Effectively Increasing Pt Utilization Efficiency of the Membrane Electrode Assembly in Proton Exchange Membrane Fuel Cells Through Multiparameter Optimization Guided by Machine Learning

Rui Ding,Wenjuan Yin,Gang Cheng,Yawen Chen,Jiankang Wang,Xuebin Wang,Min Han,Tianren Zhang,Yinliang Cao,Haimin Zhao,Shengping Wang,Jia Li,Jianguo Liu

DOI: https://doi.org/10.1021/acsami.1c23221

IF: 9.5

2022-01-01

ACS Applied Materials & Interfaces

Abstract:Although proton exchange membrane fuel cells have received attention, the high costs associated with Pt-based catalysts in membrane electrode assemblies (MEAs) remain a huge obstacle for large-scale applications. To solve this urgent problem, the utilization efficiency of Pt in MEAs must be increased. Facing numerous interacting parameters in an attempt to keep experimental costs as low as possible, we innovatively introduce machine learning (ML) to achieve this goal. Nine different ML algorithms are trained on the experimental dataset from our laboratory to precisely predict the performance and Pt utilization (maximum R2 = 0.973/0.968). To determine the best synthesis conditions, black-box interpretation methods are applied to provide reliable insights from both qualitative and quantitative perspectives. The optimized choices of ionomer/catalyst ratio, water content, organic solvent, catalyst loading, stirring method, solid content, and ultrasonic spraying flow rate are properly made with few experimental attempts under ML results' guidance. Promising Pt utilization of 0.147 gPt kW-1 and a power density of 1.02 W cm-2 are achieved at 0.6 V in a single cell (H2/air) at an ultralow total loading of 0.15 mg Pt cm-2. Therefore, this work contributes to the economy of hydrogen energy by paving the way for MEA optimization with complex parameters by ML.