The preference of 1-methylallyl polar organometallics and carbanions for cis rather than for trans geometries

In agreement with experimental equilibrium studies in solution, ab initio calculations, using diffuse function augmented split valence basis sets, favor cis geometries of 1-methylallyl lithium, sodium, potassium, and rubidium by 0.7, 1.1, 2.5, and 2.6 kcal/mol, respectively. These results, and the even greater cis preference of the 1-methylallyl anion over trans, 4.7 kcal/mol (MP2/6-31+G//3-21G) show the contrary conclusions from an early ICR study and from semiempirical theory to be incorrect. Later gas phase experiments demonstrated that stereointegrity was not maintained under the ICR conditions. Semiempirical theory is shown here to be inherently deficient for the examination of alkyl-substituted carbanions. The surprisingly large cis preferences of the 1-methyl-allyl anion systems are attributed to the largely electrostatic attraction between the positively charged methyl hydrogens and the remote allyl anion terminus. This attraction is largest in the free anion and attenuated to an increasingly greater extent the smaller the alkali metal gegenions become. The allyl moieties in the metal 1-methylallyl derivatives are calculated to be quite unsymmetric, in agreement with NMR observations.