The structure and properties of molecular crystals have long been of great interest, not just for fundamental reasons but also due to their diverse applications ranging from drug design and food chemistry to hydrogen storage and nonlinear optics. The importance of molecular crystals has led to the employment of various computational methods to elucidate and understand their properties. While crystal structure prediction (CSP) is a very active research area given its importance for pharmaceutical applications, many computational studies are also focusing on vibrational properties such as phonons and thermal displacement parameters, proton-transfer as well as electronic properties. However, our understanding of most of these phenomena in molecular crystals is still far from complete. <br/>Density functional theory supplemented by a high-level van der Waals (vdW) model, such as the many-body dispersion (MBD) method has shown to provide accurate results for determining the lattice energies, vibrational contributions and relative energetics of molecular crystals (PRL 113, 055701; Phys. Chem. Chem. Phys. 21, 24333). More specifically, PBE0+MBD yields remarkably accurate structures and stability rankings for systems studied in the CSP blind test and other molecular crystals (Sci. Adv. 5, eaau3338). However, the high computational cost of PBE0+MBD hinders a high throughput screening of systems of large unit cells. In this work, we assess the accuracy of Density Functional Based Tight Binding method including many body dispersion (DFTB-MBD) for the investigation of molecular crystals’ properties, both with and without a neural network correction for the Pauli repulsion (J. Phys. Chem. Lett. 11, 6835). In particular, since the reliable calculation of free energy contributions is one of the key factors of CSP studies, we are also focusing on the computation of vibrational spectra. Besides, low-frequency terahertz (THz) phonon modes are crucial for an accurate treatment of the vibrational free energy. The reduction in computational cost that our approach provides, allows us to extensively study systems that consist of large unit cells, paving the way for a better understanding of the structural characteristics and properties of molecular crystals.

Reliable density-functional approach for the modeling of molecular crystal polymorphs

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.