Computational Modeling and Optimization of Large-Area Showerhead Configurations in PECVD Reactors

This Plasma enhanced chemical vapor deposition (PECVD), a key process in semiconductor manufacturing, typically requires optimization of device geometry and operating conditions to achieve uniform deposition across large substrates, ensuring high-quality production. Computational fluid dynamics (CFD) serves as a critical tool for simulating and designing flow fields; however, the complexity of device geometries, particularly in large-scale PECVD reactors with thousands to hundreds of thousands of nozzles, often poses significant challenges to numerical modeling due to substantial computational cost. In this study, we propose a novel series porous resistance model (SPRM) that significantly reduces computational cost by representing the complex nozzle geometry as simplified porous media. The SPRM provides an analogous mathematical representation for each individual nozzle without requiring the actual geometric modeling, thereby enabling efficient simulation and optimization of nozzle arrays without recreating a mesh model for each design case. The efficiency of the proposed SPRM is demonstrated through 28 distinct showerhead configurations, with flow fields and deposition characteristics evaluated and compared. The results show that the SPRM achieves up to a 90% reduction in computational load compared to conventional CFD approaches and eliminates the need for remeshing during optimization. Thus, the proposed method offers a computationally efficient pathway for the design and performance optimization of industrial-scale PECVD reactors, with potential applications in semiconductor manufacturing and other thin-film deposition processes.