MD Modeling

Molecular dynamics (MD) modeling is the process of modeling the atoms in a material as charged particles. Each of the charged particles interact with the surrounding particles through "force fields" by writing equations for the behavior of the force field for each atom and then allowing the computer to calculate the optimum location of each particle to minimize the energy in the system of particles, we can determine a lot about the structure and behavior of materials. MD has been around for a while and software such as LAMMPS is free which can be used to model crystalline structures very easily. MD is also used on molecules, especially in the medical field. For example the commercial software Materials Studio is often used for biomolecular systems. Many times the output of the analysis is tables and table of data. For example the output could include the 3D position of each particle as well as its energy state etc. Therefore software also exists for visualizing the output. For example VMD is a free program for molecular visualization that displays and animates the results of a MD analysis.

Performing MD on polymers has a number of significant challenges. The software described above work really well on crystalline materials that have long-range-order or on molecules, even very complex molecules found in biochemistry research. However, for a general polymer, like polyethylene that has short-range-order but no long-range-order, as well as many different molecules of varying lengths and sizes results in a structure that is not well suited for the software described above. In addition, using the above software (at the time this was written) would require developing extensive initial configurations with the locations of the atoms making up the polymer.

A code, written in C++, was developed for the simulation of polymers. This code is designed to run on a distributed parallel computing system and includes the capability to generate the polymer chain definitions before running an energy minimization routine utilizing the Lennard-Jones Potential to create feasible initial position of the polymer components. A simple visualization of the polymer material was created as a post-processor (running on a single processor) to quickly show the results of the simulation. The polymer chain creation using Rotational Isomeric State models effectively creating a random-walk procedure utilizing the most likely angles associated with that particular polymer chain. The number of chains and target density are used as input data and the code would randomly distribute chain initiation points throughout the 3D space. Each chain would grow with a random walk (probabilities were assigned to each possible angle) and if the placement of an atom would interfere with an existing chain (self or neighboring) the chain would back up and try a different angle for a previous link. This would occur until the target density was achieved resulting in entangled, various chain-length, polymer segments with highly probable configurations. After the entangle polymer chains were defined, the energy minimization routine incorporating bond torsion forces, bond angle forces, and LJ potentials for non-bonded forces would determine the final configuration of the polymer structure. Note: the bond length is not held constant and allowed to fluctuate according to an inverse LJ potential attraction. The video below shows an example of the chain creation portion of the code. Near the end of the video you can see where particles of new chains are trying to be placed, fail, and the chain shrinks and tries to find a more favorable configuration.