Home // CENICS 2025, The Eighteenth International Conference on Advances in Circuits, Electronics and Micro-electronics // View article

Optical-System-Aware Feature Extraction for Lithography Hotspot Detection

Authors:
Masahiro Yamamoto
Masato Inagi
Shinobu Nagayama

Keywords: lithography; hotspot; feature vector; optical system.

Abstract:
This paper proposes a new feature vector for machine learning-based hotspot detection in lithography for Large-Scale Integration (LSI) fabrication, which incorporates the optical characteristics of exposure systems. Unlike existing features that focus only on local layout sub-patterns, the proposed feature takes into account optical behavior essential to accurate pattern transfer. In LSI fabrication, a hotspot is a region in the layout where an undesired open or short circuit may occur, even if the design rules are satisfied. Hotspots can significantly reduce manufacturing yield, and the cost of reworking after fabrication begins is substantial. Therefore, it is crucial to detect and remove hotspots at the pre-fabrication stage. Although several feature vectors have been developed for hotspot detection, most of them ignore the optical system's influence, which is critical in the lithography process. By incorporating optical characteristics, our proposed feature aims to improve detection accuracy and reduce the need for time-consuming lithography simulations.

Pages: 8 to 13

Copyright: Copyright (c) IARIA, 2025

Publication date: October 26, 2025

Published in: conference

ISSN: 2308-426X

ISBN: 978-1-68558-308-8

Location: Barcelona, Spain

Dates: from October 26, 2025 to October 30, 2025