Speaker
Description
Molecular crystals represent a large entity of materials as most organic and numerous inorganic chemical compounds exist in the form of a molecular crystal at low temperatures or high pressures. Materials with extremely different properties can be mentioned in this context – weakly-bound noble-gases crystals, solids as common as water ice, or strongly bound zwitterionic or molecular-ionic crystals. Existence of polymorphs, cocrystals or crystalline solvates, so typical for molecular crystals, further enrich the already vast field of research. Probably the most extensively investigated group of molecular crystals currently are the active pharmaceutical ingredients.
A delicate interplay of intermolecular interactions (electrostatic, induction, dispersion, hydrogen-bonding) in the crystal lattice predetermines the properties of the material. A key property of each crystal is its cohesive energy, expressing how strongly are the molecules in a crystal bound together. Adding this property to vibrational properties of the crystal (phonons), many valuable thermodynamic characteristics of the crystal, such as the heat of sublimation, phase transition temperature or the stability of the given phase at given pressure and temperature can be predicted.
The recent advent of computational chemistry consisting in development of high-performance quantum-chemical methods, statistical-thermodynamic models and a broad availability of the computational resources has enabled to predict many properties of various molecular crystals with a sufficient accuracy, sometimes even approaching the uncertainty level of the experimental determinations of such properties. To capture the interplay of different intermolecular short-range and long-range forces acting in the crystals, computational methodologies need to be robust and versatile enough to ensure their broad applicability and transferability without any dramatic loss of accuracy. A combination of periodic density functional theory (DFT) calculations, ab initio wavefunction methods and the quasi-harmonic approximation has established as the workhorse of our first-principles predictions of cohesive properties of molecular crystals.
This contribution gives an account on our recent first-principles predictions of the heat of sublimation, sublimation pressures or phase diagrams and their computational uncertainties. The development of our computational methodologies, based on the hybrid many-body interaction model combining quantum-chemical short-range and cheaper long-range regimes, over time and the appropriate range of applicability is emphasized in this context. At the beginning, only non-polar hydrocarbons were investigated and a primitive point-charge embedding scheme was used to treat their long-range electrostatics. Next, more polar species were considered and we opted for a polarizable force-field instead of mere point charges. Here, we managed to qualitatively predict the phase diagram of three polymorphs of methanol, with the underlying computational accuracy in terms of the Gibbs energy well in the sub-kJ/mol accuracy. Only recently, we have started using cheap periodic quantum-chemical methods to treat the long-range forces which proved to be essential for crystals of proteinogenic amino acids or alkylmethylimidazolium-based ionic liquids. Counterintuitively, the increasing rate of ionicity in these crystals does not grant a higher accuracy of the calculations since the induction or charge transfer effects become too important to be neglected or simply described by a generalizing empiric force-field.
