Tersoff and LJ hybrid

Hi All,

I tried to describe the interaction of C atoms of a nanotube with both
Tersoff and LJ potential, and made the commands as follows (units

pair_style hybrid/overlay tersoff lj/cut 10.0
pair_coeff * * tersoff SiC.tersoff C
pair_coeff * * lj/cut 4.2038e-3 3.37

however, as I began to relax the system under nvt, it showed an error
of atom loss, or if I instead minimize the system, atoms tended to
expand in disorder.When I previously used other potentils, for
instance, Tersoff only, or used AIREBO only, all the things went well.
So is there any problem with these coefficients for hybrid style? By
the way, I used the Nov 2010 version.

Thanks in advance.


First, I would use the current version.
Second, your epsilon is tiny, so the LJ term
is essentially repulsive, consistent with
expansion. Third if you are losing atoms,
something else is wrong, like your timestep
or reneighboring criteria, or initial config.