Wave Front (Japanese)Machine Learning (Japanese) > Particle-PULS/DL
[Japanese]   [English]

Plasma Simulator with Surrogate Models from Deep Learning

- collaboration of CAE and ML -

News

  • Particle-PLUS/DL website is open!
  • We are scheduled to release the application this July.
  • This page includes contents under development.
  • They might change in future release.
  • (July 1, 2021)

About Particle-PLUS ⁄ DL

Particle-PLUS/DL is an application to simulate various plasmas (such as magnetron, capacitive coupling, and inductive coupling) with surrogate models from deep learning. The simulation is enough quick to help parametric studies to optimize deposition and etching processes.

Procedure of Plasma Simulation by Particle PLUS ⁄ DL

Procedure of Plasma Simulation by Particle PLUS/DL

Input Variables
Device type
Magnetron
plasma
CCP ICP
Chambar size (D) (D)
DC electrodes
RF electrodes (D)
Magnets
Coils
Gas pressure (C) (C) (C)
Gas mixture (C) (C) (C)

Output Variables
Device type
Magnetron
plasma
CCP ICP
Electron density
Ion flux
Electron temperature (C) (C) (C)

  • (C) : under consideration
  • (D) : under development

Development Steps of Deep Surrogate Model

Some surrogate models have already built by learning results of our CAE simulations using neural networks. Therefore users need not to prepare sample data for the machine learning. Particle-PLUS/DL including the models is released after verification.

Development Steps of Deep Surrogate Model

Minimum System Requirements

  • OS: Windows 10
  • RAM: 4 GB
  • Disk space: 2 GB
  • Display resolution: 1152 x 864
  • API: .NET Framework 4.6 (which is pre-installed to Windows 10)

Further Simulations

Some parameters are limited in Particle-PLUS/DL for reasons of learned values. If you want to simulate more complex plasma systems (, for example, magnetron CCP, magnetron ICP, multi-frequency CCP, and mixture gas plasma), please consider following simulators.