Virtual design of multilayered SOFC microstructures based on particle packing and spherical harmonics

Abstract Realistic three-dimensional reconstructions of solid oxide fuel cell (SOFC) microstructures are essential for understanding structure-property relationships and optimizing performance. This work presents a physics-informed, stochastic microstructure generator that extends existing particle-packing and spherical harmonics-based methods designed for lithium-ion battery electrodes to the more complex, multilayered structures of SOFCs. The approach integrates a discrete element method for realistic sphere packing, an enhanced neighbor graph for inter-particle connectivity, and a spherical harmonics-based shape generation constrained by contact points. The resulting structures are represented using a multiphase-field method and calibrated through a multi-objective Bayesian optimization framework against morphological metrics derived from focused ion beam–scanning electron microscopy tomography, including volume fraction, tortuosity, specific surface area, and equivalent radius distributions. The method accurately reproduces both porous and dense SOFC layers, as well as complete multilayer structures, achieving low deviations from experimental data and strong visual agreement. This generator enables the efficient and systematic generation of virtual SOFC microstructures with controllable morphological properties and multiphase-field representations suitable for subsequent morphology-resolved simulations. It therefore provides a versatile framework for microstructure design and for preparing calibrated virtual structures for further electrochemical, electro-chemo-mechanical, or degradation studies.