Performance of lithium-ion batteries (LIBs) is strongly influenced by the porous microstructure of their electrodes. In this work, the microstructure of a non-calendared LiNi_1/3Mn_1/3Co_1/3O₂ (NMC) electrode has been investigated in order to extract relevant properties useful for battery modeling. At first, the image level of detail has been evaluated (using original criteria based upon edge detection) to assess the overall data quality available for the study. Then, transport properties (volume fraction, connectivity, particle size and tortuosity) and electrochemical property (specific surface area) have been calculated for the pore and the active material. Figure 1 shows concentration field calculated within the pore (used to determine the effective diffusion coefficient) and the active material particle size.

Special attention has been paid to determine (i) the size of the so-called representative volume element (RVE) required to be statistically representative of the heterogeneous medium and (ii) the voxel size dependence for each properties. Both analyses are essential since they allow checking the relevance of the calculated values, before actually using them in macroscopic models. It has been found that the investigated properties exhibit different RVE sizes and image resolution dependences. Several properties have been calculated using a panel of different numerical methods in order to compare their results. Indeed, some of these methods rely on different assumptions and exhibit intrinsic bias, leading as a consequence to different outcomes as established in this work.

Properties have been calculated on the whole domain and have been then plotted along the electrode thickness to evaluate their overall heterogeneity and assess the presence of edge effects, if any. It is then possible to introduce subsequently these spatially-dependent properties into 1D macroscopic battery model in order to take into consideration the heterogeneity, between each slice, of the electrode material while retaining the advantage of an inexpensive CPU time macroscopic model. This approach is particularly relevant for graded electrode materials that exhibit microstructure gradients. Edge effects, as well as property gradients, are shown for the investigated electrode material.

An additional microstructure analysis has been performed on a calendared NMC electrode. Indeed, in industrial LIB electrode manufacturing, the cast electrode undergoes a compression step (calendaring) in order to improve particle connectivity and reduce the electrode’s porosity which increases battery capacity. The effect of this compression step on the material properties have been quantified in this work and significant microstructure properties change are shown.