The interpretation of the EPR spectra of radicals can be a powerful tool in
obtaining structural information. There are many examples of this in the f
ield of molecular biology, and we are particularly interested in a reliable
method for modeling G tensors as a function of different structures. We be
gin this study by examining simple organic radicals and testing how accurat
e the intermediate neglect of differential overlap (INDO/S) semiempirical m
odel can be when put to this purpose. The INDO/S model is fast and has been
used with some success in examining the photochemistry of very large syste
ms. Since this model is accurate in predicting electronic spectra, we suspe
ct also that the propagator needed to estimate the orbital Zeeman-spin-orbi
t cross terms might be quite accurate. In addition, since this model is als
o parameterized for many elements of the periodic table, we expect this mod
el to be of general utility. We begin here with a preliminary study of G te
nsors calculated for simple radicals, testing the accuracy of the calculate
d results and their sensitivity to geometry. We also report preliminary res
ults on calculated G tensors for the special pair radical cation formed as
the first step in the electron transfer in photosynthesis. (C) 1999 John Wi
ley & Sons, Inc.