I am facing the problem with modified airebo potential in LAMMPS.
To change the cutoff for C-C interaction for pristine graphene 2.0 in both input file and CH.airebo file rcmin_CC. This is work fine for me and gives a good stress-strain curve for graphene.
But I face a problem when I used the same input file and modified airebo file for graphene with grain boundary. I am getting Lost atom error after a few step of pressure equilibrium. But its working well with original rcmin_CC=1.7 in original CH.airebo file. But this does not give an actual Stress-strain curve because for this there is a secondary elastic zone in the curve.
it is ill advised to change AIREBO parameters wantonly. they are a
complex set of parameters and parameter derived spline functions.
there have been extended discussions on the mailing list and github,
to get it to work correctly and reliably and often subtle details have
mattered. even changing the scaling factor 3.0 in the input file is
best not be changed, if you want reliable results.
whether a random change gives a favorable stress-strain curve is not a
good measure of the resulting accuracy.
in short. it is best to not modify anything.
You might be doing this already, but it's worth pointing out: The
modified AIREBO uses a different parameter file format.
So if you're using your own modified CH.airebo with airebo/mod, it has
no chance to work.
You need to modify CH.airebo-m in the same way you modified CH.airebo,
and use that.
If you're already doing that, I guess I have to refer you to Axel's answer