Metal-organic frameworks (MOFs) are microporous materials of emergent technological importance for gas storage and separation, catalysis, sensing devices and other advanced applications. Computational methods are becoming increasingly important for exploring structure-property relationships in MOF materials and providing tools for target-specific design of new generation materials.

The broad range of applications stems from the modular nature of MOFs which are constructed from metal nodes (individual atoms or clusters) interconnected by organic linkers. The large choice of available nodes and linkers leads to essentially infinite number of topologies and, ultimately, properties. The question is, how do we find the right combinations of metals and linkers to build MOFs for specific applications?

Currently, the most common answer is experimental screening, which is not an optimal solution due to high costs and excess chemical waste. Our aim is to optimise the design of new MOFs by utilising computational methods. We have developed a method for ab initio crystal structure prediction of MOFs, demonstrating that crystal structure of MOFs can be predicted solely based on the knowledge of chemical structure of nodes and linkers.

This presentation will outline our recent advances in using periodic density functional theory (DFT) calculations to characterizing and even predicting MOF structures. First, we will show how periodic DFT calculations can be used to explain the thermodynamic pathways observed in mechanochemically-induced topological transformations of MOFs, demonstrating that such reactions proceed towards thermodynamically most stable structures.

The talk will continue with a report of the computational screening of a previously not synthesized class of microporous materials based on the recently isolated pentazolate (pnz^-) ligand. The energy landscapes for Zn(pnz)₂ and Cd(pnz)₂ frameworks were calculated, reliveving formation of several previously unreported topologies. Finally, energetic properties of putative pentazolate frameworks were evaluated.

Finally, the first example of a truly ab initio structure prediction will be presented, with examples from several MOF classes, including hexafluorosilicate, azolate, and carboxylate frameworks. In all cases, experimental structures were used to confirm and evaluate theoretical expectations, providing an important step forward in theoretical understanding as well as modelling of MOFs.