Regularization of mask tone quantization for inverse lithography

One fundamental problem of inverse lithography is to evolve the optimized mask function which represents the mask using continuous tones to assume allowed mask values. To this end, a solution-dependent inverse lithography framework is proposed to penalize the typical continuous-tone mask (CTM) toward transparent and opaque tones that prints similarly well. The core of this framework is the mixing-in of a regularization cost functional which favors solutions with optimal quantization of the main feature and the sub-resolution assist features (SRAFs) in a quasistatic way instead of simplistic quantization with single thresholding in all regions. This solution-based strategy not only boosts the flexibility of the quantization process but also markedly enhances the process compatibility of mask fabrication. Simulation results show that, compared to conventional methods, our proposed optimization technique significantly improves both the pattern fidelity and mask manufacturability, thereby expanding the solution spectrum for mask layout design.