Experimental equilibrium structures: application of molecular dynamics simulations to vibrational corrections for gas electron diffraction.

A general method is described that allows experimental equilibrium structures to be determined from gas electron diffraction (GED) data. Distance corrections, starting values for amplitudes of vibration and anharmonic "Morse" constants (all required for a GED refinement) have been extracted from molecular dynamics (MD) simulations. For this purpose MD methods have significant advantages over traditional force-field methods, as they can more easily be performed for large molecules, and, as they do not rely on extrapolation from equilibrium geometries, they are highly suitable for molecules with large-amplitude and anharmonic modes of vibration. For the test case Si(8)O(12)Me(8), where the methyl groups rotate and large deformations of the Si(8)O(12) cage are observed, the MD simulations produced results markedly superior to those obtained using force-field methods. The experimental equilibrium structure of Si(8)O(12)H(8) has also been determined, demonstrating the use of empirical potentials rather than DFT methods when such potentials exist. We highlight the one major deficiency associated with classical MD--the absence of quantum effects--which causes some light-atom bonded-pair amplitudes of vibration to be significantly underestimated. However, using C(3)N(3)Cl(3) and C(3)N(3)H(3) as examples, we show that path-integral MD simulations can overcome these problems. The distance corrections and amplitudes of vibration obtained for C(3)N(3)Cl(3) are almost identical to those obtained from force-field methods, as we would expect for such a rigid molecule. In the case of C(3)N(3)H(3), for which an accurate experimental structure exists, the use of path-integral methods more than doubles the C-H amplitude of vibration.