Abstract
The pi‐electronic structures of the trans and cis diastereomers of acrolein are studied within the framework of the Pople—Pariser—Parr approximation in which: (1) Resonance integrals are evaluated by a formula originally derived by Ohno, (2) valence‐state ionization potentials of non‐carbon atoms are chosen by the constraint that Koopmans' theorem is satisfied literally for the first ionization potential of the molecule, and (3) all penetration integrals are neglected. The effects of including nonnearest‐neighbor resonance integrals and of using ``fixed'' effective atomic charges in lieu of ``self‐consistent electronegativity'' values are investigated. Also examined is the effect of modifying Ohno's resonance integral formula when used for heteropolar bonds. The optimum agreement between calculations and the limited experimental evidence available occurs when ``self‐consistent electronegativity'' effective atomic charges are used along with nonnearest‐neighbor resonance integrals computed from a modified Ohno formula. Application of the ``best'' acrolein procedures to the glyoxal molecule suggests either that one cannot use the modified Ohno formula to obtain good spectroscopic intervals and reasonable charge densities from a single set of parameters unless the Koopmans' theorem constraint is relaxed or else that the experimental upper limit to the ionization potential is in error by 0.5 eV or more.