Systematic study of the structure-property relationship of a series of ferrocenyl nonlinear optical chromophores

Article abstract of DOI:10.1021/ja0449123

A series of novel nonlinear optical (NLO) chromophores 1-4 incorporating the ferrocenyl (Fc) group as an electron donor and 2-dicyanomethylene-3-cyano-4- methyl-2,5-dihydrofuran (TCF) derivatives as electron acceptors are presented. The use of a constant Fc donor and varied acceptors and bridges makes it possible to systematically determine the contribution of the conjugated bridge and the acceptor strength to chromophore nonlinear optical activity. The X-ray crystal structures of all four chromophores allow for the systematic investigation of the structure-property relationship for this class of molecules. For example, the crystal structures reveal that both cyclopentadienyl groups in the ferrocenyl donor contribute to the electron donating ability. The first-order hyperpolarizabilities β of these chromophores, measured by hyper-Rayleigh scattering (HRS) relative to p-nitroaniline are reported. These β values are compared to those calculated by density functional theory (DFT). The excellent agreement between the theoretical and experimental β values demonstrates that a linear relation exists between the hyperpolarizability and the bond length alternation. An electrooptic coefficient, r₃₃, of ～25 pm/V at 1300 nm, for compound 4, incorporated into a polymer matrix, is competitive with organic chromophores. Moreover, this r₃₃ is more than 30 times larger than the previously reported value for an organometallic chromophore in a poled polymer matrix. This work not only underscores the potential for Fc donor moieties, which have been underutilized, but also demonstrates that experimental characterization and theoretical simulations are now congruent, viable methods for assessing potential performance of NLO materials.

Full text of DOI:10.1021/ja0449123

Published on Web 02/03/2005
Systematic Study of the Structure-Property Relationship of a
Series of Ferrocenyl Nonlinear Optical Chromophores
Yi Liao,† Bruce E. Eichinger,† Kimberly A. Firestone,† Marnie Haller,‡ Jingdong Luo,‡
Werner Kaminsky,† Jason B. Benedict,† Philip J. Reid,† Alex K-Y Jen,†,‡
Larry R. Dalton,† and Bruce H. Robinson*,†
Contribution from the Department of Chemistry and Department of Materials Science and
Engineering UniVersity of Washington, Seattle, Washington 98195-2120
Received August 23, 2004; E-mail: robinson@chem.washington.edu
Abstract: A series of novel nonlinear optical (NLO) chromophores 1-4 incorporating the ferrocenyl (Fc)
group as an electron donor and 2-dicyanomethylene-3-cyano-4-methyl-2,5-dihydrofuran (TCF) derivatives
as electron acceptors are presented. The use of a constant Fc donor and varied acceptors and bridges
makes it possible to systematically determine the contribution of the conjugated bridge and the acceptor
strength to chromophore nonlinear optical activity. The X-ray crystal structures of all four chromophores
allow for the systematic investigation of the structure-property relationship for this class of molecules. For
example, the crystal structures reveal that both cyclopentadienyl groups in the ferrocenyl donor contribute
to the electron donating ability. The first-order hyperpolarizabilities â of these chromophores, measured by
hyper-Rayleigh scattering (HRS) relative to p-nitroaniline are reported. These â values are compared to
those calculated by density functional theory (DFT). The excellent agreement between the theoretical and
experimental â values demonstrates that a linear relation exists between the hyperpolarizability and the
bond length alternation. An electrooptic coefficient, r33, of ∼25 pm/V at 1300 nm, for compound 4,
incorporated into a polymer matrix, is competitive with organic chromophores. Moreover, this r33 is more
than 30 times larger than the previously reported value for an organometallic chromophore in a poled
polymer matrix. This work not only underscores the potential for Fc donor moieties, which have been
underutilized, but also demonstrates that experimental characterization and theoretical simulations are now
congruent, viable methods for assessing potential performance of NLO materials.
redox switching ability.4 Moreover, the Fc chromophore is
bulkier than typical organic NLO donor groups, making the Fc
Introduction
Second-order nonlinear optical chromophores with a ferro-
cenyl (Fc) group as the electron donor have been of considerable
interest.1 The molecular hyperpolarizability of Fc chromophores
has been studied by electric field-induced second harmonic
generation (EFISH) and hyper-Rayleigh scattering (HRS).2 A
Fc chromophore with µâ as large as 11200 × 10-48esu has been
achieved.2b The potentially attractive NLO properties are
coupled with good thermal and photochemical stability3 and
moiety a donor with distinct properties, resulting in novel
chromophore architectures.
Wright et al.5 studied second harmonic generation of main-
chain, side-chain, and guest-host polymers of Fc chromophores.
A second harmonic intensity (SHG) 4 times that of y-cut quartz
crystal was measured. Marder et al.6 studied the electrooptic
effect of a Fc chromophore in a polymer matrix. Specifically,
1-ferrocenyl-4-E,E(4-nitrophenyl)butadiene was doped in poly-
(methyl methacrylate) (PMMA) at 2% molar loading; after
† Department of Chemistry.
poling at 100 MV/m the resulting film demonstrated an r33
)
‡ Department of Materials Science and Engineering.
0.8 pm/V as measured at 820 nm. This is the first and only
report of an r33 for an organometallic chromophore in a polymer
matrix. Since then, presumably because of such poor perfor-
mance, the Fc moiety has not been actively used in electrooptical
materials.
(1) Di Bella, S. Chem. Soc. ReV. 2001, 30, 355. Barlow, S.; Marder, S. R.
Chem. Commun. 2000, 1555. Heck, J. Dabek, S.; Meyer-Friedrichsen;
Wong, H. Coord. Chem. ReV. 1999, 190-192, 1217. Verbiest, T.;
Houbrechts, S.; Kauranen, M.; Clays, K.; Persoons, A. J. Mater. Chem.
1997, 7, 2175. Long, N. J. Angew. Chem., Int. Ed. Engl. 1995, 34, 21.
Kanis, D. R.; Ratner, M. A.; Marks, T. J. Chem. ReV. 1994, 94, 195.
(2) For example: (a) Janowska, I.; Zakrzewski, J.; Nakatani, K.; Delaire, J.
A.; Palusiak, M.; Walak, M.; Scholl, H. J. Organomet. Chem. 2003, 675,
35. (b) Alain, V.; Blanchard-Desce, M.; Chen, C.; Marder, S. R.; Fort, A.;
Barzoukas, M. Synth. Met. 1996, 81, 133. (c) Blanchard-Desce, M.; Runser,
C.; Fort, A.; Barzoukas, M.; Lehn, J.-M.; Bloy, V.; Alain, V. Chem. Phys.
1995, 199, 253. (d) Calabrese, J. C.; Cheng, L. T.; Green, J. C.; Marder,
S. R.; Tam, W. J. Am. Chem. Soc. 1991, 113, 7227.
(3) Rosenbulm, M. Chemistry of the Iron Group Metallocenes; Weiley: New
York, 1965. Harwood, J. W. Industrial Applications of Organometallic
Compounds; Reinhold: New York, 1963. Johnson, J. C., Jr. Metallocene
Technology; Noyes Data Corp.: Park Ridge, NJ, 1973. Neuse, E. W.;
Woodhouse, J. R.; Montaudo, G.; Puglis, C. Appl. Organomet. Chem. 1988,
2, 53.
(4) (a) Malaun, M.; Reeves, Z. R.; Paul, R. L.; Jeffery, J. C.; McCleverty, J.
A.; Ward, M. D.; Asselberghs, I.; Clays, K.; Persoons, A. Chem. Commun.
2001, 49. (b) Asselberghs, I.; Clays, K.; Persoons, A.; McDonagh, A. M.;
Ward, M. D.; McCleverty, J. A. Chem. Phys. Lett. 2003, 368, 408. (c)
Marder, S. R.; Cheng, L. T.; Tiemann, B. G.; Friedli, A. C.; Blanchard-
Desce, M.; Perry, J. W.; Skindhoej, J. Science 1994, 263, 511.
(5) Wright, M. E.; Toplikar, E. G.; Lackritz, H. S.; Kerney, J. T. Macromol-
ecules 1994, 27, 3016. Wright, M. E.; Toplikar, E. G.; Kubin, R. F.; Seltzer,
M. D. Macromolecules 1992, 25, 1838.
(6) Alain, V.; Fort, A.; Barzoukas, M.; Chen, C. T.; Blanchard-Desce, M.;
Marder, S. R.; Perry, J. W. Inorg. Chim. Acta 1996, 242, 43.
9
2758
J. AM. CHEM. SOC. 2005, 127, 2758-2766
10.1021/ja0449123 CCC: $30.25 © 2005 American Chemical Society

Structure−Property Relationship of Optical Chromophores
A R T I C L E S
Growing noncentrosymmetric crystals is another approach to
achieve electrooptically active materials. Most published crystal
structures of Fc chromophores2a,4a,7,8 have centrosymmetrical
structures. When coupled to strong acceptors, the chromophores
tend to pack antiparallel because of the strong dipole-dipole
interactions. Occasionally a ferrocenyl-based chromophore will
generate a crystal structure that is noncentrosymmetric;9 how-
ever, such occurrences are rare. Therefore, doping or linking
Fc chromophores into a polymer matrix followed by poling to
form an NLO active material presents a more practical and
reliable approach to realizing noncentrosymmetric NLO materi-
als.
Figure 1. UV-vis spectra of 1-4.
to study the effects of acceptor strength. First, to clearly compare
the acceptor strength, the two acceptors must have similar
structures. Also, to isolate the acceptor effect from the bridge
effect, it is preferred that the functional group that increases
the acceptor strength is not conjugated to the π system. Third,
the effects should be strong and unambiguous. Comparing the
electron-withdrawing ability of CF3PhTCF acceptor 6 with that
of the TCF acceptor 5 (see below), it is seen that 6 is enhanced
by a nonconjugated CF3 group. These two acceptors, 5 and 6,
fulfill all three requirements and are good models for studying
the acceptor effect. The electrooptic properties of amorphous
polycarbonate doped with 3 and 4 is also studied for comparison.
Results and Discussion
In addition to the electron donor, the electron acceptor can
have a pronounced effect on the molecular hyperpolarizability.
In recent years, a number of strong electron acceptors have been
developed to increase the hyperpolarizability of NLO chro-
mophores. Among those are the 2-dicyanomethylene-3-cyano-
4-methyl-2,5-dihydrofuran (TCF) type acceptors. Chromophores
containing the TCF acceptor have been used to achieve sub-1-
volt halfwave voltage polymeric electrooptic modulators10 and
an r33 electrooptic coefficient higher than 45 pm/V.11 Mating
acceptors with distinct properties to the Fc donor may provide
a new strategy to develop even better NLO materials. Therefore,
we present the synthesis, crystal structures, and experimental
and theoretical characterization of some Fc chromophores
coupled to the TCF-type acceptor, 1-4. This series of chro-
mophores is suitable for isolating the effects of increasing
acceptor strength and π bridge length conjugation. The effects
of longer π bridges have been studied extensively.2 However,
it is not straightforward to design a series of model compounds
Synthesis. Chromophores 1-4 were synthesized as shown
in Scheme 1. The TCF acceptor 5 was synthesized following
literature methods.12 Its derivative acceptor 6, CF3PhTCF, was
prepared by a three-step synthesis, illustrated in Scheme 1.
Compound 6 had been synthesized by He, et al.13 However,
we were not able to repeat the last step of their procedure.
Instead, we utilized a microwave-assisted condensation method
developed by Liu et al.11 and used lithium ethoxide instead of
sodium ethoxide as the catalyst. This microwave-assisted
synthesis produced 6 with 42% yield. The synthesis of Fc
compounds 7 and 8 followed literature methods except that
diethyl-thiophenyl-phosphonate instead of triphenyl-thiophenyl-
phosphonium bromide was used as the starting material in the
Wittig reactions.8 Knoevenagel condensation of the aldehydes
with the acceptors yielded the chromophores. The condensation
of the Fc aldehydes with CF3PhTCF, acceptor 6, in ethanol was
extremely efficient. No base or heating was necessary to
accelerate the reaction.
(7) For example: Shin, D. M.; Lee, I. S.; Chung, Y. K. Eur. J. Inorg. Chem.
2003, 12, 2311. Chiffre, J.; Averseng, F.; Balavoine, G. G. A.; Daran, J.-
C.; Iftime, G.; Lacroix, P. G.; Manoury, E.; Nakatani, K. Eur. J. Inorg.
Chem. 2001, 9, 2221. Balavoine, G. G. A.; Daran, J.-C.; Iftime, G.; Lacroix,
P. G.; Manoury, E.; Delaire, J. A.; Maltey-Fanton, I.; Nakatani, K.; Di
Bella, S. Organometallics 1999, 181, 21.
UV-Visible Absorbance Spectra. The spectra of the Fc
chromophores show two strong transitions in the UV-vis region
(as shown in Figure 1). We found that increasing either the
acceptor strength or the length of the π bridge caused a red-
shift for both the higher energy transition (HE) band and the
lower energy transition (LE) band. However, the strength of
(8) Thomas, K. R. J.; Lin, J. T.; Wen, Y. S. J. Organomet. Chem. 1999, 575,
301.
(9) Green, M. L. H.; Marder, S. R.; Thompson, M. E.; Bandy, J. A.; Bloor,
D.; Kolinsky, P. V.; Jones, R. J. Nature 1987, 330, 360.
(10) Shi, Y.; Zhang, C.; Zhang, H.; Bechtel, J. H.; Dalton, L. R.; Robinson, B.
H.; Steier, W. H. Science 2000, 288, 119.
(11) Liu, S.; Haller, M. A.; Ma, H.; Dalton, L. R.; Jang, S.-H.; Jen, A. K.-Y.
AdV. Mater. 2003, 15, 603.
(12) Melikian, G.; Rouessac, F. P.; Alexandre, C. Synth. Commun. 1995, 25,
3045.
(13) He, M.; Leslie, T. M.; Sinicropi, J. A. Chem. Mater. 2002, 14, 2393.
9
J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005 2759

A R T I C L E S
Scheme 1
Liao et al.
Table 1. UV-Vis Data on the HE and LE Bands of 1-4 (in
and the LUMO. The π orbital is formed from a combination of
the highest occupied cyclopentadienyl orbital and the highest
occupied π bridge orbital. The LUMO is largely localized on
the acceptor with a small contribution from the bridge; thus, it
represents the characteristics of the acceptor. The HOMOs
correspond to the nonbonding nearly degenerate dz /dx -y /dxy
orbitals of the metal. The idealized model predicts that changing
acceptor strength will affect the HE and LE bands equally,
whereas changing the conjugated bridge will affect only the
HE.
chloroform)
compound
λ
max [nm] (
ꢀ
[M-1 cm-1])
hν [eV]
1
2
3
4
424 (30058), 630 (11908)
442 (29001), 684 (13090)
522 (34107), 664 (20464)
562 (36225), 738 (23760)
2.92, 1.97
2.81, 1.81
2.38, 1.87
2.21, 1.68
2
2
2
the acceptor shifts the LE more than the HE, while the length
of the π bridge shifts the HE more than the LE. This behavior
is clearly demonstrated by comparing 2 with 3 (Table 1).
Chromophore 2 has a stronger acceptor, while 3 has a longer
conjugated bridge. UV-vis spectra showed that the λmax of the
HE band of 2 is 79 nm shorter than that of 3, while for the LE
band, λmax of 2 is 21 nm longer than that of 3. When the
conjugated bridge gets longer, the acceptor does not preferen-
tially shift the LE more than the HE. From 1 to 2, the red-shift
of the LE band (0.16 ev) is about 50% more than that of the
HE band (0.11 ev), while from 3 to 4, the red-shift of the LE
band (0.19 ev) is only a little larger than that of the HE band
(0.17 ev). The UV-vis data of 1-4 are listed in Table 1, and
the spectra are in Figure 1.
As previously mentioned, compounds 1-4 are excellent
model compounds for the study of the effect of both the acceptor
and the conjugated system. Different models have been proposed
to assign the transitions and explain the effects of acceptors
and conjugated bridges2d,14 on NLO activity. A recent model
developed by Barlow et al.15 assigns the LE band to a transition
between a set of degenerate HOMOs and a LUMO, and the
HE band to a transition between an occupied orbital labeled π
Our results deviate from the prediction of the idealized model.
The interaction between the conjugated bridge and the metal-
locene and the interaction between the acceptor and the
conjugated bridge must both be considered to correctly predict
the effects. When the acceptor strength increases, the energy
of the π orbital will decrease with the decrease of the LUMO,
while the HOMO will remain the same. Therefore, the decrease
of the energy gap between the LUMO and the π orbital will be
smaller than between that of the LUMO and HOMO. That is
the reason that the acceptor strength affects the LE band more
than the HE band. When the length of the conjugated bridge
increases, the acceptor will affect the π orbital relatively less,
and the two gaps (HOMO-LUMO and π-LUMO) will become
closer. Simple Hu¨ckel arguments predict that an increase in
conjugation length will increase the energy of the π orbital.
Because of the interactions of the bridge with the ferrocene,
this will modestly increase the HOMO. Given the electron-
donating nature of the thiophene bridge, the interaction between
the bridge with the acceptor will also increase the LUMO which
is mostly localized on the acceptor. The combination of these
two effects will not significantly change the gap between the
LUMO and the HOMO. Therefore, changing the conjugated
(14) Kanis, D. R.; Ratner, M. A.; Marks, T. J. J. Am. Chem. Soc. 1990, 112,
8203. Kanis, D. R.; Ratner, M. A.; Marks, T. J. J. Am. Chem. Soc. 1992,
114, 10338.
9
2760 J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005

Structure−Property Relationship of Optical Chromophores
A R T I C L E S
Figure 4. ORTEP labeling (50%) of 3. (CCDC 249700. Some selected
bond distances are listed in ref 16.)
Figure 2. ORTEP labeling (50%) of 1. (CCDC 249699. Some selected
bond distances are listed in ref 16.)
Figure 5. ORTEP labeling (50%) of 4. (CCDC 249702. Some selected
bond distances are listed in ref 16.)
Table 2. Average Bond Lengths (ABL) of the Two Cp
Rings in 1-4
compound
ABL of Cp1 (Å)
ABL of CP2 (Å)
1
2
3
4
1.417
1.416
1.425
1.413
1.407
1.391
1.417
1.411
to the bridge, significantly contributes to the electron-donating
ability (i.e. whether the charge-transfer interaction passes
through the metal). We examined the average bond length
(ABL) of the two Cp rings of 1-4. The Cp that is directly linked
to the acceptor through the π bridge is labeled Cp1, and the
other is labeled Cp2. The ABL of Cp2 is more sensitive than
that of Cp1 toward the electron-withdrawing strength applied
to the ferrocene. The ABL of the two Cp rings in 1-4 are listed
in the Table 2. The ABL of Cp1 does not significantly change
for 1-4. The ABL of Cp2 clearly depends on the electron-
withdrawing strength applied to the ferrocene, i.e. the strength
and the distance of the acceptor. The stronger the electron-
withdrawing strength on the ferrocene, the smaller is the ABL
of Cp2. For example, the ABL of 2 is smaller than that of 1,
while the ABL of 1 is smaller than that of 3 because the acceptor
is closer to the ferrocene in 1 than in 3. Moreover, in all four
chromophores the ABL of Cp1 is larger than Cp2. This evidence
suggests that charge transfer interactions significantly pass
through the iron to Cp2, suggesting that an improved design of
Cp2 might improve NLO performance.
Strong nonlinear optical chromophores generally have large
dipole moments leading to large dipole-dipole interactions
between chromophores. Normally such strong interactions
induce antiparallel packing that reduces the optical anisotropy
of the bulk material. Hence, noncentrosymmetric crystals of
strong NLO chromophores are rare. These intermolecular
interactions are an important factor in determining the elec-
trooptic efficiency in material.10,17 In the four crystals, the
neighboring chromophores align antiparallel; but the dimeriza-
tion patterns are quite different.
Figure 3. ORTEP labeling (50%) of 2. (CCDC 249701. Some selected
bond distances are listed in ref 16.)
tobridge structure or length will affect the HE band more than
the LE band.
Crystal Structures. Single-crystal X-ray diffraction provides
a reliable measure of molecular structure, providing information
for the study of the structure-property relationship and inter-
molecular interactions among chromophores. Crystals of all four
chromophores were obtained by slow diffusion of a heptane
layer into a concentrated solution of the corresponding chro-
mophore in dichloromethane. All four compounds form cen-
trosymmetric crystal structures. Compound 1 forms a monoclinic
P lattice. The space group was found to be P21/c. The crystal
structure of chiral compound 2 revealed two enantiomers
cocrystallized in a triclinic lattice. The space group was found
to be P1h. A disordered solvent molecule, CH2Cl2, was found in
each unit cell. The presence of the solvent causes the two
enantiomers inside the asymmetric unit to be slightly different.
The inversion symmetry adds two more molecules to complete
the unit cell. The whole cell consists of two pairs of enantiomers
and one disordered CH2Cl2. Compound 3 also forms a mono-
clinic P lattice. The space group is P1h. Four solvent molecules,
CH2Cl2, were found in a unit cell together with two chro-
mophores. Compound 4 also forms cocrystals of two enanti-
omers in a monoclinic C lattice. The space group was found to
be C2/c. The solved molecular structures are given in Figures
2-5.
For designing better metallocene chromophores, it is impor-
tant to know whether cyclopentadiene (Cp) not directly linked
Dimerization is one of the most important aspects of
antiparallel packing of NLO chromophores. In a recent paper,
9
J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005 2761

A R T I C L E S
Liao et al.
Wurthner, Wortmann, et al.18 carefully examined the dimeriza-
tion of merocyanine dyes. Centrosymmetric dimers formed in
all the investigated merocyanines in concentrated solutions and
in the solid state. Similarly, here all four Fc chromophores form
dimers in the crystalline state with 1, 2, and 4 forming fully
overlapped antiparallel dimers. The acceptor moiety of one
molecule is adjacent to the Cp ring that links with the π bridge
in the other molecule. Chiral compound 2 forms a centrosym-
metric dimer of a pair of enantiomers. However, 4 forms a
noncentrosymmetric dimer of the same enantiomers, which
means there are two types of dimers with different chirality.
The dimer of the S enantiomer is shown in Figure 6.
Compound 3 has a unique dimer structure (Figure 7). The
acceptor group partially covers the thiophenyl ring rather than
Figure 6. The asymmetric dimer of the S enantiomer of 4.
(15) Barlow, S.; Bunting, H. E.; Ringham, C.; Green, J. C.; Bublitz, G. U.;
Boxer, S. G.; Perry, J. W.; Marder, S. R J. Am. Chem. Soc. 1999, 121,
3715.
(16) Selected bond distances of 1 (Å) (Figure 2): Fe1-C4 2.017(5), Fe1-C2
2.021(5), Fe1-C6 2.027(5), Fe1-C7 2.027(5), Fe1-C3 2.031(5), Fe1-
C1 2.032(4), Fe1-C8 2.032(5), Fe1-C10 2.036(5), Fe1-C5 2.038(5),
Fe1-C9 2.040(5), O1-C15 1.331(5), O1-C16 1.494(4), N1-C19 1.141-
(5), N2-C21 1.146(6), N3-C22 1.132(5), C13-C14 1.347(5), C13-C12
1.426(5), C13-C16 1.517(5), C14-C19 1.421(6), C14-C15 1.451(5), C3-
C4 1.370(7), C3-C2 1.410(6), C1-C2 1.423(6), C1-C5 1.435(6), C1-
C11 1.446(5), C16-C17 1.520(6), C16-C18 1.528(6), C15-C20 1.362(6),
C11-C12 1.344(6), C21-C20 1.395(6), C5-C4 1.429(6), C20-C22
1.441(6), C10-C9 1.400(7), C10-C6 1.417(7), C8-C7 1.403(6), C8-
C9 1.411(7), C7-C6 1.405(7). Selected bond distances of 2 (Å) (Figure
3): S enantiomer (the top molecule in Figure 3): Fe1-C13 2.019(6), Fe1-
C20 2.036(6), Fe1-C18 2.038(5), Fe1-C12 2.039(5), Fe1-C16 2.045-
(5), Fe1-C14 2.046(6), Fe1-C19 2.046(5), Fe1-C15 2.054(6), Fe1-C17
2.061(6), Fe1-C21 2.067(6), F1-C5 1.341(6), F2-C5 1.349(6), F3-C5
1.335(6), O1-C1 1.355(6), O1-C4 1.455(6), N1-C7 1.139(7), N2-C8
1.153(7), N3-C9 1.145(7), C1-C6 1.357(7), C1-C2 1.427(7), C2-C3
1.360(7), C2-C9 1.446(8), C3-C10 1.416(7), C3-C4 1.536(7), C4-C5
1.549(8), C6-C8 1.423(8), C6-C7 1.439(8), C10-C11 1.347(7), C11-
C12 1.443(7), C12-C16 1.448(8), C12-C13 1.453(7), C13-C14 1.386-
(7), C14-C15 1.405(8), C15-C16 1.415(7), C17-C21 1.406(7), C17-
C18 1.410(7), C18-C19 1.399(7), C19-C20 1.424(7), C20-C21 1.432(7);
R enantiomer (the bottom molecule in Figure 3): Fe2-C37 2.017(6), Fe2-
C41 2.021(7), Fe2-C40 2.022(6), Fe2-C34 2.023(5), Fe2-C38 2.031-
(6), Fe2-C33 2.033(6), Fe2-C39 2.033(6), Fe2-C42 2.033(7), Fe2-C35
2.034(5), Fe2-C36 2.041(5), F4-C26 1.330(6), F5-C26 1.339(6), F6-
C26 1.336(6), O2-C22 1.352(6), O2-C25 1.457(6), N4-C28 1.148(7),
N5-C29 1.136(7), N6-C30 1.148(7), C22-C27 1.335(7), C22-C23
1.448(7), C23-C24 1.379(7), C23-C30 1.437(8), C24-C31 1.425(7),
C24-C25 1.537(7), C25-C51 1.522(7), C25-C26 1.530(8), C27-C29
1.442(8), C27-C28 1.445(8), C31-C32 1.340(6), C32-C33 1.438(7),
C33-C37 1.419(7), C33-C34 1.425(7), C34-C35 1.419(7), C35-C36
1.415(8), C36-C37 1.402(8), C38-C39 1.370(9), C38-C42 1.400(9),
C39-C40 1.397(8), C40-C41 1.370(9), C41-C42 1.417(10). Selected
bond distances of 3 (Å) (Figure 4): Fe1-C22 2.034(3), Fe1-C21 2.037-
(3), Fe1-C23 2.037(3), Fe1-C27 2.038(3), Fe1-C26 2.041(3), Fe1-C20
2.044(3), Fe1-C25 2.045(3), Fe1-C24 2.049(3), Fe1-C28 2.054(3), Fe1-
C19 2.058(3), S1-C16 1.733(3), S1-C13 1.737(3), O1-C1 1.328(3), O1-
C4 1.482(3), N1-C6 1.146(4), N2-C7 1.149(4), N3-C8 1.152(4), C1-
C5 1.370(4), C1-C2 1.439(4), C2-C3 1.375(4), C2-C6 1.423(4), C3-
C11 1.423(4), C3-C4 1.509(4), C4-C9 1.509(4), C4-C10 1.523(4), C5-
C8 1.423(4), C5-C7 1.429(4), C11-C12 1.352(4), C12-C13 1.431(4),
C13-C14 1.381(4), C14-C15 1.398(4), C15-C16 1.383(4), C16-C17
1.437(4), C17-C18 1.340(4), C18-C19 1.446(4), C19-C20 1.435(4),
C19-C23 1.442(4), C20-C21 1.415(4), C21-C22 1.419(5), C22-C23
1.416(4), C24-C25 1.416(4), C24-C28 1.420(5), C25-C26 1.421(4),
C26-C27 1.419(5), C27-C28 1.410(5). Selected bond distances of 4 (Å)
(Figure 5): Fe1-C2 2.016(5), Fe1-C4 2.025(5), Fe1-C3 2.028(6), Fe1-
C5 2.034(5), Fe1-C10 2.037(6), Fe1-C9 2.039(6), Fe1-C1 2.041(5),
Fe1-C8 2.045(6), Fe1-C6 2.049(6), Fe1-C7 2.050(6), S1-C13 1.725-
(5), S1-C16 1.728(5), F1-C23 1.346(5), F2-C23 1.340(5), F3-C23
1.351(6), O1-C21 1.347(6), O1-C22 1.461(6), N1-C30 1.168(7), N2-
C32 1.165(7), N3-C33 1.158(8), C1-C2 1.417(7), C1-C5 1.429(7), C2-
C3 1.413(8), C3-C4 1.395(9), C4-C5 1.409(7), C6-C10 1.389(8), C6-
C7 1.411(9), C7-C8 1.413(9), C8-C9 1.426(9), C9-C10 1.415(8), C11-
C12 1.336(7), C12-C13 1.443(7), C13-C14 1.387(7), C14-C15 1.380(7),
C15-C16 1.385(7), C16-C17 1.437(7), C17-C18 1.354(7), C18-C19
1.413(7), C19-C20 1.390(7), C19-C22 1.519(7), C20-C30 1.402(8),
C20-C21 1.444(7), C21-C31 1.352(7), C22-C24 1.493(7), C22-C23
1.525(8), C31-C32 1.422(8), C31-C33 1.442(9)
Figure 7. The dimer of 3.
Scheme 2. Model for an Antiparallel Dimer of Two Dipoles.17 (d is
the distance between planes and R Is the Slip Angle that Results
from the Translational offset.)
the Fc donor. For a simple model of a centrosymmetric dimer,
see Scheme 2. The dimerization energy may be calculated from
the equation17 for two antiparallel dipoles
NA
∆Edim ) -
µ2gM(1 - 3 cos2 R)sin3 R
(1)
4πꢀ0d3
where NA is Avogadro’s constant, ꢀ0 is the dielectric constant
of the vacuum, d is the distance between planes, defined in
Scheme 2, and R is the slip angle that results from the
translational offset. For a given d, the absolute value of the
dimerization energy increases as R f π/2. The distance between
the two π planes in the dimer of 3 is about the same as that of
the other three compounds (d ≈ 3.4 Å). If only the electrostatic
interaction is considered, the inclined dimer of 3 should be less
stable than the fully overlapped conformation. Apparently,
crystal packing forces dominate the electrostatic interaction
between the dimers. It is not clear why only 3 forms this dimer
structure.
First Hyperpolarizabilities Measured by HRS. The first-
order hyperpolarizabilities of 1-4 and p-nitroaniline (pNA) were
measured by HRS using the solvent (chloroform) as internal
reference. The excitation wavelength, λ ) 1000 nm, was chosen
to reduce the self-absorption of the scattered light by the
chromophores and to minimize the impact of resonant enhance-
ment on the measured hyperpolarizabilities. Self-absorption of
the scattered light was of particularly importance in the case of
1, where the low hyperpolarizability of this chromophore
(17) Robinson, B. H.; Dalton, L. R. J. Phys. Chem. A 2000, 104, 4785-4795.
Dalton, L. R.; Steier, W. H.; Robinson, B. H.; Zhang, C.; Ren, A.; Garner,
S.; Chen, A.; Londergan, T.; Irwin, L.; Carlson, B.; Fifield, L.; Phelan, G.;
Kincaid, C.; Amend, J.; Jen, A. K.-Y. J. Mater. Chem. 1999, 9, 1905-
1920.
(18) Wurthner, F.; Yao, S.; Debaerdemaeker, T.; Wortmann, R. J. Am. Chem.
Soc. 2002, 124, 9431.
9
2762 J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005

Structure−Property Relationship of Optical Chromophores
A R T I C L E S
Table 3. Relative Hyperpolarizability of 1-4 Measured by HRS
accuracy of DFT-based computations for properties, such as
electronic transition frequencies, that are associated with the
excited states. All calculations were spin restricted, as ferrocene
is known to have spin zero. The DMol3 code returns the energy
and dipole vector for each field setting; the 13 dipole vectors
were analyzed using the technique of Sim et al.24 to extract the
permanent dipole vector (the null field result), the polarizability,
and the hyperpolarizability. The value of the angular averaged
hyperpolarizability, measured by HRS, was calculated using the
equations derived by Cyvin, Rauch, and Decius.26
compound
1
2
3
4
pNA
â/âchloroform 279 ( 31 450 ( 148 2633 ( 340 3333 ( 202 79 ( 7
necessitated the use of elevated chromophore concentrations.
It has been demonstrated that high chromophore concentrations,
leading to increasingly significant self-absorption, can attenuate
the measured hyperpolarizability values.19 Compound 1 dem-
onstrates relatively modest extinction at 1000 and 500 nm;
therefore, this excitation wavelength was chosen to minimize
any attenuation of scattered light. Furthermore, for 1, this
excitation wavelength should provide minimal resonant en-
hancement, allowing for direct comparison to pNA. For com-
pounds 2-4, resonant enhancement may result in elevated â
values. The extent of enhancement can be determined by
measuring the excitation-wavelength dependence of the HRS
intensity, and such experiments are currently underway. The
details of the HRS experiment are provided in the Experimental
Section.
The chromophore â values, reported relative to âchloroform, can
be converted to the absolute values using pNA as the standard.
The âHRS of pNA at 1064 nm in chloroform was measured by
Shelton et al.20 to be 22.7 × 10-30esu. Using the relative â of
the Fc chromophores, reported here, and a measured value of
âHRS pNA, the absolute values may be calculated (see Table
3).
DFT Calculations of the First Hyperpolarizability. DFT
calculations were utilized to study the molecular hyperpolariz-
ability. All calculations were performed with DMol,3 a density
functional code developed by Freeman, Delley, and co-work-
ers.21 Preliminary calculations using a variety of generalized
gradient approximation (GGA) functionals on p-nitroaniline
(pNA) showed that two of the most recent functionals, that of
the Perdew, Burke, Ernzerhof22 (PBE) and the revised version
of this functional, the RPBE of Hammer, et al.,23 gave results
in excellent agreement with the MP2 calculations of Sim, et
al.24 In particular, Sim, et al., determined the dipole moment of
pNA to be in the range 6.87-7.82 D, calculated with MP2 and
their most extensive basis set, with variations in the calculated
moment depending on whether the geometry of the (constrained)
planar molecule was optimized with HF or MP2. Our calcula-
tions were done primarily with the PBE functional, with two
structures being calculated with RPBE for comparison. Both
the PBE and RPBE functionals yield â for pNA in excellent
agreement with experimental gas-phase EFISH measurements.25
The energies and dipole vectors of a given molecule were
calculated with zero field, together with 12 different field E
settings of (0.001 au and (0.002 au along the three coordinate
axes. Therefore, the polarizabilities so computed are ground-
state properties of the molecule. No claim is made for the
Dixon et al. calculated second-order hyperpolarizabilities of
metallocenes using DFT .27 The feasibility of calculating the
first-order hyperpolarizability by DFT was verified by calculat-
ing the µâ and âzzz of 7-9, which were studied by Alain et
al.2b who measured the µâ by electric-field-induced second-
harmonic generation (EFISH). The results of our calculations
are listed in Table 4. Our results showed that the DFT calculation
of the Fc chromophores yielded lower µâ values than the
experimental values (by a factor of approximately 3, in this
series of Fc chromophores). This is at least partially due to the
fact that DFT calculates the molecules in the gas phase while
the experimental values were taken in the solution phase. In
fact, the â value of pNA measured by EFISH in gas phase is
9.26 × 10-30 esu, while that in solution phase is 33.6 × 10-30
esu.20,28 When the calculated values are compared with the
experimental values of 7-9, not only the trends but also the
ratios of the hyperpolarizabilities are well predicted. This can
best be seen by comparing the normalized values of experi-
mental and calculated µâ and âzzz in Table 4. Therefore, DFT
calculations are suitable for systematically studying the present
series of Fc chromophores.
We calculated both âHRS and âzzz using the molecular
structures obtained from the X-ray analysis. These calculations
have been done with the PBE/dnp Hamiltonian/basis set
combination. The results are listed in Table 5. As mentioned
above, due to the difference between the gas-phase and solution-
phase â, directly comparing the DFT calculated values and the
experimental values measured in solution is inappropriate. As
a result we used pNA as a standard. We found that DFT
calculated values for the first hyperpolarizabilities relative to
pNA agreed well with the experimental values. (Table 5).
Because the crystal structures and hyperpolarizabilities of 1-4
are available, bond-length alternation (BLA) theory can be used
to analyze the structure-property relationship for these mol-
ecules. It is arbitrary whether the two double bonds and one
single bond in the TCF acceptor should be included in the
analysis. In fact, we found that the differences between the
bridge BLA and the acceptor BLA are very small for 1, 2, and
4. Only in 3 is the bridge BLA (0.088 Å) significantly larger
(19) Pauley, M. A.; Wang, C. H.; Jen, A. K.-Y. J. Chem. Phys. 1995, 102,
6400.
(20) Kaatz, P.; Shelton, D. P. J. Chem. Phys. 1996, 105, 3918.
(21) Delley, B. DMol, a Standard Tool for Density Functional Calculations:
Review and Advances. In Modern Density Functional Theory: A Tool for
Chemistry; Seminario, J. M., Politzer, P., Ed.; Elsevier Science Publ.:
Amsterdam, 1995; Vol. 2. DMol3 is a product of Accelrys, Inc., San Diego,
CA.
(22) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77, 3865.
(23) Hammer, B.; Hansen, L. B.; Norskov, J. K. Phys. ReV. B 1999, 59, 7413.
(24) Sim, F.; Chin, S.; Dupuis, M.; Rice, J. E. J. Phys. Chem. 1993, 97, 1158.
(25) Davidson, E. R., Eichinger, B. E.; Robinson, B. H. Manuscript in
preparation.
(26) Cyvin, S. J.; Rauch, J. E.; Decius, J. C. J. Chem. Phys. 1965, 43, 4083.
(27) Matsuzawa, N.; Seto, J.; Dixon, D. A. J. Phys. Chem. A 1997, 101, 9391.
(28) Kaatz, P.; Donley, E. A.; Shelton, D. P. J. Chem. Phys. 1998, 108, 849.
9
J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005 2763

A R T I C L E S
Liao et al.
Table 4. Experimental and Calculated Hyperpolarizabilities of 7-9
experimental
µâ/10-48 esu (normalized valueb)
DFT
DFT
molecule
na
µâ/10-48 esu (normalized valueb)
â
zzz/10-30esu (normalized valueb)
7
8
9
1
2
3
92 (0.082)
420 (0.37)
1120 (1.0)
28.6 (0.073)
125 (0.32)
392 (1.0)
4.18 (0.090)
16.5 (0.36)
41.2 (1.0)
a Number of the double bonds in the bridge. b Values are normalized using the largest value in the corresponding column as 1.0.
Figure 8. Bonds used for calculating BLA. (Highlighted bonds with label
“b” are used for calculating the bridge BLA and “a” for acceptor BLA.)
Table 5. Experimental and DFT Calculated Data of 1-4
experimental âHRS
DFT âHRS
/10-30esu
DTF âzzz
/10-30esu
molecule
â/
âCHCl3
â/
âpNA
â
â/
âpNA
â
â/
âpNA
pNA
79 ( 7
1.0
3.5
5.7
33.3
42.4
3.08
14.34
14.75
110.1
134.1
1.0
4.6
4.8
35.7
43.5
7.20
17.39
26.66
82.34
318.1
1.0
2.4
3.7
11.4
44.2
1
2
3
4
279 ( 31
450 ( 148
2633 ( 340
3333 ( 202
Figure 9. Linear relationship between BLA and first-order hyperpolariz-
ability. (The data are fit by a linear equation of âzzz ) 19.1 × (0.0925-
BLA) × 10-27 esu with a linear correlation coefficient of R ) 0.9993.)
The data lie within the region of linearity predicted by Marder et al.
Table 6. BLA Values of 1-4
The molecular structure of 4 is unique: In 1-3, all the ene
structures adopt a trans conformation (as expected). While in
4, the diene structure C17-C18-C19-C20 has a cis conforma-
tion (Figure 5). To understand whether this is due to the crystal
packing, we calculated the energies of cis and trans conforma-
tions in the gas phase with DMol3. In both cases, optimized
geometries were used. Indeed, the cis conformation is 25.4 kcal/
mol more stable than the trans conformation. For comparison,
cis and trans conformations of 3 were also calculated. In this
case, the trans conformation is 22 kcal/mol more stable. Our
results are consistent with the crystal structure. This suggests
that the cis conformation of 4 is not solely the result of crystal
packing forces.
compound
bridge BLA (Å)
acceptor BLA (Å)
average BLA (Å)
1
2
3
4
0.092
0.091
0.088
0.076
0.097
0.092
0.067
0.073
0.095
0.091
0.078
0.074
than the acceptor BLA (0.067 Å). This might correlate with
the unique dimerization conformation of 3 discussed above. The
bonds used for calculating BLA are shown in Figure 8. The
bridge BLA, the acceptor BLA, and the average BLA are listed
in Table 6. The BLA values were compared with hyperpolar-
izability âzzz calculated by DFT as described below.
The correlation between the hyperpolarizabilities and BLA
of 1-4 was studied. The bridge BLA was used to avoid the
effect of dimerization discussed above. Because the BLA is
related to the bonds along the long axis of the molecule,
theoretical values of âzzz in Table 6 were used. A linear
relationship is obtained (Figure 9). A quasi-linear relationship
between âzzz and BLA values over the range from 0.05 to 0.1
Å was predicted by Marder et al. more than 10 years ago and
has since been a guide for designing nonlinear optical chro-
mophores.29 However, because of the difficulty of obtaining
crystals suitable for X-ray analysis, the previous systematic
studies were based on calculated BLA values. Also, the strengths
of donor and acceptor groups were modeled by an external
field.30 Our results showed that in a well-designed system, such
a linear relationship does exist even when chromophores with
different acceptors are compared.
Electrooptic Property. Compound 4, with a strong acceptor,
a long bridge, and a chiral center, is expected to have a large
electrooptic coefficient. A guest-host system of 4 in amorphous
polycarbonate (APC) with 20 wt % loading was prepared by
spin-coating a solution of 4 and APC in cyclopentanone on a
glass slide possessing an indium-tin-oxide (ITO) overlayer. The
samples were spin-coated at a spread speed of 500 rpm and a
20 s spin at 700 rpm. The film was then baked overnight at 85
°C. The film was contact poled between the temperatures of
140 °C and 150 °C at 1.0 MV cm-1 for 5 min. The resulting
r
33 value was 25 pm V-1 at 1300 nm. For comparison, a guest-
host system of 3 was also prepared and poled at the same
condition. An r33 of 5 pm V-1 was obtained at 1300 nm. The
poling conditions consisted of a current of 2 µA for 3 and 11-
14 µA for 4. Because the weight loading is the same and 3 has
a smaller molecular weight than 4, the film of 3 contained a
higher concentration of Fc groups. The result shows that, at
least in this case, the organometallic species did not cause any
unusual conduction problems in the poling process.
(29) Marder, S. R.; Gorman, C. B.; Tiemann, B. G.; Cheng, L.-T. J. Am. Chem.
Soc. 1993, 115, 3006.
(30) Marder, S. R.; Gorman, C. B.; Meyers, F.; Perry, J. W.; Bourhill, G.; Bredas,
J.-L.; Pierce, B. M. Science 1994, 265, 632.
9
2764 J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005

Structure−Property Relationship of Optical Chromophores
A R T I C L E S
the course of successive scans to monitor the onset of bulk photodeg-
radation. No measurable photodegradation was observed during these
experiments. Additionally, solution concentrations were carefully chosen
to avoid self-absorption of the scattered light by the chromophores.
No measurable self-absorption effects were observed at the concentra-
tions employed.
Data were analyzed in the MATLAB environment by fitting the HRS
emission peaks to a Gaussian functional form to determine the
intensities. Intensities were converted to â by comparing spectra of
the sample solutions to corresponding spectra of neat chloroform. â
values are expressed relative to âCHCl3, as determined through the
following equation (where I is the signal intensity and N is the number
density):
To the best of our knowledge, there is only one previously
reported value of r33 for an organometallic chromophore in a
poled film. In that work, incorporation of 1-ferrocenyl-4-E,E-
(4-nitrophenyl)butadiene into poly(methyl methacrylate) at 2%
molar loading (0.19 mmol g-1) yielded an r33 ) 0.8 pm V-1
measured at 820 nm. The loading of 4 is 20 wt % corresponding
to a loading density of 0.32 mmol‚g-1. The molar loading
density is only 68% higher than the previous one, while the
electrooptic coefficient is more than 30 times higher. Such large
increases could be due to a combination of high hyperpolariz-
ability of the chromophore, good compatibility with the polymer
matrix, and optimized poling condition.
The resulting r33 is comparable to values recently reported,11,31
using chromophores containing bulky side groups designed to
prevent aggregation.17,32 However, this strategy has a drawback
in that it limits the loading density. Compound 4 achieved a
high r33 at 20 wt % loading in the absence of bulky substituents.
The electrooptic efficiency may be further increased by adding
periphery structures, optimizing loading density, choosing a
more compatible polymer matrix, or linking the chromophore
to a polymer backbone.17
Isample Nsample âs2ample + Nsolvent â2solvent
Nsample â2sample
)
)
+ 1
(
)
Nsolvent âs2olvent
â2solvent
Isolvent
Nsolvent
Synthesis. 2-Dicyanomethylene-3-cyano-4-methyl-5-phenyl-5-per-
fluoromethyl-2,5-dihydrofuran (6).13 In a 25 mL round-bottom flask
were mixed 1.8 g of 4,4,4-trifluoro-3-hydroxy-3-phenyl-2-butanone
(8.25 mmol), 1.16 g of malonitrile (17.6 mmol), and 1 mL of absolute
ethanol. The mixture was heated in a microwave reactor for 1 h at 35
W. The reaction temperature was measured by an internal thermometer
to be between 95 and 100 °C. After the reaction, a couple of drops of
water were added to quench the reaction. The crude product was purified
by chromatography on silica gel with CH2Cl2 as the eluant to yield 1.1
Experimental Section
General. All commercially available compounds were used as
supplied. Tetrahydrofuran, dichloromethane, and diethyl ether were
distilled over drying agents under nitrogen before use. 2-Dicyanometh-
ylene-3-cyano-4-methyl-5-dimethyl-2,5-dihydrofuran (5), 4,4,4-triflu-
oro-3-hydroxy-3-phenyl-2-butanone (the precursor of 6), and [(1E)-2-
(5-formyl-2-thienyl)ethenyl]ferrocene (8), were synthesized following
literature methods.8,12,13 UV-visible spectra were recorded on a
1
g of 6 (42% yield). H NMR δ (CDCl3): 7.58-7.54 (3H, m), 7.45-
7.41 (2H, m), 2.48 (3H, s). MS (ESP): 338.05 (M + Na).
Chromophore 1. Into a mixture of 16 mL of CHCl3 and 4 mL of
THF were dissolved 0.37 g of ferrocenecarboxaldehyde (1.7 mmol)
and 0.50 g of 5 (2.5 mmol) . To this mixture was added three drops of
triethylamine. The solution was refluxed overnight and then concen-
trated under reduced pressure. The residue was chromatographed on
silica gel using 1:3 hexane/ethyl acetate as the eluant. The yield was
0.38 g (57% yield). 1H NMR δ (CDCl3): 7.68 (1 H, d, J ) 15.4 Hz),
6.50 (1 H, d, J ) 15.4 Hz), 4.85 (2 H, t, J ) 1.8 Hz), 4.70 (2 H, t,
J ) 1.8 Hz), 4.25 (5 H, s), 1.74(6 H, s). HRMS (ESP): 396.0800
(M + H), 418.0613 (M + Na). Anal. Calcd for C22H17FeN3O: C, 66.86;
H, 4.34; N, 10.63. Found: C, 66.52; H, 4.54; N, 10.56.
Chromophore 2. Into about 0.5 mL of CH2Cl2 were dissolved 0.1
g (0.317 mmol) of 6 and 0.067 g (0.313 mmol) of ferrocenecarboxal-
dehyde. To this mixture was added 2 mL of ethanol. The mixture was
heated at 60 °C for 1 h (or left at room temperature overnight). After
cooling to room temperature, the mixture was concentrated to about 1
mL. The crystalline precipitate was collected and washed with a little
ethanol. The crude product was recrystallized in a mixture of heptane
and dichloromethane to yield 116 mg of pure product (72% yield). 1H
NMR δ (CDCl3): 7.64 (1 H, d, J ) 15.5 Hz), 7.44-7.61 (5 H, m),
6.53 (1 H, d, J ) 15.5 Hz), 4.90 (2 H, s, br), 4.63(2 H, s, br), 4.16, (5
H, s). HRMS (ESP): 534.0485 (M + Na). Anal. Calcd for C27H16F3-
FeN3O‚0.25(CH2Cl2): C, 61.46; H, 3.12; N, 7.89. Found: C, 61.94;
H, 3.48; N, 7.95.
1
SHIMADZU 1601 UV spectrometer. H NMR spectra were recorded
on a Bruker AM-300. Crystals of 1-4, suitable for X-ray diffraction
analysis, were obtained by slow diffusion of a heptane layer to a
concentrated solution of the corresponding chromophore in dichlo-
romethane. Prevalere Life Sciences, Inc., performed all elemental
analyses.
Hyper-Rayleigh Scattering Measurement. Hyper-Rayleigh scat-
tering (HRS) measurements were performed to quantify the first
hyperpolarizability (â) of 1-4, and pNA. Solutions were prepared in
chloroform (Fisher: ACS Spectranalyzed). The concentrations em-
ployed were as follows: 1 (23 µM), 2 (939 nM), 3 (348 and 697 nM)
4 (220 and 330 nM), pNA (173 and 346 µM).
A mode-locked Ti:sapphire oscillator (Spectra-Physics Tsunami)
provided excitation at 1000 nm, consisting of 100-fs pulses (full-width
at half-maximum, fwhm), an 80 MHz repetition rate, and 550 mW of
incident power. The laser spectrum (fwhm ≈ 12 nm) was monitored
throughout the experiment. The incident light was focused into the
middle of a low-volume flow cell connected to an in-line 0.1-µm PTFE
filter. Scattered light was imaged onto the slits of a spectrograph (Acton
300i) through a 700-nm short-pass filter and a piece of blue-green BK7
glass. Detection was accomplished with a 1340 × 100 pixel, red-edge
enhanced, back-thinned, LN2-cooled CCD camera (Roper Scientific),
using exposure times of 240-500 s. A more detailed discussion of the
experimental apparatus is described elsewhere.33
Chromophore 3. In a mixture of 16 mL of CHCl3 and 4 mL of
THF were dissolved 0.40 g of 8 (1.2 mmol) and 0.40 of 5 (2.0 mmol).
To this mixture was added three drops of triethylamine. The solution
was refluxed overnight and then concentrated under reduced pressure.
The residue was chromatographed on silica gel using 1:1 hexane/ethyl
HRS was confirmed by a quadratic dependence of the signal intensity
on incident power. Furthermore, the spectral-width of the HRS peak
was equal to the incident field and tracked with changes in excitation
wavelength. Signal amplitudes were monitored for consistency over
1
acetate as the eluant. The yield was 0.51 g (84% yield). H NMR δ
(CDCl3): 7.76 (1 H, d, J ) 14.7 Hz), 7.31 (1 H, d, J ) 4.0 Hz), 6.93-
6.95 (2 H, m), 6.78 (1 H, d, 16.0 Hz), 6.59 (d, J ) 14.7 Hz), 4.58 (2
H, s, br), 4.50 (2 H, s, br), 4.23 (5 H, s), 1.75 (6 H, s). HRMS (ESP):
503.0740 (M + H), 526.0638 (M + Na). Anal. Calcd for C28H21FeN3-
OS: C, 66.81; H, 4.20; N, 8.35. Found: C, 66.75; H, 4.44; N, 8.44.
Chromophore 4. In about 0.5 mL of CH2Cl2 were dissolved 0.1 g
(0.310 mmol) of 8 and 0.1 g (0.317 mmol) of 6. To this mixture was
added 2 mL of ethanol. The mixture was heated at 60 °C for 1 h (or
(31) For recent reviews: Dalton, L. R.; Robinson, B. H.; Jen, A. K.-Y.; Steier,
W. H.; Nielsen, R. Opt. Mater. 2003, 21, 19. Ma, H.; Jen, A. K.-Y.; Dalton,
L. R. AdV. Mater. 2002, 14, 1339.
(32) Ma, H.; Chen, B.; Sassa, T.; Dalton, L. R.; Jen, A. K.-Y. J. Am. Chem.
Soc. 2001, 123, 986.
(33) Firestone, K. A.; Reid, P. J.; Lawson, L. R.; Jang, S.-H.; Dalton, L. R.
Inorg. Chim. Acta. 2004, 357, 3957-3966.
9
J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005 2765

A R T I C L E S
Liao et al.
left at room temperature overnight). After cooling to room temperature,
the mixture was concentrated to about 1 mL. The crystalline precipitate
was collected and washed with a little ethanol. The crude product was
recrystallized in a mixture of heptane and dichloromethane to yield
Acknowledgment. Support from the Air Force Office of
Scientific Research, the Office of Naval Research and the
National Science Foundation are gratefully acknowledged.
1
146 mg of pure product (76%). H NMR δ (CDCl3): 7.77 (1 H, d,
Supporting Information Available: X-ray crystallographic
files in CIF format with the coordinates of molecules 1-4. This
material is available free of charge via the Internet at
http://pubs.acs.org.
J ) 15.0 Hz), 7.50-7.56 (5 H, m), 7.27 (1 H, d, J ) 3.6 Hz), 7.01
(1 H, d, J ) 16.1 Hz), 6.95 (1 H, d, J ) 3.5 Hz), 6.75 (1 H, d, J )
15.0 Hz), 6.65 (1 H, d, J ) 16.1 Hz), 4.54 (2 H, s, br), 4.50 (2 H, s,
br), 4.20 (5 H, s). HRMS (ESP): 642.0508 (M + Na). Anal. Calcd for
C33H20F3FeN3OS‚0.6(CH2Cl2): C, 60.20; H, 3.19; N, 6.27. Found: C,
60.11; H, 3.66; N, 6.05.
JA0449123
9
2766 J. AM. CHEM. SOC. VOL. 127, NO. 8, 2005