[lammps-users] same group-->two regions --> two potentials

Hi

Thanks for your answers.

Actually, what I am trying to simulate is the growth of nanotubes on a Fe cluster in an atmosphere of vaporized C. I use a Teresoff potential for the C-C interaction. The problem is that in "real" conditions the density of C atoms is low and you would need a huge computational box and very long times of simulations. A way to bypass this issue is to artificially increase the density of C in the box. There is, however, a problem because now the C atoms in the atmosphere are too close and they start to interact each other before reaching the cluster. Thus, the trick is to use a simple Lennard Jones for the C-C not on the cluster just to prevent them from bonding together.
Basically, I suppose if LAMMPS has any way to suppress C-C aggregation in some regions of the space it would do the job. Is this possible?

Ciao

Alessio

Hi

ciao alessio,

Basically, I suppose if LAMMPS has any way to suppress C-C aggregation in some regions of the space it would do the job. Is this possible?

i am not aware of any current feature that would allow this, but i can
see your point.
i still believe that you have to make this transition smooth or else
you'll get strange
effects due to the discontinuity at the region boundary.

what i would suggest to do would be to implement a variant of the LJ
potential that would
have a position dependent epsilon, set sigma and cutoff so that it
would be only repulsive
and keep the atoms sufficiently apart and then use a hybrid overlay
construct to add it to
the "real" potential. this should be rather straightforward to
implement and would hopefully
be generic enough to be reused for other purposes as well.

in a simplest case i would implement a z-axis ramped epsilon via tanh() for
a smooth transition area.

hope that helps,
    axel.

I would probably write a fix that changed the types
of C atoms once they become part of a nanotube. Then
they wouldn't interact with each other while in gas phase.

Steve