Title:Operator-Consistent Physics-Informed Learning for Wafer Thermal Reconstruction in Lithography

Abstract:Thermal field reconstruction in post-exposure bake (PEB) is critical for advanced lithography, but current methods lack consistent accuracy and physical alignment. To bridge this gap, we introduce a physics-informed neural model that unifies coordinates, fields, and differential operators on a single computation graph using gated liquid layers. Applied to a 2D PEB scenario with internal heat generation and convective boundaries, the model formulates residuals via differential forms and a loss functional integrating closure, balance, and trace misfits. The results demonstrate rapid convergence, uniformly low errors, and strong agreement with FEM benchmarks, achieving stable training without oscillation. Our framework outperforms existing baselines in accuracy and robustness, establishing a reliable foundation for high-fidelity thermal modeling and offering a transferable strategy for operator-consistent neural surrogates in other physical domains.