MaskOpt: A Large-Scale Mask Optimization Dataset to Advance AI in Integrated Circuit Manufacturing

2025-12-18

AI Summary

Optical lithography faces severe mask optimization challenges at advanced technology nodes due to diffraction effects and process variations; conventional model-based approaches (e.g., OPC/ILT) suffer from high computational cost and poor scalability. Existing deep learning methods rely on synthetic data, neglect standard-cell topology and contextual dependencies, and thus fail to generalize to real manufacturing. This work introduces the first large-scale, real-world IC manufacturing benchmark for mask optimization—covering over 220,000 standard-cell slices (metal + via layers) at the 45 nm node—with explicit modeling of cell-level topology and multi-scale optical proximity context. We propose the first “cell-aware + context-aware” mask optimization paradigm, enabling variable-window contextual modeling. Experiments demonstrate that our dataset significantly improves simulation fidelity of deep learning–based mask generation, and both cell-structure and contextual inputs are indispensable for performance.

Technology Category

• Machine Learning: Optimization
• Search and Optimization: Learning to Search
• Computer Vision: Learning & Optimization for CV

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

As integrated circuit (IC) dimensions shrink below the lithographic wavelength, optical lithography faces growing challenges from diffraction and process variability. Model-based optical proximity correction (OPC) and inverse lithography technique (ILT) remain indispensable but computationally expensive, requiring repeated simulations that limit scalability. Although deep learning has been applied to mask optimization, existing datasets often rely on synthetic layouts, disregard standard-cell hierarchy, and neglect the surrounding contexts around the mask optimization targets, thereby constraining their applicability to practical mask optimization. To advance deep learning for cell- and context-aware mask optimization, we present MaskOpt, a large-scale benchmark dataset constructed from real IC designs at the 45$mathrm{nm}$ node. MaskOpt includes 104,714 metal-layer tiles and 121,952 via-layer tiles. Each tile is clipped at a standard-cell placement to preserve cell information, exploiting repeated logic gate occurrences. Different context window sizes are supported in MaskOpt to capture the influence of neighboring shapes from optical proximity effects. We evaluate state-of-the-art deep learning models for IC mask optimization to build up benchmarks, and the evaluation results expose distinct trade-offs across baseline models. Further context size analysis and input ablation studies confirm the importance of both surrounding geometries and cell-aware inputs in achieving accurate mask generation.

Problem

• Advancing deep learning for cell- and context-aware mask optimization
• Addressing limitations of synthetic datasets in practical IC manufacturing
• Providing a large-scale real-design benchmark for optical proximity correction

Innovation

• Large-scale real IC design dataset for mask optimization
• Cell-aware and context-aware tile clipping approach
• Multiple context window sizes to capture optical proximity effects