Title: Modeling and Simulations of Polymers: A Roadmap
Authors: Thomas E. Gartner and Arthi Jayaraman
In this modern era, chemists have been harnessing advanced computational technology to predict, design, and understand complex chemistry. In particular, simulating polymer behaviour, structures, and reactions can guide the discovery and design of macromolecular materials. Yet, without careful considerations in selecting and developing appropriate models, simulations could get wild and unrealistic. In fact, there are skeptics labelling computational simulations as “unreal” in contrast to real experiments. So how could chemists choose and/or design a realistic model to ensure the computational results are valid and reliable? In this article, we will look into the common practices and pitfalls in choosing an appropriate model to reach meaningful conclusions for material design.
Polymers have always been an important class in material chemistry. From plastic to clothing to proteins, polymers play a crucial role. They are a class of large molecules with many repeating subunits called monomers, just like architecture built from Lego blocks. There are two commonly used models for simulating polymers. If one wants to understand the specific chemistry of each Lego block or specific chemistry at a particular position of this polymer, one could choose the atomistic model. On the other hand, if one wants to have an overall picture of the polymer architecture, its structure, and morphology, one could choose the coarse-grained (CG) model to look at it as a bigger picture.
Atomistic models are the better option for studying arrangements, fluctuations, and interactions of the composing monomers of a polymer. These models can calculate interactions between monomers and the block polymer and/or show the arrangement of a specific particle surface within the polymer composite. Since these models investigate specific interactions of small molecules with block polymers, they can model the solubility, absorption, and diffusion within a polymer membrane. They are also useful for studying charged polymer systems and electrostatic interactions.
The setback of such models is mainly due to their demanding computational time. The accuracy of these atomistic models depends heavily on the choice of bonded and nonbonded interaction constants used, called the force fields. The calculations are also highly dependent on the initial configuration one inputs into the simulation.
In order to reduce computational time, united atom models may be used. These models neglect the hydrogen atoms and only take care of the heavier atoms they are attached to. Yet, care has to be taken if hydrogen bonding plays an important role in the structure and thermodynamics of the system.
To implement the atomistic model, one has to select the force field that could represent the experimental observations at conditions, such as temperature and pressure, that are relevant to the problem of interest. Since a proper atomistic initial condition is essential to obtaining accurate and reasonable results, one can make use of a database of previously obtained crystallographic structures for the polymer as a starting point. If the structure of the molecule at hand is unknown, one can use molecular builder and energy minimization software with the selected force field to create the starting configuration.
If one is interested in the bigger picture of polymer chemistry, for instance, the structure or morphology of polymers, CG models may be a better choice. In these models, the atoms are considered as a group instead of individual atoms. These groups are called CG “beads”. With carefully designed parameters and well-tested atomistic force fields, calculations with CG beads could reduce computational time required and improve efficiency. These models are extremely useful in investigating polymer design parameters like molecular weights and system conditions such as temperatures and solvents.
Since CG models view atoms as a group, they could not provide in-depth details at atomic level. Yet, they are useful in obtaining universal trends that explore structure and thermodynamics.
After a model has been selected, the next step is to find the suitable simulation method. In general, the two popular simulation methods are called the Monte Carlo (MC) and molecular dynamics (MD).
MC simulation scatters atoms or beads randomly and then accepts or rejects the trial move according to a function with certain criteria. On the other hand, MD makes use of Newton’s equations of motion to calculate the movement of atoms or beads in a certain time period. The averages of these movements result in structural and thermodynamic quantities of the molecule. In general, MD is more computationally efficient for large systems as they are simpler to implement. However, for dense and systems with many atoms, they could require long simulation times to achieve equilibrium configurations and become less efficient.
After the simulation is completed, it is essential to analyse the results to ensure they are realistic and meaningful in order to provide useful and insightful data. As no simulations are perfect, referring to theory is a good way to overcome the limitations of simulations. Comparing simulations results with experimental observations is also a good approach to validate the chosen model, simulation method, and analysis.
In short, computational modelling could provide valuable insights into polymer design and chemistry. This article has provided a roadmap for beginners to peek into the computational world and considerations involved to kick start our understanding molecules with computers!