there is  packmol http://www.ime.unicamp.br/~martinez/packmol/ which as far as i understand is free, and there is the commercial Materials Studio which comes with a similar tool for polymers - but i believe none of them will manage to construct an amorphous structure of a complex polymer at a reasonably high density without generating major overlaps (or requiring an enormous amount of time).... so the best way to go here is to construct a sample with a very low density (a large box) and then compress your sample in an NPT simulation. ususally you want to do this in steps: 1) build a low density sample 2) compress (hight P and room temp) 3) heat (high T ambient pressure) 4) goto step 2. in my experience a few cycles of (2->3)  in a few nano seconds simulation time are enough to obtain an amorphous sample at a reasonably high density. You have to play around a bit with the values for high temperature/pressure that you use, and perhaps reduce the time step of your NPT simulation somewhat to make sure your system won't blow up though.

alternatively (to make sure that your polymer chains are sufficiently entangled) you might want to start with a structure in which all polymers are straight and aligned in a way so that you have no overlaps but still a relatively high density, the annealing (hight T/high P) will then make sure that the order in your system is reduced to obtain a properly amorphous  state...

good luck!

michael

It is worth trying any of these procedures:

1. place your styrene molecules in a simulation cell via PACKMOL software. create bonds between the reactive atoms of styrene molecules using LAMMPS. After this, refine your system. 

2. Use MOLTEMPLATE (which is distributed under LAMMPS) to create your polymer and place it in a simulation box via VMD, then write your data file.

Hope this helps.

Michael: This is what I am trying now. I started from density of 0.01 and now it is about 0.98 close to the exp value. In heating cycles NPT is absolutely necessary or can I apply NVT instead? But molecules are not (or weakly) mixed so maybe NPT will be better.

You can try using the procedures reported in the paper Macromolecules, 2011, 44, 6944-6951. For obtaining the equilibrium density you have to run NPT ensemble MD. Also, you know the forcefield you are using should model well your system.

To ensure the distribution is homogeneous, you can apply simulated annealing procedure via NVT ensemble MD, in which the temperature of your system is increased up to high temperatures and then cooled back to the target temperature. To check the homogeneity, you can calculate mass density profiles after each simulated annealing cycle. 

Hi, no need to install any software, just create an account to nanoHUB.org and use the polymer modeler tool, you can setup the final density, relax the structure, etc. https://nanohub.org/tools/polymod

Nicolas: So my problem is that Polymer Modeller does not work fine for me for the case of polystyrene. If I set 10 chains having 13 monomers of atactic PS with the density of 1.00 (the experimental density) and use default procedure (not change the other parameters) for creating and relaxing by Lammps I got wrong structure (some of  phenyl rings are open or even not connected to the main chain).

Could you please try these setting by your own and tell me if I am right. If you succeed this means that I made something wrong but it is so easy tools so it's unbelievable to my how can I make a misteake.

Even if I set density to 0.001 (lowest possible) I need to set more molecules (more than 10 even if I want 10) and then eliminate the wrong one.

1. What is the theoretical background behind using 10 chains with 13 monomers each?

2. Try to decrease the number of polymer chains and increase the monomer units in each chain.

3. Also, have tried any of the procedures I recommended you?

4. Before relaxing your structure in LAMMPS you have to ensure that the structure is modeled properly and you can visualize it in VMD. If not, it is pointless running equilibration simulations. Try always first the baby steps.

cheers.

Lukasz, first I apologies I haven't seen your entire question. We actively develop PolyMod in our research group here at Purdue and your feedback is of importance. Could you please report the bug on nanoHUB with all the details and I copy your message to the main developer. The problem will be fixed in no time. Thank you

Nicolas: I reported the bug (as a ticket). Here I put the part of the explanation of the problem and the zip file with .pdb structures. You can look at it.

"I attached zip archivum in which there are some .pdb files. They can be open in Avogadro software and you can take a closer look into the molecules. File with name modeller_15xPS13.pdb is the main file containing 15 PS molecules inside the big box (density=0.01) and the files named modellerX.pdb contain molecule nr X. Molecules 1(open ring), 5 (open ring), 7 (molecule broken into 2 pieces), 10 (single ring near the rest of molecules), 13 (single ring near the rest of molecules) failed."

Baris: Could you attach this paper? I dont have access to Macromolecules Journal.

Lukasz, unfortunately I cannot attach anyone's published paper here, because this would be a breach of copyright.

Materials Studio is a good software for model building, and lammps is a good software for running simulations.