Excited-state absorption (ESA) spectra of π-conjugated
compounds
are frequently calculated by (quadratic response) time-dependent density
functional theory, (QR) TD-DFT, often giving a reasonable representation
of the experimental results despite the (known) incomplete electronic
description. To investigate whether this is inherent to the method,
we calculate here the ESA spectra of small-to-medium-sized oligophenylenevinylenes
(nPV) and oligothiophenes (nT) using
QR TD-DFT as well as CASPT2 based on CASSCF geometries. CASPT2 gives
indeed a reliable, theoretically correct description of the ESA features
for all compounds; the computational effort can be reduced without
significant loss of accuracy using TD-DFT geometries. QR TD-DFT, based
on BHandHLYP and CAM-/B3LYP functionals, fails on short
n
Ts but provides a reasonable description
for spectral positions of
n
PVs and long
n
Ts. The failure
on short
n
Ts is, however,
only partly due to the incomplete configuration description but, in
particular, related to an improper MO description, resulting in an
asymmetric energy spacing of the occupied vs unoccupied MOs in the
DFT scheme. Longer
n
Ts,
on the other side, adapt approximately the MO scheme for alternant
hydrocarbons just like in
n
PVs, while contributions by two triplet excitations combined to a singlet
(which inhibits an accurate treatment of polyenes with standard TD-DFT)
do not play a relevant role in the current case. For such “well-behaved”
systems, a reasonable representation of ESA spectra is found at the
QR TD-DFT level due to the rather small energy shifts when including
higher-order excitations.