Proceedings of XVIII Tchugaev Conference on the Chemistry of Coordination Compounds, Moscow, Nauka, p. 88 (1996)
Among the methods of theoretical investigation of coordination compounds, molecular mechanics seems to be an optimal choice. However, conventional MM force fields typically do lack metal-related force field parameters. The common approach consists in trial-and-error calibration of these parameters; this is time-consuming and sometimes unreliable. We developed an alternative approach. Suggest that we know "organic"force field parameters as well as experimental structures of a series of compounds of interest. Assuming that experimental structures are "equilibrium" ones, the energy gradient should be zero. Let us calculate (MM) energy gradient for experimental structures using purely organic force field with no metal-dependent terms. Calculated gradient would be, naturally, non-zero as it represents only ligand-driven forces which act on metal center and ligating atoms; however, calculated forces should be exactly compensated by metal-driven forces. In other words, thus calculated forces give an "imprint" for the derivation of metal-dependent forces. Technically, it transforms the parametrization problem into the problem of least-squares solution of the system of (sometimes, linear) equations. Using this approach, we found a force field parameters for complexes with aza- and thia-ligands for both conventional and Gillespie-Kepert parametrization. For a broad range of ligands, using novel Gillespie-Kepert force field, we reproduced geometry of Ni(II) and Fe(II) complexes with pretty good accuracy.