the hybrid of tersoff/zbl and airebo

LAMMPS LAMMPS Mailing List Mirror
1

Hello everyone

I want to do the simple simulation about the collision of Si and graphene. I want to use C-Si
parameters from the tersoff/zbl, the C-C from the airebo. I use the command as follows:

pair_style hybrid tersoff/zbl airebo 3.0

pair_coeff * * tersoff/zbl SiC.tersoff.zbl C Si
pair_coeff 1 1 none
pair_coeff * * airebo CH.airebo C NULL

​it can run in lammps​, but I find the results are sort of wrong. So I search the lammps website, I find it writes as follows:

​pair_style hybrid/overlay tersoff/zbl airebo 3.0​

pair_coeff * * tersoff/zbl SiC.tersoff C Si
pair_coeff * * airebo CH.airebo C NULL
(Note that to prevent the Tersoff potential from computing C/C interactions, you would need to modify the SiC.tersoff file to turn off C/C interaction, i.e. by setting the appropriate coefficients to 0.0.)

I donot know how to go to the SiC.tersoff file to turn off the interaction of C-C. Could you help me to change the SiC.tersoff file turning off the interaction of C-C? Or other ways to simulate my model. Thank you very much!

1 Like
2

Aidan or Ray can likely tell you. The format of the Tersoff file

is described on the tersoff doc page.

Steve

3

I want to do the simple simulation about the collision of Si and graphene.

That does not sound at all simple to me. I think we should add a line
to the first sentence of the doc page for pair hybrid, as follows:

"The hybrid and hybrid/overlay styles enable the use of multiple pair
styles in one simulation. They do not yet perform miracles."

Seriously, there are fundamental theoretical issues with defining
manybody potentials that operate on subsets of atoms. hybrid and
hybrid/overlay implement two simple combining rules, but these may or
may not correspond to good physical/chemical models. In addition to
the fundamental theoretical difficulties, it is in practice
non-trivial to tell exactly what model a particular set of
pair_style/pair_coeff/neigh_modify exclude commands are implementing.

Therefore, before shooting off a bunch of simulations, it is a good
idea to interrogate the model that you have created by examining some
simple cases e.g. the energy landscape sampled by a silicon atom
moving on a graphene substrate.

With all those caveats, I can tell you what I think the two models
below are doing.

>> pair_style hybrid tersoff/zbl airebo 3.0

pair_coeff * * tersoff/zbl SiC.tersoff.zbl C Si
pair_coeff 1 1 none
pair_coeff * * airebo CH.airebo C NULL

"With the hybrid style, exactly one pair style is assigned to each
pair of atom types."
In the absence of an Si atom, this model is just AIREBO for graphene.
When an Si atom is present, the Si energy will be based on the
Tersoff/ZBL model of Si interacting with graphene, which is good.
However, the C atom energies will be based on the Tersoff model for a
set of isolated C-Si dimers in vacuum. These carbon energies will
then be added to the AIREBO energy for the pure graphene sheet. Is
this good or bad? Hard to say.

pair_style hybrid/overlay tersoff/zbl airebo 3.0
#pair_coeff * * tersoff/zbl SiC.tersoff C Si # this won't work
pair_coeff * * tersoff/zbl SiC.tersoff.zbl C Si
pair_coeff * * airebo CH.airebo C NULL

"With the hybrid/overlay style, one or more pair styles can be
assigned to each pair of atom types."

Let's assume that you have modified SiC.tersoff.zbl or used
neigh_modify exclude to turn off the C-C energy. There are multiple
ways to do this. You should verify correctness by running the modified
Tersoff/ZBL potential by itself with pure graphene. The energy should
be zero. Also verify that the Tersoff/ZBL energies for pure Si and a
single C in Si are unchanged.

The key difference between the hybrid/overlay and the hybrid example
is that instead of replacing the C-C Tersoff/ZBL interaction with
AIREBO, it adds them together. Since we have set the Tersoff/ZBL
interaction to zero, the result is identical in both cases, but
clearly the hybrid version is more straightforward.

Aidan