Hello lammps users,
I have a question regarding radiation damage cascade. I have modelled Silicon Dioxide and have created a damage cascade via a PKA that I gave an initial x velocity to simulate 250, 500, 750, and 1000eV. My results seemed to be pretty linear when it came to analysing displacement/atom and mean square displacement.
My temperature is thermostatted at 300K and for all four cases equilibriate back to 300K after about 1ps.
However, when I simulate a 2000eV impact, the temperature explodes up to thousands of K, the msd explodes to 100s of angstroms (whereas for the others, msd < 1).
Can someone help explain to me why this happens? Could it be an error with my code or is it something physical that I am not properly understanding?
are you using a dynamic timestep? The higher the PKA energy and thus
velocity the smaller the timestep has to be chosen.
Without seeing your input it is just a guess, but you may be venturing
into a bad dynamics regime.
Rather than using a thermostat I suggest making the sample as big as
possible to provide for a more physical dissipation of energy.
The timestep was the problem. If I do not use the thermostat, will the temperature not explode? My simulation size is ~59,000 atoms. I am also using a barostat. Will that produce flaws in my results too?
The timestep was the problem. If I do not use the thermostat, will the
temperature not explode? My simulation size is ~59,000 atoms. I am also
2000K/60000atoms will result in a temperature increase of about 400K
if no thermostat is used. Hardly a temperature "explosion".
A strong thermostat will introduce an artificial drag/friction on the
atoms which may screw up your dynamics. As I said, if you increase
system size you can do without the thermostat.
using a barostat. Will that produce flaws in my results too?
Depends on the barostat settings. It is most likely useless for this
type of simulation. At high cascade energies you may see pressure
waves going through your simulation cell. Those are not damped by a
barostat and their wrapping at the boundaries may introduce artifacts.
Again, large samples help.
I am running cascade simulation for amorphous silica and alpha-quartz.
For both I am using PKA energy of 250, 500, 750, and 1000 eV.
For alphaquartz, the mean square displacement (msd) increases with increasing PKA energy.
For amorphuous, the msd increases with increasing PKA energy but for some reason, the 1000eV case msd is actually lower than 750eV case. I do not think the issue is my timestep since the temperature is not exploding. In fact, the temp of system after 1000eV PKA is only slightly greater than 750eV.
Could this have to do with the properties of amorphous structures or is more likely an error in my code?
Could this have to do with the properties of amorphous structures or is more
likely an error in my code?
It is unlikely to be an effect of the amorphous structure. In
amorphous materials the 'cleanest' results (i.e. good agreement with
binary collision Monte Carlo) can be expected.
It's crystalline samples that introduce additional effects such as
channeling and anisotropic displacement energies.
Have you tried a comparison with TRIM/SRIM results?