Rational enhancement of second-order nonlinearity: Bis-(4-methoxyphenyl) hetero-aryl-amino donor-based chromophores: Design, synthesis, and electrooptic activity

Article abstract of DOI:10.1021/ja8007424

Two new highly hyperpolarizable chromophores, based on N,N-bis-(4- methoxyphenyl)aryl-amino donors and phenyl-trifluoromethyl-tricyanofuran (CF₃-Ph-TCF) acceptor linked together via π-conjugation through 2,5-divinylenethienyl moieties as the bridge, have been designed and synthesized successfully for the first time. The aryl moieties on the donor side of the chromophore molecules were varied as to be thiophene and 1-n-hexylpyrrole. The linear and nonlinear optical (NLO) properties of all compounds were evaluated in addition to recording relevant thermal and electrochemical data. The properties of the two new molecules were comparatively studied. These results are critically analyzed along with two other compounds, reported earlier from our laboratories and our collaborator's, that contain (i) aliphatic chain-bearing aniline and (ii) dianisylaniline as donors, keeping the bridge (2,5-divinylenethienyl-), and the acceptor (CF₃-Ph-TCF), constant. Trends in theoretically (density functional theory, DFT) predicted, zero-frequency gas-phase hyperpolarizability [β(0;0,0)] values are shown to be consistent with the trends in [β_HRS(-2ω;ω, ω), as measured by Hyper-Rayleigh Scattering (HRS), when corrected to zero-frequency using the two-level model (TLM) approximation. Similarly, trends in poling efficiency data (r₃₃/E_p) and wavelength dispersion measured by reflection ellipsometry (using a Teng-Man apparatus) and attenuated total reflection (ATR) are found to fit the TLM and DFT predictions. A 3-fold enhancement in bulk nonlinearity (r₃₃) is realized as the donor subunits are changed from alkylaniline to dianisylaminopyrrole donors. The results of these studies provide insight into the complicated effects on molecular hyperpolarizability of substituting heteroaromatic subunits into the donor group structures. These studies also demonstrate that, when frequency dependence and electric-field-induced ordering behavior are correctly accounted for, ab initio DFT generated β(0;0,0) is effective as a predictor of changes in r₃₃ behavior based on chromophore structure modification. Thus DFT can provide valuable insight into the electronic structure origin of complex optical phenomena in organic media.

Full text of DOI:10.1021/ja8007424

Published on Web 07/19/2008
Rational Enhancement of Second-Order Nonlinearity:
Bis-(4-methoxyphenyl)hetero-aryl-amino Donor-Based
Chromophores: Design, Synthesis, and Electrooptic Activity
Joshua A. Davies, Arumugasamy Elangovan,* Philip A. Sullivan,*
Benjamin C. Olbricht, Denise H. Bale, Todd R. Ewy, Christine M. Isborn,
Bruce E. Eichinger, Bruce H. Robinson, Philip J. Reid, Xiaosong Li, and
Larry R. Dalton*
Department of Chemistry, UniVersity of Washington, 109 Bagley Hall, Box 351700, Seattle,
Washington 98195
Received January 29, 2008; E-mail: easamy@u.washington.edu; psull76@u.washington.edu; dalton@chem.washington.edu
Abstract: Two new highly hyperpolarizable chromophores, based on N,N-bis-(4-methoxyphenyl)aryl-amino
donors and phenyl-trifluoromethyl-tricyanofuran (CF3-Ph-TCF) acceptor linked together via π-conjugation
through 2,5-divinylenethienyl moieties as the bridge, have been designed and synthesized successfully
for the first time. The aryl moieties on the donor side of the chromophore molecules were varied as to be
thiophene and 1-n-hexylpyrrole. The linear and nonlinear optical (NLO) properties of all compounds were
evaluated in addition to recording relevant thermal and electrochemical data. The properties of the two
new molecules were comparatively studied. These results are critically analyzed along with two other
compounds, reported earlier from our laboratories and our collaborator’s, that contain (i) aliphatic chain-
bearing aniline and (ii) dianisylaniline as donors, keeping the bridge (2,5-divinylenethienyl-), and the acceptor
(CF3-Ph-TCF), constant. Trends in theoretically (density functional theory, DFT) predicted, zero-frequency
gas-phase hyperpolarizability [ꢀ(0;0,0)] values are shown to be consistent with the trends in ꢀHRS(-2ω;ω,ω),
as measured by Hyper-Rayleigh Scattering (HRS), when corrected to zero-frequency using the two-level
model (TLM) approximation. Similarly, trends in poling efficiency data (r33/Ep) and wavelength dispersion
measured by reflection ellipsometry (using a Teng-Man apparatus) and attenuated total reflection (ATR)
are found to fit the TLM and DFT predictions. A 3-fold enhancement in bulk nonlinearity (r33) is realized as
the donor subunits are changed from alkylaniline to dianisylaminopyrrole donors. The results of these studies
provide insight into the complicated effects on molecular hyperpolarizability of substituting heteroaromatic
subunits into the donor group structures. These studies also demonstrate that, when frequency dependence
and electric-field-induced ordering behavior are correctly accounted for, ab initio DFT generated ꢀ(0;0,0)
is effective as a predictor of changes in r33 behavior based on chromophore structure modification. Thus
DFT can provide valuable insight into the electronic structure origin of complex optical phenomena in organic
media.
Introduction
molecular hyperpolarizabilities (ꢀ), low optical loss, and excel-
lent thermal-, chemical-, and photochemical stabilities.5–7
When considering the design of dipolar π-conjugated chro-
mophores as organic second-order nonlinear optical materials,
optimization of electron donor and electron acceptor substituents
is necessary to obtain maximum nonlinearity at the molecular
level. Moreover, chemical-, photochemical-, and thermal stabil-
ity in turn depend on the nature of these groups.
In general, dipolar second-order NLO chromophores consist
of a π-conjugated bridge end-capped with strong electron donor
and acceptor substituents (D-π-A). It is well established that
extremely large ꢀ values can be achieved by optimizing the
ground-state polarization by carefully considering which com-
Development of highly efficient organic electrooptic (EO)
materials has been the focal point of recent research among
many organic materials research groups. This research is driven
by the attractive potential of improving the efficiency of
nonlinear optical (NLO) material while lowering production cost
for applications including electrooptic modulation (the Pockels
effect), optical rectification, and terahertz radiation generation
and detection.1–4 In order for the material to display very large
bulk EO response (characterized by its EO coefficient, r33), non-
centrosymmetrically aligned active materials must have large
(1) Baehr-Jones, T.; Hochberg, M.; Wang, G.; Lawson, R.; Liao, Y.;
Sullivan, P. A.; Dalton, L.; Jen, A. K. Y.; Scherer, A. Opt. Express
2005, 13, 5216–5226.
(5) He, M.; Leslie, T. M.; Sinicropi, J. A.; Garner, S. M.; Reed, L. D.
Chem. Mater. 2002, 14, 4669–4675.
(2) Dalton, L. R. AdV. Polym. Sci. 2002, 158.
(6) Marder, S. R.; Kippelen, B.; Jen, A. K. Y.; Peyghammbarian, N. Nature
(London) 1997, 388, 845–851.
(3) Schneider, A.; Guenter, P. Ferroelectrics 2005, 318, 83–88.
(4) Sinyukov, A. M.; Leahy, M. R.; Hayden, L. M.; Haller, M.; Luo, J.;
Jen, A. K. Y.; Dalton, L. R. Appl. Phys. Lett. 2004, 85, 5827–5829.
(7) Zhang, C.; Dalton, L. R.; Oh, M.-C.; Zhang, H.; Steier, W. H. Chem.
Mater. 2001, 13, 3043–3050.
9
10.1021/ja8007424 CCC: $40.75
2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 10565–10575 10565

A R T I C L E S
Davies et al.
bination of these three components can be used to impart a
necessary directional bias.8 Several D-π-A systems are known
in the literature, and the EO properties of the best of these
materials is improved by an order of magnitude as compared
to the best known inorganic material, LiNbO3. Yet, the full
potential of organic electrooptic materials remains largely
unrealized. In order to optimize molecular hyperpolarizability,
much effort has been focused on the design and synthesis of
optimized conjugated bridge and electron acceptor structures
while, in the newest generation of EO chromophores, the donor
units have remained relatively unchanged. Building upon this
knowledge, we envisaged that a further enhancement of mo-
lecular hyperpolarizability could be achieved by introducing
stronger electron donors. Unfortunately, the relative inflexibility
of the structural make up of the currently available chromophore
electron donor units have presented hurdles in this path for
achieving durable materials with enhanced electrooptic activity,
possessing a balance of all desirable qualities.
few examples of second-order NLO chromophores making use
of dialkylamino-heteroaryl donors,19–21 and fewer of diary-
lamino-heteroaryl donors.22–24 An explanation for this void in
the literature probably has to do with the reported difficulty in
functionalizing a heteroaromatic ring with an arylamine. Al-
though amination of 2- and 3-halo-thiophene is known,22–28
Hartwig and co-workers indicated that the reluctance to attempt
such reactions with 2- or 3-bromo-pyrroles is owed to the
instability of the starting material.25 In fact, to the best of the
authors’ knowledge, such an analogous functionalization has
not been previously reported.
In this work, aminoheteroarmomatic (thienyl- and pyrrolyl-)
groups were explored as a replacement for the phenyl (i.e.,
aniline) unit in the donor part of the donor-π-acceptor molecule.
For this study, two new compounds based on N,N-bis-(4-
methoxyphenyl)aryl-amine donors were synthesized anew.
These chromophores contain the key 2,5-divinylthienyl (FTC)
bridge and the powerful 2-dicyanomethylen-3-cyano-4-methyl-
5-phenyl-5-trifluoromethyl-2,5-dihydrofuran (CF3-Ph-TCF) ac-
ceptor. The N,N-bis-(4-methoxyphenyl)phenyl-amino donor-
containing chromophore A (Chart 1) was chosen as a reference
for comparison and further enhancement of ꢀ by synthetic
modification not only because it has a larger calculated π-donor
strength, but it is also experimentally shown to have a 2-fold
improvement in molecular hyperpolarizability (ꢀ) and a 3-fold
macroscopic EO response (r33) over the simple triphenylamino
analogue.9
The use of complementary experimental and theoretical
methods enables us to explore and evaluate the effects of
structural modifications in this series of molecules. Density
functional theory (DFT) quantum mechanical calculations were
used to model ground and exited state electron-density con-
figurations and evaluate the changes in zero-frequency hyper-
polarizability, ꢀ(0;0,0), to be expected with donor group
modification. These data were then used in concert with linear
optical property measurements to deconvolute the frequency
dependence of hyper-Rayleigh scattering (HRS) measurements
as well as attenuated total reflection (ATR) measurements of
electric field poling-induced r33. In addition to nonlinear optical
measurements, relevant linear optical, electrochemical, and
thermal properties are reported and discussed herein.
Apart from a few recent works,9,10 very little experimental
or theoretical research has been focused on the use of the
4-(diarylamino)aryl group as the electron donor in NLO
chromophores despite very comparable π-donor strengths with
respect to their 4-(dialkylamino)aryl analogues, and demon-
strated superior thermal stability and photostability.9,10 More-
over, the π-donor strength of 4-(diarylamino)phenyl moieties
can be tuned by the structural variation of the substituents on
the aryl rings. Much of the reluctance in incorporating these
donors into dipolar chromophores stems from the general lack
of success in translating large microscopic hyperpolarizability
(ꢀ) into large macroscopic EO response (r33), which may be
caused by π-π stacking interactions and charge transport
between chromophores. Recently, Jen and some of us were able
to overcome some of these challenges to achieve an order of
magnitude improvement in r33 over previous triarylamino
chromophores.9
Conceptually, a strategy that could be used to increase
π-donor strength is to replace the phenyl ring, commonly used
in the donor end of the main conjugation, with heteroaromatic
molecules having reduced aromaticity and ionization potential.10
Despite several accounts of experimental and theoretical work
suggesting that the use and location of heteroaromatic units may
have the potential to dramatically increase ꢀ,10–18 there are very
Results and Discussion
(8) Marder, S. R.; Gorman, C. B.; Meyers, F.; Perry, J. W.; Bourhill, G.;
Bredas, J.-L.; Pierce, B. M. Science (Washington) 1994, 265, 632–
635.
Synthesis. The syntheses of donors 5a and 5b are outlined in
Scheme 1. The literature method for making 5a involves a two-
(9) Cheng, Y.-J.; Luo, J.; Hau, S.; Bale, D. H.; Kim, T.-D.; Shi, Z.; Lao,
D. B.; Tucker, N. M.; Tian, Y.; Dalton, L. R.; Reid, P. J.; Jen, A. K. Y.
Chem. Mater. 2007, 19, 1154–1163.
(18) Raposo, M. M. M.; Sousa, A. M. R. C.; Kirsch, G.; Cardoso, P.;
Belsley, M.; de Matos Gomes, E.; Fonseca, A. M. C. Org. Lett. 2006,
8, 3681–3684.
(10) Kwon, O.; Barlow, S.; Odom, S. A.; Beverina, L.; Thompson, N. J.;
Zojer, E.; Bredas, J.-L.; Marder, S. R. J. Phys. Chem. A 2005, 109,
9346–9352.
(19) Jen, A. K. Y.; Rao, V. P.; Wong, K. Y.; Drost, K. J. J. Chem. Soc.,
Chem. Commun. 1993, 90–92.
(11) Albert, I. D. L.; Marks, T. J.; Ratner, M. A. J. Am. Chem. Soc. 1997,
119, 6575–6582.
(20) Rao, V. P.; Cai, Y. M.; Jen, A. K. Y. J. Chem. Soc., Chem. Commun.
1994, 1689–1690.
(12) Breitung, E. M.; Shu, C.-F.; McMahon, R. J. J. Am. Chem. Soc. 2000,
122, 1154–1160.
(21) Rao, V. P.; Jen, A. K. Y.; Wong, K. Y.; Drost, K. J. Tetrahedron
Lett. 1993, 34, 1747–1750.
(13) Leclercq, A.; Zojer, E.; Jang, S. H.; Barlow, S.; Geskin, V.; Jen,
A. K. Y.; Marder, S. R.; Bredas, J. L. J. Chem. Phys. 2006, 124,
044510/1-044510/7.
(22) Bedworth, P. V.; Cai, Y.; Jen, A.; Marder, S. R. J. Org. Chem. 1996,
61, 2242–2246.
(23) Hu, Z.-Y.; Fort, A.; Barzoukas, M.; Jen, A. K. Y.; Barlow, S.; Marder,
S. R. J. Phys. Chem. B 2004, 108, 8626–8630.
(24) Jen, A. K. Y.; Cai, Y.; Bedworth, P. V.; Marder, S. R. AdV. Mater.
1997, 9, 132–135.
(14) Rao, V. P.; Jen, A. K. Y.; Chandrasekhar, J.; Namboothiri, I. N. N.;
Rathna, A. J. Am. Chem. Soc. 1996, 118, 12443–12448.
(15) Batista, R. M. F.; Costa, S. P. G.; Malheiro, E. L.; Belsley, M.; Raposo,
M. M. M. Tetrahedron 2007, 63, 4258–4265.
(25) Hooper, M. W.; Utsunomiya, M.; Hartwig, J. F. J. Org. Chem. 2003,
68, 2861–2873.
(16) Facchetti, A.; Beverina, L.; van der Boom Milko, E.; Dutta, P.;
Evmenenko, G.; Shukla, A. D.; Stern, C. E.; Pagani Giorgio, A.; Marks
Tobin, J. J. Am. Chem. Soc. 2006, 128, 2142–21453.
(17) Raposo, M. M. M.; Sousa, A. M. R. C.; Kirsch, G.; Ferreira, F.;
Belsley, M.; de Matos Gomes, E.; Fonseca, A. M. C. Tetrahedron
2005, 61, 11991–11998.
(26) Ogawa, K.; Rasmussen, S. C. J. Org. Chem. 2003, 68, 2921–2928.
(27) Wu, I.-Y.; Lin, J. T.; Tao, Y.-T.; Balasubramaniam, E.; Su, Y. Z.;
Ko, C.-W. Chem. Mater. 2001, 13, 2626–2631.
(28) Yamashita, M.; Hartwig, J. F. J. Am. Chem. Soc. 2004, 126, 5344–
5345.
9
10566 J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008

Design and Synthesis of Hyperpolarizable Chromophores
A R T I C L E S
Chart 1. Structures of the Chromophores YLD156, A, B, and C
pyrrole was dibrominated selectively and quantitatively using
N-bromosuccinimide in tetrahydrofuran at -78 °C, then mono-
formylated with dimethylformamide following lithium exchange
at -78 °C using n-butyllithium. 5-Bromo-1-hexyl-1H-pyrrole-
2-carbaldehyde 4 was then aminated with the dianisylamine to
produce 5c in 59% yield over four steps. Thus, we have
successfully devised a novel synthetic protocol for the prepara-
tion of the previously unknown compound 5c.
Following the synthetic exploration invested into the prepara-
tion of donors 5a-5c, the effort put forth toward the syntheses
of chromophores A, B, and C, as outlined in Scheme 3, was
more or less the same in comparison to the synthesis of other
FTC type chromophores. In a parallel fashion, donors 5a-5c
were extended to the donor-bridges 6a-6c, all in 80 to 95%
yield, using efficient Horner-Wadsworth-Emmons olefination
methodology using diethyl (2-methylthiophene) phosphonate.
Donor-bridge intermediates were then formylated in excellent
yields at the 5′ position using n-butyllithium and DMF at -78
°C. Finally, the Knoevenagel condensation reaction with the
CF3Ph-TCF acceptor proceeded smoothly in ethanolic solution
in the presence of an inorganic base to give final chromophores
A, B, and C in 70-75% yield. Thus, the overall yields of A,
B, and C were 51%, 19%, and 34%, respectively.
Linear optical and electrochemical properties. UV-visible
absorption spectral and electrochemical methods were employed
to explore the differing electron-donating properties provided
by the modified donor moieties. The recorded UV-visible
absorption spectra of compounds A, B, C, and YLD156 in
chloroform are displayed in Figure 1 and the photophysical and
electrochemical data are recorded in Table 1.
When comparing the λmax of bis-(4-methoxyphenyl)ami-
nophenyl donor-containing chromophore A (745 nm) with its
dialkylaminophenyl counterpart YLD156 (753 nm), A is slightly
blue-shifted and the two compounds exhibit similar charge-
transfer band shape.
step process wherein 4-bromobenzonitrile is first coupled with
4,4′-dimethoxydiphenylamine using Hartwig-Buchwald pal-
ladium-catalyzed amination, then reduced using DIBAL-H to
afford 5a in a 73% yield overall.9 While this yield is satisfactory,
it was improved to 89% and reduced to one step by directly
coupling 4,4′-dimethoxydiphenylamine to 4-bromobenzalde-
hyde. Although donor 5b is reported in the literature its
preparation has yet to be disclosed. It was prepared in a manner
analogous to that for 5a in 60% yield.
The preparation of 5c, as shown in Scheme 2, was much more
challenging and involved several attempts by different routes
before the synthesis was satisfactorily completed. To facilitate
incorporation of the pyrrole into an NLO chromophore, and to
increase solubility, the labile nitrogen proton was replaced with
an n-hexyl chain following a literature procedure.29 As shown
in Route A, 1-hexyl-1H-pyrrole was then monobrominated at
the 2 position producing the desired product with good
selectivity (90% as determined by 1H NMR). This bromide was
then reacted with 4,4′-dimethoxydiphenylamine, following typi-
cal Pd-catalyzed coupling procedures, to produce 2 in 85% yield.
The instability of the 2-pyrrole-containing compounds, as cited
by Hartwig, was realized during the attempted Vilsmeier
formylation of 2, which resulted in decomposition and failed
to yield 5c. After a second attempt at the formylation of 2 using
n-butyllithium resulted in the same fate, a newly revised plan
was followed. As shown in Route B, Scheme 2, 1-hexyl-1H-
It may be noted that the presence of the dianisyl group in A,
B, and C gives rise to the high-energy band at 500 nm which
is absent in YLD156. Replacing the phenyl donor group with
the thiophene donor, or the N′-hexylpyrrole donor, has a
profound effect on the charge-transfer bands resulting in 112
nm, and 108 nm, respectively, red-shifted values for chro-
mophores B and C. While the charge-transfer band of B is
slightly broader than that of A, that of C is by far the broadest,
particularly the low-energy tail of the transition, which extends
all the way out to about 1200 nm. This could indicate that the
CT state is populated over a wide range of energy levels,
suggesting that the vibrational contribution is disfavored or that
the vibrational components are masked under the broad CT
band. This can be explained on the basis that the pyrrole moiety,
coupled with the strong electron-donating ability of the dianisyl
units, augments the electron-donating ability of the donor. The
HOMO and LUMO orbitals for structure C show that the pyrrole
group contributes roughly equally to both, which suggests that
it fosters electron transfer from the donor to the bridge (see
Figure 2C). Broadening can also be caused by an increased
density of states in the excited state. Such a situation can arise
from high excitation into the vibrational manifold. This cir-
cumstance would require a drastic change in Franck-Condon
overlap, suggesting a large geometry change, which would be
common for a molecule consisting of a loose, long, and efficient
conjugation structure versus a rigid polynuclear aromatic system.
In spite of the fact that the charge transfer state is spread over
500 nm, the oscillator strength appears to be satisfactory as
(29) Lion, C.; Baudry, R.; Hedayatullah, M.; Da Conceicao, L.; Hocquaux,
M.; Genard, S.; Maignan, J. J. Heterocycl. Chem. 2000, 37, 1635–
1640.
9
J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008 10567

A R T I C L E S
Davies et al.
Scheme 1. Syntheses of Donors 5a, and 5ba
a Reagents and conditions: (a) 2 mol % Pd2(dba)3, 4 mol % (t-Bu)3P, t-BuONa, toluene, 125 °C, 18 h.
Scheme 2. Synthesis of Donor 5ca
a Reagents and conditions: (a) 1-bromo-hexane, NaH, THF, 65 °C, 12 h. (b) NBS, THF, -78 to -15 °C 4 h. (c) 2 mol % Pd2(dba)3, 4 mol % (t-Bu)3P,
t-BuONa, toluene, reflux, 18 h. (d) POCl3, DMF, diethylether, 0 °C to rt, 6 h, then 1 N HCl. (e) n-BuLi, THF, -78 °C, 1 h, then DMF, -78 °C to rt, 1 N
HCl.
compared with the rest of the series as seen from the molar
absorptivity in chloroform.
Chromophore B, having two thiophene rings in the bridge,
exhibits a minor contribution from a vibrational coupling
appearing as a slight shoulder at about 800 nm in addition to
the major charge transfer band at 857 nm. This shoulder could
originate from the fact that, although there is sufficient charge
transfer transition possible in this chromophore, the somewhat
symmetrical nature of the bridge impacts the molecule to reduce
its CT nature, making it a slightly unperturbed system favoring
vertical excitation to populate the Franck-Condon state,
especially around the bridge units. This is supported by the
narrowness of the band structure as well. Additional support
for this contention comes from the DFT-calculated HOMO
structures (Figure 2), wherein one can see a strong delocalization
of both HOMO and LUMO on the dithienylethene bridge.
(30) Accelrys: San Diego, 2002.
(31) Delley, B. In Modern Density Functional Theory: A Tool for
Chemistry; Seminario, J. M., Politzer, P. , Eds.; Elsevier Science
Publishers: Amsterdam, 1995; Vol. 2.
(32) Hammer, B.; Hansen, L. B.; Norskov, J. K. Phys. ReV. B: Condens.
Mater. Phys. 1999, 59, 7413–7421.
(33) Frisch, M. J.; et al. Gaussian03, B.04 ed.; Gaussian, Inc: Wallingford,
CT, 2004.
The UV-vis absorption spectra were also measured in
dioxane so that solvatochromic behavior of the chromophores
could be explored (Table 1). Compounds B and C showed very
large hypsochromic shifts of 107 and 95 nm, respectively, from
chloroform to dioxane, whereas compounds A and YLD156
(34) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B: Condens. Mater. Phys.
1988, 37, 785–789.
(35) Hariharan, P. C.; Pople, J. A. Theor. Chim. Acta 1973, 28, 213–22.
(36) Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972, 56,
2257–2261.
9
10568 J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008

Design and Synthesis of Hyperpolarizable Chromophores
A R T I C L E S
Scheme 3. Syntheses of Chromophores A, B, and Ca
a Reagents and conditions: (a) t-BuOK, THF, 0 °C to rt, 8 h. (b) n-BuLi, THF, -78 °C, 1 h, then DMF, -78 °C to rt, 1 N HCl. (c) CF3Ph-TCF acceptor
(8) EtOH, 8 h rt.
Figure 1. (a) Absorption spectra of chromophores A, B, C, and YLD156 in chloroform. (b) Solvatochromic behavior of C recorded in different solvents
of varying dielectric constants (dioxane: 2.25; toluene: 2.38; CHCl3: 4.81; acetone: 20.7; acetonitrile: 37.5) and as film in amorphous polycarbonate (APC).
It may be noted that the spectrum recorded for the film (black) appears very similar (with slight broadening) to that recorded in chloroform (blue).
Table 1. Summary of Optical Absorption and Electro-chemical Data
absorption λmax (nm)
ꢀ ( × 104 M-1 cm-1
toluene
)
b
Eox (V)
c
Ered (V)
cmpd
dioxane
toluene
CHCl3
dioxane
CHCl3
∆E (UV)a (eV)
∆E (CV)d (eV)
YLD156
687
682
750
758
710
713
811
786
753
745
857
853
5.64
5.06
5.05
4.37
5.16
3.95
5.41
4.12
6.21
5.10
6.23
4.85
1.38
1.37
1.23
0.395
0.189
0.174
-0.815
-0.952
-0.822
-0.824
1.03
0.95
0.84
0.71
A
B
C
e
c
1.09
0.075
a Estimated from the onset of the absorption at the low-energy edge in chloroform. b Referenced to ferrocene standard (E1/2 ) 0.39 in CH2Cl2).
d
c Arrived at using peak potential as the reactions are irreversible. EHOMO ) -e[4.8 V + (Eox
- Fcox (onset))]; ELUMO ) -e[4.8 V + (Ered
-
(onset)
(onset)
Fcred (onset))]. e Estimated from the onset of the absorption at the low-energy edge in solid film in APC.
had moderate blue-shifts of 63 and 66 nm, respectively. An
extended solvatochromic study was attempted for C; the spectra
are overlain in Figure 1b. While the charge transfer band showed
decent solvatochromic red-shift upon increasing the solvent
dielectric constant, there appears to be a threshold for this
behavior for such a highly polarizable compound as C that there
occurs a reversal of CT upon increasing the dielectric constant
to beyond 20.7 [acetone and acetonitrile (37.5) appear to favor
excessive charge transfer in this compound] leading to back
electron donation causing blue-shift from the absorption in
chloroform. This confirms the concept that chromophores B and
C are more strongly polarizable than even YLD156 and A.
In order to determine the electrochemical properties of the
compounds under study, cyclic voltammetry (CV) experiments
were performed in degassed anhydrous dichloromethane solu-
tions containing Bu4NPF6 as the supporting electrolyte. Volta-
mmograms of ∼10-3 M solutions of chromophores A, B, C,
and YLD156 were recorded, and the Eox and Ered for each
compound, relative to ferrocene standard, were determined from
each data set and furnished in Table 1. Cyclic voltammetric
traces recorded for all compounds are shown in Figure 3. The
one-electron oxidation reaction was found to be typically
Nernstian (fully reversible) for A and YLD156, whereas that
for compounds B was quasi-reversible and that for C was found
to be completely irreversible, as was the reduction reaction for
all four compounds. The observed electrochemical behavior can
be explained by taking into account the orbital structures of
the HOMO and LUMO for the compounds studied (Figure 2).
In the case of compounds B and C the transfer of an electron
from the donor to the anode occurs adiabatically during the
forward scan as the one-electron oxidation. As the potential is
switched, the electron added to the molecule probably does not
9
J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008 10569

A R T I C L E S
Davies et al.
Figure 2. HOMO (bottom) and LUMO (top) surfaces of chromophores. Geometries were optimized with DMol30,31 using PBE32 functional with the dnp
basis set under medium quality convergence criteria. Electrostatic moments and HOMO-LUMO energy gaps were calculated on the basis of these optimized
geometries using Gaussian0333 employing B3LYP34 functional with 3-21G*35,36 basis set (the pyrrole n-alkyl substituent was limited to an ethyl group for
simplicity).
Figure 3. Cyclic voltammogram of chromophores recorded in CH2Cl2 with a scan rate of 50 mV/s using tetrabutylammonium hexafluorophosphate supporting
electrolyte.
land on the donor, but instead reaches the acceptor. Following
this, a chemical reaction probably occurs, making a neutral
molecule that in turn accepts an electron from the electrode
during the reduction cycle. The occurrence of a LUMO that is
vastly delocalized from the heteroaryl part of the donor through
the acceptor end is suggestive of this reactivity. It is possible
that the oxidized species in these two cases possess lower
LUMO energy than that of the neutral molecule. Evidence of
this kind of chemical reaction may be noticeable just before
the onset of the reductions of both compounds B (a shallow
onset) and C (a small spike) close to 0.2 V relative to Fc.
The HOMO and LUMO energy levels were determined from
the oxidation and reduction potentials, respectively. The ioniza-
tion potential (IP) was calculated from the oxidation onset (Eonset
ox) according to IP ) Eonset ox (Ag/Ag+) + 4.8, where Eonset ox
(Ag/Ag+) is the oxidation potential onset in volts versus the
Ag/Ag+ electrode. The LUMO energy level was similarly
determined from the electron affinity (EA) from the reduction
potential onset (Eonset red) according to IP ) Eonset red (SEC) +
4.4. The difference between these two values provides the
HOMO - LUMO energy difference (∆E (CV)). From the
calculated values reported in Table 1, it can be concluded that
the LUMO energy levels are approximately the same for all
the compounds. Conversely, the HOMO energy levels are
systematically lowered with an increase in the donor strength
which influences the HOMO to a greater extent than the LUMO
9
10570 J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008

Design and Synthesis of Hyperpolarizable Chromophores
A R T I C L E S
it is still quite comparable to many commonly used chro-
mophores like AJC146 and YLD124.9
Nonlinear Optical Properties
Molecular nonlinearity. Molecular ꢀijk, as described by
quantum mechanical modeling, is usually reported as the
nonlinear component of electron density polarization that is
induced along the molecular dipolar axis when the molecule is
acted upon by a static, DC (zero frequency) electric field
[ꢀzzz(0;0,0)]. For dipolar chromophores, this is the largest
component of the hyperpolarizability tensor.37 Experimentally
observed second-order nonlinear optical properties, such as
second harmonic generation (SHG) and the electrooptic (EO)
effect, often exhibit pronounced frequency dependence. This
occurs when the frequency of the AC optical field approaches
a molecular resonance. Therefore, when measured at a finite
operational wavelength, λexp, the quantity that defines the
electrooptic coefficient, r33, is not zero frequency, ꢀzzz(0;0,0),
but is instead the frequency dependent ꢀzzz(-ω;0,ω). For
solution-based hyper-Rayleigh scattering measurements the
important term is ꢀHRS(-2ω;ω,ω). For this reason, in order to
compare chromophore systems having quite different values for
∆Eeg(optical) and thus different ωeg in terms of DFT generated
ꢀ(0;0,0), frequency dispersion effects must be accounted for
when analyzing experimental data. This deconvolution allows
direct comparison of experimental measurements of r33 and
Figure 4. DSC thermograms of chromophores A, B, C, and YLD156.
Table 2. Thermal Properties
chromophore
Tg (°C)
Td (°C)
YLD156
158
138
206
89
211
261
228
179
A
B
C
ꢀ
ꢀ
HRS(-2ω;ω,ω) with quantum mechanical predictions of
level. The electronic effect concomitant with the increased
degree of conjugation moreover narrows the energy gap between
the HOMO and the LUMO energy levels leading to ∆E values
of 0.7 eV for C. The electrochemical values are corroborated
by the spectroscopically measured values and also by the DFT
calculations.
Trends in energy gaps between ∆E (optical) and ∆E
(electrochemical) for the four compounds were found to be
consistent. As is commonly observed, ∆E (optical) is signifi-
cantly overestimated when compared to the data obtained by
CV. The observed trends in optical and electrochemical ∆E
support the idea that as the aromatic stabilization energy in the
donor is reduced in moving from phenyl-, through thienyl-, to
pyrrole-based donors, electron-donating strength is indeed
enhanced. As ∆E is reduced, resulting in a bathochromic shift
of λmax within the series of compounds, it is noted that Ered
remains very stable and Eox is lowered.
zzz(0;0,0), made to correlate structural modification with
enhancements in molecular nonlinearity.
In order to estimate the differences in frequency dependence
among the chromophore series studied herein, the two-level
model (TLM) may be employed. The simplified version of the
TLM optical frequency dispersion formula needed for this study
is taken from work by Singer, et al.38 which can be written as
(3 + r r - r )
√
1
2
3
ꢀijk(-ω3;ω1, ω2) )
ꢀijk(0;0, 0)
(1)
3
3
1 - r
( )
k
∏
k)1
where the factor rR ) ωR2/ω2o, 1 e R e 3, is the square of the
ratio of the radiation frequency, ωR, to the frequency of the
lowest-energy transition, ωo ) ∆E/p. Converting to these ratios
to wavelengths gives
Thermal Properties. Thermal properties of chromophores A,
B, C, and YLD156 were explored by differential scanning
calorimetry and are shown and tabulated in Figure 4 and Table
2. Compounds C, A, YLD156, and B have a range of glass
transition temperatures (Tg) of 89, 138, 158, and 206 °C,
respectively. The relatively low Tg of C is likely due to the
flexible hexyl chain on the pyrrolesmolecular flexibility is
known to reduce Tg. The thermal decomposition temperatures
do not follow the same trends as Tg. A has by far the highest
decomposition temperature (261 °C), followed by B (228 °C),
YLD156 (211 °C), and C (178 °C). These results are thought
to be caused by two opposing effects. The enhanced thermal
stability of the diarylamino donors (A and B) over the
dialkylamino analogue, YLD156, is due to the absence of a
sp3 hybridized R-hydrogen adjacent to the donor nitrogen in
the latter. This effect is partially offset by lower thermal
stabilization of the thiophene and pyrrole moieties relative to
the phenyl. Furthermore, Td of C is likely further decreased due
to the introduction of an R-hydrogen adjacent to the pyrrole
nitrogen. Although the Td of C is about 30 °C less than YLD156,
ꢀijk(-ω;0, ω)
ꢀijk(0;0, 0)
3 - (λx ⁄ λ)2
3(1 - (λx ⁄ λ)2)2
)
(2)
as appropriate for r33 and
ꢀijk(-2ω;ω, ω)
1
)
(3)
(1 - (λx ⁄ λ)2)(1 - 4(λx ⁄ λ)2)
ꢀijk(0;0, 0)
to be used with HRS. Here λx locates the absorption maximum,
and λ is the excitation wavelength.
From the above relationships and the experimental values for
max, the frequency dependence of the nonlinear properties of
each chromophore can be estimated.
λ
Density functional theory (DFT) quantum mechanical meth-
ods were employed to determine theoretical values for the
(37) Cyvin, S. J.; Rauch, J. E.; Decius, J. C. J. Chem. Phys. 1965, 43,
4083–4095.
(38) Singer, K. D.; Kuzyk, M. G.; Sohn, J. E. J. Opt. Soc. Am. B: 1987, 4,
968–976.
9
J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008 10571

A R T I C L E S
Davies et al.
Table 3. DFT Predictions of ꢀHRS(0;0,0) and HRS Measurement Data Corrected by TLM
theoretical data (DFT)
experimental data (HRS)
a
b
c
cmpd
ꢀHRS(0;0,0)DFT
µ(D)DFT
ꢀ
HRS(-2ω;ω,ω) exp. × 10-30esu
ꢀ
HRS(-2ω;ω,ω)/ꢀHRS(0;0,0)
ꢀ
HRS(-2ω;ω,ω) corrected
ꢀ
HRS(TLM) normalized
YLD156
298
710
660
606
24
25
25
24
928 ( 106
1129 ( 42
2654 ( 209
2334 ( 140
3.19/2.31
3.07/2.31
6.74/2.31
6.46/2.31
672 ( 77
849 ( 31
909 ( 72
837 ( 50
1
A
B
C
1.26
1.35
1.24
b
a TLM proportionality factor derived using equation 3 divided by the TLM factor for the FTC standard; TLMFTC ) 2.31(λx ) 674nm). ꢀHRS divided
by TLM factor. c ꢀHRS(corrected)/ꢀHRS(YLD156).
hyperpolarizability39–41 for comparison with experiment. This
data was then used to help correlate experimental measurements
with structural effects as they are related to the electronic
properties. Figure 2, depicts the electron density distribution of
the HOMO and LUMO structures of the four chromophores.
In addition to explaining the steady-state photophysical proper-
ties of the systems studied here, this illustration also demon-
strates the substituent-induced ground- and excited-state electron
density asymmetry along the dipolar axis of the chromophore
that is essential for the induction of large ꢀzzz(0;0,0) molecular
nonlinearity.
Hyper-Rayleigh scattering (HRS) experiments were per-
formed on chromophores YLD156, A, B, and C. The HRS
technique unlike EFISH (electric field induced second harmonic
generation), 42 allows direct evaluation of ꢀHRS(-2ω;ω,ω). The
HRS measurements were performed using chloroform chro-
mophore solutions at a fundamental wavelength of λexp ) 1.9
µm, employing the Dalton group chromophore FTC 43–45 as an
external standard.46 The FTC reference value has been deter-
mined in house as; ꢀHRS(-2ω;ω,ω)FTC ) 445 ( 230. Because
FTC also exhibits pronounced frequency dependence, using this
standard allows the partial cancelation of the TLM factor,
reducing the severity of this correction. Normalizing the
resulting values against results for YLD156, and applying the
proportionality factors derived using the TLM equation (eq 3),
divided by TLMFTC ) 2.31(λx ) 674 nm), we can compare
hyperpolarizability values with those predicted by DFT. These
data are presented in Table 3.
much interest. Since DFT modeling considers an isolated
chromophore molecule in the gas phase, experimental measure-
ments performed in chloroform solution are expected to be
enhanced by a factor of this magnitude due to the reaction field
of the solvent.
To advance the computer model one step beyond the simple
TLM discussed above, the implications of the TLM can be
combined with DFT in order to directly assess frequency
dependence combined with reaction field calculations. In this
manner, the TLM provides a means to estimate the frequency
and solvent dependence of measurable properties of NLO
molecules simultaneously. We illustrate the method here for
molecule C, the pyrrole variant. According to Singer, et al.,34
the frequency dependence of the hyperpolarizability (in atomic
units) can be expanded as follows
ꢀijk(-ω3;ω1, ω2) ) sym[µi;egµj;eg(µk;ee
-
ω02(3ω20 + ω1ω2 - ω32)
(ω02 - ω12)(ω20 - ω22)(ω20 - ω32)
µk;gg)]
in the approximation that the perturbation series is dominated
by just the one state above the ground state. Here sym[*]
signifies symmetrization of all tensor components. The matrix
element of the kth component of the dipole operator between
states σ and F is denoted by µk;σF, the frequencies of the stimuli
electric fields are ω1 and ω2, while the response frequency is
ω3 ) ω1 + ω2, and ω0 ) Ee - Eg is the energy gap between
the ground and excited states in atomic units.
Using the TLM derived factor to estimate the expected
frequency dependence, it is seen that although the exact
magnitude of ꢀHRS(0;0,0)enhancement expected from theory may
not be observed, experimental values compare quite favorably
with DFT predictions considering measurement error. This also
suggests that in this case, the simple TLM may be reasonably
predictive. Within the A, B, and C chromophore series,
An ability to calculate excited-state dipole vectors, as well
as transition moments, is available in the developmental version
of Gaussian (GDV). We have used this functionality (with
B3LYP and 3-21G*) to evaluate µk;ee as well as µj;eg for
molecule C; the ground-state dipole µk;gg was done with
Gaussian03. The advantage of the method is that all dipole
matrix elements can be calculated concurrently with the PCM
solvent calculation, which would be prohibitively expensive with
the RT-TDDFT method described elsewhere.45 Gaussian default
parameters were used for the chloroform solvent. As shown in
Figure 5, the results obtained using this method seem to be quite
consistent with the uncorrected experimental values. Because
this methods takes into account the dependence of
ꢀ
HRS(0;0,0)values are found to be quite similar, in agreement
with DFT. The discrepancy between absolute experimental and
DFT ꢀHRS(0;0,0), a factor of 2 to 3, is currently the subject of
(39) Isborn, C. M.; Leclercq, A.; Vila, F. D.; Dalton, L. R.; Bredas, J. L.;
Eichinger, B. E.; Robinson, B. H. J. Phys. Chem. A 2007, 111, 1319–
1327.
ꢀ
HRS(-2ω;ω,ω) on both measurement frequency and dielectric
(40) Kinnibrugh, T.; Bhattacharjee, S.; Sullivan, P.; Isborn, C.; Robinson,
B. H.; Eichinger, B. E. J. Phys. Chem. B 2006, 110, 13512–13522.
(41) Liao, Y.; Eichinger, B. E.; Firestone, K. A.; Haller, M.; Luo, J.;
Kaminsky, W.; Benedict, J. B.; Reid, P. J.; Jen, A. K. Y.; Dalton,
L. R.; Robinson, B. H. J. Am. Chem. Soc. 2005, 127, 2758–2766.
(42) Teng, C. C.; Garito, A. F. Phys. ReV. B. 1983, 28, 6766.
(43) Dalton, L. R.; Harper, A.; Ren, A.; Wang, F.; Todorova, G.; Chen, J.;
Zhang, C.; Lee, M. Ind. Eng. Chem. Res. 1999, 38, 8.
(44) Kaatz, P.; Shelton, D. P. Opt. Commun. 1998, 157, 177–198.
(45) Takimoto, Y.; Vila, F. D.; Rehr, J. J. J. Chem. Phys. 2007, 127,
154114.
(solvent) environment for each chromophore, the result should
correspond to the uncorrected experimental results. Figure 5a
illustrates that at λexp ) 1.9 µm indeed the experimental value
of ꢀHRS ) 928 ( 106 for YLD156 compares quite favorably
with the modified DFT predictions. Similar comparison of the
experimental value for chromophore C, ꢀHRS ) 2334 ( 140,
and Figure 5c, confirms that these values are also consistent.
Macroscopic Nonlinearity. Bulk material EO activity (rij
coefficients) is governed by first nonlinear susceptibility, ꢁ(2)
.
(46) Akelaitis, A. J. P.; Olbricht, B. C.; Sullivan, P. A.; Liao, Y.; Lee,
S. K.; Bale, D. H.; Lao, D. B.; Kamisky, W.; Eichinger, B. E.; Choi,
D. H.; Dalton, L. R. Opt. Mater. 2007, ; http/dx.doi.org/10.1016/
j.optmat.2007.09.007.
Realization of nonlinear susceptibility requires the removal of
centrosymmetry both at the molecular and bulk levels. Provided
9
10572 J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008

Design and Synthesis of Hyperpolarizable Chromophores
A R T I C L E S
Figure 5. Plots of calculated molecular hyperpolarizabilities in vacuum and in the presence of solvent (CHCl3) showing frequency dependence: (a) and (b)
of YLD 156; (c) and (d) of chromophore C.
that the molecular hyperpolarizability tensor is dominated by
the component that is oriented along the dipolar molecular axis,
ment for comparison, the ATR technique was modified to
accommodate high-refractive index polymers through the use
of a rutile prism. The same sample format that is used for the
TMT measurement is also employed for ATR. In this report
samples were taken directly from poling and measurement, using
the in situ TMT, to a supplementary measurement on the ATR
setup. The ATR technique is a modification of the widely used
prism-coupling refractive index measurement.48
A change in the prism-coupling angle, due to the electrooptic
effect, is measured by monitoring the change in intensity of
the laser light reflected from the prism/sample interface as a
function of applied voltage. A schematic representation of the
experimental apparatus is shown in Figure 6. The sample is
mounted on the prism using a pneumatic plunger. The plunger,
prism, and detector are mounted atop a stepper motor to allow
precise control over the incidence angle of the measurement
laser. The electrooptic coefficient, r33, is calculated using the
relation
ꢀzzz, the first nonlinear susceptibility may be described by the
relation
(2)
zzz
ꢁ
) Nꢀzzz(-ω;0, ω) cos3 θ g(ω)
(4)
where N represents chromophore number density, θ is the angle
between the chromophore dipolar axis and the poling vector,
and cos3θ is the acentric order parameter. The term, g(ω),
denotes the Lorentz-Onsager local field factors. For the
materials considered herein, g(ω), should only differ by no more
than (10% from system to system, and will be neglected for
the sake of simplicity.38 The Pockels tensor component, r33, of
interest to device performance is related to ꢁ(2)by
r33 ) -2ꢁ(2) ⁄ n4
(5)
in which n is the refractive index of the material at the
operational wavelength.
To complete the analysis presented herein, two complemen-
tary experimental techniques were used for evaluation of
nonlinear optical behavior. A Teng-Man technique reflection
ellipsometry apparatus (TMT), modified to perform real-time
monitoring of electrooptic activity, was used to optimize poling
conditions and perform initial r33 measurements.47 In order to
obtain r33 data at two different wavelengths (λexp ) 1310 nm
and λexp ) 1550 nm), and to provide an alternate r33 measure-
δneff
δneff
δR
2d∆R
r33
)
⁄
(6)
(
)
n3TM
V
δθ δθ δnTM
where d is the thickness of the polymer film and nTM is the film
refractive index. As the sample is rotated past the coupling angle,
the intensity of the DC reflected signal, R, is recorded.
Simultaneously, the change in R, related to the change in
(48) Chen, A.; Chuyanov, V.; Garner, S.; Steier, W. H.; Dalton, L. R.
Organic Thin Films for Photonics Applications, OSA Technical Digest
Series; 1997; Vol. 14, pp 158-160.
(47) Sullivan, P. A.; Olbricht, B. C.; Akelaitis, A. J. P.; Mistry, A. A.;
Liao, Y.; Dalton, L. R. J. Mater. Chem. 2007, 17, 2899–2903.
9
J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008 10573

A R T I C L E S
Davies et al.
Figure 7. Plot of the ATR measurement data taken for a series of samples
poled using increasing Ep of chromophore B loaded into APC at 25% by
weight.
Figure 6. Schematic representation of the attenuated total reflection (ATR)
electrooptic coefficient measurement apparatus modified to use a high index
rutile prism, operating at λexp ) 1310 nm and λexp ) 1550 nm.
Table 4. Comparison of the Calculated and the Experimental
Nonlinear Properties with Order Measurements
r33/Epol (nm/V)2 (ATR)c
refractive index caused by application of an AC modulation
ꢀ
zzz(0) × 10-30 r33/Epol (nm/V)2
TLM
factord
e
voltage, V, is recorded using a lock-in amplifier. Here, neff
)
cmpd
1310 nm 1550 nm
P2 /Epol
esu (DFT)a
(Teng-Man)b
np sin θ where np is the prism refractive index. Rutile TiO2 has
a refractive index of no(1310 nm) ) 2.462 and no(1550 nm) )
2.453. The current setup includes two laser sources for measure-
ment of r33 at λexp ) 1310 nm and λexp ) 1550 nm for the
purpose of determining poling efficiency, r33/Ep, as a function
of measurement wavelength. From eqs 4 and 5, assuming the
order produced by poling, cos3 θ /Ep, chromophore number
density, N, and refractive index, nTM, are the same in all samples,
it follows that any difference in r33/Ep between λexp ) 1310 nm
and λexp ) 1550 nm must be due to the dependence of
YLD156
340
715
631
606
0.44b
0.60b
1.11
0.42
0.36
0.99
1.24
0.35
1
0.00041
A
B
C
0.28 0.98 0.00012
0.66 1.32
- 1.30 0.00046
-
1.21
a ꢀzzz(0;0,0)calculated by DFT. b Literature value measured by TMT.
c r33/Ep values measured by ATR. d TLM factors normalized against
YLD156; TLM(chrom)/TLM(YL156). e Centrosymmetric order parame-
ter as a function of poling field measured by VAPAS.
the simplicity of the comparison among chromophores. Also
tabulated are DFT values for ꢀzzz (0;0,0), and centrosymmetric
order parameter measurements, P2 /Ep.
ꢀzzz(-ω;0,ω) on measurement wavelength. This difference
should be well approximated using the proportionality factor
obtained from applying the TLM, eq 2. Performing this
experiment during evaluation of the chromophores under study
provides a simple but valuable check on the applicability of
the TLM.
When different samples composed of different materials are
compared, it may not be sufficient to assume that cos3 θ /Ep is
the same or even comparable. If the order induced by poling is
indeed very different, comparison of ꢀzzz(0;0,0) between materi-
als based on r33/Ep and TLM factors will be rendered unhelpful
without direct comparison of polar order. No direct measurement
of dipolar order is known. However, the centrosymmetric order
parameter P2 ) 1/2(3 cos2 θ - 1) may be evaluated though
simple linear dichroism measurements using the ratio variable
angle polarized absorption spectroscopy method.49,50 These
measurements were performed for poled samples of the chro-
mophores under study here to gain information about possible
differences in poling induced order between materials. Analo-
gous to r33 data, P2 was evaluated for a number of poled
samples and plotted as a function of poling voltage for each
material. In this manner poling induced dichroism, P2 /Ep, can
be compared among the chromophores. This data is also listed
in the last column of Table 4.
Shown in Figure 7 is r33 data corresponding to chromophore
B, as measured by ATR at both wavelengths, plotted as a
function of poling voltage. The error shown is the standard error
of the fit. The slope of this plot is the poling efficiency, r33/
Ep(nm/V)2. From eq 4, using λmax (CHCl3) ) 753 nm, the
difference in ꢀzzz(-ω;0,ω) between λexp ) 1310 nm and λexp
) 1550 nm is predicted to be (1.41)[ꢀzzz (1550 nm)] ≈
[ꢀzzz(1310 nm)] by the TLM approximation. Measured experi-
mentally, (1.5)[r33/Ep (1550 nm)] ) [r33/Ep (1310 nm)]. Holding
all other quantities equal, an assumption that should be valid
for the case where the values at both wavelengths are measured
on the same samples by simply changing laser sources, it is
suggested that the TLM is reasonably predictive. Similar
analyses were performed for chromophores YLD156 and A.
Consistent results were again obtained.
Finally, with measurements of r33/Ep, P2 /Ep, and TLM
frequency correction factors, we are ready to make qualitative
Table 4 lists the results obtained through the TMT technique
and ATR for chromophores B and C. Also listed are the
previously reported TMT results corresponding to chromophores
YLD156 and A, along with the corresponding TLM propor-
tionality factors. In this case, the TLM factors for ꢀzzz(-ω;0,ω)/
(49) Graf, H. M.; Zobel, O.; East, A. J.; Haarer, D. J. Appl. Phys. 1994,
75, 3335–3339.
(50) Rommel, H. L.; Robinson, B. H. J. Phys. Chem. C 2007, 111, 18765–
ꢀzzz(0;0,0) are normalized to chromophore YLD156 to aid in
18777.
9
10574 J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008

Design and Synthesis of Hyperpolarizable Chromophores
A R T I C L E S
comparisons of intrinsic hyperpolarizability as predicted by DFT
within a group of structurally similar, but electronically dis-
similar, EO chromophores. For example, ꢀzzz(0;0,0) predicted
by DFT calculations can be compared with experimental results
using the relation
oxidation potentials of the heteroaromatic donor chromophores
are significantly reduced, supporting the argument that such
modification indeed enhances electron-donor strength and is the
major cause of the reduced optical ∆Eeg.
Modeling of the nonlinear optical properties of the chro-
mophore structures reveals that when the dialkyl-substituted
amine donor is replaced by a di-p-methoxyphenyl-substituted
amine, a factor of approximately twice enhancement in
ꢀ(0;0,0)can be expected regardless of the nature of the central
donor aromatic ring. However, simple analysis of these predic-
tions was not sufficient for the purpose of rationalizing
experimental differences in r33 and ꢀHRS(-2ω;ω,ω), as cor-
related with molecular structure modification.
r
33(chrom)
Ep
≈
(
)
ꢀ
zzz(0)(chrom)
p2 ⁄ Ep chrom
[TLMfactor
{
[
]
][
]
ꢀzzz(0)(YLD156)
p2 ⁄ Ep(YLD156)
r
33(YLD156) (7)
Ep
Chemical alteration of the fundamental chromophore structure
results in pronounced variation in the optical ∆Eeg, molecular
shape, and solubility characteristics, as well as other complicat-
ing factors that must be addressed in order for chromophores
to be intercompared. In the case of HRS evaluation of
Substituting the data from Table 4 into eq 7, we can evaluate
the agreement between DFT and experiment for chromophore
C as follows:
1.24
) 0.48
(8)
606
340
0.00046
0.00041
ꢀHRS(-2ω;ω,ω), measurement frequency (excitation wave-
(1.30)
{(
)
(
)}
length) dependence of molecular nonlinearity must be taken into
account. In this report it is shown that the simple two-level
model, in which HOMO and LUMO level contributions are
assumed to dominate the relevant molecular electronic proper-
ties, is relatively satisfactory for this purpose. Correction of
experimental data using the TLM approximation results in good
agreement between theory and experiment. To extend this TLM-
based frequency dependence analysis to solid-state electrooptic
activity, the applicability of the TLM was confirmed by r33
measurements at multiple wavelengths using a newly modified
ATR experiment. For comparison of r33 measurements with
quantum mechanical predictions, experimental evaluation of
poling induced order was also required. Using a variable angle
polarized absorption spectroscopy technique, also modified in
our laboratories, P2 /Ep was evaluated. The ratio P2 /Ep is
found to correlate directly with the ratio r33/Ep. Combining DFT
and TLM predictions with P2 /Ep measurements, systematic
donor modification has been successfully correlated with
changes in poling induced r33. This study emphasizes the need
for simultaneous consideration of many interrelated effects
resulting from seemingly straightforward chromophore modi-
fication. Molecular structure must be optimized not only for
hyperpolarizability, but also for chromophore shape and solubil-
ity. Combined theoretical analysis and thorough experimental
characterization of systematically modified materials allows the
continued development of improved electrooptic materials
through a heightened understanding of increasingly detailed
structure-property relationships.
This comparison also yields quite good agreement when
applied to the remaining chromophore results. Such a good
agreement suggests that DFT calculations are reasonably
predictive of trends in r33 when the frequency dependence of
nonlinear effects is accounted for and poling-induced order can
be estimated. Applying the above qualitative relation, the
argument can be supported that dianisyl donor-based chro-
mophores A, B, and C indeed deliver ꢀzzz(0;0,0)values that are
enhanced by as much as a factor of 2, as predicted by DFT,
compared to the YLD156 chromophore standard. Chromophore
B, with its relatively rigid and overall planar structure, does
not respond well to the poling field, showing a low value for
P2 /Epcompared to the other chromophores in the series. The
data above suggests that even though chromphore B should
possess almost double the ꢀzzz(-ω;0,ω) of YLD156, the acentric
order produced by poling is only half as large, resulting in a
cancelation of effects and a very similar r33/Ep. Such poor poling
induced ordering behavior may be explained by differences in
chromophore-host compatibility and chromophore shape.50
Therefore, in order to realize improvements in r33/Ep such factors
must be taken into account in addition to hyperpolarizability as
illustrated in the recent work by Cheng et al.9
Conclusions
Two new theory-inspired, highly hyperpolarizable chro-
mophores, based on the bis-(4-methoxyphenyl)hetero-aryl-amino
donor (thiophene and N-hexylpyrrole) with 2,5-divinylenethienyl
bridges, and CF3-Ph-TCF acceptors have been synthesized.
Efficient synthetic approaches were developed to reach these
heteroaryl-amino donors. In addition to recording thermal,
electrochemical, and analytical data, linear and nonlinear optical
(NLO) properties were thoroughly investigated for all com-
pounds. The newly prepared dianisylaminothienyl and diani-
sylaminopyrrolyl chromophores both show nearly a 3-fold
enhancement in EO activity as compared to the previously
reported dianisylaminophenyl analogues.
Experimental Section
Complete synthetic details of all compounds and character-
ization details are available in the Supporting Information.
Acknowledgment. We acknowledge the support of the STC-
MDITR Program of the National Science Foundation (DMR0120967)
and NSF-(DMR-0092380). Support from the DARPA MORPH
Program Phase II ((N) 14-04-10094), is also gratefully acknowledged.
Several optical effects contribute to enhanced, poling-induced,
electrooptic activity. Evaluation of the absorption spectra within
the compound series revealed that as the donor is systematically
modified from dialkyl (YLD156), and diaryl (A), to diaryl
thiophene (B), and diarylpyrrole (C), a significant reduction in
∆Eeg is observed. Electrochemical data confirmed that the
Supporting Information Available: Experimental methods,
procedures for the synthesis of new compounds, complete
characterization data, and complete ref 33. This material is
available free of charge via the Internet at http://pubs.acs.org.
JA8007424
9
J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008 10575