Source optimization using hybrid genetic algorithm in lithography

Abstract

Source Optimization (SO), as a key technology of Resolution Enhancement Technologies (RETs), is employed to improve the performance of lithographic imaging in the advanced node. It is critical to guarantee the superior convergence ability of the pixelated SO method with the optimization algorithm. In this paper, a hybrid Genetic Algorithm (GA) is proposed as the method of optimizing the intensity distribution of the lithographic source according to the different mask pattern. Although the obvious performance of quick convergence and global search can be well represented in GA, keeping the local optimum in the optimization process is easily caused due to the complexity mathematical model for lithographic imaging process. Tabu search algorithm is combined with GA to widen the search range and break away from the local optimum for approximately generating the global optimum results of lithographic source. The valid pixels in the pixelated source are utilized as the optimized variables, which are encoded as a 1-D matrix in the optimized model. For ensuring the continuous gray-level changes in the intensity distribution of lithographic source, the Gaussian filter method is employed to blur the source shape. The simulation results demonstrate that the proposed hybrid-GA have the superior convergence performance in optimizing the intensity distribution of the lithographic source for the different mask patterns.