Computational modeling of low temperature plasmas (LTPs) is critical for many technology areas including manufacturing of computer chips, medical treatments, and the synthesis of nanomaterials. For LTPs, particle-in-cell Monte Carlo collision simulations are the highest-fidelity technique commonly used, in which electrons and ions are treated kinetically, meaning there are no assumptions about their velocity distributions. However, even the massively parallel kinetic codes rarely reach the timescales, length scales, and dimensionality of most real problems. As a result, many of these problems have been examined using fluid models with approximations (e.g. Maxwellian distributions, drift-diffusion fluxes, local mean energy approximation). In this project, we will develop and verify methods to use machine learning to bridge the gap between kinetic Monte Carlo simulations and fluid-based models for LTPs. These new techniques will enable simulations that extend to fluid time and length scales, but include physics that are normally only present in kinetic simulations which are several orders of magnitude more expensive. These methods would enable an improved understanding of the underlying physics and chemistry that is critical for semiconductor processing, medicine, chemical processing and many other areas.

DC pulsers are becoming available in plasma-based semiconductor processing equipment. These tailored waveforms open new degrees of freedom for process engineers, and also new questions on how the plasma chemistry and physics responds to such pulses. In this work, a combined experimental and modeling approaches will be used to investigate the impact of these pulsers on the plasma and how to best utilize them.

The primary objective is to investigate the underlying physics of breakdown which occasionally occurs at anamolously low applied voltages. The Subcontractor shall measure electron emission thresholds on surfaces perturbed with physical modifications, dielectric particles, or inclusions. The Subcontractor shall also identify and measure chemical interactions at electrode surfaces (e.g. etching, oxidation) in the presence of a plasma. Lastly, the Subcontractor shall perform computational modeling of this reactive chemical environment to associate observed surface changes with gas composition and conditions within the switch.

"The objective is to develop PIC capabilities for the plasma periphery and to integrate these capabilities with the rest of the FENIX framework through the underlying MOOSE ecosystem."

This research proposal targets the development of a prototypic radio frequency (RF), inductively coupled plasma (ICP) source for the DIII-D neutral beam injection (NBI) system. The envisioned research tasks will build the capabilities to inform critical design choices and, ultimately, deliver a full-scale prototype ICP positive ion source that provides an uniform positive hydrogenic (hydrogen or deuterium) ion density at the ion extraction area for 10 s, which translates to at least 85 A of positive ion current to the accelerator. A successful completion of the project will demonstrate the feasibility to deploy an ICP positive ion source on DIII-D, which will reduce the maintenance needs and will enable the envisioned power rise from 16 to 24 MW of total injected NBI power.

Capacitively coupled plasmas are used throughout industry for computer chip manufacturing. One of the most challenging of these processes is high aspect ratio etching, and custom voltage waveforms (non-sinusoidal waveforms) may provide more control over ion energy and angular distributions to address many of these challenges. In this work, we will use computational modeling to explore how the presence of electronegative gases (which can form negative ions), can influence the plasma dynamics in conditions relevant to high aspect ratio etching. The impact on ionization processes, sheath dynamics, and fluxes of reactive species will be analyzed.