Revolutionizing material design for protonic ceramic fuel cells: Bridging the limitations of conventional experimental screening and machine learning methods

The commercial viability of protonic ceramic fuel cells (PCFCs) is contingent upon developing highly active and stable cathode materials. The conventional trial-and-error process is time-consuming and costly for cathode material development, while the availability of sufficient and reliable datasets limits the recently emerging machine learning (ML) method. Here, we propose a novel approach based on the experimental design paradigm (EDP) to efficiently facilitate PCFC cathode materials’ development with a minimal dataset. As a rigorous systematic statistical approach, we employ the EDP for strategic variation of multiple elements and measure their effect on desired performance characteristics. We generate empirical models that reveal the optimal concentrations and interactions of the elemental composition and performance characteristics. In this study, we select the BaCo_αCe_βFe_γY_ζO_3-δ series as a proof-of-concept, and the optimal composition, BaCo_0.667Ce_0.167Fe_0.083Y_0.083O_3-δ, was promptly determined—guided by the EDP—using only 16 independent conditions and 32 randomized experimental runs. We further demonstrate the EDP's versatility by optimizing the widely-used and high-performing Ba_0.5Sr_0.5Co_0.8Fe_0.2O_3-δ cathode material for solid oxide fuel cells. Our results highlight the potential of the EDP for effectively designing superior materials for solid-state electrochemical power generation systems, offering a reliable and practical alternative to conventional trial-and-error screening and ML methods.