Hi all,
I’ve been sifting through a lot of the LAMMPS archives on quenching quartz into amorphous silica and noticed that most of the people doing this are using either a reaxFF or BKS paramaters. I’m just wondering if it is possible to use CHARMM style paramaters and still be able to use langevin and nph aniso fixes, or will this result in significant artifacts in the surface model, particularly if I am using a siloxane terminated surface?
Doubt it. CHARMM parametrizations typically include contribution from bonded terms thus you just cannot grab the non-bonded terms and place them on their own somewhere else or combined with other alien FF terms. Obviously, using a bonded FF for you problem makes little sense as you will not be allowing bond breaking…
Sandro Scandolo and Paul Tagney developed a while ago an induced dipole potential that did much better than BKS when modeling crystalline phases of silica. Yet, I don’t think its form is implemented in Lammps. Changes are will beat BKS as well simulating amorphous phases…
Check the lit, my silica knowledge is quite outdated.
Carlos
Thanks Carlos, I guess my main issue is not with generating an amorphous silica surface so much as being able to correctly paramaterise it with CHARMM. Currently the model that I have previously used with other force-fields (Athanasopoulos, D. C.; Garofalini, S. H.; J. Chem. Phys. 1992, 97, (5), 3775-3780) would result in significantly large vdW and coul energies. Since my main purpose for using such a surface is to look at the interaction with protein and water on this surface it, when modelled with water alone it forces water molecules away from the surface, which would result in interactions with the bottom of the surface through pbc.
Thanks Carlos, I guess my main issue is not with generating an amorphous
silica surface so much as being able to correctly paramaterise it with
CHARMM. Currently the model that I have previously used with other
force-fields (Athanasopoulos, D. C.; Garofalini, S. H.; J. Chem. Phys. 1992,
97, (5), 3775-3780) would result in significantly large vdW and coul
energies. Since my main purpose for using such a surface is to look at the
interaction with protein and water on this surface it, when modelled with
water alone it forces water molecules away from the surface, which would
result in interactions with the bottom of the surface through pbc.
now, these are multiple different issues that must not be confused.
a) if you don't care about the dynamics or structural changes of the
surface itself, you can simply leave those atoms in place by not time
integrating them or not computing their forces and initializing them
to no velocity. you can them focus on modeling the interaction with
the water and the protein.
b) LAMMPS allows you to use different force fields for different
parts of a system. so all you need are suitable parameters for the
surface protein interactions
c) is your surface *supposed* to be hydrophobic or not? are you
certain that you have no typo in the force field parameters?
this is best tested by using very small test systems and comparing
with numbers computed by hand and then to other aggregate properties
that should be presented in the publication(s) you take the force
field from.
d) retaining atoms on one part of a system can be achieved through
placing a repulsive only wall into your box. depending on whether you
need to use long-range electrostatics or not, this can be simple or
require a little bit more ingenuity.
axel.
In addition to Axel’s topical review 
you may also want to check if your slab/surface has a net dipole moment oriented along the surface normal. This could introduce an artificial electric field due to the PBCs that could mess up the dynamics of your system.
Carlos
Thanks Axel and Carlos. The error was indeed with my data file and not the model (I was missing fix shake in my input). Sorry for sending you on a wild goose chase.