Answers are grounded in this paper and cite their sources.
Optical Proximity Correction (OPC) is a critical step in semiconductor manufacturing, as it mitigates lithographic distortions and improves the pattern transfer fidelity from masks to silicon wafers. With the adoption of multi-beam mask writers, mask geometries are evolving from traditional Manhattan shapes to curvilinear patterns, offering improved resolution and process robustness. However, existing curvilinear OPC methods still suffer from limited optimization efficiency and accuracy, largely due to the difficulty of contour parameterization. To address this issue, we propose a gradient-based framework for vertex-based curvilinear OPC, where the mask is represented by connected Bézier curve segments controlled by trainable vertices, and the resulting rasterized mask is optimized to reduce lithographic errors. We evaluate the proposed method on the industry-standard dataset. Experimental results show that our method consistently outperforms state-of-the-art baselines on L 2 , PVBand, and VPE, demonstrating the effectiveness of the proposed gradient-based curvilinear optimization framework.
No takes yet. Share an insight, caveat, or question.
Zeng et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: