MS33-P02 Applying fast, accurate lattice energies for molecular crystal structure prediction using CrystalExplorer model energies Peter Spackman (School of Chemistry, University of Southampton, Southampton, United Kingdom) Graeme Day (School of Chemistry, University of Southampton, Southampton, United Kingdom)email: interaction energy between two molecules can be understood as a sum of several discrete terms [1]:  generally electrostatic, polarisation (or induction), dispersion and exchange-repulsion (Etot = Eele + Epol + Edis + Erep). CrystalExplorer (CE) model energies [2], calibrated for use in intermolecular interactions, have been shown to estimate experimental crystal lattice energies [4] showing a favourable mean absolute deviation of only 6.6 kJ/mol from reference values in 110 crystal structures (the X23, G60, K7 and Z20 benchmark sets).
The relative energies of sets of crystal structures composed of the same molecular building blocks are of particular importance in crystal structure prediction (CSP), as lattice energy ranking is used as a primary predictor for the likelihood of a predicted structure being realized in the laboratory.

We investigate the potential use of CE lattice energies (and variations thereof) in CSP protocols – both in terms of relative energy ordering for different polymorphs and in terms of absolute energies – with reference to density functional and force-field based methods. Using CE lattice energies on the landscape of predicted crystal structures may constitute an accurate technique to improve predicted results, without the associated computational cost (and scaling) of periodic density functional calculations.

Keywords: Lattice Energies, CrystalExplorer, Polymorphism