Help needed with REAXFF simulation.

Hi,
I am using the REAXFF potential to add O molecules atop a Si substrate at 300K to create a few monolayers of Oxygen atoms atop the substrate. As expected, when the O2 molecule touches the Si surface, it breaks up and the atoms try to fly away. In my setup, the atoms seem to be so thermal excited that they fly out of the box in the second timestep of the simulation.

I have a thermal bath group of atoms which I rescale every step to absorb the head generated atop the substrate surface when the O molecule hits. Could somebody please advise what may be causing this and how I may control the atoms from flying away from the box. When I use periodic boundary conditions, they reenter the box and penetrate the fixed layers at the bottom of the subtrate. When I used p p f,I lose the 2 atoms.

Script:

units real
atom_style charge

read_data ConfigAtomsproperslabnew.txt

pair_style reax
pair_coeff * * ffield.reax 6 3

group fixed id <> 1 288
group moving id <> 289 1154
group rescale id <> 289 792
group top id <> 793 1152
group adatoms id <> 1153 1154

timestep 0.5
compute mytemp top temp
neighbor 0.5 bin
neigh_modify every 10 delay 0 check no

fix 1 moving nve
fix 2 rescale temp/rescale 1 300.0 300.0 10.0 1.0

dump FinalCoordsandVel all custom 100 dump300K.startingall id type x y z vx vy vz
dump FinalCoords all xyz 100 dump300K.Startingxyz.xyz

thermo_style custom step temp press pe etotal
thermo_modify temp mytemp
thermo 100

run 4000

The masses are being read from the atom coords file.

Any help would be appreciated.

Regards,
Saketh.

Have you monitored the velocities and forces on the atoms that are
flying out of the box (e.g. shrink wrap it for now with "s" instead of "f").

If they are huge (which they probably are), then you are creating
a bogus system. You can't thermostat bad systems and expect
them to become good.

Steve