Presentation Information
[7p-S202-1]Autonomous synthesis of polycrystalline TiO2 thin films with desired crystal structure
〇Sota Hasebe1, Kazunori Nishio5, Akira Aiba2, Yota Suzuki3, Ryo Nakayama1, Tomohito Sudare1, Shigeru Kobayashi1, Ryota Shimizu4, Taro Hitosugi1,5 (1.Univ. Tokyo, 2.Rigaku Corp, 3.CFC, 4.IMS, 5.Science Tokyo)
Keywords:
Autonomous experiment,Bayesian optimization,X-ray diffraction
The autonomous synthesis of polycrystalline TiO2 thin films with desired crystal structures was demonstrated by leveraging machine learning and robotics. An evaluation metric, defined as the correlation coefficient between the X-ray diffraction (XRD) pattern of the as-grown thin film and that of a powder reference from a database, was established. With this metric as the objective variable, Bayesian optimization efficiently explored the synthesis conditions, namely substrate temperature, leading to the selective formation of the rutile and anatase phases.