Mask optimization based on conditional generative adversarial nets

In order to ensure the manufacturability of the mask while optimizing the exposure pattern, a way to optimize mask based on conditional generative adversarial nets (CGAN) is proposed. By establishing the connection between the lithography model and the nets, we define the model of the generator and the discriminator, and input the target images into them as additional information in the network. After sufficient training, the result under this simple network is able to obtain the optimal mask, which has better mask quality and ensures better manufacturability compared to the unoptimized mask.