Lanthanide induced 1H and 13C NMR shifts and their use for geometry analysis with alicyclic compounds

Based on computations with conformationally rigid substrates the ambiguities involved in the geometrical analysis of pseudocontact shifts are demonstrated. The minimal agreement factors R and the corresponding lanthanide positions (usually more than one) are extremely dependent upon the chosen structural data for the model substrate and on the errors of the experimental shift values. It is observed, that a difference or improvement of R factors of less than 3% is not significant in most cases. This is so, even on the basis of Ytterbium induced ¹³C shifts, which are found to be more accurate than ¹H shifts and free from contact contributions within the experimental error. Using lanthanide induced shifts some ¹³C signal assignment problems are discussed. The computed minima for five substituted norbornanols indicate a lanthanide position with an L . . . O distance of d = 2·5 Å for the secondary alcohols and an orientation avoiding gauche interactions with the two neighbouring CC bonds. Similar computed results are obtained with five bicyclic ketones, except for an L . . . O distance from 3 to 4 Å increasing with steric hindrance. The particular problems with the analysis of symmetrical compounds like cyclohexyl derivatives are pointed out.