Diquat derivatives: Highly active, two-dimensional nonlinear optical chromophores with potential redox switchability

Article abstract of DOI:10.1021/ja103289a

In this article, we present a detailed study of structure-activity relationships in diquaternized 2,2′-bipyridyl (diquat) derivatives. Sixteen new chromophores have been synthesized, with variations in the amino electron donor substituents, π-conjugated bridge, and alkyl diquaternizing unit. Our aim is to combine very large, two-dimensional (2D) quadratic nonlinear optical (NLO) responses with reversible redox chemistry. The chromophores have been characterized as their PF₆^- salts by using various techniques including electronic absorption spectroscopy and cyclic voltammetry. Their visible absorption spectra are dominated by intense π → π* intramolecular charge-transfer (ICT) bands, and all show two reversible diquat-based reductions. First hyperpolarizabilities β have been measured by using hyper-Rayleigh scattering with an 800 nm laser, and Stark spectroscopy of the ICT bands affords estimated static first hyperpolarizabilities β₀. The directly and indirectly derived β values are large and increase with the extent of π-conjugation and electron donor strength. Extending the quaternizing alkyl linkage always increases the ICT energy and decreases the E_1/2 values for diquat reduction, but a compensating increase in the ICT intensity prevents significant decreases in Stark-based β₀ responses. Nine single-crystal X-ray structures have also been obtained. Time-dependent density functional theory clarifies the molecular electronic/optical properties, and finite field calculations agree with polarized HRS data in that the NLO responses of the disubstituted species are dominated by 'off-diagonal' β_zyy components. The most significant findings of these studies are: (i) β₀ values as much as 6 times that of the chromophore in the technologically important material (E)-4′-(dimethylamino)-N-methyl-4- stilbazolium tosylate; (ii) reversible electrochemistry that offers potential for redox-switching of optical properties over multiple states; (iii) strongly 2D NLO responses that may be exploited for novel practical applications; (iv) a new polar material, suitable for bulk NLO behavior.

Full text of DOI:10.1021/ja103289a

Published on Web 07/09/2010
Diquat Derivatives: Highly Active, Two-Dimensional Nonlinear
Optical Chromophores with Potential Redox Switchability
Benjamin J. Coe,*,† John Fielden,† Simon P. Foxon,† James A. Harris,†
Madeleine Helliwell,† Bruce S. Brunschwig,‡ Inge Asselberghs,§ Koen Clays,§
Javier Gar´ın,| and Jesu´s Orduna|
School of Chemistry, UniVersity of Manchester, Oxford Road, Manchester M13 9PL, U.K.,
Molecular Materials Research Center, Beckman Institute, MC 139-74, California Institute of
Technology, 1200 East California BouleVard, Pasadena, California 91125, Department of
Chemistry, UniVersity of LeuVen, Celestijnenlaan 200D, B-3001 LeuVen, Belgium, and
Departamento de Qu´ımica Orga´nica, ICMA, UniVersidad de Zaragoza-CSIC,
E-50009 Zaragoza, Spain
Received April 19, 2010; E-mail: b.coe@manchester.ac.uk.
Abstract: In this article, we present a detailed study of structure-activity relationships in diquaternized
2,2′-bipyridyl (diquat) derivatives. Sixteen new chromophores have been synthesized, with variations in
the amino electron donor substituents, π-conjugated bridge, and alkyl diquaternizing unit. Our aim is to
combine very large, two-dimensional (2D) quadratic nonlinear optical (NLO) responses with reversible redox
-
chemistry. The chromophores have been characterized as their PF6 salts by using various techniques
including electronic absorption spectroscopy and cyclic voltammetry. Their visible absorption spectra are
dominated by intense π f π* intramolecular charge-transfer (ICT) bands, and all show two reversible diquat-
based reductions. First hyperpolarizabilities ꢀ have been measured by using hyper-Rayleigh scattering
with an 800 nm laser, and Stark spectroscopy of the ICT bands affords estimated static first hyperpolar-
izabilities ꢀ0. The directly and indirectly derived ꢀ values are large and increase with the extent of
π-conjugation and electron donor strength. Extending the quaternizing alkyl linkage always increases the
ICT energy and decreases the E1/2 values for diquat reduction, but a compensating increase in the ICT
intensity prevents significant decreases in Stark-based ꢀ0 responses. Nine single-crystal X-ray structures
have also been obtained. Time-dependent density functional theory clarifies the molecular electronic/optical
properties, and finite field calculations agree with polarized HRS data in that the NLO responses of the
disubstituted species are dominated by ‘off-diagonal’ ꢀzyy components. The most significant findings of
these studies are: (i) ꢀ0 values as much as 6 times that of the chromophore in the technologically important
material (E)-4′-(dimethylamino)-N-methyl-4-stilbazolium tosylate; (ii) reversible electrochemistry that offers
potential for redox-switching of optical properties over multiple states; (iii) strongly 2D NLO responses that
may be exploited for novel practical applications; (iv) a new polar material, suitable for bulk NLO behavior.
Introduction
lead to the creation of all-optical computing devices. For
molecular species, quadratic (second-order) NLO behavior arises
In addition to purely academic incentives, a wealth of diverse
applications such as advanced telecommunications and biologi-
cal imaging makes organic nonlinear optical (NLO) materials
of great current interest.1 For example, NLO effects can allow
the exchange of information between laser beams and thus may
from the first hyperpolarizability ꢀ that translates into ꢁ(2) in
bulk materials. In order to achieve nonzero values of both ꢀ
and ꢁ(2), noncentrosymmetric structures are essential.
Synthetic chemistry drives much fundamental NLO research,
which has involved a wide variety of compounds, including
organic salts. An important feature of salts is that counterion
variations can be used to modify crystal packing, potentially
giving polar structures showing quadratic NLO effects such as
second harmonic generation (SHG) and linear electrooptic
behavior. Solubility can also be thus tuned, having implications
for imaging applications. Other attractive aspects of crystalline
salts are their inherently greater stabilities and higher chro-
mophore number densities when compared with alternative NLO
materials such as poled polymers. In this context, special
attention has been given to compounds such as (E)-4′-(dim-
ethylamino)-N-methyl-4-stilbazolium tosylate (DAST)2 and
related species.3 Notably, crystals of DAST have recently been
† University of Manchester.
‡ California Institute of Technology.
§ University of Leuven.
| Universidad de Zaragoza.
(1) (a) Zyss, J. Molecular Nonlinear Optics: Materials, Physics and
DeVices; Academic Press: Boston, 1994. (b) Bosshard, Ch.; Sutter,
K.; Preˆtre, Ph.; Hulliger, J.; Flo¨rsheimer, M.; Kaatz, P.; Gu¨nter, P.
AdVances in Nonlinear Optics; Organic Nonlinear Optical Materials,
Vol. 1; Gordon & Breach: Amsterdam, The Netherlands, 1995. (c)
Nonlinear Optics of Organic Molecules and Polymers; Nalwa, H. S.,
Miyata, S., Eds.; CRC Press: Boca Raton, FL, 1997. (d) Marder, S. R.
Chem. Commun. 2006, 131–134. (e) Nonlinear Optical Properties of
Matter: From Molecules to Condensed Phases; Papadopoulos, M. G.;
Leszczynski, J.; Sadlej, A. J., Eds.; Springer: Dordrecht, 2006. (f)
Kuzyk, M. G. J. Mater. Chem. 2009, 19, 7444–7465.
9
10498 J. AM. CHEM. SOC. 2010, 132, 10498–10512
10.1021/ja103289a  2010 American Chemical Society

Diquat Derivatives: 2D Nonlinear Optical Chromophores
A R T I C L E S
commercialized for terahertz (THz) wave generation via non-
linear frequency mixing.4 Terahertz radiation lies between the
microwave and IR regions and has many applications including
security scanning, biomedical analysis, and space communica-
tions.5 When compared with semiconductors such as GaAs,
organic materials with large ꢁ(2) values can produce broader
bandwidths and also allow higher-power outputs without
damage. The major limitation of DAST is a gap in the spectrum
produced between 0.9 and 1.3 THz,5c and it is likely that the
versatility of organic synthesis can be used to produce better
materials giving more complete spectral coverage and/or
superior crystal growth or stability properties.
Molecules with large ꢀ responses contain π-electron donor
(π-ED) and acceptor (π-EA) groups linked via polarizable
π-systems, with pyridinium rings being strong π-EAs in
stilbazolium dyes. Such chromophores also show intense π f
π* intramolecular charge-transfer (ICT) absorptions. Hyperpo-
larizabilities are tensor quantities that can have several nonzero
components. Although most NLO chromophores have simple
one-dimensional (1D) dipolar structures, multidimensional spe-
cies including 2D dipoles6 and 2D or 3D octupoles7 are also
very interesting. Such molecules offer significant potential
advantages over their 1D counterparts, including increased NLO
responses without undesirable losses of visible transparency.
In V-shaped species, the presence of more than one significant
component of ꢀ is important in various ways, for example
preventing reabsorption of harmonic light due to its polarization
being perpendicular to the ICT transition dipole moment (µ12)
and facilitating phase-matching between the fundamental and
harmonic waves in SHG.6a
We have previously studied salts of stilbazolium-type chro-
mophores with N-arylpyridinium groups8 and also related
N-methylbenzothiazolium species.9 Two salts have been dis-
covered that crystallize in the acentric space group Cc and have
powder SHG efficiencies similar to that of DAST (using a 1907
nm laser). The compound (E)-4′-(dimethylamino)-N-phenyl-4-
stilbazolium hexafluorophosphate shows polymorphism,10 and
Maker fringe SHG measurements on large, single crystals of
the active form afford a huge diagonal NLO susceptibility of
d111 ≈ 290 pm V-1 at 1907 nm (cf., DAST gives d111 ≈ 210
(6) Selected examples: (a) Wortmann, R.; Kra¨mer, P.; Glania, C.; Lebus,
S.; Detzer, N. Chem. Phys. 1993, 173, 99–108. (b) Moylan, C. R.;
Ermer, S.; Lovejoy, S. M.; McComb, I.-H.; Leung, D. S.; Wortmann,
R.; Krämer, P.; Twieg, R. J. J. Am. Chem. Soc. 1996, 118, 12950–
12955. (c) Di Bella, S.; Fragala`, I.; Ledoux, I.; Diaz-Garcia, M. A.;
Marks, T. J. J. Am. Chem. Soc. 1997, 119, 9550–9557. (d) Wolff,
J. J.; La¨ngle, D.; Hillenbrand, D.; Wortmann, R.; Matschiner, R.;
Glania, C.; Kra¨mer, P. AdV. Mater. 1997, 9, 138–143. (e) Averseng,
F.; Lacroix, P. G.; Malfant, I.; Lenoble, G.; Cassoux, P.; Nakatani,
K.; Maltey-Fanton, I.; Delaire, J. A.; Aukauloo, A. Chem. Mater. 1999,
11, 995–1002. (f) Hilton, A.; Renouard, T.; Maury, O.; Le Bozec, H.;
Ledoux, I.; Zyss, J. Chem. Commun. 1999, 2521–2522. (g) Lacroix,
P. G. Eur. J. Inorg. Chem. 2001, 339–348. (h) Ostroverkhov, V.;
Petschek, R. G.; Singer, K. D.; Twieg, R. J. Chem. Phys. Lett. 2001,
340, 109–115. (i) Di Bella, S.; Fragala`, I.; Ledoux, I.; Zyss, J.
Chem.sEur. J. 2001, 7, 3738–3743. (j) Yang, M.-L.; Champagne, B.
J. Phys. Chem. A 2003, 107, 3942–3951. (k) Wortmann, R.; Lebus-
Henn, S.; Reis, H.; Papadopoulos, M. G. J. Mol. Struct. (THEOCHEM)
2003, 633, 217–226. (l) Cui, Y.-Z.; Fang, Q.; Huang, Z.-L.; Xue, G.;
Yu, W.-T.; Lei, H. Opt. Mater. 2005, 27, 1571–1575. (m) Rigamonti,
L.; Demartin, F.; Forni, A.; Righetto, S.; Pasini, A. Inorg. Chem. 2006,
45, 10976–10989. (n) Li, H.-P.; Han, K.; Tang, G.; Shen, X.-P.; Wang,
H.-T.; Huang, Z.-M.; Zhang, Z.-H.; Bai, L.; Wang, Z.-Y. Chem. Phys.
Lett. 2007, 444, 80–84. (o) Zrig, S.; Koeckelberghs, G.; Verbiest, T.;
Andrioletti, B.; Rose, E.; Persoons, A.; Asselberghs, I.; Clays, K. J.
Org. Chem. 2007, 72, 5855–5858. (p) Liu, C.-G.; Qiu, Y.-Q.; Su, Z.-
M.; Yang, G.-C.; Sun, S.-L. J. Phys. Chem. C 2008, 112, 7021–7028.
(q) Li, H.-P.; Han, K.; Tang, G.; Li, M.-X.; Shen, X.-P.; Huang, Z.-
M. Mol. Phys. 2009, 107, 1597–1603. (r) Muhammad, S.; Janjua,
M. R. S. A.; Su, Z.-M. J. Phys. Chem. C 2009, 113, 12551–12557.
(7) Selected examples: (a) Verbiest, T.; Clays, K.; Samyn, C.; Wolff, J.;
Reinhoudt, D.; Persoons, A. J. Am. Chem. Soc. 1994, 116, 9320–
9323. (b) Zyss, J.; Ledoux, I. Chem. ReV. 1994, 94, 77–105. (c)
Dhenaut, C.; Ledoux, I.; Samuel, I. D. W.; Zyss, J.; Bourgault, M.;
Le Bozec, H. Nature 1995, 374, 339–342. (d) McDonagh, A. M.;
Humphrey, M. G.; Samoc, M.; Luther-Davies, B.; Houbrechts, S.;
Wada, T.; Sasabe, H.; Persoons, A. J. Am. Chem. Soc. 1999, 121,
1405–1406. (e) Vance, F. W.; Hupp, J. T. J. Am. Chem. Soc. 1999,
121, 4047–4053. (f) Wolff, J. J.; Siegler, F.; Matschiner, R.; Wortmann,
R. Angew. Chem., Int. Ed. 2000, 39, 1436–1439. (g) Cho, B. R.; Piao,
M. J.; Son, K. H.; Lee, S. H.; Yoon, S. J.; Jeon, S.-J.; Cho, M.-H.
Chem.sEur. J. 2002, 8, 3907–3916. (h) Le Bouder, T.; Maury, O.;
Bondon, O.; Costuas, K.; Amouyal, E.; Ledoux, I.; Zyss, J.; Le Bozec,
H. J. Am. Chem. Soc. 2003, 125, 12284–12299. (i) Maury, O.; Viau,
L.; Se´ne´chal, K.; Corre, B.; Gue´gan, J.-P.; Renouard, T.; Ledoux, I.;
Zyss, J.; Le Bozec, H. Chem.sEur. J. 2004, 10, 4454–4466. (j) Maury,
O.; Le Bozec, H. Acc. Chem. Res. 2005, 38, 691–704. (k) Coe, B. J.;
Harris, J. A.; Brunschwig, B. S.; Asselberghs, I.; Clays, K.; Gar´ın, J.;
Orduna, J. J. Am. Chem. Soc. 2005, 127, 13399–13410. (l) Le Floc’h,
V.; Brasselet, S.; Zyss, J.; Cho, B. R.; Lee, S. H.; Jeon, S.-J.; Cho,
M.-H.; Min, K. S.; Suh, M. P. AdV. Mater. 2005, 17, 196–200. (m)
Hennrich, G.; Omenat, A.; Asselberghs, I.; Foerier, S.; Clays, K.;
Verbiest, T.; Serrano, J. L. Angew. Chem., Int. Ed. 2006, 45, 4203–
4206. (n) Jeong, M.-Y.; Kim, H. M.; Jeon, S.-J.; Brasselet, S.; Cho,
B. R. AdV. Mater. 2007, 19, 2107–2111. (o) Liu, Y.; Xu, X.; Zheng,
F.; Cui, Y. Angew. Chem., Int. Ed. 2008, 47, 4538–4541. (p) Akdas-
Kilig, H.; Roisnel, T.; Ledoux, I.; Le Bozec, H. New J. Chem. 2009,
33, 1470–1473. (q) Kim, H. M.; Cho, B. R. J. Mater. Chem. 2009,
19, 7402–7409.
(2) Selected examples: (a) Marder, S. R.; Perry, J. W.; Schaefer, W. P.
Science 1989, 245, 626–628. (b) Marder, S. R.; Perry, J. W.;
Yakymyshyn, C. P. Chem. Mater. 1994, 6, 1137–1147. (c) Lee, O.-
K.; Kim, K.-S. Photonics Sci. News 1999, 4, 9–20. (d) Sohma, S.;
Takahashi, H.; Taniuchi, T.; Ito, H. Chem. Phys. 1999, 245, 359–
364. (e) Kaino, T.; Cai, B.; Takayama, K. AdV. Funct. Mater. 2002,
12, 599–603. (f) Mohan Kumar, R.; Rajan Babu, D.; Ravi, G.; Jayavel,
R. J. Cryst. Growth 2003, 250, 113–117. (g) Geis, W.; Sinta, R.;
Mowers, W.; Deneault, S. J.; Marchant, M. F.; Krohn, K. E.; Spector,
S. J.; Calawa, D. R.; Lyszczarz, T. M. Appl. Phys. Lett. 2004, 84,
3729–3731.
(3) Selected examples: (a) Marder, S. R.; Perry, J. W.; Schaefer, W. P. J.
Mater. Chem. 1992, 2, 985–986. (b) Yitzchaik, S.; Marks, T. J. Acc.
Chem. Res. 1996, 29, 197–202. (c) Alain, V.; Blanchard-Desce, M.;
Ledoux-Rak, I.; Zyss, J. Chem. Commun. 2000, 353–354. (d) Cariati,
E.; Ugo, R.; Cariati, F.; Roberto, D.; Masciocchi, N.; Galli, S.; Sironi,
A. AdV. Mater. 2001, 13, 1665–1668. (e) Abbotto, A.; Beverina, L.;
Bradamante, S.; Facchetti, A.; Klein, C.; Pagani, G. A.; Redi-Abshiro,
M.; Wortmann, R. Chem.sEur. J. 2003, 9, 1991–2007. (f) Kim, H. S.;
Lee, S. M.; Ha, K.; Jung, C.; Lee, Y.-J.; Chun, Y. S.; Kim, D.; Rhee,
B. K.; Yoon, K. B. J. Am. Chem. Soc. 2004, 126, 673–682. (g) Yang,
Z.; Aravazhi, S.; Schneider, A.; Seiler, P.; Jazbinsˇek, M.; Gu¨nter, P.
AdV. Funct. Mater. 2005, 15, 1072–1076. (h) Ruiz, B.; Yang, Z.;
Gramlich, V.; Jazbinsˇek, M.; Gu¨nter, P. J. Mater. Chem. 2006, 16,
2839–2842. (i) Kim, H. S.; Sohn, K. W.; Jeon, Y.; Min, H.; Kim, D.;
Yoon, K. B. AdV. Mater. 2007, 19, 260–263. (j) Yang, Z.; Jazbinsˇek,
M.; Ruiz, B.; Aravazhi, S.; Gramlich, V.; Gu¨nter, P. Chem. Mater.
2007, 19, 3512–3518. (k) Yang, Z.; Mutter, L.; Stillhart, M.; Ruiz,
B.; Aravazhi, S.; Jazbinsˇek, M.; Schneider, A.; Gramlich, V.; Gu¨nter,
P. AdV. Funct. Mater. 2007, 17, 2018–2023. (l) Nunzi, F.; Fantacci,
S.; De Angelis, F.; Sgamellotti, A.; Cariati, E.; Ugo, R.; Macchi, P. J.
Phys. Chem. C 2008, 112, 1213–1226. (m) Xu, X.-L.; Qiu, W.-W.;
Zhou, Q.; Tang, J.; Yang, F.; Sun, Z.-R.; Audebert, P. J. Phys. Chem.
B 2008, 112, 4913–4917.
(4) Selected examples: (a) Kawase, K.; Mizuno, M.; Sohma, S.; Takahashi,
H.; Taniuchi, T.; Urata, Y.; Wada, S.; Tashiro, H.; Ito, H. Opt. Lett.
1999, 24, 1065–1067. (b) Kawase, K.; Hatanaka, T.; Takahashi, H.;
Nakamura, K.; Taniuchi, T.; Ito, H. Opt. Lett. 2000, 25, 1714–1716.
(c) Taniuchi, T.; Okada, S.; Nakanishi, H. J. Appl. Phys. 2004, 95,
5984–5988. (d) Taniuchi, T.; Ikeda, S.; Okada, S.; Nakanishi, H. Jpn.
J. Appl. Phys. 2005, 44, L652–L654. (e) Schneider, A.; Neis, M.;
Stillhart, M.; Ruiz, B.; Khan, R. U. A.; Gu¨nter, P. J. Opt. Soc. Am. B
2006, 23, 1822–1835. (f) Schneider, A.; Stillhart, M.; Gu¨nter, P. Opt.
Express 2006, 14, 5376–5384.
(5) (a) Terahertz Sources and Systems; Miles, R. E., Harrison, P., Lippens,
D., Eds. NATO Science Series II, Vol. 27; Kluwer: Dordrecht, 2001.
(b) Ferguson, B.; Zhang, X.-C. Nat. Mater. 2002, 1, 26–33. (c)
McEntee, J. Chem. World 2007, 52–56. (d) http://www.teraview.
com.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10499

A R T I C L E S
Coe et al.
pm V-1).11 We have hence already produced materials with
superior bulk NLO responses when compared with DAST. Our
present studies are focused on chromophores featuring the
diquaternized 2,2′-bipyridyl (diquat) unit. Except for one brief
theoretical study concerning two-photon absorption,12 the diquat
group has not been exploited for NLO purposes previously. We
chose this motif because it is a π-EA stronger than a simple
pyridinium group, acts as a core for the construction of 2D
dipoles, and shows reversible reductions that may be used to
switch the optical properties, thus incorporating metal-like
properties into low-cost, purely organic chromophores. Prelimi-
nary investigations revealed that such molecules exhibit two
substantial components of the ꢀ tensor, ꢀzzz and ꢀzyy;13 here we
describe major extensions of this work that aim to establish
structure-activity relationships for these interesting new NLO
chromophores.
Emax(µ12)
∆µab
Hab
)
(4)
|
|
|
|
If the hyperpolarizability tensor ꢀ0 has only nonzero elements along
the ICT direction, then this quantity is given by
3∆µ12(µ12)2
ꢀ0 )
(5)
2
(Emax
)
A relative error of (20% is estimated for the ꢀ0 values derived
from the Stark data and using eq 5, while experimental errors of
(10% are estimated for µ12, ∆µ12, and ∆µab, (15% for Hab and
(50% for cb . Note that the (20% uncertainty for the ꢀ0 values is
merely statistical and does not account for any errors introduced
by two-state extrapolation.
2
Results and Discussion
Experimental Section
Synthetic Studies. Most of the new diquat chromophores
3-18 (Figure 1) were prepared in one step from commercial
or easily accessible diquat and aldehyde precursors, and isolated
as their PF6- salts. Knoevenagel condensations between a 2,11/
12-dimethyl-diquat derivative and either one equivalent (for
monosubstituted (MS) compounds) or 8-20 equiv (for disub-
stituted (DS) compounds) of the appropriate aldehyde were
originally used to access 3 and 11.13 This approach also afforded
4-7, 9, 10, 12-15, 17, and 18. While very large excesses (20
equiv) of aldehyde are necessary to achieve high yields of DS
compounds in reactions with 4-(dimethylamino)benzaldehyde,
thin-layer chromatography indicates that reactions with (E)-4-
(dimethylamino)cinnamaldehyde or julolidinyl aldehydes are
essentially complete when using only 8-10 equiv. Notably, the
synthesis of [15][PF6]2 gave a substantial quantity (∼20% yield)
of the corresponding MS compound [7][PF6]2, but no MS
products were isolated from the preparations of [17][PF6]2 and
[18][PF6]2, despite the use of smaller relative quantities of
aldehyde.
Compounds [3][PF6]2 and [11][PF6]2 were purified simply via
reprecipitation, exploiting the greatly different solubilities of
the corresponding bromide salts in methanol. Using the same
method was successful for [15][PF6]2, but not its MS analogue
[7][PF6]2. The latter and all the other salts, were purified by
column chromatography on silica gel with mixtures of acetone,
water and saturated aqueous KNO3. The isolated yields are
relatively low (7-32% for the MS species and 13-51% for
their DS counterparts), with substantial quantities of the colored
products becoming trapped on the silica gel.
Syntheses and Measurements. All of the details relating to the
synthetic chemistry, general characterization, X-ray crystallography,
hyper-Rayleigh scattering and other physical measurements, and
theoretical calculations can be found in the Supporting Information.
Stark Spectroscopy. Full details are in the Supporting Informa-
tion, while the equations and definitions used are included below
because they are integral to the subsequent discussion. A two-state
analysis of the ICT transitions gives
∆µ2ab ) ∆µ212 + 4µ212
(1)
where ∆µab is the dipole-moment change between the diabatic states
and ∆µ12 is the observed (adiabatic) dipole-moment change. The
value of µ12 can be determined from the oscillator strength fos of
the transition by
1/2
fos
|µ12| )
(2)
-5
(
)
1.08 × 10
E
max
where Emax is the energy of the ICT maximum (in wavenumbers)
and µ12 is in e ·Å. The latter is converted into Debye units upon
multiplying by 4.803. The degree of delocalization cb2 and electronic
coupling matrix element Hab for the diabatic states are given by
∆µ212
1/2
1
2
c2b ) 1 -
(3)
2
2
(
)
[
]
∆µ12 + 4µ12
(8) (a) Coe, B. J.; Harris, J. A.; Asselberghs, I.; Clays, K.; Olbrechts, G.;
Persoons, A.; Hupp, J. T.; Johnson, R. C.; Coles, S. J.; Hursthouse,
M. B.; Nakatani, K. AdV. Funct. Mater. 2002, 12, 110–116. (b) Clays,
K.; Coe, B. J. Chem. Mater. 2003, 15, 642–648. (c) Coe, B. J.; Harris,
J. A.; Asselberghs, I.; Wostyn, K.; Clays, K.; Persoons, A.; Brun-
schwig, B. S.; Coles, S. J.; Gelbrich, T.; Light, M. E.; Hursthouse,
M. B.; Nakatani, K. AdV. Funct. Mater. 2003, 13, 347–357. (d) Coe,
B. J.; Harris, J. A.; Brunschwig, B. S.; Gar´ın, J.; Orduna, J.; Coles,
S. J.; Hursthouse, M. B. J. Am. Chem. Soc. 2004, 126, 10418–10427.
(e) Coe, B. J.; Beljonne, D.; Vogel, H.; Gar´ın, J.; Orduna, J. J. Phys.
Chem. A 2005, 109, 10052–10057. (f) Coe, B. J.; Hall, J. J.; Harris,
J. A.; Brunschwig, B. S.; Coles, S. J.; Hursthouse, M. B. Acta
Crystallogr., Sect. E 2005, 61, o464–o467.
Using the Knoevenagel approach gave unsatisfactory results
for 8 or 16, so these species were prepared by reacting the
respective 2,2′-bipyridyl derivatives 1 and 2 (Figure 1) with 1,3-
bis(triflyloxy)propane.14 1 is a known compound,15 while 2 was
prepared in high yield via a Wadsworth-Emmons reaction
between (E)-4-(dimethylamino)cinnamaldehyde and 4,4′-bis-
[(diethoxyphosphinyl)methyl]-2,2′-bipyridyl.16 These diquater-
(9) Coe, B. J.; Harris, J. A.; Hall, J. J.; Brunschwig, B. S.; Hung, S.-T.;
Libaers, W.; Clays, K.; Coles, S. J.; Horton, P. N.; Light, M. E.;
Hursthouse, M. B.; Gar´ın, J.; Orduna, J. Chem. Mater. 2006, 18, 5907–
5918.
(13) Coe, B. J.; Harris, J. A.; Brunschwig, B. S.; Gar´ın, J.; Orduna, J. J. Am.
Chem. Soc. 2005, 127, 3284–3285.
(14) Lindner, E.; von Au, G.; Eberle, H.-J. Chem. Ber. 1981, 114, 810–
813.
(10) Ruiz, B.; Coe, B. J.; Gianotti, R.; Gramlich, V.; Jazbinsˇek, M.; Gu¨nter,
P. CrystEngComm 2007, 9, 772–776.
(15) Jang, S.-R.; Lee, C.-C.; Choi, H.-B.; Ko, J. J.; Lee, J.-W.; Vittal, R.;
Kim, K.-J. Chem. Mater. 2006, 18, 5604–5608.
(16) Gillaizeau-Gauthier, I.; Odobel, F.; Alebbi, M.; Argazzi, R.; Costa,
E.; Bignozzi, C. A.; Qu, P.; Meyer, G. J. Inorg. Chem. 2001, 40, 6073–
6079.
(11) Figi, H.; Mutter, L.; Hunziker, C.; Jazbinsˇek, M.; Gu¨nter, P.; Coe,
B. J. J. Opt. Soc. Am. B 2008, 25, 1786–1793.
(12) Zhou, Y.-F.; Meng, F.-Q.; Zhao, X.; Feng, S.-Y.; Jiang, M.-H. Chem.
Phys. 2001, 269, 441–445.
9
10500 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Diquat Derivatives: 2D Nonlinear Optical Chromophores
A R T I C L E S
Figure 1. Chemical structures of the new diquat derivatives and precursors. Shorthand: EDQ2+ ) ethylene diquat; PDQ2+ ) propylene diquat. Except for
-
some of the X-ray crystal structure determinations, all of the measurements on 3-18 were made with PF6 salts.
nizations followed our precedent set with EDQ2+ species,17
designed to avoid the need for relatively forcing reaction
conditions with dibromo precursors. The salts [8][PF6]2 and
[16][PF6]2 were thus obtained pure in respective yields of 32
and 39%; these are lower than those reported previously17 due
to the poor solubility of 2 and the need for chromatography.
1H NMR Spectroscopy Studies. The new compounds give
doublets and 2 singlets), while the DS species show only 3
signals (2 doublets and 1 singlet) (Table 1). The diquat CH2
protons afford a 4H singlet in the symmetrical DS EDQ2+
species. In all of the other compounds, the CH2 protons give
multiplets; two or three 2H signals for MS EDQ2+ and DS
PDQ2+ species, respectively, but five signals (four 1H, one 2H)
for MS PDQ2+ chromophores. Upfield shifts occur, up to 0.17
ppm in the C5H3N protons (Supporting Information, Figure S1),
and up to 0.09 ppm in the diquat CH2 protons, on replacing a
4-(dimethylamino)phenyl (Dap) group with julolidinyl (Jd).
These shifts arise from greater shielding18 caused by the stronger
1
diagnostic H NMR spectra, and the signals for the protons of
the diquat units show several noteworthy trends. Selected data
are collected in Table 1 and representative spectra of salts
[7][PF6]2, [9][PF6]2, [13][PF6]2, and [17][PF6]2 are shown in
Figure S1 (Supporting Information) and Figure 2.
For the diquat aryl (C5H3N) protons, the low symmetry of
the MS compounds generally leads to 6 resolved signals (4
(18) Coe, B. J.; Foxon, S. P.; Harper, E. C.; Harris, J. A.; Helliwell, M.;
Raftery, J.; Asselberghs, I.; Clays, K.; Franz, E.; Brunschwig, B. S.;
Fitch, A. G. Dyes Pigments 2009, 82, 171–186.
(17) Coe, B. J.; Curati, N. R. M.; Fitzgerald, E. C. Synthesis 2006, 146–
150.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10501

A R T I C L E S
Table 1. Selected H NMR Data for the Salts [3-18][PF6]2
Coe et al.
a
1
dication
C5H3N signals
CH2 signals
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
8.70d 8.68s
8.70d 8.68s
8.69d 8.66s
8.68d 8.66s
8.74d 8.46d 8.20s
8.74d 8.48d 8.19s
8.73d 8.31d 8.18s
8.75d 8.40d 8.19s
8.58s
8.54s
8.46s
8.45s
8.43d
8.45d
8.29d
8.35d
8.12s
8.11-8.07
8.07d
8.07-8.01b 7.90d
4.98-4.93 4.84-4.80
8.05d
8.05d
8.03d
8.09d
7.96-7.89b 4.98-4.93 4.86-4.81
7.77d
7.80d
4.98-4.92 4.80-4.74
4.96-4.91 4.80-4.76
7.96-7.90b 4.80-4.74 4.64-4.58 4.40-4.32 4.16-4.08 2.75-2.67
7.94d
4.80-4.74 4.65-4.58 4.38-4.30 4.17-4.08 2.75-2.67
7.98s
8.02s
7.83-7.77b 4.80-4.72 4.58-4.49 4.42-4.30 4.11-3.98 2.79-2.60c
8.08d
7.86-7.78b 4.80-4.74 4.60-4.54 4.39-4.32 4.13-4.05 2.74-2.65c
8.63s
8.57s
8.46s
8.47s
8.40d 7.89d
4.79s
4.80s
4.70s
4.74s
8.43d 7.98-7.88b
8.24d 7.73d
8.31d 7.79d
8.45d 8.21s
8.47d 8.17s
8.30d 8.06s
8.36d 8.08s
7.99-7.91b
7.92d
4.65-4.57 4.27-4.17 2.70-2.62
4.66-4.58 4.27-4.16 2.71-2.62
4.55-4.48 4.20-4.12 2.63-2.54
4.60-4.53 4.20-4.14 2.64-2.58
7.77d
7.94-7.85b
a Recorded at 400 or 500 MHz in CD3CN; all values are given in ppm with respect to TMS. b Overlapped with CH signals. c Overlapped with Jd
CH2 signals.
Figure 3. UV-vis absorption spectra of [7][PF6]2 (green), [12][PF6]2 (red),
[15][PF6]2 (blue), and [16][PF6]2 (gold) at 293 K in acetonitrile.
and Hb are much closer together for [Me2PDQ2+][PF6]2 (at 8.15
Figure 2. Aromatic regions of the 1H NMR spectra of the salts [13][PF6]2
and 8.12 ppm) indicates that these two environments become
more similar as the diquat unit twists. The new substituted diquat
derivatives show behavior similar to that of the model species
(Figure 2).
and [17][PF6]2 recorded at 400 MHz in CD3CN at 293 K. The arrow
indicates the largest shift observed for one of the C5H3N signals.
π-ED ability of the Jd unit,19 and are generally larger when n
) 1 because the influence of the π-ED group(s) is communicated
more strongly over shorter distances. Notably, increasing n also
makes the chromophores more electron rich, but does not
produce consistent effects on the chemical shifts of the diquat
groups.
Large changes (up to 0.5 ppm) in the chemical shifts of the
C5H3N signals occur on extending the diquaternized bridge
(replacing EDQ2+ with PDQ2+, Table 1). These environments
are identified according to their splitting patterns, as confirmed
via H,C-HMQC correlation studies with the model compounds
[Me2EDQ2+][PF6]2 and [Me2PDQ2+][PF6]2. For the latter sym-
metric species, the largest difference is observed for the singlet
(or finely split doublet) signal associated with the protons
adjacent to the bridge between the two pyridyl rings (Hc); this
signal shifts upfield from 8.63 ppm in [Me2EDQ2+][PF6]2 to
8.15 ppm in [Me2PDQ2+][PF6]2. In contrast, the other two
signals (Ha and Hb) show smaller, accompanying downfield
shifts of only 0.02-0.03 ppm. The fact that the signals for Hc
Electronic Spectroscopy Studies. The UV-vis absorption data
for salts [3-18][PF6]2 in acetonitrile are shown in Table 2, and
representative spectra of [7][PF6]2, [12][PF6]2, [15][PF6]2, and
[16][PF6]2 in Figure 3. All of the new compounds show intense
absorptions in the visible region, attributable to Dap/Jd f diquat
ICT transitions. For all of the EDQ2+ compounds, the main
visible bands are accompanied by weaker features to high
energy, while the PDQ2+ species show only one visible
maximum. The less intense UV absorptions are due to π f π*
excitations that are expected to have lower directionality when
compared with the ICT bands (i.e., processes largely confined
within an aryl ring or ethenyl/butadienyl unit). It should be noted
that the combination of relatively high energies and low
intensities with small or neglible dipole-moment changes means
that these excitations are not expected to contribute significantly
to the NLO responses. The much lower ICT Emax values for
8a
[3][PF6]2 and [7][PF6]2 when compared with [DAS]PF6
confirm that the diquat units are considerably stronger π-EAs
than a N-methylpyridinium group.13
(19) Kwon, O.; Barlow, S.; Odom, S. A.; Beverina, L.; Thompson, N. J.;
Zojer, E.; Bre´das, J.-L.; Marder, S. R. J. Phys. Chem. A 2005, 109,
9346–9352.
The ICT energies Emax of the new chromophores show certain
trends that are also observed with simpler pseudolinear mono-
9
10502 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Diquat Derivatives: 2D Nonlinear Optical Chromophores
A R T I C L E S
Table 2. UV-Vis Absorption and Electrochemical Data for Salts [3-18][PF6]2 and [DAS]PF6 in Acetonitrilea
E, V vs Ag-AgCl (∆Ep, mV)b
cation
λ
max, nma (ε, 103 M-1 cm-1
)
Emax (eV)
assignment
Epa, oxidations E1/2/Epc, reductions
3c
572 (38.0)
426 (12.7)
314 (22.7)
600 (39.4)
437 (13.5)
311 (20.4)
633 (39.2)
466 (11.2)
314 (18.0)
680 (42.6)
480 (12.9)
309 (20.3)
544 (46.9)
288 (17.9)
572 (43.8)
343 (10.1)
286 (19.4)
602 (52.0)
286 (16.4)
644 (54.2)
376 (10.4)
289 (17.8)
560 (45.1)
442 (12.1)
318 (18.7)
596 (50.2)
460 (28.9)
313 (23.0)
615 (45.5)
477 (24.6)
317 (19.6)
669 (56.6)
507 (32.9)
316 (25.9)
535 (79.8)
297 (24.2)
566 (70.6)
341 (21.4)
297 (23.2)
588 (82.5)
292 (20.3)
633 (71.3)
374 (18.3)
298 (16.5)
470 (42.8)
270 (11.1)
2.17
2.91
3.95
2.07
2.84
3.99
1.96
2.66
3.95
1.82
2.58
4.01
2.28
4.31
2.17
3.62
4.34
2.06
4.34
1.93
3.30
4.29
2.21
2.81
3.90
2.08
2.70
3.96
2.02
2.60
3.91
1.85
2.45
3.92
2.32
4.18
2.19
3.64
4.18
2.11
4.25
1.96
3.32
4.16
2.64
4.59
ICT
ICT
π f π*
ICT
ICT
π f π*
ICT
ICT
π f π*
ICT
ICT
π f π*
ICT
π f π*
ICT
π f π*
π f π*
ICT
π f π*
ICT
π f π*
π f π*
ICT
ICT
π f π*
ICT
ICT
π f π*
ICT
ICT
π f π*
ICT
ICT
π f π*
ICT
π f π*
ICT
π f π*
π f π*
ICT
π f π*
ICT
π f π*
π f π*
ICT
0.98
0.80
0.76
0.60
-0.41 (100)
-0.82 (90)
4
5
6
-0.41 (80)
-0.80 (130)
-0.47 (80)
-0.86 (110)
-0.43 (100)
-0.84 (110)
7
8
0.97
0.78
-0.61 (110)
-0.90 (100)
-0.58 (80)
-0.89 (70)
9
0.77
0.62
-0.64 (110)
-0.91 (100)
-0.57 (120)
-0.89 (100)
10
11c
12
13
14
0.98
0.78
0.76
0.60
-0.42 (90)
-0.77 (80)
-0.38 (80)
-0.76 (120)
-0.49 (100)
-0.81 (125)
-0.42 (80)
-0.74 (160)
15
16
0.97
0.79
-0.61 (80)
-0.88 (100)
-0.56 (100)
-0.82 (90)
17
18
0.76
0.58
-0.65 (80)
-0.88 (80)
-0.58 (100)
-0.83 (80)
d
DAS+
0.95e
-1.10e
π f π*
b
a Solutions ca. 2-3 × 10-5 M. Solutions ca. 10-3 M in analyte and 0.1 M in [N(C4H9-n)4]PF6 at a 2 mm disk glassy carbon working electrode with
a scan rate of 200 mV s-1. Ferrocene internal reference E1/2 ) 0.46 V, ∆Ep ) 80-100 mV. c Data taken from ref 13. d Data taken from refs 8a and 8c.
e Conditions as for other compounds, but using a 2 mm Pt disk working electrode.
cations.18 Thus, in keeping with the 1H NMR data (see above),
replacing Dap with the stronger π-ED Jd decreases Emax by
0.19-0.26 eV, while moving from n ) 1 to 2 gives smaller
decreases of 0.07-0.17 eV (Table 2, Figure 3). In contrast,
moving from an EDQ2+ chromophore to its PDQ2+ analogue
causes blue shifts in the ICT bands (Figure 3); when considering
only the dominant low energy bands for the EDQ2+ species,
the increase in Emax is almost constant at ca. 0.1 eV. This trend
is consistent with the lower relative π-EA strength of a PDQ2+
unit, noted in previous electrochemical studies20 (and see below).
Often small but significant changes in Emax are also found on
moving from a MS chromophore to its DS counterpart (Figure
3). For the EDQ2+ species, Emax of the dominant band increases
by 0.01-0.06 eV, while the accompanying higher energy band
shifts to lower energy by 0.06-0.14 eV. For the PDQ2+
chromophores, the energies of the single ICT bands increase
by 0.02-0.05 eV as a result of this structural change.
Trends are also observed in the molar extinction coefficients
ε of the ICT bands. Moving from n ) 1 to 2 increases ε (but
not always greatly) for all the EDQ2+ species, whereas the
opposite occurs for the PDQ2+ chromophores (Figure 3), except
the pair [9/10][PF6]2. In our previous studies,18 replacing Dap
with Jd always increased ε; the same is generally observed for
[3-18][PF6]2, excepting the minor, high energy ICT bands of
the pairs [3/5][PF6]2 and [4/6][PF6]2. However, these intensity
increases are generally only quite small. Moving from an EDQ2+
chromophore to its PDQ2+ analogue always increases ε sub-
stantially (Figure 3); with regards to NLO properties, it may be
anticipated that this effect will compensate for the relatively
higher Emax values of the PDQ2+ species (see below). The ε
(20) Kuzuya, M.; Kondo, S.-I.; Murase, K. J. Phys. Chem. 1993, 97, 7800–
7802.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10503

A R T I C L E S
Coe et al.
Figure 4. Cyclic voltammograms for the salts [5][PF6]2 (red) and [9][PF6]2
(blue) recorded at 200 mV s-1 in acetonitrile with a glassy carbon working
electrode (Fc ) ferrocene internal standard). The arrow indicates the
direction of the initial scans.
values for the DS species are always rather larger than those of
their MS counterparts (Figure 3), except for the minor ICT band
for the pair [3/11][PF6]2.
Figure 5. Representation of the molecular structure of [8][PF6]2 ·Me2CO
Electrochemical Studies. Cyclic voltammetric data for salts
[3-18][PF6]2 in acetonitrile are shown in Table 2, and repre-
sentative voltammograms of [5][PF6]2 and [9][PF6]2 in Figure
4. Although they are always irreversible, the Epa values ascribed
to oxidation of the Dap/Jd units display trends that are also found
with related monocations,18 and are consistent with the 1H NMR
and ICT data (see above). Moving from n ) 1 to 2 causes Epa
to decrease by 150-200 mV, while similar changes of 160-230
mV are observed on replacing Dap with Jd. Oxidation hence
becomes easier as both structural modifications make the
molecules more electron rich. In contrast, moving from an
EDQ2+ chromophore to its PDQ2+ analogue or from a MS
compound to its DS counterpart leaves the Epa values almost
or completely unchanged.
The salt [DAS]PF6 shows only one, irreversible reduction
wave at relatively low potential,8a consistent with the moderate
π-EA strength of a single pyridinium ring (Table 2). The
reductive electrochemical behavior of the diquat units in
[3-18][PF6]2 is more interesting, with each compound showing
two reversible or quasireversible one-electron waves. Moving
from n ) 1 to 2 causes the E1/2 values to increase slightly by
up to 70 mV (and the shifts are generally largest for the first
wave), whereas moving from a MS compound to its DS
analogue has little effect on the first E1/2 value but increases
the second by up to 100 mV. The reduction processes thus
become easier, and these changes can be attributed to more
effective delocalization of the extra electron(s) in the reduced
forms due to extension of the π-conjugated systems. Replacing
Dap with Jd either barely affects the E1/2 values or produces
small decreases of up to 70 mV because the presence of the
stronger π-ED group(s) makes reduction more difficult. The
lower π-EA strength of a PDQ2+ as opposed to a EDQ2+ unit,
noted in previous electrochemical studies20 (and evidenced by
the ICT energies) leads to relatively large decreases in the E1/2
values (Figure 4). For the first wave, shifts of 140-200 mV
are observed, while the second wave shows smaller changes of
50-110 mV. The main cause of this weaker relative π-EA
ability is the propylene bridging unit that prevents the pyridyl
rings from becoming coplanar and thus hinders effective
delocalization of the extra electron(s) in the reduced species.
X-ray Crystallography. While this study concerns primarily
molecular properties, bulk NLO behavior is required for
(50% probability ellipsoids).
potential applications, and for quadratic effects polar structures
are essential. We have obtained single-crystal X-ray structures
for nine salts of our new chromophores. Six of these (with data
99% complete to 20.82° or better) are presented here, with
crystallographic data and refinement details in Table 3, while
the others are included in the CIF provided as Supporting
Information. Eight of the structures contain PDQ2+-based
chromophores, which seem to crystallize better than their EDQ2+
counterparts, perhaps due to more efficient packing that allows
the formation of larger crystals.
The salt [8][PF6]2 ·Me2CO (Figure 5) forms large crystals and
provides the highest quality structure achieved in this study.
The dihedral angle (Table 4) between the planes of the two
pyridyl rings is slightly lower than those in published structures
of PDQ2+ units (51.4 and 59.7°).21 Other geometric parameters
are unremarkable (Table S1, Supporting Information), as for
all the other structures. The 8 dications pack in layers parallel
to the crystallographic ab plane. Although the packing within
each layer is polar, the layers oppose each other so that the
overall structure is centrosymmetric. In the salt
[13][OTf]2 · 2MeCN (Figure 6), the py/py angle (Table 4) is
much lower than that in [8][PF6]2 ·Me2CO, but slightly above
the range of 15-24° found in comparable published structures.22
The 13 dications form polar chains parallel to the crystal-
lographic b axis; these chains align antiparallel, giving a
centrosymmetric structure.
The structure of [15][NapSO3]2 ·2H2O (Figure 7a) is espe-
cially notable, as its noncentrosymmetric Aba2 space group and
polar packing (Figure 7b) mean that this material will show
bulk NLO activity. This structure contains H-bonded chains of
-
NapSO3 anions and water molecules parallel to the crystal-
lographic c axis. The H-bond distances between the water
-
oxygen O1W and the NapSO3 oxygens O3 and O2 are
2.751(14) Å and 2.957(16) Å, respectively. The V-shaped
cations form chains along the c axis, all with the same polar
(21) (a) Mori, Y.; Matsuyama, Y.; Yamada, S.; Maeda, K. Acta Crystallogr.,
Sect. C 1992, 48, 894–897. (b) Knoch, F.; Smauch, G.; Kisch, H. Z.
Kristallogr. 1995, 210, 225.
(22) Cambridge Structural Database, Version 5.30, May 2009 (update).
9
10504 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Diquat Derivatives: 2D Nonlinear Optical Chromophores
A R T I C L E S
Table 3. Crystallographic Data and Refinement Details for the Salts [8][PF6]2 ·Me2CO, [13][OTf]2 ·2MeCN, [15][NapSO3]2 ·2H2O,
[16][NapSO3]2 ·2H2O, [17][PF6]2 ·2MeCN, and [18][PF6]2 ·2Me2CO
[8][PF6]2 ·Me2CO
[13][OTf]2 ·2MeCN
[15][NapSO3]2 ·2H2O
[16][NapSO3]2 ·2H2O
[17][PF6]2 ·2MeCN
[18][PF6]2 ·2Me2CO
formula
M
crystal system
space group
a, Å
C29H35F12N3OP2
731.54
C46H48F6N6O6S2
959.02
monoclinic
C2/c
18.315(2)
16.102(1)
15.462(1)
90
103.33(1)
90
C53H54N4O8S2
939.12
orthorhombic
Aba2
10.310(4)
35.92(1)
12.402(5)
90
90
90
4592(3)
4
C57H58N4O8S2
991.19
orthorhombic
Pbcn
10.055(5)
12.875(7)
37.90(2)
90
90
90
4906(4)
4
C45H50F12N6P2
964.85
monoclinic
C2/c
19.870(5)
15.248(5)
15.184(5)
90
106.185(5)
90
C51H60F12N4O2P2
1050.97
monoclinic
C2/c
33.971(3)
8.5208(9)
20.299(2)
90
124.043(1)
90
4868.8(8)
4
triclinic
j
P1
8.3822(4)
11.8966(6)
16.7255(8)
97.812(4)
99.103(4)
104.236(4)
1569.42(13)
2
b, Å
c, Å
R, deg
ꢀ, deg
γ, deg
V, Å3
4437.0(7)
4
4418(2)
4
Z
T, K
100(2)
0.240
100(2)
0.202
100(2)
0.178
100(2)
0.171
100(2)
0.191
100(2)
0.181
µ, mm-1
crystal size, mm3
0.35 × 0.18 ×
0.50 × 0.35 ×
0.50 × 0.30 ×
0.50 × 0.30 ×
0.40 × 0.10 ×
0.13 × 0.05 ×
0.15
0.20
0.05
0.05
0.06
0.01
crystal appearance
reflections collected
dark purple block
15497
dark blue block
6696
red plate
11761
brown rod
25106
black needle
15664
dark blue plate
18357
independent reflections (Rint)
6397 (0.0371)
26.37 (99.7%)
3787
429
0.947
0.0405, 0.0930
0.0758, 0.1064
N/A
2297 (0.0536)
20.82 (99%)
1462
331
1.270
0.1053, 0.2963
0.1431, 0.3305
N/A
2794 (0.1155)
21.97 (99.9%)
2154
274
1.091
0.0995, 0.2342
0.1239, 0.2494
0.3(3)b
3491 (0.1821)
23.26 (99%)
2578
459
1.131
0.1143, 0.2491
0.1370, 0.2598
N/A
3891 (0.1148)
25.03 (99.8%)
2421
297
1.107
0.1026, 0.2071
0.1554, 0.2315
N/A
4436
25.35 (99.4%)
3261
323
1.023
0.0447, 0.1208
0.0677, 0.1090
N/A
θ
max, deg (completeness)
reflections with I > 2σ(I)
parameters
goodness-of-fit on F2
final R1, wR2 [I > 2σ(I)]a
(all data)
Flack parameter
peak and hole (e ·Å-3
)
0.370, -0.365
0.573, -0.352
0.513, -0.277
0.301, -0.274
0.802, -0.357
0.567, -0.291
a
2
The structures were refined on Fo using all data; the values of R1 are given for comparison with older refinements based on Fo with a typical
threshold of Fo > 4σ(Fo). b Two data sets were obtained, but neither was strong enough to allow accurate determination of the absolute structure given
the absence of any atoms heavier than sulfur.
Table 4. Selected Twist Angles (deg) for the Diquat Salts
salt
py/pya
bridge/pyb
bridge/Phc
[8][PF6]2 ·Me2CO
48.6
27.3
57.4
58.9
49.0
46.6
8.5
2.8
9.5
6.6
1.0
3.8
5.3
14.1
20.0
6.4
[13][OTf]2 ·2MeCN
[15][NapSO3]2 ·2H2O
[16][NapSO3]2 ·2H2O
[17][PF6]2 ·2MeCN
[18][PF6]2 ·2Me2CO
13.0
2.8
a Dihedral angle between the planes of the two pyridyl rings.
b Torsion angle between the ethenyl/butadienyl unit and the attached
pyridyl ring. c Torsion angle between the ethenyl/butadienyl unit and the
attached phenyl ring.
antiparallel (Figure 8b). The H-bonds in [16][NapSO3]2 ·2H2O
occur between the water O1W positions and disordered oxygens
on the NapSO3- anion, with D · · ·A distances of 2.69(1), 2.79(1),
2.80(2), and 2.94(2) Å to O3, O1, O2B, and O3B, respectively.
Clearly, the increase in cation size induces substantial changes
in the packing arrangement. However, the salts
[17][PF6]2 · 2MeCN (Supporting Information, Figure S3) and
[18][PF6]2 ·2Me2CO (Supporting Information, Figure S4) adopt
very similar packing structures, with the dications in antiparallel-
aligned polar chains parallel to the crystallographic b axis. The
lower resolution structures of [7][NapSO3]2, [7][PF6]2 ·MeCN
and [9][PF6]2 ·MeCN are included in the Supporting Information
(Figures S5-S7 and CIF).
Figure 6. Representation of the molecular structure of [13][OTf]2 ·2MeCN,
showing only the single acetonitrile molecule in the asymmetric unit (30%
probability ellipsoids).
Hyper-Rayleigh Scattering Studies. The ꢀ values of salts
[3-18][PF6]2 measured in acetonitrile by using the HRS
technique23,24 with a 800 nm Ti3+:sapphire laser are shown in
Table 5. This wavelength was chosen because the ICT bands
orientation, separated by layers of anions. X-ray powder
diffraction (Supporting Information, Figure S2) confirms that
this material can be obtained in substantial quantities, offering
potential for bulk NLO measurements.
In contrast to [15][NapSO3]2 ·2H2O, [16][NapSO3]2 ·2H2O
(Figure 8a) forms a centrosymmetric structure. Although both
structures contain NapSO3- · · · H2O H-bonded chains,
[16][NapSO3]2 ·2H2O has alternating chains of dications aligned
(23) (a) Clays, K.; Persoons, A. Phys. ReV. Lett. 1991, 66, 2980–2983. (b)
Clays, K.; Persoons, A. ReV. Sci. Instrum. 1992, 63, 3285–3289. (c)
Hendrickx, E.; Clays, K.; Persoons, A. Acc. Chem. Res. 1998, 31,
675–683.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10505

A R T I C L E S
Coe et al.
Figure 7. Representations of the structure of [15][NapSO3]2 ·2H2O. (a)
Molecular structure (50% probability ellipsoids); the cation has been
completed by generating symmetry equivalent atoms, but only the single
NapSO3- anion and water molecule in the asymmetric unit are shown. (b)
Crystal packing and H-bonding, viewed along the crystallographic a axis;
cation C ) black; other C ) gray; N ) blue; O ) red; S ) yellow (H
atoms omitted except for the crystallographically located water atoms;
H-bonds are shown as green lines).
Figure 8. Representations of the structure of [16][NapSO3]2 ·2H2O. (a)
Molecular structure, with the disorder of the NapSO3- anions removed (50%
probability ellipsoids). (b) Crystal packing and H-bonding. All other details
are as for Figure 7.
lie between 800 nm and the second harmonic (SH) at 400 nm
(Figure 3). However, the compounds do absorb relatively
strongly at 400 nm, and also at 800 nm for most of the Jd
species, and thus the results obtained are generally markedly
resonance enhanced. Other available lasers, e.g., 1064 or 1300
nm, would be unsuitable for most of the samples due to very
strong absorption at their SH wavelengths. The data shown are
orientationally averaged ꢀ〈ꢀ2HRS〉 values derived from the total
HRS intensity, regardless of molecular symmetry.
Although many of the chromophores show an additional ICT
maximum to high energy of the main band and there is
significant resonance enhancement, we have used the simple
two-state model25 and the major λmax values (Table 2) to estimate
ꢀ0 values (Table 5). It should be noted that this analysis
constitutes a major oversimplification because the diquat deriva-
tives, especially the DS ones, cannot be accurately described
as two-state systems, as confirmed by the theoretical studies
that reveal multiple low-energy excited states (see below).
Nevertheless, it is notable that all of the chromophores do
display one dominant visible maximum (and only one for the
PDQ2+ species), and most of the derived ꢀ0 values are larger
than that determined under the same conditions for [DAS]PF6,8b,c
and in several cases ca. 2-3-fold increased when compared with
this benchmark. As also observed with related 1D species,18 ꢀ0
increases substantially either when moving from n ) 1 to 2
(except for the pair [9/10][PF6]2), or on replacing Dap with Jd
(excepting the pair [8/10][PF6]2). While moving from a EDQ2+
chromophore to its PDQ2+ analogue apparently causes ꢀ0 to
decrease, this is not always the case. Likewise, the ꢀ0 responses
of the DS compounds are mostly larger than for their MS
counterparts, but several exceptions to this trend are found.
Given the caveats mentioned above and since these comments
also relate to the ꢀ0 values determined without considering
molecular symmetry differences, they should be viewed with
caution.
The electronic structures and thus hyperpolarizabilities of
these chromophores are in fact substantially 2D in nature, and
the main value of our HRS measurements is not actually to
estimate ꢀ0 values, but rather to probe the relative importance
of different tensor components. A chromophore with C2V
symmetry has five nonzero components of the ꢀ tensor, ꢀzzz,
(24) (a) Olbrechts, G.; Strobbe, R.; Clays, K.; Persoons, A. ReV. Sci.
Instrum. 1998, 69, 2233–2241. (b) Olbrechts, G.; Wostyn, K.; Clays,
K.; Persoons, A. Opt. Lett. 1999, 24, 403–405. (c) Clays, K.; Wostyn,
K.; Olbrechts, G.; Persoons, A.; Watanabe, A.; Nogi, K.; Duan, X.-
M.; Okada, S.; Oikawa, H.; Nakanishi, H.; Vogel, H.; Beljonne, D.;
Bre´das, J.-L. J. Opt. Soc. Am. B 2000, 17, 256–265. (d) Franz, E.;
Harper, E. C.; Coe, B. J.; Zahradnik, P.; Clays, K.; Asselberghs, I.
Proc. SPIE-Int. Soc. Opt. Eng 2008, 6999, 699923-1–699923-11.
(25) (a) Oudar, J. L.; Chemla, D. S. J. Chem. Phys. 1977, 66, 2664–2668.
(b) Oudar, J. L. J. Chem. Phys. 1977, 67, 446–457.
ꢀzyy, ꢀzxx, ꢀyyz, and ꢀxxz. Assuming Kleinman symmetry, ꢀzyy
)
ꢀyyz and ꢀzxx ) ꢀxxz, and if the structure is essentially 2D, then
9
10506 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Diquat Derivatives: 2D Nonlinear Optical Chromophores
A R T I C L E S
Table 5. HRS Data and Depolarization Ratios for Salts [3-18][PF6]2 and [DAS]PF6 in Acetonitrile
a
c
F
d
d
cation
ꢀ〈ꢀ2HRS
〉
(10-30 esu)
ꢀ0b (10-30 esu)
k
ꢀzzz (10-30 esu)
ꢀzyy (10-30 esu)
3
4
5
6
7
8
9
115 ( 50
272 ( 9
58 ( 29
148 ( 5
233 ( 14
261 ( 23
58 ( 5
130 ( 12
144 ( 12
126 ( 9
68 ( 28
163 ( 10
179 ( 54
256 ( 46
92 ( 9
2.9 ( 0.4
3.0 ( 0.5
3.5 ( 0.5
4.1 ( 0.6
3.4 ( 0.5
3.6 ( 0.5
3.3 ( 0.5
3.3 ( 0.5
2.7 ( 0.4
3.2 ( 0.5
2.5 ( 0.4
2.7 ( 0.4
2.9 ( 0.4
3.4 ( 0.5
2.7 ( 0.4
2.3 ( 0.3
277 ( 120
655 ( 22
1000 ( 60
1200 ( 100
310 ( 30
615 ( 60
630 ( 55
540 ( 40
25 ( 10
150 ( 25
50 ( 15
85 ( 20
70 ( 10
414 ( 25
498 ( 43
128 ( 12
255 ( 25
263 ( 22
224 ( 16
140 ( 58
300 ( 18
321 ( 97
473 ( 86
212 ( 20
276 ( 25
190 ( 15
603 ( 32
440 ( 30
10
11
12
13
14
15
16
17
18
DAS+
9
230 ( 100
460 ( 75
520 ( 175
770 ( 180
335 ( 60
420 ( 75
310 ( 50
980 ( 150
3
10
9
5
3
138 ( 13
101 ( 8
339 ( 18
110 ( 7
140 ( 25
35 ( 6
100 ( 15
9
10
e
a Orientationally averaged ꢀ without any assumption of symmetry or contributing tensor elements, measured by using an 800 nm Ti3+:sapphire laser.
b
The quoted cgs units (esu) can be converted into SI units (C3 m3 J-2) by dividing by a factor of 2.693 × 1020. Static ꢀ estimated from ꢀ〈ꢀH2 RS〉 via the
two-state model.25 Note that the uncertainties quoted are merely statistical and do not account for any errors introduced by two-state extrapolation.
c Depolarization ratio. d ꢀ tensor components derived from the HRS intensity and F. e Data taken from refs 8b and c.
ꢀzxx ) ꢀxxz ) 0, so only ꢀzzz and ꢀzyy are significant. These DS
diquat derivatives have a somewhat distorted C2V geometry that
retains C2 symmetry. The calculated tensor components are
nevertheless analogous to those of C2V molecules, being
dominated by the ꢀzzz and ꢀzyy terms. In order to assess the
importance of “off-diagonal” tensor components, we have
measured HRS depolarization ratios F for salts [3-18][PF6]2
and these are included in Table 5. The parameter F is the ratio
of the intensities of the scattered SH light polarized parallel
and perpendicular to the polarization direction of the funda-
mental beam.26 A F value of 5 is the upper limit for purely
dipolar symmetry, corresponding with a single dipolar ꢀzzz tensor
component, and under ideal experimental conditions. Realistic
experimental conditions, such as the use of a condenser system
to collect the SH light, will induce a lowering of the observed
F value. The octupolar reference compound crystal violet gives
a F value of 1.5.
Treatment of our experimental data (for the method, see the
Supporting Information) affords the ꢀzzz and ꢀzyy values shown
in Table 5. All of the MS compounds [3-10][PF6]2 exhibit a F
value of at least 2.9, and up to 4.1, with an average of 3.4 (
0.3 clearly confirming the dominance of a single ꢀzzz tensor
component. The DS chromophores in [11-18][PF6]2 show F
values between 2.3 and 3.4, with an average of 2.8 ( 0.3,
significantly lower than for their MS counterparts. Because the
MS species are predominantly 1D dipoles, we have not
attempted to derive their ꢀzyy values, but the shapes of the DS
species do justify such analyses. In each case, ꢀzyy dominates
markedly over ꢀzzz (Table 5). Given that the two types of
chromophore have been treated differently, assuming purely 1D
dipolar symmetry (C∞V) for the MS species but simplified C2V
symmetry for the DS derivatives, comparing the ꢀzzz values for
MS/DS pairs is inappropriate. However, for the DS species, both
Stark Spectroscopic Studies. Stark spectroscopy27 affords
dipole-moment changes that can be used with other parameters
to estimate ꢀ0 values via the two-state model.25 Data obtained
for salts [3-18][PF6]2 in butyronitrile at 77 K are shown in
Table 6, and representative spectra for [9][PF6]2, [10][PF6]2 and
[18][PF6]2 are shown in Figure 9.
The ICT bands of [3-18][PF6]2 generally show slight blue
shifts on moving from acetonitrile solutions to butyronitrile
glasses (Tables 2 and 6). All of the trends in Emax noted at room
temperature are maintained at 77 K. The band intensities at 77
K, quantified by fos and µ12 (summed for [11-18][PF6]2) show
trends that mostly agree with those already noted for the ε values
(see above). Moving from n ) 1 to 2 always increases µ12, while
fos generally increases only for the EDQ2+ species. Replacing
Dap with Jd gives broadly similar behavior. In every case, either
replacing EDQ2+ with PDQ2+ or moving from a MS to its DS
counterpart increases fos and µ12, and the latter structural change
always gives by far the largest increases in absorption intensity.
The dipole-moment changes and electron-transfer distances
(∆µ12, ∆µab, r12, and rab; averaged for [11-18][PF6]2) logically
always increase with molecular length on moving from n ) 1
to 2 (Table 6). However, swapping Dap for Jd gives no clear
trends for these parameters. In most cases, replacing EDQ2+
with PDQ2+ gives small decreases in ∆µ12, ∆µab, r12, and rab,
while moving from a MS to the related DS dication slightly
2
increases these parameters. The terms cb and Hab (again,
averaged for [11-18][PF6]2) quantify π-ED-EA coupling and
orbital mixing and always decrease, or show no significant
change, on moving from n ) 1 to 2. These observations are
consistent with the 1H NMR studies (see above), and predictable
because π-orbital overlap weakens over longer distances.
2
Generally, cb and Hab also decrease on moving from a MS
chromophore to its DS analogue. Replacing Dap with Jd does
not generally change cb2 or Hab significantly. Comparing EDQ2+
and PDQ2+ pairs reveals similar cb values, but significantly
2
ꢀ
zzz and ꢀzyy generally increase significantly when moving from
higher Hab values for the EDQ2+ species.
n ) 1 to 2. Although these tensor components show no clear
dependence on the nature of the π-ED or diquat units, k (i.e.,
ꢀzyy/ꢀzzz) always increases on replacing Dap with Jd.
(27) (a) Liptay, W. In Excited States; Lim, E. C., Ed.; Academic Press,
New York, 1974, Vol. 1, pp 129-229. (b) Bublitz, G. U.; Boxer, S. G.
Annu. ReV. Phys. Chem. 1997, 48, 213–242. (c) Vance, F. W.;
Williams, R. D.; Hupp, J. T. Int. ReV. Phys. Chem. 1998, 17, 307–
329. (d) Brunschwig, B. S.; Creutz, C.; Sutin, N. Coord. Chem. ReV.
1998, 177, 61–79.
(26) Heesink, G. J. T.; Ruiter, A. G. T.; van Hulst, N. F.; Bo¨lger, B. Phys.
ReV. Lett. 1993, 71, 999–1002.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10507

A R T I C L E S
Coe et al.
Table 6. ICT Absorption and Stark Spectroscopic Data for Salts [3-18][PF6]2 and [DAS]PF6
i
Hab
(103 cm-1
ꢀ0j
Σꢀ0
a
a
a
Emax (eV)
b
fos
c
d
e
f
r12 (Å)
g
rab (Å)
h
cation
νmax (cm-1
)
λmax (nm)
µ12 (D)
∆µ12 (D)
∆µab (D)
cb2
)
(10-30 esu)
(10-30 esu)
3k
4
5
6
7
8
9
17762
563
596
617
656
537
574
585
623
2.20
0.50
0.51
0.57
0.63
0.66
0.73
0.65
0.74
7.7
8.0
8.6
9.4
8.7
9.5
9.0
9.9
11.0
7.1
7.1
7.7
6.9
3.3
10.9
6.4
6.9
8.9
6.6
5.1
10.9
5.9
6.6
10.1
7.0
3.7
11.5
5.8
6.2
10.1
7.7
9.1
20.6
26.4
20.5
26.6
18.1
24.7
17.5
24.4
23.6
25.7
30.0
30.1
34.0
16.0
20.4
23.6
27.6
30.2
32.3
19.2
20.5
23.2
24.0
26.3
24.9
19.6
21.5
24.9
23.6
27.4
25.5
16.3
25.8
31.0
26.8
32.9
25.3
31.1
25.3
31.5
32.2
29.4
33.1
33.8
36.7
17.3
29.9
26.8
30.9
35.1
34.9
21.8
29.9
26.1
27.4
33.1
28.6
21.0
31.5
27.5
26.7
34.1
29.8
24.4
4.3
5.5
4.3
5.5
3.8
5.1
3.6
5.1
4.9
5.6
6.2
6.3
7.1
3.3
4.3
4.9
5.8
6.3
6.7
4.0
4.3
4.8
5.0
5.5
5.2
4.1
4.5
5.2
5.8
6.1
6.9
3.4
5.4
6.5
5.6
6.9
5.3
6.5
5.3
6.6
6.7
6.1
6.9
7.0
7.6
3.6
6.2
5.6
6.4
7.3
7.3
4.5
6.2
5.4
5.7
6.9
6.0
4.4
6.6
5.7
6.4
6.7
7.3
5.1
0.10
0.07
0.12
0.09
0.13
0.10
0.15
0.11
0.13
0.06
0.05
0.05
0.04
0.04
0.16
0.06
0.05
0.07
0.04
0.06
0.16
0.05
0.06
0.10
0.06
0.03
0.16
0.05
0.05
0.10
0.07
0.17
5.3
4.3
5.2
4.4
6.5
5.3
6.1
5.1
6.0
5.5
3.4
4.0
4.0
2.7
6.0
5.0
3.1
4.0
3.8
4.2
7.0
5.1
4.0
5.5
5.1
2.8
6.4
4.4
3.5
4.9
4.9
7.7
297
456
439
778
305
525
372
705
688
188
462
445
266
63
679
163
500
725
275
119
498
121
289
617
220
79
16779
16212
15244
18631
17421
17099
16050
2.08
2.01
1.89
2.31
2.16
2.12
1.99
10
11k
17911 (17730) 558 (564) 2.22 (2.20) 1.00
22518 (22883) 444 (437) 2.79 (2.84) 0.54
16857 (15728) 593 (636) 2.09 (1.95) 0.37
876
12
13
14
15
16
17
18
1173
(17502)
(571)
(2.17)
0.49
21212 (21454) 471 (466) 2.63 (2.66) 0.48
16615 (14437) 602 (693) 2.06 (1.79) 0.07
(16494)
20890 (21172) 478 (472) 2.59 (2.63) 0.40
15486 (14126) 646 (708) 1.92 (1.75) 0.31
(15888)
19455 (19757) 514 (506) 2.41 (2.45) 0.41
18954 (17825) 528 (561) 2.35 (2.21) 0.20
(19196)
(22584)
17664 (16615) 566 (602) 2.19 (2.06) 0.34
(18147)
(20608)
17422 (16091) 574 (622) 2.16 (2.00) 0.11
(17462)
(20648)
16292 (15244) 614 (656) 2.02 (1.89) 0.28
(16615)
(18873)
905
1500
738
(607)
(2.05)
0.92
(629)
(1.97)
0.60
(521)
(443)
(2.38)
(2.80)
1.10
0.40
1126
943
(551)
(485)
(2.25)
(2.56)
0.86
0.48
(573)
(484)
(2.17)
(2.56)
1.09
0.33
712
152
297
770
323
236
1390
(602)
(530)
480
(2.06)
(2.34)
2.58
0.80
0.53
0.80
DAS+ l 20833
a In butyronitrile at 77 K; observed absorption maxima, maxima for Gaussian fitting functions for [11-18][PF6]2 in brackets. Data in all subsequent
b
columns relate to fitted curves if used. For [3-10][PF6]2, obtained from (4.32 × 10-9 M cm2)A where A is the numerically integrated area under the
absorption peak; for [11-18][PF6]2, obtained from (4.60 × 10-9 M cm2)εmax × fw1/2, where εmax is the maximal molar extinction coefficient and fw1/2 is
the full width at half height (in wavenumbers). c Calculated using eq 2. d Calculated from fint∆µ12 using fint ) 1.33. e Calculated from eq 1. f Delocalized
electron-transfer distance calculated from ∆µ12/e. g Effective (localized) electron-transfer distance calculated from ∆µab/e. h Calculated from eq 3.
i Calculated from eq 4. j Calculated from eq 5. k Data taken in part from ref 13. l Data taken in part from ref 8c.
Figure 9. Stark spectra and calculated fits for salts [9][PF6]2, [10][PF6]2, and [18][PF6]2 in an external electric field of 3.51 × 107 V m-1. (Top panel)
Absorption spectrum illustrating Gaussian curves used in data fitting for [18][PF6]2. (Bottom panel) Electroabsorption spectrum, experimental (blue) and fits
(green) according to the Liptay equation.27a
Given the strongly limited validity of the ꢀ0 values derived
via HRS data (see above), due to resonance and the inherent
complexity of the chromophores under study, and the fact that
no other direct experimental method for determining such data
exists, Stark measurements provide a useful means to assess
the intrinsic NLO responses of our new molecules. Therefore,
ꢀ0 values have been estimated by using eq 5 (Table 6). It should
be emphasized that the two-state model is applied in a different
sense here when compared with the HRS studies. The latter
are inherently able to probe only the NLO responses of whole
9
10508 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Diquat Derivatives: 2D Nonlinear Optical Chromophores
A R T I C L E S
chromophores, while in contrast, the Stark observations allow
the system to be broken down into its component electronic
excitation processes which may be viewed as having two-state
nature. While including the important benefit of completely
avoiding complications due to resonance effects, this indirect
analytical approach is nevertheless only an approximation that
involves three assumptions: (i) all of the visible absorption arises
from ICT transitions, so that even if a band is resolved into
components, each of these can be adequately described as
involving a ground and single excited state; (ii) each Gaussian
curve corresponds with a single ICT process; (iii) the relative
directionality of µ12 is not important in the two-state model.25
Assumption (ii) may be somewhat tentative given that the TD-
DFT calculations (see below) predict up to 5 ICT transitions of
substantial intensity for some of the DS species, but experi-
mentally this represents a “best case approximation” scenario.
Literature precedent indicates that assumption (iii) is reasonable
(see eqs 40 and 41 in ref 6a).28
For [11-18][PF6]2, the total NLO response (Σꢀ0) is taken as
the sum of those for the Gaussian curves. Notably, related
studies have shown that using such an approach usually gives
total ꢀ0 values similar to those obtained without spectral
deconvolution.18,29 It is also important to note that other two-
state descriptions exist, which would change the quoted ꢀ0
values by constant factors of 0.5 or 2.30 Therefore, these data
are useful primarily for revealing structure-activity relation-
ships, rather than necessarily indicating the absolute magnitudes
of ꢀ0. Nevertheless, it is remarkable that previous studies with
1D RuII complexes show that using the two-state eq 5 with a
prefactor of 3 (the “perturbation series convention”)30 often gives
excellent quantitative agreement with the ꢀ0 values determined
via HRS with a 1064 nm laser.31 Although the general
applicability of this basic two-state model is certainly limited
and refinements have been proposed,32 recent theoretical studies
confirm that good estimates of ꢀ0 can be obtained for 1D dipoles
via even such a simplified approach.33 Furthermore, previous
experimental observations such as our own31 show that eq 5
remains a viable tool when considering the molecular NLO
responses associated with single ICT transitions. The novelty
of the present study lies in extending our Stark-based treatment
to more complicated 2D systems, following the original
precedent set by Vance and Hupp.7e While we acknowledge
that our approach is only an approximation from a physical
perspective, the fact that treating all the compounds in the same
manner affords consistent and plausible trends (see below) lends
further credence to the results.
The Stark-derived ꢀ0 values show several very clear trends,
reinforcing those noted in the HRS data (see above). First,
extending the π-conjugation always markedly increases ꢀ0, with
the largest (ca. 2-fold) change for the pair [9/10][PF6]2. Such
behavior is expected by comparison with simpler organic
molecules,1 and attributable to mainly the combination of
decreasing Emax and increasing ∆µ12 values (Table 6). Interest-
ingly, the largest relative ꢀ0 increases are found for the MS
compounds. Our previous studies with 1D pyridinium and
related species showed that replacing Dap with the stronger
π-ED group Jd usually (and logically) increases ꢀ0.18 A similar
pattern is observed for the diquat compounds, although the
difference is not significant for the pair [11/13][PF6]2. For the
other seven pairs, the increases in ꢀ0 lie in the range ca.
20-70%, and are largely due to the decreases in Emax. In
contrast, the ꢀ0 responses of the EDQ2+/PDQ2+ pairs do not
differ significantly, because the blue-shifting of the ICT transi-
tions on extending the quaternized bridge is offset by their
increased intensities. The most substantial increases in ꢀ0 (ca.
45-195%) are generally observed on moving from a MS
compound to its DS counterpart. This trend is attributable
primarily to the greatly increased µ12 values for the DS
chromophores (see above).
Since DAST is now a technologically valuable material, it is
important to compare the NLO responses of our new chro-
mophores with that of the DAS+ cation. Many of the diquat-
based salts show HRS-derived ꢀ0 values larger than that of
[DAS]PF6,8b,c with the value for [18][PF6]2 being about 3 times
larger (Table 5). However, given the limitations introduced by
resonance effects and electronic structural differences in these
HRS data, it is probably more instructive to draw comparisons
using the Stark-derived results. By using this approach, we
obtained a ꢀ0 value for [DAS]PF6 that is significantly smaller
than any of those derived for the new compounds (Table 6).
Indeed, the Σꢀ0 responses estimated for [12][PF6]2, [14][PF6]2,
[16][PF6]2, and [18][PF6]2 are ca. 5-6-fold increased when
compared with [DAS]PF6. These observations provide a strong
incentive to pursue studies on polar materials incorporating these
new symmetrical diquat cations, such as [15][NapSO3]2 ·2H2O
(see above), in which exceptionally large and 2D bulk NLO
effects may be anticipated.
Theoretical Studies. The electronic structures of a selection
of the new diquat chromophores (3, 7, 11, 12, 14-16, and 18)
have been studied by using density functional theory (DFT)
methods. These calculations give py/py angles of 48-51° in
the PDQ2+ derivatives, while smaller twists of 23-24° are
predicted for the EDQ2+ species. The latter are only a little
smaller than the value determined crystallographically for the
salt [13][OTf]2 ·2MeCN (Table 4), while the predictions for the
PDQ2+ species are close to the average angle (52.9°) observed
for the eight crystal structures we report. The parameters
involved in eq 5 were determined by time-dependent DFT (TD-
DFT) calculations, and ꢀ values were calculated from the
numerical second derivative of the dipole moment with respect
to the applied electric field by using the finite field (FF)
approach. The results of these calculations are shown in Table
7. While a recent comparative study of quantum mechanical
methods for predicting ꢀ0 (including FF-DFT) concluded that
such approaches are relatively reliable for pseudolinear dipoles,34
(28) Coe, B. J.; Harris, J. A.; Jones, L. A.; Brunschwig, B. S.; Song, K.;
Clays, K.; Gar´ın, J.; Orduna, J.; Coles, S. J.; Hursthouse, M. B. J. Am.
Chem. Soc. 2005, 127, 4845–4859.
(29) Coe, B. J.; Foxon, S. P.; Harper, E. C.; Helliwell, M.; Raftery, J.;
Swanson, C. A.; Brunschwig, B. S.; Clays, K.; Franz, E.; Gar´ın, J.;
Orduna, J.; Horton, P. N.; Hursthouse, M. B. J. Am. Chem. Soc. 2010,
132, 1706–1723.
(30) Willetts, A.; Rice, J. E.; Burland, D. M.; Shelton, D. P. J. Chem. Phys.
1992, 97, 7590–7599.
(31) Coe, B. J.; Harris, J. A.; Brunschwig, B. S. J. Phys. Chem. A 2002,
106, 897–905.
(32) Selected examples: (a) Meshulam, G.; Berkovic, G.; Kotler, Z. Opt.
Lett. 2001, 26, 30–32. (b) Woodford, J. N.; Wang, C. H.; Jen, A. K.-
Y. Chem. Phys. 2001, 271, 137–143. (c) Di Bella, S. New J. Chem.
2002, 26, 495–497. (d) Tai, O. Y.-H.; Wang, C. H.; Ma, H.; Jen, A. K.-
Y. J. Chem. Phys. 2004, 121, 6086–6092. (e) Kuzyk, M. G. Phys.
ReV. A 2005, 72, 053819-1–053819-5. (f) Campo, J.; Wenseleers, W.;
Goovaerts, E.; Szablewski, M.; Cross, G. H. J. Phys. Chem. C 2008,
112, 287–296.
(34) Isborn, C. M.; Leclercq, A.; Vila, F. D.; Dalton, L. R.; Bre´das, J. L.;
Eichinger, B. E.; Robinson, B. H. J. Phys. Chem. A 2007, 111, 1319–
1327.
(33) Champagne, B.; Kirtman, B. J. Chem. Phys. 2006, 125, 024101-1–
024101-7.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10509

A R T I C L E S
Coe et al.
Table 7. Calculated Electronic Transitions (TD-B3P86/6-31G*) and Finite Field Hyperpolarizabilities for the Diquat Cations 3, 7, 11, 12,
14-16, and 18 and DAS+
a
Emax (eV)
a
a
fos
a,b
∆µ12 (D)
c
c
cation
symmetrya
µ12 (D)
major contributionsa
ꢀzzz (10-30 esu)
ꢀzyy (10-30 esu)
Σ[ꢀFF]c(10-30 esu)
ꢀ
zyy/ꢀzzz
3
7
1.45
2.69
1.51
2.34
2.79
1.73
1.86
2.59
2.68
2.77
2.95
1.58
1.70
2.31
2.47
2.57
2.72
1.53
1.64
2.25
2.41
2.51
2.66
1.87
1.95
2.50
2.72
1.69
1.76
2.24
2.50
2.84
2.89
1.63
1.70
2.17
2.45
2.77
2.82
2.65
6.61
11.52
5.98
5.76
10.55
10.66
3.86
6.93
6.25
7.29
9.26
14.06
3.68
9.98
8.81
7.81
9.28
14.82
3.53
9.97
8.95
7.96
9.65
8.75
2.90
11.06
9.39
11.18
2.70
14.25
11.38
5.18
5.29
11.72
2.56
14.64
11.72
5.13
5.38
11.5
0.24
1.35
0.20
0.30
1.18
0.75
0.11
0.47
0.40
0.56
0.96
1.18
0.09
0.87
0.73
0.60
0.89
1.27
0.08
0.85
0.73
0.60
0.94
0.54
0.06
1.16
0.91
0.80
0.05
1.72
1.23
0.29
0.31
0.85
0.04
1.77
1.28
0.28
0.31
1.32
26.07
7.83
H f L
H f L+1
H f L
277
32.17
33.36
12.67
10.64
10.53
8.10
11.13
14.06
10.56
10.45
10.47
6.95
10.78
14.60
10.68
9.90
9.88
6.35
10.43
14.38
9.62
9.98
9.82
7.08
8.94
9.79
9.69
407
353
H f L+1
H f L+2
H f L
H-1 f L
H-1 f L+1
H f L+1
H f L+2
H-1 f L+2
H f L
H-1 f L
H-1 f L+1
H f L+1
H f L+2
H-1 f L+2
H f L
H-1 f L
H-1 f L+1
H f L+1
H f L+2
H-1 f L+2
H f L
H-1 f L
H-1 f L+1
H f L+1
H f L
H-1 f L
H-1 f L+1
H f L+1
H f L+2
H-1 f L+2
H f L
H-1 f L
H-1 f L+1
H f L+1
H f L+2
H-1 f L+2
H f L
11
12
14
B
A
B
A
B
A
B
A
B
A
B
A
B
A
B
A
B
A
B
A
B
A
B
A
B
A
B
A
B
A
B
A
B
A
64
88
78
289
527
573
4.5
6.0
7.3
615
650
15
16
56
80
213
382
269
462
3.8
4.8
6.60
9.12
14.53
12.44
9.20
9.09
5.91
8.32
18
75
408
482
150
5.4
14.38
11.92
13.9
d
DAS+
a TD-B3P86/6-31G*, only transitions with excitation energies below 3.0 eV are included; H ) HOMO, L ) LUMO. b Derived from the excited state
dipole moment calculated by using the RhoCI density. c Static hyperpolarizability calculated by using FF-B3P86/6-31G*. d Data taken from ref 8d
(B3P86/LanL2DZ model Chemistry).
we note that the FF-DFT method has been applied only rarely
to 2D species.6p,13,28,29
ꢀ0 (ꢀFF) values are very large and show a substantial increase
for the PDQ2+ derivative when compared with its EDQ2+
analogue. This result contrasts with the HRS and Stark studies
which afford very similar ꢀ0 values for [3][PF6]2 and [7][PF6]2
(Tables 5 and 6). For comparison, the DAS+ cation gives ꢀFF
) 150 × 10-30 esu,8d confirming the expected enhancements
for the new diquat chromophores. This result provides confir-
mation of the (at least approximate) validity of our experimen-
tally derived ꢀ0 data because the FF method calculates the first
hyperpolarizability as the second derivative of the dipole
moment with respect to an applied electric field, and is
completely independent of any potential concerns over the
applicability of the two-state model.
The MS cations 3 and 7 resemble stilbazolium species, with
lowest energy (LE) electronic transitions of HOMO f LUMO
character.8e Figures S8 and S9 (Supporting Information) show
the topologies of the MOs involved. In each case, the HOMO
is primarily Dap-based, but with significant contributions from
the ethenyl bridge and attached pyridyl ring. The LUMO is
located mostly on the diquat units (and especially the picolyl
ring), while the LUMO+1 and LUMO+2 also feature some
contributions from the rest of the molecule. TD-DFT does not
accurately predict the observed ICT energies (Tables 2 and 6),
underestimating Emax by ca. 0.7-0.8 eV, but the trend of Emax
increasing on replacing EDQ2+ with PDQ2+ is reproduced. These
calculations indicate that both the HOMO and LUMO are
destabilized on moving from 3 to 7, but this effect is larger for
the LUMO. The second lowest energy (SLE) transition predicted
has HOMO f LUMO+1 character for both cations, with an
Emax somewhat closer to (but higher than) the experimental
value. However, these transitions show a reverse trend of
decreasing Emax on moving from 3 to 7. The predicted values
of µ12 and ∆µ12 for the LE transitions agree only moderately
with those measured at 77 K (Table 6). The FF-DFT-derived
The overall picture arising from our calculations on the six
DS cations is similar to that described previously for V-shaped
RuII complexes.28,29 Thus, all of the chromophores have C2
symmetry and their MOs are symmetric (a) or antisymmetric
(b) with respect to the 2-fold axis (z in the standard orientation).
The electronic transitions between orbitals with the same
symmetry (A transitions) are polarized along the z axis and
contribute to the ꢀzzz tensor component, while the transitions
9
10510 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Diquat Derivatives: 2D Nonlinear Optical Chromophores
A R T I C L E S
Figure 10. Illustrations of the contour surface diagrams of the molecular
orbitals of 14 involved in the six lowest energy transitions (isosurface value
0.04 au), showing the axis convention used in all of the calculations.
Figure 11. Illustrations of the contour surface diagrams of the molecular
orbitals of 18 involved in the six lowest energy transitions (isosurface value
0.04 au).
between orbitals with different symmetries (B transitions) are
polarized perpendicular to the z axis and therefore contribute
to ꢀzyy.
empirical trends in ICT energies are all predicted by these
calculations. Thus, Emax for both transitions decreases on moving
from n ) 1 to 2 (pairs 11/12 and 15/16) or on replacing Dap
with Jd (pairs 12/14 and 16/18), but increases on moving from
an EDQ2+ derivative to its PDQ2+ counterpart (pairs 11/15, 12/
16, and 14/18). While the measurements show only small
increases in Emax (e0.06 eV) for the DS species when compared
with their MS analogues, more significant increases are predicted
for the LE transition energies of the pairs 3/11 (0.28 eV) and
7/15 (0.36 eV). Again, the extent of agreement between the
predicted and measured values of µ12 and ∆µ12 (Table 6) is
generally poor, and in particular, the TD-DFT-derived ∆µ12
values are all 2-3 times smaller than those determined from
the Stark studies and also those predicted for 3 and 7.
The main point of note concerning the FF-DFT-derived ꢀ
values is that the off-diagonal tensor component ꢀzyy is always
larger than ꢀzzz, by a factor of ca. 4-7. Coupled perturbed
Hartree-Fock calculations on 11 predicted the same general
behavior,13 although according to this method the ꢀzyy/ꢀzzz ratio
is smaller (2.1) than that derived via FF-DFT (4.5). The results
for the pairs 11/12 and 15/16 indicate that the relative dominance
of ꢀzyy increases as the π-conjugation extends. Furthermore, the
ratio ꢀzyy/ꢀzzz also increases on replacing Dap with Jd (pairs 12/
14 and 16/18), but decreases on moving from an EDQ2+
derivative to its PDQ2+ analogue (pairs 11/15, 12/16, and 14/
18). The Dap vs Jd trend, but only this one, is also shown by
the HRS data (Table 5).
The calculated electronic structures and excitation properties
of 11, 12, 14-16, and 18 are rather more complicated than for
3 and 7. For each cation, the LE transition is again of HOMO
f LUMO character and B-polarized, but an A-polarized SLE
(HOMO-1 f LUMO) transition is also predicted, with Emax
being only ca. 0.1 eV above that of the LE transition (Table 7).
Another pair of transitions appears at rather higher energies; a
B-polarized HOMO-1 f LUMO+1 and an A-polarized
HOMO f LUMO+1, separated by 0.07-0.28 eV. For all of
the chromophores except 15, a third pair of transitions 0.05-0.18
eV apart is predicted at even higher energies (B-polarized
HOMO f LUMO+2 and A-polarized HOMO-1 f LU-
MO+2). Figures 10 and 11 show the topologies of the MOs
involved for 14 and 18, while the corresponding diagrams for
the other species are included in the Supporting Information
(Figures S10-S13). The HOMO and HOMO-1 are always
mostly located on the electron-rich Dap or Jd phenyl groups,
but with significant contributions from the ethenyl/butadienyl
bridges and the diquat units. The LUMO, LUMO+1 and
LUMO+2 are concentrated on the electron-accepting diquat
fragments, with the LUMO+2 having especially limited con-
tributions from the rest of the molecule. In qualitative terms,
the electronic structure is not greatly influenced by the various
structural changes made.
As for 3 and 7, the TD-DFT Emax values for the LE and SLE
transitions in 11, 12, 14-16, and 18 are somewhat lower than
those measured experimentally (Tables 2 and 6). However, the
When considering the total FF-DFT-derived ꢀ0 (Σ[ꢀFF])
values, the following trends are predicted: (i) Σ[ꢀFF] increases
on moving from n ) 1 to 2, or on replacing Dap with Jd, but
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10511

A R T I C L E S
Coe et al.
the effect of the latter change is much less significant; (ii) Σ[ꢀFF]
decreases on moving from an EDQ2+ derivative to its PDQ2+
counterpart for the DS pairs 11/15, 12/16, and 14/18, but the
opposite is found for the MS pair 3/7. These predictions
generally agree qualitatively with the conclusions from the Stark
studies (see above), providing further validation of our experi-
mental approach, but the EDQ2+ vs PDQ2+ comparisons are
more clear-cut according to FF-DFT. However, the effect of
moving from a MS derivative to its DS counterpart is not
reproduced consistently. While the Stark results indicate that
this change always markedly increases the total ꢀ0 response,
the FF-DFT results show an increase in moving from 3 to 11,
but a decrease for the pair 7/15.
TD-DFT computations on selected chromophores reveal a
relatively high multiplicity of visible ICT transitions for the DS
species. FF-DFT calculations predict for the DS chromophores
that ꢀzyy is always larger than ꢀzzz, by a factor of ca. 4-7. In
terms of the trends in the total ꢀ responses, the FF-DFT
approach also gives a generally good qualitative level of
agreement with the Stark studies. In addition to these detailed
insights, our studies afford several conclusions of potentially
wide-reaching significance. First, ꢀ0 values ca. 6 times larger
than that of the chromophore in the commercial NLO material
(E)-4′-(dimethylamino)-N-methyl-4-stilbazolium tosylate are
achieved, by joining two π-EDS to a powerful diquat acceptor.
Second, the new species have strongly 2D NLO responses that
may be exploited for novel practical applications. Third, we
have demonstrated that such chromophores can form polar
crystalline materials, suitable for bulk NLO behavior. Finally,
the presence of two reversible reductions indicates potential for
redox-switchability over multiple states. Notably, redox-switch-
ing of NLO responses has been achieved hitherto only with
transition metal complexes.35
Conclusion
We have synthesized the first family of diquat-based NLO
chromophores. Their visible absorption spectra are dominated
by intense ICT bands, the energies of which show both
predictable (Dap vs Jd, n ) 1 f 2) and previously unobserved
(MS vs DS, EDQ2+ vs PDQ2+) trends. Cyclic voltammograms
show two reversible diquat-based reductions, and single-crystal
X-ray structures have been determined for nine salts. HRS
studies with an 800 nm laser reveal relatively large ꢀ responses,
with depolarization experiments showing dominant ‘off-
diagonal’ ꢀzyy tensor components for the DS compounds. Stark
spectroscopy allows the estimation of ꢀ0 values in the absence
of complications arising from the variable effects of resonance,
and affords structure-activity correlations according to the two-
state model. The main new insights emerging from the Stark-
based analyses are that the ꢀ0 responses of the EDQ2+ and
PDQ2+ pairs do not differ significantly, and the largest increases
in ꢀ0 occur on moving from a MS species to its DS counterpart.
Acknowledgment. We thank the EPSRC for support (Grants
EP/E000738 and EP/D070732) and also the Fund for Scientific
Research-Flanders (FWO-V, G.0297.04), the University of Leuven
(GOA/2006/3), MCyT-FEDER (CTQ2005-01368, CTQ2008-
02942), the NSF (Grant CHE-0802907, ‘Powering the Planet: an
NSF Center for Chemical Innovation’) and Gobierno de Aragon-
Fondo Social Europeo (E39). We thank Dr. Emma C. Fitzgerald
for assistance with obtaining the data for Figure 4, Kurt De Mey
for assistance with the HRS studies, and Dr. John E. Warren
(Synchrotron Radiation Source, CCLRC, Daresbury Laboratory,
Warrington WA4 4AD, U.K.) for assistance with the data collection
for [16][NapSO3]2 ·2H2O and [18][PF6]2 ·2Me2CO.
(35) Selected examples: (a) Coe, B. J.; Houbrechts, S.; Asselberghs, I.;
Persoons, A. Angew. Chem., Int. Ed. 1999, 38, 366–369. (b) Weyland,
T.; Ledoux, I.; Brasselet, S.; Zyss, J.; Lapinte, C. Organometallics
2000, 19, 5235–5237. (c) 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, 49–50. (d) Powell, C. E.;
Cifuentes, M. P.; Morrall, J. P.; Stranger, R.; Humphrey, M. G.; Samoc,
M.; Luther-Davies, B.; Heath, G. A. J. Am. Chem. Soc. 2003, 125,
602–610. (e) Sporer, C.; Ratera, I.; Ruiz-Molina, D.; Zhao, Y.-X.;
Vidal-Gancedo, J.; Wurst, K.; Jaitner, P.; Clays, K.; Persoons, A.;
Rovira, C.; Veciana, J. Angew. Chem., Int. Ed. 2004, 43, 5266–5268.
(f) Cifuentes, M. P.; Powell, C. E.; Morrall, J. P.; McDonagh, A. M.;
Lucas, N. T.; Humphrey, M. G.; Samoc, M.; Houbrechts, S.;
Asselberghs, I.; Clays, K.; Persoons, A.; Isoshima, T. J. Am. Chem.
Soc. 2006, 128, 10819–10832. (g) Boubekeur-Lecaque, L.; Coe, B. J.;
Clays, K.; Foerier, S.; Verbiest, T.; Asselberghs, I. J. Am. Chem. Soc.
2008, 130, 3286–3287.
Supporting Information Available: Experimental details;
additional H NMR figure; further crystallographic data and
1
figures; X-ray powder diffraction data for [15][NapSO3]2 ·2H2O;
crystallographic information in CIF format, including [7][Nap-
SO3]2, [7][PF6]2 ·MeCN, and [9][PF6]2 ·MeCN; method for
derivation of ꢀ components from HRS data; Cartesian coordi-
nates of theoretically optimized geometries for the cations 3, 7,
11, 12, 14-16, and 18; contour surface diagrams of the MOs
involved in the lowest energy transitions for the cations 3, 7,
11, 12, 15, and 16. This material is available free of charge via
the Internet at http://pubs.acs.org.
JA103289A
9
10512 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010