Time-dependent deep learning manufacturing process model for battery electrode microstructure prediction

Alejandro A. Franco,Diego E. Galvez-Aranda,Tan Le Dinh,Utkarsh Vijay,Franco M. Zanotto

DOI: https://doi.org/10.26434/chemrxiv-2024-kkn30

2024-02-05

Abstract:The manufacturing process of Lithium-ion battery electrodes directly affects the practical properties of the cells, such as their performance, durability, and safety. While computational physics-based modeling has been proved as a useful method to produce insights on the manufacturing properties interdependencies as well as the formation of electrode microstructures, their high computational costs avoid their direct utilization in electrode optimization loops. In this work, we report a novel time-dependent deep learning (DL) model of battery electrodes manufacturing process, demonstrated for calendering of NMC111 electrodes, and trained with time-series data arising from physics-based Discrete Element Method (DEM) simulations. The DL model predictions are validated by comparing evaluation metrics (e.g. MSE and R2 score) and electrode functional metrics (contact surface area, porosity, diffusivity and tortuosity factor), showing very good accuracy with respect to the DEM simulations. Our DL model can remarkably capture the elastic recovery of the electrode upon compression (spring-back phenomenon) and the main 3D electrode microstructure features without using the functional descriptors for its training. Furthermore, our DL model has a significantly less computational cost that the DEM simulations, paving the way towards quasi-real time optimization loops of the 3D electrode architecture predicting the calendering conditions to adopt in order to obtain the desired electrode performance.

Chemistry

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve

This paper aims to address the issue of microstructure prediction during the manufacturing process of lithium-ion battery electrodes. Specifically, the authors propose a deep learning (DL) model based on time series data to simulate and optimize the microstructural changes of lithium-ion battery electrodes during the pressing process.

#### Main Research Content:
1. **Problem Background**:
- The manufacturing process of lithium-ion battery electrodes directly affects the actual performance, durability, and safety of the battery.
- Current physics-based computational models, while providing a deep understanding of the manufacturing process characteristics, have high computational costs that limit their direct application in electrode optimization.

2. **Solution**:
- A new deep learning model based on time series data is proposed to simulate the manufacturing process of lithium-ion battery electrodes.
- The model is trained using discrete element method (DEM) simulation data and applied to the pressing process of NMC111 electrodes.
- Model validation is performed by comparing evaluation metrics (such as mean squared error MSE and R² score) and electrode functional metrics (contact surface area, porosity, diffusion coefficient, and tortuosity factor).

3. **Model Features**:
- The model can capture the elastic recovery phenomenon (springback effect) of the electrode during compression and predict the main 3D electrode microstructural features without using functional descriptors.
- Compared to traditional DEM simulations, this model has significantly lower computational costs, enabling near-real-time electrode architecture optimization cycles.

4. **Objective**:
- Develop an efficient deep learning model to predict and optimize microstructural changes during the electrode manufacturing process, particularly to achieve the desired electrode performance under pressing conditions.

Through this work, the authors hope to provide an efficient and accurate tool for the lithium-ion battery electrode manufacturing process to optimize electrode microstructure and enhance the overall performance of the battery.