Reaxff with NPT and Silicon parameters in Reaxff

Dear all,

I try to simulate Silicon Carbide (SiC) with Reaxff potential. I found that there is no potential parameter file considering Si-C bond provided in the Lammps potential directory. Even though I found a series of parameter in a paper by Chenoweth et al., 2005. , 2005, 127 (19), pp 7192–7202. I just copy there supplemental information to make a potential file. The average energy per atom is about -6.8 eve for SiC. I wonder whether someone simulate SiC with that potential? Furthermore, is there any tested potential for SiO2 and SiC with Reaxff?

I tried to use NPT to control the pressure for a structure described with Reaxff, but always encountered “Segmentation fault” error information. I tried several different potential and also changed the atom types both to “C”, and still had that problem. Is there anyone can help me about this?

(sic_reax.lam is the data file for one SiC unit cell
in.reaxc is the input file for lammps
ffield.reax is the default parameters file provided by lammps
ffield.reax.sic is the supplemental information from the 2005 paper



ffield.reax.budzienSi (21.6 KB)

ffield.reax.sic (7.6 KB)

in.reaxc (462 Bytes)

sic_reax.lam (2.27 KB)

If you've created your own potential file and want
to verify it is good for you model, you are on your
own. No one else is going to do that for you.

Re: the NPT problem. Have you verified you
can run your system stably with NVE and NVT
before trying NPT? Have you tried running NPT
with one of the ReaxFF examples scripts in examples/reax?
Those are good preliminary steps to diagnose your
problem, which likely isn't about ReaxFF, but about
your use of fix npt and the geometry of your model.


Thanks a lot.

I will try to read related literatures about reaxff for more knowledge.

As for the NPT control, it seems something goes wrong for my input file or data file. As Ray helped me, it seems that the NPT parameters needs more carefully chosen during the simulations.

Thanks again.



2012/2/13 Steve Plimpton <sjplimp@…92…>