Machine Learning for Fast Prediction of Ion Kinetics in Capacitively Coupled Plasmas at Industry Relevant Conditions

Machine learning (ML) techniques offer a powerful approach to assimilate previously collected data to improve engineering design and even accelerate high-fidelity numerical simulations. We apply deep neural-networks and convolutional neutral networks to model the kinetic properties of ion velocity distribution functions (IVDFs) in capacitively coupled plasma (CCP) discharges over a wide range of gas pressures, driving frequencies and voltages. This includes industry relevant voltages up to 10s of kV. Using data collected from 1D-3V parallel and GPU accelerated particle-in-cell (PIC) simulations of an Argon CCP discharge, the ML models are trained to map these inputs to the IVDF within the discharge and at the wafer surface. This model can then be directly applied to predict IVDFs without running new and expensive PIC simulations. Additionally, when higher accuracy is required, these new predictions of the plasma profiles can be applied to accelerate full fidelity PIC simulations. The plasma profiles generated from the ML models are used as initial conditions for new simulations, which converge, an order of magnitude faster than those initialized with uniform plasma profiles provided by a global model.