LIC-CGAN: fast lithography latent images calculation method for large-area masks using deep learning

Latent image calculation for large-area masks is an indispensable but time-consuming step in lithography simulation. This paper presents LIC-CGAN, a fast method for three-dimensional (3D) latent image calculation of large-area masks using deep learning. Initially, the library of mask clips and their corresponding latent images is established, which is then used to train conditional generative adversarial networks (CGANs). The large area layout is divided into mask clips based on local pattern features. If a mask clip matches one from the training library, its latent image can be obtained directly. Otherwise, the CGANs are employed to calculate its local latent image. Finally, all local latent images are synthesized to simulate the entire latent image. The proposed method is applied to lithography simulations for display panels, demonstrating high accuracy and a speed-up of 2.5 to 4.7 times compared to the rigorous process.