Tuning the nature of the fluorescent state: A substituted polycondensed dye as a case study

Article abstract of DOI:10.1002/chem.201202154

An extensive spectroscopic analysis is presented of an elongated polycondensed dye with a donor-acceptor substitution. The charge-transfer (CT) state, polarized along the long molecular axis, is close in energy to a local excitation (LE) of the polycondensed system, roughly polarized along the short molecular axis, which makes this system particularly suitable to investigate the subtle LE/CT interplay. An essential-state model is presented that quantitatively reproduces absorption and fluorescence spectra, as well as fluorescence emission and excitation anisotropy spectra collected in solvents of different polarity and viscosity, which sets a sound basis for the understanding of how solvent polarity and solvent relaxation affect the nature of low-lying excitations. The markedly different fluorescence emission and excitation anisotropy spectra measured in glassy and liquid polar solvents unambiguously demonstrate the major role played by solvent relaxation in the definition of fluorescence properties of the dye. Charge-transfer or local excitations? Fluorescence emission from a substituted azachrysene chromophore is shown to change its nature according to solvent polarity. The different polarization of the low-energy local and charge-transfer excitations (see scheme) is responsible for highly informative fluorescence anisotropy spectra. Experimental results are interpreted and reproduced based on an original three-state model. Copyright

Full text of DOI:10.1002/chem.201202154

DOI: 10.1002/chem.201202154
Tuning the Nature of the Fluorescent State: A Substituted Polycondensed
Dye as a Case Study
Cristina Sissa,[a] Valentina Calabrese,[b] Marco Cavazzini,[b] Luca Grisanti,[a]
Francesca Terenziani,[a] Silvio Quici,[b] and Anna Painelli*[a]
Abstract: An extensive spectroscopic
analysis is presented of an elongated
polycondensed dye with a donor–ac-
ceptor substitution. The charge-transfer
(CT) state, polarized along the long
molecular axis, is close in energy to a
local excitation (LE) of the polycon-
densed system, roughly polarized along
the short molecular axis, which makes
this system particularly suitable to in-
vestigate the subtle LE/CT interplay.
An essential-state model is presented
that quantitatively reproduces absorp-
tion and fluorescence spectra, as well
as fluorescence emission and excitation
anisotropy spectra collected in solvents
of different polarity and viscosity,
which sets a sound basis for the under-
standing of how solvent polarity and
solvent relaxation affect the nature of
low-lying excitations. The markedly dif-
ferent fluorescence emission and exci-
tation anisotropy spectra measured in
glassy and liquid polar solvents unam-
biguously demonstrate the major role
played by solvent relaxation in the def-
inition of fluorescence properties of
the dye.
Keywords: donor–acceptor
sys-
tems · dyes/pigments · essential-
state models · fluorescence · solva-
tochromism
Introduction
so-called push–pull chromophores,[13–21] an interesting class
of elongated p-conjugated molecules bearing an electron-
donor (D) and -acceptor (A) group at the two molecular
ends. Intramolecular CT degrees of freedom in these p-con-
jugated systems share the same physics as intermolecular
CT in molecular complexes, even if some differences are
conveniently highlighted. In particular, in D-p-A chromo-
phores the two resonating states are coupled via virtual in-
termediate states in which charges are located in the bridge:
the two-state model is therefore an effective model and, at
variance with DA complexes, the D+-p-Aꢀ dipole length
cannot be estimated from the D-to-A physical distance.[9,22]
The two-state model belongs to the more general class of es-
sential-state models,[13–21,23–31] a family of parametric models
developed to describe several families of CT chromophores.
Essential-state models are parameterized and validated
against experimental spectroscopic data, but recent attempts
to extract relevant parameters from high-quality ab initio
calculations appear very promising.[32–35]
The two-state model captures most of the physics of CT
complexes but, already in the early days of the Mulliken
model, the interaction between the CT state and other excit-
ed states, localized either on the D or the A molecule,[36,37]
had been discussed to explain some anomalous spectral fea-
tures. The distinction between CT and local excited (LE)
states is clear-cut in molecular complexes, for which LE
states can be mapped on the states of the two isolated neu-
tral molecules, whereas the CT state describes the situation
in which one electron is transferred from the D to the A
molecule. The LE versus CT distinction is somewhat blurred
in CT chromophores. However, based on a valence-bond de-
scription of the molecular electronic structure, the diabatic
Basic understanding of spectroscopic signatures of charge-
transfer (CT) states and processes dates back to the classical
work of Robert Mulliken in 1952 on molecular complexes.[1]
When two different molecules, D and A, come together in
solution, a CT complex can be stabilized by the resonance
between the neutral state, DA, and the zwitterionic state,
D+Aꢀ, in which D labels the molecule that behaves as an
electron donor with respect to the electron-accepting (A)
molecule. A simple two-state model predicts not only the
stabilization of the ground state due to resonance, but also
the appearance of a low-lying excited state, responsible for
the occurrence of a so-called CT transition in the visible or
near-infrared spectral region, depending on the ionization
potential of D, the electronic affinity of A, and the hopping
integral between frontier molecular orbitals.[1–5]
The same Mulliken Hamiltonian has been successfully
adopted to describe absorption spectra of mixed-valence
systems,[6–12] and more recently to model optical spectra of
[a] Dr. C. Sissa, Dr. L. Grisanti, Prof. F. Terenziani, Prof. A. Painelli
Dipartimento di Chimica
Universitꢀ di Parma and INSTM UdR-Parma
Parco Area delle Scienze 17A, 43124 Parma (Italy)
Fax : (+39)0521-905556
E-mail: anna.painelli@unipr.it
[b] Dr. V. Calabrese, Dr. M. Cavazzini, Dr. S. Quici
Istituto di Scienze e Tecnologie Molecolari (ISTM)
Consiglio Nazionale delle Ricerche (CNR)
via C. Golgi 19, 20133 Milano (Italy)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201202154.
924
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 924 – 935

FULL PAPER
CT state corresponding to D+-p-Aꢀ is easily defined as the
state having the maximum charge separation or the maxi-
mum dipole moment along the D–A direction.[38] The defini-
tion of CT transitions is extremely useful in spectroscopy.
CT transitions involve electronic states with a large contri-
bution from the D+-p-Aꢀ state, and are characterized by a
large variation of the molecular dipole moment upon excita-
tion. Experimentally, absorption and fluorescence bands as-
sociated with CT transitions show a strong solvatochromism
and a considerable inhomogeneous broadening in polar sol-
vents.[39]
Scheme 1. Molecular structures of the investigated dyes.
The interplay between CT and LE states in CT chromo-
phores shows up in the most spectacular way with dual-fluo-
rescence phenomena.[40–46] Strictly speaking, dual fluores-
cence applies to systems, like N,N-dimethylaminobenzoni-
trile (DMABN), that show in the same solvent two distinct
fluorescence bands. In these systems a detectable fluores-
cence signal is observed not only from the lowest singlet
state, according to the Kasha rule,[47] but also from a higher-
energy state. Breaking the Kasha rule is of course possible,
provided that the lifetime of the higher-energy fluorescent
state is long enough: double fluorescence then not only re-
quires CT/LE competition driven by polar solvation, but
also the coupling to some slow (conformational) degree of
freedom, to slow down the conversion between the two
forms and give time for the system to fluoresce from both
states.[40–46] In a more relaxed definition, double fluorescence
describes systems for which the fluorescence can be ob-
tained from different states according to experimental con-
ditions.[48] For LE/CT systems, in particular, the fluorescent
state can switch from LE to CT upon increasing the solvent
polarity.[48–50] In this case, a single fluorescence band is ob-
served in each solvent, the Kasha rule is fulfilled, and there
are no special requirements about the relative lifetime of ex-
cited states.
ed benzene ring, which in turn acts as an acceptor (A). A
low-lying CT excitation is therefore expected for this dye,
polarized along the D–A axis, roughly coincident with the
long molecular axis. According to the Platt description of
electronic excitation in polyacenes,[69] the lowest electronic
1
excitation of chrysene is assigned to a Lb state, polarized
along the short molecular axis, and with a very weak intensi-
1
ty.[70] A localized state with Lb character is therefore also
expected in aza-substituted chrysene and related molecules,
even if its precise position is difficult to guess. In any case,
for the molecules discussed in this study a low-lying LE
state is expected, roughly polarized along the short molecu-
lar axis and hence making a sizable angle with the direction
of polarization of the CT absorption. Under these condi-
tions, fluorescence anisotropy spectra offer detailed informa-
tion on the nature of excited states. Based on recent advan-
ces in the theoretical simulation of anisotropy spectra in
polar solvents,[71] a detailed study of the spectral properties
of 1 and related dyes offers a clear-cut demonstration of the
evolution of the nature of the fluorescent state from LE in
nonpolar solvents to CT in polar solvents.
Results
The possibility to switch the fluorescent state from a LE
to a CT state is of interest for all applications, such as natu-
ral or artificial light-harvesting systems, for which charge
separation is sought after photoexcitation.[51–53] At the same
time, environment-dependent fluorescence is of interest for
applications in sensing and biosensing.[54–57] The broad emis-
sion from CT states can be usefully exploited for organic
light-emitting devices: the charge separation, which influen-
ces the intersystem-crossing yield, can strongly affect the
overall device efficiency.[58] Recognizing the nature of the
fluorescent state is, however, a nontrivial task: a strong sol-
vent dependence of fluorescence spectra, in terms of fre-
quency and band shapes, is in fact expected for several fami-
lies of organic chromophores with a pure CT character,
even in the absence of LE/CT competition.[23,59–65]
Materials: The materials used in the present study were pre-
pared in good yields by careful modification of a known lit-
erature procedure.[67] The compound 8-(N,N-dibutylamino)-
2-azachrysene 1 (Scheme 2) was obtained through a modi-
fied procedure previously reported[72] by some of us, which
consists in a Wittig condensation of (3-N,N-dibutylamino)-
benzyltriphenylphosphonium chloride 4 with isoquinoline-5-
carbaldehyde 6, followed by a photochemical cyclization of
Herein, we present a detailed study of a 2-azachrysene
substituted in position 8 by a dibutylamino group (1 in
Scheme 1). This dye is related to an interesting family of
polycondensed dyes extensively studied by Fromherz and
co-workers for voltage-sensing applications in biologically
relevant environments.[66–68] The dibutylamino group acts as
a good electron donor (D) with respect to the aza-substitut-
Scheme 2. a) Isoquinoline-5-carbaldehyde 6, tBuOK, dry THF, N2, reflux,
overnight; b) CH2Cl2, hn, 16 h, 158C.
Chem. Eur. J. 2013, 19, 924 – 935
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
925

A. Painelli et al.
the mixture of diastereoisomeric alkenes (E)- and (Z)-7 so
obtained. The corresponding methylated salt 3 was simply
prepared by reaction of compound 1 with MeI in THF.[72]
Similarly, the unsubstituted 2-azachrysene 2 was synthesized
by photochemical cyclization of a mixture of (E)- and (Z)-5-
aminostyrylisoquinoline 8 that was obtained by Wittig reac-
tion between benzyltriphenylphosphonium chloride (5) and
isoquinoline-5-carbaldehyde (6).
Optical properties: The solvents used for spectroscopic stud-
ies are listed in Table 1. We selected a set of solvents with
similar refractive index and increasing dielectric constant, to
Figure 1. Absorption (left panel, full lines) and emission spectra (right
panel, dashed lines) of 1 in solvents of different polarity. The molar ex-
tinction coefficient e axis refers to CH2Cl2 solutions; absorption spectra
in other solvents are normalized to the same maximum.
Table 1. Properties of solvents used for spectroscopic studies.
Solvent
Dielectric
constant
Refractive
index
Kamlet–Taft
a parameter
To understand the physical origin of these weak features,
we collected absorption and fluorescence spectra of the un-
substituted azachrysene dye 2. The absorption spectra of 2
(top panel of Figure 2) show an intense absorption at
ꢁ300 nm (not shown here; see the Supporting Information
in which other spectra of 2, including fluorescence, are re-
ported), which can be ascribed to allowed transitions of the
polycondensed rings polarized along the long molecular
axis. More relevant to our discussion is the weak absorption
feature (effi3000mꢀ1 cmꢀ1 in CH2Cl2 and DMSO) with a well-
resolved vibronic structure centered around 350 nm. This
weak absorption can be assigned to a nominally forbidden
transition (1Lb in the Platt scheme for unsubstituted poly-
cyclohexane (CH)
toluene (Tol)
2-MeTHF
CH2Cl2
propylene glycol (PrGly)
DMSO
2.0
2.4
7.0
8.9
32.1
46.7
1.43
1.50
1.41
1.42
1.43
1.48
0.00[73]
0.00[73]
0.00[73],[a]
0.13[73]
0.83[74]
0.00[73]
[a] The parameter refers to tetrahydrofuran.
span a wide range of solvent polarity. To avoid specific inter-
actions, namely H-bonding interactions involving the aza
group in 1 and 2,[75] solvents were selected with poor H-do-
nating characteristics (see the Kamlet–Taft a parameter in
Table 1). Indeed propylene glycol (PrGly), the only solvent
with a mild H-donating character, was considered because it
is one of the few polar solvents that can be frozen in glassy
matrices, as needed for fluorescence anisotropy spectra. In
any case, as discussed below, all spectra measured at ambi-
ent conditions in PrGly practically coincide with those meas-
ured in DMSO, thus confirming that, with reference to the
molecules of interest in this study, PrGly behaves as a polar
solvent with a similar polarity to that of DMSO. In PrGly a
secondary emission appears at ꢁ620 nm that, according to
lifetime data (see the Supporting Information), is safely as-
cribed to an excimer emission and will not be further dis-
cussed.
AHCTUNGTRENNUNG
acenes) polarized along the short molecular axis.[69,70] The
transition does not imply any important variation of the mo-
lecular dipole moment; accordingly, the absorption band is
nonsolvatochromic and does not broaden in polar solvents.
Absorption and fluorescence spectra of the dibutylamino-
substituted azachrysene dye (1) dissolved in solvents of dif-
ferent polarity are reported in Figure 1. The most interesting
feature of the absorption spectrum is the intense band locat-
ed at 350–370 nm: this band, which shows a pronounced sol-
vatochromism and an important broadening in polar sol-
vents, can be safely ascribed to a CT absorption polarized
along the D–A axis (roughly parallel to the long molecular
axis). At lower energy, two weak absorption features are
clearly seen in cyclohexane and in mildly polar solvents (up
to 2-methyltetrahydrofuran (2-MeTHF)). In more polar sol-
vents, the CT band moves to the red and overlaps these
small features.
Figure 2. Absorption spectra of 2 (top panel) and 3 (bottom panel) in dif-
ferent solvents.
926
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 924 – 935

Tuning the Nature of the Fluorescent State
FULL PAPER
We therefore tentatively assign the small features observed
to the red of the main (CT) absorption band of 1 to a local-
ized (vibrationally resolved) transition, polarized along the
short molecular axis.
For comparison, the bottom panel of Figure 2 shows the
absorption spectra of dye 3, a derivative of 1 with a meth-
porting the description of the relevant excitation as due to a
LE state. With increasing solvent polarity, the radiative life-
time of 1 decreases fast, reaching a value of ꢁ20 ns in
DMSO, as expected for a CT state. Indeed the strong CT
character of the low-lying transition of dye 3 leads to an
even shorter t0 (ꢁ10 ns).
ylated aza group. This group is a much stronger electron ac-
The different direction of polarization of the LE (roughly
parallel to the short molecular axis) and CT transition
(roughly parallel to the long molecular axis) in 1 suggests
that further experimental support of the intriguing hypothe-
sis of an evolution of the fluorescence band from LE to CT
with solvent polarity can be gained through detailed meas-
urements of fluorescence anisotropy spectra. Emission and
excitation anisotropy spectra collected for 1–3 in glassy ma-
trices obtained upon fast cooling of 2-MeTHF and PrGly
solutions are shown in Figure 3.
ceptor than the non-methylated group; accordingly, the ab-
sorption band related to the CT transition is redshifted with
respect to 2, falling in a spectral region well separated from
the LE transition. The low-energy spectral features observed
for 3 are therefore safely described as a pure CT transition.
Fluorescence spectra of 1 (Figure 1) show the typical sol-
vatochromism of CT states, with an important redshift in
polar solvents and an increasing Stokes shift with solvent
polarity. The band-shape evolution also resembles what is
often observed for polar dyes,[59–63] with a well-resolved vi-
bronic structure in cyclohexane and a progressive increase
of inhomogeneous broadening with increasing solvent polar-
ity. Based on the similarity with the behavior of polar dyes,
it is tempting to assign the fluorescence band to the CT
transition. However, in cyclohexane the 0–0 emission band
is exactly superimposed on the weak lowest-energy absorp-
tion band, assigned to the LE (0–0 vibronic) transition.
Moreover, according to the Kasha rule,[47] one expects fluo-
rescence from the lowest-energy singlet state. On this basis,
the fluorescence signal measured in cyclohexane can be as-
signed to the LE transition. However, the character of the
fluorescence band apparently changes to CT in more polar
solvents, as demonstrated by its pronounced solvatochrom-
ism, by the large Stokes shifts, and by the conspicuous inho-
mogeneous broadening, strongly suggesting a double-fluo-
rescence behavior for 1.
A thorough understanding of the complex behavior of
anisotropy spectra of 1 requires a detailed model of the LE/
CT interplay and is deferred to the next section. Here we
Partial support of this hypothesis is obtained from a de-
tailed analysis of fluorescence lifetimes. Table 2 collects the
fluorescence quantum yields, F, and fluorescence lifetimes,
tf, measured for the three compounds in solvents of differ-
ent polarity. From these data we can estimate the radiative
lifetime, t0 =tf/F (i.e., the inverse emission radiative rate).
Being unaffected by the rate of nonradiative processes, simi-
lar t0 values are expected for similar processes occurring in
different media. According to the data in Table 2, t0 is mar-
ginally solvent dependent for 2. The corresponding value is
very similar to that estimated for 1 in cyclohexane, thus sup-
Table 2. Wavenumber at maximum fluorescence intensity (n¯f), fluores-
cence quantum yields (F), fluorescence lifetimes (tf), and radiative life-
times (t0).
Molecule
Solvent
n¯f [cmꢀ1
]
F
tf [ns]
t0 [ns]
1
CH
Tol
24900
24000
22600
21100
27300
27300
27100
16200
15600
0.22
0.25
0.25
0.52
0.20
0.18
0.33
0.61
0.36
7.6
6.7
5.1
9.9
7.9
6.7
10.3
5.5
3.6
34.5
26.8
20.4
19.0
39.5
37.2
31.2
9.0
Figure 3. Fluorescence excitation (circles) and emission (diamonds) ani-
sotropy spectra in glassy 2-MeTHF at 77 K (green) and PrGly at 200 K
(violet), and the corresponding fluorescence excitation (full lines) and
emission (dashed lines) spectra of compounds 1–3. For 1 and 3, fluores-
cence emission/excitation spectra were collected exciting/detecting at the
maximum of absorption/emission; for 2, due to the superposition of the
vibronic 0–0 lines in absorption and emission, spectra were excited/de-
tected at the maximum of the 0–1 line.
CH2Cl2
DMSO
Tol
CH2Cl2
DMSO
CH2Cl2
DMSO
2
3
10.0
Chem. Eur. J. 2013, 19, 924 – 935
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
927

A. Painelli et al.
observe that fluorescence spectra collected in glassy polar
solvents are qualitatively different from those collected in
the same liquid solutions. In the liquid phase, after the
solute photoexcitation the surrounding molecules of the
polar solvent reorient themselves in response to the varia-
tion of the charge distribution in the solute, which favors the
emission from a CT state. This reorientation is not possible
in glassy matrices so that, irrespective of solvent polarity,
emission occurs from a LE state. Indeed, fluorescence spec-
tra measured in glassy polar solvents (2-MeTHF and PrGly,
Figure 3) closely resemble spectra measured in nonpolar or
mildly polar liquid solvents (see Figure 1). In line with this
interpretation, the fluorescence excitation anisotropy is
close to the limiting value, r=0.4, on the red edge of the ab-
sorption band, that is, in the region of the LE absorption,
and decreases fast in the region of the CT absorption. Al-
though this interpretation sounds appealing and is correct to
a first approximation, the detailed modeling in the next sec-
tion suggests a somewhat more complex picture.
Figure 4. Sketch of the proposed model. The top-left panel defines the
energies of the three electronic basis states and the corresponding mixing
matrix elements (ꢀt and ꢀb). The top-right panel shows the contour
plots for the diabatic potential energy surfaces associated with the three
electronic states in the Q,q plane. The two bottom panels show cuts of
the same potential energy surfaces along the q=0 (left) and Q=0 (right)
planes.
Anisotropy spectra of 2, with rffi0.4 in the region of the vi-
bronic 0–0 transition, confirm that the emission signal
comes from the same state involved in the absorption pro-
ACHTUNGTRENNUNGcess. The progressive decrease of r on moving away from
the 0–0 line suggests that the vibrational modes responsible
for the vibronic progression are polarized along a different
direction with respect to the electronic transition. Spectra of
3 are easily interpreted: in the region of the main absorption
and fluorescence bands, both emission and excitation aniso-
tropy are close to the limiting r=0.4 value, which confirms
that absorption and fluorescence involve the same state,
with dominant CT character.
the problem as a classical coordinate, and polar solvation is
fully described by eor, the solvent reorganization energy (as-
sociated with the D+-p-Aꢀ basis state).
This very simple model proved successful for the calcula-
tion of linear (absorption and fluorescence) and nonlinear
(two-photon absorption, second- and third-harmonic genera-
tion, etc.) spectra of several push–pull chromophores,[59–65]
and is therefore expected to capture the basic physics of the
CT transition in 1. However, it cannot describe the LE state.
To model the complex interplay between the LE and CT
states, we therefore extend the standard model to introduce
a third state, D-p*-A, which describes the LE state related
to the low-energy excitation observed in the substituted
azachrysene dye. We assign this state an energy 2z and, ac-
cording to early suggestions,[49,76] we account for the cou-
pling between this state and D+-p-Aꢀ: hD-p*-AjHjD+-p-
Aꢀi=ꢀb. We neglect the direct coupling between the LE
state D-p*-A and the neutral D-p-A state, in view of the
high energy difference between the two states. We assume
that the D-p*-A state is displaced with respect to D-p-A
along an effective vibrational coordinate, q, with frequency
w* and relaxation energy e*. A sketch of the relevant model
is given in Figure 4.
Modeling the LE/CT interplay: Optical spectra of polar D-
p-A chromophores (in which p is a conjugated bridge) can
be quantitatively described based on a two-state model that
is an extension of the Mulliken model,[1,2] also accounting
for molecular vibrations and polar solvation.[59–65] As in the
Mulliken model, the electronic Hamiltonian is written on
the basis of two diabatic states corresponding to the limiting
valence-bond structures, D-p-A and D+-p-Aꢀ. The two
states are separated by an energy 2h, and mixed by a matrix
element hD-p-AjHjD+-p-Aꢀi=ꢀt. The dipole moment as-
sociated with D+-p-Aꢀ, m0, is very large, and all other
matrix elements of the dipole-moment operator on the
chosen basis are neglected. An effective vibrational coordi-
nate, Q, is introduced to describe the vibronic structure of
optical spectra: the two basis states are assigned two har-
monic potential energy surfaces with the same frequency, wv,
but displaced minima, to introduce a linear dependence on
Q
of the energy gap between the two basis states:
The total Hamiltonian reads [Eq. (1)]:
pffiffiffiffiffiffi
2hꢀ 2evwvQ, in which ev is the vibrational relaxation
energy associated with the charge-separated basis state.
Polar solvation is described by introducing the polar compo-
nent of the reaction field F, that is, the electric field that is
generated at the solute location due to the reorientation of
polar solvent molecules around the polar solute. If the
solute is described as a continuum elastic medium, F enters
928
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 924 – 935

Tuning the Nature of the Fluorescent State
FULL PAPER
in which P and p are the conjugated momenta of Q and q,
respectively.
lowest states, to make the problem numerically tractable.
The Hamiltonian matrix (dimension 3M2) is diagonalized to
get nonadiabatic eigenstates that represent a numerically
exact solution provided M is large enough to guarantee con-
vergence (M=8–10 was used in this work). Transition ener-
gies and transition and permanent dipole moments are then
easily evaluated, as needed for the calculation of optical
spectra, according to the equations reported as Supporting
Information. Optical spectra calculated for each point of the
F-grid are finally mediated over all F values, thereby weight-
ing each spectrum for the relevant Boltzmann distribution
(see the Supporting Information for more details).
According to Mulliken,[1,2] in the two-state model the
dipole-moment operator is defined just in terms of m0, the
dipole moment of the zwitterionic D+-p-Aꢀ state, which is
very large if compared with the other matrix elements of
the dipole-moment operator, so that one can safely approxi-
mate hD-p-AjmjD-p-Ai=hD+-p-Aꢀ jmjD-p-Ai=0. We
stick to this approximation but, to account for the small in-
tensity of the LE transition, we introduce an additional
matrix element of the dipole-moment operator: hD-p-Ajmj
D-p*-Ai=m*. This dipole moment makes an angle f with
the D–A axis, parallel to m0. Fixing the x direction along m0,
the two matrices representing the x and y components of
the dipole-moment operator then read [Eqs. (2) and (3)]:
Discussion
Figure 5 shows absorption and fluorescence spectra calculat-
ed for 1 with optimized model parameters (reported in
Table 3) and with increasing eor values (reported in Table 4)
Strictly speaking, the total dipole-moment operator, de-
fined by the two matrices in Equations (2) and (3), should
be used in the effective Hamiltonian to describe polar solva-
tion. Accordingly, two components of the reaction field (Fx
and Fy) should be introduced as independent variables, thus
leading to a numerically intensive calculation. However, on
a physical basis, we expect m* !m0, so that solvation effects
related to m* are negligible. Estimated model parameters
fully confirm this working hypothesis, with m0 >20m* (see
Table 3).
Figure 5. Calculated absorption and emission spectra of 1 according to
parameters reported in Table 3 and for different eor values [eV] corre-
sponding to solvents of different polarity. The extinction coefficient axis
refers to eor =0.55 eV, as relevant to CH2Cl2; other spectra are normal-
ized to the same maximum.
Table 4. Solvent parameters entering the calculation of the optical spec-
tra of 1.
Solvent
eor [eV]
Solvent
eor [eV]
The Hamiltonian in Equation (1) describes three electron-
ic states coupled to two vibrational modes and to a solvation
coordinate. The solvation coordinate is related to a very
slow motion and is treated in the adiabatic approximation.
Specifically, the Hamiltonian is defined on a grid of F
values. For each F value, the Hamiltonian describing the
coupled electron-vibration problem is written on the basis
obtained as the direct product of the three electronic basis
states and the eigenstates of the two harmonic oscillators as-
sociated with Q and q. The nominally infinite bases associat-
ed with the two oscillators are truncated to the first M
CH
Tol
2-MeTHF
CH2Cl2
0.00
0.40
0.50
0.55
PrGly
DMSO
glycerol
0.60
0.65
0.75
to mimic increasing solvent polarity. The LE feature is clear-
ly seen in the low-energy wing of the calculated absorption
spectrum at eor =0. The progressive redshift and broadening
of the CT band with increasing eor nicely reproduces the
evolution of experimental spec-
tra with increasing solvent po-
Table 3. Molecular model parameters entering the calculation of the optical spectra of 1, as discussed in the
text. G=0.075 eV is the effective intrinsic width assigned to each transition (see the Supporting Information).
larity. Emission frequencies and
band shapes and their evolution
with solvent polarity are simi-
larly well reproduced.
h
t
z
b
m0
m*
f
ev
wv
e*
w*
[eV]
[eV]
[eV]
[eV]
[D]
[D]
[8]
[eV]
[eV]
[eV]
[eV]
1.51
1.07
1.46
0.04
19.6
0.85
50
0.28
0.18
0.15
0.14
Chem. Eur. J. 2013, 19, 924 – 935
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
929

A. Painelli et al.
To understand the concept of dual fluorescence and to
shed light on the optical spectra of 1, Figure 6 shows the en-
ergies of the ground and fluorescent states calculated as a
function of F. The color of the curve relevant to the fluores-
cent state gives information about its nature: the red versus
essentially pure LE, but upon increasing F it changes
abruptly to CT (corresponding curves in Figure 6 change
color abruptly from red to green). For medium to large eor
values (middle and right panels, Figure 6), the fluorescent-
state energy shows two minima along the reaction field co-
ordinate, one with LE character and the other with CT char-
acter. For intermediate eor values the two minima are within
thermal energy and one observes a bimodal probability dis-
tribution. Turning to the spectra, for low eor values the po-
tential energy surfaces in the left panels of Figure 6 show
that the lowest-energy absorption, which occurs on the verti-
cal, has a pure LE character and hence a small transition
dipole moment (related to m*). In agreement with experi-
mental data, a second much more intense vertical absorp-
tion is expected towards a higher-energy state (its potential
energy curve is not shown in Figure 6) with a pure CT char-
acter. The relaxed excited state responsible for steady-state
fluorescence has again a pure LE character. At intermediate
eor values (middle panel, Figure 6), the vertical excitation
leads again mainly to a state with pure LE character. How-
ever, due to the finite width of the ground-state distribution,
states with at least a partial CT nature give a small yet siza-
ble contribution to the absorption spectrum. In view of the
high intensity associated with the CT process if compared
with the LE transition, the marginal contribution of CT
transitions actually dominates the observed (and calculated)
absorption spectrum in solvents of intermediate polarity.
Emission comes from the equilibrated fluorescent state that,
for eor =0.5, has a comparable population of both the LE
and CT states. In other terms, due to thermal disorder, fluo-
rescence in intermediate-polarity solutions arises with simi-
lar proportions from the LE state of molecules experiencing
a small reaction field and from the CT state of molecules ex-
periencing a sizable reaction field. This is not clearly evident
from experimental (and calculated) spectra that are again
largely dominated by the CT emission, in view of its large
transition dipole moment.
In strongly polar solvents (eor =0.75, right panel in
Figure 6) the vertical absorption is again mainly driving the
system towards a LE state, yet the spectrum is largely domi-
nated by the (marginal) contribution from the CT transition,
mainly because of its comparatively large transition dipole
moment. The relaxed fluorescent state, however, is now a
pure CT state, with a negligible thermal population of the
LE state: emission in this case has an essentially pure CT
character.
This complex theoretical picture leads to an impressive
agreement with experimental spectra in Figure 1, a nontrivi-
al achievement, strongly supporting the reliability of the
proposed model. An even more stringent validation of the
model and of relevant parameters can be obtained from the
analysis of fluorescence anisotropy spectra. According to a
recently devised computational strategy,[71] briefly summar-
ized in the Supporting Information, we are in a position to
calculate fluorescence excitation and emission anisotropy
spectra, fully accounting for polar solvation. The spectra in
Figure 7 have been calculated for 1 by adopting exactly the
Figure 6. Full lines: the energy of the ground and fluorescent states calcu-
lated as functions of the reaction field F for the model parameters report-
ed in Table 3 and three eor values. The color scale on the right refers to
the curve relevant to the fluorescent state and measures the weight of
the CT state (pure red corresponds to pure LE character, pure green to
pure CT character). Dashed lines: probability distributions of the reac-
tion field Pgs(F) and Pfluo(F) calculated at 300 K.
green proportions are related to the weight of the LE state
versus the weight of the CT state (the CT state is a mixture
of D-p-A and D+-p-Aꢀ, so its weight is simply obtained as
the complement to 1 of the weight of the LE state). The
change in color, and hence in the nature of the lowest excit-
ed state, is fairly abrupt, in line with the small value of b,
the matrix element that mixes LE and CT states (see
Table 3), which suggests that the direct mixing between the
two states can be safely neglected except in a narrow region
of F values in which the two diabatic CT and LE states are
almost degenerate.
Figure 6 also reports the Boltzmann distributions associat-
ed with the ground and fluorescent states: Pgs(F)/exp-
A
U
Egs and Efluo are the energies of the ground and excited
states, respectively, k is the Boltzmann constant, and T the
absolute temperature. The ground-state distribution, Pgs,
enters the calculation of absorption spectra as well as of flu-
orescence spectra in glassy matrices, whereas the fluores-
cent-state distribution, Pfluo, is relevant to the calculation of
(steady-state) fluorescence spectra in liquid solutions.
The “regular” form of the ground-state energy curve im-
plies a typical bell-shaped ground-state distribution, the
width of which increases with solvent polarity, thus support-
ing increasing inhomogeneous spectral broadening in polar
solvents. More interesting is the behavior of the fluorescent
state: for all eor values, the nature of the vertical excitation is
930
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 924 – 935

Tuning the Nature of the Fluorescent State
FULL PAPER
same model used for the spectra in Figure 5 and fixing the
temperature (entering the Boltzmann distribution, see the
Supporting Information) to the glass transition temperature
of 2-MeTHF (90 K) and PrGly (200 K).
In glassy matrices, the solvent molecules cannot change
their orientation in response to the photoexcitation of the
solute, so that the distribution of the reaction field stays
fixed at the distribution relevant to the solute in the ground
state. Therefore, as sketched in Figure 8, in the calculation
Figure 8. Full lines: energy of the ground and fluorescent states calculat-
ed as functions of the reaction field F for the model parameters reported
in Table 3 and two eor values. The color scale on the right refers to the
curve relevant to the fluorescent state and measures the weight of the
CT state (pure red corresponds to pure LE character, pure green to pure
CT character). Dashed lines: probability distributions of the reaction
field relevant to the ground state Pgs(F) (as needed for the calculation of
optical spectra in glassy matrices) calculated at 90 K for eor =0.5, as rele-
vant to 2-MeTHF glassy matrices (left panel), and at 200 K for eor =0.75,
as relevant to undercooled glycerol (right panel).
Figure 7. Calculated fluorescence excitation anisotropy (circles) and fluo-
rescence emission anisotropy (diamonds) of 1 in 2-MeTHF (green) at
90 K and PrGly (violet) at 200 K, and the corresponding fluorescence ex-
citation (full lines) and emission (dashed lines) spectra. Fluorescence
emission/excitation spectra are calculated by fixing the excitation/detec-
tion wavelengths at 357/400 nm.
of fluorescence emission and excitation spectra and of the
relevant anisotropies in glassy matrices, the same distribu-
tion of the reaction field must be adopted as introduced in
the calculation of absorption spectra (of course at the rele-
vant temperature). As a result, irrespective of the solvent
polarity, in glassy solvents fluorescence mainly comes from
the LE state: in line with experimental data, the calculated
fluorescence in glassy solvents is nonsolvatochromic. More-
over, fluorescence excitation anisotropy is high (rffi0.4) in
the red edge of the fluorescence excitation profile, dominat-
ed by the LE absorption, and decreases at higher energies,
at which the excitation acquires a CT nature.
In glassy solvents fluorescence spectra are largely domi-
nated by the LE transition, mainly because under these con-
ditions the large reaction fields needed to stabilize the CT
state have negligibly small probabilities. It is therefore inter-
esting to study fluorescence anisotropy spectra in polar non-
glassy solvents, to gain information about fluorescence from
the CT state. Reliable fluorescence anisotropy data can be
obtained only in liquid solvents viscous enough to hinder
the solute reorientation in the timescale of the excited-state
lifetime, thus making the solute reorientation a slower pro-
The main differences between the spectra in 2-MeTHF
and PrGly are related to the different glass transition tem-
peratures. 2-MeTHF vitrifies at 90 K, so that the corre-
sponding reaction-field distribution is much narrower than
for PrGly, with a glass transition temperature of 200 K. At
200 K a sizable population is calculated of states with a
mixed LE/CT character, which leads to a smoother switch-
ing from LE to CT than in 2-MeTHF, as supported by the
slope of the fluorescence excitation anisotropy. Although
the calculated anisotropy spectra in 2-MeTHF compare very
well with experimental data, the fluorescence excitation
spectrum calculated in PrGly is somewhat redshifted. This is
most probably related to an overestimated intensity from
the mixed CT/LE region. Indeed, better results would be
obtained by fixing the temperature to 90 K also in this sol-
vent. A temperature difference of 110 K corresponds to
ꢁ0.01 eV, a fairly acceptable energy mismatch within the
scope of the present model.
AHCTUNGTREGcNNNU ess than fluorescence. As discussed in references [26] and
[77], glycerol satisfies these requirements. The top panels of
Figure 9 show the fluorescence excitation and relevant an-
AHCTUNGTREGiNNNU sotropy spectra of 1 in liquid glycerol (room temperature,
left panel) and in an undercooled glycerol matrix (200 K,
right panel). As expected, due to the similar polarity of the
two solvents, spectra collected in glassy glycerol resemble
closely those collected in glassy PrGly (see Figure 3). More
interesting are spectra collected in liquid glycerol (room
temperature). In liquid polar solvents the CT state is stabi-
lized by the solvent relaxation, so that, as seen in the right
panel of Figure 6, fluorescence arises from a pure CT state.
Indeed, the observed anisotropy is fairly large (r>0.3) and
basically flat in the region of the CT absorption, showing
just a small dip in the red wing of the excitation band, at
which the LE transition contributes to the excitation spec-
trum. The corresponding calculated spectra, reported in the
Chem. Eur. J. 2013, 19, 924 – 935
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
931

A. Painelli et al.
lies of systems are distinctively different. The first obvious
difference is that when dealing with intramolecular CT, that
is, with systems in which the D and A groups are bound to-
gether in a single molecular structure, there are no spectra
of the isolated D and A species for comparison: singling out
spectral features due to the CT excitation is then more deli-
cate than for CT complexes. Most often and fairly reliably,
CT excitations are recognized based on their distinctive de-
pendence on solvent polarity. CT transitions in fact induce
large variations of the molecular dipole moment leading to
an important solvatochromism that was recognized
early.[39,78,79]
At variance with intermolecular CT, intramolecular CT
occurs in well-defined molecular structures and, apart from
a few cases in which electrons are strongly coupled to low-
frequency conformational degrees of freedom,[80,81] electron-
ic energies are modulated by high-frequency vibrations that
appear in optical spectra with well-defined vibronic progres-
sions rather than with inhomogeneous broadening phenom-
ena. Polar solvation is a specific source of inhomogeneous
broadening in push–pull chromophores:[21,59–65,82] thermal dis-
order in the orientation of polar solvent molecules is in fact
responsible for a distribution of solvent reaction fields at the
solute location, with inhomogeneous broadening effects that
increase fast with the solvent polarity. In nonpolar solvents,
however, there are no effective mechanisms for inhomoge-
neous broadening of intramolecular CT transitions, and
many examples are known of CT absorption (and fluores-
cence) bands with a well-resolved vibronic structure.[59–65]
The possibility of controlling the nature of the excited
state, switching it from a LE state (typically a p–p* excita-
tion marginally affected by solvent polarity) to a CT state
with its distinctive solvatochromism, opens an interesting
scenario with important applicative implications,[51–58] which
Figure 9. Experimental (top panels) and calculated (bottom panels) fluo-
rescence excitation spectra (full lines) and fluorescence excitation anisot-
ACHTUNGTRENNUNGropy spectra (circles) of 1 in glycerol at room temperature (liquid phase,
left panels) and at 200 K (glassy phase, right panels). Fluorescence excita-
tion anisotropy spectra were collected by detecting the fluorescence
signal at the maximum of the emission spectra. The model parameter eor
relevant to the solvent relaxation is set to 0.75 eV for glycerol. Fluores-
cence emission spectra are reported in the Supporting Information.
AHCTUNGTRENNUNG
bottom panels of Figure 9, quantitatively reproduce the ex-
perimental spectra, hence offering very strong support of
the proposed picture.
Conclusion
The concept of CT states and excitations was introduced
60 years ago with reference to molecular complexes formed
in solution by the encounter between an electron donor (D)
and an acceptor (A) molecule.[1,2] The absorption spectra of
the complex show bands that have no counterpart in solu-
tions of either the isolated D or A molecules. These bands,
assigned to CT processes, are usually located to the red of
local excitations characteristic of either the D or A mole-
cules: their comparatively low energy is related to the low
ionization potential of D, the high electron affinity of A,
and the (usually very small) intermolecular hopping integral.
CT bands of molecular complexes are very broad and struc-
tureless: thermal disorder in the relative orientation and dis-
tance between the two molecules forming the complex mod-
ulates the intermolecular hopping integral, thus representing
a very effective source of inhomogeneous spectral broaden-
ing. The sum rule for the oscillator strength imposes that the
intensity associated with the CT absorption is borrowed
from LE states. However, the success of the Mulliken
model[1,2] demonstrates that, in most cases, the LE–CT inter-
actions need not be explicitly accounted for.
offers an intriguing challenge to theoreticians and spec
ACHTUNGTRENNUNGtros-
AHCTUNGTREGcNNUN opists. Dual fluorescence, with the concomitant appearance
of two fluorescence bands in the same spectrum, gives a
clear indication of LE/CT interplay, even if the detailed in-
terpretation of the phenomenon is somewhat controver-
sial.[40–46] Dual fluorescence is a rare phenomenon that re-
quires a very precise balance between the fluorescence life-
times of the two involved states and their interconversion
velocity, to guarantee the observation of sizable steady-state
fluorescence from both states.
Although not so striking as dual fluorescence, the obser-
vation of the evolution of fluorescence spectra from a LE
emission to a CT emission upon increasing solvent polarity
is a clear signature of LE/CT interplay. Double fluorescence
in this more relaxed definition is a rare phenomenon as
well: though somewhat independent of excited-state life-
times, it requires a LE state lower in energy than the CT
state in nonpolar solvents, but not too far from the CT state,
so that polar solvents can stabilize the CT state below the
LE state. The delicate issue is how to assign, in these sys-
tems, the emission to the CT or the LE transition. The evo-
lution of spectral band shapes with the solvent polarity is
not a guarantee for LE-to-CT evolution: in fact, at variance
Concepts and models developed to describe intermolecu-
lar CT processes in molecular complexes proved extremely
useful to set up the basis of current understanding of intra-
molecular CT in p-conjugated D-p-A (or push–pull) chro-
mophores. Although sharing many similarities, the two fami-
932
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 924 – 935

Tuning the Nature of the Fluorescent State
FULL PAPER
with the common paradigm, established in the field of CT
complexes, narrow CT bands with a well-resolved vibronic
structure are routinely observed for push–pull chromo-
phores in nonpolar solvents.[59–65]
both states are thermally populated. Under these conditions
the observed fluorescence is actually a superposition of the
signals due to the CT and the LE emission, actually leading
to a dual fluorescence in the strict definition. The two sig-
nals, however, occur in the same spectral region so that the
two contributions show an important spectral overlap.
In conclusion, a thorough experimental and theoretical
analysis proves unambiguously that the nature of the fluo-
rescent state of dye 1 evolves from a pure LE state in non-
polar solvents to a pure CT state in strongly polar solvents.
Solvent relaxation after photoexcitation plays a major role
in the stabilization of the CT state, so that in glassy matrices
emission mainly comes from the LE state, irrespective of
the solvent polarity. Thermal population of both states is ex-
pected in solvents of intermediate polarity, which leads to
the simultaneous emission from both states, even if relevant
spectral effects are difficult to single out due to the spectral
overlap of the two signals.
Herein, we have presented an extensive analysis of a
push–pull dye based on a polycondensed ring (azachrysene)
structure, which exhibits a delicate balance between a low-
lying LE state of the polyaromatic system and the CT state.
Clear signatures of the LE state are recognized in absorp-
tion spectra in which, at least in nonpolar or weakly polar
solvents, a weak and vibronically resolved band is observed
just to the red of the more intense CT transition. Upon in-
creasing solvent polarity, the CT band redshifts and broad-
ens overlapping the small LE features. In the nonpolar sol-
vent (cyclohexane) the fluorescence band is characterized
by a negligible Stokes shift and is therefore safely ascribed
to a LE transition. The important fluorescence solvato-
chromism definitely suggests that in polar solvents the fluo-
rescent state acquires a CT nature.
The resulting picture is fairly complex and efforts to build
reliable models must rely on an extensive set of experimen-
tal data. The noncollinear polarization of the CT and LE
transitions in this system makes fluorescence anisotropy
spectra extremely informative. Anisotropy spectra collected
in two glassy matrices obtained from solvents of different
polarity as well as in a liquid viscous solvent quite unambig-
uously point to the important role of the reorientation of
the polar solvent molecules after the solute photoexcitation
as the driving mechanism stabilizing the CT state.
The extensive set of spectroscopic data collected on dye 1
allowed us to define and validate a reliable model fully ac-
counting for the complex interplay between LE and CT
states. The proposed model shares the same basic physics
with early models for the LE/CT competition (or somewhat
analogous spectral phenomena) based on three electronic
states.[13–19,76] Herein, the model is extended to account for
the coupling of electronic degrees of freedom with molecu-
lar vibrations and with an effective polar solvation coordi-
nate. Moreover, by exploiting numerical techniques recently
Experimental Section
General: All commercially available solvents and reagents were used as
received without further purification. 8-(N,N-Dibutylamino)-2-azachry-
sene (1) and the corresponding methylated salt (3) were prepared accord-
ing to the procedure previously reported by us. For a detailed description
of their synthesis and characterization, see ref. [72]. Thin-layer chroma-
tography (TLC) was conducted on plates precoated with silica gel Si 60-
F254 (Merck, Darmstadt, Germany). Column chromatography was car-
ried out on silica gel Si 60, mesh size 0.040–0.063 mm (Merck, Darmstadt,
Germany). 1H and 13C NMR spectra were recorded on a Bruker Avance
400 spectrometer (400 and 100.6 MHz, respectively). Chemical shifts are
indicated in parts per million downfield from SiMe4, with the residual
proton (CHCl3, d=7.26 ppm) and carbon (CDCl3, d=77.0 ppm) solvent
resonances as internal reference. ESI mass spectra were recorded on a
Thermo Finnigan LCQ Advantage spectrometer equipped with an elec-
trospray ion trap and an electrospray ion-trap mass spectrometer ICR-
FTMS APEX II (Bruker Daltonics) by the Centro Interdipartimentale
Grandi Apparecchiature (C.I.G.A.) of the University of Milano.
Benzyltriphenylphosphonium chloride (5): PPh3 (2.48 g, 9.4 mmol) was
added to a solution of benzyl chloride (1.00 g, 7.9 mmol) in heptane
(10 mL) and the mixture was heated at reflux for 24 h. After cooling to
room temperature, the white precipitate was isolated by filtration and
washed with heptane and Et2O, then dried in vacuo to afford pure 5
(2.73 g, 7.03 mmol, 89%). 1H NMR (400 MHz, CDCl3): d=7.78–7.73 (m,
9H; Ar-H), 7.63–7.27 (m, 6H; Ar-H), 7.20 (m, 1H; Ar-H), 7.10 (m, 4H;
developed
in
the
framework
of
essential-state
models,[23,59–65,71] for the first time we are able to quantita-
tively reproduce the complete spectral evolution of the
system with solvent polarity. With a minimal set of opti-
mized model parameters we calculate absorption, fluores-
cence, and fluorescence anisotropy spectra of the chromo-
phore, as well as their solvent dependence, and reproduce
not only the spectral position but also—which is much more
challenging—the relevant band shapes. The proposed model
is general, and indeed we explicitly verified its applicability
to an analogue of chromophore 1 with three rather than
four condensed rings (a substituted azaphenanthrene). Rele-
vant results are reported in the Supporting Information.
An important and nontrivial issue emerging from the
analysis concerns the role of thermal disorder associated
with polar solvation. The LE/CT interplay implies that, at
least in some solvents, the energy gap between the two ex-
cited states is comparable to the thermal energy, so that
Ar-H), 5.53 ppm (d, 1J
CDCl3): d=134.86 (J(C,P)=2.3 Hz; CH), 134.42 (J
131.60 (J(C,P)=5.4 Hz; CH), 130.10 (J(C,P)=12.4 Hz; CH), 128.78 (J-
(C,P)=2.7 Hz; CH), 128.28 (J(C,P)=3.7 Hz; CH), 127.43 (J(C,P)=
8.9 Hz; C), 118.06 (J(C,P)=85.4 Hz; C), 30.67 ppm (J(C,P)=46.3 Hz;
ACHTUNGTRENNUNG
T
ACHTUNGTRENNUNG
G
ACHTUNGTRENNUNG
E
N
ACHTUNGTRENNUNG
G
ACHTUNGTRENNUNG
CH2); HRMS (ESI): m/z: calcd for C25H22P+: 353.15 [MꢀCl]+; found:
353.2; elemental analysis calcd (%) for C25H22PCl: C 77.22, H 5.70;
found: C 77.19, H 5.71.
AHCTUNGTERG(NNUN E,Z)-5-Styrylisoquinoline (8): Benzyltriphenylphosphonium chloride 5
(2.00 g, 5.14 mmol) was dissolved in dry THF (40 mL) under a nitrogen
atmosphere, tBuOK (656 mg, 5.37 mmol) was added, and the solution
was heated at reflux. Isoquinoline-5-carbaldehyde 6 (735 mg, 4.68 mmol)
dissolved in dry THF (5 mL) was then added and the reaction mixture
was heated at reflux overnight. The solvent was concentrated in vacuo,
then a solution of hexane/AcOEt (4:1 v/v, 20 mL) was added to promote
the precipitation of triphenylphosphine oxide. The solid formed was iso-
lated by filtration, and the solution was evaporated in vacuo. The crude
product was purified by column chromatography (silica gel, 1:1 hexane/
Chem. Eur. J. 2013, 19, 924 – 935
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
933

A. Painelli et al.
AcOEt) to afford 8 (615 mg, 2.66 mmol, 51.7%) as a mixture of diaster-
eoisomers. Minimum amounts of pure (E)-5-styrylisoquinoline and (Z)-5-
styrylisoquinoline were isolated only for NMR characterization. (Z)-8:
(after rapid cooling at 200 K). Further information on fluorescence aniso-
tropy measurements can be found in ref. [71].
1H NMR (400 MHz, CDCl3): d=9.27 (s, 1H), 8.50 (d, 3J
ACHTUNGTRENNUNG
1H), 7.89 (d, 3J
(d, 3J(H,H)=7.1 Hz, 1H), 7.48 (t, 3J
3H), 7.02–7.05 (m, 2H), 6.96 (d, 3J
ACHTUNGTRENNUNG ACHTUNGTRENNUNG
(H,H)=8.1 Hz, 1H), 7.81 (d, 3J
G
ACHTUNGTRENNUNG
Acknowledgements
AHCTUNGTRENNUNG
ACHTUNGTRENNUNG
This work was partially supported by the Fondazione Cariparma through
the Project 2010.0329 and by the Italian Ministry of University and Re-
search (MIUR) through the Project FIRB-Futuro in Ricerca
RBFR10Y5VW. C.S. thanks the University of Parma and INSTM for fi-
nancial support.
136.44, 136.30, 134.40, 134.18, 133.30, 130.67, 128.96, 128.20, 127.44,
127.05, 126.99, 126.46, 117.83 ppm. (E)-8: 1H NMR (400 MHz, CDCl3):
3
d=9.27 (s, 1H), 8.58 (d, J
G
3J
ACHTUNGTRENNUNG ACHTUNGTRENNUNG
(H,H)=16 Hz, 1H), 7.60–7.65 (m, 3H), 7.42 (t, 3J
7.31–7.35 (m, 2H), 7.22 ppm (d, 3J
AHCTUNGTRENNUNG
(100.6 MHz, CDCl3): d=153.19, 143.32, 137.12, 134.08, 133.96, 132.80,
128.86, 128.23, 127.38, 127.25, 127.11, 126.80, 123.84, 116.67 ppm; HRMS
(ESI): m/z: calcd for C17H14N+: 232.1126 [M+H]+; found: 232.1120; ele-
mental analysis calcd (%) for C17H13N: C 88.28, H 5.67, N 6.06; found: C
88.38, H 5.65, N 6.08.
[1] R. S. Mulliken, J. Am. Chem. Soc. 1952, 74, 811–824.
[2] R. S. Mulliken, W. B. Person, Molecular Complexes, Wiley, New
York, 1969.
[3] Z. G. Soos, D. J. Klein, in Molecular Associations, Vol. 1, (Ed. R.
Foster), Academic Press, New York, 1971, p.1.
[4] A. Painelli, A. Girlando, J. Chem. Phys. 1987, 87, 1705–1711.
[5] C. Pecile, A. Painelli, A. Girlando, Mol. Cryst. Liq. Cryst. 1989, 171,
69–87.
[6] N. S. Hush, Prog. Inorg. Chem. 1967, 8, 391–444.
[7] N. S. Hush, Coord. Chem. Rev. 1985, 64, 135–157.
[8] N. S. Hush, J. R. Reimers, Chem. Rev. 2000, 100, 775–786.
[9] M. Bixon, J. Jortner, Adv. Chem. Phys. 1999, 106, 35–202.
[10] J. Jortner, J. Ulstrup, J. Chem. Phys. 1973, 63, 4358–4368.
[11] J. Jortner, M. J. Bixon, J. Chem. Phys. 1988, 88, 167–170.
[12] J. Cortes, H. Heitele, J. Jortner, J. Phys. Chem. 1994, 98, 2527–2536.
[13] D. R. Kanis, M. A. Ratner, T. Marks, J. Chem. Res. 1994, 94, 195–
242.
2-Azachrysene (2): A magnetically stirred solution of (E)- and (Z)-5-styr-
ylisoquinoline 8 (518 mg, 2.24 mmol) in CH2Cl2 (500 mL) was irradiated
by a Hg high-pressure lamp in a quartz reactor for 16 h at 158C. The mix-
ture was concentrated to a small volume, which allowed the precipitation
of 2-azachrysene (2) that was collected by filtration. After evaporation to
dryness of the remaining solution, the solid residue was rinsed with Et2O,
thus affording a second crop of compound 2 (410 mg, total yield 80%).
1H NMR (400 MHz, CDCl3): d=9.36 (s, 1H), 8.78–8.84 (m, 3H), 8.64 (d,
3J(H,H)=8.8 Hz, 1H), 8.49 (d, 3J
ACHTUNGTRENNUNG ACHTUNGTREN(NNGU H,H)=6 Hz, 1H), 8.02–8.10 (m, 3H),
7.69–7.78 ppm (m, 2H); 13C NMR (100.6 MHz, CDCl3): d=152.20,
144.78, 134.55, 133.09, 130.64, 130.21, 128.78, 128.16, 127.39, 127.14,
126.61, 125.58, 123.47, 122.74, 120.79, 116.35 ppm; HRMS (ESI): m/z:
calcd for C17H12N+: 230.0970 [M+H]+; found: 230.0962; elemental analy-
sis calcd (%) for C17H11N: C 89.06, H 4.84, N 6.11; found: C 89.20, H
4.85, N 6.09.
[14] S. R. Marder, B. Kippelen, A. K.-Y. Jen, N. Peyghambarian, Nature
1997, 388, 845–851.
[15] J. Brꢃdas, K. Cornil, F. B. D. Meyers, In Handbook of Conducting
Polymers, 2nd ed. (Eds: T. A. Skotheim, R. L. Elsenbaumer, J. R.
Reynolds), Marcel Dekker, New York, 1998, p.1.
[16] J. L. Oudar, D. S. Chemla, J. Chem. Phys. 1977, 66, 2664–2668.
[17] S. R. Marder, D. N. Beratan, L.-T. Cheng, Science 1991, 252, 103–
106.
Spectroscopic measurements: Solvents used for spectroscopic measure-
ments: cyclohexane (CH), Sigma–Aldrich, Chromasolv plus ꢂ99.9%; tol-
uene (Tol), Sigma–Aldrich, Chromasolv plus ꢂ99.9%; 2-methyltetrahy-
drofuran (2-MeTHF), anhydrous ꢂ99%; dichloromethane (CH2Cl2),
Sigma–Aldrich, Chromasolv plus ꢂ99.9%; propylene glycol (PrGly),
Riedel-de Haꢂn, 99.5%; dimethyl sulfoxide (DMSO), Riedel-de Haꢂn,
99.5%; glycerol, Sigma–Aldrich, anhydrous ꢂ99.5%. All solvents were
used as received, except 2-MeTHF, which was stored overnight over mo-
lecular sieves (0.3 nm).
[18] G. Chen, J. W. Perry, W. A. Goddard, J. Am. Chem. Soc. 1994, 116,
10679–10685.
[19] W. H. Thompson, M. Blanchard-Desce, V. Alain, J. Muller, A. Fort,
M. Barzoukas, J. T. Hynes, J. Phys. Chem. A 1999, 103, 3766–3771.
[20] A. Painelli, Chem. Phys. Lett. 1998, 285, 352–358.
[21] A. Painelli, Chem. Phys. 1999, 245, 185–197.
[22] L. Grisanti, G. D’Avino, A. Painelli, J. Guash, I. Ratera, J. Veciana,
J. Phys. Chem. B 2009, 113, 4718–4725.
[23] F. Terenziani, A. Painelli, C. Katan, M. Charlot, M. Blanchard-
Desce, J. Am. Chem. Soc. 2006, 128, 15742–15755.
[24] F. Terenziani, O. V. Przhonska, S. Webster, L. A. Padilha, Y. L. Slo-
minsky, I. G. Davydenko, A. O. Gerasov, Y. P. Kovtun, M. P. Shan-
dura, A. D. Kachkovski, J. Phys. Chem. Lett. 2010, 1, 1800–1804.
[25] C. Sissa, F. Terenziani, A. Painelli, A. Abbotto, L. Bellotto, C. Mar-
inzi, E. Garbin, C. Ferrante, R. Bozio, J. Phys. Chem. B 2010, 114,
882–893.
Absorption spectra were collected on a Perkin–Elmer Lambda 650 spec-
trometer on solutions of concentration ꢁ10ꢀ5 m, which verified the Beer–
Lambert law for measurements of the molar extinction coefficient. Fluo-
rescence emission and excitation spectra were measured on a Horiba Jo-
binYvon Fluoromax-3 spectrofluorometer equipped with a xenon arc-
lamp as light source, on dilute solutions (ꢁ10ꢀ6 m) to minimize self-ab-
sorption. Fluorescein in NaOH 0.1m (fluorescence quantum yield=0.9)
was used as the standard for the determination of the fluorescence quan-
tum yields. Fluorescence decays were recorded with the same spectro-
fluorometer, in time-correlated single-photon counting (TCSPC) mode,
by using different nano-LEDs as pulsed excitation sources. The same in-
strument, equipped with emission and excitation Glan–Thompson auto-
matic polarizers, was used to measure fluorescence anisotropy, defined as
[Eq. (4)]:
[26] C. Sissa, F. Terenziani, A. Painelli, R. B. Kanth Siram, S. Patil, J.
Phys. Chem. B 2012, 116, 4959–4966.
[27] C. Sissa, P. Mohamadzadeh Jahani, Z. G. Soos, A. Painelli, Chem-
PhysChem. 2012, 13, 2795–2800.
[28] F. Terenziani, C. Sissa, A. Painelli, J. Phys. Chem. B 2008, 112,
5079–5087.
[29] C. Sissa, V. Parthasarathy, D. Drouin-Kucma, M. H. V. Verts, M.
Blanchard-Desce, F. Terenziani, Phys. Chem. Chem. Phys. 2010, 12,
11715–11727.
[30] J. Campo, A. Painelli, F. Terenziani, T. Van Regemorter, D. Bel-
jonne, E. Goovaerts, W. Wenseleers, J. Am. Chem. Soc. 2010, 132,
16467–16478.
in which Ik is the fluorescence intensity when emission and excitation po-
larizers are parallel, and I? is the fluorescence intensity when the two po-
larizers are orthogonal.[47] Fluorescence anisotropy spectra were meas-
ured in viscous media or in rigid matrices, to avoid depolarization caused
by rotational diffusion. Glycerol at room temperature is viscous enough
to avoid diffusion,[26,77] whereas other measurements were performed in
glassy 2-MeTHF (after rapid cooling at 77 K) and undercooled PrGly
[31] F. Terenziani, G. D’Avino, A. Painelli, ChemPhysChem 2007, 8,
2433–2444.
934
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 924 – 935

Tuning the Nature of the Fluorescent State
FULL PAPER
[32] S. Olsen, R. H. McKenzie, J. Chem. Phys. 2009, 130, 184302–184314.
[33] S. Olsen, J. Chem. Theory Comput. 2010, 6, 1089–1103.
[34] S. Olsen, R. H. McKenzie, J. Chem. Phys. 2011, 134, 114520–114532.
[35] S. Olsen, J. Phys. Chem. A 2012, 116, 1486–1492.
[36] J. N. Murrel, J. Chem. Phys. 1959, 81, 5037–5043.
[37] S. Masaki, T. M. N. Okada, Y. Sakat, S. Misumi, Bull. Chem. Soc.
Jpn. 1976, 49, 1277–1283.
[38] T. Van Voorhis, T. Kowalczyk, B. Kaduk, L.-P. Wang, C.-L. Cheng,
Q. Wu, Annu. Rev. Phys. Chem. 2010, 61, 149–170.
[39] C. Reichardt, Chem. Rev. 1994, 94, 2319–2358.
[40] Z. R. Grabowski, K. Rotkiewicz, W. Rubaszewska, E. Kirror-Kamin-
ska, Acta Physica Polonica 1978, A54, 767–776.
[62] L. Grisanti, C. Sissa, F. Terenziani, A. Painelli, D. Roberto, F. Tes-
sore, R. Ugo, S. Quici, I. Fortunati, E. Garbin, Phys. Chem. Chem.
Phys. 2009, 11, 9450–9457.
[63] F. Todescato, I. Fortunati, S. Carlotto, C. Ferrante, L. Grisanti, C.
Sissa, A. Painelli, A. Colombo, C. Dragonetti, D. Roberto, Phys.
Chem. Chem. Phys. 2011, 13, 11099–11109.
[64] L. Grisanti, F. Terenziani, C. Sissa, M. Cavazzini, F. Rizzo, S. Orlan-
di, A. Painelli, J. Phys. Chem. B 2011, 115, 11420–11430.
[65] C. Sissa, F. Terenziani, A. Painelli, A. K. Manna, S. K. Pati, Chem.
Phys. 2012, 404, 9–15.
[66] C. Rçcker, A. Heilemann, P. Fromherz, J. Phys. Chem. 1996, 100,
12172–12177.
[41] K. A. Zachariasse, Chem. Phys. Lett. 2000, 320, 8–13.
[42] W. Rettig, S. Lutze, Chem. Phys. Lett. 2001, 341, 263–271.
[43] A. B. J. Parusel, W. Retting, K. Rotkiewicz, J. Phys. Chem. A 2002,
106, 2293–2299.
[44] Z. R. Grabowski, K. Rotkiewicz, W. Retting, Chem. Rev. 2003, 103,
3899–4031.
[67] G. Hꢅbener, A. Lambacher, P. Fromherz, J. Phys. Chem. B 2003,
107, 7896–7902.
[68] B. Kuhn, P. Fromherz, J. Phys. Chem. B 2003, 107, 7903–7913.
[69] J. R. Platt, J. Chem. Phys. 1949, 17, 484–495.
[70] P. N. Skancke, Acta Chem. Scand. 1965, 19, 401–413; H. B. Klevens,
J. R. Platt, J. Chem. Phys. 1949, 17, 470–481; N. A. Borisevich, G. G.
Dyachenko, V. A. Petukhov, M. A. Semenov, Optics and Spectrosco-
py 2008, 105, 859–866.
[45] T. Yoshihara, S. I. Druzhinin, K. A. Zachariasse, J. Am. Chem. Soc.
2004, 126, 8535–8539.
[46] A. Demeter, S. I. Druzhinin, A. A. Kovalenko, T. A. Senyushkina,
K. A. Zachariasse, J. Phys. Chem. A 2011, 115, 1521–1537.
[47] J. R. Lakowicz, Principles of Fluorescence Spectroscopy, Kluwer
Academic/Plenum Publishers, New York, 1999.
[48] F. B. Dias, S. Pollock, G. Hedley, L.-O. Pꢄlsson, A. P. Monkman, I.
Perepichka, I. F. Perepichka, M. Tavalsi, M. R. Bryce, J. Phys.
Chem. B 2006, 110, 19329–19339.
[71] C. Sissa, A. Painelli, M. Blanchard-Desce, F. Terenziani, J. Phys.
Chem. B 2011, 115, 7009–7020.
[72] S. Quici, V. Calabrese, E. Rossi, E. Cariati, C. Dragonetti, D. Rob-
erto, E. Tordin, F. De Angelis, S. Fantacci, Chem. Commun. 2010,
46, 8374–8376.
[73] C. Reichardt, Solvents and Solvents Effects in Organic Chemistry,
Wiley-VCH, Weinheim, 2003, p.433.
[49] J. Herbich, A. Kapturkiewicz, J. Am. Chem. Soc. 1998, 120, 1014–
1029.
[50] N. Banerji, G. Angulo, I. Barabanov, E. Vauthey, J. Am. Chem. Soc.
2008, 130, 9665–9674.
[74] R. M. C. GonÅalves, A. M. N. Simoes, L. M. P. C. Albuquerque, M.
Roses, C. Rafols, E. Bosch, J. Chem. Res. 1993, 214–215.
[75] L. Brooker, R. Sprague, C. Smyth, G. Lewis, J. Am. Chem. Soc.
1940, 62, 1116–1125.
[51] S. Gꢅnes, H. Neugebauer, N. Sariciftci, Chem. Rev. 2007, 107, 1324–
1338.
[76] P. Pasman, F. Rob, J. W. Verhoeven, J. Am. Chem. Soc. 1982, 104,
5127–5133.
[52] M. Grꢆtzel, Acc. Chem. Res. 2009, 42, 1788–1798.
[53] J. Roncali, Acc. Chem. Res. 2010, 43, 1716–1730.
[54] M. S. T. GonÅalves, Chem. Rev. 2009, 109, 190–212.
[55] M. Y. Berezin, S. Achilefu, Chem. Rev. 2010, 110, 2641–2684.
[56] H. Kobayashi, M. Hogawa, R. Alford, P. U. Y. Choike, Chem. Rev.
2010, 110, 2620–2640.
[57] G. Signore, R. Nifosꢇ, L. Albertazzi, B. Storti, R. Bizzarri, J. Am.
Chem. Soc. 2010, 132, 1276–1288.
[58] S. M. King, R. Matheson, F. B. Dias, A. P. Monkman, J. Phys. Chem.
B 2008, 112, 8010–8016.
[77] S. Webster, J. Fu, L. A. Padilha, O. V. Przhonska, D. J. Hagan, E. W.
Van Stryland, M. V. Bondar, Y. L. Slominsky, A. D. Kachkovski,
Chem. Phys. 2008, 348, 143–151.
[78] W. Liptay, Angew. Chem. 1969, 81, 195–206; Angew. Chem. Int. Ed.
Engl. 1969, 8, 177–188.
[79] W. Liptay, In Excited States (Ed.: E. Lim), Academic Press, New
York, 1974, p.129.
[80] A. Painelli, F. Terenziani, J. Phys. Chem. A 2000, 104, 11041–11048.
[81] A. Painelli, F. Terenziani, D. Comoretto, J. Phys. Chem. A 2000, 104,
11049–11054.
[59] B. Boldrini, E. Cavalli, A. Painelli, F. Terenziani, J. Phys. Chem. A
2002, 106, 6286–6294.
[82] A. Painelli, F. Terenziani, Chem. Phys. Lett. 1999, 312, 211–220.
[60] F. Terenziani, A. Painelli, A. Girlando, R. M. Metzger, J. Phys.
Chem. B 2004, 108, 10743–10750.
[61] A. Painelli, F. Terenziani, L. Angiolini, T. Benelli, L. Giorgini,
Chem. Eur. J. 2005, 11, 6053–6063.
Received: June 18, 2012
Revised: September 25, 2012
Published online: November 23, 2012
Chem. Eur. J. 2013, 19, 924 – 935
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
935