Chiral interactions in azobenzene dimers: A combined experimental and theoretical study

Article abstract of DOI:10.1002/chem.200401081

To investigate interchromophore interactions in azobenzene polymers, we have undertaken a thorough spectroscopic analysis of the azodye [(5)-3-pivaloyloxy-1-(4'-nitro-4-azobenzene)pyrrolidine] by modeling the repeating unit of poly[(S)3-methacryloyloxy-1-(4'-nitro-4-azobenzene) pyrrolidine) and its dimeric derivative whose synthesis is presented here. The analysis of the electronic and Raman spectra of the azodye in several solvents is based on a previously proposed model for polar chromophores in solution. Electronic and CD spectra of the dimeric unit are collected and analyzed within the framework of a new model. On the basis of the information collect ed from the spectroscopic analysis of the solvated dye, this model accounts for interchromophore interactions in the dimer. The large CD signal measured for the dimer (amounting to about a third of the signal measured for the polymer) suggests the presence of important chiral interactions in the dimeric unit, and is modeled in terms of a right-handed relative orientation of the two chromophores.

Full text of DOI:10.1002/chem.200401081

FULL PAPER
DOI: 10.1002/chem.200401081
Chiral Interactions in Azobenzene Dimers: A Combined Experimental and
Theoretical Study
Anna Painelli,*[a] Francesca Terenziani,[a, b] Luigi Angiolini,[c] Tiziana Benelli,[c] and
Loris Giorgini[c]
Abstract: To investigate interchromo-
phore interactions in azobenzene poly-
mers, we have undertaken a thorough
spectroscopic analysis of the azodye
[(S)-3-pivaloyloxy-1-(4’-nitro-4-azoben-
zene)pyrrolidine] by modeling the re-
peating unit of poly[(S)-3-methacryloy-
loxy-1-(4’-nitro-4-azobenzene)pyrroli-
dine) and its dimeric derivative whose
synthesis is presented here. The analy-
sis of the electronic and Raman spectra
of the azodye in several solvents is
based on a previously proposed model
for polar chromophores in solution.
Electronic and CD spectra of the di-
meric unit are collected and analyzed
within the framework of a new model.
On the basis of the information collect-
ed from the spectroscopic analysis of
the solvated dye, this model accounts
for interchromophore interactions in
the dimer. The large CD signal mea-
sured for the dimer (amounting to
about a third of the signal measured
for the polymer) suggests the presence
of important chiral interactions in the
dimeric unit, and is modeled in terms
of a right-handed relative orientation
of the two chromophores.
Keywords:
azo compounds
·
circular
dichroism Raman
·
spectroscopy · solvatochromism ·
UV/Vis spectroscopy
Introduction
dispersed-red 1 is a common reference chromophore.[2] Poly-
mers functionalized with azobenzene side groups (azopoly-
mers) have been synthesized as potentially interesting mate-
rials for several applications: a fairly large concentration of
chromophores can in fact be loaded into the polymers to
obtain films of good optical quality. The properties of the
films can be finely tuned by adjusting the chromophore con-
centration, by changing the D and A substituents in the azo
dye, or by altering the nature of the polymeric backbone.[3]
Azodyes undergo a photoinduced trans–cis isomerization
that is responsible for the intriguing photochromic behavior
of azopolymers.[4] In particular, all-optical poling has been
demonstrated by shining linearly polarized light on azopoly-
mer films.[5] Even more impressive is the possibility of writ-
ing relief gratings over the surface of a film illuminated by
interference patterns of linearly polarized laser beams.[6]
Conformational chirality can be optically induced in azo-
polymer films. Elliptically polarized light induces chirality
on amorphous achiral films.[7] Moreover the resulting chiral
material can be switched between the two enantiomeric
structures by simply alternating irradiation with left and
right circularly polarized light.[8] In achiral films of liquid-
crystalline azopolymers, circularly polarized light (without a
linear component) can induce chirality.[9] In chiral polymers,
in which the azodye is grafted onto the polymeric backbone
through an enantiomerically pure chiral bridge, the native
Donor–acceptor (D–A)-substituted azobenzenes are a well-
known family of organic dyes whose color can be tuned by
adjusting the strength of the D and/or A end groups.[1]
These molecules, which belong to the wider class of D–p-A
(or push–pull) chromophores, have been extensively investi-
gated for second-order NLO applications. In this context,
[a] Prof. Dr. A. Painelli, Dr. F. Terenziani
Dipartimento di Chimica GIAF and INSTM UdR-Parma
Università di Parma
viale delle Scienze 17/a, 43100 Parma (Italy)
Fax : (+39)0521-905-556
E-mail: anna.painelli@unipr.it
[b] Dr. F. Terenziani
Synth›se et Electrosynth›se Organiques (CNRS, UMR 6510)
Institut de Chimie
UniversitØ de Rennes 1, Campus de Beaulieu
Bât. 10 A, 35042 Rennes Cedex (France)
[c] Prof. Dr. L. Angiolini, Dr. T. Benelli, Dr. L. Giorgini
Dipartimento di Chimica Industriale e dei Materiali and INSTM
UdR-Bologna, Università di Bologna
viale Risorgimento 4, 40136 Bologna (Italy)
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the authors.
Chem. Eur. J. 2005, 11, 6053 – 6063
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
6053

polymers show
a
well-pro-
nounced circular dichroism
(CD) signal in the region of the
visible absorption of azodyes,
demonstrating that the chirality
of the bridge induces interchro-
mophoric interactions with
a
predominant helicity.[10] Succes-
sive cycles of irradiation with
circularly polarized light with
alternating L and R character
force the system into two enan-
tiomeric states with opposite
CD spectra.[11] The two enantio-
mers are temporally and ther-
mally stable, suggesting that cir-
cularly polarized light drives
chiral azopolymer films into a
genuine bistable regime.[12]
Photoinduced chirality, and
particularly the observation of
a bistable regime in chiral azo-
polymer films, makes these ma-
terials very promising for appli-
cations in optical computing
and more generally in the opti-
cal storage and manipulation of
information.[4] However, so far,
there is no clear understanding
of the mechanism of optically
induced chirality or of chiral in-
teractions in azopolymers. It is
generally accepted that chirality
Scheme 2. Synthesis of dimeric azodye D.
is induced in these materials as a consequence of chiral in-
terchromophore interactions, but no detailed modeling of
these interactions is available. Herein, we address this fun-
damental problem in a step-by-step procedure. We first con-
centrate on an isolated chromophore in solution, (S)-3-piva-
loyloxy-1-(4’-nitro-4-azobenzene)pyrrolidine (M, Scheme 1),
smallest section of the polymer where interchromophore in-
teractions are relevant. Electronic absorption and CD spec-
tra were collected for D in several solvents to gain informa-
tion on interchromophore interactions. Accordingly, we
present a model for the dimeric unit and for its spectroscop-
ic behavior in solution that takes advantage of information
collected from the spectroscopic analysis of M. The resulting
picture for interchromophore interactions is the first funda-
mental step toward the understanding of chiral interactions
in azopolymers and of supramolecular interactions in molec-
ular materials, in general.
Results and Discussion
Scheme 1. Molecular structure of azodye M.
The solvated chromophore: electronic and vibrational spec-
tra: The UV/Vis spectra of M (prepared as reported in ref-
erence [10a]) in CCl4, CHCl3, and DMSO are shown in the
top panel of Figure 1. The solvents have been selected as
representative of nonpolar, slightly polar, and strongly polar
media, respectively. The intense absorption in the visible
region is characteristic of azodyes and undergoes a well-pro-
nounced red shift with increasing solvent polarity. This
normal solvatochromism (bathochromism) is typically ob-
that models the repeating unit of the poly[(S)-3-methacryl-
oyloxy-1-(4’-nitro-4-azobenzene)pyrrolidine] polymer. An
extensive spectroscopic study of M in different solvents
leads to the definition of a reliable molecular model for this
chromophore. We synthesized the relevant dimeric unit, 2,4-
dimethyl-glutaric acid bis(S)-3-[1-(4’-nitro-4-azobenzene)]-
pyrrolidine ester (D, Scheme 2), which corresponds to the
6054
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6053 – 6063

Azobenzene Dimers
FULL PAPER
Figure 2. Raman spectra of M dissolved in different solvents, with excita-
tion line l = 647 nm. Left and right panels show experimental and calcu-
lated spectra, respectively. The experimental Raman intensities are re-
scaled to keep the height of the band at about 1140 cmꢀ1 approximately
constant. The calculated intensities have the same (arbitrary) units in all
the spectra. The vertical dotted lines are drawn as a visual guide.
Figure 1. a) Experimental absorption spectra of D (heavy lines) and M
(light lines) dissolved in CCl4, CHCl3, and DMSO. b) CD spectra of D in
CCl4, CHCl3, and DMSO.
Figure 3. The same as in Figure 2, but with excitation line at l = 476 nm.
served for D–p-A chromophores with a neutral ground
state, that is, for chromophores whose polarity increases
upon photoexcitation.[13]
Azodyes are nonfluorescent. Therefore, to collect addi-
tional spectroscopic information, we also measured Raman
spectra of M in the same solvents with several excitation
lines. The complete assignment of the vibrational bands is
beyond the scope of this work, here we discuss the Raman
spectra just to obtain information about the coupling be-
tween electronic and vibrational degrees of freedom, as re-
quired for the modeling of the electronic and CD spectra of
M and D.
The left-hand panels of Figure 2 and Figure 3 show
Raman spectra of M in the three solvents, collected with red
(l = 647 nm) and blue (l = 476 nm) excitation lines, re-
spectively. The red excitation falls, for all solvents, below or
just at the lowest energy edge of the absorption band so
that the spectra in Figure 2 are nonresonant, or at most pre-
resonant. The blue line corresponds to a resonant excitation
for all solvents. The comparison between nonresonant and
resonant spectra supports resonant amplification of the
group of bands located at about 1340 cmꢀ1 and of the band
at about 1590 cmꢀ1 with respect to other bands, suggesting
that these modes have the largest coupling with the elec-
tronic degrees of freedom involved in the absorption in the
visible region. Indeed, as the solvent polarity increases, the
lowest energy band in the group of bands located at about
1400 cmꢀ1 becomes more prominent. The dominant role of
Figure 4. Raman spectra of M in DMSO, collected with different excita-
tion lines. Left and right panels show experimental and calculated spec-
tra, respectively. The experimental Raman intensities are rescaled to
keep the height of the band at ꢁ1140 cmꢀ1 approximately constant. The
calculated intensities have the same (arbitrary) units in all the spectra.
The vertical dotted lines are drawn as a visual guide.
this band at about 1340 cmꢀ1 and of the band at 1590 cmꢀ1 is
nicely confirmed by data in Figure 4, which shows Raman
spectra collected with several excitation lines in DMSO,
where all excitation lines are resonant or preresonant.
Moreover, as the solvent polarity increases, the two bands at
about 1340 and 1590 cmꢀ1 undergo an appreciable red-shift
of about 7 and 5 cmꢀ1, respectively, while all other bands are
unaffected (within 2 cmꢀ1). Spectra collected with the yellow
(l = 568 nm) and green (l = 521 nm) lines in different sol-
vents are reported in the Supporting Information.
Chem. Eur. J. 2005, 11, 6053 – 6063
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6055

A. Painelli et al.
Modeling the spectroscopic properties of the isolated chro-
mophore: Electronic solvatochromism of D–p-A chromo-
phores is a very well-known phenomenon with a good cover-
age in recent literature.[13–17] Its basic physics is simple and is
related to the variation of the dipole moment of D–p-A
chromophores upon photoexcitation: the ground and excited
states are stabilized by differing amounts in polar solvents
and the transition energy acquires a large dependence on
the solvent polarity.[13] Whereas this simple mechanism de-
scribes the basic physics of electronic solvatochromism, it
fails to account for some more subtle phenomena, including
the dependence of absorption intensities and band shapes
on the solvent polarity, or the common observation that the
emission bands are narrower than the absorption bands. In
recent years, two of the authors have proposed a model for
D–p-A chromophores that nicely reproduces their spectro-
scopic behavior.[18,15] The model describes the electronic
structure of the chromophore in terms of two basis states,
jDAi and jD+Aꢀi, corresponding to the neutral and charge-
separated (zwitterionic) resonating structures of D–p-A
chromophores. The relevant electronic Hamiltonian is given
fractive index of the solvent, which shows minor variations
in common organic solvents, thus justifying the use of sol-
vent-independent molecular parameters.[21]
In polar solvents, a second component of the reaction
field appears owing to reorientation of the polar solvent
molecules around the polar solute.[21] To describe polar sol-
vation, we introduce an effective solvation coordinate, Q0,
whose coupling to the electronic degrees of freedom is mea-
sured by the relaxation energy eor.[21,18,22] At variance with
internal vibrations, the solvation coordinate describes a very
slow motion and is best treated in the classical approxima-
tion. An equilibrium position for Q0 is defined for each
chromophore in solution and depends in a self-consistent
way on the chromophore polarity.[22] However, thermal dis-
order is responsible for deviations of Q0 from the equilibri-
um. A solution of push–pull chromophores in polar solvents
can then be described as a collection of chromophores, each
one in equilibrium with the local configuration of the sur-
rounding solvent (i.e. with the local Q0), and the probability
of each configuration is weighted by the Boltzmann energy
distribution.[22,15] Local fluctuations of the reaction field in
polar solvents are therefore responsible for inhomogeneous
spectral broadening, which shows up in electronic and vibra-
tional spectra with a broadening of the absorption and non-
resonant Raman bands, and with the anomalous dispersion
with the excitation line of the Raman frequencies of strong-
ly coupled modes.[22,23]
by Equation (1),[20] where 1 = (1ꢀsz)/2 measures the weight
ˆ
ˆ
ˆ
of the zwitterionic state, and sx/z is the x/z Pauli matrix. The
two basis states mix to give a ground and an excited state.
pffiffiffi
ðMÞ
ð1Þ
^
^
H
¼ 2z1ꢀ 2tsx
el
The polarity of the chromophore is measured by 1, the
To describe electronic and vibrational spectra of solvated
M, a set of molecular parameters must be fixed. Following
[20]
ˆ
ground state expectation value of 1. It depends on 2z, the
the same procedure successfully applied to phenol blue,[23,24]
energy difference between the two basis states, and on
pffiffiffi
pffiffiffi
ꢀ
2t, the matrix element responsible for the mixing of the
two states.
we fix 2t = 1 eV, as a typical value. To account for the
two Raman bands at about 1340 and 1590 cmꢀ1, we intro-
duce two molecular vibrations with reference frequencies w1
= 1351 and w2 = 1597 cmꢀ1, and small polaron binding en-
ergies e1 = 0.09 and e2 = 0.05 eV. Finally, the energy gap
between the zwitterionic and the neutral states is fixed as 2z
= 2 eV. With these molecular parameters fixed (see Table 1),
the evolution of electronic and Raman spectra with the sol-
vent polarity is modeled by varying just a single parameter,
eor, the solvent relaxation energy, which vanishes for nonpo-
lar solvents and increases with the solvent polarity (eor = 0,
0.57, 0.95 eV for CCl4, CHCl3 and DMSO, respectively).
Visible spectra in Figure 5 are calculated along the same
lines as those discussed in reference [15], and by adopting
the same intrinsic electronic width, G = 2900 cmꢀ1, for all
spectra. The spectra nicely reproduce the evolution of the
absorption and emission frequencies as well as the intensi-
ties with the solvent polarity. This is a nontrivial result be-
cause all molecular parameters are kept fixed, and the sol-
vent effects are accounted for by tuning eor. In particular,
the experimental oscillator strength smoothly increases with
increasing solvent polarity. This behavior is reproduced
within our model and is understood in terms of a slight in-
crease of the molecular polarity (from 1 = 0.15 to 0.20
from CCl4 to DMSO).[25] The quantitative comparison of ab-
solute intensities fixes the dipole moment of jD+Aꢀi to
m0 ꢁ20 D.
To describe the optical spectra of D–p-A chromophores
in solution, the model must be extended to account for the
coupling of electrons with molecular vibrations and solva-
tion degrees of freedom. As for electron-vibration coupling,
we introduce a set of coupled coordinates, Qi. The two basis
states are assigned two harmonic potential energy surfaces
with exactly the same frequencies, wi, but with displaced
minima to induce a linear dependence of 2z on the Qi
values. The strength of the coupling of the i-th coordinate is
measured by the relevant small polaron binding energy, ei,
corresponding to the relaxation energy of the jD+Aꢀi state
along the coordinate.[21,22,24]
Solvation effects are twofold. A polar solute polarizes the
electronic clouds of the surrounding solvent molecules and
an electric field is generated at the solute site that screens
the molecular dipole moment.[21,15,17] The electronic degrees
of freedom of the solvent (typically in the ultraviolet region)
are faster that the electronic degrees of freedom responsible
for the chromophore absorption spectrum in the visible.
Therefore, the electronic component of the solvation reac-
tion field does not need to be treated explicitly because it
enters the model by a renormalization of the model parame-
ters.[21] Strictly speaking, different molecular parameters
then apply to the same chromophore in different solvents.
However, the electronic polarization is governed by the re-
6056
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6053 – 6063

Azobenzene Dimers
FULL PAPER
tivity with the sodium D line were hindered by the strong
absorption of the azoaromatic chromophore at that wave-
length. However, previous data concerning the optical
purity of the 4’-unsubstituted pivaloyl or methacryloyl deriv-
atives prepared through a similar synthetic pathway[10,28,29]
suggest an enantiomeric excess larger than 95% in those
compounds, thus excluding the racemization of the pyrroli-
dine chiral ring in the synthetic process. We therefore attrib-
ute a similar optical purity to the (S)-HAP-N residue linked
to the side chain of D. The H NMR spectra, reported in
Figure 6, provide information on the amount of meso and
racemic stereoisomers of the 2,4-dimethylglutarate residue.
The methylene protons of the glutaric moiety are magnet-
ically equivalent in the racemic isomer and produce a well-
defined double triplet centered at d = 1.75 ppm. In the case
of the meso isomer, these protons are nonequivalent and
appear in the NMR spectrum as two double doublets of
doublets centered at d = 2.10 and 1.45 ppm. From the inte-
grated intensities of these signals, we estimate the molar
amount of the meso form of the 2,4-dimethyglutarate resi-
due in the sample to be 32%. Analogous values are ob-
tained by analyzing the resonances of the methyl groups in
the 13C NMR spectra; the resulting stereoisomeric composi-
tion of D is well determined (ratio SRSS/(SSSS + SRRS) 1/
2.1). The stereoisomeric composition of D is by chance simi-
lar to that usually obtained for polymethacrylate derivatives
prepared by radical polymerization (70% of syndiotactic
dyads of the stereogenic centers in the main chain).[10,29,30] In
fact, the SRSS stereoisomer is the structural model for the
isotactic dyad and the SSSS and SRRS stereoisomers are
models for the syndiotactic dyad of poly[(S)-MAP-N]. The
content of syndiotactic dyads in this polymer amounts to
about 72%.[10] Therefore, the study of D in its native stereoi-
someric composition offers important information on the
behavior of the relevant polymer. The synthesis and separa-
tion of enantiomerically pure dimers and oligomers is diffi-
cult; however, we are currently working on this with the aim
of collecting more detailed information on stereochemical
interactions in azobenzene polymers.
Figure 5. Calculated a) absorption spectra of D (heavy lines) and M
(light lines), and b) circular dichroism spectra of D in different solvents.
The same parameters used for electronic spectra were ap-
plied to the calculation of Raman spectra shown in the
right-hand panels of Figure 2, Figure 3, and Figure 4 (other
results are reported in the Supporting Information). Spectra
are calculated as described in the literature,[22–24] and by
fixing the intrinsic vibrational linewidth as g = 10 cmꢀ1. The
model reproduces the red shift of the Raman bands of the
two coupled modes with increasing solvent polarity, as well
as the variation of the relative Raman intensities. This is
particularly evident for data in Figure 4, which shows
Raman spectra collected in DMSO with different excitation
lines from preresonant (red excitation) to fully resonant
(blue excitation). Figure 2 and Figure 3 show instead spectra
collected with a single excitation line (red and blue, respec-
tively), in different solvents. The variations of the intensities
are clearly related to the electronic solvatochromism, so
that, in both cases, spectra collected in DMSO are more res-
onant with the chosen excitation line than those in CCl4.[26]
Overall, the agreement between the experimental and the
calculated spectra for M is very good, giving us confidence
in the adopted model and on the estimated molecular pa-
rameters.
Electronic absorption and circular dichroism (CD) spectra
of D in different solvents (see the Experimental Section)
are shown in Figure 1. Dimerization barely affects the elec-
tronic absorption spectra, except for a small decrease in the
intensity. The negligible excitonic effects in absorption spec-
tra of D point to weak interchromophore interactions. The
sizeable CD signal observed in the visible region instead
demonstrates a finite chiral interaction between the chromo-
phores. Specifically, when normalized to the concentration
The dimer: synthesis and characterization: The synthesis of
D (Scheme 2) starts from commercial dimethyl 2,4-dimeth-
yl-glutarate (ratio meso/dl 1/0.7), which is converted to 2,4-
dimethyl-glutaric acid[27] (ratio meso/dl 1/0.7), and succes-
sively esterified with the azoic alcohol (S)-(ꢀ)-3-hydroxy-1-
(4’-nitro-4-azobenzene)pyrrolidine, (S)-HAP-N, prepared as
previously reported.[10] The chemical structure was con-
of the chromophoric units, the CD signal measured for D is
1
ꢁ = of the signal measured for the corresponding poly-
3
mer,[10] suggesting that the CD signal exhibited by the poly-
mer may be attributable to the presence of fairly short chain
segments with a well-defined skewness, and does not neces-
sarily imply long-range chiral order.
1
firmed by H NMR, 13C NMR, and FT-IR (see the Experi-
The dimer: interchromophore interactions from electronic
mental Section). Accurate measurements of the optical ac-
and CD spectra: To describe D, we assume that each chro-
Chem. Eur. J. 2005, 11, 6053 – 6063
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6057

A. Painelli et al.
This Hamiltonian must be ex-
tended to account for electron-
vibration coupling and for sol-
vation effects. For D, we only
discuss electronic (absorption
and CD) spectra so that we can
adopt a simplified description
of electron–vibration coupling.
To describe the Raman spectra
of M in the section on the sol-
vated chromophore above, we
accounted for two coupled
modes per chromophore (the
two modes with frequency w1
and w2 in Table 1). However,
such a detailed description of
the vibrational degrees of free-
dom is not actually required to
model the electronic spectra.
Electronic spectra of polar
chromophores in solution are
fairly broad and structureless,
so that accounting for just one
effective coupled vibration is
most often enough.[15,17,23] In
particular, the electronic ab-
sorption spectra of
M
in
Figure 5, calculated by account-
ing for both coupled vibrations,
is superimposed, at the scale of
the figure, with the spectrum
calculated by accounting for
just the most strongly coupled
mode (that with frequency w1
in Table 1). Therefore, in the
Figure 6. 1H NMR spectra of the azodye D in CDCl3. The resonances marked with an asterisk belong to the
solvent.
mophore is surrounded by its own solvation sphere. This ap-
proximation is justified by the fairly large interchromophore
distance and is supported by the strong similarity of absorp-
tion spectra measured for M and D (Figure 1). Within this
approximation, the model parameters extracted from the
study of M can be transferred to the analysis of D. We
therefore describe the electronic structure of D based on a
model recently proposed to describe interacting D-p-A
chromophores.[31,32] The relevant Hamiltonian is given by
Equation (2), where the first term describes the two chro-
mophores based on the Hamiltonian in Equation (1), with
the index i running on the two chromophores.
Table 1. Molecular parameters fixed for M and the three eor correspond-
ing to the three solvents of interest. Units: eV.
eor
pffiffiffiffi
1.0
z
2t
w1
e1
w2
e2
CCl4
0
CHCl3
0.57
DMSO
0.95
1.0
0.17
0.09
0.20
0.05
following, we adopt a simplified description for the elec-
tron–vibration coupling by only accounting for a single cou-
pled vibration per chromophore. The relevant vibrational
Hamiltonian is then represented by Equation (3), where i
runs on the two chromophoric sites, and Q1,i and P1,i are the
coordinate and conjugated momentum that describe the
coupled harmonic oscillator relevant to the i-th chromo-
phore. Finally, w1 and e1 are the frequency of the oscillator
and the corresponding small polaron binding energy, respec-
tively: they are the same for both chromophores, and their
values are taken from Table 1.
X
pffiffiffi
^
^
^ ^
Hel
¼
ð2z1iꢀ 2tsx,iÞ þ V1112
ð2Þ
i¼1,2
The last term accounts for interchromophore interactions,
with V measuring the electrostatic interaction between two
zwitterionic, jD+Aꢀi, chromophores.
X
pffiffiffiffiffiffi
1
2
2
2
ðw21Q1,iꢀP1,iÞꢀ 2e1w1Q1,i1i
^
Hvib
¼
ð3Þ
i¼1,2
6058
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6053 – 6063

Azobenzene Dimers
FULL PAPER
Similarly, we model polar solvation by introducing two ef-
fective solvation coordinates Q0,i, with i running on the two
chromophoric sites. The solvation relaxation energy, eor, rel-
evant to each solvent is taken from Table 1, which results
from the spectroscopic analysis of M. The two solvation co-
ordinates enter as classical coordinates, the relevant Hamil-
tonian is given by Equation (4), where the frequency w0 as-
signed to the effective solvation coordinate is irrelevant.[21]
The CD spectra are calculated with Equation (6),[35,36]
3
X
~
2
80 L
3 ln 10ꢀhc2e0
n Gn0
2
DeðnÞ½L molꢀ1 cmꢀ1Š ¼
Rn0
2
~
2
2
~
~
~
ðnn0ꢀn Þ þ n Gn0
n
ð6Þ
where Rn0 is defined according to Equation (7):
!
ðEnꢀE0Þ
ð7Þ
^
^
Rn0
¼
R ꢂ ðhnjm1j0i ꢃ h0jm2jniÞ
2ꢀh
X
pffiffiffiffiffiffiffiffi
1
2
!
Hsolv
¼
w20Q20,iꢀ 2eorw0Q0,i1i
ð4Þ
R is the vector that joins the two chromophores. Both
absorption and CD spectra calculated for each Q0,1, Q0,2 are
weighted according to a Boltzmann distribution and are
then summed to calculate the total spectra that fully account
for electron–vibration coupling and for solvation effects, in-
cluding inhomogeneous broadening from polar solvation.
The geometry of D, and more precisely the relative orien-
tation of the chromophores, enters the calculation of the ab-
sorption spectra in the definition of the total dipole moment
i¼1,2
The total Hamiltonian H = Hel + Hvib + Hsolv can be di-
agonalized exactly for fixed Q0,1, Q0,2, to span the distribu-
tion of the reaction fields surrounding the two chromo-
phores.
Whereas solvation coordinates are treated as classical var-
iables, vibrational coordinates must be treated quantum me-
chanically. For fixed Q0,1 and Q0,2, the total Hamiltonian de-
scribes a complex system with coupled electrons and molec-
ular vibrations. Herein, we present results obtained from the
direct (nonadiabatic) diagonalization of the Hamiltonian.
Specifically, a complete basis for the problem is obtained
from the direct product of the electronic basis multiplied by
the vibrational basis. As for the electronic basis, the two-
states per chromophore lead to four basis states: jDA,DAi,
jD+Aꢀ,DAi, jDA,D+Aꢀi, jD+Aꢀ,D+Aꢀi. As for the vibra-
tional states, an oscillator must be considered for each chro-
mophore, and an infinite basis set is associated with each os-
cillator, composed of states with 0, 1, ···, N vibrational
quanta. The matrix elements of the total Hamiltonian for
fixed Q0,1, Q0,2, can be easily written on the complete basis
and a numerically exact diagonalization of the problem can
be obtained by truncating the vibrational basis to a finite N,
such that a further increase in N does not affect the results.
The dimension of the resulting basis set, 4N2, increases rap-
idly with N. In the present case of weak coupling, N = 4 is
typically sufficient for the calculation of the optical spectra.
The nonadiabatic diagonalization of the Hamiltonian in
Equation (4) goes along the lines already discussed for a
single chromophoric unit.[33,34] The resulting eigenstates, jni,
with energies En, enter the expression for the absorption
spectra according to Equation (5),[15] where the sum runs
over all the excited states, and j0i represents the ground
state.
ˆ
operator, m, the vectorial sum of the dipole moments of
each chromophore. Both the orientation and the distance
between the chromophores enter the calculation of the CD
spectra, through the Rn0 factor in Equation (7). On the
other hand, the geometry of D implicitly enters the Hamil-
tonian through V, in Equation (1), which measures the
energy of the electrostatic interactions between the two
chromophores in the zwitterionic state. To relate V to the
dimer geometry, we model each jD+Aꢀi as a rigid segment
of length l, with unit positive and negative charges at the
two extremes. The two chromophores are attached perpen-
dicularly to the oligomeric chain, modeled as a rigid seg-
ment of length R. Based on typical bond lengths, we fix R =
4 Š and l = 10 Š, so that V is fully defined by q, the dihe-
dral angle between the chromophores (Figure 7). The explic-
it expression for V is discussed in the Experimental Section.
Figure 8 shows the evolution with the dihedral angle, q, of
the absorption and CD spectra calculated for D in CCl4.
Figure 7. Schematic representation of the considered dimer. Polar mole-
cules are represented by arrows; l is the dipole length, R the interchro-
mophore distance, and q is the angle formed by the two dipoles.
eðnÞ½L molꢀ1 cmꢀ1
Š
~
ꢀ
ꢁ
ꢂ ꢃ
2
X
ð5Þ
~
~
10pL
3 ln 10ꢀhce0
1
1
2
nn0ꢀn
2
pffiffiffiffiffi
2p
~
^
¼
nn0jh0jmjnij exp ꢀ
s
Only angles between 0 and 1808 are shown. Spectra for
ꢀ180<q<08 can be obtained easily: upon changing the sign
of q, the absorption spectra are invariant, whereas CD spec-
tra change their sign. The results in Figure 8 were obtained
from the complete diagonalization of the electronic and vi-
brational Hamiltonian; however, the basic physics underly-
ing the spectroscopic behavior of D can be understood
n
L is the Avogadro number, c the speed of light, e0 the
vacuum permittivity (SI units), and n˜ is the wavenumber in
cmꢀ1. The dipole moment operator is defined as the vectori-
al sum of the dipole moments relevant to the isolated chro-
[32]
ˆ
ˆ
ˆ
mophores: m = m1 + m2
.
Chem. Eur. J. 2005, 11, 6053 – 6063
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6059

A. Painelli et al.
band. Our model for D in CCl4 gives an integrated g factor
increasing from 0 at q = 0 to qꢁ154”10ꢀ5 at q = 908 and
then decreasing again to vanish at q = 1808. The experi-
mental g value for D in CCl4, gꢁ6.0”10ꢀ5, then imposes q
close to 0 or 1808. Based on the similarity between absorp-
tion spectra of D and M, we safely locate q close to 1808 in
CCl4. Specifically, adopting the model parameters for the
chromophore as obtained from the spectroscopic analysis of
M above, and with the specific choice of the dimer geometry
discussed above, the best fit of the CD spectrum in CCl4
fixes qꢁ1788. The resulting absorption and CD spectra are
shown in Figure 5. The same or very similar angles lead to
the best fits of absorption and CD spectra in CHCl3 and
DMSO (Figure 5) with the integrated g factor ꢁ4.2”10ꢀ5
and ꢁ6.5”10ꢀ5, respectively.
Figure 8. Absorption (c) and CD (a) spectra calculated for D in
CCl4, for q values shown in each panel. The panel marked with M shows
the spectra calculated for the isolated M. Absorption and CD spectra are
reported in arbitrary units, but the corresponding units are the same in
all panels.
A preliminary discussion of chiral interchromophore inter-
actions in the polymer side chains: The success of the pro-
posed model in the description of absorption and CD spec-
tra of D invites us to extend the discussion to the relevant
polymeric material. A detailed description of chiral inter-
chromophore interactions in the polymer requires a system-
atic spectroscopic analysis of trimeric and oligomeric units.
Oligomers are challenging in view of both their chemical
synthesis and theoretical analysis. However, some prelimina-
ry discussion of interchromophore interactions in the macro-
molecular chain is presented based on information collected
for D. A crude model for an oligomer is obtained by extend-
ing the dimer model to NM sites (i.e. considering a perfectly
ordered chiral structure made up of NM chromophores): The
i index in Equation (2) then runs up to NM and electrostatic
interaction, Vij, among all sites are accounted for. Solving
the complete electronic and vibrational problem on the
based on
a simple excitonic (or coupled oscillator)
model.[36,37,38] The M!M* excitation responsible for the visi-
ble absorption in M (topmost left panel) corresponds to two
excitations in the dimer because each one of the two chro-
mophores can be excited. The two chromophores are equiv-
alent, and it is convenient to combine the two singly excited
states as follows: jEꢄi = (jMM*iꢄjM*Mi), where jMMi
represents the dimeric unit and the asterisk marks the excit-
ed chromophore within the dimer. Interchromophore inter-
actions split the two states by 2 V1(1ꢀ1), and, because inter-
chromophore interactions are always repulsive for the
adopted geometry, jEꢀi corresponds to the lowest excita-
(2N)N basis is challenging for NM >2; however, the solution
M
tion.[37,38] For aligned molecules (q
=
0), all oscillator
of the electronic Hamiltonian on the 2N basis is trivial up
M
strength resides on jE+i, and the observed spectrum is
largely blue-shifted with respect to M. For antiparallel orien-
tation (q = 1808), only Eꢀ is active in absorption: indeed
the red-shift with respect to M is negligible in this case since
the effects of intermolecular interactions are smaller for this
orientation. Of course, no CD signal is observed for aligned
molecules (either parallel or antiparallel) owing to the lack
of chirality. For nonaligned molecules, both E+ and Eꢀ have
a finite intensity for absorption: for q = ꢄ908 the two tran-
sitions have the same intensity, and the resulting absorption
spectrum is broad. For ꢀ90<q<908 (90<q<2708) E+ (Eꢀ)
dominates the absorption spectrum: asymmetric absorption
spectra are calculated for q = 308 and 1508.
A common measure of the strength of the CD spectrum is
given in terms of the so-called chiral anisotropy coefficient,
g, defined as the ratio between the CD signal and the ab-
sorbance measured for the same sample at the same fre-
quency (usually the maximum of the CD spectrum).[41] To
avoid ambiguities and uncertainties caused by the choice of
a specific frequency, we define the integrated g factor to be
the ratio of the area underlying the absolute value of the
CD signal divided by the area underlying the absorption
to fairly large NM. The resulting CD spectra cannot give any
precise description of experimental spectra, and particularly
of band shapes, but they do offer qualitative information on
the NM dependence of the CD signal. Just as an example, if
we construct a trimer in DMSO with the same dihedral
angle between adjacent chromophores as for the dimer
(1778) we get a CD response of approximately the same
magnitude as for the dimer but with the opposite sign. This
result can be easily rationalized: the CD spectra of NM inter-
acting chromophores roughly corresponds to the sum of the
CD spectra resulting from all interactions between pairs of
chromophores. In a trimer, we then must sum up two equal
contributions from the dimeric interaction between nearest
neighbors plus the next-nearest neighbor interaction be-
tween two chromophores located 8 Š apart and with a dihe-
dral angle of ꢁ3548. This last interaction leads to a CD
signal that is larger than the sum of the previous two, and
has the opposite sign, thus justifying the inversion of the cal-
culated CD signal when going from a dimeric to a trimeric
unit. Further increases in NM give larger and larger CD re-
sponses, but still leads to a CD signal with the wrong sign,
with respect to the experimental results. Therefore, we must
6060
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6053 – 6063

Azobenzene Dimers
FULL PAPER
conclude that the dihedral angle between adjacent chromo-
phores changes on going from D to the polymeric system.
Experimental spectra for the macromolecular chain can be
simulated quite nicely for qꢁ1508. This estimate, based on
an oversimplified description of the polymer, is very rough
and represents a preliminary result to be confirmed by fur-
ther experimental and theoretical studies. However, we un-
derline that a decrease of the dihedral angle when going
from a dimeric to an oligomeric unit is expected, based on
simple electrostatic arguments, which suggest reduced
angles in order to release electrostatic repulsion between
non nearest-neighbor chromophores. Overall, within our ap-
proach, the experimental negative Cotton effect observed in
the CD spectra of D is related to a right-handed chiral con-
formation of the two chromophores. The observation of a
Cotton effect with the same sign for the polymer can be sim-
ilarly explained based on a right-handed conformational
structure, such as that relevant to a right-hand helix.
spectra of D can be understood in terms of classical electro-
static interactions, based on a realistic description of the D
geometry.
The CD signal in our model is attributed to the intrinsical-
ly chiral interaction between the two chromophoric units for
any dihedral angle other than 0 or 1808. The presence of a
chiral C center in the side chain is instrumental in favoring
the chiral orientation of the chromophores, but, by itself, it
is irrelevant to the spectroscopic behavior in the visible
region. In this respect, our model shares the same physics
with the standard excitonic picture that is usually adopted to
describe CD spectra of interacting chromophores.[36,39,40]
However, a major difference must be noted. In fact, the
electrostatic interchromophore interactions are modeled in
the electric dipole approximation in the standard picture.
This approximation, strictly valid only in the long-distance
regime (l !R), describes each chromophore as an electric
dipole: it corresponds to repulsive/attractive interactions for
parallel/antiparallel molecules, and predicts vanishing inter-
actions for perpendicularly oriented chromophores. In our
more realistic picture, where the two molecules are anch-
ored through their D site to the rigid polymeric backbone
(Figure 7) interchromophore interactions are always finite
and repulsive, having their maximum (minimum) value for
parallel (antiparallel) molecules. This has, of course, impor-
tant spectroscopic consequences that are most evident in
CD spectra: in our geometry, the CD spectrum only vanish-
es for parallel or antiparallel chromophores, whereas in the
standard model the CD signal vanishes for parallel, antipar-
allel, and perpendicular chromophores. The vanishing of the
CD signal for parallel (or antiparallel) chromophores is triv-
ially related to the lack of chirality in both models. Struc-
tures with perpendicular chromophores stay chiral unless
the two chromophores are anchored in the middle, which is
always the case for point-dipole molecules. Indeed, in the
dipole approximation, perpendicular chromophores do not
interact and no CD signal is expected in any case.
Our model for CD spectra fully accounts for the coupling
of electronic degrees of freedom with vibrational and solva-
tion degrees of freedom. This allows us to discuss CD and
absorption spectra on the same footing and gives informa-
tion not only on the amplitude of the CD signal, but also on
the band-shape of the CD spectra. In particular, coupling to
slow degrees of freedom (including internal vibrations and
polar solvation coordinates) is responsible for the observa-
tion of broad absorption and CD spectra. In this respect, we
underline that the amount of the exciton splitting (E+ꢀEꢀ)
cannot be directly extracted from the energy difference be-
tween the positive and negative CD peaks in the case of broad
spectra. CD spectra are, in fact, the difference of spectra rel-
evant to the E+ and Eꢀ absorption. In the case of broad
spectra, that is, when E+ꢀEꢀ is smaller than the bandwidth,
the apparent exciton splitting, obtained from the frequency
difference between the positive and negative CD peaks,
measures the amount of broadening and not the true split-
ting. A reliable estimate of the exciton splitting can only be
obtained via a detailed fit of both absorption and CD spectra.
Conclusions
We have presented a combined theoretical and experimental
investigation of interchromophore interactions in azoben-
zene derivatives. The analysis starts from the quantitative
description of the chromophoric unit, based on an extensive
spectroscopic study of the chromophore in solution. The di-
meric unit was then synthesized as an interesting model
system to investigate interchromophore interactions in azo-
polymers. The analysis of CD spectra of D in several sol-
vents proved particularly useful in this respect.
The model proposed for optical spectra of D is very
simple and only accounts for electrostatic interchromophore
interactions. Moreover, the adopted geometry is somewhat
arbitrary, leading to an oversimplified expression for V (see
the Experimental Section). In spite of these limitations, the
model nicely reproduces experimental absorption and CD
spectra. In particular, in both CHCl3 and DMSO, in agree-
ment with experimental data, we calculated somewhat less
intense absorption spectra for D than for M, as a conse-
quence of 1) the decrease of the chromophore polarity
caused by interchromophore interactions, 2) the increased
broadening in D. The effect is much reduced in CCl4, lead-
ing to some discrepancy with respect to the experimental
data. This minor discrepancy could be cured by assigning a
small eor to this nonpolar solvent;[25] however, we do not be-
lieve the accuracy of the model is high enough to justify the
effort. In any case, our aim is not the detailed fit of the ex-
perimental features, nor a precise estimate of microscopic
parameters: indeed, the estimated dihedral angles depend
on the specific choice of l and R. Instead, our results demon-
strate that, starting with an extensive spectroscopic analysis
of the solvated M chromophore, enough information can be
obtained to construct a model for the dimeric unit, D. This
bottom-up analysis of molecular materials is a first step in
the understanding of interchromophore interactions. Specifi-
cally, we demonstrate here that both electronic and CD
Chem. Eur. J. 2005, 11, 6053 – 6063
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6061

A. Painelli et al.
of Equation (9), where e0 is the vacuum dielectric permittivity, and n2,
the squared refractive index of the solvent, accounts for the dielectric
screening of electrostatic interactions at optical frequencies.
Experimental Section
Synthesis of model compound M: This was synthesized as previously re-
ported.[10]
ꢄ
ꢅ
e2
4pe0n2
1
R
1
2
Synthesis of D: An excess of KOH in water (4.8 g, 5 mL) was added to a
solution of dimethyl 2,4-dimethylglutarate (Aldrich, ratio meso/dl 1/0.7)
in ethanol (7.41 mmol, 9.5 mL), and the mixture was kept under reflux
for 4 h. The solvent was eliminated under reduced pressure and the re-
sulting white solid was dissolved in water. The solution was acidified
(pH 1) with concentrated aqueous HCl, and the precipitated material
was filtered, dissolved in diethyl ether, and dried with Na2SO4. The aque-
ous solution was repeatedly washed with ether, and the organic phases
were collected and dried. The solvent was evaporated, and the pure prod-
uct (ratio meso/dl 1/0.7) was immediately used for the following reaction.
The diacidic derivative (0.622 mmol) was dissolved in dry CH2Cl2 under a
nitrogen atmosphere with the azoic alcohol (S)-(ꢀ)-3-hydroxy-1-(4’-nitro-
4-azobenzene)-pyrrolidine [(S)-HAP-N, 2.47 mmol], prepared as previ-
ously reported.[10] 4-(Dimethylamino)pyridinium 4-toluenesulfonate
(DPTS, 1.24 mmol) and 1,3-diisopropylcarbodiimide (DIPC, 1.60 mmol)
were added to the solution, and the mixture was stirred for five days at
room temperature. The product was purified by column chromatography
(SiO2, CHCl3) followed by crystallization in hot toluene (yield 29%).
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
V
¼
þ
ꢀ
ð9Þ
l2 þ R2
R2 þ 4 l2sin2ðq=2Þ
Modeling the screening of electrostatic interactions caused by the solvent
is nontrivial, and deserves some discussion. Interchromophore interac-
tions and their screening in solution haven been extensively covered in
recent literature; however, most often with reference to centrosymmetric
chromophores or to molecules with no permanent dipole moment.[42] In
this case, the only relevant electrostatic interactions involve transition
dipole moments and should therefore be screened by the dielectric con-
stant at optical frequencies, that is, by the squared refractive index. The
problem is more complex in the case of polar chromophores: electrostat-
ic interactions in the ground state involve static charge distributions and
should be screened by the static dielectric constant, whereas transient in-
teractions involving either transition or excited state dipole moments
should be screened by the squared refractive index. The electrostatic in-
teraction V in Equation (1) enters into the definition of the ground-state
chromophore polarity,[32] and in this respect it should account for the
static dielectric screening. But the very same quantity also enters the def-
inition of the exciton splitting[32] where the dielectric constant at optical
frequencies should play a role. Certainly, there is no ambiguity with re-
spect to in nonpolar solvents, where the static dielectric constant is virtu-
ally equal to the squared refractive index. In contrast, the difference is
large in strongly polar solvents. For example, in DMSO the static dielec-
tric screening is about 10 times larger than the screening at optical fre-
quencies. To the best of our knowledge, a general solution to this inter-
esting problem is lacking. For the specific system we are discussing in this
paper, we take advantage of the comparatively small value of the
ground-state dipole moment of M and then adopt the same screening
model for interchromophore interactions as is usually adopted for nonpo-
lar chromophores, or, in other words, we screen V in Equation (1) by the
squared refractive index of the solvent. We can appreciate the quality of
this approximation by comparing the ground-state polarity calculated for
the chromophores in D in the hypothesis of optical or static dielectric
screening for interchromophore interactions. In CCl4, the two screenings
are virtually identical and lead to the same ground-state polarity 1ꢁ0.16.
In CHCl3, static screening leads to 1ꢁ0.17, to be compared with a value
of 0.16 obtained in the adopted approximation. By the way, the largest
difference was found in DMSO, with 1 = 0.20 and 0.18 for static and op-
tical screening, respectively. In any case, the two results differ by no
more than 10%, well within the uncertainties of the proposed model.
M.p. 209–2118C; FT-IR: n˜ = 3068 (nCH, arom.), 2982 and 2949 (nCH
aliph.), 1733 (nCO, ester), 1605 and 1516 (nC=C, arom.), 1140 (nCꢀO), 861
and 823 (dCH 1,4-disubst. arom. ring) cmꢀ1 1H NMR: d = 8.35 (dd, 4H,
,
;
arom 3’-H), 7.90 (m, 8H, arom metha to amino group and 2’-H), 6.60
(dd, 4H, arom ortho to amino group), 5.45 (m, 2H, 3-CH), 3.80–3.35 (m,
8H, 2- and 5-CH2), 2.45 (m, 2H, backbone CH), 2.25 (m, 4H, 4-CH2),
2.10 and 1.45 (2ddd, 2H, backbone CH2 meso form), 1.75 (2t, 2H, back-
bone CH2 racemic form), 1.15 (d, 6H, CH3 meso form), 1.10 ppm (d, 6H,
CH3 racemic form); 13C NMR: d = 176.4 (CO), 156.5 (arom C-NO2),
151.1, 148.1, 144.7 (arom C-N=N-C and C-NCH2), 126.9, 125.4, 123.3
(arom 3’-C, 2’-C and 3-C), 112.4 (arom 2-C), 74.1 (CH-O), 54.3 (CH-
CH2-N), 46.5 (CH2-CH2-N), 38.0 (main chain CH2-CH), 31.8 (CH2-CH2-
N), 18.4 (CH3 racemic form), 18.0 ppm (CH3 meso form).
General procedures: 1H and 13C NMR spectra were obtained at room
temperature in 5–10% CDCl3 solutions with a Varian NMR Gemini300
spectrometer. Chemical shifts are given with respect to tetramethylsilane
(TMS) as the internal reference. 1H NMR spectra were recorded at
300 MHz with the following experimental conditions: 24000 data points,
4.5 kHz spectral width, 2.6 s acquisition time, 64 transients. 13C NMR
spectra were recorded at 75.5 MHz, under full proton decoupling, with
the following experimental conditions: 24000 data points, 20 kHz spectral
width, 0.6 s acquisition time, 64000 transients. UV/Vis absorption spectra
of CHCl3, CCl4, and DMSO solutions were recorded at 258C in the 700–
250 nm spectral region with a Perkin-Elmer Lambda19 spectrophotome-
ter. The cell path length was 0.1 cm. Concentrations of azobenzene chro-
mophore of ꢁ3”10ꢀ4 molLꢀ1 were used. CD spectra were recorded at
258C on a Jasco810A dichrograph with the same path lengths and solu-
tion concentrations as for UV measurements. De values, expressed as
Lmolꢀ1 cmꢀ1, were calculated with Equation (8), where the molar elliptic-
ity [q] in degcm2 dmolꢀ1 refers to one azobenzene chromophore.
Acknowledgements
A.P. is grateful to Aldo Brillante for useful correspondence on CD spec-
tra. F.T. acknowledges support by a Marie Curie Intra-European Fellow-
ship within the 6th European Community Framework Programme. Finan-
cial support by MIUR (through FIRB2001 and PRIN2004) and INSTM
is gratefully acknowledged.
De ¼ ½qŠ=3300
ð8Þ
Raman spectra were collected with a Renishow System-1000 Raman mi-
croscope, equipped with a Kr laser. The spectral resolution was about
1 cmꢀ1. The following excitation lines were used: 647 nm (red), 568 nm
(yellow), 521 nm (green), 476 nm (blue). Spectra obtained with the
568 nm and the 521 nm excitation lines for the solutions in CCl4, CHCl3,
and DMSO are reported in the Supporting Information.
[1] Proceedings of the Symposium on Azobenzene-Containing Materi-
als, Boston, MA, 1998 (Ed.: A. Natansohn), Macromolecular Sym-
posia 1999, 137, 1–165.
[2] D. R. Kanis, M. A. Ratner, T. J. Marks, Chem. Rev. 1994, 94, 195.
[3] a) Z. Sekkat, W. Knoll, Photoreactive organic thin films, Amsterdam
Academic Press, Amsterdam 2002; b) O. Nuyken, C. Scherer, A. R.
Brenner, U. Dahn, R. Gärtner, S. Kaiser-Rçhrich, R. Kollefrath, P.
Matusche, B. Voit, Prog. Polym. Sci. 1997, 22, 93–183; c) S. K. Yes-
odha, C. K. Sadashiva Pilai, N. Tsutsumi, Prog. Polym. Sci. 2004, 29,
45–74.
Modeling electrostatic interactions: V represents the electrostatic interac-
tion between two zwitterionic chromophores. Each zwitterionic chromo-
phore was modeled in terms of a pair of unit positive and negative charg-
es at the two extremes of a rigid rope of length l = 10 Š. The two chro-
mophores are perpendicularly attached to the polymeric backbone at a
distance of R = 4 Š (Figure 7) and define a dihedral angle q. Based on
this simplified picture for the D geometry, V can be calculated by means
[4] J. A. Delaire, K. Nakatani, Chem. Rev. 2000, 100, 1817–1845.
6062
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6053 – 6063

Azobenzene Dimers
FULL PAPER
[5] F. Charra, F. Devaux, J.-M. Nunzi, P. Raimond, Phys. Rev. Lett.
1992, 68, 2440–2443.
[6] a) P. Rochon, E. Batalla, A. Natansohn, Appl. Phys. Lett. 1995, 66,
136; b) D.-Y. Kim, L. Li, J. Kumar, S. K. Tripathy, Appl. Phys. Lett.
1995, 66, 1166.
[26] We underline that the strength of the coupling between electronic
and vibrational degrees of freedom is much lower for M than for
phenol blue. At the same time, M is less polar, and hence less polar-
izable than phenol blue. Both facts concur to reduce the effects of
coupling in vibrational spectra. In particular, the inhomogeneous
broadening that strongly affects Raman spectra of phenol blue[23,24]
is barely appreciable in the present spectra.
[7] M.-J. Kim, B.-G. Shin, J.-J. Kim, D.-Y. Kim, J. Am. Chem. Soc. 2002,
124, 3504–3505.
[8] G. Iftime, F. L. Labarthet, A. Natansohn, P. Rochon, J. Am. Chem.
Soc. 2000, 122, 12646–12650.
[27] H. Achenbach, W. Karl, Chem. Ber. 1975, 108, 759–771.
[28] L. Angiolini, D. Caretti, C. Carlini, L. Giorgini, E. Salatelli, Macro-
mol. Chem. Phys. 1999 200, 390.
[29] a) L. Angiolini, D. Caretti, L. Giorgini, E. Salatelli, A. Altomare, C.
Carlini, R. Solaro, Polymer 1998 39, 6621; b) L. Angiolini, D. Care-
tti, L. Giorgini, E. Salatelli, A. Altomare, C. Carlini, R. Solaro, Poly-
mer 2000, 41, 4767.
[30] a) I. R. Peat, W. F. Reynolds, Tetrahedron Lett. 1972, 13, 1359;
b) E. F. McCord, W. L. Anton, L. Wilczek, S. D. Ittel, L. T. J. Nelson,
K. D. Raffael, Macromol. Symp. 1994, 86, 47; c) L. Angiolini, D.
Caretti, L. Giorgini, E. Salatelli, Polymer, 2001 42, 4005; d) L. An-
giolini, D. Caretti, L. Giorgini, E. Salatelli, e-Polym. e-Polymers
2001, 21.
[31] A. Painelli, F. Terenziani, J. Am. Chem. Soc. 2003, 125 5624–5625.
[32] F. Terenziani, A. Painelli, Phys. Rev. B 2003, 68, 165405 (1–13).
[33] L. Del Freo, A. Painelli, Chem. Phys. Lett. 2001, 338, 208.
[34] L. Del Freo, F. Terenziani, A. Painelli, J. Chem. Phys. 2002, 116,
755–761.
[9] a) L. Nikolova, T. Todorov, M. Ivanov, F. Andruzzi, S. Hvilsted, P. S.
Ramanujam, Opt. Mater. 1997, 8, 255–258; b) I. Nayadenova, L. Ni-
kolova, P. S. Ramanujam, S. Hvilsted, J. Opt. A 1999, 1, 438–441.
[10] a) L. Angiolini, D. Caretti, L. Giorgini, E. Salatelli, J. Polym. Sci.
Part A: Polym. Chem. 1999, 37, 3257; b) L. Angiolini, L. Giorgini,
E. Salatelli, e-Polymers 2003, 37.
[11] L. Angiolini, R. Bozio, L. Giorgini, D. Pedron, G. Turco, A. Daurffl,
Chem. Eur. J. 2002, 8, 4241–4247.
[12] a) L. Angiolini, T. Benelli, R. Bozio, A. Daurffl, L. Giorgini, D.
Pedron, Synth. Met. 2003, 139, 743–746; b) L. Angiolini, R. Bozio,
L. Giorgini, D. Pedron, Synth. Met. 2003, 138, 375–378.
[13] C. Reichardt, Chem. Rev. 1994, 94, 2319.
[14] M. L. Horng, J. A. Gardecki, A. Papazyan, M. Maroncelli, J. Phys.
Chem. 1995, 99, 17311.
[15] B. Boldrini, E. Cavalli, A. Painelli, F. Terenziani, J. Phys. Chem. A
2002, 106 , 6286–6294; and references therein.
[16] D. Laage, W. H. Thompson, M. Blanchard-Desce, J. T. Hynes, J.
Phys. Chem. A 2003, 107, 6032–6046.
[17] F. Terenziani, A. Painelli, A. Girlando, R. M. Metzger, J. Phys.
Chem. B 2004, 108, 10743–10750.
[18] A. Painelli, F. Terenziani, Chem. Phys. Lett. 1999, 312, 211.
[19] R. S. Mulliken, J. Am. Chem. Soc. 1952, 74, 811.
[20] A. Painelli, Chem. Phys. Lett. 1998, 285, 352.
[35] E. U. Condon, Rev. Mod. Phys. 1937, 9, 432–456.
[36] D. P. Craig, T. Thirunamachandran, Molecular Quantum Electrody-
namics, Academic press, London 1984, Chapt. 8.
[37] V. M. Agranovich and M. D. Galanin, Electronic Excitation Energy
Transfer in Condensed Matter North-Holland, Amsterdam, 1982.
[38] J. Knoester, in Organic Nanostructures: Science and Applications
(Eds.: V. M. Agranovich, G. C. La Rocca), IOS Press, Amsterdam
2002, p. 149, and references therein.
[21] A. Painelli, Chem. Phys. 1999, 245, 185–197.
[22] A. Painelli, F. Terenziani, J. Phys. Chem. A 2000, 104, 11041.
[23] F. Terenziani, A. Painelli, D. Comoretto, J. Phys. Chem. A 2000, 104,
11049.
[39] A. Rodger, B. NordØn, Circular Dichroism and Linear Dichroism,
Oxford University Press, Oxford 1997, Chapts. 5 and 7.
[40] S. F. Mason, Molecular Optical Activity and the Chiral Discrimina-
tion, Cambridge University Press, Cambridge 1982, Chapt. 3.
[41] N. Berova, K. Nakanishi, R. W. Woody, Circular dichroism, princi-
ples and applications, Wiley-VCH Inc., New York, 2000.
[42] G. D. Scholes, Annu. Rev. Phys. Chem. 2003, 54, 87.
[24] A. Painelli, F. Terenziani, Synth. Met. 2001 116 135–138.
[25] The absorption spectrum calculated in CCl4 is somewhat narrower
than the experimental spectrum. This small discrepancy can be
easily accounted for by assigning a small eor to this solvent. Whereas
a small but finite eor in CCl4 can be justified as accounting for quad-
rupolar contributions to the orientational reaction field,[15] we feel
that this correction is not really required for the scope of the present
work.
Received: October 25, 2004
Revised: May 13, 2005
Published online: July 29, 2005
Chem. Eur. J. 2005, 11, 6053 – 6063
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6063