Crystal Structure Prediction
(173 Words, 1 Minutes)
Different packings of the same molecule (polymorphism) can lead to different physical properties, such as melting point, hygroscopicity and bioavailability. This can cause problems for the effectiveness of many materials, particularly pharmaceuticals. Hence, it is very important for the industry to know which polymorphs are likely.
Crystal Structure Prediction (CSP) generates energy/structure/function landscapes of hypothetical crystal structures. These structures are ranked based on their lattice energy, with the most stable being the most likely to form. Energy landscapes (such as the one pictured) often have many more hypothetical crystal structures than observed polymorphs. However, they are still helpful in assessing the experimental solid form landscape for small pharmaceuticals, such as galunisertib or ibuprofen.
Addressing Overprediction in CSP
The over-prediction of putative polymorphs is a well-known issue of CSP that has its roots both in the computational methods employed and in the extent of the experimental polymorph screening campaigns. For instance, one of the reasons for overprediction is that CSP neglects thermal effects that lead lattice energy minima to coalesce at finite temperatures. To tackle this problem, we have recently proposed a workflow based on molecular dynamics and enhanced sampling aimed to systematically reduce the number of putative polymorphs generated by CSP by identifying: i) Labile structures - i.e., crystal structures that, despite representing valid potential energy minima, are unstable when simulated at finite temperature and can thus be discarded. ii) Effectively equivalent structures—i.e., structures representing distinct potential energy minima that can easily interconvert and merge into the same dynamic ensemble at finite temperature.
Relevant Publications:
- Systematic finite-temperature reduction of crystal energy landscapes. NF Francia, LS Price, J Nyman, SL Price, M Salvalaglio Crystal Growth & Design 20 (10), 6847-6862 23 2020
- Reducing crystal structure overprediction of ibuprofen with large scale molecular dynamics simulations. NF Francia, LS Price, M Salvalaglio CrystEngComm 23 (33), 5575-5584 8 2021
- Rationalising the difference in crystallisability of two sulflowers using efficient in silico methods. IJ Sugden, NF Francia, T Jensen, CS Adjiman, M Salvalaglio CrystEngComm 24 (39), 6830-6838 2022