• Subscribe

Multi-scale modeling in the virtual laboratory

the virtual laboratory
A snapshot from simulation of a self-assembled stack of clay layer and polymer molecules.

Peter Coveney, director of the Centre for Computational Science at University College London, has been working with colleagues James Suter and Derek Groen on ways of connecting different representations of matter together. Coveney believes this is the first step towards speeding up the process of discovering new and useful materials.

“Imagine, for example, a material that has fractured," Coveney says. "At the molecular level, this is shown as the breaking of chemical bonds by electrons moving between atoms, whereas the manifestation on a larger scale would be the breaking of a component made of that material. These are very different representations of the same event, but both are equally correct. To simulate this event separately at different scales is relatively easy. What is not so easy is to connect the two — to extrapolate the macroscale properties of a material from its chemical composition.”

Creating a description of a material that works at all scales without having to inject ad hoc parameters at higher levels is a crucial step towards in silico materials discovery. To pull off ‘multi-scale modeling’, as it is known, the lowest level parameters must be extremely precise, and the most powerful computers are needed in order to run the simulations. But the rewards for succeeding in this task are great; if one can predict the useful physical properties of a material from its molecular structure, then costly and time-consuming trial and error experiments can be eliminated from the discovery process.

In February 2015, the journal Advanced Materials published a paper by Suter, Groen and Coveney that discusses the properties of a number of clay-polymer nanocomposites, an almagam first created by a team of Toyota researchers in the 1980s. However, it is not the specific materials that make the paper so interesting, but rather the groundbreaking methods behind the research. In the paper, they describe a method that can be used to calculate the properties of clay-polymer nanocomposites using multi-scale modeling. The only inputs needed for this ‘virtual laboratory’ are chemical composition, molecular structure, and processing conditions. In return, it provides information that has largely never been shown before in any kind of modeling — let alone in an experiment.

“By connecting all the scales together into a multi-scale model, we were able to show the process of polymers getting inside the clay layers — how it happens and how long it takes,” says Coveney. “Clay exists naturally as stacked sheets called tactoids. When you add a polymer, it will break up this natural configuration — encapsulating, exfoliating, or intercalating the stacks. Our simulation showed that the composite then arranges itself in a particular orientation, such that the material properties begin to look very different from what you might predict from a linear combination of the properties of clay and the polymer.”

the virtual laboratory
Coarse-grained molecular dynamics simulation of polyvinyl alcohol polymer intercalating between layers of clay.

The paper was considered so important by Advanced Materials that for the first time in its entire history the high-impact journal published an extended feature so that the methods behind the work could be fully explained. “The ability to model and simulate the properties of a material in this manner has opened the door for making predictions that could vastly speed up many scientific discovery processes, not just in the field of clay-polymer nanocomposites,” explains Coveney.

Graphene, for example, is a material that has long been touted as a modern wonder material that will eventually revolutionize numerous fields of research. However, delivering the practical applications of graphene has proven difficult, not least due to the challenges of producing it in large enough quantities. Multi-scale modeling could be used to model the industrial production of graphene by exfoliating two-dimensional sheets of graphene from graphite — a process fairly similar to the exfoliation of clay tactoids in the production of clay-polymer nanocomposites.

Coveney and his researchers have made extensive use of tier-0 PRACE supercomputers, including 40.5 million core hours on JUGENE BlueGene/P at Forschungszentrum Jülich in Germany. “Carrying out multi-scale simulations comes under the domain of what we call “heroic computing tasks,” he says. “I personally believe that the future of materials science lies in gaining a proper understanding of composites, and this is very much dependent on the high-fidelity nature of our models and simulations. Tier-0 supercomputers such as those provided by PRACE are absolutely essential for running these simulations in feasible time periods, and so the success of our work and any future work that uses our methods leans on the access that researchers have to these valuable resources.”

In the short term, the team’s methods have the potential to accelerate scientific discovery and understanding. In the long run, materials science will be changed for the better, by eliminating a lot of the trial and error that currently besets the development of useful materials.

This article is republished with permission from PRACE. Find out more about how PRACE is supporting scientific research in their latest annual report.

Peter Coveney is set to give a keynote speech at the upcoming ISC Cloud and Big Data Conference in Frankfurt, Germany. Find out more about this event here.

Join the conversation

Do you have story ideas or something to contribute?
Let us know!

Copyright © 2015 Science Node ™  |  Privacy Notice  |  Sitemap

Disclaimer: While Science Node ™ does its best to provide complete and up-to-date information, it does not warrant that the information is error-free and disclaims all liability with respect to results from the use of the information.


We encourage you to republish this article online and in print, it’s free under our creative commons attribution license, but please follow some simple guidelines:
  1. You have to credit our authors.
  2. You have to credit ScienceNode.org — where possible include our logo with a link back to the original article.
  3. You can simply run the first few lines of the article and then add: “Read the full article on ScienceNode.org” containing a link back to the original article.
  4. The easiest way to get the article on your site is to embed the code below.