The materials express - faster materials modelling

Materials World magazine
4 Mar 2013

In an age of
rapidly advancing technology, supercomputing is pushing materials
science into the fast lane. John Conti-Ramsden, Richard Anderson and
Sebastian Metz from the Knowledge Centre for Materials Chemistry, UK,
examine the many applications of materials modelling. 

‘A modern industry economy is no better than its best materials’ – the words of British metallurgist Robert Wolfgang Cahn ring true for materials scientists the world over. For any economy, it is crucial to efficiently make use of the available resources to develop materials with improved properties.

In the early days of materials science, enhanced materials were developed on a trial-and-error approach, with slow progress. But with the development of sound underpinning scientific theories, the progression of improved materials accelerated significantly. 

Since then, the way materials are designed and tested has become ever more efficient and the experimental methods used increasingly sophisticated. But often, materials are difficult to accurately characterise by conventional experimental techniques, owing to their heterogeneous and complex, disordered composition. Some indication of their structure and composition can be obtained by conventional experimental methods, for example X-ray diffraction. However, establishing the molecular arrangement, which is often important for material performance, can be much more difficult. 

But there is an emerging revolution taking place. Computational simulations have proven extremely valuable for gaining insight into the structure and dynamics of new materials that would not otherwise be possible. The development of computational methods coupled with ever improving computational power is becoming increasingly important for the development of industry-relevant materials in a wide range of areas, for example:

  • supporting the development of highly sophisticated membranes to purify biopharmaceuticals
  • novel intermetallic alloys to be used in turbine blades
  • complex magnetoelectric films in electronic devices  

Materials modelling – a rough guide 

Materials modelling can enhance understanding of existing materials or reactions (such as catalysts), enabling researchers to identify the most productive areas for further research. But it can also simulate the design of new materials, by helping to work out how the chemical make-up can be altered to produce new and improved properties. In this way, modelling allows the optimisation of materials without actually having to make them. Depending on the system and the property of interest, different modelling techniques (or combinations thereof) can be applied. There are three methods most likely to have the biggest impact on future developments:  

1. Quantum mechanics (QM) – The chemical and physical properties of all matter result from the electronic structure of the atoms, ions or molecules they are built from. The latter can be modelled as systems consisting of atomic cores and electrons, based purely on the underlying laws of quantum mechanics. Apart from the 3D structure, no further knowledge is needed a priori. However, QM calculations can only be applied to a limited system size due to the complexity of the underlying theory. QM calculations can be used for a broad range of applications. These include simulation of molecular structures – for example, to figure out whether a molecule fits into the pocket of an enzyme or to model the surface properties of a potential catalyst (opposite). It can also be used to determine spectra, such as identification of intermediates based on the spectral fingerprint; band gaps, for example predicting bulk properties for semiconductors of solar cells; or full reaction mechanisms, such as identification of the rate determining step of a chemical reaction (left).  

2. Molecular dynamics (MD) – Monitors the interaction between and movements of atoms or molecules, such as for a given temperature or pressure for a short period of time. Most systems are too big and complex to model their dynamics with QM methods, therefore common MD methods make use of force fields, an approximation defining the way a system of particles interacts. Force field functions and parameter sets for the system of interest must be known before simulation work can start. MD is becoming increasingly commonplace in materials science and sophisticated algorithms exist to investigate a variety of systems. The results of MD simulations may be used to monitor and determine macroscopic thermodynamic properties of the system – such as temperature, pressure, internal energy and entropy – in addition to a number of materials properties, such as elastic constants, diffusion coefficients or compressibility. This enables in-depth understanding of those molecular arrangements of materials not easily understood using conventional techniques. Due to its properties, MD has become the standard method to simulate the effects that energetic neutron and ion irradiation have on solids and solid surfaces, which is important for materials used in the nuclear sector.  

3. Mesoscale modelling – If MD is computationally too demanding and larger systems have to be modelled for a longer time, so-called mesoscale modelling techniques can be applied. These methods make use of further approximations in the description of a system, and increasingly attract interest from a range of industries dealing with, for instance, the development of formulated products, semiconductors, novel plastics and metal-based products.  

In mesoscale methods, particles do not represent single atoms – instead they are made up of molecule segments (such as functional groups), collections of whole molecules or entire regions (see below). A variety of mesoscale methods exist that can be used to study the interaction and propagation of dislocations (directly related to crack formation) fundamental to many phenomena in crystalline materials, in addition to dynamics of colloid suspensions, biological membranes, surfactants, dilute polymer solutions and polymer blends.  

Who has the power? 

The modelling of material properties requires sufficient computational power, preferably highperformance computing (HPC) facilities, which are of course everything but cheap. With modern supercomputers costing millions of pounds, this is an investment beyond what many companies can afford in-house. And while UK Government has invested heavily in HPC to give companies access to supercomputing facilities at a smaller cost, big computers are only part of the solution. There is further need for software that makes the most of these facilities, and frequently it does not.  

Most software is designed to run on a single processor, but to modelling that’s like having a nice sports car you can only drive in first gear. For software to calculate at a speed beneficial for industrial applications, its design has to be changed so that it can efficiently run on multiple processors. This allows the simulation of significantly larger systems, revealing new science and developing a detailed understanding of the inner workings of materials. The latest and most sophisticated supercomputing facility was recently installed at Daresbury Laboratory near Manchester, UK, run by the Science and Technology Facilities Council. The Hartree Centre was set up with industry in mind, as a specialist centre providing HPC facilities and producing this vital software of the future.  

Through the Knowledge Centre of Materials Chemistry (KCMC), companies can easily access the Hartree Centre’s substantial range of computational facilities as well as a full range of experimental expertise in materials chemistry at Bolton, Liverpool and Manchester Universities, to avoid time-consuming experimentation and instead use simulation and data analysis to quickly answer a given problem. Although computational methods are not likely to displace the need for experimentation altogether, it can reduce or even eliminate the laborious early stages of investigation, tweaking and re-testing.  

The materials revolution 

Materials modelling has proven successful for a number of organisations in a range of materials-based industries. Aircraft manufacturers, for instance, have used it to investigate the effect of bird strike on aircraft wing leading edges from fibre–metal laminates. Opto-electronics companies have used materials modelling to investigate the interfaces between layers in these devices, which are major sources of loss of performance. Oil companies have used modelling to investigate how additives can permeate clay to better extract oil. Computational modelling stands to help a variety of other sectors too, including formulated products, nuclear, biotechnology and energy, to name just a few. These sophisticated modelling capabilities will not only enable faster completion of research, but accelerate the end product to market.  

For more information, please contact:

John Conti-Ramsden –

Richard Anderson –

Sebastian Metz –