De novo prediction of ground state multiplicity and structural isomerism for transition metal complexes

Prediction of the ground state geometries and multiplicities for 33 transition metal tetrachlorides has been carried out using two different levels of quantum mechanics: semiempirical and density functional theory. All data regarding geometry and spin state provided by both computational methods were compared with experimental data when available. The calculations were performed for all possible spin multiplicities. The most important geometries for coordination number four (tetrahedral, square-planar, dodecahedral, and disphenoidal), as well as less symmetric structural isomers, were evaluated. A match between both computational methods in terms of predicted ground state multiplicity and geometry was found for 26 species, which translated into almost 80% agreement. Even though the PM3(tm) geometry prediction protocol involved more steps for isolating a feasible global minimum, the aggregate of these calculations was still orders of magnitude faster than DFT calculations using extended basis sets. The calculations indicate that caution is needed in the application of the PM3(tm) method to very high-spin transition metal complexes, but point to the suitability of very rapid semiempirical methods for reliable prediction of structural and 'spin' isomers, and hence their use in an efficient de novo design protocol for transition metal complexes. (C) 2004 Elsevier B.V. All rights reserved.