Theoretical computation of the heats of hydrogen abstraction reactions

Hydrogen abstraction reactions of molecules leading to highly unsaturated radicals are studied. It is shown that fairly good results may be obtained with polarized valence split basis sets at the QCI level of methodology. Contrary to the usually admitted opinion, the further extension of basis sets does not appreciably improve the results, nor does the introduction of corrections for the triple substitutions. The use of unrestricted perturbation theory, even at fourth order will generally lead to poor results. The recourse to projection techniques improves but slightly the results.