[lammps-users] AIREBO Bug

Hello LAMMPS users,

There seems to be a bug in the lammps version AIREBO potential that shows up when heating structures to high temperatures and when simulating amorphous carbon generated using other potentials. I suspect the potential might have issues in other situations where reactivity is important, though I have not checked this. I noticed that this had been reported to the list before and wanted to confirm that is its limited to the LAMMPS version of AIREBO and not the potential itself.

The problem does not show up using various versions of Steve Stuart’s AIREBO code or Don Brenner’s code for the 2nd Generation Brenner potential. The bug can be reproduced using the attached LAMMPS script. Essentially the simulation runs well for a time, then explodes spectacularly. Running the same script with the Tersoff potential gives good results.

melt test

dimension 3
boundary p p p

atom_style atomic
units metal
lattice diamond 3.57
region box block 0 5 0 5 0 5
create_box 1 box
create_atoms 1 box
mass 1 12.011

neighbor 0.5 bin

#define pair styles
pair_style airebo 3.0 0 0
#pair_style tersoff
pair_coeff * * CH.airebo C
#pair_coeff * * SiC.tersoff C

set initial velocity

velocity all create 8000.0 123456789 dist gaussian units box

set fixes

fix 1 all nve

thermo 1

timestep 0.0001
dump 1 all xyz 100 c100.xyz

fix 3 all temp/berendsen 8000. 8000. 0.005
run 25000
write_restart heat_1.restart

I'll pass this along to Steve Stuart (CCd). His
group has been looking at the AIREBO in LAMMPS and comparing
it to their code. The latest I heard from him, all was in basic agreement
except for a few minor tweaks which we'll add to LAMMPS at some point.
What you describe sounds more major, so he'd want to look at it,
esp if you have a test case. Dave - you can contact him directly
with more details ...