The deposition of uniform thin films is a critical process requirement in manufacturing precision optical components. Nonuniform thicknesses directly degrade device performance, creating a need for robust process control solutions. This paper presents an adaptive computational framework for the engineering and optimization of shadow masks, designed to correct deposition variations in physical vapor deposition systems over a large-aperture coating area of 20 cm in diameter. The simulation framework integrates a vector-based cosine power law for vapor distribution with a dynamic shadow projection model that couples mask geometry to the substrate’s radial displacement. To ensure reliable convergence to the global minimum, a two-stage optimization strategy using the Nelder–Mead algorithm is employed to minimize thickness variance over a configurable set of monitoring points. The framework was experimentally validated by depositing MgF2 over the 20 cm target zone, achieving a final thickness uniformity of 1.02%. This represents a significant improvement over the 50% nonuniformity of the uncorrected system. The simulation accurately predicts the experimental outcome (uniformity of 0.91%), demonstrating the framework’s effectiveness as a high-fidelity design-for-manufacturing tool for advanced optical coatings on extended surfaces.
Ramirez et al. (Tue,) studied this question.