Simulation of Local Electrochemistry in Solid Oxide Fuel Cell Microstructures by High Performance Computation

Recent progress in micro-scale three-dimensional (3D) characterization techniques such as focused ion beam–scanning electron microscopy (FIBSEM) and X-ray nanotomography has brought unprecedented opportunities
in linking material microstructure to performance and properties. The link between average electrode microstructure and average performance in research-grade solid oxide fuel cells (SOFCs) can be established using effective medium theory, or continuum modeling, where effective performance parameters are calculated based on effective microstructural properties. However,
microstructural degradation and failure in commercial-grade SOFCs have not been adequately captured using mean parameters from effective medium theory. This work aims to quantify distributions of electrochemical parameters throughout the microstructure in heterogeneous electrodes
to establish links between microstructure and degradation, failure, and performance. In the initial stages of this work, a novel high-throughput computational methodology was developed and implemented, on which the remaining goals
could be carried out. Using an open-source finite element framework (MOOSE, Idaho National Laboratory) on high performance computing platforms (Joule, National Energy Technology Laboratory, and Bridges, Pittsburgh Supercomputing Center), a numerical transport-and-reaction model was constructed and applied to morphology-preserving microstructural meshes. The model computes local distributions of electrochemical parameters that are coupled to morphological features such as interfaces and triple phase boundaries. Using a scriptable commercial meshing software (Simpleware ScanIP), the
microstructural meshes were obtained from the plasma-sourced FIB-SEM characterization of a commercial SOFC electrode, as well as synthetic microstructures generated using DREAM.3D that model specific types of heterogeneities known to exist within SOFC commercial cathodes. Results using the commercial microstructure are shown and discussed, which demonstrate the capabilities of the computational workflow and remove elements of infeasibilty for the future work. A series of computational investigations, on both physical and synthetic
microstructures, are described to correlate local microstructural features to local electrochemical performance distributions. Comparisons to conventional
modeling approaches (effective medium theory, or continuum modeling) bring forth additional insight/observation capabilities unique to the
novel methodology of this work. Furthermore, a synthetic infiltration algorithm was developed to model how nanoscale catalysts dispersed on the internal pore walls affect performance, to model a well-known experimental
practice. The results shed light into the design and fabrication of optimal electrodes in fuel cells.
By combining physical and synthetic microstructures in simulations, specific microstructrual traits can be examined for the distributions in nano-, micro-, and meso-scale performance characteristics, providing a new framework
to include in degradation model and enable more comprehensive statistical analyses/evaluations for the future work.