How can molecular modelling help drive materials discovery?
What is Molecular Modelling?
With a drive to digitise all aspects of science and business, computational modelling is gaining interest across industry and academia as people look to keep up with the times and accelerate their research. In the realm of materials science, molecular modelling is a physics driven approach to determining material properties - think finite element analysis on a smaller scale. This modelling can take many forms, such as Monte Carlo or Hartree-Fock simulations, but my interest is in molecular dynamics.
Molecular Dynamics
Molecular dynamics, or MD, is a method of simulating the movement of particles using classical mechanics. This is important as Newton's laws of motion are significantly cheaper to compute than quantum mechanics, allowing for much larger models that simulate a larger timeframe. This allows for simulation of macroscale systems, like polymers and metals, to determine a range of material properties. MD has been used for decades in the pharmaceutical industry to support drug design by modelling proteins and how a novel drug interacts with different active sites. By contrast, MD for materials science only started being investigated in academia around 30 years ago, and uptake in industry has been slow due to a lack of specialist computational materials scientists.
Simulated Properties
Many properties can be determined through MD simulation. As my background is in polymer science, I will leave metals to someone more specialist, but here's a good paper on the topic. The simplest properties to simulate are those inherent to the structure: density, free volume, and atom distribution. These properties can be quickly obtained for monomers or polymers, and can be exciting as it's possible to see chemistry happening in real time: polymer chains folding and back biting, or coordination of atoms around a hydrogen bond donor (seen in purple in the figure). Additional simulations can be carried out to calculate thermal properties like glass transition temperature or coefficient of thermal expansion, or mechanical properties like Young's modulus or Poisson's ratio. MD can be pushed further to visualise and calculate permeation of gases through a material - considering how permeation is vital to applications like preventing methane leakage, hydrogen storage, and carbon capture, I'd expect these simulations to be a major application for modelling.
Modelling and R&D
Whilst this hopefully sounds appealing, there are limitations to modelling. There is a trade off between speed and accuracy, and even large, long runtime models will not be perfectly accurate due to limitations in the modelling physics. However, what MD does enable is high-throughput screening of polymers. MD is often able to capture the trend of material properties for a range of simulated materials. This allows researchers to eliminate the worst candidate materials, and only take the most promising forward for laboratory testing. Additionally, it's possible to simulate hazardous materials, hard to obtain bio-based monomers, or even molecules that don't exist yet, to figure out if it is worth the effort trying to develop a material based on these chemicals. MD can help to give direction to researchers in the challenging trial-and-error world of materials discovery.
Thank you for reading the article, the first in this series on molecular modelling and materials science! The next article will cover the how to get started in MD.
