Particle model
PALADIN is a Lagrangian code: every particle is followed individually, and all quantities that act on it are evaluated at its current position from the fields that were read at start-up. This page describes what those quantities are and which parameter switches them on. The derivation — moments of the Maxwell-Boltzmann distribution, the resulting force and current expressions, and their validation — is in the dissertation and is not repeated here.
Time integration
The particle is advanced with a fixed time step DT, which is a physical time in seconds. The
step has to be short enough for two things: the particle must not cross a cell per step (the
fields are interpolated per cell, so a large step means the particle sees an averaged field), and
the charging must not overshoot — a particle that gains its whole equilibrium charge within one
step oscillates.
Note: MINOR_TIME_STEP_COUNT subdivides DT internally. It helps with the stability of
the integration, but it does not change the output interval and it does not make a coarse
spatial resolution finer.
Charging and force evaluation are per particle and per step; there is no particle-particle interaction included in the force sum except the cohesion of particles that are in contact.
Fields the particle sees
All fields are cell-centred and are interpolated at the particle position. A scalar, a vector, its square norm, its gradient and its Hessian are all available, which is what the different force terms need — the polarisation force needs a field gradient, the electric force only the field itself.
If a field was read from a file, a cell value of zero can be replaced by the constant that was
given alongside it (default_if_zero). That is how a plasma field that only covers part of the
volume keeps a sensible neutral-gas temperature outside.
Forces
Every force has its own switch in the parameter object, and the dump can write the contribution of each one separately — which is the practical way to find out what dominates a trajectory.
| Switch | Force | What it needs |
|---|---|---|
USE_GRAVITY | Weight | GRAVITY |
USE_ELECTRIC_FORCE | Coulomb force qE | ELECTRIC |
USE_MAGNETIC_FORCE | Lorentz force qv×B | MAGNETIC |
USE_POLARIZATION_FORCE | Force on the induced dipole in a field gradient | ELECTRIC, relative_permittivity |
USE_DEBYE_POLARIZATION_FORCE | Force from the polarised ion cloud around a charged particle | plasma species |
USE_COLLISIONAL_DRAG | Friction with the neutral gas | gas species, pressure, temperature |
USE_COULOMB_DRAG | Ion drag from Coulomb collisions with the plasma | plasma species, charge |
USE_COLLISIONAL_THERMOPHORESIS | Force along a temperature gradient in the neutral gas | gas temperature gradient |
USE_COULOMB_THERMOPHORESIS | The same for the plasma species | plasma temperature gradient |
Default state: gravity, electric force, magnetic force, both drag terms and both thermophoresis terms on; polarisation and Debye polarisation off. That is the set for a micrometre particle in a low-pressure plasma: the neutral drag and the electric force dominate, the magnetic force is negligible for a slow, heavy particle unless the field is strong, and the polarisation terms only matter for a dielectric particle in a strong field gradient.
The two thermophoresis and two drag terms are deliberately separate: a particle that is charged interacts with the plasma through Coulomb collisions, and a particle that is neutral only through hard collisions with gas molecules. Whether both are relevant depends on the ionisation degree.
Charging
The charge of a particle is a state variable and is written in units of the elementary charge. In a plasma it reaches thousands to tens of thousands of elementary charges for a micrometre particle, and it follows the local plasma rather than the distance to a wall.
Charging is computed from the currents that each plasma species delivers to the particle surface. For every species the code evaluates the flux of ions or electrons onto the sphere in the presence of the particle's own charge and of the drift of the species through the gas, with a shielding length that combines the Debye length with a term from the directed motion. The result is a current per species, and the sum over species moves the charge one step forward.
| Switch | Effect |
|---|---|
USE_PARTICLE_CURRENTS | Turn the current-based charging on at all. Without it the charge stays at its initial value |
MORE_ACCURATE_COULOMB_FORCE_CALCULATION | Use the more expensive evaluation of the Coulomb interaction |
DELETE_PARTICLES_IF_INVALID_CHARGE | Remove a particle whose charge became unphysical instead of carrying the value through the run |
MINIMUM_PARTICLE_SIZE | Lower limit for the radius of an evaporating droplet, see below |
Warning: the charging needs at least one charged plasma species with a density and a temperature. The species lists are not validated for physical consistency — an electron density without ions charges the particle to a potential that no plasma would produce, and the run looks perfectly healthy while doing so.
What the charge is used for is everything downstream: the Coulomb force, the ion drag, the Coulomb thermophoresis, the Debye polarisation force, and the currents again in the next step. A particle that is charged in a region with a field is therefore also accelerated there, which is the mechanism behind the EOSS® observation on the overview page.
Drag and thermophoresis in two regimes
Both the neutral drag and the neutral thermophoresis depend on the Knudsen number — the ratio of the mean free path to the particle radius. At low pressure the particle is in the free molecular regime and the momentum transfer is a sum over single molecule impacts; at higher pressure the gas behaves as a continuum around the particle and the force follows the Stokes form with a slip correction.
LOW_KNUDSEN_NEUTRAL_GAS_CALCULATION selects the continuum form for the neutral gas terms. It
is off by default, which is right for a sputter chamber at 10⁻³ to 10⁻¹ mbar and wrong for an
atmospheric case.
REFLECTION_COEFFICIENT belongs to the same family of terms: it describes how a gas molecule
leaves the particle surface and therefore how much tangential momentum it takes away. The
default of 1.0 means specular reflection; a value of 0 corresponds to full accommodation. If
the literature values for your gas-particle combination are available, use them here — the
drag term scales with this coefficient directly.
REFERENCE_TEMPERATURE and REFERENCE_PRESSURE are the state at which the gas properties
(viscosity, thermal conductivity) are evaluated before they are scaled to the local
temperature. They do not change the physical setup of the case, only the property model.
Surfaces: reflection, adhesion, rolling
A wall collision is decided in two steps. First the particle bounces with the restitution given
by the border's elasticity. Then an energy balance is evaluated for the normal component: the
kinetic energy after the impact against the van der Waals well of the surface, which is built
from the Hamaker constant of the particle species and the minimum wall distance.
- If the remaining energy is above a threshold, the particle is reflected with the reduced normal velocity.
- Otherwise it sticks. From then on the time since sticking is recorded, and the particle can roll along the surface under the tangential force balance.
That is why elasticity and hamaker_constant together decide whether a surface collects
particles or not. A perfectly elastic wall (elasticity = 1.0) has no energy loss, so the
sticking criterion is reached only by the van der Waals term — with the default Hamaker constant
of 10⁻¹⁹ J a micrometre particle usually bounces, a sub-micrometre particle usually sticks.
| Switch | Effect |
|---|---|
CHECK_FOR_WALLS | Test whether a particle is inside a wall volume, not only whether it crossed a surface |
CHECK_ALL_SURFACE_ANGLES_FOR_ROLLING_PARTICLES | For a rolling particle, check all surface normals of the element instead of only the current one |
WALL_MODULUS | Parity setting for the inside-the-wall test; flip it for a mesh whose surface orientation is inverted |
WALL_MODULUS exists because the inside/outside test counts how often a ray from the particle
to the boundary crosses the surface. For a mesh with correctly oriented normals the default of
1 is right; if a particle that is clearly outside is treated as being inside a wall, the mesh
orientation is the cause.
Cohesion and agglomerates
CALCULATE_MESH_CONNECTIONS (on by default) builds a connectivity graph of all particles that
touch each other, so that a cluster moves as one body instead of as independent spheres. The
graph is computed once and cached in a connection_caches directory inside the run directory.
For a contamination study this is the difference between a cloud of single particles and a realistic agglomerate that falls, hits the substrate and stays there.
Droplets: evaporation and condensation
A particle species can carry vapour properties: saturation pressure, latent heat, diffusivity, molar fraction and specific heat of the condensable component, plus the relative humidity of the surrounding gas species. With those, the radius of the particle changes along its path — it evaporates in dry gas and takes up vapour in a saturated one — and the heat of condensation changes its temperature.
The radius is changed by the mass flux to or from the surface, evaluated with a Nusselt number for mass transfer; the temperature follows from the heat flux with the Nusselt number for heat transfer and the latent heat term. Two knobs control the numerical side:
TEMPERATURE_DIFFERENCE_FOR_INSTANT_EQUILIBRIUM— if the computed temperature change per step exceeds this value, the particle is set to the gas temperature instead of being integrated. This avoids an oscillation of the temperature in a fast-changing environment.MINIMUM_PARTICLE_SIZE— the radius at which the droplet is set back to its initial radius and temperature. With the default of 0 the limit is the radius of the non-volatile core, so a droplet ends as its dry core and starts over as a fresh droplet.
This part of the model is why the same code is used for aerosol problems outside coating, where the carrier phase is air with a humidity and the particles are droplets.
Where the equations are
The force and current expressions, the shielding length, the collision model and the validation against the EOSS® coater are in the dissertation cited on the overview page. The switches on this page are named exactly as in the source, so the mapping from a parameter to a term in the thesis is direct.