American Chemical Society
cg1c01317_si_001.xlsx (15.37 kB)

Crystal Energy Landscape of Nifedipine by Experiment and Computer Prediction

Download (15.37 kB)
posted on 2022-01-10, 11:29 authored by Yue Gui, Yingdi Jin, Shigang Ruan, Guangxu Sun, Vilmalí López-Mejías, Lian Yu
The six polymorphs of nifedipine (NIF) known at present have been discovered over the past 50 years, and the most recent one (δ), discovered in 2020, came from an unusual route. This polymorph is ranked second in thermodynamic stability but evaded all previous workers until its melt was seeded with the crystal of a foreign substance, felodipine, in which the molecule has a different conformation from all other NIF polymorphs known at that time. Given this unusual discovery in the lab, we investigated whether crystal structure prediction (CSP) can find this and other polymorphs in a “routine” search. We show that our CSP finds all ordered polymorphs of NIF known at present as low-energy structures (Ranks 1, 3, 4, and 43), including the most recent one unveiled by pseudoseeding (Rank 4). NIF being a flexible molecule, it is of interest to learn which of its many conformers provides the best building block for crystals. An experimental investigation of this question is limited by survival; that is, information exists on the structures that are observed but not on those that are difficult to observe or not yet discovered. In this regard, our “computer experiments” access the full range of possibilities. We find that the synperiplanar (sp) conformer with respect to phenyl torsion produces lower-energy crystals than the antiperiplanar (ap) conformer, with the most stable ap crystal being 4 kJ/mol higher in energy than the most stable sp structure. Experimentally, the sp conformer dominates the ap in solution and is the only conformer observed in crystals. With respect to the ester torsions, the cis/trans conformer produces the lowest-energy crystals, followed by the cis/cis conformer and by the trans/trans conformer. Experimentally, five of the six known polymorphs contain the cis/trans conformer, one contains the cis/cis conformer, and none contain the trans/trans conformer. Overall, the CSP is remarkably successful in predicting the polymorphs of NIF in spite of its complex conformational space and provides a quantitative assessment of the relative costs of employing different conformers as units of crystal building.