Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Differentiable Edge-based OPC

Guojin Chen, Haoyu Yang, Haoxing Ren, Bei Yu, David Z. Pan

TL;DR

DiffOPC tackles the trade-off between manufacturability and imaging performance in optical proximity correction by introducing a differentiable edge-based OPC framework. It propagates true gradients through a differentiable edge-to-mask rasterization, enabling gradient-based optimization of edge segments under MRC constraints and integrating a two-stage SRAF strategy. The approach combines EBOPC’s manufacturability with ILT-like performance, achieving improved $L_2$ and $EPE$ metrics and substantially reduced shot counts and manufacturing costs compared with state-of-the-art ILT and EBOPC methods. Experimental results on metal and via layers, including large datasets, show that DiffOPC reduces EPE by up to a large margin and delivers post-MRC-friendly masks without requiring additional post-processing. The work demonstrates a practical path toward industrial adoption of differentiable OPC by delivering high fidelity printability with efficient computation and manufacturability guarantees.

Abstract

Optical proximity correction (OPC) is crucial for pushing the boundaries of semiconductor manufacturing and enabling the continued scaling of integrated circuits. While pixel-based OPC, termed as inverse lithography technology (ILT), has gained research interest due to its flexibility and precision. Its complexity and intricate features can lead to challenges in mask writing, increased defects, and higher costs, hence hindering widespread industrial adoption. In this paper, we propose DiffOPC, a differentiable OPC framework that enjoys the virtue of both edge-based OPC and ILT. By employing a mask rule-aware gradient-based optimization approach, DiffOPC efficiently guides mask edge segment movement during mask optimization, minimizing wafer error by propagating true gradients from the cost function back to the mask edges. Our approach achieves lower edge placement error while reducing manufacturing cost by half compared to state-of-the-art OPC techniques, bridging the gap between the high accuracy of pixel-based OPC and the practicality required for industrial adoption, thus offering a promising solution for advanced semiconductor manufacturing.

Differentiable Edge-based OPC

TL;DR

DiffOPC tackles the trade-off between manufacturability and imaging performance in optical proximity correction by introducing a differentiable edge-based OPC framework. It propagates true gradients through a differentiable edge-to-mask rasterization, enabling gradient-based optimization of edge segments under MRC constraints and integrating a two-stage SRAF strategy. The approach combines EBOPC’s manufacturability with ILT-like performance, achieving improved and metrics and substantially reduced shot counts and manufacturing costs compared with state-of-the-art ILT and EBOPC methods. Experimental results on metal and via layers, including large datasets, show that DiffOPC reduces EPE by up to a large margin and delivers post-MRC-friendly masks without requiring additional post-processing. The work demonstrates a practical path toward industrial adoption of differentiable OPC by delivering high fidelity printability with efficient computation and manufacturability guarantees.

Abstract

Optical proximity correction (OPC) is crucial for pushing the boundaries of semiconductor manufacturing and enabling the continued scaling of integrated circuits. While pixel-based OPC, termed as inverse lithography technology (ILT), has gained research interest due to its flexibility and precision. Its complexity and intricate features can lead to challenges in mask writing, increased defects, and higher costs, hence hindering widespread industrial adoption. In this paper, we propose DiffOPC, a differentiable OPC framework that enjoys the virtue of both edge-based OPC and ILT. By employing a mask rule-aware gradient-based optimization approach, DiffOPC efficiently guides mask edge segment movement during mask optimization, minimizing wafer error by propagating true gradients from the cost function back to the mask edges. Our approach achieves lower edge placement error while reducing manufacturing cost by half compared to state-of-the-art OPC techniques, bridging the gap between the high accuracy of pixel-based OPC and the practicality required for industrial adoption, thus offering a promising solution for advanced semiconductor manufacturing.
Paper Structure (14 sections, 17 equations, 5 figures, 5 tables, 4 algorithms)

This paper contains 14 sections, 17 equations, 5 figures, 5 tables, 4 algorithms.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Forward Lithography Model
  4. Evaluation Metrics
  5. Problem Formulation
  6. DiffOPC Algorithm
  7. Edge Segmentation and Movement
  8. Differentiable Edge-Based OPC
  9. Experimental Results
  10. Experimental Results on Metal Layer
  11. Experimental Results on Via Layer
  12. Ablation Study
  13. Summary of Experimental Results
  14. Conclusion

Figures (5)

  • Figure 1: Model-based OPC includes pixel-based OPC (ILT) and edge-based OPC (EBOPC). While ILT masks face manufacturability issues requiring significant post-processing, EBOPC masks are manufacturable but have performance limitations. DiffOPC combines the advantages of both approaches, enhancing manufacturability and performance.
  • Figure 2: DiffOPC: differentiable edge-based OPC framework.
  • Figure 3: (a) STE forward; (b) STE backward.
  • Figure 4: DiffOPC SRAF insertion and optimization.
  • Figure 5: MRC violations across methods and datasets