[lammps-users] Issue with Tersoff potential

Hi Steve,
In order to simulate multicomponent system using Tersoff potential, variable Chi is needed. Chi represents the bond strength between atom type i and j. It is 1 when i=j but not necessarily 1 otherwise. The current Tersoff code doen’t have this variable as an input nor it is there in the code. Is this the case?
Thank you,

For the benefit of the list, here is Aidan's response to Rutuparna:


I think your question about the parameter Chi is answered by the following
response to a previous query. Let me know if this is insufficient.

> We wrote the code so that *the user* could use many different parameter sets.
> The choice of parameters is up to you. The file si.tersoff contains the
> parameters from Tersoff's 1988 paper, but LAMMPS can handle many other
> parameter (we deliberately did not choose any mixing rule convention; all
> combinations of elements mus tbe explcitily specified). As a result, I think
> it should be possible to use the existing code to run Tersoff's 1989 force
> field for SiC. You just need to transorm some of the parameters using
> elementary algebra and compute the cross-parameters according to Tersoff's
> mixing rules. If you did this, perhaps you could send us the force field file.

I'm not an expert here, but looking at the formulae of both forms of
Tersoff potential, it seems that it's not possible to get T89 (Phys
Rev B 39 5566) as a parametrization of the LAMMPS / Tersoff '88

More specifically, if Chi is not equal to unity, you should scale B by Chi
in the parameter file.

in T88:
b = (1 + Beta^n * zeta^n)^(-1/2n)

in T89
b = chi * (1 + Beta^n * zeta^n)^(-1/2n)

where zeta is a function of atomic positions.
Scaling Beta will not help, that's why the additional parameter was introduced.