lammps crash with AIREBO CH potential

Dear lammps users,

We are currently experiencing some severe issues with lammps and the AIREBO potential when trying to run MD simulations on CH systems with high hydrogen content. As an example, starting from an amorphous hydrogenated carbon system with H/C ratio around 2 and relatively low density (1g/cm3), well equilibrated at 100K, we tried to increase the temperature gradually from 100K to 5000K with a reasonably low rate of 24.5K/ps. Our system behaved perfectly up to a temperature of 270 K at which it crashed (nan in all thermo data). We carefully checked the timestep (0.2 fs) and the thermostat parameters by comparing with an implementation of AIREBO in a serial in-house MD code which worked perfectly fine. As can be seen in the log file, potential and kinetic energies nicely oscillate around slowly increasing average values up to the moment at which the system crashes. We also checked the interatomic distances and all of them were absolutely reasonable.

Interestingly we only observed crashes with lammps for systems containing hydrogen. We read in the bug fixes pages that some bug regarding the attribution of C and H potential parameters has been fixed in October 2015. However, our current simulations performed with the last version of lammps (mar 2016) suggests that this issue might not be entirely fixed.

Another indication of this is that the state equations of alkanes obtained with lammps/AIREBO considerably underestimate the density at fixed P&T with respect to both our in-house code and former results with AIREBO published by S. J. Stuart (Liu & Stuart, J. Comput Chem., 2008).

Could anybody help us with this, as we definitely need a parallel MD code to run our project.

Thanks in advance,

Nicolas Capit

Here is the input file and an extract from the log:

LAMMPS input file

processors * * *

units metal

boundary p p p

atom_style charge

pair_style airebo 3.0

read_restart dataflex.restart.20000

group C type 2

group H type 1

group ALL type 1 2

pair_coeff * * /home/capit/src/CH.airebo H C

fix 1 ALL nvt temp 100 5000 0.005

timestep 0.0002

thermo 1

thermo_style custom step etotal temp pe ke

thermo_modify flush yes

dump 2 ALL atom 50 melting.lammpstrj.gz

dump_modify 2 flush yes sort id

restart 10000 melting.restart

run 1000000

Log extract :

55042 -63045.676000 272.029360 -63663.658000 617.981900

55043 -63045.368000 272.244760 -63663.839000 618.471230

55044 -63044.967000 272.408280 -63663.810000 618.842700

55045 -63044.574000 272.525540 -63663.683000 619.109100

55046 -63044.194000 272.608200 -63663.491000 619.296870

55047 -63043.821000 272.664120 -63663.245000 619.423900

55048 -63043.464000 272.707350 -63662.986000 619.522120

55049 -63043.121000 272.750450 -63662.741000 619.620040

55050 nan nan nan nan

55051 nan nan nan nan