有機半導体の結晶構造決定のための機械学習論文が Applied Physics Letters に掲載されました。
“Powder x-ray diffraction analysis with machine learning for organic-semiconductor crystal-structure determination”
Naoyuki Niitsu, Masato Mitani, Hiroyuki Ishii, Nobuhiko Kobayashi, Kenji Hirose, Shun Watanabe, Toshihiro Okamoto, Jun Takeya
Abstract
The crystal structure of organic semiconductors is an important factor that dominates various electronic properties, including charge transport properties. However, compared with the crystal structures of inorganic semiconductors, those of organic semiconductors are difficult to determine by powder x-ray diffraction (PXRD) analysis. Our proposed machine-learning (neural-network) technique can determine the diffraction peaks buried in noise and make deconvolution of the overlapped peaks of organic semiconductors, resulting in crystal-structure determination by the Rietveld analysis. As a demonstration, we apply the method to a few high-mobility organic semiconductors and confirm that the method is potentially useful for analyzing the crystal structure of organic semiconductors. The present method is also expected to be applicable to the determination of complex crystal structures in addition to organic semiconductors.