OpenILT: An Open Source Inverse Lithography Technique Framework (Invited Paper)

Semiconductor lithography is a key process for fabricating integrated circuits, but it suffers from various distortions and variations that affect the quality of the printed patterns. Optical proximity correction (OPC) is a technique to improve pattern fidelity and robustness, and inverse lithography technique (ILT) is a promising OPC method that optimizes the mask as an inverse problem of the imaging system. However, ILT is computationally expensive and challenging to implement at a full-chip scale. In this paper, we present OpenILT, an open-source ILT platform that supports the rapid development and evaluation of GPU-accelerated and AI-driven ILT methods. OpenILT provides a modular and flexible framework that integrates various ILT components, such as lithography simulation, objective functions, and evaluation metrics. It also offers a convenient interface to PyTorch, a popular deep learning library, to enable the implementation of GPU-accelerated and AI-driven ILT methods.