[15p-D511-1~11] 15.7 Crystal characterization, impurities and crystal defects

▼ [15p-D511-3] A nanoXRD Based Analysis on HVPE GaN Structure Combined with Machine Learning

〇(DC)Zhendong WU¹, Yudai Nakanish¹, Yusuke Hayashi¹, Tetsuya Tohei¹, Yasuhiko Imai², Kazushi Sumitani², Shigeru Kimura², Akira Sakai¹ (1.Grad. Sch. Eng. Sci., Osaka Univ., 2.JASRI)

Keywords:GaN substrate, NanoXRD, Machine Learning

Nanobeam X-ray diffraction (nanoXRD) is a powerful in situ crystal structure detection method utilizing synchrotron radiation. Highly collimated and monochromatic beamlines generated by the synchrotron radiation enable high throughput and rapid nanoXRD experiments. Meanwhile, we are facing a primary challenge about how to efficiently utilize enormous diffraction patterns, which include rich information on crystallinity, like microstrain and defects. Manual analysis of the experimental data for novelty or defect recognition becomes challenging since the increasing complexity while the experimental data acquisition rate increases. Here, a novel method based on machine learning (ML) is utilized to analyze the clustering properties based on enormous raw nanoXRD patterns, which help us investigate the crystal structural characteristics. Interestingly, the ML-based method successfully clustered raw nanoXRD patterns with different crystallinity. More results will be shown in the presentation.