Computational studies of crystal structure and bonding.

The analysis, prediction, and control of crystal structures are frontier topics in present-day research in view of their importance for materials science, pharmaceutical sciences, and many other chemical processes. Computational crystallography is nowadays a branch of the chemical and physicals sciences dealing with the study of inner structure, intermolecular bonding, and cohesive energies in crystals. This chapter, mainly focused on organic compounds, first reviews the current methods for X-ray diffraction data treatment, and the new tools available both for quantitative statistical analysis of geometries of intermolecular contacts using crystallographic databases and for the comparison of crystal structures to detect similarities or differences. Quantum chemical methods for the evaluation of intermolecular energies are then reviewed in detail: atoms-in-molecules and other density-based methods, ab initio MO theory, perturbation theory methods, dispersion-supplemented DFT, semiempirical methods and, finally, entirely empirical atom-atom force fields. The superiority of analyses based on energy over analyses based on geometry is highlighted, with caveats on improvised definitions of some intermolecular chemical bonds that are in fact no more than fluxional approach preferences. A perspective is also given on the present status of computational methods for the prediction of crystal structures: in spite of great steps forward, some fundamental obstacles related to the kinetic-thermodynamic dilemma persist. Molecular dynamics and Monte Carlo methods for the simulation of crystal structures and of phase transitions are reviewed. These methods are still at a very speculative stage, but hold promise for substantial future developments.