[18a-Z32-1~10] 23.1 Joint Session N "Informatics"

[18a-Z32-10] Identification of Dislocation Clusters based on Image Translation of Photoluminescence Image of Multicrystalline Silicon Wafer

〇Hiroaki Kudo¹, Tetsuya Matsumoto¹, Kentaro Kutsukake², Noritaka Usami³ (1.Grad Sch. of Info., Nagoya Univ., 2.Center for AIP, RIKEN, 3.Grad. Sch. of Eng., Nagoya Univ.)

Keywords:machine learning, multicrystalline silicon, dislocation clusters

Identification of dislocation cluster region in the photoluminescence image (PL image) of the multicrystalline silicon wafer was implemented with image translation by GAN, which is machine learning. As an input of the network, an image which is composed with 3 channels of brightness of PL images for the target wafer and the upper and lower wafers was used for experiments of image translation. An image of 1 channel of brightness of PL image for the target wafer was also used. As a result, it was confirmed that it is possible to identify the dislocation region. It was also suggested that the information on the three channels can be used to embed information on the occurrence and disappearance of dislocation clusters.