Is it possible to create a better semiempirical mo theory ?

✍ Scribed by Einar Lindholm

• Publisher: Elsevier Science
• Year: 1988
• Tongue: French
• Weight: 298 KB
• Volume: 44
• Category: Article
• ISSN: 0040-4020
• DOI: 10.1016/s0040-4020(01)86240-2

✦ Synopsis

An outline is given of a future semiempirical MO method, intended for use in organic chemistry.

Dewar 11.21 has pointed out that it seems to be impossible to cnlcuhte the total energy of a niolecula with sufficient accuracy by use of ab-initio Cl methods. For e.g. benzene the error amounts to 1031 kcal/mof It/. It is true that cance~tion of errors takes ptace when energy differences are considered, but the cancellation is never perfect. Ilrercfore, calculations of properties like excitation energies and electron affinities have nppenred to have very large errors when compared to experiments. 11 would therefore be desieble to have sufficiently accurate semiempirical methods in which the correlation cttctgies have been included.

Even if the methods, developed by Dewar and coworkers, have appeared to be extremely useful, it is of in&rest to discuss how a future MO method could be developed. We will below discuss some features of such a method.

I. Since correlation energies should be included, the theory cannot stnrt from the Ilartree-rock theory, since Ike addition nfterwards of correlation