Design Space Exploration for Hotspot Detection

Advanced technology nodes suffer from high systematic defectivity. A significant part of such defectivity is caused by Lithographic Hotspots. Hotspots are Design Rule Check (DRC) clean areas of a layout which tend to show abnormal variation due to complex design-process interactions. In the past decade, we have witnessesed several Pattern Matching (PM) and Machine Learning (ML)-based hotspot detection solutions. However, there is little research towards addressing two pressing needs of the industry: (a) Development of hotspot detection models during early stages of technology development, and (b) Detecting Truly-Never-Seen-Before (TNSB) hotspots in incoming layouts. In this work, we propose solutions to address these issues and quantitatively demonstrate the effectiveness of the proposed methodologies.