[lammps-users] Water in Polymer

Dear all,
I would like to simulate water behavior in polymer structure. I am familiar to how to handle water molecules, but my concern is to generate polymer structure.
Could anyone tell me how to handle this issue?
Thanks, Gisuk.

Dear all,
I would like to simulate water behavior in polymer structure. I am
familiar to how to handle water molecules, but my concern is to generate
polymer structure.
Could anyone tell me how to handle this issue?

in order to get help you have to be _much_ more specific
about what kind of polymer and under what conditions you
want to similar and what kind of force field you are aiming
to use. all of that has an impact on what is a reasonable
way to get started.

also, you can search through the literature and look up
what other people have done for similar systems.

cheers,
     axel.

My simulation setup would be water in Nafion. One of my project was water behavior (MD simulation) in rigid walls (nanopore, slab), and I am familiar to water molecules. For Nafion polymer, somebody has done with harmonic, dihedral, LJ, and charged atoms.
Now, I would like to replace the rigid walls with soft polymer (Nafion), but have a hard time to generate the coordinate of each atoms for monomers since it looks like complicated procedures.
I think that a good example to start with is peptide in an example folder, and it looks like the input file has been generated by chain.f in tools folder. It seems that the coordinates are randomly generated, and need to be optimized using soft potential. However, using this program, how can I determine the type of atoms and specify the bonds and angles?
Or, it looks like we can use other commercially available software or pdb, and could you tell me some guideline how I can start with?
Thanks, Gisuk.

A colleague of mine has done quite a bit of work trying to get reasonable starting configurations for Nafion simulations. Here is a link to his thesis, which describes in detail his findings.

http://content.lib.utah.edu/cdm4/document.php?CISOROOT=/us-etd2&CISOPTR=201338

The dynamic timescales for Nafion are prohibitively long. You aren't going to be able to just make a starting configuration and equilibrate it. And given the length scale of the water and polymer domains of hydrated Nafion, you will need *very* large systems. The combination of length and timescales are a killer. You will need to come up with a clever and novel approach, or make some severe approximations (like coarse graining).

Matt

My simulation setup would be water in Nafion. One of my project was water

i am not an expert on nafion, but from all i know, it can be pretty nasty
because of the many ways it can be interconnected and how entangled
the polymer structures can be and also how to get the proper amount
of water in there.

behavior (MD simulation) in rigid walls (nanopore, slab), and I am familiar
to water molecules. For Nafion polymer, somebody has done with harmonic,
dihedral, LJ, and charged atoms.

nod. that is a standard all-atom force field type appoach.

Now, I would like to replace the rigid walls with soft polymer (Nafion), but
have a hard time to generate the coordinate of each atoms for monomers since
it looks like complicated procedures.

it is.

I think that a good example to start with is peptide in an example folder,
and it looks like the input file has been generated by chain.f in tools

no. chain.f is for simple bead-spring polymers. peptide inputs are
usually prepared from specialized tools where you either fit a
crystal structure as input or a amino acid sequence and the
secondary structure. this is processed by having a databased of
templates with (idealized) internal coordinate representations of
all atoms (crystal structures are typically missing hydrogens, so their
coordinates get regenerated at this step. also with peptides it is
trivial to connect the monomers, since there is only one way.
the definitions of angles and dihedrals can be inferred from the bonding
topology and the fragment templates have special entries that
contain improper dihedrals (where needed) and also match individual
atoms with force field types, charges and more.

since you have only one or a small number of monomer types
you could write a script or program that does something equivalent
based on placing those monomers in space and connecting them.
the easiest way to do this is actually leave excessive amounts of
space between them and then use a minimization run in lammps
to right all the bond distances afterwards. keeping excessive
space, makes it easy to place a proper fraction of water in between
(it will take "forever" on the MD timescale) to wet such a polymer
wall like in an experiment. you can also "shake up" things for a while
by running at elevated temperature and then anneal slowly to
a decent equilibrium.

there is no simple "just click here" way of setting these systems
up. it is probably advisable to work on small parts and equilibrate
those and then assemble larger systems from them. however,
it will always be problematic to confirm that you have reached
a sufficient state of equilibrium.

axel.

Dear Matt,
Thanks for the useful information.
For now, I am not sure if I need to simulate larger geometrical/time scales for the physics that I wanted to look into.
If I need large scale simulation, perhaps, coarse graining might be a good idea.
Thanks, Gisuk.

Dear Matt,
Thanks for the useful information.
For now, I am not sure if I need to simulate larger geometrical/time scales
for the physics that I wanted to look into.

i think you are missing the point that matt wanted to make.
whatever the physics that you want to look into are going to
demand, it _will_ take a large system and a long time to get
a properly equilibrated _starting_ structure. i tried to be more
careful in my wording, since i have only second hand knowledge
on this.

if you don't have a good starting structure, all your other
work will be tainted by it and thus you have to carefully
consider what it is that you want to model and what kind
of result you want to extract from those simulations. it is
often better to do a (too) simple model well, rather than
doing a good model badly.

axel.