Artificial neural network approach for multiphase segmentation of battery electrode nano-CT images

Abstract

The segmentation of tomographic images of the battery electrode is a crucial processing step, which will have an additional impact on the results of material characterization and electrochemical simulation. However, manually labeling X-ray CT images (XCT) is time-consuming, and these XCT images are generally difficult to segment with histographical methods. We propose a deep learning approach with an asymmetrical depth encode-decoder convolutional neural network (CNN) for real-world battery material datasets. This network achieves high accuracy while requiring small amounts of labeled data and predicts a volume of billions voxel within few minutes. While applying supervised machine learning for segmenting real-world data, the ground truth is often absent. The results of segmentation are usually qualitatively justified by visual judgement. We try to unravel this fuzzy definition of segmentation quality by identifying the uncertainty due to the human bias diluted in the training data. Further CNN trainings using synthetic data show quantitative impact of such uncertainty on the determination of material’s properties. Nano-XCT datasets of various battery materials have been successfully segmented by training this neural network from scratch. We will also show that applying the transfer learning, which consists of reusing a well-trained network, can improve the accuracy of a similar dataset.

Data Specifications for Battery Manufacturing Digitalization: Current Status, Challenges, and Opportunities

“…However, poor accuracy and measurability can arise from differentiating materials with similar chemical or morphological environments. [119,120] The peel test measured the electrode adhesion strength offline, separating the coating from the current collector using tapes. [121] This test provides comparative information between electrodes produced in-house.…”

“…These techniques have demonstrated the ability to distinguish between the binder distribution, conductive additive, and AM. However, poor accuracy and measurability can arise from differentiating materials with similar chemical or morphological environments [119,120] . The peel test measured the electrode adhesion strength offline, separating the coating from the current collector using tapes [121] .…”

Multiscale Dense U-Net: A Fast Correction Method for Thermal Drift Artifacts in Laboratory NanoCT Scans of Semi-Conductor Chips

“…NanoCT has been widely used in chemical, biomedicine, and industrial inspection [ 1 , 2 , 3 , 4 ]. Researchers study the internal structure of objects non-destructively with the results of 3D reconstruction.…”

About the Consideration of the Inactive Materials and the Meshing Procedures in Computational Models of Lithium Ion Battery Electrodes

“…This approach is still used today in the state-of-the-art models but also in Computer Tomography (CT) characterization of the electrodes because it is challenging to spatially resolve the carbon particles and the binder locations. [6,7] In this review, the history of the LIB 3D-resolved continuum modeling field is explored through the prism of the consideration of the CBD. The CBD was either implicitly considered, meaning it was merged with either the electrolyte or the AM, and it did not have its own properties, or either explicitly considered, meaning the phase was resolved in three dimensions and had its own properties.…”

“…Later, a new generation of 3D models emerged where the carbon additive and polymeric binder phases were represented as a single phase usually called carbon binder domain (CBD). This approach is still used today in the state‐of‐the‐art models but also in Computer Tomography (CT) characterization of the electrodes because it is challenging to spatially resolve the carbon particles and the binder locations [6,7] . In this review, the history of the LIB 3D‐resolved continuum modeling field is explored through the prism of the consideration of the CBD.…”