Data-driven topology optimization of all-solid-state batteries considering conductive additive material informed by microstructure analysis
Naoyuki Ishida, Kozo Furuta, Masashi Kishimoto, Takamitsu Sasaki, Hiroshi Iwai, Kazuhiro Izui, Shinji Nishiwaki
Abstract
Abstract All-solid-state batteries (ASSBs) have emerged as a promising alternative to conventional lithium-ion batteries, offering enhanced safety and efficiency through the utilization of solid electrolytes, which simultaneously improve effective energy density and mitigate ignition risks. However, these batteries exhibit reduced conductivity and electrochemical reaction rates compared to conventional lithium-ion batteries due to insufficient contact area between constituent materials. Previous experimental analyses have demonstrated that battery performance can be enhanced through the implementation of functionally graded materials (FGM), which introduce gradients in the volume fractions of mixed materials within the battery structure. Nevertheless, existing FGM approaches have been constrained to one-dimensional design modifications and empirical trial-and-error methodologies. To address these limitations, this paper presents a novel topology optimization (TO) method for determining the optimal spatial distribution of material volume fractions, specifically electrolytes, active materials, and conductive additives, within the composite anode of ASSBs. To correlate the effective material properties with the material volume fractions, we propose a physicochemically rigorous material interpolation function derived through microstructural analysis and least-square approximation, rather than the conventional Bruggeman model. This methodology enables the precise evaluation of effective properties at the appropriate length scales while accounting for the inherent heterogeneity of composite anodes. We formulate the optimization problem as the minimization of overpotential during discharge processes, with sensitivities derived using continuous Lagrange adjoint methods. Subsequently, we develop a comprehensive optimization algorithm. Through numerical examples, we demonstrate that the optimized material distribution obtained via the proposed method achieves a 6.81% enhancement in capacity compared to conventional uniform composite distributions.