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

TL;DR

MaskOpt addresses the challenge of scalable, realistic mask optimization for IC manufacturing by introducing a large-scale, real-design dataset at the 45nm node that preserves standard-cell hierarchy and surrounding context. The dataset provides 104,714 metal tiles and 121,952 via tiles with cell-aware clipping, multiple context windows, and ground-truth MB-OPC and ILT masks generated via OpenILT, enabling robust benchmarking of deep learning approaches. Benchmark results reveal distinct trade-offs among baselines, with context and cell information enhancing mask fidelity and manufacturability, and highlight the differing needs of metal versus via layers. Overall, MaskOpt enables practical, context-rich DL development for IC mask optimization, bridging the gap between industrial layouts and learning-based methods.

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.

Figures (7)

• Figure 1: Visualization of real IC layout clipping for MaskOpt.
• Figure 2: MaskOpt dataset statistics.
• Figure 3: AOI221_X2 layout tile with mask samples.
• Figure 4: Mask prediction with cell and context awareness.
• Figure 5: Context size analysis for mask prediction.