Fast lithographic source optimization using a batch-processing sequential least square estimator

This paper proposes a fast source optimization (SO) method for lithography systems to improve the imaging performance of different hotspots on the fullchip layout. Hotspots are referred to as the critical locations on the layout that are difficult to print. A fullchip layout usually includes numerous hotspots with different geometric characteristics. Current SO approaches collect all of the data from different hotspots before the optimization, and then try to calculate the common optimal source for all hotspots. If any new data from unaccounted hotspots become available, the optimal source has to be recalculated. This paper first develops a batch-processing sequential least square estimator, and then uses it to iteratively modify the source pattern based on the ongoing hotspot data. The optimized source for one hotspot can be updated to suit others without redundant computation. Simulations show that the proposed method can significantly accelerate the SO procedure, while improving the imaging performance of multiple hotspots.