[23p-F307-1~6] Machine learning in radiation research

[23p-F307-2] Extraction of physical information from extended X-ray absorption fine structure(EXAFS) data by sparse modeling

〇Yasuhiko Igarashi¹, Hiroyuki Kumazoe², Fabio Iesari³, Kazunori Iwamitsu², Okajima Toshihiro³, Yoshiki Seno⁴, Ichiro Akai², Masato Okada⁵ (1.Univ. Tsukuba, 2.Kumamoto Univ., 3.Aichi SR, 4.SAGA-LS, 5.Univ. Tokyo)

Keywords:extended X-ray absorption fine structure(EXAFS), Sparse modeling, Bayesian inference

For the analysis of extended X-ray absorption fine structure (EXAFS) data, we describe how sparse modeling and Bayesian inference can be used to extract physical information from noise-tolerant EXAFS analysis. By applying this method to analyze the EXAFS measured at the K-edge of Y atoms in a YOxHy epitaxial thin-film crystal, the radial distances of the first nearest O atom (Y-O) and the second nearest Y atoms (Y-Y) were correctly estimated, and the ratio of the radial distances indicated that the interstitial O-site is tetrahedral in the fcc lattice structure of Y atoms.