Exploring Magnetized Plasma Sheaths through Physics-Constrained Deep Learning

This project focuses on the development of a novel framework that will leverage state-of-the-art machine learning techniques to predict the behavior of a plasma sheath. Plasma sheaths characterize the boundary between a plasma and a material, and play a critical role in numerous applications including space propulsion, materials processing, and future nuclear fusion energy reactors. Furthering the understanding of plasma sheaths will enable the prediction and control of this crucial plasma-material interface, facilitating advancements in multiple technologies. The award will also enable workforce development by supporting graduate and undergraduate student research. The project aims to develop a computationally efficient, yet high physics fidelity description of a plasma sheath. While particle-in-cell simulations are routinely employed to describe plasma-material interfaces, such simulations are computationally intensive, limiting their use to a relatively small number of cases. The project will overcome this limitation by leveraging advances in physics-constrained machine learning (ML) that enable high physics fidelity data to be integrated into reduced models of the sheath. The newly developed ML framework will treat the plasma-material interface for a low collisionality plasma, such as in a plasma thruster, along with the plasma sheath that emerges in the presence of a strong magnetic field and a low incident angle, typical of a magnetic fusion device. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.