Lithographic Source and Mask Optimization With Low Aberration Sensitivity

Source and mask optimization (SMO) is an important lithographic resolution enhancement technology. Recently, some research indicate that the lithography performance is sensitive to the errors of an actual lithography system, such as thermal aberration, thick mask effects, and mask uncertainties. Most of the errors would result in uncertain wavefront aberration, so the reduction of aberration sensitivity means the improvement of lithography stability. In this paper, we propose a low aberration sensitivity SMO (LASSMO) method to improve robustness of lithography performance against uncertain aberration. To reduce the aberration sensitivity, we build the LASSMO model via innovating new cost function including sensitivity penalty terms. Aiming at spherical aberration and coma, this method is demonstrated using two target patterns with critical dimensions of 45 nm. Taking into account the statistic characteristics of uncertain aberration, we use the normalized stochastic gradient descent algorithm to establish an iterative optimization framework. The simulation results show the benefit of LASSMO method in both high pattern fidelity and the low sensitivity of lithography imaging to aberration.