Molecular crystals are an important ingredient of many drug products, agrochemicals, optoelectronics and explosives. Key properties such as the solubility and the dissolution rate strongly depend on the crystal structure. Many organic compounds can form various crystal packings with different properties, a behaviour known as polymorphism. Since thermodynamic stability and crystallization dynamics are largely unrelated properties, it can happen that important crystal forms are discovered very late in the product development cycle. Sometimes, such late-appearing forms prevent the crystallization of previously known forms, giving rise to the dreaded phenomenon of disappearing polymorphs, as witnessed by the famous cases of Rotigotine and Ritonavir. But late-appearing forms may also be perceived as a chance if they come with improved solid-state properties such as higher solubility.
Crystal Structure Prediction (CSP) has turned from an academic challenge into a routine tool for the prediction and avoidance of disappearing polymorph cases. For compounds of 60 atoms and more, the crystal energy landscapes of anhydrates, hydrates, salts and co-crystals can be generated with one and often two chemical unites per asymmetric unit. The relative stability of crystal forms as a function of temperature, pressure and relative humidity is calculated with an accuracy approaching 1 kJ/mol. Crystal structures can be readily solved even from low quality powder diffraction data by comparison with predicted crystal structures.
Free electron lasers provide an opportunity to combine CSP with an experimental method for very high-throughput micro-crystallization that generates XRD fingerprints sufficient for structure solution by comparison. Such a combination may solve may applied problems related to the crystallization of predicted but not yet observed crystal structures.
Sakura Pascarelli / Gabriella Mulá-Mathews