Fast EUV Lithography Source Optimization Using Cascade Compressed Sensing

Abstract

Source optimization (SO) is intensively used in extreme ultraviolet (EUV) lithography to improve the image fidelity of multiple mask clips on the integrated circuit chip. However, if new layout data is inserted to the mask clip library, the optimal source pattern needs to be recalculated. This paper proposes a fast SO method based on cascade compressed sensing (CCS-SO) to circumvent the redundant computation problem. First, the optimized source patterns for different mask clips are individually calculated using compressed sensing algorithm, where the robustness to defocus is considered. Then, the global optimal source is formulated as the superposition of these pre-optimized source patterns, where the weighting coefficients are solved by the gradient-based algorithm. When new data is included in the mask clip library, we can rapidly update the weighting coefficients instead of re-optimizing the source pattern over again. The proposed CCS-SO approach is shown to achieve superior imaging performance and computational efficiency over the traditional SO approach.