Predictions of relative structural trends from ab initio derived standard geometry functions

Abstract

Some aspects are reviewed of our series “Ab initio Studies of Structural Features Not Easily Amenable to Experiment”, in which, so far, more than 300 structures of almost 100 basic organic molecules have been investigated by gradient geometry optimizations on the 4-21G level. Together with 4-21G structures from other groups, these results currently represent one of the most extensive, consistent bodies of computed structural information whose credibility and accuracy has proved to be equivalent to that of successful experimental work. It is a particular advantage of the calculated structural trends that they include information on unstable species (e.g. energy maxima and saddle points) which cannot be observed experimentally. Therefore, recurrent patterns in local geometries can be studied systematically, and standard structural functions can be derived which reproduce the dependence of typical bond distances and angles on torsional degrees of freedom. It is shown that, in composite molecules, some parameters can be modelled (i.e. their conformational dependence can be predicted) by simple combinations of component standard geometry functions.