Dear all,
I am investigating collision cascades in bcc Fe. The crystal has an open surface that is bombarded with an atom with a kinetic energy of 10 keV. This atom comes from outside the crystal. The simulation box is thus larger than the crystal so the projectile can be initially located outside the crystal.
Boundary conditions are ppf (f in the direction of the open surface).
By suggestion of Axel, I use an inner and an outer thermostat. The inner thermostat is composed of 2 atomic layers maintained at 300 K with fix Berendsen and a time constant of 0.1. The outer thermostat is composed of two immobile atomic layers with velocities maintained at 0and T=0 K. I use adaptive time with fix adapt/dt 0.1.
I calculate the vacancies and interstitials that form using Wigner-Seitz analysis (Ovito). I was observing the evolution of SIA-V as the cascade unfolds and observed a strange effect. After the number of defects reaches a maximum, their number starts to decrease by recombination. After some ps, the number remains constant. All this is expected. But suddenly, after a longer time, the number abruptly increases. Ovito shows that a large number of defects appears in the box.
If the number of defects has reached a stationary value, it is clear that defects cannot appear by magic. Thus, it must be some artefacts. Any idea why this can occur? Could this be due to the shockwave that reflects on boundaries? If so, how can this be avoided?
Many thanks in advance for any help you could bring me.
Best regards,
Christophe
What you are calling the outer thermostat appears to just
be frozen atoms, i.e. an atomic wall. You certainly could
get energy reflection off such a wall. Whether that
is producing what Ovito tags as defects at some
later time (after reflection) is unclear. You would
probably need to visualize your system by coloring
atoms by kinetic energy (or the like) to see the time
evolution of how energy is dispersed in your system.
If the reflecting boundaries are the problem, you
could look in the literature for shock simulations
with absorbing boundaries. There are methods
(more complicated than what you are doing)
for using boundary atoms that can absorb incident
energy. Or you can simply not run your simulation
long enough that the energy reaches the boundary
and reflects.
Steve
What you are calling the outer thermostat appears to just
be frozen atoms, i.e. an atomic wall. You certainly could
get energy reflection off such a wall.
I was using such thermostat from a suggestion from Axel, some time ago. In his mail he said that this type of wall would help squashing incident phonons. I thought thus that this would avoid reflection.
Whether that
is producing what Ovito tags as defects at some
later time (after reflection) is unclear.
In Ovito I identify defects using the Wigner-Seitz analysis. If occupancy=0, it is a vacancy, if occupancy= 2 (in Fe) it is an SIA.
You would
probably need to visualize your system by coloring
atoms by kinetic energy (or the like) to see the time
evolution of how energy is dispersed in your system.
Ok, I shall try that.
If the reflecting boundaries are the problem, you
could look in the literature for shock simulations
with absorbing boundaries. There are methods
(more complicated than what you are doing)
for using boundary atoms that can absorb incident
energy.
Thanks for the advice.
Or you can simply not run your simulation
long enough that the energy reaches the boundary
and reflects.
That is an idea but difficult to know at which time this occur and I cannot inspect manually each cascade. It would take too much time.
Christophe
If your simulation box is big enough I think it’s pretty
unlikely that a cascade model will reflect enough
energy at the boundaries to create new defects.
That is more typical of shocked systems where absorbing
boundaries can be needed. Normal behavior in
a cascade model is simply that the overall system
thermalizes to a slightly higher temperature based
on the small energy (averaged over all the atoms)
you have added to the system via the incident
knock-on event. And the thermostat should dissipate
that.
Steve