Gan-Based Fast Mask Near-Field Calculation

Near-field calculation of thick masks in extreme ultra-violet (EUV) lithography is one of the fundamental tasks for process modeling and physical verification. Rigorous simulations are not applicable because of the enormous computation load, whereas the perturbation models (M3D) are susceptible to inaccurate computation for mask patterns with random edge corner and feature size. In this paper, we propose a generative adversarial network (GAN) based deep-learning approach to calculate the mask near-field to address the regarding efficiency and accuracy issues. We describe the GAN structure including the encoder, the discriminator network and the training strategy explicitly with sufficient detail, so others can follow, analyze and improve. The network is trained based on a set of mask samples where the corresponding mask near-field data are obtained by the EMF simulator. Simulation results show the merits of the proposed GAN-based approach.