[20a-A406-1~9] 13.4 Si processing /Si based thin film / MEMS / Equipment technology

[20a-A406-3] Classification of surface states on chlorosilane reacted Si using machine learning

〇Hidetoshi Mimura¹, Leonid Bolotov², Yuunosuke Sakai¹, Tetsuo Yamamoto¹, Takafumi Sasaki¹, Noriyuki Uchida² (1.KOKUSAI ELECTRIC Co., 2.AIST-DTech)

Keywords:Surface coverage, Scanning tunneling spectroscopy, Machine learning

Silicon nitride is an important material for semiconductor devices, and HCDS is one of its material gases. Machine learning classification of the surface data measured by an external laboratory after exposure to HCDS revealed that the Si(100) surface coverage was broadly classified into hydrogen-terminated, chlorine-terminated, Si dangling-bond, and HCD adsorbed states. First-principles calculations identified several candidate structures that are considered to be intermediate states of the reactions.