Discussion Forum

I'm running a model the calculate the trajectories of electrons emitted from a filament and accelerated to an ionization chamber by the electric field produced by a voltage bias between the chamber the filament is in and the ionization chamber. The idea is to vary the distance from the filament to the ionization chamber and the voltage bias to see how the electron trajectories react. I have the charged particle tracing interface set up to uniformly produce 10000 electrons across all faces of the filament. For each computation that I've run, some number (that varies with each combination of the above parameters, but not in any pattern that I can understand) of the electrons seemingly disappear once the time becomes greater than 0. The total number of particles in the geometry stays constant at 10000, but the number of particles in the regions that I am interested in ( and the only regions the particles could possibly have gotten to in 0.06 ns) drops by between one third and two thirds. For instance, my last run had a voltage bias of 220V and a filament distance of less than two inches, and the number of particles in the relevant areas drops from 10000 at 0 microseconds to 3741 at 6e-5 microseconds for no reason that's apparent to me. I played with different wall conditions on the filament, but none made this strange disappearance go away. When looking at a 1000 particle sample of the particles' positions at 6e-5 microseconds (and the same particles at 0.03 microseconds, the max time value of the study), a good number ( probably about proportional to the number of missing particles) have NaN as their position making it seem like the particles were removed from the system. Those particles with NaN positions had real valued coordinates at 0 seconds.

I'd greatly appreciate it if anyone has a solution to this problem, or at least some ideas as to why this is happening.