[lammps-users] ReaxFF with Hybrid pair_style

Hello,

I am trying to include the reax force field with a separate potential by creating a hybrid or hybrid/overlay pair_style:


pair_style hybrid/overlay lj/cut 10.0 reax 10.0 1.0e-6
pair_coeff 1 * lj/cut 1.0 1.0
pair_coeff 2 2 reax ffield.reax 1

LAMMPS throws the following error when I try to run a script containing these commands:

ERROR: Incorrect args for pair coefficients

So LAMMPS seems to make it past the pair_style command without a problem, but doesn’t like the second pair_coeff command I use. I’ve tested reax by itself and the hybrid/overlay and hybrid commands with other potentials and all seems to work fine. Is it possible to use the reax force field in conjunction with other potentials in LAMMPS? If not, is this simply because this has not been implemented, or instead is this not possible to do or considered bad practice?

Thanks!

William French

Hello,

I am trying to include the reax force field with a separate potential by
creating a hybrid or hybrid/overlay pair_style:

...
pair_style hybrid/overlay lj/cut 10.0 reax 10.0 1.0e-6
pair_coeff 1 * lj/cut 1.0 1.0
pair_coeff 2 2 reax ffield.reax 1

reax always requires:

pair_coeff * *

it looks that regular hybrid should do for you, or do you
want to add the LJ term to the reaxx terms?

cheers,
    axel.

Just to clarify, if I use:

pair_style hybrid lj/cut 10.0 reax 10.0 1.0e-6
pair_coeff * * reax ffield.reax 1 1
pair_coeff 1 1 lj/cut 1.0 1.0

I am essentially overwriting the use of the reax potential with the use of the lj/cut potential for the 1-1 interaction. Is that correct? If this is true, one could, for example, simulate an organometallic system by treating the metallic-metallic interaction with one type of potential, the organic-organic interaction with a separate potential, and the cross-interaction using the reactive force field. In principle, is there anything wrong or dangerous with such an approach?

Thank you,

Will

no - this is not right. Think of ReaxFF operating among
the atoms of a subset of types (or all types if not using hybrid),
in a manybody sense. If you want ReaxFF only between
atoms of type 1 and 2 and there are 3 types you would use
something like

pair_coeff * * reax ffield.reax 1 1 NULL

The other hybrid settings could specify interactions between 1-3,2-3,
and 3-3 type pairs.

Steve