please *always* copy the mailing list on your replies. thanks.
dear Axel
my system has graphene oxide+water and it is periodic,and i can not found
mulliken charge for graphene oxide with gaussian or orca ,and i want to use
finding suitable force field parameters for a system involves more
effort than computing mulliken charges. in fact, for your kind of
system, i would expect the mulliken charges a very bad choice due to
the way how they are defined. in any case, usually you need to do some
fitting to make the point charge distribution of your system match the
distributed charge distribution computed from a quantum chemistry
calculation. most force field parameterization schemes involve some
kind of electrostatic potential fitting procedure for that purpose.
also the interaction between your molecule of interest and your
solvent (water) needs to be adapted. there are extensive
documents/publications describing the procedure for different force
fields. you may also search the literature, if some publications
relevant to your system of interest exist. since graphene oxide is
quite a popular research topic, there should be something usable.
pair_style lj/charmm/coul/charmm command. now, can i use fix qeq/reax for
graphefne oxide?
if no, my simulation is not reactive , can i use reax pair_style?
well, technically yes. however, same as for conventional force
fields, you have to use it with a suitable parameterization that is
adapted for your system of interest. by construction (reaxff is based
on using DFT calculations to determine the parameters) it may not be
giving as good a representation of water than tip4p, since regular DFT
itself has problems to model bulk water well. particularly the
handling of long range electrostatics may not be to your liking.
mind you, reax does bonding implicitly and thus you have to remove all
topology information and you cannot easily prohibit that reactions are
happening, in case they are favorable based on the force field
parameters.
in both cases, you cannot get around proper validation (best done with
small subsystems) of the choice of parameters and force field.
to summarize, you not only need *some* numbers, but you need *proper*
parameters for your input and rather than getting them in some
automated fashion you will have to verify and validate everything.
otherwise you need to use some kind of quantum chemical method, with
the drawback of being significantly more demanding in terms of
computation. yet, even in that case, there are plenty of parameters
and input data that need to be carefully checked and validated.
axel.