The manifestations of the symmetry-breaking artifact in three-electron-bond
ed systems have been investigated at several computational levels including
second-order Moller-Plesset perturbation theory (MP2), coupled cluster (CC
), and Bruckner-coupled cluster (B-CC) theories. The model systems, [H(n)X4
XH(n)]+(X=Ne, F, O, N, Ar, Cl, S, P; n=0-3) cover all types of three-electr
on bonds that can possibly take place between atoms of the second and third
rows of the Periodic Table. The critical interatomic distance beyond which
symmetry breaking begins to take place at the Hartree-Fock and Moller-Ples
set levels are determined for each model system. Their magnitude are found
to obey regular tendencies which are related to the compactness of the orbi
tals involved in the three-electron bonds. In all model systems, the onsets
of symmetry-breaking at the MP2 level are greater or equal to the equilibr
ium bonding distance between the XHn fragments. The symmetry-breaking artif
act results in severe discontinuities in the dissociation curves at the MP2
level. The CC level pushes away the occurrence of the artifact to larger d
istances but do not remove the discontinuities. The artifact is practically
cured at the B-CC level with perturbative treatment of triple excitations.
The onset of symmetry-breaking may in some cases be shortened by substitue
nt effects, to the extent that it becomes shorter than the equilibrium bond
ing distance like in the Me4O2+ and Me2F2+ cation radicals that are found t
o be symmetry-unstable even in their equilibrium geometries. The artifact c
arries over to unsymmetrical systems that display close functional resembla
nce to symmetrical systems, leading to convergence difficulties, erroneous
geometries, and unphysical localization of the electronic charge. An econom
ical alternative to the MP2 method, based on the average quadratic coupled-
clusters (AQCC), is proposed for such cases, or in cases some stretched thr
ee-electron-bonded systems or full dissociation curves are to be investigat
ed. (C) 2001 American Institute of Physics.