Session: 12-03-03: Data-Enabled Predictive Modeling, Scientific Machine Learning, and Uncertainty Quantification in Computational Mechanics

Paper Number: 113404

113404 - Multiphysics-Informed Machine Learning for Mechanical-Induced Degradation of Silicon Anode 

Li ion batteries (LIBs) have a wide range of applications in today’s world. These batteries are used in the areas of energy storage such as in car batteries, mobile phones, tablets, laptops etc. Hence, there is a constant need to continuously improve the existing LIBs and to improve their designs. One such area of design improvement is the changing of the anode material from graphite as the existing battery designs are nearing their performance limits and are failing to meet the demand for longer battery life and better energy cycles along with improved safety.

Employing a silicon (Si) based anode material instead of graphite helps in mitigating some of these issues. This Si anode material helps in increasing the specific capacity of the battery due to the difference in the intercalation mechanism of the Si for Li-ion storage (Li-ion diffusion into the interstitial sites of the host lattice). In this mechanism Si atoms react with lithium, which leads to the bonds between the Si atoms giving way for the formation of  alloy. Since there are no constraints from the atomic framework of the host material, Si is able to store much more Li as compared to other electrodes leading to dramatically increased specific capacities.

However, the use of Si anode in LIBs comes with its own set of issues. One of the major issue is that the alloying lithium storage mechanism leads to substantial increase in the volume of the Si during the lithiation/delithiation cycles. This further leads to the evolution of internal stresses within the anode that results in the development of cracks within the anode and the delamination of the anode from the metal substrate, which ultimately causes an overall decrease in the capacity of the battery.

The second major capacity fade mechanism in LIBs is the growth of the solid-electrolyte interface (SEI). SEI layer growth in LIBs can lead to the increase in the internal resistance of the cell and the removal of the active Li material from the cycling process. The SEI layer growth, between the anode and the electrolyte, mainly occur during the battery cycling phase. This layer starts off as a protective barrier between the two phases, through the reduction in the side reactions on the surface of the anode, by allowing the transfer of Li while also maintaining a physical barrier between them. However, ultimately the overall capacity of the battery is reduced as the growth of this layer increases the resistance and the removal of active lithium from the cycling system.

In this study, multi-physics based FE models are used along with Gaussian Process (GP) surrogate models to design Si anode based Li ion batteries. First, two separate FE models are built to study the solid mechanics and the electrochemical part of the battery cycling and lithiation/delithiation process. The first model studies the volumetric change induced cracking and delamination in the Si anode whereas the second model utilizes the output of the first model to simulate the SEI growth in the areas of delamination and cracking (due to seepage of electrolyte in these areas) to determine the resultant capacity loss for the battery. The developed FE model will be used to investigate the influence of Si anode thicknesses, working temperatures and charging rates. The outputs from the FE models will be used to train the GP surrogate models for the design of LIBs towards application-oriented properties such as high energy storage, fast charging or optimal life time.

Presenting Author: Parth Bansal University of Illinois at Urbana Champaign

Presenting Author Biography: Parth is a Ph.D candidate at UIUC.

Authors:

Parth Bansal University of Illinois at Urbana Champaign
Yumeng Li University of Illinois at Urbana-Champaign