L. Seijo et Z. Barandiaran, APPLICATIONS OF THE GROUP-FUNCTION THEORY TO THE FIELD OF MATERIALS SCIENCE, International journal of quantum chemistry, 60(1), 1996, pp. 617-634
The group-function theory, as proposed by McWeeny for the study of wea
k intermolecular interactions and developed by Huzinaga in the context
of valence-electron methods, is shown to be applicable to the ab init
io study of tunable solid-state laser materials made of defective ioni
c crystals. The applicability of the theory relies on the existence of
local electronic states (to which the demonstrated/potential laser ac
tivity is ascribed), which are essentially localized in a small cluste
r of atoms including the defect and whose electron correlation interac
tions with the surrounding crystal components are negligible. Accordin
g to the group-function formalism, it is possible (a) to neglect elect
ron correlation effects beyond the defect cluster and (b) to define a
quantum mechanical embedding potential which embodies the rest of the
so-called host effects. Computationally, the theory becomes applicable
as the embedding potential is approximated through ab initio model po
tentials (AIMP). The results of AIMP embedded-cluster calculations dem
onstrate that it is possible to calculate the local structure and spec
troscopy of the active defect at an ab initio level, the attainable ac
curacy being comparable to the usual one in molecular ab initio studie
s in the gas phase. Also, in this article, we present a systematic stu
dy of the local distortions produced upon doping divalent first-series
transition-metal ions in rock-salt oxides, MO:Me(2+) (M = Mg, Ca, Sr;
Me = Sc-Zn) and Tl+ in KMgF3 and KF hosts. This study leads to the ca
lculation of the local structures of the defects in these materials, w
hich have not been measured. The results suggest that the use of the m
ismatch of the empirical ionic radii of the impurity and the substitut
ed ion in order to predict local distortions in doped ionic crystals i
s not significant when it is smaller than 0.1 Angstrom, and when it is
larger, it should be weighted by a reduction factor depending on the
host. For the first-series divalent transition-metal ion impurities, t
his factor is shown to be 0.15 for SrO, 0.25 for CaO, and around 0.50
for MgO. (C) 1996 John Wiley & Sons, Inc.