Simulation is fantastic.

I'm trying to visualise how different gasses will separate out in different closed containers. e.g. if you had a container of air with some helium and some Argon, mass 29, 4, and 40 respectively, then the helium would settle out at the top of the container, air in the middle and argon at the bottom. I'm interested in how long this process would take is it minutes hours, days or weeks.

Is there a buoyancy effect on the force on each particle or is the mass only used to calculate the elastic collisions with other particles?

Each particle will have a downward force of 9.8 x real Mass. (Real mass = Mass x 1.66054 x 10E-27 Kg)  the heavier molecules will therefore have a larger force pushing them down and will effectively settle out at the bottom first. (sorry if this is so obvious that it is already in the equations)

I think maybe the scale is so small that the particles are not given the space to settle and the proximity of the walls mean that the elastic collisions far out weigh the buoyancy differences of the gas molecules.  If we were to define each dot as a unit volume of gas, say 1000 molecules, then the model would be considerably bigger and the distance between the walls would be significantly longer reducing the effect of the elastic collisions while the downward force would be considerably higher.  in real terms 1000 molecules is a fraction of a fraction of a cm3 so easily justifiable as a modelling constraint.

Can you have a button to turn off the particle particle interactions? and see if they settle better with only particle-wall interactions.