覆盖
计算机科学
分割
计算机视觉
人工智能
程序设计语言
作者
Shinya Kyogoku,Minoru Harada,Mayuka Osaki,Takahiro Nishihata,Yasunori Goto
出处
期刊:Journal of micro/nanopatterning, materials, and metrology
[SPIE]
日期:2025-07-15
卷期号:24 (03)
标识
DOI:10.1117/1.jmm.24.3.034001
摘要
Due to design rule shrinkage for semiconductor devices, layer-to-layer overlays of actual circuit patterns need to be measured directly using a critical dimension scanning electron microscope (CD-SEM). In cases where upper- and lower-layer patterns have the potential to overlap partially, the visible area of the lower-layer pattern in the SEM image differs depending on the overlay as well as other process variations. As a result, it has become difficult to distinguish between upper- and lower-layer patterns, resulting in errors in overlay measurement. In response to this problem, we developed a method for dynamically placing measurement cursors for overlay metrology using a segmentation technique. This method determines the measurement cursor by recognizing the position and size of each lower-layer pattern through a segmentation image generated from a target SEM image using unsupervised deep learning. The performance of this method is evaluated on imitated images, for which value-controlled process variation is adopted. We compared the measurement results obtained with the proposed method and a conventional method that uses fixed measurement cursors and found that the proposed method maintains measurement accuracy and repeatability even for targets where measurement accuracy and repeatability degrade with the conventional method due to process variations.
科研通智能强力驱动
Strongly Powered by AbleSci AI