Particle Tracing Minicourse

COMSOL Tokyo Conference 2014

 Outline
• What’s new in V5.0
• Overview of Particle Tracing Module physics interfaces
• Boundary conditions
• Release features
• Secondary particle emission
• Monte Carlo modeling
• Accumulators
• Results and visualization
• Model Library examples
 Particle Tracing Module
• Add-on to COMSOL Multiphysics.

• Combines with any COMSOL Multiphysics

Module.

• Includes three physics interfaces for computing

particle trajectories.
 Particle Tracing
• Particle tracing can be used
as an alternative to the finite
element method for solving
real world physics problems.
• Advantages:
– No numerical instabilities
which occur in the finite
element method due to high
Peclet numbers.
– Much simpler mathematics
involved in formulating the
problem.
– Solves a different class of
problems than continuum Ion motion in a quadrupole mass spectrometer.
methods. Ion motion is influenced by both AC and DC
fields
 Key Applications
Ion cyclotron motion
• AC/DC due to uniform
– Mass spectrometry external magnetic
– Beam physics and ion optics field computed using
– Multipaction the Lagrangian
• Fluid Flow formulation
– Fluid flow visualization
– Sprays
– Separation, filtration, and erosion
– Brownian motion and particle diffusion
– Rarefied gas dynamics
Histogram of the ion
• RF angular energy
– Ray tracing for smoothly graded materials distribution function
(limited ray tracing) in an inductively
• Plasma coupled plasma. The
– Ion angular energy distribution function x-axis is the angle of
– Etching incidence of the ions
• Acoustics striking the wafer
– Acoustophoretic forces (deg), the y-axis is the
energy (eV). The
• Mathematics height is the relative
– Classical mechanics count
 New in V5.0
• Accumulators
– Communicate information from particles to the
underlying mesh.
– Enhances multiphysics capability.
 New in V5.0
• Improvements to secondary particle emission
– Domain-based release.
– Multiple species.
 New in V5.0
• New options for built-in forces
– High Knudsen number corrections.
– New drag laws.
 Physics Interfaces
• Mathematical Particle Tracing
– Specify the equations of motion using Massless,
Newtonian, Lagrangian or Hamiltonian formulations.
– Complete freedom over the equations solved allows, for
example, ray tracing to be modeled.

• Charged Particle Tracing

– Model ion and electron trajectories in electric and
magnetic fields.
– Easy to define electric, magnetic and collisional forces.
– Allows one-way or two-way coupling to electric fields.

• Particle Tracing for Fluid Flow

– Model microscopic and macroscopic particles in a fluid.
– Includes drag, gravitational, dielectrophoretic,
acoustophoretic and many other forces.
– Can model fluid-particle interactions.
 Charged Particle Tracing
• Use this to model ion and electron trajectories in electric and
magnetic fields.
• Predefined forces.
• Monte Carlo modeling of interaction with background gases
• Typically the fields are pre-computed from one of the AC/DC
interfaces.
• Coulomb force for particle-particle interactions.

Multipaction due to
secondary electron
emission in an RF
device operating at
a very specific
frequency
 Particle Tracing for Fluid Flow
• Use this to model motion of microscopic
particles in a fluid.
• Predefined forces.
• Sticking probabilities and secondary
particle emission.
• Typically the velocity field is pre-
computed using one of the interfaces in
the CFD or Microfluidics modules.
• Particle-particle interactions.

Particle trajectories in a
laminar static mixer
visualized using a comet
tail plot
 Mathematical Particle Tracing
• Complete freedom over the equations solved for each particle.
• Analogous to the PDE modes offered in COMSOL Multiphysics.
• Many different ways of solving the same problem, for example:

Formulation Equation of motion Charged Particle in a Magnetic Field

Lagrangian

Hamiltonian

Newtonian

Massless N/A
 Boundary Conditions
• Freeze, Stick, Disappear – Stops a particle at
the boundary.
– Freeze: Retains information about the velocity.
– Stick: Sets the velocity to zero.
– Disappear: Completely removes the particle.
• Pass through – Ignores the boundary.
• General reflection – Particles are reflected in a
user-defined direction.
 Boundary Conditions, Continued
• Bounce – Specular reflection at
surfaces.

• Diffuse scattering – Reflect

particles in a random direction
based on Knudsen’s cosine law.

• Mixed diffuse and specular

reflection – Assign a probability
of specular reflection.
 Conditional Particle-Wall Interactions
• Sticking probabilities and
expressions:
– Particles making contact with the
wall are subjected to one of two
wall conditions according to an
arbitrary expression, or a
probability (between 0 and 1).
• It is also possible to change the
value of an auxiliary dependent
variable when the particle crosses
or touches a boundary.

Animation: Particles
freeze on all walls with a
probability of 0.5,
otherwise they bounce
 Releasing Particles from Domains
• Release node: release particles using the mesh. Particle
positions can be weighted using a density function.

Mesh-based, refinement Mesh-based, refinement Density proportional to 1/r

factor = 1. factor = 2.
 Releasing Particles from a Grid
• Release from Grid node: Specify initial coordinates.
The particles are then released at a grid of points.
Releasing Particles from Boundaries
• Inlet node: release particles from a boundary. The release can
be mesh-based, weighted by a density function, or at a uniform
grid of points.

Projected plane grid in 3D Uniform release in 2D

 Initializing particle velocity
• Expression – User-defined expressions for the initial velocity components.
– At inlets, the initial velocity can be specified using tangential and normal components.
• Constant speed: spherical, hemispherical, or cone – Releases a group of particles uniformly
throughout a sphere, hemisphere, or cone in velocity space .
• Maxwellian – Particle velocity components are sampled from a Maxwellian distribution at a
user-defined temperature.
• Using kinetic energy and direction.

The Constant speed, cone setting can be used

to model sprays in devices such as CVD
chambers.
 Particle Velocity Reinitialization
• Velocity
Reinitialization: Resets
the velocity to some
expression when a
condition is met.
• Elastic Collision Force,
Monte Carlo: Particles
are randomly
scattered by a
background gas.
• Auxiliary dependent
variables can be reset.
Electrodynamic ion funnel: particles are accelerated by AC and DC fields
and collide with a buffer gas, causing them to be focused into a small area.
 Additional Particle Properties
• Using the Override Properties node, it
is possible to release multiple species
of particle in a single model.
• When releasing particles, you can
decide what species the released
particles should be.

Dielectrophoretic separation: red

blood cells are separated from
platelets using dielectrophoresis.
 Secondary Particle Emission
• Secondary particles can be
created when a particle hits a
boundary or when a certain
condition is met.
• Secondary emission is available
as a subnode to many features.

Multipaction due to secondary electron

emission in an RF device operating at a very
specific frequency
 Secondary Emission at Boundaries
• Secondary Emission
subnodes can be added to
Wall and Outlet features.
• Secondary particles are
released when particles hit
the boundaries.
• Multiple species of
secondary particle can be
released at once.
• Secondary emission can be
based on a probability or a
logical expression.
Exponential electron growth due
to secondary emission in a
photomultiplier
 Secondary Emission from Domains
• Velocity Reinitialization:
Can release secondary
particles whenever the
velocity is reinitialized.
• Elastic Collision Force,
Monte Carlo: Can
release secondary
particles whenever a
collision occurs.
• Secondary particles can
be offset from the
primary particle
location.
• Multiple species can be
released.
 Accumulators
• Accumulators can be used to communicate
information from the particle position to
dependent variables defined on the underlying
mesh.
– Domain-level Accumulators
– Boundary-level Accumulators
– Nonlocal Accumulators
 Accumulator (Domain)
• Change the values of domain
variables using each particle’s current
position or the time history of particle
positions.
• Specialized features: Particle-Field
Interaction and Fluid-Particle
Interaction.

S-bend benchmark: a comparison of the number density in the s-bend

using the Free Molecular Flow interface (bottom) and a domain-level
Body load created by a spray of Accumulator (top), assuming molecules reflect diffusely at the walls.
particles in a cross-flow.
 Accumulator (Boundary)
• Subnode to Wall and Outlet nodes.
• Changes the value of a dependent variable when particles hit walls.
• Specialized accumulator features
– Etch (Charged Particle Tracing)
– Erosion (Particle Tracing for Fluid Flow)

Erosion in a pipe elbow.

 Nonlocal Accumulator
• Communicates
information about a
particle’s current
position to the location
from which it was
released.
• Can be used to model
radiative heat transfer
between diffusely and
specularly reflecting
surfaces.
Parallel plates: Rays are released from each surface and query the
temperature of whatever surface they hit. The incident heat flux at
each plate is computed using Nonlocal Accumulators.
 Advanced Modeling Tools
• Weak contributions on
particles.
• Self-consistent particle-
field and fluid-particle
interactions.
• Auxiliary dependent
variables.
– Solve additional
equations for each
particle.
• Compatible with
Deformed Geometry
and Moving Mesh. Charge exchange cell: An auxiliary dependent variable is used to change the
• User-defined forces. charge number of a particle when a collision with a background gas occurs. As
a result, some particles are deflected by a charged plate while others are not.
 Particle Trajectory Plots
• Visualize particle trajectories using:
– Lines
– Tubes
– Points
– Comet Tails
• Color particles with arbitrary expressions.

Elastic particles suspended in a fluid due to

the acoustophoretic force. The color
Particle motion in a stirred micromixer. The particle represents the radiation potential.
streams are colored differently for the 3 different inlets
 Particle Plot
• Plot a particle variable versus time for all particles, or plot two particle variables against
each other at selected time steps.
• When plotting over time, use data series operations to compute the following quantities
over all particles:
• Average
• Sum
• RMS
• Maximum
• Minimum
• Standard deviation
• Variance

Rossler attractor (left). The average particle speed is

plotted over time (right) using a Particle plot.
Other Plot Types
Poincaré map
(red dots) in a
baffled mixer

Phase portrait of
particle speed Histogram of the ion angular energy
vs. position in a distribution function in an inductively
mixer coupled plasma. The x-axis is the angle of
incidence of the ions striking the wafer
(deg), the y-axis is the energy (eV). The
height is the relative count
 Other Visualization Tools
• Filters
• Particle evaluation
• Animate Particle Trajectories plots

All particles rendered (left) and only 10% of particles rendered (right)
 Model Library Examples
• Benchmark Models
– Ion drift velocity benchmark
• Fluid Flow Rotating galaxy: A custom particle-particle
interaction force is used to model
– Dielectrophoretic separation gravitational attraction between stars.

– Laminar mixer particle

• Tutorial Models
– Brownian motion
– Ion cyclotron motion
– Rossler attractor
– Rotating galaxy
– Trapped protons
 Model Library Examples
• Ray Tracing
– Ideal cloak
– Luneburg lens
• ACDC Module
– Electron beam divergence
– Ion funnel
– Magnetic lens Ion trajectories in a quadrupole
mass spectrometer.
– Quadrupole mass filter
– Quadrupole mass spectrometer
• Acoustics Module
– Acoustic levitator
 Model Library Examples
• CFD Module
– Micromixer particle tracing
– Thermophoresis
• Heat Transfer Module
– Parallel plates diffuse specular
• Molecular Flow Module
– Charge exchange cell Space charge distribution in a
magnetic lens.
– RF coupler
– S-bend benchmark
• Plasma Module
– Ion energy distribution function
 Brownian Motion
• Particle tracing methods
can be used to avoid t=0s t=10s
instabilities due to high
Peclet numbers.
• The Particle Tracing for
Fluid Flow interface
includes a Brownian
Force for Monte Carlo t=30s t=100s

simulations of diffusion
of particles.
Diffusion of particles in a
background fluid due to
the Brownian force.
 Trapped Protons
• Imports data about the Earth’s
magnetic field using an external library.
• Computes the mirror point latitude of
protons in the magnetic field as a
function of equatorial pitch angle.

Proton trajectory in the Earth’s magnetic field.

Magnetic field lines. Data is imported using an The proton changes direction at certain
external library. latitudes known as mirror points.
 Electron Beam Divergence
• Sets up a bidirectional coupling between
particle trajectories and the electric potential.
• Uses a time-dependent solver and stationary
solver in an iterative loop.
• The self-consistent solution agrees with the
analytical expression for the shape of a
nonrelativistic beam envelope.

Solver loop used to compute a self-consistent

solution for the electron trajectories.

A beam of electrons diverging due to its self-

potential. The particle trajectories are shown as
lines while the slice plot shows the electric
potential.
 Summary
• Flexible modeling tool
with a wide variety of
built-in forces.
• Can be used in one-way or
two-way coupling to other
physics, or used on its
own for powerful Monte
Carlo simulations.
• Dedicated postprocessing Poincaré map
tools. showing residence
time in a mixer