Direct mask modelling for enhanced OPC accuracy in the era of curvilinear masks

Abstract

As advanced techniques in Optical Proximity Correction (OPC), such as curvilinear OPC and inverse lithography technology, gain traction in production flows, the accuracy and robustness of compact OPC models become increasingly critical for advanced technology nodes. Traditional mask modeling approaches typically infer effects such as corner rounding and mask loading from wafer data, which can leave residual errors and limit model fidelity. In this work, we implement a methodology that utilizes direct mask CD-SEM measurements to achieve more accurate mask characterization. By incorporating these measurements, we construct a mask lithography process (MLP) model that captures residual mask errors beyond what wafer-based inference can achieve. This mask-aware approach is designed to be embedded within a compact OPC model, improving prediction accuracy and enabling better correction strategies for complex geometries. As an initial step, the performance of the proposed MLP model is evaluated for curvilinear target patterns. The results demonstrate that the MLP model provides a high-fidelity representation of the physical mask, paving the way for improved OPC modeling in future nodes where curvilinear masks and inverse lithography are prevalent.