Source optimization (SO) is an extensively used resolution enhancement technique in optical lithography. To improve computational efficiency, compressive sensing (CS) theory was applied to SO for clip-level applications in previous works. We propose, for the first time to our knowledge, a multi-objective adaptive SO (adaptive-MOSO) with CS for full chip. The fast optimization of a pixel illumination source pattern is achieved, and the imaging fidelity of each clip is guaranteed simultaneously at full chip. Fast CS with contour sampling is applied to accelerate the SO procedure by sampling all layout patterns. Novel cost function with adaptive weight distribution for every single clip is established to guarantee the lithography imaging fidelity for full chip. The simulation results prove that the adaptive-MOSO method improves the efficiency of SO and the lithography performance for large-scale chips.

Current source and mask optimization (SMO) research tends to focus on advanced inverse optimization algorithms to accelerate SMO procedures. However, innovations of forward imaging models currently attract little attention, which impacts computational efficiency more significantly. A sampling-based imaging model is established with the innovation of an inverse point spread function to reduce computational dimensions, which can provide an advanced framework for fast inverse lithography. Simulations show that the proposed SMO method with the help of the proposed model can further speed up the algorithm-accelerated SMO procedure by a factor of 3.

Recently, a single vectorial pupil optimization (VPO) was proposed to compensate for the polarization effect induced by thick mask and image optics at one field point in a lithography system, which does not work at full field points. In this paper, we propose a multi-objective VPO (MOVPO) method to obtain a universal vectorial pupil that can compensate for the polarization aberration at full field points. A novel multi-objective cost function, to the best of our knowledge, is built and includes uneven image pattern errors causing by polarization aberration (PA) at full field points in the MOVPO method. Comprehensive simulations demonstrate that the proposed MOVPO method can effectively improve the consistency of imaging and enlarge the overlapped process window at full field points.

No abstract

No abstract