Researchers from the Shanghai Institute of Optics and Fine Mechanics of the Chinese Academy of Sciences proposed a source mask optimization (SMO) technique for extreme-ultraviolet lithography (EUVL) based on a thick mask model and a social learning particle swarm optimization (SL-PSO) algorithm. Simulations of the method suggested that the technique is more effective than similar approaches based on heuristic algorithms in optimization efficiency. With the continuous shrinking of the critical dimension of integrated circuits, the researchers said, precision and optimal performance requirements have become more stringent. Computational lithography improves performance without changing the hardware or software configurations of the lithography systems themselves, instead improving their performance by optimizing the present light source, mathematical models, and algorithms. SMO, a method of computational lithography, optimizes the source of illumination and the mask pattern simultaneously to improve the quality of imaging. EUVL has been applied to high-volume manufacturing of 5-nm process nodes, and SMO is critical to this technology’s capabilities. Seeking to improve the technique, the researchers proposed an SMO method for EUVL based on the thick mask model and SL-PSO algorithm. The method that the researchers tested in their simulation used the fast thick model based on a structure decomposition method (SDM). SDM calculates the mask spectrum by describing the propagation of incident light in the absorber and the multilayer. Though SDM has been studied extensively, the researchers said it has not been studied extensively with a pixelated phase mask — itself a resolution enhancement technique. The SL-PSO algorithm optimized the source and mask pattern, and the social learning strategy improved system efficiency. Additionally, a tuned initialization parameter controlled the initial swarm in the algorithm. This improved the mask’s manufacturability. The simulation results showed that the method was demonstrated to be faster and more efficient than other heuristic algorithms. Pattern error is significantly reduced and the imaging fidelity is improved. Simulations using patterns located at different positions of the ring slit showed that the method not only improved the imaging fidelity but also alleviated the shadowing effect. Future research will be focused on improving the optimization efficiency with increased numbers of mask pixels, as well as eliminating the pattern shift at the wafer defocus plane. The research was published in Optics Express (www.doi.org/10.1364/OE.418242).