Application of kernel convolution for complementing source mask optimizationa

Kernel convolution with pattern matching (KCPM) is shown to be an effective complementary fast-computer-aided design tool for post or real-time quantitative assessments of the robustness and richness of the source-mask optimization solution. Compared with rigorous simulation, R 2 correlation is shown to be > 0.98 . A single match takes approximately 40 μ s , enabling full chip scale image quality assurance. Additionally, KCPM includes a full complex field interaction and can model any physical behavior that can be represented as a transmission on the mask or as a path difference in the pupil function. Examples are shown for monitoring electromagnetic field effects induced by mask topography through focus for scenarios where the mask pattern is changing and for comparing different source configurations.