Density Functional Theory in Transition-Metal
Chemistry: Relative Energies of Low-Lying States of Iron
Compounds and the Effect of Spatial Symmetry Breaking
Anastassia Sorkin, Mark A. Iron, and Donald G. Truhlar*
Department of Chemistry and Supercomputing Institute, UniVersity of Minnesota,
Minneapolis, Minnesota 55455-0431
Received September 25, 2007
Abstract: The ground and lower excited states of Fe
2
, Fe
2
-
, and FeO
+
were studied using a
number of density functional theory (DFT) methods. Specific attention was paid to the relative
state energies, the internuclear distances (r
e
), and the harmonic vibrational frequencies (ω
e
). A
number of factors influencing the calculated values of these properties were examined. These
include basis sets, the nature of the density functional chosen, the percentage of Hartree-
Fock exchange in the density functional, and constraints on orbital symmetry. A number of
different types of generalized gradient approximation (GGA) density functionals (straight GGA,
hybrid GGA, meta-GGA, and hybrid meta-GGA) were examined, and it was found that the best
results were obtained with hybrid GGA or hybrid meta-GGA functionals that contain nonzero
fractions of HF exchange; specifically, the best overall results were obtained with B3LYP, M05,
and M06, closely followed by B1LYP. One significant observation was the effect of enforcing
symmetry on the orbitals. When a degenerate orbital (π or δ) is partially occupied in the
4
Φ
excited state of FeO
+
, reducing the enforced symmetry (from C
6v
to C
4v
to C
2v
) results in a
lower energy since these degenerate orbitals are split in the lower symmetries. The results
obtained were compared to higher level ab initio results from the literature and to recent PBE+U
plane wave results by Kulik et al. (Phys. Rev. Lett. 2006, 97, 103001). It was found that some
of the improvements that were afforded by the semiempirical +U correction can also be
accomplished by improving the form of the DFT functional and, in one case, by not enforcing
high symmetry on the orbitals.
1. Introduction
Transition-metal centers have great versatility in their bond-
ing. Consequently, accurate theoretical treatment of transi-
tion-metal chemistry demands a flexible theoretical frame-
work that treats all energetically accessible spin states, spin
couplings, and valence states in an even-handed fashion,
which in wave function theory requires a multiconfigurational
treatment. This poses a difficult problem for the Kohn-Sham
density functional theory (DFT) because all information on
the multiconfigurational character of the wave function is
contained in the exchange-correlation energy E
XC
, which is
computed from an electron density that is in turn obtained
from a single Hartree product of orbitals.
1,2
Even when a
single antisymmetrized product (Slater determinant) of
orbitals does not describe the true electronic wave function
well, the Kohn-Sham ground-state energy is correct if one
solves the Kohn-Sham equations for the orbitals that give
the lowest-energy solution,
1
even though the Kohn-Sham
orbitals and eigenvalues, except for the highest orbital
eigenvalue,
3
do not have any strict physical significance,
4-6
and the Kohn-Sham Slater determinant that generates the
accurate electron density may have different spin properties
than the true wave function.
7-9
Nevertheless, great progress
has been made in understanding transition-metal chemistry
in terms of the Kohn-Sham theory.
10-13
* Corresponding author e-mail: truhlar@umn.edu.
307 J. Chem. Theory Comput. 2008, 4, 307-315
10.1021/ct700250a CCC: $40.75 © 2008 American Chemical Society
Published on Web 01/01/2008