How to build a polymer with molecular simulation?

With molecular modelling you can visualise how a polymer chain spirals it grows

Previously, we've looked at how to setup the pieces for a simulation, and how to design the simulation to get useful results. Now we get to run the simulation! In this article we cover methods of creating a polymer network, and some of the tools in the LAMMPS ecosystem that can help you to create your own networks. This is continuing my series on molecular modelling for materials science, so follow along for more tutorials.

Chain or Network?

Firstly, there are two ways of simulating a polymer: as a single chain, or as a network of chains. Whilst single chain models are much easier to setup, they tend to lack accuracy for most properties as they are less able to capture inter-chain interactions and can't form crosslinks.

Networks are more challenging to create, but offer improved accuracy and insight into how chains entangle and interact. In order to create a network, it is best to start from an equilibrated unit cell of monomers. Monomers are then reacted together by an algorithm that typically uses a cutoff distance to determine bonding pairs. Over time a cured polymer forms, with the potential for crosslinks if the monomers are suitably functionalised. Reacting all these monomers together too rapidly, or with too large a cutoff distance, can create an unrealistic geometry that cannot be easily relaxed. Some patience is required to generate a suitable network. However, you may need to raise the cutoff as the polymers start to form as atoms become less mobile, or it may take a very long simulation to reach a desired degree of cure.

LAMMPS for Polymer Models

Creating an algorithm to manage this bonding process is normally a huge task as there are many different ways different monomers can bond. Recognising all the potential reactive sites, and determining if there is a suitable reaction available, requires a large code base that has to be tweaked every time a new polymerisation reaction is modelled. However, within LAMMPS the fix bond/react command is capable of handling any polymerisation process you may wish to model! It works by using templates of the reactive site before and after a reaction takes place. If a matching site is found, within the specified cutoff distance, it converts the current structure to the one in the post-reaction template. In my opinion, bond/react is an essential innovation for polymerisation models and means that LAMMPS is the software of choice for modelling polymer networks.

AutoMapper

The difficulty with using bond/react lies in preparing the pre- and post-reaction template files, and an atom map that explains how to convert between the templates. Fortunately, I had this problem myself and created AutoMapper to tackle this challenge with minimal input from the user. With only the atom IDs of the atoms doing the bonding, and the pre- and post-reaction templates, AutoMapper will create all the files needed to use bond/react to create a polymer network. This makes setting up a polymerisation simulation quick and easy, hopefully helping to enable higher throughput modelling for materials science.

You should now know how to create a realistic polymer network, and why I believe LAMMPS is the tool of choice for doing so. In the next article, we'll look at how to get some more complex thermo-mechanical properties from our polymer network. Thanks for reading!