Data Driven Discovery of Functional Molecular Co-crystals

This is an excellent opportunity for a well-motivated student to participate in a joint project between the Leverhulme Research Centre for Functional Material Design (https://www.liverpool.ac.uk/leverhulme-research-centre/ ) and the Cambridge Crystallographic Data Centre (https://www.ccdc.cam.ac.uk/ ). The student will be based at the University of Liverpool, with a supervisory team taken from the departments of Computer Science (Dr V Kurlin, Dr F Oliehoek) and Chemistry (Dr M S Dyer, Dr N Berry, Prof M J Rosseinsky), and will work closely with colleagues in Cambridge. The studentship is fully funded for a period of 42 months starting in October 2018.

The intercalation of alkali metals (Li, Na, K, Rb, Cs) into molecular crystals of fullerides (e.g. C60) and polyaromatic hydrocarbons (PAHs) can give rise to superconductivity, magnetism and the existence of exotic electronic states (e.g. 3D quantum spin-liquids). To date, these studies have focussed on materials with a single molecular species, however this restricts the chemical flexibility available for the optimisation of material properties or the discovery of novel electronic states. Therefore, we propose to investigate the intercalation of alkali metals into molecular co-crystals. One or more of the molecules in the co-crystal will be a fulleride or PAH. The other molecule(s) must be chosen such that it will pack with the fulleride or PAH in a co-crystal, while also fulfilling particular electronic criteria.

We propose to use the information stored within the Cambridge Structure Database (CSD) to identify the best molecular candidates to form co-crystals with fullerides and PAHs. Although few co-crystals involving fullerides and PAHs have been reported, there are many more co-crystals in the database which contain information about which pairs of molecules do form co-crystals together. The first step is to design a geometric code that is invariant under all rigid motions in and uniquely represents any periodic co-crystal structure. The second step is to define a similarity measure between resulting codes. The third step is to apply Topological Data Analysis for build a topological map of all existing co-crystals from the CSD. The fourth step is to use Machine Learning on a smaller subset of known co-crystals with good properties in this map as a guide for searching new co-crystals with better properties. As well as crystal structures, the CSD will also provide molecular identifiers of the molecules in co-crystals (e.g. InChI, SMILES). These will be used to compute descriptors and measures of molecular similarity commonly used in pharmaceutical research, which in turn become features in machine learning and informatics algorithms. This allows for alternative searches for molecules likely to form co-crystals with fullerides and PAHs which will complement the structural based method outlined above.

Qualifications: Applications are welcomed from students with a 2:1 or higher masters degree or equivalent in Computer Science or Chemistry, particularly those with some of the skills directly relevant to the project outlined above.

When applying through the online application form, please quote reference: LRC102.