[18a-A401-1~9] 23.1 Joint Session N "Informatics"

△ [18a-A401-1] Evaluation of RHEED intensity oscillations by luminance histogram analysis using machine learning technique

〇(M2)Asako Yoshinari^1,2, Yusuke Kozuka², Yasunobu Ando³, Tarojiro Matsumura³, Masato Kotsugi¹, Naoka Nagamura^1,2,4 (1.Tokyo Univ. of Science, 2.NIMS, 3.AIST, 4.JST-PRESTO)

Keywords:RHEED, RHEED intensity oscillations, machine learning

RHEED (Reflection High-Energy Electron Diffraction) is widely used to evaluate surface conditions during thin film fabrication, especially for estimating film thickness using intensity oscillations. But only a part of the diffraction pattern information is available. In this study, we applied high-throughput peak fitting analysis using the EM algorithm, which is a kind of maximum likelihood estimation method, to the luminance histograms of RHEED images of the growth process of Nb-doped STO thin films. We tried to propose a new skill-independent evaluation method of RHEED intensity oscillation by using the information of the entire pattern.