[15p-MH-1~7] Innovation Inspired by Informatics in Crystal Growth Processes

[15p-MH-4] Suggestion of the method of fast search of optimized condition in SiC solution growth

〇Yosuke Tsunooka¹, Nobuhiko Kokubo¹, Shunta Harada¹, Miho Tagawa¹, Toru Ujihara¹ (1.Nagoya Univ.)

Keywords:solution growth, simulation, machine learning

In SiC solution growth, solution flow and supersaturation are important. Therefore, thermal fluid analysis is frequently done, but it is difficult to calculate comprehensively about parameters. In this study, thermal fluid analysis was performed and trained its data with machine learning to enable comprehensively calculation, and attempted to method of fast optimizing. In the result, machine learning calculates about 10000 times faster than thermal fluid analysis. It is suggested that machine learning enable to search optimized condition by investigating velocity under seed crystal on two parameters.