2025-11-11

高加速SEMのシースルーBSE画像のための高度なパターン輪郭抽出機能

• Event: 第45回ナノテスティングシンポジウム(NANOTS2025)
• 大家政洋, 岡本陽介, 中澤伸一, 丸山浩太郎, 山崎裕一郎, 御堂義博, 三浦典之
  
  第45回ナノテスティングシンポジウム(NANOTS2025), pp.35–39, Nov. 2025.

ABSTRACT

Backscattered electron (BSE) images acquired by high-voltage SEM contain information from both upper and lower device layers, while lower-layer patterns exhibit reduced contrast and sharpness due to electron diffusion. To enable accurate contour extraction, we developed a novel method combining deep learning with the Active Contour Model. A supervised training dataset was generated by modeling pattern and image variations observed in actual SEM images and producing synthetic images with corresponding ground-truth contours. Using the generated dataset, the proposed method successfully extracted lower-layer patterns and achieved accurate overlay measurements on actual semiconductor device structures.

Key words: High voltage, EBI, SEM, D2DB, CD, EPE, Overlay, CVNet