Design of proton exchange membranes with high durability for fuel cells: From the perspective of machine learning

Proton exchange membranes (PEMs)() with high chemical durability have been generally researched in recent years. Free radical scavengers such as ceria can significantly suppress the chemical degradation of proton exchange membranes when added to proton exchange membranes. However, proton conductivity may decline in the presence of ceria. Thus, a trade-off between performance and durability in ceria-containing membranes emerges. To address this challenge, we developed a novel machine learning methodology called SPARK (Smart Prediction of Advanced Research on PEMs using Knowledge-based machine learning) to quickly predict the performance and durability of proton exchange membranes in fuel cell applications with high precision. Moreover, with the definition of the mixed index, the performance and durability were coupled, and the trade-off of performance and durability could be easily made. The SPARK methodology not only streamlines the design process for ceria-containing proton exchange membranes but also has broader implications for the development of other proton exchange membrane systems.

Automated summary

Proton exchange membranes (PEMs) with high chemical durability have been researched, and the addition of ceria as a scavenger can suppress chemical degradation but leads to a decline in proton conductivity. To address this trade-off, a machine learning methodology called SPARK was developed to predict the performance and durability of ceria-containing PEMs, with broader implications for other membrane systems.