistance variations are related to mechanical issues of the jelly roll. As observed by others, 2 capacity retention is a function of depthof-discharge (DOD) and state-of-charge (SOC) with higher capacity retentions (up to 5000 equivalent full cycles with 80% remaining capacity) at values of around 50% SOC and 50% DOD (i.e. 25% SOC when discharged and 75% when charged). 31 By comparison the cell of the current study has undergone a rather demanding cycling profile and accordingly shows a rapid capacity fade.

Microstructure and geometry of electrodes -new vs aged cell.-The dimensions and construction details of the investigated
Sanyo18650 cell have been published previously. 30 The main characteristics of the jelly roll arrangement are visible in the cross section image (Figure 2). The cell contains an electrode jelly roll wound around a steel pin, which itself has a groove opened to the centre. The jelly roll itself consists of an anode (dark gray), a separator (not visible) and a cathode (light gray). Furthermore, the current collecting tab of the cathode can be recognized by the dark area left of the center of the jelly roll ( Figure 2). This feature can be denoted as mechanical inhomogeneity, because of the broken rotational symmetry of the jelly roll. Also, most strikingly, a deformation of the jelly roll could be identified in cells, which have been exposed to extensive charge/discharge cycling. 30 A relationship between the charge/discharge history of the cells and the appearance of the deformation has been established, but could not yet be fully explained. 30 A potential explanation for the appearance of mechanical deformation as a function of the charge/discharge-history is the volume variation of electrodes. In order to illuminate the interplay between volume variations and deformation, a detailed high-resolution X-ray computed tomography study of the electrode morphologies and geometries was performed.
High resolution images of parts of the electrode jelly roll are shown in Figure 5. The qualitative impression of the electrode layers of a fresh cell appears very homogeneous, independent of the location of the electrode speaking in favor of a high quality coating and winding process during manufacturing. The light gray areas representing the cathode appear rather solid and only a very thin dark stripe indicating the aluminum current collector interrupts the homogeneous cathode area. In contrast, the images of the jelly roll from a cell, which had been exposed to almost 1700 charge/discharge-cycles ( Figure 5, bottom), exhibit several features indicating structural detoriation of the electrodes: The light gray parts (cathode) of the inner layers (bottom left - Figure 5) appear less compact, more porous and it seems that cathode coating has partly separated from the current collector in the deformed areas. This indicates that cycle ageing does not take place homogeneously across the battery cell, but rather depends on the position within the cell and therefore on the local geometrical restrictions This may have strong implications on ageing models with predictive power and is generally not taken into account. 33 Jelly roll geometry.-In order to quantify geometric variations in an aged cell, a detailed analysis according to Figure 2 was undertaken. The result is plotted in Figure 6 which shows the thickness of jelly roll sections as a function of the cross section height, which ap-pears to affect the general behavior reflecting a vertical geometrical inhomogeneity including strong variations of jelly roll deformations. Analyzing multiple layers of the jelly roll was useful, because the variations in geometry of single layers of the anode might be too small to become visible with the given voxel size of 5 μm (see Figure 7 for a separate analysis of the thickness of each double-coated electrode). The left diagram in Figure 6 depicts the thickness of the 11 layers of the jelly roll found between the cathode tab and the outer can, whereas the right diagram shows the thickness of 8 jelly roll layers located between the cathode tab and the inner pin for the charged and discharged state of the cell. The thickness of the jelly roll sections depends strongly on the position within the cell. The strongest variation in thickness is observed between the cathode tab and the inner pin at the area of the strongest deformation (0 degree, inner 8 layers), followed by the outer layers and with the least variation for the inner 8 layers at 120 degrees and 180 degrees. When charging the cell, the largest expansion of the jelly roll takes place also at the place of strongest deformation, whereas the inner 8 layers at 180 degrees expand very little and at 120 degrees seem not to expand at all within the resolution of our method. A more  homogeneous expansion behavior appears for the section of the outer 11 layers of the jelly roll ( Figure 6 -left).
Since volume expansions in lithium ion batteries during charging are dominated by expansion of the anode, 26 a detailed geometrical analysis of the anode around the deformed area was performed according to Figure 3. The result for a distance of 24.7 mm to the top of the cell casing is plotted in Figure 7.
The strongest expansion was observed for layers 4 and 5, whereas for the other layers almost no change in thickness could be detected. The CT image shown in Figure 7 also reveals clearly visible damage of the cathode coating between the anode layers 5 and 4 as well as 4 and 3. The delamination of the cathode coating was confirmed by visual inspection during post-mortem analysis, which indicated a major mechanical stress at these positions. This is caused by bending of the electrode and formation of a sharp edge resulting in delamination of the cathode coating. This effect may be compared to winding a coated electrode with a small radius -a common test for adhesion strength when producing electrodes. From the CT images only the damage of the cathode becomes visible due to the low contrast of the graphitic anode, but it is reasonable to assume that delamination of the anode coating happens as well by the same cause. Post-mortem analysis observations support this assumption: As published before, the delamination of the cathode relating to the deformation was seen in post-mortem inspection. On the other hand, as the entire anode coating was delaminated no correlation to the CT images could be seen anymore after separating the jelly roll layers. 28 In summary, geometrical variations as a function of the state of charge are highly inhomogeneous with respect to the location within the jelly roll. The most significant effects are found in the areas close to clearly visible jelly roll deformations, whereas there are other areas where no jelly roll expansion was visible in the charged cell.

Discussion
Anode expansion.-The expansion of carbon-based anode materials has been measured, investigated and discussed in a number of publications and can be accepted as valid explanation for our findings of variations in the dimensions of the anode and jelly roll during charging/discharging of the cell. 33 The intercalation of lithium into graphitic layers itself (C6 + Li → C6Li) is accompanied by a volume increase of slightly more than 10% together with several phase transitions. [30][31][32] Obviously, the expansion of randomly distributed graphitic particles leads to expansion in all directions with growth into the porous structure and the thickness of the electrode. Accordingly, dilatometry revealed reversible expansions of a single anode electrode by about 5% in height. 26 The dimensional change of the cathode with the degree of lithiation is still under debate in the literature. Some authors even claim an expansion of the cathode material with de-intercalation of lithium. 35 Overall, the cathode volume changes are significantly smaller than for the anode and therefore we focus our interpretation on the geometrical changes observed at the anode. 22,[36][37][38] In addition, it is also reported that in-plane stress in few MPa range in graphite electrode may build up. 39 If this in-plane stress is not compensated by compressive forces (e.g. from adjacent layers), it may also bend the layer especially in those areas which inherited a non-homogeneous mechanical stress distribution from manufacturing. An example for this is the cathode tab, which interrupts the even round shape of the jelly roll with the according tension and pressure due to the thicker layer in case the cathode tab is thicker than the other parts of the cathode. The enhanced mechanical deformation may accelerate during cycling and may create more uneven stress distribution in the cell.
Based on literature data indicating a 5% thickness increase for anode electrodes without geometrical restrictions the anode thickness change within an 18650 cell should be visible either by analyzing the geometry of the jelly roll or the single layer anode. With this hypothetical increase -neglecting geometrical restrictions that could limit expansion and lead to stress buildup -a value of about 5 μm would be expected, because the average thickness of the discharged anode layers in areas without deformation is approximately 90 μm as determined by CT imaging. However, the accuracy of the applied method of our study is rather at the limit when looking at 5 μm variations of single layer electrode dimensions. Therefore cumulative thickness variations over several jelly roll layers have been evaluated as described and plotted above ( Figure 6) and a dependency of position was found, which is further illustrated in Figure 8 (right): The light blue section of the jelly roll (11 layers) appears to expand and contract rather homogeneously. A maximum variation was found for the area indicated by the white elipse whereas almost no expansion was found after charging in the gray area right of the inner pin. It is a very interesting result that the charge-induced expansion of the jelly roll significantly varies throughout the cell. This should be taken into account when discussing degradation mechanisms and moreover when attempting to model and predict the lifetime of a cell. We suggest that when modeling, the battery cell could be divided into sections of similar geometrical characteristics and that the overall model should be the sum of the sections. Nevertheless, it must also be considered that much localized degradation features (e.g. damaged separators) can have a big impact resulting from inhomogeneous mechanical properties and are not well described by models. Similar considerations regarding inhomogeneities within cells have also been discussed by other authors. 27,40 Though detecting such inhomogeneous mechanical stress is a challenging task, we find in our case that it leads to detectable variations in jelly roll expansion and eventually to an inhomogeneous jelly roll deformation. Both might not be visible in the early period of A few consequences with respect to ageing mechanisms are rather obvious from our results: First of all, the mechanical stress throughout the cell is uneven due to an asymmetric packing of the jelly roll. Eventually, this causes serious jelly roll deformation, which leads to delamination of the electrode coatings as shown by X-ray computed tomography analysis and confirmed by post-mortem analysis. It is speculated that it may also cause somewhat reversible "micro" short circuits between anode and cathode by damaging the separator and sharp bends of the electrodes in combination with lithium plating leading to significantly lower impedance, but not to catastrophic failure. The plated lithium would then be dissolved during continuous charge/discharge cycling, which then would lead to increased impedance as observed. A direct observation of lithium-induced shorting by CT imaging is impossible due to the low contrast of lithium. Nevertheless, recovery of capacity by redistribution of lithium was described in literature [24][25][26] and will be a matter of future investigations. Furthermore, synchrotron X-ray tomography has shown that a structural collapse of the jelly roll however induced can lead to shorts and catastrophic failure. 27 Our investigation indicates that such structural collapse can be caused by charge/discharge-cycling and not only by massive gassing or other mechanical macroscopic damage.
Secondly, such expansion/contraction patterns are either an indication of an inhomogeneous charge distribution throughout the jelly roll, or unequal pore distribution throughout the cell, meaning that the graphite expands into the pores. Such an effect would cause an inhomogeneous in-plane current distribution especially at higher C-rates when the diffusion of the ions in the liquid phase? becomes a rate limiting step. This would become more serious in an aged cell, because pore clogging by SEI and layer growth by electrolyte decomposition limit the mobility of the ions. 11 Reducing locally the porosity and the mobility of ions eventually leads to local overcharging or overdischarging of the anode with the corresponding degradation features including Cu dissolution and plating respectively, 41 which can lead to potentially catastrophic failure challenging lifetime and safety of cells. This is obviously most severe when applying high currents so that the power capability of the cell fades with lifetime.

Conclusions
A detailed geometrical analysis of an 18650 lithium ion cell was performed by micro X-ray computed tomographic analysis. Overall, our investigation showed strong geometrical inhomogeneities throughout the cell. Most strikingly, the position of the cathode tab interrupts the rotational symmetry of the jelly roll. When ageing the cell by charging and discharging, the inhomogeneities of the starting geometry lead to macroscopic features like the jelly roll deformation between the cathode tab and the inner pin, which eventually results in delamination of active material in the most bent parts of the jelly roll and potentially to localized short circuits by mechanical damages.
The analysis of the variation of jelly roll thickness as a function of state-of-charge and position revealed significant variations of the expansion/contraction behavior throughout the cell with the according consequence of an uneven current distribution. This has obvious implications on lifetime and safety. Such findings must be taken into account for modeling lithium ion battery cells with the objective to predict lifetime and safety.
Another mechanism, which becomes apparent due to the impeded growth of anode layers in most areas of the jelly roll is pore clogging leading to rapid loss of capacity when areas of the electrode are not available for electrolyte anymore. 11 Overall, our study points out that degradation does not happen homogeneously throughout the cell, which is generally not taken into account in ageing models. Interpreting mechanical inhomogeneities, hence pressure, as source for accelerated capacity fade is supported by investigations of inhomogeneous external mechanical pressure leading to lithium plating. 8,12,15 On the other hand, homogeneous mechanical pressure can indeed also be beneficial to the cells lifetime. This applies in particular to new cell chemistries with e.g. Silicon, which has an significantly larger expansion during charging than graphite. 42 It can be concluded that a well-balanced and even mechanical pressure distribution is mandatory for a maximized cell lifetime. This has to be seen complementary to the chemical and thermal design of a lithium ion battery cell.