Image Based Modeling of Li-ion Battery Electrodes

Overview

The heterogeneous microstructure of lithium ion batteries (LIBs) remarkably influences performance by providing certain interfacial surface area, diffusion path, and active material connectivity. In this work, the LiFePO4 (LFP) electrode microstructure of a LIB is reconstructed based on the nano-XCT images in Simpleware software. The 3D model was then meshed and exported to COMSOL Multiphysics® for Finite Element Analysis (FEA). The results show that the electrode heterogeneity causes a wider range of physical and electrochemical properties compared to common homogenous models.

Characteristics:

  • Model based on the real 3D microstructure data
  • Sample of a commercial LFP/graphite cell scanned using nano-XCT (5 μm each side)
  • Image data quickly segmented in Simpleware ScanIP using thresholding
  • Mesh generated with Simpleware FE and exported to COMSOL Multiphysics®
  • LFP performance more accurately predicted than with traditional homogenous 1D model

Acknowledgements:

This study was financially supported by the University of Akron and the Natural Sciences and Engineering Research Council of Canada (NSERC) through grants to Z.C. and the University of Waterloo.

Thanks to:

Electrode Structure Reconstruction

An LiFePO4 (LFP) sample from a commercial LFP/graphite cell which was disassembled and scanned using nano-XCT. The obtained 2D stack was imported in Simpleware ScanIP and segmented utilizing thresholding technique to convert the greyscale stack to a binary stack. The process included segmenting the active material particles and pore-PVDF-carbon regions from the scan. If the weight percentage of the active material is high, the carbon material and polymer binder are randomly distributed in the electrode. To reconstruct the connected solid matrix, it was assumed that the carbon material is randomly distributed among the active material to provide electronic connectivity. A morphological close filter was used in Simpleware ScanIP on the active material region to fuse the neighboring active material together.

Reconstruced nano-XCT image data in Simpleware ScanIP
Reconstruced nano-XCT image data in Simpleware ScanIP

Reconstructed nano-XCT image data in Simpleware ScanIP

A mesh of the model was then generated using Simpleware FE and exported directly to COMSOL Multiphysics® for solving governing partial differential equations related to developed LIB multiscale model. In microscale, the model is based on the real 3D microstructure data, taking advantage of the traditional homogenous 1D model in macroscale to characterize discharge/charge performance. This framework was used for the Multiscale-Multiphysics study of LIB.

Simulation-ready volume mesh in Simpleware FE
Simulation-ready volume mesh in Simpleware FE

Simulation-ready volume mesh generated in Simpleware FE and exported directly to COMSOL Multiphysics®

Computer Simulation Results

It is shown that this model can predict the experimental performance of LiFePO4 cathode at different discharge rates more accurately than the conventional homogenous models. The simulation results could predict the experimental discharge voltage of LFP cathodes at different rates. The simulation showed that the lithium ion concentration in the electrode active material structure is much higher in the region with smaller cross-section area perpendicular to the lithium intercalation pathway. Such low area regions would intercalate ca. 10 times higher than the area with an average concentration. The approach used in this study can provide valuable insight into the spatial distribution of lithium ions inside the microstructure of LIB electrodes. The inhomogenous microstructure of LFP causes a wide range of physical and electrochemical properties compared to the homogenous model.

Distribution of lithium concentration inside electrode microstructure

Distribution of lithium concentration (mol m-3) inside the electrode microstructure during discharge at c-rate = 1 for different SOCs (3D electrode microstructure represents geometry in microscale and 1D x-coordinate describe geometry in macroscale along the electrode thickness direction).
(Reproduced with permission from Journal of Power sources, doi:10.1016/j.jpowsour.2015.12.134)

Distribution of the overpotential (unit:V) on the solid/electrolyte interface during discharge at c-rate = 1 for different SOCs.
(Reproduced with permission from Journal of Power sources, doi:10.1016/j.jpowsour.2015.12.134)