Abstract
We have shown that the MNDO‐PM3 semiempirical molecular orbital method is capable of yielding calculated structures of oxaphosphetanes and ylides that are comparable with the experimentally known structures.
The MNDO‐PM3 method also reproduces the geometric and thermodynamic parameters of the “mythyical Witting half‐reaction”, previously calculated using a more elaborate and costly ab initio molecular orbital approach. In addition, using the SADDLE routine, we were able to find a transition state for this half‐reaction. This transition state resembles geometrically the transition state found using the ab initio approach (4‐31G* level). The energy of activation was calculated to be 9.2 kcal/mole, which is fairly comparable with the ab initio reslt of 5.2 kcal/mole.
According to our calculations, and in general agreement with findings of Volatron and Eisenstein [11], the mythical Witting reaction reaction would be expected to proceed through a planar transition state for the cycloaddition of the ylide and the aldehyde in a process that superficially seems to be a synchronous one. However, an analysis of the evolution of the bond orders and the localized molecular orbitals throughout the reaction indicates that the PO bond has not been formed in the transition state, whereas the CC bond is about 40% formed. Therefore, the mythical Wittig half‐reaction is best described as a very asynchronous cycloaddition (a bordeline two step mechanism) with a planar geometry in the transition state, composed of a multi‐centered mPCCO bond orbital and with the phosphorus and the oxygen atoms sharing a strong attractive charge interaction. A similar description results from computations of the Wittig reactions of H~3~CHCH~3~ and (C~6~H~5~)~3~PCHCH~3~ with H~3~CCHO, which is inconsistent with the Vedejs geometry.