Emi Kawamura

Contact:

UCB Phone: (510) 642-1297
UCB Fax: (510) 642-6330
Email: emi@langmuir.eecs.berkeley.edu

Biographical:

Finished Ph.D. dissertation in Physics at the University of California at Berkeley on December 17, 2003.

Research:

THE EFFECT OF COULOMB SCATTERING ON LOW PRESSURE HIGH DENSITY ELECTRONEGATIVE DISCHARGES

For electronegative plasmas with low gas pressure and high ion densities, we expect Coulomb collisions between positive and negative ions to dominate over collisions between ions and neutrals. We incorporated Nanbu's method (Phys. Rev. E 55:4642, 1997) into our 1d3v particle-in-cell (PIC) code PDP1 in order to study the effect of Coulomb collisions on low pressure high density electronegative discharges. Nanbu's method treats a succession of small-angle binary collisions as a single binary collision with a large scattering angle, which is far faster than treating each individual small-angle collision. We find that Coulomb collisions between positive and negative ions in electronegative discharges significantly modify the negative ion flux, density and temperature profiles.

PARTICLE IN CELL SIMULATIONS OF LOW PRESSURE SMALL RADIUS POSITIVE COLUMN DISCHARGES

Improved positive column simulation techniques are needed because of the nonlocal nature of typical low pressure discharges used for lighting. In a local model, power balance between Joule heating and collisional losses must hold for each volume element of the discharge separately while a nonlocal model requires only a global power balance. The departure from locality increases as either gas density ng or radius R is decreased. Despite this, most current fluorescent lamp software is based on the local concept. We present a nonlocal kinetic particle-in-cell Monte-Carlo collisions (PIC-MCC) code to simulate low pressure, small radius, positive column discharges. This code is also compared to a nonlocal fluid code, a nonlocal kinetic Monte-Carlo code and to experimental data. The PIC-MCC code made the least approximations and assumptions and was accurate and stable over a wider parameter regime than the other codes. Also, 1d3v PIC-MCC simulation speeds are quite competitive even on moderate workstations. Finally, we analyze the PIC-MCC simulation results in detail, especially the power balance and the radial electron kinetic energy flux Hr(r). We found that for low ngR < 1e15 cm-2, the electron kinetic energy flux is directed radially outward while for higher ngR, it is directed radially inward except right near the wall.

GE Project

PARALLELING AND OTHER METHODS OF SPEEDING UP PARTICLE CODES APPLIED TO RF DISCHARGES

Particle-in-cell codes provide detailed descriptions of plasma behavior but run much slower than fluid codes We demonstrate means for speeding up particle-in-cell (PIC) simulations of RF discharges. In electrostatic 1d3v PIC simulations of RF discharges, the field solve is typically less than 1 percent of the work load. Even for 2d3v PIC simulations, the field solve can be a small percentage of the work load especially when FFT methods are used to solve for the field. Thus, we can obtain significant gains by just paralleling the particle processing (e.g., pushing/accumulating) without paralleling the field solve. We applied this simple scheme to conduct 1d3v and 2d3v PIC simulations of argon RF discharges on 2 and 4 CPU symmetric multiprocessor (SMP) machines and on a distributed network of workstations (NOW). For a fixed number of grid points, the speedup for this parallel particle processing became more linear with increasing particle number. Other speedup methods include implicit coding (longer time steps); subcycling of electrons (many electron steps per ion step); lighter mass ions for acceleration to intermediate equilibrium, then return to full mass; different weights for electrons and ions; improved initial density profiles. Tables are provided showing the gains achieved by each method, and six global cautions (e.g., a particle should not travel more than one grid spacing in one timestep), used to determine the time step and grid spacing for each simulation. Explanations are also provided for any observed gains being less than simple expectations.

Download ERL report "Physical and Numerical Methods of Speeding Up Particle Codes and Paralleling as Applied to RF Discharges"
Postscript (436 Kbytes)

ION ENERGY DISTRIBUTIONS IN RF PLASMA SHEATHS

We present a review and analysis of ion energy distributions (IEDs) arriving at the target of an rf discharge. We mainly discuss the collisionless regime, which is of great interest to experimentalists and modelers studying the high density discharges in which the sheath is much thinner than in conventional RIE systems. We assess what has been done so far and determine what factors influence the shape of the IEDs. We also briefly discuss collisional effects on the IEDs. Having determined the important parameters, we perform some particle-in-cell (PIC) simulations of a collisionless current-driven rf sheath which show that ion modulations in an rf sheath significantly affect the IEDs when the ratio of the ion transit time over the rf period is less than one.

Download paper "Ion Energy Distributions in RF Sheaths; Review, Analysis and Simulation"
Postscript (6.7 MBytes)
PDF (4.8 MBytes)

MODELING FIELD ENHANCEMENT IN AN RF GAP

The high electric fields applied to microwave cavities induces field emission of electrons. This field emission current I_FE combined with neutral desorption at the nose cones of a microwave cavity can lead to surface damage. I_FE heats metal surfaces, leading to desorption of contaminant neutrals on the surfaces. The field emitted electrons ionize the desorbing neutrals. The resulting positive ions enhance the field at the emitter, increasing I_FE. The higher I_FE increases the power dissipation and temperature of the emitter, leading to more neutral desorption. More neutrals implies more positive ion creation and field enhancement, etc., leading to a positive feedback loop. Eventually, the emitter will melt and is ``rf processed''. We use a 1D particle-in-cell/Monte-Carlo collisions model to show the effect of positive space charge on I_FE.

Download MURI report "Modeling Field Enhancement in an RF Gap"
(Note that for some browsers you may have to hold down the shift key while clicking on the file below.)
Microsoft Word document (583 KBytes)

A CHLORINE MCC MODEL FOR THE PIC METHOD

In order to study the interactions of the charged particles with the neutral background in a particle-in-cell (PIC) simulation, V. Vahedi and M. Surendra developed a Monte Carlo Collision (MCC) model which included the null collision method (V. Vahedi and M. Surendra, Comp. Phys. Comm. 87, 170-198, 1995). This involved identifying the important species in the plasma which have relatively high densities and selecting a set of reactions between those species with highest reaction rates which give the proper energy peakloss and particle creation and loss mechanisms. Vahedi and Surendra applied their MCC model to argon and oxygen discharges. We also wish to study other processing gases and have recently developed an MCC package for chlorine. We intend to compare our PIC-MCC simulations of chlorine discharges with both theory and experiment.

[ Return to the Beginning ] [ Return to the PTSG Homepage ] [ Return to the Berkeley Homepage ]