Optimizing non-uniform pt distribution in the cathode catalyst layer for high-performance PEM fuel cells

Abstract

This study presents a multi-objective optimization framework to enhance proton exchange membrane (PEM) fuel cell performance through spatially tailored platinum (Pt) loading in the cathode catalyst layer (CL). A multilayer perceptron (MLP) surrogate model, trained on 3D simulation data, demonstrated high predictive accuracy (R² = 0.99, RMSE < 10 mV), enabling efficient optimization via particle swarm optimization (PSO). The optimized through-plane Pt gradients increased cell voltage by up to 8.3 mV at 1.5 A/cm² by reducing ohmic losses within the cathode CL. Electrolyte potential and current density analyses confirmed that Pt gradients primarily affect proton transport resistance. Multi-objective optimization using the NSGA-II algorithm revealed a clear trade-off between cell performance and stack cost, with gradient Pt loading designs providing greater advantages at higher total loadings. Notably, uniform Pt loading designs closely approached the Pareto front at lower Pt loadings (<0.2 mg/cm²), emphasizing the importance of controlled Pt distribution for cost-effective CL fabrication. These findings offer valuable design guidelines for next-generation PEM fuel cells.