LithoDreamer: A Physics-Informed World Model for Multi-Stage Computational Lithography

2026-06-25

AI Summary

Existing computational lithography models struggle to accurately capture the continuous, multi-stage physical evolution from layout to developed image. This work proposes a physics-informed world model framework that formulates lithography as a decision-driven, multi-step state evolution system. By modeling feature transitions between consecutive states in a latent space, the approach enables controllable process intervention and state transition. The method innovatively introduces a contrastive variational optimization paradigm, allowing for interpretable intervention policy learning without requiring dense supervision. It achieves state-of-the-art performance in both forward simulation and inverse lithography planning tasks and releases the first multi-stage lithographic evolution dataset to support future research.

Abstract

As semiconductor technology nodes scale, computational lithography is essential for ensuring yield and performance. However, lithography is a continuous physical process involving mask optimization, optical imaging, resist exposure, and development, which existing models fail to capture. To overcome this limitation, we present LithoDreamer, the first physics-informed World Model (WM) framework for computational lithography, which formulates the ``Layout-Mask-Resist Image-After Development Image (ADI)'' pipeline as a decision-driven multi-step evolution system. LithoDreamer captures feature changes between adjacent states to model stage-specific physics-informed latent spaces, in which it controls process intervention exploration and drives subsequent state transitions. To achieve interpretable intervention optimization without continuous supervision, we propose a contrastive variational optimization paradigm that contrasts the latent differences between intervention paths with variational evolution constraints, guiding the model to generate evolutions consistent with real lithography physics. Experiments show LithoDreamer achieves state-of-the-art performance in forward evolution and inverse planning. Our lithography dataset is publicly available at GitHub (https://github.com/7jiangyq/lithodreamer.git).

Problem

• computational lithography
• physics-informed modeling
• multi-stage process
• world model
• lithography simulation

Innovation

• Physics-Informed World Model
• Computational Lithography
• Multi-Stage Evolution
• Contrastive Variational Optimization
• Inverse Planning