Date Published: August 01, 2016
Publisher: International Union of Crystallography
Author(s): Edward O. Pyzer-Knapp, Hugh P. G. Thompson, Graeme M. Day.
An empirically parameterized intermolecular force field is developed for crystal structure modelling and prediction. The model is optimized for use with an atomic multipole description of electrostatic interactions.
The role of computational modelling in understanding the molecular organic solid state is developing rapidly, and computer simulations are key to understanding a wide range of properties of molecular solids, such as lattice energies (Nyman & Day, 2015 ▸), mechanical properties (Karki et al., 2009 ▸), solubility (Palmer et al., 2008 ▸, 2012 ▸), lattice dynamics (Li et al., 2010 ▸; King et al., 2011 ▸) and molecular dynamics (Gavezzotti, 2013 ▸), disorder (Habgood et al., 2011 ▸), conformational preferences (Thompson & Day, 2014 ▸) and polymorphism (Cruz-Cabeza & Bernstein, 2014 ▸). The field of crystal engineering is concerned with relationships between molecular structure and crystal structure, whose computational embodiment is the ever-developing field of crystal structure prediction (Day et al., 2009 ▸; Bardwell et al., 2011 ▸; Day, 2011 ▸; Price, 2014 ▸).
Our aim in empirically determining the best set of parameters to describe intermolecular interactions remains the same as that stated by Williams: ‘Our optimum intermolecular force field is one which gives the best fits to observed crystal structures and heats of sublimation. The goodness-of-fit is determined by minimization of the crystal energies using the force field to be tested, and comparing the resulting relaxed structures with the observed ones’ (Williams, 1999 ▸). Here, we describe the form of the force field, our strategy in optimizing the adjustable parameters and the selection of structures and energies to which we have parameterized.
We present a revision of the W99 intermolecular force field for modeling molecular organic crystals. The force-field parameters describing hydrogen-bond interactions have been optimized to work optimally with an atomic multipole model of electrostatic interactions. We also parameterize versions of the force field that are compatible with using polarized multipoles, derived from the charge density of a molecule embedded in a continuum dielectric (PCM) approximation of the crystalline environment. Low-temperature crystal structures have been used in the re-parameterization to minimize the extent to which thermal expansion is incorporated into the empirical parameters, making the resulting force field suitable for including thermal effects, via lattice or molecular dynamics methods.