Prediction of Microstructural Degradation in SOEC Oxygen Electrodes with COMSOL
Solid oxide electrolysis cells (SOECs) are a promising pathway for green hydrogen production, but their commercial viability is limited by oxygen-electrode delamination that accelerates performance degradation. This work presents a microstructure-resolved electro-chemo-mechanical model implemented in COMSOL Multiphysics to elucidate degradation mechanisms within the SOEC oxygen electrode (OE). Mixed electronic and ionic conductor (MIEC) microstructures are synthetically generated with DREAM.3D based on particle size distribution (PSD) and volume fractions. These synthetic images are processed and meshed in MATLAB using iso2mesh tool (>1.8 million tetrahedra) and imported into COMSOL with automated tagging of two-phase (2PB: LSCF/GDC, LSCF/air) and three-phase boundaries (3PB: LSCF/GDC/air) using internally developed MATLAB code. Charge transport in LSCF phases is modeled using Poisson-Nernst-Planck (vacancies and holes), while GDC and YSZ are modeled using Poisson equation for ionic conduction. Interface kinetics at 2PB and 3PB follow Butler-Volmer expressions, and oxygen evolution at LSCF/air uses a Chang-Jaffé boundary. The governing equations are solved using general PDE form in the mathematics module. For further investigation, physics-based EIS under DC bias is modeled to link impedance arcs to physical processes, such as diffusion, reaction kinetics at 3PBs (low frequency) and charge transfer at 2PBs (mid/high frequency). Chemical induced stresses are computed using Structural Mechanics module in COMSOL Multiphysics, providing a one-way coupled model to predict delamination risk due to chemical stress. We benchmark a baseline MIEC electrode with lanthanum cobalt iron oxide (LSCF) and gadolinium doped ceria (GDC) electrode against a novel bilayer design featuring a dense nanometric tantalum doped strontium cobalt oxide (SCT) coating. Previous studies[1,2] indicate that this nanolayer coating enhances kinetic at the LSCF/Gas interface and reduces the activity of LSCF/GDC/gas (3PB) interface. The models for both electrode designs were validated by fine tuning process parameters through COMSOL-MATLAB interface against experimental data at 700°C, showing <10% overpotential deviation. At 1 A/cm2, the bilayer design lowers the cell voltage by ≈160 mV while maintaining near-equilibrium vacancy concentrations at the OE/electrolyte (EL) interface. This reduces the ionic potential buildup and shifts the majority of the current from 3PB to 2PB pathways. The maximum principal stress at the interface drops from ≈378 MPa (baseline) to ≈89 MPa (bilayer), a ~77% reduction, indicating lower delamination propensity due to chemical induced stress. Sensitivity analyses of the EIS highlight the dominant influence of 2PB exchange current density on polarization resistance, the distinct 3PB imprint on low-frequency arcs, and the role of 2PB capacitance in positioning the arcs. Crack simulations (25–75% interfacial loss) reproduce concurrent increases in ohmic and polarization resistances and a redistribution of total current toward 3PB currents. This workflow DREAM.3D-to-COMSOL microstructure import with automated boundary mapping, simultaneous steady/perturbed solution, and stress post-processing-offers a practical, extensible route to link microstructural design, impedance fingerprints, and mechanically driven failure in SOEC oxygen electrodes.
