Machine Learning for Accelerating Materials Development for Electrochemical Systems: Fuel Cells and Electrochemical Separations
Date:
Abstract
Fuel cells are ubiquitous subplot devices in science fiction films and novels (e.g., Apollo 13, Star Trek, the Martian, etc.). Their common appearance in this genre probably stems from the fact that they are powered on hydrogen – the same fuel that powers the stars. In this talk, the impetus behind the re-emergence of high-temperature polymer electrolyte membrane (HT-PEM) fuel cells for vehicular applications will be presented. New polymer architectures based upon polycation-acid anion interactions have resulted in superior HT-PEMs in terms of ionic conductivity and stability over the classic phosphoric acid imbibed polybenzimidazole. Despite the advent of more functional membranes, gas reactant transport and reaction kinetic limitations in electrode layers still stymie the power density of HT-PEM fuel cells. Our lab has engaged in machine learning and high-throughput experimental methods to overcome the said problems through new electrode ionomer binders. These materials are examined as thin films on interdigitated electrode arrays featuring nanoscale electrocatalysts afforded through the process of block copolymer templating. The machine learning bridges molecular scale attributes of the ionomers to bulk material properties and even HT-PEM fuel cell device performance. Overall, we envision a new paradigm for streamlining materials discovery and development for achieving high power HT-PEM fuel cells.