THE USE OF POTENTIAL-ENERGY CALCULATIONS IN THE SOLUTION OF CRYSTAL STRUCTURES STUDIED BY POWDER DIFFRACTION METHOD

A.V.Dzyabchenko¹ and V.Agafonov²

¹Karpov Institute of Physical Chemistry, 10 Obukha, 103064 Moscow
²Laboratoire de Chimie Physique, Faculte de Pharmacie, 31 avenue de Monge, 37200 Tours, France

Determination of organic crystal structures by powder diffraction (PD) experiments is a new reality. Of the methods proposed to use in phase problem solution to provide initial model for further refinement by Rietveld techniqie, one of most promising is that based on the modelling optimal crystal packing.

We present our tools for realization of the above ideas. They include the PMC program of crystal packing optimization and the CRYCOM program of structure similarity searching. We provide numerical examples of structure solutions made by energy minimization. They are remarkably different with respect to molecular shape and interaction potential determining crystal packing:

- A metastable form of piracetam (2-oxo-1-pirrolidine- acetamide) whose packing was governed to a great extent by intermolecular hydrogen bonds. As a result of global search we found a lot of piracetam molecular packings corresponding to different H-bonded motifs of chain and layer configuration, in addition to two previously known polymorphs presenting dimer configuration. The model further confirmed by PD data and refined by Rietveld method was one of the lowest ones in potential energy.

- C₆₀ fullerene pressure-induced crystal phases. As a result of energy calculations we suggested revised models of orthorhombic (O), tetragonal (T), and rhombohedral (R) polymerized phases which, on one hand, showed at least no poorer agreement with the observed PD data and, on the other hand, provided much more energetically favourable packing of C₆₀ chains (O phase), or layers of tetragonal or hexagonal configuration (T and R phases, resp.) than the models existed in the literature (Nunez et al. 1995). Moreover, the new models were found more easy to explain their origination from normal C₆₀

- via hypothetical ('precursor') packings predicted as local energy minima with favourable orientation of double C-C bonds regarding the [2+2] cycloaddition reaction and configurationally similar to the corresponding polymerized phases. Finally, we studied the crystal packings of dimeric C₆₀ molecules and suggested a model of crystal disorder in the pressure-induced dimerized C₆₀ phase discovered quite recently to form in the region of low pressures and temperatures.