Two-way molecular switches with large nonlinear optical contrast

Article abstract of DOI:10.1002/chem.200801967

To optimize the nonlinear optical (NLO) contrast, a series of indolinooxazolidine derivatives with electron-withdrawing substituents in the para position on the indolinic residue have been synthesized. Their linear and nonlinear optical properties have been characterized by UV-visible absorption and hyper-Rayleigh scattering measurements, as well as by ab initio calculations. The two-way photo- or pH-triggered switching mechanism has been demonstrated by comparing the absorption spectra of the zwitterionic and protonated open forms (POF). Hyper-Rayleigh measurements have revealed that the second-order NLO contrast between the closed indolinooxazolidine and the open it-conjugated colored forms remain very large upon substitution. Theory and measurements show that for the POFs the amplitude of the first hyperpolarizability follows the Hammett parameters of the withdrawing groups. However, because the measurements are performed in resonance, to recover this behavior, elaborate procedures including homogeneous and inhomogeneous broadenings, as well as single-mode vibronic structures are necessary to extrapolate to the static limit.

Full text of DOI:10.1002/chem.200801967

DOI: 10.1002/chem.200801967
Two-Way Molecular Switches with Large Nonlinear Optical Contrast
Fabien ManÅois,[a] Jean-Luc Pozzo,[a] Jianfeng Pan,[a] Frꢀdꢀric Adamietz,[a]
Vincent Rodriguez,[a] Laurent Ducasse,[a] Frꢀdꢀric Castet,*[a] Aurꢀlie Plaquet,[a, b] and
Benoꢁt Champagne*[a, b]
Abstract: To optimize the nonlinear
optical (NLO) contrast, a series of in-
dolinooxazolidine derivatives with elec-
tron-withdrawing substituents in the
para position on the indolinic residue
have been synthesized. Their linear
and nonlinear optical properties have
been characterized by UV-visible ab-
sorption and hyper-Rayleigh scattering
measurements, as well as by ab initio
calculations. The two-way photo- or
pH-triggered switching mechanism has
been demonstrated by comparing the
absorption spectra of the zwitterionic
and protonated open forms (POF).
Hyper-Rayleigh measurements have
revealed that the second-order NLO
contrast between the closed indoli-
nooxazolidine and the open p-conju-
gated colored forms remain very large
upon substitution. Theory and meas-
urements show that for the POFs the
amplitude of the first hyperpolarizabili-
ty follows the Hammett parameters of
the withdrawing groups. However, be-
cause the measurements are performed
in resonance, to recover this behavior,
elaborate procedures including homo-
geneous and inhomogeneous broaden-
ings, as well as single-mode vibronic
structures are necessary to extrapolate
to the static limit.
Keywords: ab initio calculations ·
hyper-Rayleigh measurements
indolinooxazolidine derivatives
·
·
molecular switches
optics
·
nonlinear
Introduction
cules with the typical advantages of organic materials, such
as tailored synthesis, easy processing, and low cost. The effi-
ciency of such NLO switches results from a subtle compro-
mise between several aspects, which are sometimes contra-
dictory. First, the hyperpolarizability (b) of the ON state
must be as large as possible, whereas it should ideally be
zero for the OFF state, in order to maximize the contrast of
the property. Second, the two forms must be stable, energet-
ically equiprobable, and have a significant activation barrier
to avoid thermally activated reverse reactions. Third, the ge-
ometry changes in the molecule during the transformation
should be limited to obtain a commutation as quickly as
possible, as well as the possibility of switching in the solid
state.
Within this framework, the NLO properties of keto/enol
systems have been extensively studied by experimental and
theoretical means, owing to, in particular, their ability to
commute in the crystalline state, by using either light or
temperature.[2] Hence, Sliwa and co-workers have recently
highlighted that the small difference between the N-(3,5-di-
tert-butylsalicylidene)-2-aminopyridine and N-(3,5-di-tert-bu-
tylsalicylidene)-4-aminopyridine isomers lead to very differ-
ent switching mechanisms. In the latter, the switching mech-
anism is triggered by light, whereas in the former the trigger
In recent years, the development of optoelectronic devices
for optical information transport and storage has motivated
the design of molecular systems with commutable nonlinear
optical (NLO) properties.[1] Among them, organic photo-
chromic compounds are targeted as potential candidates for
integration into functionalized materials that combine the
unique electronic and optical properties of conjugated mole-
[a] F. ManÅois, Prof. J.-L. Pozzo, Dr. J. Pan, F. Adamietz,
Prof. V. Rodriguez, Dr. L. Ducasse, Dr. F. Castet, A. Plaquet,
Dr. B. Champagne
Institut des Sciences Molꢀculaires
UMR 5255 CNRS-Universitꢀ de Bordeaux,
Cours de la Libꢀration, 351, 33405 Talence CEDEX (France)
Fax : (+33)540006645
E-mail: f.castet@ism.u-bordeaux1.fr
[b] A. Plaquet, Dr. B. Champagne
Laboratoire de Chimie Thꢀorique Appliquꢀe
Facultꢀs Universitaires Notre-Dame de la Paix (FUNDP)
Rue de Bruxelles, 61, B-5000 Namur (Belgium)
Fax : (+32)81724567
E-mail: benoit.champagne@fundp.ac.be
2560
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2560 – 2571

FULL PAPER
is thermochromic. This difference has been related to the
relative position of the pyridine ring, which, for the former
but not for the latter, is in the same plane as the benzene
ring with the enol function. However, the relatively small
NLO responses of the two forms (c(2), the second-order non-
linear susceptibility, which is
various styrylic residues.[17] The switching mechanism, which
ꢀ
consists of light-induced cleavage of the s-C O bond on the
oxazolidine moiety of the closed form (CF) that leads to a
zwitterionic open form (OF; Scheme 1), is associated with a
large variation in the NLO response. In particular, a con-
the macroscopic equivalent of
b, is only a few times that of
urea crystals) and their weak
NLO contrast upon switching
limit their potential applica-
tions.
Nitrobenzylpyridines
represent another class of
compounds that display photo-
tautomerization and thermo-
ACHTUNGTRENNUNGswitchable behavior, which
leads to substantial variations
of the quadratic NLO re-
Scheme 1. Photo/acidochromic equilibrium for the dimethylaminophenylethylenylindolino
derivatives. The Cartesian frame used in the calculations is also displayed.
ACHTUNGTREN[NUNG 2,1-b]oxazolidine
sponse.[3]
Dithiazolylethene-
based derivatives incorporating
a push–pull structural motif[4] have been investigated for
their photoswitchable NLO properties. Spiropyran/merocya-
nine systems have also been investigated and ab initio calcu-
lations predict large hyperpolarizability contrasts when ap-
propriate substitutions by donor and acceptor groups are re-
alized.[5] In fact, spiropyran- and spirooxazine-derivatives
have already been incorporated in poly(methyl methacry-
late) (PMMA) films to obtain materials that exhibit both
photochromism and photoswitchable NLO responses.[6]
Zinc(II) complexes that contain a pair of dithienylethene
moieties have also demonstrated photoswitchable activity
associated with a variation of the mb contrast by up to a
factor of 20.[7]
A large contrast of the first hyperpolarizability has also
been combined with other mechanisms of commutation, in
particular, those based on redox or protonation/deprotona-
tion reactions. Reversible redox-switching NLO behaviors
have been investigated in detail in RuII- and FeII-based chro-
mophores,[8] as well as in substituted helicenes[9] and in
stacks of tetrathiafulvalene.[10] In the case of commutations
triggered by the pH, 6s-cis-retinal and its 13-cis isomer,[11] as
well as pyridine-based octupolar systems[12] have shown
large b contrasts. Moreover, other studies have shown that
large NLO variations can be triggered by an electric field,
as in the case of a push–pull bisboronate chromophore,[13] or
simply by doping with different alkaline atoms (which is a
kind of redox reaction) as in the case of polyacetylene.[14] In
this last case, the switching property is the second hyperpo-
larizability (g); the third-order NLO response.
trast of about 10 is observed in the first hyperpolarizability
associated with the second harmonic generation (SHG) pro-
cess, making these compounds highly efficient switchable
frequency doublers. Another interesting property of these
systems lies in the fact that the reversible transformation
may be triggered either by light or pH variations. Indeed,
acid addition generates a protonated open form (POF), the
absorption spectrum of which is perfectly superimposable
on that of the zwitterionic form.
The efficiency of these compounds in terms of NLO con-
trast between the closed and open forms is currently under
optimization within an iterative approach that combines mo-
lecular synthesis, hyper-Rayleigh scattering (HRS) measure-
ments, and theoretical simulations. Structure–property rela-
tionships deduced from the latter constitute a guide for the
synthesis of new compounds, and optical characterizations,
in turn, highlight the qualities and deficiencies of the com-
putational methods. Optimal systems could then be compu-
tationally designed by performing structural modifications,
which aimed to improve the electron conjugation within the
open forms, and these systems could be used to plan further
syntheses. After investigating the impact of the styrylic resi-
due linked to the indolinic core, and selecting p-N,N-dime-
thylaminophenyl as the most efficient donor group, the
effect of replacing the indolinic unit by a benzACHTNUTRGNEiNUG midazolic or
a benzothiazolic residue was addressed, with a view to in-
crease the number of delocalizable electrons in the
system.[18] Very high b values were indeed obtained in the
POF with the benzothiazolic moiety, but they were accom-
panied by a loss of stability of the light-induced OF. Prelimi-
nary theoretical investigations also predicted that a signifi-
cant enhancement of the NLO response in the POF should
also be achieved by grafting electron-attracting substituents
onto the indolinic residue.[19] These two ways (substituents
and linkers) for designing more efficient switchable NLO
materials follow the general strategy that was established
several years ago to maximize the NLO responses of organic
Substituted diarylethenes have been found to exhibit pho-
toswitching-based contrasts of both b and g.[15] Applying a
magnetic field to induce spin transition is another means of
changing the NLO responses, but, to date, this has been
mostly used in the case of g compounds.[16]
As systems of interest in this context, several years ago
we presented a new family of organic molecular switches
based on the association of an indolinooxazolidine core with
Chem. Eur. J. 2009, 15, 2560 – 2571
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2561

F. Castet, B. Champagne et al.
1-(2-Hydroxyethyl)-5-bromo-2-[2-(4-dimethylaminophenyl)ethenyl]-3,3-
trimethylindoleninium iodide (2c): A mixture of 2b (2.56 g, 6.24 mmol)
and 4-dimethylaminobenzaldehyde (0.93 g, 6.24 mmol) was dissolved in
absolute EtOH (10 mL) under nitrogen. N-Methylmorpholine (0.7 mL,
6.24 mmol) was added to the solution under reflux over 60 min. After the
addition was complete, the reaction was maintained at reflux for 8 h
before the medium was allowed to cool, concentrated in vacuo, and the
residue was purified by column chromatography (SiO2; CH2Cl2/CH3OH,
9:1) to afford 2c as a purple solid (2.1 g, 62%).
molecules and leads to rationalizing b in terms of geometri-
cal parameters, charge distribution, excited-state properties,
donor/acceptor strengths, electronic versus vibrational con-
tributions, and effects of the surroundings. Note, however,
that the situation is more complex for switches because, as
discussed above, in addition to the NLO responses and con-
trasts, the switching characteristics need to be optimized.
This paper focuses on the substituent aspect and reports on
1) the synthesis of a series of compounds with an aldehyde,
bromine, or nitro as electron-attracting groups on the indo-
linic moiety, 2) the measurement of their linear and nonlin-
ear optical properties, and 3) the interpretation of these re-
sults in light of quantum chemical calculations.
Experimental and Computational Details
Synthesis: IndolinoACHTUNGTRENNUNG[2,1-b]oxazolidines are usually obtained by using
2,3,3-trialkylindolenines as starting materials. These compounds are then
quaternarized with 2-iodoethanol to afford 1-(2-hydroxyethyl)-2,3,3-trial-
kylindoleninium iodides, which are subsequently treated with aromatic
aldehydes in ethanol, and the process is completed by a basic treatment
with sodium hydroxide or N-methylmorpholine. This three-step synthetic
route to form the desired two-way targets has been previously described
for compound 1.[17]
10-[2-(4-Dimethylaminophenyl)ethenyl]-9,9-trimethyl-7-bromoindolino
[2,1-b]oxazolidine (2): A mixture of 2c (1.5 g, 2.8 mmol) and NaOH
(0.4 g, 10.0 mmol) was dissolved in CH2Cl2 (40 mL) and stirred at ambi-
ent temperature overnight. The solution was then concentrated in vacuo,
and the residue was washed with water (2ꢂ25 mL) and hexane (20 mL)
to afford 2 as a purple solid (0.9 g, 70%). 1H NMR (250 MHz, CDCl3):
d=7.33–7.16 (m, 4H; ArH), 6.79–6.64 (m, 4H; ArH, =CH-), 6.38 (d, J=
15.9 Hz, 1H; =CH-) 3.77–3.46 (m, 4H; -CH2-), 2.97 (s, 6H; -NCH3), 1.40
(s, 3H; -CH3), 1.15 ppm (s, 3H; -CH3); MS (EI) m/z (%): 413 (48), 411
(51), 383 (26), 381 (27), 254 (97), 252 (100); elemental analysis calcd (%)
for C22H25N2OBr: C 63.93, H 6.10; found: C 64.26, H 5.87.
5-Bromo-2,3,3-trimethylindolenine (2a): A mixture of 4-bromophenylhy-
drazine hydrochloride (1.0 g, 4.47 mmol) and 3-methyl-2-butanone
(0.6 mL, 5.61 mmol) was dissolved in glacial acetic acid (15 mL), and
then heated under reflux for 8 h under nitrogen. The solvent was evapo-
rated in vacuo. The residue was dissolved in CH2Cl2 (30 mL) and washed
with 10% aqueous Na2CO3 (2ꢂ30 mL), dried over Na2SO4, and the sol-
vent was evaporated to afford 2a as a brown oil (0.96 g, 90%). The prod-
uct was used in the next reaction without further purification. 1H NMR
(250 MHz, CDCl3): d=7.38 (m, 3H; ArH), 2.25 (s, 3H; -CH3), 1.28 ppm
(s, 6H; -CH3).
10-[2-(4-Dimethylaminophenyl)ethenyl]-9,9-trimethylindolino
zolidine-7-carboxaldehyde (3): nBuLi (2.5m) in hexane (0.60 mL,
1.50 mmol) was added dropwise to stirred solution of (0.50 g,
ACHTUNGTRENUN[NG 2,1-b]oxa-
a
2
1.21 mmol) in THF (15 mL) at ꢀ788C over 10 min. After a further
10 min dry DMF (0.12 mL, 1.55 mmol) was added swiftly, and the reac-
tion mixture was allowed to reach room temperature over 1 h. The reac-
tion continued for 30 min before water (1 mL) was added. Most of the
solvent was removed in vacuo and CH2Cl2 (15 mL) and water (10 mL)
were added. The aqueous phase was extracted by using CH2Cl2 (2ꢂ
10 mL) before the organic phase was collected, dried and concentrated to
afford compound 3 as brown solid (0.37 g, 83%). 1H NMR (250 MHz,
CDCl3): d=9.84 (s, 1H; -CHO), 7.69 (d, J=8.25 Hz, 1H; ArH), 7.63 (s,
1H; ArH), 7.35 (d, J=8.25 Hz, 1H; ArH), 6.87–6.68 (m, 4H, ArH; =
CH-), 6.00 (d, J=15.9 Hz, 1H; =CH-), 3.82–3.55 (m, 4H; -CH2-), 2.97 (s,
6H; -NCH3), 1.46 (s, 3H; -CH3), 1.17 ppm (s, 3H; -CH3); MS (EI) m/z
(%): 362 (31), 332 (26), 202 (100); elemental analysis calcd (%) for
C23H26N2O2: C 76.21, H 7.23; found: C 75.96, H 7.07.
1-(2-Hydroxyethyl)-5-bromo-2,3,3-trimethylindoleninium iodide (2b): A
mixture of 2a (1.5 g, 6.3 mmol) and 2-iodoethathol (0.74 mL, 9.5 mmol)
was dissolved in toluene (10 mL) and heated under reflux for 8 h under
nitrogen before the medium was allowed to cool and washed with Et2O
(3ꢂ10 mL). The precipitate was filtered and washed with Et2O/EtOH
(10:0.1 mL) to afford 2b as
a
brown solid (4.7 g, 91%). 1H NMR
(250 MHz, [D6]DMSO): d=8.19 (s, 1H; ArH), 7.93 (d, J=8.5 Hz, 1H;
ArH), 7.84 (d, J=8.6 Hz, 1H; ArH), 4.87 (s, 1H; -OH), 4.58 (t, J=
4.6 Hz, 2H; -NCH2-), 3.84 (t, J=4.6 Hz, 2H; -OCH2-), 2.81 (s, 3H;
-CH3), 1.56 ppm (s, 6H; -CH3).
5-Nitro-2,3,3-trimethylindolenine (4a): A mixture of 4-nitrophenylhydra-
zine (containing 30% H2O, 3.0 g, 13.7 mmol), 3-methyl-2-butanone
(1.91 mL, 17.8 mmol), and H2SO4 (2 mL) was dissolved in glacial acetic
acid (20 mL) and then heated under reflux for 8 h under nitrogen. After
it was cooled to room temperature, the solvent was evaporated in vacuo
before the residue was dissolved in CH2Cl2 (30 mL), washed with 10%
2562
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2560 – 2571

Molecular Switches
FULL PAPER
lecular symmetry, in which the mole-
cule lies in a mean (xz) plane (with z
as the two-fold symmetry axis), and
assuming Kleinman symmetry, two in-
dependent components bzxx and bzzz
remain and are determined by com-
aqueous Na2CO3 (2ꢂ30 mL), dried with Na2SO4, and the solvent was re-
moved in vacuo. The crude product was obtained as a black solid, which
was purified by column chromatography (SiO2; CH2Cl2/CH3OH, 60:1) to
afford 4a as a yellow solid (2.05 g, 73%). 1H NMR (250 MHz, CDCl3):
d=8.24 (d, J=8.3 Hz, 1H; ArH), 8.15 (s, 1H; ArH), 7.60 (d, J=8.3 Hz,
1H; ArH), 2.35 (s, 3H; -CH3), 1.37 ppm (s, 6H; -CH3).
bining measurements in HV (horizontally polarized incident light and
vertically polarized scattered light) and VV (vertically polarized incident
and scattered lights) configurations. More details about the procedure
can be found in reference [17].
1-(2-Hydroxyethyl)-5-nitro-2,3,3-trimethylindoleninium iodide (4b):
A
mixture of 4a (0.9 g, 4.41 mmol) and 2-iodoethanol (0.52 mL, 6.62 mmol)
was dissolved in toluene (8 mL) and heated under reflux for 8 h under ni-
trogen, before the medium was allowed to cool and was washed with
Et2O (3ꢂ10 mL). The precipitate was filtered and washed with Et2O/
EtOH (10:0.1 mL) to afford 4b as a black solid (0.98 g, 59%). 1H NMR
(250 MHz, CD3CN): d=8.60 (s, 1H; ArH), 8.49 (d, J=8.2 Hz, 1H;
ArH), 8.00 (d, J=8.5 Hz, 1H; ArH), 4.59 (s, 2H; -NCH2-), 4.03 (s, 2H;
-OCH2-), 2.88 (s, 3H; -CH3), 1.67 ppm (s, 6H; -CH3).
Calculations: The molecular structures were optimized in vacuo at the
density functional theory (DFT) level by using the B3LYP hybrid ex-
change-correlation functional and the 6-31G(d) basis set. Time-depen-
dent DFT (TDDFT) calculations were carried out at the same level of
theory
to
determine
the
vertical
excitation
energies
ꢀ
ꢁ
(DEge ¼ ꢀhwge ¼ ꢀh we ꢀ wg ) and the excited-state properties of the com-
pounds. This approach provides UV/Vis spectra in good qualitative
agreement with respect to experimental data for both excitation energies
and oscillator strength.[20] The time-dependent Hartree–Fock (TDHF)[21]
and the coupled-perturbed Hartree–Fock (CPHF) schemes, respectively,
were applied to obtain dynamic and static first hyperpolarizabilities by
using an incident wavelength of 1064 nm. Solvent effects were included
by using the polarizable continuum model within the integral equation
formalism (IEFPCM)[22] by taking e0 =36.640 (e =1.806) for acetonitrile,
1
and e0 =5.621 (e =2.320) for chlorobenzene. The cavity of the solute is
1
built up by using atomic radii from the universal force field (UFF) force
field by placing individual spheres around each heavy and hydrogen
atom. To account for correlation effects, the second-order Møller–Plesset
(MP2) method was employed in combination with a finite field (FF) pro-
cedure,[23] implying a Romberg scheme to improve the accuracy of the
numerical derivatives.[24] FF/MP2 calculations have already provided reli-
able first hyperpolarizabilities for organic chromophores.[17–18] Moreover,
in many cases the MP2 method recovers the largest part of the electron
correlation effects, as estimated from higher-order methods.[25] To account
for frequency dispersion at the MP2 level, the multiplicative correction
scheme was applied.[26] It consists of multiplying the static MP2 value by
the TDHF/CPHF ratio:
10-[2-(4-Dimethylaminophenyl)ethenyl]-9,9-trimethyl-7-nitroindolino-
ACHTUNGTRENNUNG[2,1-b]oxazolidine (4): A mixture of 4b (0.50 g, 1.33 mmol) and 4-dime-
thylaminobenzaldehyde (0.24 g, 1.61 mmol) was dissolved in absolute
EtOH (10 mL) under nitrogen. N-Methylmorpholine (0.16 mL,
1.46 mmol) was added over 60 min into the solution under reflux. After
the end of the addition, the solution was maintained at reflux for 8 h.
The medium was then allowed to cool and was concentrated in vacuo
before the residue was purified by column chromatography (SiO2;
CH2Cl2/CH3OH, 25:1) to afford
4 as a yellow solid (0.34 g, 68%).
1H NMR (250 MHz, CDCl3): d=8.13 (d, J=7.6 Hz, 1H; ArH), 7.96 (s,
1H; ArH), 7.35 (d, J=8.3 Hz, 2H; ArH), 6.80–6.69 (m, 4H, ArH; =CH-),
5.98 (d, J=15.9 Hz, 1H; =CH-), 3.83–3.55 (m, 4H; -CH2-), 2.98 (s, 6H;
-NCH3), 1.47 (s, 3H; -CH3), 1.19 ppm (s, 3H; -CH3); 13C NMR
(300 MHz, CDCl3): d=157.3, 150.6, 142.6, 141.1, 133.0, 127.8, 125.1,
124.2, 119.1, 118.9, 112.3, 111.1, 110.3, 63.5, 49.4, 47.4, 40.4, 28.1, 20.2
ppm; MS (EI) m/z (%): 379 (22), 349 (16), 219 (100); elemental analysis
calcd (%) for C22H25N3O3: C 69.64, H 6.64; found: C 69.98, H 6.27.
bTDHFðꢀ2w; w; wÞ
ð1Þ
bMP2ðꢀ2w; w; wÞ ꢁ bMP2ð0; 0; 0Þ ꢂ
bCPHFð0; 0; 0Þ
DFT methods were not used to compute b because conventional DFT
methods have demonstrated serious drawbacks.[27] Improved approaches,
such as optimized effective potential for exact exchange (OEP-EXX)
and long range (LR)-DFT have recently appeared, but still need to be
optimized.[28] For example, we have recently compared the NLO respons-
es of anil derivatives calculated by using DFT (employing the B3LYP ex-
change-correlation functional) to those determined by using the CPHF
and FF/MP2 methods.[2e] It appears that FF Kohn–Sham calculations pro-
vide b values that are sometimes in good agreement with the MP2 re-
sults, sometimes close to the CPHF values, and also sometimes far away
from both CPHF and MP2 results. Moreover, the solvent effects on the
HRS experiments: HRS experiments were performed on dilute solutions
(between 10ꢀ4 and 10ꢀ6 m) in acetonitrile or chlorobenzene, which corre-
spond to typical concentrations used to check the linear dependency of
absorbance and to determine the extinction coefficient e. In this NLO
scattering technique, the intensity of the incoherent scattered light at the
second harmonic frequency of an IR Nd:YAG pulsed laser (nanosecond
regime) is used to determine the first hyperpolarizability b. The scattered
harmonic light is related to quadratic products between components of
the molecular b tensor that correspond to isotropic averaging over the
molecular motions (noninteracting molecules). Assuming pseudo-C2v mo-
first hyperpolarizabilities are determined by inserting bTDHF
bMP2(0;0,0) quantities, which are evaluated within the IEFPCM method,
into Equation (1).
(ꢀ2w;w,w) or
AHCTUNGTRENNUNG
Chem. Eur. J. 2009, 15, 2560 – 2571
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2563

F. Castet, B. Champagne et al.
Comparisons between experimental measurements and theoretical esti-
mates are made by considering the square root of the HRS intensity for
plane-polarized incident light observed perpendicularly to the propaga-
tion direction, as well as the associated depolarization ratios, as shown in
Equation (2):
Table 2. Absorption maxima (lge, nm), transition energies (DEge), oscilla-
tor strength (fge), and electronic transitions implied in the dominant low-
energy charge-transfer states calculated at the PCM/B3LYP/6-31G(d)
level.
lge
DEge [eV]
fge
lge
DEge [eV]
fge
ꢂ
ꢂ
ꢃ
ꢃ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
CF in acetonitrile
CF in chlorobenzene
b2ZZZ
b2ZXX
2w
VV
2w
HV
Èꢂ
ꢃ
ꢂ
ꢃÉ
I
I
bHRSðꢀ2w; w; wÞ ¼
b2ZZZ
þ
b2ZXX
DR ¼
¼
ð2Þ
1 (R=H)
304
280
305
280
308
294
280
354
306
280
4.08
4.43
4.06
4.42
4.02
4.22
4.44
3.50
4.05
4.43
0.923
0.203
0.988
0.194
1.337
0.152
0.119
0.583
0.747
0.159
304
280
306
280
309
294
279
351
307
280
4.07
4.43
4.05
4.42
4.01
4.22
4.44
3.53
4.03
4.43
0.970
0.206
1.035
0.152
1.384
0.169
0.099
0.617
0.776
0.152
ꢂ
ꢃ
ꢂ
ꢃ
2 (R=Br)
3 (R=CHO)
in which b2ZZZ and bX2 ZZ correspond to orientational averages of the b
tensor and are calculated without assuming Kleinmanꢃs conditions (full
expressions of these terms can be found in reference [29]). All calcula-
tions were performed by using Gaussian 03,[30] as well as homemade
codes to carry out the FF–Romberg differentiation scheme.
4 (R=NO2)
POF in acetonitrile
POF in chlorobenzene
Results and Discussion
1 (R=H)
2 (R=Br)
3 (R=CHO)
4 (R=NO2)
474
481
486
503
2.62
2.58
2.55
2.46
1.523
1.625
1.675
1.500
483
489
492
505
2.57
2.54
2.52
2.46
1.568
1.670
1.723
1.643
Linear optical properties: The linear optical properties of
the compounds are given by UV-visible absorption spectros-
copy. As already mentioned, the switching mechanism is
two-way for all compounds: the absorption spectra arising
from UV irradiation and acidic addition are totally superim-
posable, indicating that photo-induced and acido-generated
colored forms adopt similar structures. Total back photo-
bleaching could be achieved by irradiating the zwitterionic
OFs with an appropriate visible light, or upon base addition
to the POFs. The experimental data discussed in this section
were obtained by considering the pH-induced switching pro-
cess. The absorption properties of the POF molecules were
obtained by addition of a large excess of acetic acid to CF
molecules. Experimental and theoretical results for com-
pounds 1–4 are gathered in Tables 1 and 2, respectively. For
Table 1. Absorption maxima (lmax, nm), transition energies (DEge) and
molar extinction coefficient (e, Lmolꢀ1 cmꢀ1) of substituted indolinooxa-
zolidine derivatives in their CFs and POFs measured in acetonitrile and
chlorobenzene.
lmax DEge [eV] e (lmax
)
lmax DEge [eV] e (lmax)
CF in acetonitrile
CF in chlorobenzene
1 (R=H)
2 (R=Br)
3 (R=CHO) 310
298
300
4.16
4.13
4.00
4.10
3.73
27000
30000
30000
23000
20000
302
306
312
306
336
4.10
4.05
3.97
4.05
3.69
18000
23000
32000
12500
11000
4 (R=NO2)
302
332
POF in acetonitrile
POF in chlorobenzene
1 (R=H)
2 (R=Br)
3 (R=CHO) 548
4 (R=NO2) 582
544
546
2.28
2.27
2.26
2.13
93000
554
2.24
2.20
2.21
2.09
56500
74000
80000
42500
130000 564
96000 562
106000 592
all compounds, a good agreement between the experimental
and simulated excitation spectra is obtained, as illustrated in
Figure 1 for compounds 1 and 4 in acetonitrile. Indeed, de-
spite the systematic blueshift observed in the simulated
spectra of the POF, the relative position of the low-energy
charge-transfer bands is well reproduced within the sets of
substituted CFs and POFs, as well as their relative intensi-
ties in the POF and closed forms.
Figure 1. Absorption spectra of compounds 1 (top, R=H) and 4 (bottom,
R=NO2) in their CFs (c) and POFs (c) measured in acetonitrile.
The simulated spectra (dotted lines) have been calculated at the
IEFPCM/B3LYP/6-31G(d) level and each transition has been enlarged
by using a Gaussian function that has a full width at half maximum
(FWHM) value of 0.3 eV for the POF and 0.5 eV for the CF. The intensi-
ties at the absorption maxima in the experimental and simulated spectra
were adjusted to coincide in the POF.
2564
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2560 – 2571

Molecular Switches
FULL PAPER
In all POFs, the most absorbing charge-transfer excited
As shown in Table 1, the magnitude of the experimental
absorption intensity of the CF compounds is weakened
when going from acetonitrile to chlorobenzene, except for
compound 3, which has the aldehyde group, for which the
molar extinction coefficients are similar in the two solvents.
This effect seems to be generalized for all POFs because in
chlorobenzene the magnitude of the absorption is always
smaller than in acetonitrile. This result is not reproduced by
the calculations, which predict oscillator strengths in chloro-
benzene in the same range as, or slightly higher than, those
in acetonitrile. Interactions between chromophore molecules
that lead to the formation of J aggregates, or between the
chromophore and chlorobenzene, can explain such differen-
ces between theory and experiment.
*
state is dominated by a p ! p transition between the high-
est occupied molecular orbital (HOMO) and the lowest un-
occupied molecular orbital (LUMO), with these two MOs
being delocalized over the whole molecule (Figure 2). The
Nonlinear optical properties
Experimental results: Table 3 reports the longitudinal (bzzz),
transverse (bzxx), and hyper-Rayleigh (bHRS) hyperpolariza-
bilities, as well as the depolarization ratios (DR) for com-
Table 3. Longitudinal (b1zz0z64), transverse (b1zx06x4), and HRS (bHRS) hyperpo-
larizabilities, depolarization ratios (DRs); and normalized efficiencies (t,
in parentheses), obtained from HRS measurements at l=1064 nm. All b
values are given in atomic units (1 au=3.62 10ꢀ42 m4 Vꢀ1 =3.2063ꢂ
10ꢀ53 C3 m3 Jꢀ2 =8.641ꢂ10ꢀ33 esu).
Figure 2. Molecular orbitals of compound 4 (R=NO2) in its closed and
protonated open forms involved in the main UV-visible electronic transi-
tions, as calculated at the PCM/B3LYP/6-31G(d) level in acetonitrile.
1064
zzz
1064
zxx
DR
jb
j
b1z0z6z4/b
bHRS
CF in acetonitrile
ꢀ0.096
1 (R=H)
2 (R=Br)
3 (R=CHO)
4 (R=NO2)
4.1
4.0
3.8
4.4
6000
4700
16800
12700
2200
2300
5600
5100
+0.109
ꢀ0.136
maximum absorption wavelengths are significantly impacted
by the nature of the substituent, and range in acetonitrile
from 544 nm for the nonsubstituted compound 1 to 582 nm
for compound 4, which includes the most electron-withdraw-
ing group. When changing the solvent to chlorobenzene, the
maximum absorption band is redshifted by 10–18 nm for all
compounds. Although slightly underestimated, this redshift
is well reproduced by the TDDFT calculations.
In their CFs, the four compounds have absorption
maxima at about 4 eV and the associated oscillator strengths
depend on the substituent R, especially in chlorobenzene.
The spectrum of compound 4 displays a characteristic
second band of similar intensity at 3.7 eV, which is well re-
produced by the calculations. The lowest-energy absorption
band is dominated by a [HOMOꢀ1!LUMO] transition (lo-
calized on the oxazolidine moiety), whereas the peak at
4 eV is mainly associated with [HOMO!LUMO+1] (local-
ized on the dimethylaminophenyl part, as shown in
Figure 2). In compound 3 (R=CHO), the absorption maxi-
mum at 4 eV is also associated with a [HOMO!LUMO+
1] transition, whereas this band is dominated by the
[HOMO!LUMO] transition in compounds 1 (R=H) and
2 (R=Br). For compounds 1–3, the MOs involved in the
transition at 4 eV are fully localized on the dimethylamino-
phenyl moiety. Besides, the impact of the solvent is smaller
than for the OFs, with the redshift of the main band being
observed between 2–6 nm.
ꢀ0.069
POF in acetonitrile
ꢀ0.110
1 (R=H)
2 (R=Br)
3 (R=CHO)
4 (R=NO2)
4.0
4.0
3.6
3.7
117800
175000
84000
41600 (94.8)
60500 (96.2)
26600 (79.1)
35700 (85.7)
ꢀ0.110
ꢀ0.155
110000
ꢀ0.142
POF in chlorobenzene
1 (R=H)
2 (R=Br)
3 (R=CHO)
4 (R=NO2)
4.5
4.2
4.1
3.8
ꢀ50000
ꢀ0.056
ꢀ0.085
ꢀ0.093
ꢀ0.132
19100
41600
17700
26500
ꢀ114600
ꢀ49000
ꢀ67000
pounds 1–4 issued from HRS measurements at 1064 nm.
The normalized efficiency of the HRS hyperpolarizability,
defined as t ¼ ½bHRSðPOFÞ ꢀ bHRSðCFÞꢃ=bHRSðPOFÞ ꢂ 100,
differentiates between the POF and the CF, and is also re-
ported for measurements performed in acetonitrile.
Measurements performed in acetonitrile reveal that the
first hyperpolarizabilities of the POFs are one order of mag-
nitude larger than those of the CFs, giving rise to very high
contrasts of the NLO responses under the switching reac-
tion. Indeed, the normalized efficiency of the HRS hyperpo-
larizability varies from 79.1 to 96.2%, demonstrating that
these compounds act as highly powerful two-way NLO
switches. Although it was not possible to measure the re-
sponses for the CFs in chlorobenzene, owing to signals
Chem. Eur. J. 2009, 15, 2560 – 2571
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2565

F. Castet, B. Champagne et al.
Table 4. Static longitudinal first hyperpolarizability (bzzz), HRS first hy-
perpolarizability (bHRS), and DRs of POFs evaluated by using the CPHF
scheme. All values are given in atomic units. Values relative to the non-
substituted compound are given in parentheses.
lower than the detection limit of our experimental setup, the
very high values obtained for the POF demonstrate that the
commutation mechanism is also accompanied by a large
contrast of the second-order NLO property in this solvent.
The HRS responses of the CF in acetonitrile of compounds
1 and 2 are two to three times smaller than those of com-
pounds 3 and 4, in agreement with the relative magnitude of
their maximum absorption wavelengths. Besides, it has been
noticed that solvent effects induce changes in the relative or-
dering of the bHRS values of the POF, which is CHO!H!
NO2!Br in chlorobenzene and CHO!NO2!H!Br in
acetonitrile. Also note that the bHRS values obtained in ace-
tonitrile for the POF of compounds 1–3 (and to a lesser
extent of compound 4) were measured under resonant con-
ditions. Hence, the NLO responses for the first three com-
pounds are likely to be enhanced because the two-photon
scattering frequency lies within their main absorption band.
On the contrary, the measurements for compound 4 were
also made out of resonance, so that its large hyperpolariza-
bility should result mainly from the presence of the nitro
substituent, ensuring a better charge transfer along the mol-
ecule. The DR values are all close to four and do not exhibit
a specific dependency on the nature of the solvent or on the
open/closed form of the chromophore. It can, however, be
observed that the DR of the NO2-substituted compound is
slightly smaller (larger) in its POF (CF).
1 (R=H)
2 (R=Br)
3 (R=CHO)
4 (R=NO2)
in the gas phase
CPHF/6-31G*
bzzz
5384 (1.000)
2192 (1.000)
4.30
5697 (1.058)
2339 (1.067)
4.33
6308 (1.172)
2615 (1.193)
4.36
6318 (1.173)
2614 (1.193)
4.29
bHRS
DR
CPHF/6-31+G*
bzzz
5755 (1.000)
2401 (1.000)
4.45
5922 (1.029)
2438 (1.015)
4.40
6524 (1.134)
2707 (1.127)
4.40
6827 (1.186)
2834 (1.180)
4.37
bHRS
DR
in acetonitrile
CPHF/6-31G*
bzzz
12736 (1.000)
5173 (1.000)
4.25
13673 (1.074)
5598 (1.082)
4.29
15591 (1.224)
6453 (1.247)
4.34
17305 (1.359)
7188 (1.390)
4.36
bHRS
DR
CPHF/6-31+G*
bzzz
13522 (1.000)
5508 (1.000)
4.31
14278 (1.056)
5854 (1.063)
4.32
16478 (1.219)
6827 (1.239)
4.36
18437 (1.363)
7670 (1.393)
4.38
bHRS
DR
in chlorobenzene
CPHF/6-31G*
bzzz
10841 (1.000)
4493 (1.000)
4.32
11601 (1.070)
4753 (1.058)
4.33
13133 (1.211)
5436 (1.210)
4.37
14336 (1.322)
5950 (1.324)
4.38
bHRS
DR
CPHF/6-31+G*
bzzz
11426 (1.000)
4750 (1.000)
4.40
12100 (1.059)
4965 (1.045)
4.37
13812 (1.209)
5722 (1.205)
4.40
15167 (1.327)
6304 (1.327)
4.40
Theoretical results: In the first paper of this series,[17] quan-
tum chemical calculations were carried out on the nonsub-
stituted compound 1 both in its CF and POFs. The changes
in terms of geometrical and electronic structures, as well as
in terms of (non) linear optical properties during the switch-
ing reaction, were extensively discussed and the reader is re-
ferred to this previous article for further details. In the pres-
ent work, we focused our computational efforts on the POF
rather than on the CF, with the aim of rationalizing the sig-
nificant impact of the substituent R, as well as solvent and
frequency dispersion effects on the NLO responses of the
optically active state. Static and dynamic hyperpolarizabili-
ties were first calculated at the CPHF and TDHF levels of
calculation. Tables 4 and 5 report the results obtained for
molecules in the gas phase, as well as when including the
solvent effects through the PCM scheme. At the CPHF/6-
31G* level in the gas phase, the static bHRS values increase
with the acceptor strength of R in the order H!Br!
CHO!NO2, whereas the DR values are similar for all com-
pounds and amount to 4.35ꢄ0.10, accounting for the pre-
dominance of one component in the b tensor (i.e., charge-
transfer character). Adding diffuse Gaussian functions in
the basis set increases bHRS by less than 10% and does not
change the relative ordering of the values. The enhancement
of bHRS due to frequency dispersion effects, calculated in the
gas phase by comparing the CPHF and TDHF values, is sim-
ilar for all compounds and ranges between 95 and 110%.
bHRS
DR
0.2–0.3 for all compounds up to 4.68ꢄ0.07, which is even
closer to the asymptotic limit of a purely 1D character than
in the static limit. When accounting for solvent effects,
strong increases (103–111% with 6-31G* and 129–171%
with 6-31+G*) are observed for the static bHRS values,
whereas the enhancement of the dynamic bHRS is less pro-
nounced (about 50% less than for the static b values),
owing to the reduction of the dielectric constant of acetoni-
trile at 1064 nm.
Summarizing these results, it can be observed that bHRS in-
creases with the acceptor strength of R, and follows a quasi-
linear relationship with respect to the Hammett constants of
the substituents R. As shown in Figure 3, the solvent has a
direct effect on the slope, revealing higher solute–solvent in-
teractions for compounds with strong electron-withdrawing
substituents. On the contrary, frequency dispersion effects
induce a global shift towards higher values but have a negli-
gible impact on the slope.
To evaluate the impact of electron correlation on the
static hyperpolarizabilities, FF MP2/6-31G* calculations
were carried out, both in the gas phase and in solution
(Table 6). The latter results were obtained by calculating the
third-order derivatives of the IEFPCM/MP2/6-31G* ener-
gies with respect to the applied field. For molecules in the
gas phase, electron correlation effects lead to an increase of
the HRS first hyperpolarizabilities by 81–88%, except for
However, for molecules in the gas phase, bHRS
slightly larger than bHRS(R=NO2) at the TDHF level with
both basis sets. The DR values increase by approximately
ACHTUNGTRENN(UNG R=CHO) is
AHCTUNGTRENNUNG
2566
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2560 – 2571

Molecular Switches
FULL PAPER
Table 5. Dynamic (SHG, l=1064 nm) longitudinal first hyperpolarizabil-
ity (bzzz), HRS first hyperpolarizability (bHRS), and DRs of POFs evaluat-
ed by using the TDHF scheme. All values are given in atomic units.
Values relative to the nonsubstituted compound are given in parentheses.
Table 6. Static longitudinal first hyperpolarizability (bzzz), HRS first hy-
perpolarizability (bHRS), and DRs of POFs evaluated at the MP2/6-31G*
level. All values are given in atomic units. Values relative to the nonsub-
stituted compound are given in parentheses.
1 (R=H)
2 (R=Br)
3 (R=CHO)
4 (R=NO2)
1 (R=H)
2 (R=Br)
3 (R=CHO)
4 (R=NO2)
in the gas phase
in the gas phase
TDHF/6-31G*
bzzz
bHRS
DR
10070 (1.000)
4112 (1.000)
4.68
10376 (1.030)
4257 (1.035)
4.68
11467 (1.139)
4730 (1.150)
4.71
10222 (1.015)
4208 (1.023)
4.59
bzzz
10926 (1.000)
4437 (1.000)
4.61
11986 (1.097)
4909 (1.106)
4.64
13289 (1.216)
5477 (1.234)
4.66
13186 (1.207)
5420 (1.222)
4.61
bHRS
DR
TDHF/6-31+G*
in acetonitrile
bzzz
11489 (1.000)
4678 (1.000)
4.66
12539 (1.091)
5144 (1.100)
4.69
13910 (1.211)
5737 (1.226)
4.69
13837 (1.204)
5688 (1.216)
4.63
bzzz
bHRS
DR
30203 (1.000)
12335 (1.000)
4.69
32557 (1.078)
13372 (1.084)
4.71
36893 (1.221)
15207 (1.233)
4.73
36535 (1.210)
15063 (1.221)
4.65
bHRS
DR
in acetonitrile
in chlorobenzene
TDHF/6-31G*
bzzz
bHRS
DR
24276 (1.000)
9918 (1.000)
4.70
26065 (1.074)
10710 (1.080)
4.73
29304 (1.207)
12083 (1.218)
4.74
27994 (1.153)
11540 (1.163)
4.65
bzzz
16027 (1.000)
18032 (1.125)
7380 (1.134)
4.68
20719 (1.293)
8539 (1.312)
4.70
22215 (1.386)
9162 (1.408)
4.70
bHRS
6507 (1.000)
4.64
DR
TDHF/6-31+G*
bzzz
17120 (1.000)
6967 (1.000)
4.69
18991 (1.109)
7783 (1.117)
4.71
22128 (1.293)
9125 (1.310)
4.73
24057 (1.405)
9931 (1.425)
4.72
the DR value for this compound is slightly smaller than for
the three other compounds.
bHRS
DR
When accounting for solvent effects (acetonitrile), the
bHRS responses are enhanced by 200, 214, 222, and 258% for
compounds 1 to 4, respectively, again indicating that solvent
effects are larger if R is a strong acceptor group. The same
tendency is observed, although weaker, in chlorobenzene;
therefore, the bHRS ordering becomes H!Br!NO2!CHO
when accounting for both solvent and electron correlation
effects.
in chlorobenzene
TDHF/6-31G*
bzzz
19604 (1.000)
21459 (1.099)
8781 (1.087)
4.69
24484 (1.249)
10080 (1.248)
4.71
27227 (1.389)
11236 (1.390)
4.73
bHRS
8080 (1.000)
4.68
DR
TDHF/6-31+G*
bzzz
20591 (1.000)
22631 (1.099)
9270 (1.074)
4.72
26145 (1.270)
10770 (1.248)
4.74
29301 (1.423)
12101 (1.402)
4.75
bHRS
DR
8633 (1.000)
4.72
For computational reasons, it is currently not possible to
calculate bHRS responses, which simultaneously account for
frequency dispersion, solvent, and MP2 electron correlation
effects. Nevertheless, approximate values were estimated by
using adaptations of Equation (1) to properties in solutions
[Eqs. (3) and (4)].
Solv
b
Solv
MP2
TDHF
Vacuo
ð3Þ
ð4Þ
b
ðꢀ2w; w; wÞ ¼ bMP2ð0; 0; 0Þ ꢂ
b
CPHF
Solv
b
Solv
MP2
Solv
MP2
TDHF
Solv
b
ðꢀ2w; w; wÞ ¼ b ð0; 0; 0Þ ꢂ
b
CPHF
In Equation (3), the static bHRS response obtained at the
MP2 level for the molecule in the gas phase is multiplied by
a correction factor by taking into account the impact of the
solvent and frequency dispersion. In Equation (4), the refer-
ence bMP2 response includes both correlation and solvent ef-
fects, whereas the multiplicative ratio provides a correction
accounting for frequency dispersion effects (evaluated from
b values determined by taking into account the solvent ef-
fects). The bHRS values estimated within the two schemes are
reported in Table 7. Using Equation (4) gives rise to larger
values, but the two sets of results follow very similar trends.
Within this best theoretical level of approximation, the bHRS
values of compounds 1–4 follow the ordering H!Br!
NO2!CHO in both solvents.
Figure 3. Evolution of the HRS first hyperpolarizability as a function of
the Hammett constant of the substituents. *: CPHF/6-31+G* in the gas
^
phase, &: CPHF/6-31+G* in acetonitrile, : TDHF/6-31+G* in the gas
phase, and ~: TDHF/6-31+G* in acetonitrile.
compound 4 (R=NO2) in which the increase is smaller
(61%). This results in a change of the bHRS value ordering
with respect to the CPHF scheme: H!NO2!Br!CHO.
The DR values also increase when including electron corre-
lation, but again to a lesser extent for compound 4, so that
Chem. Eur. J. 2009, 15, 2560 – 2571
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2567

F. Castet, B. Champagne et al.
Table 7. Dynamic (SHG, l=1064 nm) longitudinal first hyperpolarizabil-
ity (bzzz) and HRS first hyperpolarizability (bHRS) of POFs evaluated
from Equations (3) and (4). All values are given in atomic units. Values
relative to the nonsubstituted compound are given in parentheses.
1 (R=H)
2 (R=Br)
3 (R=CHO) 4 (R=NO2)
in acetonitrile
Equation (3) 11932 (1.00) 13590 (1.14) 15945 (1.34) 14746 (1.24)
Equation (4) 15516 (1.00) 17439 (1.12) 20119 (1.30) 19797 (1.28)
in chlorobenzene
Equation (3) 14785 (1.00) 16186 (1.09) 18819 (1.27) 17968 (1.22)
Equation (4) 17510 (1.00) 19787 (1.13) 22410 (1.28) 21367 (1.22)
Discussion: Despite accounting for frequency dispersion,
solvent, and electron correlation effects, the bHRS (or bzzz
)
values predicted for the POFs by ab initio calculations do
not follow the same ordering as the measured ones. From
the former (Table 7), one obtains H!Br!NO2!CHO in
both solvents, whereas the experimental b values follow dif-
ferent ordering depending on the solvent: CHO!H!
NO2!Br in chlorobenzene, and CHO!NO2!H!Br in
acetonitrile (Table 5). In addition, none of these experimen-
tal results follow the ordering of the Hammett parameters.
It is therefore important, within our aim of designing new
NLO switchable compounds, to point out possible origins of
this difference and to see how much this divergence is ap-
parent and results only from the differences in probing tech-
niques. First, as mentioned above, the experimental data are
impacted by resonance conditions because the measure-
ments are carried out at a frequency within the absorption
band, slightly above the maximum. This is, however, not
true in the case of the theoretical values because the poles
of the TDHF responses are underestimated. However, eval-
uating the exact magnitude of this effect is not straightfor-
ward.[31] The simplest way to estimate the impact of frequen-
cy dispersion in pseudodipolar chromophores is to consider
the two-states approximation,[32] which assumes that only
one excited state contributes to the second-order NLO re-
sponse. Considering a homogeneous damping g and neglect-
ing the nonresonant terms,[18b,33] the frequency dispersion
factor for the longitudinal component of the first hyperpo-
Figure 4. Frequency dispersion factor as a function of the maximum ab-
sorption wavelength of the compounds within the two-state approxima-
tion and different homogeneous damping values. The squares indicate
the wavelength positions for compounds 1–4 in acetonitrile. g=0 (c),
1000 (b), and 2000 cmꢀ1 (a).
approximately 21 to 30 for the three other compounds. The
resulting static bzzz responses (abbreviated as b0) are then
3900, 7100, 3900, and 14900 a.u. for compounds 1–4, respec-
tively, and thus follow the ordering HꢁCHO!Br!NO2.
Owing to the global redshift of the absorption maximum
(Tables 1 and 2), frequency dispersion effects are slightly re-
duced in chlorobenzene and, consequently, the calculated b0
values are quite different from the values obtained in aceto-
nitrile (3100, 10200, 4000 and 10800 a.u. for compounds 1–
4, respectively), but they follow the same hierarchy as in
acetonitrile. Though extrapolating to the static limit im-
proves the correspondence between theory and experiment,
differences remain. Indeed, when accounting for both sol-
vent (acetonitrile in this case) and electron correlation, the-
oretical calculations predict that the b0 values of the H-, Br-,
CHO-, and NO2-substituted compounds are in the ratio of
1.00:1.08:1.22:1.21, whereas the corresponding experimental
values extrapolated to infinite wavelength are in the ratio of
1.00:1.82:1.00:3.82.
larizability tensor F
G
This first set of extrapolated b0 values can be improved in
several ways, including 1) by considering a finite value for
the homogeneous damping associated with the optical tran-
sition [Eq. (5)],[34] 2) by incorporating an inhomogeneous
broadening based on the absorption spectrum, which implic-
itly contains information on the distribution of the transition
frequencies, 3) by taking into account the vibronic structure
of the excited states, and 4) by including higher-energy excit-
ed states in the sum-over-state expression of b. In this work,
to provide a better estimate of b0, we have employed three
additional approximations, which should also guide us in as-
sessing their accuracy.
bzzzðꢀ2w;w,wÞ
bzzzð0;0,0Þ
Fðw,wge,gÞ¼
ð5Þ
¼
ð½wgeꢀ igꢃ ꢀ4w2Þð½wgeꢀigꢃ ꢀ w2Þ
in which g and e are the two electronic (ground and excited)
states contributing to the first hyperpolarizability, and ꢀhwge
is the corresponding excitation energy. This dispersion
factor, reported in Figure 4 as a function of the positions of
the main absorption band, varies substantially with both wge
and g. In the absence of homogeneous damping (g=0)
when the solvent is acetonitrile the value of the dispersion
factor is about 7.4 for compound 4, whereas it ranges from
The first approach makes use of Equation (5). However,
the difficulty arises from the fact that the value of g is a
priori not known and it depends on the compound, the ex-
cited states, and on the solvent. In line with previous stud-
2568
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2560 – 2571

Molecular Switches
FULL PAPER
ACHTUNGTRENNUNG
ies,[33a,35] the g value was chosen to correspond to approxi-
mately 1.2 times the half width at half maximum (HWHM)
of the corresponding main absorption band, that is, it corre-
AHCTUNGTRENNUNG
[(w–wge)/g]2
sponds to the HWHM of the eꢀ
Gaussian function.
The g values estimated from the absorption spectra of com-
pounds 1–4 in acetonitrile amount to 0.193, 0.178, 0.208, and
0.149 eV, respectively. Using these results, the extrapolated
b0 values are now in the ratio of 1.00:1.41:0.79:1.23. By
keeping the same g values for results measured in chloro-
benzene, the extrapolated b0 values are in the ratio of
1.00:2.39:1.11:1.85.
An inhomogeneous broadening is employed in the second
approach. According to Campo et al,[31] for an incoherent
process like HRS, the dispersion factor is given by Equa-
tion (6):
ꢄ
ꢄ
ꢄ
ꢄ
ꢄ
ꢄ
ꢄ
ꢄ
ꢇ
ꢈ
Z
ꢄ
ꢄ
ꢄ
ꢄ
HRS
ꢅ
ꢆ
ꢅ
ꢆ
1=2
2
b
ðꢀ2w; w; wÞ
Figure 5. Evolution of the dispersion factor with the second harmonic
energy relative to the absorption spectrum. These were obtained for com-
pound 4 by using the three approaches and parameters described in the
text.
¼
N w0ge F w; w0 ; g dw0
ꢄ
ꢄ
ge
ge
bHRSð0; 0; 0Þ
ð6Þ
Here, a value of g=100 cmꢀ1 was chosen for the homoge-
neous damping, whereas N(w’ge), the normalized distribution
of the transition frequencies, is approximated by a Gaussian
function that reproduces the experimental HWHM of the
corresponding main absorption band. By applying Equa-
tion (6), the extrapolated b0 values of compounds 1–4 in ace-
tonitrile are in the ratio of 1.00:1.44:0.75:1.94.
In the last approach employed in this study, the distribu-
tion of the transition frequencies is improved by considering
the single-mode vibronic model so that NACHTNUGTRNEG(UN w’ge) of Equa-
the chromophore in its POF, whereas p–p or dipole–dipole
interactions between chlorobenzene and the chromophore
could favor J-aggregates. Such solute/solvent interactions
should affect the electron density and symmetry of the chro-
mophores and consequently their NLO responses. The
impact of such effects is underestimated when using a con-
tinuum solvation model. Including explicit solvent molecules
in the calculation by combining, for example, molecular dy-
namics and quantum chemistry tools, would possibly be a
good starting point to further improve the quality of the the-
oretical investigations. Additionally, we should not forget
that the experiments are performed in the nanosecond
regime, which affects the medium quite a lot. In particular,
the input laser beam induces collisions between molecules
that reduce the lifetime of the excited states. The use of a ps
or fs laser source should reduce these collective effects and
make the comparison between theory and experiments
easier.
AHCTUNGTRENNUNG
tion (6) is replaced by Equation (7):
ꢉ
ꢊ
vib: levels
ꢅ
ꢆ
SneꢀS
n!
X
2
w0 ꢀw =g
N w0ge
¼
eꢀ
½ð
nÞ vibꢃ
ge
ð7Þ
n
in which wn =wge +nwvib corresponds to the transition from
the ground state to the nth vibrational level of the excited
state. A Huang–Rhys factor (S) of 0.4 was adopted because
it reproduces the shape of the absorption band. Values of
wvib and gvib were also chosen to best reproduce the absorp-
tion spectra.[36] Because the number of parameters is already
large, no attempts were made to include additional vibra-
tional normal modes or excited states. The corresponding
extrapolated b0 values of compounds 1–4 in acetonitrile then
become in the ratio of 1.00:1.50:0.80:1.63. These last two
models are the most elaborate and, with the exception of
R=CHO, reproduce the theoretical ordering for the b0 am-
plitudes. In addition, as shown in Figure 5 for R=NO2, they
provide similar dispersion factors as a function of w, whereas
in the model including only homogeneous broadening, the
resonance enhancement of b is about a factor of five small-
er.
Conclusion
This paper is the last in a series aiming to optimize the NLO
contrast of multiple-way indolinooxazolidine-based molecu-
lar switches. Following a theoretical study that predicted a
significant NLO enhancement upon addition of electron-
withdrawing substituents in the para position on the indolin-
ic residue, we have reported herein on the synthesis and
NLO characterization of three new indolinooxazolidine de-
rivatives. In addition, the experimental results have been ra-
tionalized by means of ab initio calculations, including fre-
quency dispersion, solvent, and electron correlation effects.
Like the nonsubstituted compound (R=H), these three
new systems (R=Br, CHO, NO2) can commute reversibly,
upon pH variation or UV-light irradiation, between the
closed indolinooxazolidine form and the open p-conjugated
As mentioned in the Linear Optical Properties Section,
specific interactions between the chromophore and the sol-
vent might explain the remaining discrepancies. These in-
clude favorable electrostatic interactions between acetoni-
trile molecules and the positively charged nitrogen atom of
Chem. Eur. J. 2009, 15, 2560 – 2571
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2569

F. Castet, B. Champagne et al.
[4] M. Giraud, A. Lꢀaustic, R. Guillot, P. Yu, P. G. Lacroix, K. Nakata-
ni, R. Pansu, F. Maurel, J. Mater. Chem. 2007, 17, 4414–4425.
[5] A. Plaquet, M. Guillaume, B. Champagne, F. Castet, L. Ducasse, J.-
L. Pozzo, V. Rodriguez Phys. Chem. Chem. Phys. 2008, 10, 6223.
[6] Y. Atassi, J. A. Delaire, K. Nakatani, J. Phys. Chem. 1995, 99,
16320–16326.
[7] V. Aubert, V. Guerchais, E. Ishow, K. Hoang-Thi, I. Ledoux, K. Na-
katani, H. Le Bozec, Angew. Chem. 2008, 120, 587–590; Angew.
Chem. Int. Ed. 2008, 47, 577–580.
colored form. This two-way switching mechanism has been
substantiated by the similarity of the photo- and acido-trig-
gered absorption spectra of the zwitterionic and POFs. In
both measurements and calculations, the maximum of ab-
sorption is redshifted upon adding electron-withdrawing
substituents and the effect is stronger for the POFs than for
the CFs.
HRS measurements performed at 1064 nm have revealed
that 1) the second-order NLO contrasts between the two
forms remains very large upon substitution and 2) the de-
pendence of bHRS of the open form with the substituent is
difficult to explain owing to the fact that the measurements
are performed in resonance conditions. On the other hand,
calculations accounting for solvent effects predict a global
increase of bHRS of the POF with the Hammett parameter of
the substituent: bHRS (R=H)<bHRS (R=Br)<bHRS (R=
CHO)<bHRS (R=NO2). When electron correlation effects
are included at the Møller–Plesset second-order level of ap-
proximation, the bHRS values for R=CHO and R=NO2
become very similar, but their relative b ordering is revert-
ed. Extrapolations of the measured bHRS values to the static
limit has also been carried out by using models including ho-
mogeneous and inhomogeneous broadenings as well as
single-mode vibronic structures. With the exception of the
R=CHO compound, the extrapolated bHRS values are con-
sistent with the calculations and their linear optical proper-
ties.
[8] a) B. J. Coe, S. Houbrechts, I. Asselberghs, A. Persoons, Angew.
Chem. 1999, 111, 377–380; Angew. Chem. Int. Ed. 1999, 38, 366–
369; b) I. Asselberghs, K. Clays, A. Persoons, A. M. McDonagh,
M. D. Ward, J. A. McCleverty, Chem. Phys. Lett. 2003, 368, 408;
c) B. J. Coe, Acc. Chem. Res. 2006, 39, 383–393; d) L. Boubekeur-
Lecaque, B. J. Coe, K. Clays, S. Foerier, T. Verbiest, I. Asselberghs,
J. Am. Chem. Soc. 2008, 130, 3286.
[9] E. Botek, M. Spassova, B. Champagne, I. Asselberghs, A. Persoons,
K. Clays, Chem. Phys. Lett. 2005, 412, 274–279; E. Botek, M. Spas-
sova, B. Champagne, I. Asselberghs, A. Persoons, K. Clays, Chem.
Phys. Lett. 2006, 417, 282.
[10] J. F. Lamꢅre, I. Malfrant, A. Sournia-Saquet, P. G. Lacroix, J. M.
Fabre, L. Kaboub, T. Abbaz, A. K. Gouasmia, I. Asselberghs, K.
Clays, Chem. Mater. 2007, 19, 805–815.
[11] K. Clays, E. Hendrickx, M. Triest, T. Verbiest, A. Persoons, C.
Dehu, J. L. Brꢀdas, Science 1993, 262, 1419.
[12] a) I. Asselberghs, G. Hennrich, K. Clays, J. Phys. Chem. A 2006, 110,
6271–6275; b) M. M. Oliva, J. Casado, J. T. Lopez Navarrete, G.
Hennrich, M. C. Ruiz Delgado, J. Orduna, J. Phys. Chem. C 2007,
111, 18778.
[13] M. Giraud, A. Lꢀaustic, R. Guillot, P. Yu, P. G. Lacroix, K. Nakata-
ni, R. Pansu, F. Maurel, J. Mater. Chem. 2007, 17, 4414–4425.
[14] a) B. Champagne, M. Spassova, J. B. Jadin, B. Kirtman, J. Chem.
Phys. 2002, 116, 3935–3946; b) M. Spassova, B. Champagne, B. Kirt-
man, Chem. Phys. Lett. 2005, 412, 217–222.
[15] a) S. L. Gilat, S. H. Kawai, J.-M. Lehn, Chem. Eur. J. 1995, 1, 275;
b) D. Majumdar, H. M. Lee, J. Kim, K. S. Kim, B. J. Mhin, J. Chem.
Phys. 1999, 111, 5866–5872; c) M. Del Zoppo, A. Lucotti, G. Bertar-
elli, G. Zerbi, Vib. Spectrosc. 2007, 43, 249–253.
Acknowledgements
[16] a) M. Nakano, R. Kishi, N. Nakagawa, S. Ohta, H. Takahashi, S. Fur-
ukawa, K. Kamada, K. Ohta, B. Champagne, E. Botek, S. Yamada,
A.P. is grateful to the Fonds pour la Formation ꢄ la Recherche dans lꢃIn-
dustrie et dans lꢃAgriculture (F.R.I.A.) for financial support. V.R. is in-
debted to the Rꢀgion Aquitaine for financial support in optical, laser,
and computer equipment. B.C. thanks the Fund for Scientific Research
(FRS-FNRS) for his research director position as well as the University
of Bordeaux I for his visiting professorship. The calculations have been
performed at the Interuniversity Scientific Calculation Facility (ISCF) in-
stalled at the Facultꢀs Universitaires Notre-Dame de la Paix (Namur,
Belgium), for which the authors gratefully acknowledge the financial sup-
port of the FNS-FRFC, the Loterie Nationale for grant no. 2.4578.02,
and the FUNDP, as well as on the intensive calculation pole “M3PEC-
MESOCENTRE” of the University Bordeaux I financed by the Conseil
Rꢀgional dꢃAquitaine and the French Ministry of Research and Technol-
ogy.
K. Yamaguchi, J. Phys. Chem.
A 2006, 110, 4238–4243; b) M.
Nakano, R. Kishi, S. Ohta, H. Takahashi, T. Kubo, K. Kamada, K.
Ohta, E. Botek, B. Champagne, Phys. Rev. Lett. 2007, 99, 033001.
[17] L. Sanguinet, J. L. Pozzo, V. Rodriguez, F. Adamietz, F. Castet, L.
Ducasse, B. Champagne, J. Phys. Chem. B 2005, 109, 11139–11150.
[18] a) L. Sanguinet, J. L. Pozzo, M. Guillaume, B. Champagne, F. Castet,
L. Ducasse, E. Maury, J. Souliꢀ, F. ManÅois, F. Adamiez, V. Rodri-
guez, J. Phys. Chem. B 2006, 110, 10672–10682; b) F. ManÅois, L.
Sanguinet, J. L. Pozzo, M. Guillaume, B. Champagne, V. Rodriguez,
F. Adamietz, L. Ducasse, F. Castet, J. Phys. Chem. B 2007, 111,
9795–9802.
[19] F. ManÅois, V. Rodriguez, J. L. Pozzo, B. Champagne, F. Castet,
Chem. Phys. Lett. 2006, 427, 153–158.
[20] a) V. Cavillot, B. Champagne, Chem. Phys. Lett. 2002, 354, 449–457;
b) A. Masunov, S. Tretiak, J. Phys. Chem. B 2004, 108, 899–907;
c) D. Rappoport, F. Furche, J. Am. Chem. Soc. 2004, 126, 1277; d) Y.
Tawada, T. Tsuneda, S. Yanagisawa, T. Yanai, K. Hirao, J. Chem.
Phys. 2004, 120, 8425; e) M. Guillaume, B. Champagne, F. Zutter-
man, J. Phys. Chem. A 2006, 110, 13007–13013; f) M. Miura, Y.
Aoki, B. Champagne, J. Chem. Phys. 2007, 127, 084103.
[21] a) H. Sekino, R. J. Bartlett, J. Chem. Phys. 1986, 85, 976–989;
b) S. P. Karna, M. Dupuis, J. Comput. Chem. 1991, 12, 487–504.
[22] a) J. Tomasi, M. Persico, Chem. Rev. 1994, 94, 2027; b) J. Tomasi, B.
Mennucci, R. Cammi, Chem. Rev. 2005, 105, 2999.
[23] H. D. Cohen, C. C. J. Roothaan, J. Chem. Phys. 1965, 43, S34.
[24] a) P. J. Davis, P. Rabinowitz, Numerical Integration, Blaisdell Pub-
lishing Company, London, 1967, 166; b) for an example of the Rom-
berg scheme in the case of push–pull p-conjugated systems, see: B.
Champagne, B. Kirtman, Nonlinear Optical Materials, Handbook of
[1] a) B. J. Coe, Chem. Eur. J. 1999, 5, 2464–2471; b) J. A. Delaire, K.
Nakatani, Chem. Rev. 2000, 100, 1817–1845.
[2] a) K. Nakatani, J. Delaire, Chem. Mater. 1997, 9, 2682–2684; b) M.
Sliwa, S. Lꢀtard, I. Malfrant, M. Nierlich, P. G. Lacroix, T. Asahi, H.
Matsuhara, P. Yu, K. Nakatani, Chem. Mater. 2005, 17, 4727–4735;
c) M. Sliwa, K. Nakatani, T. Asahi, P. G. Lacroix, R. B. Pansu, H.
Masuhara, Chem. Phys. Lett. 2007, 437, 212; d) M. Guillaume, B.
Champagne, N. Markova, V. Enchev, F. Castet, J. Phys. Chem. A,
2007, 111, 9914–9923; e) A. Plaquet, M. Guillaume, B. Champagne,
L. Rougier, F. ManÅois, V. Rodriguez, J. L. Pozzo, L. Ducasse, F.
Castet, J. Phys. Chem. C, 2008, 112, 5638–5645; f) M. Sliwa, A.
Spangenberg, I. Malfant, P. G. Lacroix, R. Mꢀtivier, R. B. Pansu, K.
Nakatani, Chem. Mater. 2008, 20, 4062–4068.
[3] S. Houbrechts, K. Clays, A. Persoons, Z. Pikramenou, J.-M. Lehn,
Chem. Phys. Lett. 1996, 258, 485.
2570
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2560 – 2571

Molecular Switches
FULL PAPER
Advanced Electronic and Photonic Materials and Devices, Vol. 9
(Ed.: H. S. Nalwa), Academic Press, San Diego, 2001, Chapter 2,
pp. 63–126.
Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari,
J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cio-
slowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaro-
mi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng,
A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W.
Chen, M. W. Wong, C. Gonzalez, J. A. Pople, Gaussian, Inc., Wall-
ingford CT, 2004.
[25] B. Champagne, B. Kirtman, J. Chem. Phys. 2006, 125, 024101.
[26] a) M. J. Rice, N. Handy, Int. J. Quantum Chem. 1992, 43, 91; b) H.
Sekino, R. J. Bartlett, Chem. Phys. Lett. 1995, 234, 87; c) D. Jacque-
min, B. Champagne, C. Hꢆttig, Chem. Phys. Lett. 2000, 319, 327–
334.
[27] B. Champagne, E. A. Perpꢅte, D. Jacquemin, S. J. A. van Gisbergen,
E. J. Baerends, C. Soubra-Ghaoui, K. Robins, B. Kirtman, J. Phys.
Chem. A 2000, 104, 4755–4763.
[31] J. Campo, W. Wenseleers, E. Goovaerts, M. Szablewski, G. H. Cross,
J. Phys. Chem. C 2008, 112, 287–296.
[32] J. L. Oudar, D. S. Chemla, J. Chem. Phys. 1977, 66, 2664.
[33] a) G. Berkovic, G. Meshulam, Z. Kotler, J. Chem. Phys. 2000, 112,
3997–4003; b) C. H. Wang, J. Chem. Phys. 2000, 112, 1917–1924;
c) J. N. Woodford, C. H. Wang, A. E. Asato, R. S. H. Liu, J. Chem.
Phys. 1999, 111, 4621–4628.
[28] a) M. Kamiya, H. Sekino, T. Tsuneda, K. Hirao, J. Chem. Phys.
2005, 122, 234111; b) F. A. Bulat, A. Toro-Labbꢀ, B. Champagne, B.
Kirtman, W. Yang, J. Chem. Phys. 2005, 123, 014319.
[29] R. Bersohn, Y. H. Pao, H. L. Frisch, J. Chem. Phys. 1966, 45, 3184.
[30] Gaussian 03, Revision D.02, M. J. Frisch, G. W. Trucks, H. B. Schle-
gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgom-
[34] a) P. Kaatz, D. P. Shelton, J. Chem. Phys. 1996, 105, 3918; b) M. A.
Pauley, C. H. Wang, Chem. Phys. Lett. 1997, 280, 544.
[35] a) B. F. Levine, C. G. Bethea, E. Wasserman, L. Leenders, J. Chem.
Phys. 1978, 68, 5042; b) P. M. Lundquist, S. Yitzchaik, T. J. Marks,
G. K. Wong, S. DiBella, R. Cohen, G. Berkovic, Phys. Rev. B 1997,
55, 14055.
ACHTUNGTRENNUNGery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S.
Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani,
N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K.
Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda,
O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian,
J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E.
Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W.
Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J.
Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
[36] wvib =1300, 1400, 1500, and 1100 cmꢀ1; gvib =0.14, 0.13, 0.15, and
0.11 eV for compounds 1–4, respectively.
Received: October 23, 2008
Published online: January 29, 2009
Chem. Eur. J. 2009, 15, 2560 – 2571
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2571