Cw. Murray et al., A STUDY OF O-3, S-3, CH-2, AND BE-2 USING KOHN-SHAM THEORY WITH ACCURATE QUADRATURE AND LARGE BASIS-SETS, The Journal of chemical physics, 98(9), 1993, pp. 7145-7151
Density functional calculations with accurate quadrature and large bas
is sets are reported on ozone, thiozone, methylene, and the beryllium
dimer. The Kohn-Sham self-consistent procedure is used throughout and
nonlocal gradient corrected functionals (B-LYP, B-P) are used. The sys
tems and properties chosen are notoriously difficult problems for accu
rate computation, nevertheless, the density functional methods perform
well. It is found, in line with calculations on more straightforward
problems, that the functionals involving gradient corrections are more
reliable than local density functionals. The B-LYP prediction is with
in 5 kcal/mol of the most accurate ab initio data on the C2v-D3h energ
y gap in ozone and thiozone. The B-LYP functional performs reasonably
well on the vibrational frequencies of ozone whilst the B-P functional
gives very accurate results for the structural parameters of ozone. B
-LYP is used to estimate the singlet-triplet splitting in methylene wi
th the triplet treated in both a restricted and unrestricted formalism
. In either case the correct energy gap is reproduced within 2 kcal/mo
l. The bond length and low frequency of the beryllium dimer are reprod
uced accurately by the gradient corrected functionals but the binding
energy is overestimated. However, B-LYP only overestimates the binding
energy by 4 kcal/mol, which is an encouraging result given the diffic
ulty of the problem. This paper can therefore be seen as further evide
nce that density functionals methods, and in particular methods based
on nonlocal functionals, can predict molecular properties reasonably a
ccurately for systems that have multireference character.