Application of Machine Learning Techniques to Model-Based Optical Proximity Correction Algorithms


Shangliang Jiang and Avideh Zakhor

Lithography is a patterning method capable of structuring material on a fine scale, which is often applied to semiconductor manufacturing industry. Due to the resolution limits of optical lithography systems, the electronics industry has relied on resolution enhancement techniques (RET) to compensate and minimize mask distortions as they are projected onto semiconductor wafers. One of the main techniques is optical proximity correction (OPC), which modifies the mask amplitude by the addition of sub-resolution features to the mask pattern [1]. There are two typical OPC approaches: “rule-based” OPC and “model-based” OPC. Rule-based methods, heuristically and empirically modulating the amplitudes of the masks, are simple to implement, however, can just compensate the warping in local features. On the other hand, model-based methods use mathematical models to seek the global optimal solution, which improves the output pattern fidelity on the wafer. However, the model-based methods are computationally complex, especially for the full-chip OPC optimization [1]. This project concentrates on the research of fast OPC approach based on nonparametric regression, which inherits the merits from both rule-based and model-based OPCs. The proposed OPC method is tailored for the full-chip OPC optimization within reasonable running time. At the beginning, the training data are collected from the OPC pattern of a real chip. Subsequently, the mask to be optimized is divided into three kinds of regions: convex corners, concave corners and edges. Then, the nonparametric regression is independently applied to these three kinds of regions. Finally, the regression results are pieced up to compose the OPC pattern. The proposed algorithm is expected to speed up the current model-based OPC methods, and provide a practical way to optimize the OPC pattern in full-chip scale.

•A. Gu and A. Zakhor, "Optical Proximity Correction with Linear Regression," IEEE Transactions on Semiconductor Manufacturing, Volume 21, Number 2, pp 263 - 271, May 2008. (2008 Best Paper Award) [Adobe PDF]