problem with flying atoms in Si crystal


I am trying to create a silicon cantilever beam of dimensions L=217.2, w=86.88, h=86.88 (in Angstroms)
However one or two atoms from the corners of the beam free end are flying away.
I tried some different values of length ‘L’ and found that at ‘L = 217’ no atoms are flying.

Is there a better way to fix this problem (OR)
doing trail and error with the dimensions of the beam is the only way to fix this problem?

I am using Tersoff potential.


It may be that atoms at corners are not tightly bound by the Tersoff
potential. Sharp corners don’t tend to exist in nature.