Presentation Information
[24p-P07-1]Prediction for photocatalytic activity of hematite photoelectrodes by machine learning of experimental data
〇Takuma Nishimura1, Yoshitaka Kumabe2, Yosuke Harashima3, Mikiya Fujii3, Takashi Tachikawa1,2 (1.Grad. Sch of Sci., Kobe Univ., 2.Mol. Photsci. Res. Center., Kobe Univ., 3.NAIST)
Keywords:
photocatalyst,machine learning,materials informatics
Hematite (α-Fe2O3), known as a red rust, is one of the visible-light responsive photocatalysts. Through photoelectrochemical (PEC) water splitting reaction, hematite can produce clean hydrogen, which is attracting much attention as a next generation energy resource. In this study, we synthesized hematite samples doped with various elements through solvothermal reaction and made photoelectrodes through spin-coating with suspensions of the hematite photocatalysts on a fluorine-doped tin oxide (FTO) substrate to evaluate the PEC performance. We investigated how to create the machine learning models to predict their photocatalytic activity using the elemental features as explanatory variables.