Unimolecular electrical rectification in hexadecylquinolinium tricyanoquinodimethanide

Article abstract of DOI:10.1021/ja971811e

Macroscopic and nanoscopic current-voltage measurements reveal asymmetries in the DC electrical conductivity through Langmuir-Blodgett multilayers and even monolayers of γ-(n-hexadecyl)quinolinum tricyanoquinodimethanide, C₁₆H₃₃Q-3CN

Full text of DOI:10.1021/ja971811e

J. Am. Chem. Soc. 1997, 119, 10455-10466
10455
Unimolecular Electrical Rectification in Hexadecylquinolinium
Tricyanoquinodimethanide
Robert M. Metzger,*,† Bo Chen,† Ulf Ho1pfner,†,| M. V. Lakshmikantham,†
Dominique Vuillaume,†, Tsuyoshi Kawai,†,# Xiangli Wu,† Hiroaki Tachibana,†,∇
Terry V. Hughes,† Hiromi Sakurai,† Jeffrey W. Baldwin,† Christina Hosch,†,O
Michael P. Cava,† Ludwig Brehmer,‡ and Geoffrey J. Ashwell§
Contribution from the Laboratory for Molecular Electronics, Chemistry Department, UniVersity of
Alabama, Tuscaloosa, Alabama 35487-0336, Solid State Physics Institute, UniVersity of Potsdam,
Am Neuen Palais 10, D-14469 Potsdam, Germany, and Centre for Molecular Electronics,
Cranfield UniVersity, Cranfield, Bedfordshire MK43 0AL, United Kingdom
ReceiVed June 2, 1997X
Abstract: Macroscopic and nanoscopic current-voltage measurements reveal asymmetries in the DC electrical
conductivity through Langmuir-Blodgett multilayers and even monolayers of γ-(n-hexadecyl)quinolinum tricyano-
quinodimethanide, C16H33Q-3CNQ (5). These asymmetries are due to a transition of the ground-state zwitterion to
an excited-state conformer which is probably neutral. Unimolecular electrical rectification by monolayers of 5 is
unequivocally confirmed.
1. Introduction
Electrical rectification by Langmuir-Blodgett multilayers
(doped with electron donors (D), topped by insulating LB
monolayers, then topped by multilayers “doped” with electron
acceptors (A)) was demonstrated by Kuhn and co-workers8 and
confirmed by other groups.9,10 An electrochemical Langmuir-
Blodgett photodiode was produced by Fujihira;11 electrochemical
rectification at a monolayer-modified electrode has been re-
ported.12,13 Aviram, Pomerantz, and co-workers attached a
porphyrin covalently to a carboxylated highly oriented pyrolytic
graphite (HOPG) surface and studied it by scanning tunneling
microscopy (STM).14 An unsymmetrical STM tunneling current
was also seen through an alkylated hexabenzocoronene depos-
ited on graphite15 (this may be due to some salt formation?)
and an oligophenylethynyl)benzenethiol on Au(111) and Ag-
(111).16
A major challenge has been to provide experimental verifica-
tion of the Aviram-Ratner (AR) ansatz17 of unimolecular
rectification, or asymmetrical electrical conduction, through the
molecular orbitals of a single D-σ-A molecule (“through-bond
tunneling”). Here D ) good one-electron donor with low
ionization potential, A ) good electron acceptor with high
electron affinity, and σ ) insulating saturated covalent “bridge”.17
This unimolecular rectification would work because the excited
zwitterionic state D+-σ-A- would be relatively accessible
from the ground neutral state D-σ-A, while the opposite
Molecular rectification of electrical current can be defined
as a strongly asymmetric flow of electrons through a molecule
because the conductivity from, say, the left “side” of the
molecule to its right “side” is very different from the conductiv-
ity from “right” to “left”.
Rectification of electrical current was first achieved almost
a century ago in vacuum diodes,1 then later in junctions of
p-doped Ge accosted to n-doped Ge.2 For inorganic semicon-
ductors, rectification will occur either (i) at an interface,
whenever a “Schottky barrier” is formed at the boundary
between phases with dissimilar work functions,3 or (ii) within
a much wider “depletion layer”, when the charge compensation
between differently doped regions extends several micrometers
within the semiconductor.2
Rectification by junctions of micrometer-thick layers of
n-doped organic semiconductors with micrometer-thick layers
of p-doped organic semiconductors is well-known.4-7
* Author to whom correspondence should be addressed.
† University of Alabama.
‡ University of Potsdam.
§ Cranfield University.
| Present address: Solid State Physics Institute, University of Potsdam,
Am Neuen Palais 10, D-14469 Potsdam, Germany.
Permanent address: Institut d’EÄ lectronique et de Microe´lectronique
du Nord, CNRS, P.O. Box 69, Avenue Poincare´, F-59652 Villeneuve d’Ascq
cedex, France.
(8) Polymeropoulos, E. E.; Mo¨bius, D.; Kuhn, H. Thin Solid Films 1980,
68, 173-190.
(9) Sugi, M.; Sakai, K.; Saito, M.; Kawabata, Y.; Iizima, S. Thin Solid
Films 1985, 132, 69-76.
# Permanent address: Department of Electronic Engineering, Osaka
University, 2-1 Yamada-oka, Suita, Osaka 565, Japan.
∇ Present address: National Institute of Materials and Chemical Research,
1-1 Higashi, Tsukuba, Ibaraki 305, Japan.
(10) Fischer, C. M.; Burghard, M.; Roth, S.; v. Klitzing, K. Europhys.
Lett. 1994, 28, 129-134.
(11) Fujihira, M.; Nishiyama, K.; Yamada, H. Thin Solid Films 1986,
132, 77-82.
O National Science Foundation Summer Research Employment for
Undergraduates Participant from Clarke College, Dubuque, IA.
X Abstract published in AdVance ACS Abstracts, October 1, 1997.
(1) Fleming, J. A. U. K. Patent 24,850 (appl. Nov. 16, 1904); U.S. Patent
803,684 (granted Apr. 19, 1905).
(12) Sato, Y.; Itoigawa, H.; Uosaki, K. Bull. Chem. Soc. Jpn. 1993, 66,
1032-1037.
(13) Alleman, K. S.; Weber, K.; Creager, S. E. J. Phys. Chem. 1996,
100, 17050-17058.
(2) Shockley, W. Bell System Technol. J. 1949, 28, 435-489.
(3) Schottky, W. Z. Phys. 1942, 118, 539-592.
(14) Pomerantz, M.; Aviram, A.; McCorkle, R. A.; Li, L.; Aschott, A.
G. Science 1992, 255, 1115-1118.
(4) Meinhard, J. E. J. Appl. Phys. 1964, 35, 3059-3060.
(5) Anderson, T. L.; Komplin, G. C.; Pietro, W. J. J. Phys. Chem. 1993,
97, 6577-6578.
(15) Stabel, A.; Herwig, P.; Mu¨llen, K.; Rabe, J. P. Angew. Chem., Int.
Ed. Engl. 1994, 34, 1609-1611.
(16) Dhirani, A.; Lin, P.-H.; Guyot-Sionnest, P.; Zehner, R. W.; Sita, L.
R. J. Chem. Phys. 1997, 106, 5249-5253.
(6) Pietro, W. J. AdV. Mater. 1994, 6, 239-242.
(7) Hamm, S.; Wachtel, H. J. Chem. Phys. 1994, 103, 10689-10695.
(17) Aviram, A.; Ratner, M. A. Chem. Phys. Lett. 1974, 29, 277-283.
S0002-7863(97)01811-8 CCC: $14.00 © 1997 American Chemical Society

10456 J. Am. Chem. Soc., Vol. 119, No. 43, 1997
Metzger et al.
zwitterion D--σ-A+ would lie several electronvolts higher in
energy and, therefore, be inaccessible, except at dielectric
breakdown: good organic one-electron donors (e.g., tetrathi-
afulvalene, TTF, 1) are poor acceptors, while good acceptors
(e.g., tetracyanoquinodimethane, TCNQ, 2) are very poor one-
electron donors:
odimethanide (P-3CNQ, 4), whose crystal structure23 revealed
that 4 is indeed a zwitterion, with a θ ) 30° twist23 between
the pyridinium ring and the central six-membered ring of
3CNQ-, and a ground-state calculated dipole moment of 26 D
(INDO; atom coordinates from the crystal structure).23 These
molecules, when made suitably amphiphilic by addition of a
long alkyl terminations, yield Z-type Langmuir-Blodgett films
that are excellent ø(2) materials.24 This class of molecules has
a very narrow and solvatochromic absorption band.24
The most interesting member of this class is (Z)-â-N-
hexadecyl-4-quinolinium-R-cyano-4-styryldicyanomethanide or
γ-(n-hexadecyl)quinolinium tricyanoquinodimethanide, C16H33Q-
3CNQ (5), which is blue in acetonitrile solution.24 The
D-σ-A f D+-σ-A- + (1-2 eV)
D-σ-A f D--σ-A+ + (9 eV)
(1)
(2)
Several such D-σ-A molecules suitable for LB film
formation were prepared and studied by simple methods, but
evidence of rectification was not found.18-20
zwitterionic nature of 5 can be seen by calling it a T-D+-π-
A- molecule, where T is the hexadecyl “tail” (needed to help
form good LB films), D+ is the quinolinium moiety, π is the
π-electron bridge, and A- is the tricyanoquinodimethanide
(3CNQ-) moiety. 5 forms Z-type Langmuir-Blodgett (LB)
multilayers with a very high ø(2) ) 180 pm V-l.25 The blue
zzz
narrow absorption band in solution may be due to an interva-
lence transfer band (IVT); when several molecules are packed
in a multilayer, an additional intermolecular charge transfer band
(ICT) may also occur. For a single crystal of 4, these two bands
had been identified securely by polarized spectroscopy.26 It is
believed24-26 that the ground state of 5 is very polar, while its
first excited state is less polar, of the type T-D-π-A (6) and
that the efficiency for frequency doubling is due to the transition
T-D+-π-A- f T-D-π-A
(4)
However, one such D-σ-A molecule, namely DDOP-C-
BHTCNQ, 3 (where D ) DDOP ) dodecyloxyphenyl ) weak
one-electron donor, C ) carbamate ) covalent σ bridge, A )
BHTCNQ ) bromo(hydroxyethoxy)tetracyanoquinodimethane
) strong one-electron acceptor) showed electrical asymmetries
in sandwiches
In 5, the quinolinium and tricyanoquinodimethanide rings may
be twisted by some unknown angle θ (which was 30° for 423).
The neutral excited state T-D-π-A might be planar (θ ) 0°);
if θ ) 90° then IVT ) 0, and the blue color should disappear.
When LB multilayers of 5 were sandwiched between a noble
metal electrode (Pt or Ag) and a magnesium electrode (protected
by an overlayer of Ag), then asymmetric currents through the
laterally macroscopic multilayer indicated electrical rectification
through the molecule27
Pt|LB multilayer of 3|Mg|Ag
(3)
despite the fact that DDOP is a weak donor.21
A Schottky barrier could easily form22 between Mg and
TCNQ (the salt Mg2+(TCNQ-)2 is known, and Mg2+TCNQ2-
can also form); therefore, the molecular origin of asymmetries
seen in the measured currents of 3 was put into doubt.22
A new family of zwitterionic molecules was discovered by
Ashwell and co-workers, a quaternary picolinium or lepidinium
halide, when reacted with an alkali salt of TCNQ, can yield a
class of ground-state zwitterions, whose first example is
picolyltricyanoquinodimethane or picolinium tricyanoquin-
Pt|LB multilayer of 5|Mg|Ag
and even in the sandwich:
(5)
Pt|LB monolayer of 5|Mg|Ag
(23) Metzger, R. M.; Heimer, N. E.; Ashwell, G. J. Mol. Cryst. Liq.
Cryst. 1984, 107, 133-149.
(24) Ashwell, G. J.; Szablewski, M.; Kuczynski, A. P. In Lower-
Dimensional Systems and Molecular Electronics; Metzger, R. M., Day, P.,
Papavassiliou, G. C., Eds.; NATO ASI Series; Plenum: New York, 1991;
Vol. B248, p 647.
(25) Ashwell, G. J. In Organic Materials for Nonlinear Optics; Ashwell,
G. J., Bloor, D., Eds.; Royal Soc. of Chem.: Cambridge, 1993; pp 31-39.
(26) Akhtar, S.; Tanaka, J.; Metzger, R. M.; Ashwell, G. J. Mol. Cryst.
Liq. Cryst. 1986, 139, 353-364.
(27) Ashwell, G. J.; Sambles, J. R.; Martin, A. S.; Parker, W. G.;
Szablewski, M. J. Chem. Soc., Chem. Commun. 1990, 1374-1376.
(6)
(18) Metzger, R. M.; Panetta, C. A. New J. Chem. 1991, 15, 209-221.
(19) Metzger, R. M. In Molecular and Biomolecular Electronics; Birge,
R. R., Ed.; ACS Advances in Chemistry Series 240; American Chemical
Society: Washington, DC, 1994; pp 81-129.
(20) Metzger, R. M. Mater. Sci. Eng. 1995, C3, 277-285.
(21) Geddes, N. J.; Sambles, J. R.; Jarvis, D. J.; Parker, W. G.; Sandman,
D. J. Appl. Phys. Lett. 1990, 56, 1916-1918.
(22) Geddes, N. J.; Sambles, J. R.; Jarvis, D. J.; Parker, W. G.; Sandman,
D. J. J. Appl. Phys. 1992, 71, 756-768.

Hexadecylquinolinium Tricyanoquinodimethanide
J. Am. Chem. Soc., Vol. 119, No. 43, 1997 10457
To obtain the dipole moment of 5, the dielectric constants of dilute
solutions of 5 in dichloromethane were measured, as a function of
temperature, at 1 kHz, using a coaxial capacitance cell (three concentric
Pt cylinders, thickness 0.5 mm each, length 2.5 mm, diameters 8
(ground), 6, and 2 mm), a Hewlett-Packard 4263B LCR meter, and a
Cole-Palmer refrigerated temperature bath. Because 5 is insoluble in
nonpolar solvents, we used dichloromethane (gas-phase dipole moment
µ ) 1.60 D35 ) and low solute concentrations (<7.5 × 10-2 mg/mL);
results with acetonitrile (µ ) 3.92 D36 ) were irreproducible or
physically meaningless. From the Kirkwood-Fro¨hlich equation,37 an
effective dipole moment, g1/2µ, can be extracted from temperature-
dependent measurements:
Doubts raised about the Schottky barrier role of the Mg
electrode22 prompted the insertion of an insulating monolayer
of a fatty acid between the metal electrodes and 3 monolayers
of C16H33Q-3CNQ;28 I-V asymmetries were still seen and
ascribed to molecular rectification.28
The electrical rectification seen28 for 5, regardless of mech-
anism, has generated theoretical attention.29,30 One important
issue is whether the ground state is “zwitterionic”, as shown in
5, or “neutral”, as shown in 6. The theoretical studies29,30
suggested that in the ground state a mixture of neutral and
zwitterionic resonance forms may co-exist30 and allow for
intramolecular electron transfer in a monolayer in an adiabatic
regime in the sense of Marcus-Hush theory.31,32
gµ2 ) [9kT(ꢀ - n2)(2ꢀ + n2)]/[4πNꢀ(n2 + 2)2] ) (9kTΣ)/4πN)
(8)
kET ) νa{π2∆2[2h(πRTλ)1/2]-1} exp(-∆G*/RT). (7)
In eq 8, ꢀ ) dielectric constant, n ) refractive index, T ) absolute
temperature, k ) Boltzmann constant, N ) number of molecules of 5
per unit volume, Σ ) Onsager function (ꢀ - n2)(2ꢀ + n2)/ꢀ(n2 + 2)2,
and g ) Kirkwood correlation factor, discussed below. For n we used
the refractive index of the pure solvent (n ) 1.42). Measurements
were done in the range -10 to 30 °C. For pure dichloromethane, we
found µ ) 1.67 ( 0.83 D.
Here kET is the rate of electron transfer; νa is the frequency of
formation of the activated complex, whose free energy of
activation is ∆G*; λ is the total Franck-Condon reorganization
energy, i.e. the energy difference between the molecule before
electron transfer (with its solvent atmosphere) and the molecule
after the electron transfer (with its modified solvent atmosphere);
∆ is the electronic coupling energy between the vibrational
energy envelopes of the reactant and of the product at the point
of their avoided crossing; h is Planck’s constant; R is the gas
constant; and T is the absolute temperature. The factor in { }
is for “non-adiabatic” electron transfer; it becomes unity for an
“adiabatic” transfer.
For a single molecule (monolayer) of 5, ∆ is large and
rectification is explicable,30 while for a multilayer, ∆ is very
small and rectification may be due to defect conduction.30 Of
course, if the twist angle θ in 5 becomes 90°, then the molecule
must become a zwitterion.30 It was suggested that 4 and 5
belong to a class of twisted internal charge transfer (TICT)
systems.30 However, such twisting would make the electron
transfer process quite slow. The energy-minimized “gas-phase”
theoretical calculations give for 4 a smaller ground-state dipole
moment29 than calculated for the geometry of 4 in the crystal,23
unless efforts are made to account approximately for “solvent
effects”.30 Theory suggested that in the first excited state θ ≈
90°,30 while for the ground state (“gas phase”) θ ) 9°-11°.30
After two earlier preliminary studies of 5,33,34 we report here
a novel, very convenient and efficient synthesis of 5, confirm
its ground-state zwitterionic structure, confirm electrical recti-
fication by both LB monolayers and multilayers by macroscopic
and nanoscopic methods, and identify incontrovertibly the
direction of higher conductivity relative to the orientation of
the molecules in the multilayers. We assert that 5 is a
unimolecular rectifier.
Considering the molar polarization P ) 4πNAµ2/9kT ) ΣM/d, where
M is the molar mass, NA is Avogadro’s number, and d is the density,
another approach for calculating the dipole moment of the solute38 is
to assume that the molar polarization of the solution P12 is given by
P12 ) x1P1 + x2P2
(9)
where x is the mole fraction and the subscripts 1 ) solvent and 2 )
solute.
Theoretical calculations were performed using the CAChe program
package on a Macintosh PowerPC 8100AV microcomputer.
Pockels-Langmuir and Langmuir-Blodgett films were studied using
both an analog Lauda film balance and also a microcomputer-controlled
Nima Model 622D2 trough, both connected to a Lauda constant-
temperature bath (5-30 °C), in a room with HEPA-filtered air, and
“conductivity” water (Millipore Milli-Q, 18 MΩ cm). The Volta
(Kelvin) electrostatic potential difference at the air-film interface was
measured using a Monroe “Isoprobe” Model 162 electrostatic voltmeter,
interfaced to the NIMA microcomputer; a monolayer of arachidic acid
had ∆V ) 0.31 V. Substrates used were HOPG (Union Carbide Grade
ZYA) or a single-crystal Au(111) (Goodfellow) and Al evaporated on
glass.
Ellipsometric measurements were performed on a Rudolph Auto-
EL-III ellipsometer (λ ) 632.8 nm). X-ray diffraction measurements
were performed using a Rigaku DX-1B powder diffractometer and Cu
KR X-radiation. Fourier transform infrared (FTIR) spectra were
obtained on a Bruker IFS-88 spectrometer, using a grazing-angle
accessory (Specac; angle of incidence 85°) and a wire polarizer
(Specac). Scanning tunneling micrographs were obtained on Digital
Instruments Nanoscope II and Nanoscope IIIa scanned probe micro-
scopes, using a Pt/Ir tip and the type A head, the scanning tunneling
spectroscopy (STS) software, and the CITS software.
2. Experimental Methods
An Edwards EL306A evaporator, with a multiple-source rotating
stage, a liquid nitrogen-cooled cold finger, and a quartz film thickness
monitor, was used to make both the base electrode (Al on quartz) and
also the Al and/or Mg pads placed on top of the Langmuir-Blodgett
NMR spectra were measured using Bruker 360 or 500 MHz
spectrometers. UV spectra were measured on a Perkin-Elmer Sigma
3 spectrophotometer. Cyclic voltammograms were measured on a PAR
270 potentiostat, using a 10-4 M solution of 5 in CH2Cl2, a Pt wire
working electrode, an SCE reference electrode, 0.1 M NBu4ClO4
electrolyte, and a scan rate of 20 mV/s.
monolayers and multilayers. The base pressure was 10-6
Torr; the
distance from the evaporating source to substrate was 21 cm. After
deposition of the LB film atop the Al base electrode, the substrate was
dried at least 72 h in a desiccator in the presence of P2O5. Al, Mg, or
Ag was evaporated on top of the LB film. The electrodes placed on
(28) Martin, A. S.; Sambles, J. R.; Ashwell, G. J. Phys. ReV. Lett. 1993,
70, 218-221.
(29) Broo, A.; Zerner, M. C. Chem. Phys. 1996, 196, 407-422.
(30) Broo, A.; Zerner, M. C. Chem. Phys. 1996, 196, 423-426.
(31) Marcus, R. A. J. Chem. Phys. 1965, 43, 679-701.
(32) Hush, N. S. Trans. Faraday Soc. 1961, 57, 557-580.
(33) Wu, X.-L.; Shamsuzzoha, M.; Metzger, R. M.; Ashwell, G. J. Synth.
Met. 1993, 55-57, 3836-3841.
(35) Weast, R. C., Ed. Handbook of Chemistry and Physics, 70th ed.;
CRC Press: Boca Raton, FL, 1989; pp E-59-61.
(36) Ghosh, S. N.; Trambarulo, R.; Gordy, W. J. Chem. Phys. 1953, 21,
308-310.
(37) Bo¨ttcher, C. J. F.; Van Belle, O. C.; Bordewijk, P.; Rip, A. Theory
of Electric Polarization; Elsevier: Amsterdam, 1973; Vol. 1, pp 245-284.
(38) Davies, M. Some Electrical and Optical Aspects of Molecular
BehaViour; Pergamon: Oxford, 1965.
(34) Metzger, R. M.; Tachibana, H.; Wu, X.; Hopfner, U.; Chen, B.;
Lakshmikantham, M. V.; Cava, M. P. Synth. Met. 1997, 85, 1359-1360.

10458 J. Am. Chem. Soc., Vol. 119, No. 43, 1997
Metzger et al.
both the Al base and the Al pads were Au wires (diameter 0.05 mm)
wetted either by a Ga/In eutectic or by Ag paste (GC Controls; better
contacts were obtained by heating this paste to evaporate the butyl
acetate solvent and redissolving it in p-xylene).
The current-voltage profiles were obtained both in the PAR 270
potentiostat (shorting the reference electrode to the counter electrode)
and also in a Gateway 2000 microcomputer-controlled AC and DC
conductivity measuring system, using a Hewlett-Packard Model 3245A
universal source and an Hewlett-Packard Model 3457A multimeter.
3. Synthesis, NMR Spectrum, and Solution Visible-UV
Spectra
Several syntheses of 5 have appeared in the literature,39-41
all of which are erratic and tend to be unreliable as far as
reproducibility. We now report a reproducible synthesis of 5,
starting from N-cetyllepidinium tosylate and the lithium salt of
TCNQ radical anion, in dry DMSO solution in the presence of
pyridine. We believe that this transformation occurs via an
electron-transfer process. The lepidinium salt loses H• to the
Li+TCNQ•-, being converted to a radical cation salt, which then
couples with free Li+TCNQ•-, to give an intermediate; spon-
taneous loss of HCN leads to the product.
The NMR spectrum of the compound is very diagnostic and
corroborates previously cited IR and UV data indicative of a
dipolar ground state.40 The H-2 proton in 5 resonates at δ 9.39
ppm, giving a doublet with coupling constant JH-2H-3 of 5.80
Hz in DMSO-d6 solution. Upon protonation to 7, there is very
little change in the H-2 resonance, which occurs at δ 9.23
(JH-2H-3 ) 5.97 Hz), and a new singlet signal appears at δ
5.23 for the benzylic proton between the two cyano groups.
The H-2 resonances are analogous to the R protons in a
quinolinum salt. Furthermore, the H-2 proton in a 4-cyanovinyl
N-alkylquinoline methide 8 resonates at δ 7.90 ppm.42 The
visible-ultraviolet spectrum of 5 is solvent-dependent, particu-
larly the long-wavelength maximum in the visible region. In
particular, this maximum is at 720 nm in CH3CN, at 884 nm in
CH2Cl2, and at 838 nm in CHCl3. For the first two solvents
the absorbance is linear with concentration, while in CHCl3,
there is a deviation (see Figure 1), presumably due to some
intermolecular association at the higher concentrations (incipient
admicelle or micelle formation?).
Figure 1. Absorbance vs concentration plot for a solution of 5 in CH3-
CN (λmax ) 720 nm), in CH2Cl2 (λmax ) 884 nm), and in CHCl3 (λmax
) 838 nm).
in dry DMSO (4 mL) was added Li+TCNQ•- (0.42 g, 1.99
mmol). This was heated on a steam bath for 1 h and cooled,
and the blue crystalline solid (0.15 g, 28%) was filtered off.
The mother liquor was heated further on the steam bath for 20
h and cooled, and the blue crystalline solid (0.17 g, 31%) filtered
off. Total yield of product: 0.32 g, 59%. Recrystallization
from pyridine furnished pure 5 as blue microcrystals: mp >
1
290 °C; UV-vis (DMSO) broad peak at λmax ) 687 nm; H-
NMR (DMSO-d6) (330 K) δ 9.39 (d, 1H, J ) 5.80 Hz), 8.74
(d, 1H, J ) 8.50 Hz), 8.52 (d, 1H, J ) 8.96 Hz), 8.45 (d, 1H,
J ) 6.35 Hz), 8.36 (s, 1H), 8.24 (t, 1H, J ) 7.86 Hz), 8.01 (t,
1H, J ) 7.72 Hz), 7.72 (d, 2H, J ) 8.27 Hz), 6.90 (d, 2H, J )
8.59 Hz), 4.96 (t, 2H, J ) 7.16 Hz), 2.00 (t, 2H, J ) 7.40 Hz),
1.41-1.23 (m, 26H), 0.85 (t, 3H, J ) 6.76 Hz); 13C-NMR
(DMSO-d6) (330 K) δ 150, 147.34, 137.36, 134.84, 133, 129.21,
127.69, 127, 126.79, 122.90, 122.59, 121.34, 118.85, 118.18,
56.73, 30.91, 29.02, 28.63, 28.58, 28.48, 28.40, 28.27, 28.08,
25.50, 21.66; 1H-NMR (CDCl3/TFA) (protonated form 7) δ 9.23
(d, 1H, J ) 5.97 Hz), 8.45 (s, 1H), 8.37 (m, 2H), 8.30 (m, 2H),
8.07 (t, 1H, J ) 6.31 Hz), 8.01 (d, 2H, J ) 8.11Hz), 7.74 (d,
2H, J ) 8.06 Hz), 5.23 (s, 1H), 5.0 (t, 2H, J ) 5.45 Hz), 2.15
(m, 2H), 1.26 (m, 26H), 0.88 (t, 3H, J ) 6.70 Hz). Anal. Calcd
for C37H44N4: C, 81.58; H, 8.14; N, 10.28. Found: C, 81.32;
H, 8.19; N, 10.26. The blue microcrystals gave a superposition
of diffraction patterns from several crystallites, which could not
be resolved to provide an indexable diffractogram.
Synthetic Details. To a solution of N-cetyllepidinium
tosylate (0.54 g, 1.00 mmol) and pyridine (0.27 g, 3.41 mmol)
(39) Ashwell, G. J. Thin Solid Films 1990, 186, 155-165.
(40) Ashwell, G. J.; Dawney, E. J. C.; Kuczynski, A. P.; Szablewski,
M.; Sandy, I. M.; Bryce, M. R.; Grainger, A. M.; Hasan, M. J. Chem. Soc.,
Faraday Trans. 1990, 86, 1117-1121.
4. Cyclic Voltammogram
(41) Ashwell, G. J.; Jefferies, G.; Dawnay, E. J. C.; Kuczynski, A.;
Lynch, D. E. J.; Gongda, Y.; Bucknall, D. G. J. Mater. Chem. 1995, 5,
975-980.
A cyclic voltammogram has been measured for 5, and is
shown in Figure 2. There is a one-electron quasi-reversible
reduction (Ep,r ) -0.545 V, Ep,o ) -0.480 V, E1/2 ) -0.513
V vs SCE, ∆E ) 0.065 V), a second irreversible reduction at
(42) Hughes, T. V. Unpublished observation.
(43) Warren. B. L. X-Ray Diffraction; Addison-Wesley: Reading, MA,
1969; p 251.

Hexadecylquinolinium Tricyanoquinodimethanide
J. Am. Chem. Soc., Vol. 119, No. 43, 1997 10459
Figure 4. Pressure-area isotherms for 5 at 14 and 20 °C and surface
(Volta or Kelvin probe) potentials at 14 °C.
Figure 2. Cyclic voltammogram of a 10-4 M solution of 5 in CH2Cl2,
measured using a Pt wire working electrode, an SCE reference electrode,
0.1 M NBu4ClO4 electrolyte, N2 gas bubbled through the solution, and
a scan rate of 20 mV/s.
extrapolation of the results to zero concentration (Figure 3)
yields a dipole moment of 40-45 D (let us say 43 ( 8 D) for
5 in dichloromethane at infinite dilution.
This result strongly supports the zwitterionic nature of the
ground state of 5, as also established by NMR (see section 3).
Theoretical calculations have determined the ground-state dipole
moments of 4 (crystal geometry) to be µ ) 26.16 D;23 for 4
and 5, “gas-phase” calculations yield µ ) 8-11 D;29 for 5,
including “solvent effects”, µ ) 29.6 D.30 A factor of ca. 1.5
between theory and the present experiments seems reasonable,
considering the inherent inaccuracy of theoretical methods, and
the approximations used in analyzing the experimental data.
Interactions between 5 and the solvent molecules can increase
the real gas-phase dipole moment of 5. The decrease of the
effective dipole moment, when increasing the concentration,
indicates that molecules of 5 tend to align with anti-parallel
dipoles (g < 1).
6. Pockels-Langmuir Monolayers at the Air-Water
Figure 3. Effective dipole moments g1/2µ of compound 5 in dichlo-
romethane versus concentration, using the Kirkwood-Fro¨hlich equation,
eq 8. (0) Data obtained directly from the temperature dependence of
the dielectric constant of the solution (compound 5 + dichloromethane);
at zero concentration the Kirkwood correlation factor should become
g ) 1. (9) Data obtained by “subtracting” the contribution of the slightly
polar solvent, using eq 9. The solid lines are least-squares fits to the
data.
Interface
The pressure-area (∏-A) isotherm of 5, when measured in
sunlight or under fluorescent light, shows a collapse area Ac )
34 ( 1 Å2 molecule-1 at the collapse pressure ∏c ) 43.5 mN
m-1 and a limiting area Ao ) 125 ( 1 Å2 molecule-1 at zero
film pressure [39], but with additional intermediate monolayer
ordering transitions at ∏ ) 27 mN m-1 (A ) 58 Å2 molecule-1
)
and ∏ ) 30 mN m-1 (A ) 43 Å2 molecule-1), where the
Ep,r ) -1.16 V, and an irreversible oxidation at Ep,o ) 0.49 V
vs SCE. Impurities in very small concentrations show an
oxidation at 0.44 V and reductions at Ep,r ) -0.675 V and Ep,r
) 0.383 V vs SCE. Under identical conditions a solution of
TCNQ, 2, has a first reduction at E1/2 ) +0.214 V vs SCE
(TCNQ0 f TCNQ-) and a second reduction at E1/2 ) -0.384
V vs SCE (TCNQ- f TCNQ2-); a sample of hexadecyl
lepidinium bromide shows a reversible oxidation at E1/2 ) +0.70
V vs SCE.
molecules are inclined to the water surface.39,44
The ∏-A isotherm of 5, when measured under a green
safelight, shows very clearly a collapse point at 50 Å2
molecule-1 and 34 mN m-1 (Figure 4): this isotherm is
moderately temperature-dependent; the collapse point at 50 Å2
molecule-1 and 34 mN m-1 was not noticed in the first report.39
The Volta potential difference ∆V is set to zero before spreading
the monolayer, rises to about 0.18 V at the start of the
compression, and becomes about ∆V ) 0.46 V at about 120
Å2/molecule, i.e. before the monolayer is fully compressed: the
dipolar layer is more polar than, e.g., cadmium arachidate.
5. Dipole Moment
The effective dipole moments for various concentrations of
5 were calculated, using eq 8, from the slope of Σ versus 1/T,
and also from the partial molar polarization of the solute in the
same temperature range. Figure 3 shows that equivalent results
were obtained from both methods. The Kirkwood correlation
factor g takes into account all intermolecular interactions, which
lead to some mutual orientation of the molecular dipoles; a
simple expression37 is g ) 1 + z cosθij , where z is the average
number of neighboring molecules around a given dipole and
cosθij is the average of the cosine of the angle between two
dipoles; for infinitely dilute solutions, g tends to 1.37 Thus, the
7. Langmuir-Blodgett Films: X-ray and UV-Visible
Spectra of Multilayers
The LB transfer of monolayers and multilayers of 5 was
performed at 15 mm min-1 using a film pressure of 20 mN
m-1
. Molecule 5 first transfers on the downstroke onto
hydrophobic HOPG and on the upstroke onto a freshly cleaned
hydrophilic Au(111) single crystal and on an Al electrode. The
hydrophobic hexadecyl “tail” of D+ is thought to be the end of
(44) Wu, X. Ph.D. Dissertation, University of Alabama, 1995.

10460 J. Am. Chem. Soc., Vol. 119, No. 43, 1997
Metzger et al.
the molecule closest to HOPG, while the hydrophilic 3CNQ-
(A-) end should be closest to Au(111). At the lower transfer
speed of 15 mm min-1 the LB multilayers (past the first
monolayer) are Z-type, while at the faster transfer speed of 45
mm min-1, the LB multilayers become Y-type.
X-ray diffraction spectra, taken several days after film transfer
onto a glass substrate, shows that the Y-type films apparently
reorder to Z-type after transfer, with a thickness of 23 Å per
monolayer and a longitudinal crystallite correlation length43 of
200 Å (2θ ) 3.81°; Scherrer linewidth at half-height 0.4° in
2θ). Thus, for multilayers, the Z-type ordering is more stable.
LB multilayers of 5 transferred onto HOPG at 15 mm min-1
have orientation
HOPG|f r r r r r r r etc.
(10)
Figure 5. Visible spectrum of an 11-layer LB film of 5 transferred
onto quartz; the narrow absorbance has a peak at 570 nm (2.17 eV;
spectral width at half-maximum ) 0.19 eV); the absorbance per
monolayer at 633 nm is estimated as 0.0021.
Here the arrow f indicates the rough molecular orientation; it
points from the hexadecyl tail T and the quinolinium ring D+
(tail of arrow) toward the dicyanomethylene head of A- (head
of arrow). As shown in eq 10, the first monolayer adheres with
the hexadecyl tail T next to HOPG (X-type orientation), while
the other layers (second through 15th) have Z-type orientation
(tails away from HOPG). The transfer ratios were measured
to be about 50% for the first layer, and from 50% to 100% for
the Z-type layers.
The film transfer to Al is Z-type throughout
Al|r r r r r r r etc.
(11)
but the transfer ratio for the first layer is now over 90%.
A sharp and narrow absorption band peaked at 570 nm is
found for LB films transferred onto a quartz substrate at 14 °C
in the dark and measured at 25 °C (Figure 5). This narrow
band disappears if the films are protonated but returns when
exposed to ammonia vapor;41 it disappears irreversibly if the
film, or the solution, is exposed to bright light and air for some
time.
Figure 6. p-Polarized grazing-angle FTIR spectrum (θ ) 85°) of LB
monolayer of 5 on Al.
nm the molar absorptivity is not negligible (see Figure 4), so
the assumption κ ) 0 is not valid.
8. Theoretical Calculations
A geometry-optimized PM3 calculation was performed for
5 (heat of formation 6.02 eV). The molecule 5 is twisted; the
angle between the quinolinium ring and the central phenyl ring
of 3CNQ is ≈30°. The computed dipole moment for the ground
state singlet of 5 is 10.7 D, with the highest occupied molecular
orbital (HOMO) at -7.98 eV (MO #106, localized on quino-
linium, the π bridge, and the 3CNQ end) and the lowest
unoccupied molecular orbital (LUMO) at -2.42 eV (MO #107,
localized on 3CNQ-); however, the bond lengths in the molecule
are more quinonoid than benzenoid. When using a shorter alkyl
chain (octyl instead of hexadecyl) and AM1,44 the ground state
has dipole moment of 10.56 D and a heat of formation of 7.61
eV, while the excited singlet state has a smaller dipole moment,
6.36 D, and a heat of formation of 10.83 eV, i.e. 3.22 eV
higher.44 A recent PM3 calculation for the excited singlet state
of 5 yielded a heat of formation of 8.469 eV, a twisted geometry
(θ ≈ 90°), and a high dipole moment of 45.4 D. These results
agree with ref 30. The inferred approximate molecular size for
5 is 33 Å × 7 Å × 4.5 Å.
10. Infrared Spectra
Grazing-angle Fourier transform infrared spectra, using
p-polarized light for a monolayer of 5 on a glass substrate
covered by Al (Figure 6), shows absorbances at 2850 cm-1
(symmetric C-H), 2920 cm-1 (asymmetric -CH2-), 2139
cm-1 (-CtN), and 2175 cm-1 (-CtN), indicating that the
cyano groups and the methylene groups do not lie in the plane
of monolayer but are somewhat tilted from it. The CN stretches
show evidence of shifts due to partial charge: negatively
charged CtN groups (2175 cm-1) and a shifted CtN group
(2139 cm-1). For a monolayer of 5 on either Au(111) or HOPG,
the C-H and -CH2- signals can be seen clearly; a -CtN
signal at 2175 cm-1 is barely visible for the monolayer on Au-
(111).44
11. Macroscopic Conductivity of LB Films
An Edwards 306A evaporator and a PAR 270 potentiostat
were used to study macroscopic “metal|organic|metal” sand-
wiches. The glass or quartz or silicon substrates were covered
by 25 mm × 75 mm × 100 nm thick Al layer, then LB
monolayer or multilayer films were transferred under a green
safelight, then the sample was dried for 2 days in a vacuum
desiccator containing P2O5. Twelve cylindrical “pads” of Al
(four each of areas 2.8, 4.5, and 6.6 mm2, all either 100 or 300
nm thick, depending on the run) were finally deposited per
microscope slide, atop the organic layer, with the substrate
9. Ellipsometry
The thickness per monolayer, measured by ellipsometry (λ
) 670 nm) and by surface plasmon resonance, is 22 ( 2 Å;41
thus the approximate molecular axis may be tilted 48 ( 5° from
the normal to the monolayer plane. A larger apparent thickness
(26 ( 2 Å, n ) 1.50 ( 0.05, κ ≡ 0), measured by ellipsometry
(HeNe laser, λ ) 632.8 nm), is less reliable because at 632.8

Hexadecylquinolinium Tricyanoquinodimethanide
J. Am. Chem. Soc., Vol. 119, No. 43, 1997 10461
Figure 7. Conductivity of a sandwich “Ga/In eutectic|Al (100 nm)|4LB
monolayers of 5 (Z-type)|Al (100 nm)|Ga/In eutectic”. The scan rate
was 100 mV/s. The counter and reference electrodes (shorted together)
are connected to one Al electrode; the working electrode is connected
to the other Al electrode. The direction of easier electron flow (I > 0
for V > 0) is from the bottom electrode through the film to the top
electrode.
Figure 9. Conductivity of a sandwich “Ga/In eutectic|Al (75 nm)|1LB
monolayer of 5 (Z-type)|Al (150 nm)|Ga/In eutectic”. V > 0 means
electron flow from the bottom electrode through the LB monolayer to
the top electrode; the rectification ratio is low.
Of 39 “pads” checked (20 with areas of 2.8 mm2, 15 with
areas of 4.5 mm2, 4 with areas of 6.6 mm2) with two thicknesses
of the Al pad (12 with thickness 300 nm, 27 with thickness
100 nm), 17 were electrical short circuits, either because of
monolayer defects or because the Ga/In or Ag paste “made”
defects. The 22 “good” junctions had DC resistances from 1
to 400 MΩ; of these, only four exhibited rectifying behavior.
Figure 11a shows a typical rectification curve. Above a
threshold voltage Vt, the monolayer shows higher currents for
positive voltage than for the corresponding negative voltage.
This Vt varies from junction to junction in the range Vt ) 0.8-
1.3 V.
The rectification ratio RR, defined by
RR ) (current at Vo)/(current at -Vo)
(14)
Figure 8. Conductivity of a sandwich “Ga/In eutectic|Al (100 nm)|4LB
monolayers of 5 (Z-type)|Mg (110 nm)|Al (100 nm)|Ga/In eutectic”.
The counter and reference electrodes (shorted together) are connected
to one Al electrode; the working electrode is connected to the other Al
electrode. The direction of easier electron flow (I > 0 for V > 0) is
from the bottom electrode through the film to the top electrode.
where Vo is the highest positive bias used, ranges from 2.4 to
26.4 the first time the film is measured. As the cycle is repeated,
RR drops steadily and disappears after 4-6 cycles. Similarly,
by applying a constant positive or negative bias, we observed
a slow decrease, over 15-20 min, of the current for both
polarities. It appears that, under the intense electric fields, the
molecular dipoles reorient to minimize energy. For some of
the junctions, short circuits form across the monolayer during
the measurement.
cooled to 77 K, to form the sandwich
glass|Al|4 LB monolayers of 5|Al
(12)
as shown in Figure 7. Also, to duplicate the work of ref 28,
Mg was coated atop the organic film (110 nm), and then the
sample was covered by the top Al pad without breaking vacuum,
thus achieving the sandwich
In the range -0.5 to 0.5 V, the I-V characteristic is ohmic.
In the reverse regime (negative bias), the logarithm of the current
is linear with the applied bias (through-space tunneling) from
-0.5 V to the highest negative bias applied (-1.5 V or -1.8
V). In the forward regime, we again observed lne(I) V for V
> 0.5, up to a threshold voltage (in the range 0.8-1.3 V, sample
dependent), above which the current rises severalfold. This
glass|Al|4 LB monolayers of 5|Mg|Al
(13)
Contact to the Al electrodes was achieved using either a Ga/In
eutectic, or Ag paste. Note that the electrodes (Al) are
symmetrical in eq 12. Note also that, while drying multilayers
of arachidic acid before depositing the top electrode was
essential (spurious rectification is seen with wet multilayers),
this problem is not serious for 5: samples strongly dried in a
vacuum desiccator with P2O5 showed essentially the same I-V
characteristics as samples kept in the desiccator over CaSO4
for a longer time. In all cases, reversing the electrode
connections reversed the direction of rectification.
deviation from the lne(I)
V law may be ascribed to an
enhanced through-bond electron transfer. This behavior re-
sembles that observed by Martin et al. for multilayers of 5.28
The results on both monolayers and multilayers show that
the electrons preferentially flow by IVT from 3CNQ (A-) to
quinolinium (D+), in agreement with the AR ansatz. In the
case of multilayers, electron transfers between successive
monolayers from the D+ of layer N to the A- of layer N-1 is
possible (by tunneling through the alkyl chains), provided that
the electric field is high enough to move the electron between
the relevant molecular orbitals.
The work with multilayers gave equivalent results using the
potentiostat or the I-V measurement setup. The work with
monolayers was, predictably, much more difficult.

10462 J. Am. Chem. Soc., Vol. 119, No. 43, 1997
Metzger et al.
Figure 10. Orientation of the LB monolayer (a) or multilayer (b) on the glass, quartz, or Si substrate; the electrode (+) for positive bias, and the
direction of “easy” electron flow for V > 0 are marked. In some cases Ag paste was used instead of the Ga/In eutectic.
Figure 12. STM micrograph of Langmuir-Schaefer monolayer of 5
on HOPG, viewed with a Pt/Ir nanotip (Nanoscope II). Scan size ) 2
nm × 2 nm, nominal setpoint currrent ) 1.3 nA (actual ) 130 nA),
and bias ) 92.5 mV. Adapted with permission from ref 33. Copyright
1993 Elsevier Science.
monolayer of 5 deposited on HOPG an image with surprising
resolution (Figure 12) was seen: this seemed to show the
molecules, viewed with a Pt/Ir nanotip, seen from the dicya-
nomethylene end.33 The repeat area of 4.5 × 13 Å consists of
a “light” region (closer to the nanotip) of about 4.5 × 7 Å, plus
by a dark region (4.5 × 6 Å). A solution drop of 5 in dimethyl
sulfoxide also shows a repeat area of 4.5 × 13 Å.33,44 Using a
low-current STM (Nanoscope III) an LB monolayer of 5 showed
a similar image, albeit at lower resolution, with repeat distance
of 6 × 12 Å (Figure 13). As discussed below, these image are
consistent with the molecules viewed end-on.
Figure 11. Rectification through a single monolayer of 5 sandwiched
between Al electrodes (top Al pad area 4.5 mm2, thickness 100 nm),
using Ga/In eutectic and Au wires. The dc voltage is swept at a rate of
10 mV s-1. (a) Linear plot of the DC current I versus the dc applied
voltage V. (b) Plot of log10 I versus V.
13. Nanoscopic Conductivity of LB Films
An earlier effort by scanning tunneling microscopy of LB
monolayers of 5 on HOPG showed no reproducible asymmetries
in the scanning tunneling spectroscopy (STS) I-V plots.33
Multilayers of 5 on HOPG were studied using a Pt/Ir nanotip
at room temperature in air by STM and STS. Clear asymmetries
12. Nanoscopic Structure of LB Films
With a modified Nanoscope II (preamplifier modified to allow
for 100 times more current than usual) a Langmuir-Schaefer45
(46) Jin, C.; Haufler, R. E.; Hettich, R. L.; Basich, C. M.; Compton, R.
N.; Puretzky, A. A.; Demyanenko, A. V.; Tuinman, A. A. Science 1994,
263, 68.
(45) Langmuir, I.; Schaefer, V. J. J. Am. Chem. Soc. 1938, 60, 1351-
1360.

Hexadecylquinolinium Tricyanoquinodimethanide
J. Am. Chem. Soc., Vol. 119, No. 43, 1997 10463
good electron donor, A ) good electron acceptor, σ ) covalent
linkage between D and A) between two macroscopic metal
electrodes M1 and M2 with work functions φ1 and φ2, respec-
tively. Under “forward bias” an electron is transferred from
M2 to M1 in two steps:
step 1, ET
-
+
M1|D-σ-A|M2
8 M1 |D+-σ-A-|M2 (15)
step 2, IVT
-
+
-
+
M1 |D+-σ-A-|M2
8 M1 |D-σ-A|M2
(16)
In eq 15 one electron hops (elastic tunneling, ET) from the
HOMO of the neutral donor molecule, through a chemisorptive
barrier, onto the Fermi level of the metal electrode M1 whose
φ1 at the same energy as, i.e. in resonance with, the HOMO.
Thus one electron is added to the metal conduction band,
-
forming, formally, M1 and leaving the donor as the donor
radical cation D+. Similarly, an electron from the Fermi level
of the metal electrode M2 hops through a chemisorptive barrier
onto the LUMO of the neutral acceptor molecule, which is at
resonance with φ2. This forms the acceptor radical anion A-
Figure 13. STM image of a LB monolayer of 5 on HOPG, with Pt/Ir
tip (Nanoscope III). Scan size ) 4.5 nm × 4.5 nm, Z-range ) 2.3 pA,
bias ) -316 mV, and setpoint current ) 3.2 pA.
and removes one electron from the metal conduction band to
+
form M2
. This leaves the molecule as an excited-state
zwitterion D+-σ-A-.
In eq 16, an intervalence transfer (IVT) occurs: since the
LUMO of the neutral acceptor A (A ) TCNQ) is about 3 eV
higher than the HOMO of the neutral donor D (D ) TTF),
therefore down-hill inelastic tunneling ensues through the
molecule (through-bond tunneling), and the molecule returns
to its ground state D-σ-A, but one electron has been moved
by steps 1 and 2 from M2 to M1. It is guessed that the D+-
σ-A- state lies about 1-2 eV above the D-σ-A ground state.
Since good organic one-electron donors are terrible acceptors
and good electron acceptors are terrible donors, therefore, the
“reverse bias” process (step 3, eq 17; not shown in Figure 17)
would involve an excited state that is estimated to be g4 eV
higher than the excited state D+-σ-A- and can be thus
neglected.
Figure 14. STS I-V curve for an LB film of 15 Z-type monolayers
of 5 on HOPG, with Pt/Ir tip. The higher current for V < -1.35 V
corresponds to electron flow from HOPG through molecules to the Pt
tip.
in the I-V plot was seen in 15-layer films (Figure 14), but the
corresponding STM images were of a graphite superstructure.
In agreement with the I-V characteristics of multilayers
sandwiched between Al electrodes, the STS current is consistent
with electrons flowing by intramolecular intervalence electron
transfer (IVT) between LB layers 2-15, and with “jumping
over” layer 1 (Figure 15), but as discussed above, the electrons
must somehow jump between adjacent monolayers. The current
must somehow also jump over the first monolayer (which is
directed the “wrong way” and is afflicted with a poor transfer
ratio).
However, STM images (using a Pt/Ir nanotip, and setpoint
currents of 1 nA) of a saturated solution of 5 dissolved in
dimethyl sulfoxide (DMSO) on HOPG (not an LB film) show
an asymmetric STS spectrum (Figure 16): higher currents are
seen at negative bias (40 nA at -1.0 V) than at positive bias
(<2 nA at +1.0 V). An STS spectrum for a monolayer of 5
shows a slightly higher current at positive bias, in the expected
direction for the first (partial) monolayer on HOPG, adhering
with the alkyl tail onto the HOPG surface (Figure 17).
step 3, ET
N
+
-
M1|D-σ-A|M2
M1 |D--σ-A+|M2
(17)
(The only “hermaphroditic” donor that is an equally good
acceptor is graphite, for which no rectification would occur).
Since step 3, eq 17, is unlikely, but steps 1 and 2, eqs 15 and
16, are favored, molecular rectification should occur.
A
competing “reverse-bias” process is autoionization (step 4, eq
18), followed by charge migration to the two metal electrodes
(step 5, eq 19):
M1|D-σ-A|M2 step 48 M1|D+-σ-A-|M2
(18)
(19)
step 5, IVT, ET
+
-
M1|D+-σ-A-|M2
8 M1 |D-σ-A|M2
The AR process will work if step 1, eq 15, is much more
likely than step 4, eq 18, and if the through-molecule tunneling
in step 2, eq 16, is much larger than elastic through-space
tunneling. The situation is reviewed diagrammatically in the
top half of in Figure 18.
14. Review of the AR Ansatz
Before discussing whether an AR mechanism, or a modifica-
tion of it, is operative for C16H33Q-3CNQ, we should review
the essentials of the AR ansatz.17 The AR model of unimo-
lecular rectification places a single molecule D-σ-A (D )
15. The Case of a Ground-State Zwitterion
The AR ansatz can be adapted (Figure 18, bottom six frames)
to the ground-state zwitterion D+-π-A-, whose first excited

10464 J. Am. Chem. Soc., Vol. 119, No. 43, 1997
Metzger et al.
Figure 15. Direction of observed STS electron current through 15-layer LB multilayer of 5.
Figure 16. STS I-V curve for a DMSO solution drop of 5 on HOPG,
with Pt/Ir tip (Nanoscope II). The higher current for V < -1.35 V
corresponds to electron flow from HOPG through molecules to the Pt
tip.44
Figure 18. Depiction of electron transfers within D-σ-A molecules
and D+-π-A- zwitterions and to and from two metal electrodes,
according to the AR model for neutral D-σ-A molecules (top six
frames) and to the modified AR model for D+-π-A- zwitterions
(bottom six frames). Steps 1 and 2 are the preferred direction of electron
flow (right to left) for D-σ-A, and steps 6 and 7 are the preferred
direction of electron flow (right to left) for D+-π-A- zwitterions.
Figure 17. STS I-V curve for an LB monolayer of 5 on HOPG, with
Pt/Ir tip (Nanoscope III). The higher current for V > 0.5 V corresponds
to electron flow from Pt/Ir nanotip through molecules to HOPG.
16. Discussion
state is the neutral molecule D0-π-A0. Under “forward bias”
We have vastly improved on the synthesis of 5 by using a
2:1 molar ratio of TCNQ to quinolinium salt: this ensures the
activation of the quinolinium ring needed for coupling. The
ground state of the molecule is clearly zwitterionic, as demon-
strated by its NMR spectrum and its measured solution dipole
moment µ ) 43 ( 8 D ) (1.4 ( 0.3) × 10-24 C m; this value
corresponds to an electron-hole pair separated by 9 ( 1 Å, a
large separation indeed: the estimated interatomic distance from
the bridgehead C atom in the dicyanomethylene end of 5 to the
quinolinium N atom is 10.5 Å. The monolayer thickness (23
Å by X-ray diffraction, 22 ( 2 Å by ellipsometry and surface
plasmon resonance) suggest molecules inclined by 45-55° from
the surface normal. The oxidation of 5 in dichloromethane
solution is irreversible; its first reduction is reversible, but 5 is
not a powerful electron acceptor. There is no electrochemical
activity between 0.4 and -0.4 V vs SCE in CH2Cl2 solution.
The distance between the two N atoms of the dicyanometh-
ylene end of 423 is 4.385 Å; adding 2N van der Waals radii
an electron is transferred from M2 to M1 in two steps
M1|D+-π-A-|M2 step 6, IVT8 M1|D0-π-A0|M2
(20)
(21)
step 7, ET
-
+
-
M1|D0-π-A0|M2
8 M1 |D+-π-A-|M2
The reverse bias process is
step 8, ET
+
M1|D+-π-A-|M2
8 M1 |D0-π-A0|M2
(22)
(23)
step 9, IVT
+
-
+
-
M1 |D0-π-A0|M2
8 M1 |D+-π-A-|M2
The system would rectify (steps 6 and 7, eqs 20 and 21), so
that electrons are moved from M2 to M1, provided that step 8,
eq 22, is less likely than step 6, eq 20.

Hexadecylquinolinium Tricyanoquinodimethanide
J. Am. Chem. Soc., Vol. 119, No. 43, 1997 10465
Table 1. Estimates of the Energy Level ELUMO for 5 (Relative to
the Vacuum Level, Using the Work Functions of M2) That Must Be
Populated before Significant Through-Molecule Tunneling Can
Occur (Eqs 20 and 21)a
LB layers
M2
φ2 (eV) M1 (V) |∆V| (eV) ELUMO
Figure 7
Figure 8
Figure 9
Figure 11a
Figure 14
4
4
1
1
15
Al
Mg
Al
Al
HOPG
4.2
3.7
4.2
4.2
4.4
Al
Al
Al
Al
Pt
1.0
1.0
1.5
1.3
1.2
3.2
2.7
2.8
2.9
3.2
a The average is ELUMO ) 3.0 ( 0.3 eV.
We have confirmed that 5 is not only a multilayer rectifier
but also a monolayer rectifier, by an improvement of Sambles
and co-workers’ original technique27,28 and in accord with a
plausible modification of the Aviram-Ratner ansatz.17 We have
eliminated the uncertainty due to the previous use27,28 of two
dissimilar metal electrodes with different work functions (Mg
and Al).
The work functions φ of the relevant metals are 5.7 eV for
Pt, 4.4 eV for graphite, 4.2 eV for Al, and 3.7 eV for Mg. We
have several measures for the bias ∆V required for the onset of
asymmetric conduction through LB monolayers and multilayers
of 5:
(a) through 4 monolayers sandwiched between Al and Al:
∆V ≈ 1 V (Figure 7);
(b) through 4 monolayers between Mg and Al: ∆V ≈ 1 V
(Figure 8);
Figure 19. Approximate work functions of electrodes, electron affinity
of TCNQ, and of HOMO and LUMO levels (experiment and theory)
of molecule 5, all relative to the vacuum level.
(c) through 1 monolayer between Al and Al: ∆V ≈ 1.5 V
(Figure 9);
(d) through 1 monolayer between Al and Al: ∆V ≈ 1.3 V
yields a calculated width of 7.4 Å for the dicyanomethylene
end of 5. Adding 5 Å for the part of the quinolinium ring not
eclipsed by 3CNQ yields a total width of 12.4 Å, not far from
the STM width of 13 Å (Figures 12 and 13). The van der Waals
thickness of an aromatic ring (about 3.5 Å) must be increased
by the presumed noncoplanarity of the quinolinium ring and
the 3CNQ ring, so a thickness of about 4.5 Å is reasonable.
Thus, STM repeat areas of 4.5 × 13 Å (Figure 12) or 6 × 12
Å (Figure 13) are roughly consistent with the calculated
molecular cross-section and with the observed collapse area of
50 Å2 at 20 °C (Figure 4).
(Figure11a);
(e) through 15 monolayers between HOPG and Pt: ∆V ≈
-1.2 V (Figure 14).
Using the metal work functions as references, these data yield
several estimates of the excited state molecular energy level
ELUMO that must be reached before the onset of down-hill
intramolecular tunneling (Figure 19 and Table 1); the average
is ELUMO ) 3.0 ( 0.3 eV.
The experimental electron affinity of TCNQ, 2, is 3.3 ( 0.3
eV:46 the through-film conductivity data (Figures 7, 8, 9, 11a
and 14) suggest that molecule 5 is a poorer acceptor than TCNQ
by 0.3 eV in the gas phase; this agrees approximately with the
CV data (Figure 2), which show that 5 is more difficult to reduce
than TCNQ by 0.7 V in CH2Cl2 solution. The optical band
centered at 570 nm (Figure 5) provides an estimate of ∆E )
ELUMO - EHOMO ) 2.17 eV for the energy gap; this suggests
EHOMO ) 5.2 eV below the vacuum level.
The rectification by a monolayer of 5 on HOPG (Figure 17),
with higher electron flow at positive bias from nanotip to HOPG,
is seen often and is consistent with IVT across a molecule
oriented with the dicyanomethylene end closer to the Pt/Ir
nanotip (cf. LB layer 1 in Figure 15), but the monolayer adheres
poorly to HOPG and is easily moved across the surface by the
tip, so other, symmetric STS curves (through-space tunneling)
are also seen.
Published PM3 semiempirical calculations for 5,30 and the
ones presented here, give lower ground-state dipole moments
than experiment: the geometry-optimized molecule is less polar,
and its bond lengths are quinonoid rather than benzenoid. A
very twisted zwitterion with high dipole moment is obtained
for the PM3 excited singlet state of 5. For a shorter alkyl chain,
AM1 calculations yield a lower dipole moment for the excited
singlet than for the ground state.44 High calculated dipole
moments can be obtained when the Hamiltonian is modified to
model the presence of a solvent, both for the ground state30
and also for the excited singlet state. A large change in dipole
moments between the ground and the optically excited state is
expected from the large second-order optical nonlinearity of 5,
but is not found easily in the theoretical calculations.
The theoretical (PM3) estimates EHOMO ) -7.98 eV and
ELUMO ) -2.42 eV yield ∆E ) 5.56 eV, a very high result,
affected by the usual errors in using Koopmans’ theorem. Using
the differences in the AM1-computed enthalpies of formation
The STS rectification across an LB multilayer (Figure 14) is
very reproducible; the direction of higher electron flow at V <
-1.35 V corresponds to IVT electron flow within layers 2-15
of a 15-layer film, as explained in Figure 15 (boxed IVT), but
the contrary flow across layer 1 (IVT) and the electron flow
mechanism between layers (IMT) are not explained.
The STS rectification by a solution drop (Figure 16) and its
STM structure at the HOPG surface suggest that the molecules
of 5 orient on the HOPG surface and show the same electrical
behaviour as an LB multilayer of 5. Monolayer rectification
between Al electrodes (Figure 11) can be seen reproducibly in
10% of the pads examined (one must find regions that are defect-
free), but repeated cycling decreases the rectification ratio, as
the molecules must reorient in the intense electrical fields.
Temperature-dependent studies are in progress.

10466 J. Am. Chem. Soc., Vol. 119, No. 43, 1997
Metzger et al.
for the ground-state singlet and the excited singlet (of a
modification of 5 with an octyl chain instead of a hexadecyl
chain), we estimate ∆E ) 3.21 eV, which is closer to
experiment. Thus, for an isolated molecule 5, the D-π-A state
may lie about 3.2 eV above the D+-π-A- state.
We have improved on the initial claim of rectification by
monolayers27 and multilayers of 528 by showing that even a
single monolayer of 5, can rectify when sandwiched between
two electrodes of the same metal (Al, plus its inevitable covering
of oxide).
We have detected asymmetric electron flow through a single
monolayer, and demonstrated that electrical rectification involves
the molecular energy levels of 5, as expected in a suitable
modification of the Aviram-Ratner mechanism.
Acknowledgment. We are grateful to Prof. J. Roy Sambles
(U. Exeter) for many discussions, to the National Science
Foundation (Grant DMR-94-20699), to the Department of
Energy (Grant DE-FC02-91-ER-75678), to the Department of
Energy for a Predoctoral Traineeship (X.W.), and to the
Scientific Affairs Division of the North Atlantic Treaty Orga-
nization (Grant CRG 960656) for their support.
Note Added in Proof: Two articles have been brought to
our attention: resonant tunneling through a porphyrin (Tao, N.
J. Phys. ReV. Lett. 1996, 76, 4066-4069) and asymmetric
tunneling through halogen adatoms on Ag (Schott, J. H.; White,
H. S. J. Phys. Chem. 1994, 98, 297-302). Also, since a 4.5
mm2 Al pad, covering a monolayer of 5 (area ≈ 60 Å2
molecule-1), generates a current of 0.0004 mA at ≈ 1.5 V
(Figure 11a), therefore, one can estimate the electron transfer
rate as kET ) 4.0 × 10-7 C pad-1 s-1/(1.6 × 10-19 C electron-1
× 4.5 × 10-2 cm2 pad-1/6.0 × 10-15 cm2 molecule-1) ) 0.33
17. Conclusions
We have proved that a monolayer of molecule 5 can rectify
by intramolecular tunneling, and that monolayers and multilayers
rectify both as macroscopic films (Figures 7, 8, 9 and 11) and
on a nanoscopic level (Figures 14 and 17).
electrons molecule-1 s-1
JA971811E
.