question on NEB with vacancy

I would like to estimate vacancy diffusivity in a crystalline material, using NEB. Because I am not sure of the mechanism, I would like to model the vacancy as an atom with no interactions, so I can easily test a variety of reaction pathways. Is this feasible in LAMMPS? Does it make sense? If I use the lj_cut pair coeff with zeros for all the coefficients, will this work?

pair_style lj/cut 2.5
pair_coeff 3 * 0 0 0 
where 3 is the vacancy atom type and 2.5 is an arbitrary cut-off. 


Michael Shaughnessy
R&D Engineer, Synopsys

No, that won't work. A particle with no interactions will essentially
be ignored by NEB. A vacancy is actually represented implicitly be
the nearby atoms. Nearby in this case means all atoms whose positions
change "significantly" when the vacancy undergoes a hop. So, if you
include all of these atoms in your final state file, NEB should work.
One sticky issue is that even though initial and final states are
equivalent, modulo a lattice shift, there can be more than one
non-equivalent mapping of initial atoms to final atoms. If so, you
should try to enumerate all non-equivalent mappings and run NEB for
each for each of them.

Hi Aidan,
Thanks for the reply. Won’t the inter-image spring forces still operate on the non-interacting particle? I guess the easiest thing to do is just try it.


Yes, the spring forces will still operate, so your vacancy particle
will form an equally-spaced chain in the replicas, independent of what
the real particles are doing.