Human–Machine Collaboration for Accelerated Discovery of Promising Oxygen Evolution Electrocatalysts with On-Demand Elements
Ken Sakaushi, Watcharaporn Hoisang, Ryo Tamura
Abstract
High Resolution Image Download MS PowerPoint Slide A drastically efficient method for identifying electrocatalysts with desirable functionality is a pressing necessity for making a breakthrough in advanced water-electrolyzers toward large-scale green hydrogen production and addressing the significant challenge of carbon neutrality. Despite extensive investigations over the last several centuries, it remains a time-consuming task to identify even one promising affordable electrocatalyst without platinum-group-metal (PGM) for one electrochemical reaction due to its great complexities, particularly for the key anode reaction in the water-electrolyzer of the oxygen evolution reaction (OER). In this study, we demonstrate that a human–machine collaboration based on stepwise-evolving artificial intelligence (se-AI) can significantly shorten the development period of PGM-free multimetal OER electrocatalysts with performance beyond a PGM of RuO 2 . We were able to reach optimized materials only after 2% experimental trials of the entire candidate pool. The best PGM-free electrocatalyst discovered exhibited excellent activity comparable to RuO 2 and, surprisingly, also demonstrated superior stability with a high current density of up to 1000 mA/cm 2 at even pH 9.2, which condition is a thermodynamically challenging for typical PGM-free materials. This work illustrates that human’s material discovery can be significantly accelerated through collaboration with AI.