[lammps-users] initial charge distribution used for reax

LAMMPS LAMMPS Mailing List Mirror
1

Dear all,

I’m simulating a CHON system using reax package, which requires to set the charge of every atoms in the system. but how can I know the initial atomic charge of the structure I used for simulation?

best regards

Shenglong

2

ReaxFF will re-assign the charges so it doesn't really matter.
Set them to values close to what you expect them to be.

Steve

3

Prof. Steve Plimpton

Thanks for the reply. But I have not idea what the charge distribution should be. My simulation last quite long time.(~30 ps) Will the initial set of charge effect the finial result? I don’t really understand how the re-assignment works.Will it make the system unreal and unphysical if I set the charge zero for all atoms?

Best regards

Shenglong

4

If you don't know what to set it to, then let ReaxFF do it.
I don't think it matters what you set the initial charges to.

Steve

5

Prof. Steve Plimpton

Thanks for the reply. But I have not idea what the charge distribution

just run a quick and dirty quantum chemistry calculation
and take the mulliken charges. that will be a reasonable
guess for any simple molecule.

should be. My simulation last quite long time.(~30 ps) Will the initial set
of charge effect the finial result? I don't really understand how the
re-assignment works.Will it make the system unreal and unphysical if I set
the charge zero for all atoms?

the charge equilibration is very obviously going to
mimic the redistribution of the electron density that
is essentially behind all chemical reactions (only that
it is calculated based on quantum chemistry, whereas
reaxx tries to emulate this with an effectively empirical
model with simple parameters). on a more general note,
you should _never_ use a method without knowing how
the essential principles work. how do you expect to be
able to tell whether the calculation is going correctly or
not? at the very least you should start with a simulation
of a system where you _do_ know the results and that is
ideally somewhat similar to what you so you can cross-
check whether your estimates and settings are working.

cheers,
    axel.