More power, more materials
Advances in computational power can aid the discovery of new materials through sophisticated computer modelling, as Ellis Davies reports.
A chemist’s approach to developing new materials is a complex procedure – a game of chance navigating the optimised positions of atoms over an energy surface represented in many dimensions. Information is a valuable tool in this endeavour, and is now in greater supply due to computer modelling and increased computational power. Projects facilitating the discovery of new materials through an increase in accessibility and quantity of data have begun to emerge, including the Organic Materials Database (OMDB). Researchers at the University of Southampton, UK, have also made ground with a new method that maps how molecules assemble and crystallise to form new materials.
Multiple surface dimensions can be difficult to conceptualise, but by combining an array of computer modelling methods, the team at Southampton believes that its new method is able to predict how molecules will form crystal structures, and what properties they will have. Focusing on porous materials, Graeme Day, of the School of Chemistry at the University of Southampton, told Materials World, ‘We combine the structure prediction methods [also developed at Southampton] with geometric analysis calculations, which assess the size and connectivity of pores within the predicted porous structures. In addition, we use Monte Carlo methods [computational algorithms reliant on repeated random sampling] to predict some properties of the structures, such as methane uptake and selective adsorption from mixtures of gases.’
The use of several types of modelling allowed the team to assess all potential crystal structures, as well as predict the energetic stability and properties. Day said that given the point to which computer modelling methods have now been developed, the results can be used to inform experimental decisions with confidence.
Modelling produces a ‘map’ that allows chemists to find specific properties that, when combined, could produce a new material. ‘In theory, we can apply the approach to many types of properties, so we hope to be able to accelerate the discovery of materials with improved properties in several application areas, such as nanoporous materials, organic electronics and catalysis,’ Day explained. The team tested the method by conducting a series of simulations, in which they performed structure and property predictions on a set of hypothetical molecules designed to form porous crystals. Using the results, a molecule was chosen (triptycene), in this case one that showed a high-predicted methane storage capacity. The molecule was made by collaborators at the University of Liverpool, UK, and crystallised into the predicted structure, which also delivered the predicted properties.
Day intends to continue this work by demonstrating how the method can be used for the discovery of materials with other applications – not just relating to porous crystals. These include structures with other types of properties, such as high electron or hole mobilities in organic semiconductors or electronic band gap for photocatalytic materials.
Searching for properties
The OMDB is an online electronic structure database for various organic and organometallic materials. The result of a collaboration between the Nordic Institute for Theoretical Physics (NORDITA) and the Royal Institute of Technology, Sweden, the project represents the ‘convergence of data science, data mining and other tools with materials science synthesis and characterisation to facilitate the search for new materials,’ according to Professor Alexander Balatsky of NORDITA. The database currently contains electronic structure calculations for 6,688 materials and has institutional commitment for continued expansion until 2021.
Organic crystals are not investigated as frequently as inorganic materials because of the large unit cells that contain up to several hundred atoms. Advances in computational resources have helped to investigate these large amounts of data over the past decade. ‘The database was created by performing large-scale electronic structure calculations within the Condensed Matter research group at NORDITA. The calculations are based on ab initio codes in the framework of density functional theory,’ Balatsky told Materials World. The online interface allows the user to search for materials with specific properties with queries about their electronic structure, as well as an advanced search for pattern recognition and chemical and physical properties.
Data on the electronic and magnetic properties of crystals can be used for a wide range of problems, including working towards functional materials. Balatsky gave the example of those with special electrical, optical and magnetic properties, such as the 2016 Nobel Prize-winning topological states of matter, an important building block of a quantum computer. The database has already led to various discoveries in organic functional materials, including organic semiconductors and Dirac materials.
Both developments show how computational power is providing materials science with a tool for the discovery of materials, allowing for ‘a significant transformation in the field,’ according to Balatsky. Both projects are looking to expand, further increasing their potential influence on future materials.