Fast optical proximity correction method adopting an iterative threshold algorithm of log-sum regularization compressive sensing

Mask optimization is the key step of the advanced technology node in the VLSI manufacturing process. As one of the most representative techniques, optical proximity correction (OPC) is a resolution enhancement technique widely used in lithography. Striking the right balance between lithographic accuracy and computational efficiency poses a significant challenge for OPC. In this paper, a logarithmic and regularized iterative threshold algorithm with the RMSprop method for optical proximity correction (RMS-log-sum-OPC) is proposed for the first time, to our knowledge, to achieve the goal of fast OPC and high-fidelity mode at the same time. The OPC process can be implemented by solving a series of logarithmic and regularized reconstruction problems, and the weights can be updated with each iteration. The algorithm uses the dynamic learning rate of RMSprop to balance the convergence speed and stability while retaining the sparse enhancement characteristic of the logarithm, and is suitable for high-dimensional non-convex sparse problems. The results show that under two target layouts, the RMS-log-sum-OPC method improves the lithographic fidelity by 44.7% and 19.6%, respectively, compared with the existing BCS-OPC method and the running time of each iteration is reduced by 6.67% and 2.48%, respectively.