Skip to main content

PALADIN

PALADIN (PlasmA LAttice Dust INtegrator) follows individual microparticles through a plasma or a gas flow. It is not part of PICMC and it does not compute a plasma by itself: it takes the fields of a plasma or flow calculation that has already been done and uses them to move particles, charge them, let them collide with walls and record where they end up.

The code was written for the question that decides yield in an optical coating line: where do the particles come from that damage the layer, and what would have to change in the machine to keep them away from the substrate.

What it answers​

  • Where does the contamination come from. A particle track can be followed back to the surface element where it started. Over many particles this turns into a map of sources — target, shields, chamber wall, a flange — instead of a guess.
  • Why a particle reaches the substrate at all. The charge a microparticle collects in a plasma moves it, and the plasma is not homogeneous: near the substrate a particle is charged much more strongly than in the bulk and is accelerated towards the part. A trajectory shows that directly, and a change of geometry or pressure shows whether it still does.
  • What the contamination looks like. Not one trajectory but statistics: how many particles of which size arrive where, which force dominates along a path, what the size distribution of the adhering particles is.
  • What a design change would do. Chamber geometry, gas flow, pressure, target-to-substrate distance, magnetic field: everything that enters the fields or the wall list can be varied in the model before it is varied on the machine.

What it needs​

InputShape
GeometryA Gmsh mesh of the volume, with physical groups for every wall, inlet and outlet
Plasma or gas fieldsDensity and temperature per species, electric and magnetic field, gas velocity, pressure, temperature gradient or heat flux — cell-centred .pos files
Species propertiesMass density, permittivity, Hamaker constant, size distribution, vapour properties if the particles are droplets
Boundary conditionsWall type and elasticity per physical group, open or periodic bounding box
Initial particlesNumber, species, and either a surface they are released from or explicit positions and velocities

The fields normally come from a PICMC run or from a DSMC run of the same geometry. For atmospheric problems — room air, a pump, a ventilated chamber — the gas field comes from a CFD code instead, and only the particle part is done here.

What it delivers​

  • Trajectories with the full state of each particle per time step: position, velocity, acceleration, radius, charge, mass, kinetic energy, and the individual force contributions.
  • Wall statistics: where a particle hit, whether it adhered, how long it had been in contact, and which surface element it started from.
  • Fields of particles: dumps in Gmsh .pos format, so that a snapshot can be opened in ParaView next to the plasma fields that produced it.

Where it sits in the chain​

PALADIN is an optional branch, not a stage that every coating problem passes through. The simulation pipeline shows it as a side branch of the plasma stage: the fields of PICMC or DSMC are used unchanged, and what comes out is a statement about contamination rather than about layer thickness.

PICMC ── plasma fields ──┐
DSMC ── gas field ──────┼──► PALADIN ──► contamination on the part
CFD ── flow field ─────┘

Because the fields are read from files, the particle calculation can be repeated with a changed particle species or a changed chamber wall without repeating the plasma stage.

What it does not do​

  • No plasma solve. Density, potential and temperature are read, not computed. A case without a plasma field is a pure gas-flow problem, and then the charging switches are simply irrelevant.
  • No film growth and no optics. What a particle does to a layer once it has landed is a separate question.
  • No fluid solver. The gas field has to come from DSMC or CFD.
  • Particles are spheres with a radius, not resolved surfaces. What happens on the scale of a roughness or inside a pore is not part of the model.

Examples​

The images below are from PALADIN applications at Fraunhofer IST. Each caption names the case; the link goes to the page with the full description.

Simulated path of a particle in the EOSS coater, coloured by particle charge
Path of a single particle in the EOSS® magnetron sputter coater, coloured by its charge. The particle is repeatedly accelerated between the two targets and the substrate and is charged most strongly close to the substrate — where a collision would damage the layer. Source: Fraunhofer IST.
Charge of a particle developing over its path between the targets and the substrate
Charge of one particle along its path. The targets (grey, centre) and the substrate (yellow, top) of the EOSS® are visible. Source: Fraunhofer IST.
Charge histories of 100 particles with 10 micrometre radius
Charge histories of 100 particles with a radius of 10 µm. Only one of them gained enough charge to lift off the target — the reason why a contamination problem is a problem of a few particles and not of the average. Source: Fraunhofer IST.
Simulated particle motion in a hospital room with a ventilation system
The same particle model outside coating: motion of aerosol particles in a hospital room with a ventilation system. Source: Fraunhofer IST.

More on these applications:

Availability​

PALADIN is a separate program from PICMC and is licensed separately by Fraunhofer IST. It is not part of the PICMC delivery, and the source is not distributed with this documentation. For a licence, a feasibility question or a joint project, contact the Simulation and Digital Services group at Fraunhofer IST — Dennis Barton for applications, Philipp Schulz as the author of the code.

If PALADIN is installed at your site, this documentation describes the version that is in the repository today: the Python-controlled code with MPI parallelisation. Older descriptions that start a run from a parameter file and a code called submit_emdi refer to a version that is no longer maintained and do not apply.

Reference​

The physical model, the derivation of the forces and currents, and the EOSS® study are documented in:

Schulz, Paul Philipp: Model-based examination of dust particle contamination in plasma coating processes. Dissertation, Technische Universität Braunschweig, 2022. Fraunhofer Verlag, 2023. ISBN 978-3-8396-1886-8. Open access: doi:10.24406/publica-555

In this section​

PageContent
InstallationBuild requirements, MPI, build and install, environment variables
Building a scenarioThe Python scenario: parameters, mesh, borders, fields, species, running a case, output
Particle modelWhat acts on a particle, and which switch turns it on
ParametersEvery key of the parameter object, grouped by task
Constants and unitsUnits, physical constants, wall types, distributions, dump quantities