Molecular tunnel junctions based on π-conjugated oligoacene thiols and dithiols between Ag, Au, and Pt contacts: Effect of surface linking group and metal work function

Article abstract of DOI:10.1021/ja207751w

The tunneling resistance and electronic structure of metal-molecule-metal junctions based on oligoacene (benzene, naphthalene, anthracene, and tetracene) thiol and dithiol molecules were measured and correlated using conducting probe atomic force microscopy (CP-AFM) in conjunction with ultraviolet photoelectron spectroscopy (UPS). Nanoscopic tunnel junctions (～10 nm²) were formed by contacting oligoacene self-assembled monolayers (SAMs) on flat Ag, Au, or Pt substrates with metalized AFM tips (Ag, Au, or Pt). The low bias (<0.2 V) junction resistance (R) increased exponentially with molecular length (s), i.e., R = R₀ exp(βs), where R₀ is the contact resistance and β is the tunneling attenuation factor. The R₀ values for oligoacene dithiols were 2 orders of magnitude less than those of oligoacene thiols. Likewise, the β value was 0.5 per ring (0.2 A^-1) for the dithiol series and 1.0 per ring (0.5 A^-1) for the monothiol series, demonstrating that β is not simply a characteristic of the molecular backbone but is strongly affected by the number of chemical (metal-S) contacts. R₀ decreased strongly as the contact work function (Φ) increased for both monothiol and dithiol junctions, whereas β was independent of Φ within error. This divergent behavior was explained in terms of the metal-S bond dipoles and the electronic structure of the junction; namely, β is independent of contact type because of weak Fermi level pinning (UPS revealed E_F - E_HOMO varied only weakly with Φ), but R₀ varies strongly with contact type because of the strong metal-S bond dipoles that are responsible for the Fermi level pinning. A previously published triple barrier model for molecular junctions was invoked to rationalize these results in which R₀ is determined by the contact barriers, which are proportional to the size of the interfacial bond dipoles, and β is determined by the bridge barrier, E _F - E_HOMO. Current-voltage (I-V) characteristics obtained over a larger voltage range 0-1 V revealed a characteristic transition voltage V_trans at which the current increased more sharply with voltage. V_trans values were generally >0.5 V and were well correlated with the bridge barrier E_F - E_HOMO. Overall, the combination of electronic structure determination by UPS with length- and work function-dependent transport measurements provides a remarkably comprehensive picture of tunneling transport in molecular junctions based on oligoacenes.

Full text of DOI:10.1021/ja207751w

ARTICLE
pubs.acs.org/JACS
Molecular Tunnel Junctions Based on π-Conjugated Oligoacene Thiols
and Dithiols between Ag, Au, and Pt Contacts: Effect of Surface Linking
Group and Metal Work Function
BongSoo Kim,†,‡ Seong Ho Choi,† X.-Y. Zhu,§ and C. Daniel Frisbie*,†
†Department of Chemistry and Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis,
Minnesota 55455, United States
‡Solar Cell Research Center, National Agenda Research Division, Korea Institute of Science and Technology (KIST), Seoul 136-791,
Republic of Korea
§Department of Chemistry and Biochemistry, The University of Texas at Austin, 1 University Station, A5300 Austin, Texas 78712-0165,
United States
S Supporting Information
b
ABSTRACT: The tunneling resistance and electronic structure of metalꢀ
moleculeꢀmetal junctions based on oligoacene (benzene, naphthalene, anthra-
cene, and tetracene) thiol and dithiol molecules were measured and correlated
using conducting probe atomic force microscopy (CP-AFM) in conjunction
with ultraviolet photoelectron spectroscopy (UPS). Nanoscopic tunnel junc-
tions (∼10 nm2) were formed by contacting oligoacene self-assembled mono-
layers (SAMs) on flat Ag, Au, or Pt substrates with metalized AFM tips (Ag, Au,
or Pt). The low bias (<0.2 V) junction resistance (R) increased exponentially
with molecular length (s), i.e., R = R0 exp(βs), where R0 is the contact resistance
and β is the tunneling attenuation factor. The R0 values for oligoacene dithiols
were 2 orders of magnitude less than those of oligoacene thiols. Likewise, the β value was 0.5 per ring (0.2 Åꢀ1) for the dithiol series
and 1.0 per ring (0.5 Åꢀ1) for the monothiol series, demonstrating that β is not simply a characteristic of the molecular backbone but
is strongly affected by the number of chemical (metalꢀS) contacts. R0 decreased strongly as the contact work function (Φ)
increased for both monothiol and dithiol junctions, whereas β was independent of Φ within error. This divergent behavior was
explained in terms of the metalꢀS bond dipoles and the electronic structure of the junction; namely, β is independent of contact
type because of weak Fermi level pinning (UPS revealed EF ꢀ EHOMO varied only weakly with Φ), but R0 varies strongly with
contact type because of the strong metalꢀS bond dipoles that are responsible for the Fermi level pinning. A previously published
triple barrier model for molecular junctions was invoked to rationalize these results in which R0 is determined by the contact barriers,
which are proportional to the size of the interfacial bond dipoles, and β is determined by the bridge barrier, EF ꢀ EHOMO. Currentꢀvoltage
(IꢀV) characteristics obtained over a larger voltage range 0ꢀ1 V revealed a characteristic transition voltage Vtrans at which the current
increased more sharply with voltage. Vtrans values were generally >0.5 V and were well correlated with the bridge barrier EF ꢀ EHOMO
.
Overall, the combination of electronic structure determination by UPS with length- and work function-dependent transport measurements
provides a remarkably comprehensive picture of tunneling transport in molecular junctions based on oligoacenes.
’ INTRODUCTION
examination of contact effects are required in which the nature
of the contact is changed, e.g., by switching the type of linking
groups, or by measuring the dependence of the IꢀV character-
istics on molecular length. The length dependence approach that
we7,8,13,14 and others10,15ꢀ18 have emphasized is preferable
because it can be quantitative: from plots of differential resistance
(at a given voltage) versus molecular length the effective contact
resistance can be extrapolated. For example, in the nonresonant
tunneling regime the junction resistance follows,
Understanding the role of metalꢀmolecule contacts in molec-
ular electronics experiments remains a major challenge.1ꢀ12 In a
typical molecular junction, a molecule or an ensemble of molecules
is contacted at either end by metal electrodes, Figure 1. The goal
is to understand the currentꢀvoltage (IꢀV) characteristics in
terms of the molecular structure and the type of chemical linking
groups (X) employed to connect the molecules to the electrodes.
However, in these two-terminal measurements one does not have
independent knowledge of the potential profile across the junction,
and thus the relative contribution of the contacts to the junction
IꢀV characteristic (or total resistance) is generally unknown.
Because the role of the contacts cannot be discerned by IꢀV
measurements on any single molecular junction, systematic
R ¼ R0 expðβsÞ
ð1Þ
Received: August 16, 2011
Published: October 21, 2011
r
2011 American Chemical Society
19864
dx.doi.org/10.1021/ja207751w J. Am. Chem. Soc. 2011, 133, 19864–19877
|

Journal of the American Chemical Society
ARTICLE
where R0 is the contact resistance, s is the molecular length (or
barrier thickness), and β is the tunneling attenuation factor that
depends on the average barrier height. Plots of log R versus s
yield estimates of β from the slope and R0 from the zero length
intercept. Thus, length-dependent transport measurements de-
couple the contacts from the “bulk” transport through the
molecule, and comparison of R0 values for different junctions
clearly indicates the relative roles of the contact resistance.19
We have shown previously with conducting probe atomic
force microscopy (CP-AFM) that length-dependent measure-
ments can determine contact resistances in alkane thiol and
dithiol junctions as a function of the metal work function.7,13
These measurements revealed a very strong dependence of R0 on
metal type for alkane monothiols; the contact resistance varied
by 4 orders of magnitude as the work function changed from
4.3 (Ag contacts) to 5.7 eV (Pt contacts). β was found to be
independent of the work function. In addition, these measure-
ments confirmed prior reports2 that chemical contacts (AuꢀS
bonds) have significantly lower resistance than physical contacts
(e.g., Au/CH3). The CP-AFM technique is ideal for such
experiments because junctions are readily formed by soft contact
of the tip to well-characterized self-assembled monolayers
(SAMs), and the contact work function can be changed easily
by using tips and substrates coated with different metals.
molecules are the longest oligoacenes to be probed in a molecular
junction. In general, the oligoacenes are a good model system
for molecular electronics because they are rigid, planar, and aromatic.
We initiated these studies with a systematic investigation of
the electronic structure of oligoacene SAMs on Ag, Au, and Pt by
ultraviolet photoelectron spectroscopy (UPS). Very few molec-
ular junction transport experiments have been correlated with
experimentally determined electronic structure, and the possibi-
lity of such a correlation with junctions based on SAMs is an
advantage for developing a comprehensive picture of the junc-
tion behavior.7,23,24 We find by UPS that the Fermi level EF, or
more specifically the Fermi level to HOMO level offset (EF ꢀ
EHOMO), is weakly pinned in these oligoacene systems, i.e., it
does not depend strongly on metal work function Φ (Ag to Au
to Pt) because of the presence of a large bond dipole associated
with the metalꢀS contact. The origin of the bond dipole is
electron donation from S to the metal, and it increases with
increasing Φ, as expected, which then mitigates the variation of
EF ꢀ EHOMO with Φ. Further, the bond dipole results in a change
in the metal work function ΔΦ that clearly impacts the junction
transport characteristics. In particular, there is excellent correla-
tion between ΔΦ and R0 for Ag, Au, and Pt contacts to
oligoacene thiol SAMs. We also find that the other key tunneling
parameter, β, is essentially independent of Φ and ΔΦ within
error. The divergent dependences of β and R0 on the metal type
can be explained self-consistently with a multibarrier model for
the junction based on prior theoretical work by Heimel et al.25,26
In this model, contact barriers are present at the interfaces
between the molecules and the metals due to the metalꢀS bond
dipoles. The two contact barriers are proportional to ΔΦ, and we
propose that R0 reflects these barriers. The bridge barrier, which
is associated with the oligoacene backbone and impacts β, is
determined by EF ꢀ EHOMO and is only weakly sensitive to the
work function of the metals because of the pinning effect.
As expected based on earlier results,2,7,9 we also find that
the resistance of metalꢀoligoaceneꢀmetal junctions depends
strongly on whether monothiol or dithiol molecules are em-
ployed, i.e., junctions with two “chemical contacts” have drama-
tically lowered junction resistances compared with junctions with
one chemical contact and one physical contact (this is indepen-
dent of the specific metal). Surprisingly, however, the lower
junction resistance for dithiols reflects differences in both R0 and
β, a point that is only clear when the length dependence of
transport is systematically examined. That is, β (not just R0) is
lower for oligoacene dithiol junctions than for oligoacene
monothiol junctions, highlighting the key role that contacts play
Here, we expand these earlier experiments to rigid, π-con-
jugated, thiol- and dithiol-capped oligoacenes ranging from one
to four aromatic rings (0.8ꢀ1.5 nm) in length, Figure 2. Such
π-conjugated oligoacene systems have been examined previ-
ously8,20ꢀ22 but not as a systematic function of length up to four
acene rings as reported here. The tetracene mono- and dithiol
Figure 1. Schematic representation of charge transport in me-
talꢀmoleculeꢀmetal junctions. Charge carriers (electrons or holes)
tunnel through a barrier (ϕ) with width equal to the molecular length.
X represents a surface linking group usedto bind the molecule to the metal.
Figure 2. Schematic representation of a CP-AFM junction on the left. A metal-coated (Ag, Au, or Pt) AFM tip is brought into contact with a SAM of
π-conjugated oligoacene thiols or dithiols of various lengths on a metal substrate at a load of 1 nN. Voltage is swept at the tip, and currentꢀvoltage
characteristics are recorded. The molecular structures are shown on the right.
19865
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
in junction transport properties. We propose that in the case of
dithiol junctions, the HOMO level is weakly pinned closer to EF
than in the case of monothiol junctions.
spectroscopic ellipsometer (J. A. Woolam Co., Inc.). Measurements of
the polarization angles (Ψ and Δ) were taken as a function of
wavelength (λ) between 600 and 1000 nm at an incident angle of 65°.
The indices of refraction (n(λ)) and extinction coefficients (k(λ)) of the
metal-coated substrates were determined by measurement of the
polarization angles prior to monolayer deposition. The instrument
software converted these values to n(λ) and k(λ) of metal films and
saved them as a material file. After monolayer formation on metal substrates,
the polarization angles were measured again and the film thicknesses were
determined by a built-in algorithm. n(λ) and k(λ) of the SAMs were
assumed to be 1.55 and 0, respectively, over the wavelength range.
We acquired XPS spectra on a Perkin-Elmer Phi 5400 spectrometer.
The Mg Kα X-ray (1253.6 eV) anode was operated at 200 W. Photo-
electron detection was accomplished with a hemispherical analyzer set
at a pass energy of 89.45 eV for survey scans and 17.9 eV for high-
resolution scans. Each sample was transferred into the ultrahigh vacuum
chamber immediately after drying under a stream of nitrogen to mini-
mize contamination or oxidation. To obtain monolayer thicknesses,
we carried out XPS measurements on SAMs of alkanethiols (CH3-
(CH2)nSH, n = 7, 9, 11, 13, 15) on each metal and obtained calibration
curves for the relative peak areas of the metal (Ag 3d5/2, Au 4f7/2, or
Pt 4f7/2) and the C 1s peaks, using known thicknesses of these SAMs
from the literature.34,35 These calibration curves were used to obtain the
thicknesses of the oligoacene thiol SAMs. Additional description of the
characterization of the monolayers is provided in Figures S1 and S2 and
Tables S1 and S2 in Supporting Information.
Determination of Molecular Energy Levels. To probe the
electronic structure of the SAMs on metals (Ag, Au, and Pt), we acquired
UPS spectra on a Perkin-Elmer Phi-5400 spectrometer equipped with a
He I (hv = 21.2 eV) radiation source at an incident angle of 50° from the
sample normal. Photoemitted electrons were collected at normal emis-
sion with a pass energy of 4.45 eV. All spectra were acquired at an applied
bias of ꢀ7 V on the sample, and the energy scale was referenced to the
metal Fermi level (EF). The intensities of the raw spectra were normal-
ized at EF. The onsets of the highest occupied molecular orbitals were
determined from the UPS spectra. After magnifying the first increasing
peak near the metal Fermi level, we found the onset point from the cross-
point between two trend lines (one is placed on the baseline and the
other on the slope of the first peak, see Supporting Information). Optical
gaps were determined from the onset of UVꢀvisible absorption spectra
using a procedure similar to the determinations of UPS onsets. The
optical gap corresponds to the lowest gap between the lowest occupied
molecular orbital band and the highest occupied molecular orbitals.
UVꢀvisible absorption spectra were obtained on a Hewlett-Packard 8453
UVꢀvis spectrophotometer from chloroform solutions of PhSH, NaphSH,
AnthSH, and acetyl protected molecules (in place of TetSH and dithiols) at
a concentration of ∼10ꢀ5 M (the maximum absorbance was ∼1).
Preparation of Metal-Coated AFM Tips and Flat Metal
Substrates. For Ag and Au tips, contact mode AFM tips were coated
with 250 Å Ag or Au on top of a 30 Å Cr adhesion layer using a metal
evaporator inside a glovebox and transferred without exposing to air to
another glovebox where electrical measurements were carried out. For
Pt tips, contact mode AFM tips were coated with an 80 Å Pt film on top
of a 30 Å Ti adhesion layer using an e-beam evaporator, and then immedi-
ately transferred to the measurement glovebox with ∼10 min air
exposure. Template-stripped flat metal substrates were used to grow
monolayers for reproducible electrical measurements.7,8 Flat metal
substrates were prepared by the following steps. For flat Ag or Au
substrates, 5000 Å of Ag or Au was first deposited onto clean Si wafers in
an e-beam evaporator. We then glued Si chips (0.5 cm ꢁ0.5 cm) onto
the metal surface using epoxy (EPO-TEK 377, Epoxy Technologies,
MA). The epoxy layer was cured by placing the wafers in an oven at
120 °C for 1 h. The flat metal substrates were peeled off from the silicon
surface and immersed immediately in the thiol or dithiol solution for
Collectively, these results provide a comprehensive picture of
tunneling transport in oligoacene junctions in which the roles of
the surface linkers and metal work functions on both R0 and β are
clearly delineated. The tunneling characteristics of conjugated
oligoacenes are profoundly affected by the contacts, confirming
theoretical predictions24,25 and reinforcing the conclusion that
interpretation of junction IꢀV characteristics requires careful
consideration of contact properties. We emphasize that both
photoelectron spectroscopy and length-dependent transport mea-
surements are critical for the overall analysis of contact effects, and
we believe that, in broad terms, the results we present here are
likely to be similar for junctions based on other short conjugated
molecules.24,27ꢀ31
’ EXPERIMENTAL SECTION
Materials. Synthesis of oligoacene thiols and dithiols in Figure 2
except benzenethiol (PhSH) and naphthalene-2-thiol (NaphSH) are
provided in Supporting Information. PhSH, NaphSH, and the other
chemicals used for the synthesis were purchased from Aldrich. Tetra-
hydrofuran (THF) and triethylamine (TEA) were dried following
procedures in the literature.32 Gold nuggets (99.999% pure) were
purchased from Mowrey, Inc. (St. Paul, MN). Evaporation boats and
chromium evaporation rods were purchased from R. D. Mathis (Long
Beach, CA). Platinum and titanium for e-beam evaporation were
purchased from Kamis, Inc. (Mahopac Falls, NY). Silicon (100) wafers
were obtained from WaferNet (San Jose, CA). Contact mode AFM tips
(DNP silicon nitride probes) were obtained from Veeco Instruments
(Camarillo, CA). Absolute ethanol was purchased from Fisher Scientific.
Monolayer Growth and Characterization. The Ag or Au
substrates were 1000 Å thick thin films on silicon (with a 50 Å Cr
adhesion layer) prepared in a Balzers thermal evaporator at a rate of
1.0 Å/s at a base pressure of e2 ꢁ 10ꢀ6 Torr. The Pt substrates were
1000 Å films on silicon prepared in an e-beam evaporator (SEC 600)
with a 50 Å Ti adhesion layer. The metal surfaces were immersed in
10 mL of 0.01ꢀ0.1 mM solutions of the thiol molecule in argon-purged
absolute ethanol. SAMs of PhSH, NaphSH, and AnthSH were grown for
18ꢀ24 h. For the preparation of SAMs of tetracene thiol, we added
30 μL of triethylamine (TEA) to the tetracene-2-thioacetate (TetSAc)
solution for deprotection of acetyl group and SAM growth on Ag
substrates (36 h) or 10 μL of concentrated aqueous ammonia solution
for SAM growth on Au and Pt substrates (24 h). Note that for some
reason (presumably partial oxidation of the Ag surfaces that were briefly
air-exposed before immersion in ethanol), quality SAM formation on Ag
substrates required longer immersion times. When the SAM growth
conditions for Au and Pt were applied, the SAM thickness on Ag
substrates was lower than expected. Therefore, we added more triethy-
lamine deprotecting agent and immersed Ag substrates for longer times.
In the case of dithiol molecules, 30 μL of TEA was added to the
solution of the acetyl-capped molecules for deprotection and SAM
growth (18ꢀ24 h), following a published procedure.33 After removing
the samples from the solutions and rinsing with toluene and ethanol, we
reimmersed each sample in 10 mL of pure absolute ethanol with 20 μL of
TEA or concentrated aqueous ammonia solution for another 24 h to
ensure the complete removal of acetyl groups on the outer terminus of
the SAMs. All SAM samples were rinsed thoroughly with toluene and
ethanol and stored in argon-purged absolute ethanol for less than 12 h
and dried under a stream of N2 before all measurements.
We characterized the SAMs using spectroscopic ellipsometry
and XPS. Ellipsometry measurements were carried out on a VASE
19866
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
Figure 3. UPS spectra of (A) AnthSH and Anth(SH)2 and (B) TetSH and Tet(SH)2 on Ag, Au, and Pt. Binding energies are referenced to the Fermi
level, EF. Arrows indicate the onsets of the HOMOs. The spectral intensity of SAM-coated metal substrates was normalized to the intensity of bare metal
substrates at 0 eV. “—” represents chemical contact to metal.
SAM growth. For flat Pt substrates, 3000 Å of Pt was sputter-coated onto
a clean Si wafer in a Perkin-Elmer 2400 DC sputterer at a rate of ∼3 Å/s.
On top of the Pt film, subsequent depositions of 300 Å of Cr and 2000 Å
of Au were carried out in a thermal evaporator. Note that the Cr layer
prevented the penetration of Au atoms into the Pt film. The Au film
enhanced the yield of flat Pt substrates due to better adhesion with the
cured epoxy layer. The rest of the steps were the same as for flat Ag or Au
substrates.
length dependence series. The tips were calibrated with respect to an
octanethiol SAM reference sample; only tips that gave the same
resistance (6.2 ꢁ 107 Ω) within 5% were used for the dithiol oligoacene
measurements. Moreover, tips were always checked by comparing
resistances on the same oligoacene dithiol before moving to other
oligoacene samples. Averaging of the IꢀV data was accomplished in the
same manner as described above.
CurrentꢀVoltage (IꢀV) Measurements. We formed the junc-
tions by contacting a SAM on flat Ag, Au, or Pt substrates with a Ag-, Au-,
or Pt-coated AFM tip at a load of 1 nN using a Veeco (Digital
Instruments) Multimode AFM in a glovebox (O2 < 7 ppm), as illustrated
in Figure 2.7,8 Currentꢀvoltage measurements were carried out using a
Keithley model 236 electrometer with a DC voltage applied to the AFM
tip as controlled with LabVIEW software. For the length dependence
experiments, each metalꢀmoleculeꢀmetal junction was examined over
a voltage range of (0.2 V. Ten to twenty IꢀV traces were collected for
each molecular junction. A tip radius of ∼50 nm was used in all
experiments, and in our previous work13,36,37 the corresponding contact
area was estimated to be on the order of 25 nm2, which would
correspond to roughly 100 molecules in the contact. Linear fits to each
IꢀV trace yielded low-bias (within (0.2 V) resistances that were then
averaged for each molecule. Extrapolation of this low-bias average
resistance versus the number of benzene rings to zero rings gave
the contact resistance (R0). The slope of the resistance versus length
plot gave the β value. For larger voltage sweeps (up to (0.8ꢀ1 V),
10ꢀ20 IꢀV traces were collected and averaged for each SAM.
In general, measurements on oligoacene dithiol junctions were more
difficult than on monothiol junctions because of sudden tip degradation.
‘Degradation’ means that tips suddenly showed extremely low currents
after normal behavior. Tip degradation was probably due to either
(i) ohmic heating which disrupted or fractured the tip metal coating,
(ii) contamination of the tip by bonding and extraction of molecules
from the dithiol SAM, or (iii) mechanical failure (peeling) of the tip
metal film when the tip was retracted from the surface. To facilitate
the dithiol measurements, several tips were employed to complete the
’ RESULTS AND DISCUSSION
Energy Level Alignment. Electronic structures of the oligoa-
cene SAMs were probed by UVꢀvisible absorption spectro-
scopy and UPS spectroscopy. UVꢀvisible absorption spectra
(Figure S5, Supporting Information) reveal that as the number of
benzene rings increases, the HOMOꢀLUMO gaps decrease
linearly from 4.0 to 2.5 eV because of efficient conjugation in the
planar aromatic structure. We determined the HOMO levels
with respect to the metal Fermi level (EF) from the onset of
photoemission. Figure 3 displays the UPS spectra for SAMs of
AnthSH, Anth(SH)2, TetSH, and Tet(SH)2 on Ag, Au, and Pt
surfaces (see Figure S3 in Supporting Information for details
illustrating how we determined the HOMO onset and work
function from UPS spectra). The HOMO levels (EHOMO,onset
)
of all oligoacenes are compiled in Table 1. As expected, the
EHOMO,onset levels move closer to EF as molecular length in-
creases.38 Moreover, for the oligoacene dithiols, the HOMObands
lie slightly closer to EF than for the corresponding monothiol
molecular bands, and peaks for dithiol monolayers appear
broader than those of monothiol monolayers. The peak broad-
ening could have several origins: (i) different intrinsic character-
istics, i.e., different molecular orbital mixing between metalꢀ
oligoacene monothiol and metalꢀoligoacenedithiol; (ii) some-
what lower quality of the dithiol monolayer because part of the
dithiol molecules are lying down (see Table S2 in Supporting
19867
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
Table 1. Summary of EF ꢀ EHOMO,onset Values and Modified
EF.25,26,43 These calculations are consistent with our results
(see below) and prior UPS work by others.38,40,41 Therefore,
we believe it is likely that the metalꢀS BD is mainly responsible
for ΔΦ and that the BD depends strongly on metal type (Ag <
Au < Pt).
Metal Work Function Obtained from UPS Spectra (unit: eV)a
film work function
EF ꢀ EHOMO,onset
Ag (4.26)
Au (5.20)
Pt (5.65)
Figure 4 summarizes our findings concerning energy level
alignment for oligoacene molecules on Ag, Au, and Pt based on
the UPS and UVꢀvisible data. Figure 4A is a scheme of the
vacuum level shift due to ΔΦ and the work function ΦSAM for
the three different metals. Figure 4B shows the HOMO and
LUMO levels of all oligoacene SAMs with respect to EF. The
LUMO levels are constructed from the optical HOMOꢀLUMO
gaps, assuming that the optical gaps of the SAMs are the same as
those of the molecules in solution and independent of the metal
substrate.44ꢀ46 (Note that the LUMO positions are underesti-
mated by the exciton binding energy, 0.5ꢀ1.0 eV.46) From the
estimated energy level alignment in Figure 4B, it is evident that
the HOMO levels lie much closer to EF, which indicates that the
HOMO plays an important role in facilitating tunneling in
junctions based on all of the oligoacenes. Of course, these energy
level diagrams based on UPS and UVꢀvisible data do not
account for the influence of the second metal contact on the
electronic structure; these diagrams are for single SAM-coated
metal interfaces. It is likely that the effect of a second metal
contact in a junction would be to position the HOMO band even
closer to EF, as reported in previous studies.47ꢀ51 This is because
one expects that strong coupling between the molecular orbitals
and the second metal due to the metalꢀS bonds would further
broaden the frontier orbitals, resulting in a decrease in EF ꢀ
EHOMO for the full junction.52ꢀ55 Note that this view presumes
that HOMO level broadening in a dithiol junction overrides any
effect of a second bond dipole at the second contact; i.e., the
orientation of the metalꢀS bond dipoles is such that one might
conclude that a second metalꢀS bond would actually increase
EF ꢀ EHOMO. This does not appear to happen (based on trans-
port data below), and understanding the detailed reason for this
will probably require further calculations.
Figure 4 shows clearly that for a given oligoacene length the
HOMO level position, EF ꢀ EHOMO does not vary with metal
type as much as one might expect based on the 1.4 eV span in
work functions of the unmodified metals (Ag to Pt). This point is
quantified in Figure 5A which displays EF ꢀ EHOMO versus Φ for
the oligoacene monothiols on Ag, Au, and Pt. The linear fit to the
anthracene thiol data reveals that while EF ꢀ EHOMO is correlated
with Φ, the dependence is weak (slope = ꢀ0.22), indicating weak
pinning of the HOMO level with respect to the metal. On the
other hand, as expected, EF ꢀ EHOMO does depend on the length
of the oligoacene, as shown in both Figure 5A and 5B.
The correlation of ΔΦ with Φ and molecular length, Figure 5C
and 5D, is important for our later discussion. Figure 5C reveals
that ΔΦ trends linearly with Φ, with a slope of ꢀ0.86 and a
correlation coefficient R2 = 0.89 for the linear fit to the monothiol
data. The interpretation is that the magnitude of the metalꢀS BD
scales almost one-to-one with Φ, becoming very large for
the PtꢀS bond. In contrast, ΔΦ has almost no dependence
on molecular length, Figure 5D, which is consistent with the
interpretation that ΔΦ originates primarily from the metalꢀS
BD, and that therefore ΔΦ represents a very localized barrier
height change, essentially within the metalꢀS contact; this is a
point to which we will return.
molecule
Ag
Au
Pt ΦSAM ΔΦ ΦSAM ΔΦ ΦSAM ΔΦ
PhSH
1.2
1.1
0.8
0.7
1.1
1.1
0.9
0.6
0.4
0.9
0.6
0.5
0.4
0.8
0.6
0.5
0.4
0.8
0.5
0.5
0.4
3.9 ꢀ0.4 4.4 ꢀ0.8 4.0 ꢀ1.7
4.0 ꢀ0.3 4.4 ꢀ0.8 4.0 ꢀ1.7
3.9 ꢀ0.4 4.2 ꢀ1.0 4.1 ꢀ1.6
3.9 ꢀ0.4 4.1 ꢀ1.1 4.1 ꢀ1.6
3.8 ꢀ0.5 4.3 ꢀ0.9 4.5 ꢀ1.2
4.0 ꢀ0.3 4.1 ꢀ1.1 4.3 ꢀ1.4
4.1 ꢀ0.2 4.3 ꢀ0.9 4.2 ꢀ1.5
4.1 ꢀ0.2 4.0 ꢀ1.2 4.2 ꢀ1.5
NaphSH
AnthSH
TetSH
Ph(SH)2
Naph(SH)2 0.9
Anth(SH)2 0.8
Tet(SH)2
0.6
a Experimental errors are approximately (0.1 eV.
Information); (iii) inhomogeneous molecular environments and
intermolecular coupling.39
It is also evident from the photoemission cutoff regime of the
UPS spectra (shownin Figures S3 and S4) that there is a substantial
work function change (ΔΦ = ΦSAM ꢀ Φ) for the metals due to the
presence of the adsorbed SAMs. Here Φ represents the original
metal work function, and ΦSAM is the work function of the SAM-
coated metal. The values of Φ, ΦSAM, and the work function
changes, ΔΦ, are reported in Table 1, where it is clear that there is
very weak dependence of both ΦSAM and ΔΦ on oligoacene length
or the presence/absence of a terminal SH group. However,
ΔΦ does depend strongly on metal type: ΔΦ is ꢀ0.3 ((0.1),
ꢀ1.0 ((0.1), ꢀ1.5 ((0.2) eV for Ag, Au, and Pt, respectively. ΔΦ
substantially reduces the intrinsic metal work function difference
for the SAMs; the HOMO levels of the oligocenes are ultimately
lying relatively close to each other on different metal substrates (see
EF ꢀ EHOMO,onset in Table 1). For a given oligoacene length, the
EF ꢀ EHOMO,onset values are quite similar, especially when con-
sidering the large variation in the work functions of the unmodified
metals: Ag (4.26 eV), Au (5.20 eV), and Pt (5.65 eV).
Work function changes for SAMs on metals have been investi-
gated extensively experimentally38,40,41 and theoretically.25,26,42,43
Conceptually, work function changes for metal surfaces with
oriented, adsorbed molecules are considered to be the sum of
three interfacial dipoles: (i) the intrinsic molecular dipole, (ii) the
intrinsic metal dipole that changes upon adsorption (i.e., the
surface electron “push-back” effect);40,41 (iii) the metalꢀlinker
(e.g., metalꢀS) bond dipole. Recent density functional theory
(DFT) calculations have delineated the local charge redistribu-
tion at metal/SAM interfaces that gives rise to the bond dipole
(BD).25,26,42,43 Using DFT, Heimel et al. have shown that upon
adsorption of aromatic thiolate to various metals, electron
density in the C(aromatic ring)ꢀS bond decreases while that
of the Sꢀmetal bond increases, i.e., there is a net electron
donation to the metal.25,26 The same computations reveal that
the magnitudes of the BDs for thiol linkages depend strongly on
the metal type, namely the BD increases as the work function
increases. Furthermore, the BD can result in large electrostatic
potential changes (0ꢀ1 eV) across only a few Å distance
associated with metalꢀSꢀC(aromatic ring) linkages. The crea-
tion of such bond dipoles in turn mitigates the variation of EF ꢀ
EHOMO with Φ for aromatic SAMs, resulting in partial pinning of
the molecular HOMO levels with respect to the Fermi level
General IꢀV Behavior. We turn now to the transport data.
Figure 6 displays representative IꢀV characteristics of Auꢀoligoacene
19868
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
Figure 4. Estimated vacuum level change (A) and HOMO/LUMO alignment with respect to EF (B) of oligoacene thiols and dithiols for Ag, Au, and Pt
surfaces (the data from Table 1 and Figure S5). In panel A, ΔΦ stands for the work function change, ΦSAM for the work function of SAM-coated metals,
EF for the metal Fermi level, and Evac for the vacuum level. In panel B, the LUMO levels are constructed from the optical HOMOꢀLUMO gaps.
thiolꢀAu and Auꢀoligoacene dithiolꢀAu junctions. Each IꢀV
trace has a sigmoidal shape and is nearly symmetric with respect
to zero bias. For a given voltage, the current decreases exponen-
tially with length. These attributes indicate that nonresonant
tunneling is the principal transport mechanism at low bias in
these junctions.56,57 Detailed analysis and discussion of both the
low and high voltage transport regimes follow below.
Low Bias Junction Characteristics. We have examined the
molecular length dependence of the junction resistance in the
low bias regime ((0.2 V). Figure 7 displays a semilog plot of
resistance versus number of rings for oligoacene thiol and dithiol
junctions. Each data point reflects the average of 10ꢀ20 IꢀV
traces. Note that we were not able to acquire good data for
Agꢀoligoacene dithiolꢀAg junctions because of junction instabil-
ity. This is perhaps because Ag atoms are liable to electromigration
under bias.58,59 The linear relationships in Figure 7 indicate that
the data are well explained by nonresonant tunneling (see eq 1),
with an average β value determined from the slopes of the
semilog plots of 1.0 ( 0.13 per ring or 0.5 Åꢀ1 for monothiols
and 0.5 ( 0.07 per ring or 0.2 Åꢀ1 for dithiols.
The β value in the dithiol junctions is consistent with previous
reports on tunneling in aromatic dithiols and diamines.18,60ꢀ62
However, the difference in β between monothiols and dithiols
with the same conjugated backbone has not been reported
previously (β values for alkane thiols and dithiols are the same7).
This difference is perhaps surprising given that the molecular
backbone structures are identical, and it implies that EF
ꢀ
EHOMO offsets for dithiols are lower than for monothiols.63
The HOMO level alignments measured by UPS for the two
types of SAMs are not very different, however (the difference of
19869
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
Figure 5. Correlations of the data shown in Table 1. (A) Plot of EF ꢀ EHOMO,onset versus metal work function Φ. (B) Plot of EF ꢀ EHOMO,onset versus
number of rings. (C) Plot of work function change ΔΦ versus metal work function Φ. (D) Plot of work function change ΔΦ versus number of rings.
Symbols for metals: Ag (square), Au (circle), and Pt (triangle) and colors for molecules: PhSH (black), NaphSH (green), AnthSH (blue), and TetSH
(red). Blue lines are linear least-squares fits in plots A and C and guides for the eye in plots B and D.
EF ꢀ EHOMO,onset values between monothiol and corresponding
dithiol SAMs is approximately 0ꢀ0.2 eV, see Table 1). Yet here it
is important to note that the EF ꢀ EHOMO offsets measured by
UPS do not reveal the actual offsets in completed junctions with
two contacts. That is, the scheme in Figure 4 likely does not
capture the electronic role played by the second contact in a fully
assembled junction. This conclusion is consistent with many
previous studies in which the tunneling barrier in alkanethiol
junctions has been estimated to be much lower than the spectro-
scopic EF ꢀ EHOMO offset.7,27,40,64 It is reasonable to expect that
for junctions based on dithiol molecules the tunneling barriers
will be significantly smaller than the barriers for monothiols,
because there is likely significant electronic broadening of the
HOMO level in junctions with two chemical contacts.51,52,64 The
lower β values for dithiols reflect the lower barriers and highlight
the key role of chemical contacts; two chemical contacts in the
oligoacene dithiol junctions create a smaller tunneling barrier
through the oligoacene backbone.
Figure 7 also indicates that the β values for both mono- and
dithiols, respectively, appear to be independent of metal work
function within experimental uncertainty. The results of multiple
measurements of β versus work function are summarized in
Figure 8A, and indeed there appears to be no systematic trend.
We conclude then that the average tunneling barrier height is
strongly dependent on the number of chemical contacts (1 vs 2)
but not on metal type. The lack of dependence of β on Φ is a
central result of this work, and it is consistent with our finding
that EF is weakly pinned in these systems, i.e., that the effective
tunneling barrier is approximately the same for all monothiol
junctions and for all dithiol junctions, respectively (see Figure 4B).
We will return to this below in the context of describing the
nature of the tunneling barrier.
We can also conclude from Figure 7 that the precise values of
the metalꢀSꢀC bond angles, which can be different for the Ag,
Au, and Pt substrates,65 are not critical for β, as it is clear that the
slope of the resistance vs length data are independent of the
contact type. This result is consistent with “through-bond
tunneling” and previous reports.24,37
The junction contact resistances R0 can be determined from
the y-intercepts of the plots in Figure 7. All R0 values are
displayed as a function of unmodified metal work function (Φ)
in Figure 8B. There are three important observations associated
with Figure 8B: (i) R0 decreases strongly with increasing metal
work function; (ii) dithiol contacts have much lower R0 than
monothiol contacts for a given set of metal electrodes; (iii) mole-
cular junctions incorporating a pair of mixed metal electrodes
(e.g., Ag tip/Au substrate and Au tip/Ag substrate) have R0
values that are positioned between the R0s for the homometal
junctions (e.g., Ag tip/Ag substrate and Au tip/Au substrate).
These observations are consistent with our previous report
on work function effects in junctions incorporating alkane thiols
and dithiols7 and with the work of others on mono- versus
19870
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
Figure 6. Representative linear and semilog plots of average IꢀV traces of Auꢀoligoacene thiolꢀAu (A, B) and Auꢀoligoacene dithiolꢀAu junctions
(C, D). Error bars in plots B and D are given at a single point near 0.6 V for clarity.
but the low R0 for dithiols is consistent with previous reports that
chemical contacts are better for transport than physical contacts.2,7,9,16,18
We emphasize that the dependences of R0 and β on Φ are
completely different (compare parts A and B of Figure 8); this is a
key result. The fact that the dependences are so different implies
that the two parameters are sensitive to different aspects of the
junction tunneling barrier. Any model for junction transport
must take this into account.
To clarify the origin of the strong dependence of R0 on Φ, we
have also examined the correlation of R0 with ΔΦ, Figure 9.
Figure 9 reveals that R0 correlates very well with ΔΦ (R2 = 0.9),
which is not unexpected because ΔΦ and Φ are well-correlated
as shown in Figure 5C. Like the dependence of β and R0 on Φ,
we believe the R0ꢀΔΦ correlation is an important discovery for
the overall picture of transport in these junctions, as elaborated
in the next section. In addition, we have found that R0 is not
well correlated with EF ꢀ EHOMO (data shown in Figure S6,
Supporting Information), which can be anticipated because EF is
weakly pinned in these systems, as already discussed.
Figure 7. Semilog plot of resistance versus number of rings for
oligoacene thiol and dithiol junctions. Resistances were calculated from
the average of 10ꢀ20 IꢀV traces within (0.2 V. The error bars
represent one standard deviation from the mean. “ꢀ” represents
chemical contact to metal and “/” represents physical contact.
We have also examined the bias dependence of β and R0 in
AuꢀmoleculeꢀAu junctions (Figure 8C and 8D). At all biases,
both β and R0 values for dithiol junctions are lower than those of
monothiol junctions. However, β and R0 show little voltage
dependence for dithiol junctions. In contrast, the monothiol
junctions do exhibit voltage dependence for β and R0 at higher
biases. In monothiol junctions, we found that R0 decreased by a
factor of 2 over 0ꢀ1 V while β remained initially constant but
decreased for biases greater than (0.6 V. The changes in β and
dithiols.16,17 Indeed, the metal work function has a dramatic
effect on R0 for the monothiol oligoacene junctions; R0 decreases
by 3 orders of magnitude for a ∼1.4 eV increase in metal work
function. For dithiols, the effect is not as strong (R0 decreases by
about a factor of 10 or 20 over the same range of work functions),
19871
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
Figure 8. (A) β as a function of electrode metal work function for junctions composed of oligoacene thiols and dithiols. For both thiols and dithiols,
β values appear independent of metal work function Φ. The solid line is the average of each series, and dotted lines are one standard deviation above and
below the average. (B) Semilog plot of contact resistance as a function of electrode metal work function Φ for junctions composed of oligoacene
monothiol and dithiols. The solid lines are guides for the eye. (C, D) Dependence of β (C) and contact resistance (D) on applied voltage in
Auꢀoligoacene thiolꢀAu and Auꢀoligoacene dithiolꢀAu junctions.
monothiols it is clear that β does have some voltage dependence,
as expected (the tunneling barrier decreases with applied bias).
The Nature of the Tunneling Barrier. From the above
discussion, we have seen that β for oligoacene thiols and dithiols
is independent of the metal work function, which is consistent
with the observation that EF ꢀ EHOMO (that is the tunneling
barrier through the backbone) is weakly pinned. On the other
hand, R0 strongly depends on Φ and more particularly on ΔΦ,
which is related to the presence of metalꢀS bond dipoles at the
contacts. To rationalize these collective observations, we propose
a triple barrier model for the junction, Figure 10. The model
follows the work of Heimel et al.25,26 and is similar to the triple
rectangular barriers proposed by Lee et al.,16,17 based on similar
length-dependent measurements on alkane thiols. The model
explicitly addresses electrostatic potential barriers across the
metalꢀmoleculeꢀmetal junction. Figure 10A, for example,
depicts the electrostatic potential profile across an AgꢀS-mole-
Figure 9. Semilog plot of contact resistance of homometal junctions versus
work function change (ΔΦ). Symbols for metals: Ag (square), Au (circle),
and Pt (triangle). Colors for molecules: PhSH (black), NaphSH (green),
AnthSH (blue), and TetSH (red). Blue line shows a linear least-squares fit.
cule interface, which is a result of combining the potentials for the
metal, molecule, and the AgꢀS bond dipole (BD). It is the
magnitude of the BD that changes for junctions based on Ag, Au, or
Pt (panels A, B, and C). The changing BD results in different
“chemical contact barriers” as shown. Panel D represents the
full triple barriers (left contactꢀbridgeꢀright contact) for thiol
and dithiol junctions. The heights of the contact barriers are
R0 were not dramatic (certainly in the case of R0, the Φ (and
ΔΦ) dependence is much greater), but at least in the case of
19872
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
Figure 10. Schematic diagram showing electrostatic potential change at the moleculeꢀmetal interfaces and resulting molecular energy level alignment
as well as the formation of contact barriers depending on metal substrates for (A) Ag, (B) Au, and (C) Pt contacts. Panel D shows schematic tunneling
barriers formed at the metalꢀoligoacene monothiolꢀmetal (left) and metalꢀoligoacene dithiolꢀmetal junctions (right).
proportional to ΔΦ, and, as we have shown, the absolute value of ΔΦ
becomes larger as the bare metal work function increases, Figure 5C.
We propose that large contact barriers give rise to large R0
values. Likewise, we propose that weak pinning of the HOMO
position, due to the BD formation, leads to the independence of
β on Φ. The precise value of EF ꢀ EHOMO will depend strongly
on the number of chemical contacts (1 versus 2) and thus β will
be smaller for dithiols versus monothiols (Figure 10D). Theore-
tical calculations by Bredas and Ratner also support similar
HOMO band alignments on different metal substrates, and
they attribute conductance differences of junctions mainly to
electronic coupling at the interfaces.24,25 Overall, the qualitative
description of the junction electrostatics presented in Figure 10 is
consistent with our quantitative measurements (it provides a
basis for understanding the very different dependences of β
and R0 on Φ), and thus we believe it may provide valuable
physical insight into the properties of molecular tunnel junctions.
We reiterate that motivation for the triple barrier model came
from the very different dependences of β and R0 on Φ.
voltages, there is a steeper dependence of I on V, as ex-
pected.14,27,28,45,64,66ꢀ68 Figure 11 displays logꢀlog plots of the
IꢀV characteristics for PhSH, NaphSH, AnthSH, Ph(SH)2,
Naph(SH)2, and Anth(SH)2 with Au contacts. The vertical
dotted lines correspond to the voltages (Vtrans) where there
is a change in the slopes of the plots. For convenience only
(not theoretically motivated), the currents were fit to a power
law, I µ V n; the values of n are indicated in Figure 11. Below Vtrans
for both monothiol and dithiol junctions, I depends approxi-
mately linearly on V (n = 1.1ꢀ1.4), as expected for low bias
nonresonant tunneling. Above Vtrans, n is 2.1ꢀ4.6. Basic theories
of tunneling of course predict strong increases in current as the
applied potential becomes comparable to the average tunneling
barrier,56 and qualitatively Vtrans should be a measure of when
this occurs.
Figure 12 summarizes the dependence of Vtrans for monothiol
junctions on the metal work function, molecular length, and
HOMO level position. Similar data for dithiols are shown in Figure
S8. From Figure 12A, it is clear Vtrans decreases systematically with
both increasing metal work function and increasing molecular
length. The work function dependence of Vtrans is relatively weak,
Transport at Higher Bias. The discussion so far has fo-
cused on junction resistance data at low biases. With increasing
19873
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
Figure 11. Logꢀlog plot of representative AuꢀmoleculeꢀAu junctions: (A) Auꢀbenzene thiolꢀAu, (B) Auꢀbenzene dithiolꢀAu,
(C) Auꢀnaphthalene thiolꢀAu, (D) Auꢀnaphthalene dithiolꢀAu, (E) Auꢀanthracene thiolꢀAu, and (F) Auꢀanthracene dithiolꢀAu. The dotted
line corresponds to the voltage (Vtrans) where the currentꢀvoltage relationship becomes steeper. The slopes in the logꢀlog plots are from the equation,
I µ Vn.
i.e., a 1.3 eV change in the work function results in only ∼0.2 eV
change in Vtrans. However, Vtrans is more sensitive to molecular
length, suggesting that there is a dependence on the HOMO level
The meaning of Vtrans is currently being debated in the
literature.27,28,55,64,66ꢀ73 Originally, we and our collaborators iden-
tified Vtrans as the minimum in a FowlerꢀNordheim plot (ln(I/V2)
versus 1/V) and suggested that Vtrans corresponded to the onset
of field emission in the junction.27,28 This idea was motivated by
(1) the observation that for voltages above Vtrans the Fowlerꢀ
Nordheim plot exhibited approximately linear behavior with a
negative slope, as expected for field emission, (2) the clear
correlation of Vtrans with EF ꢀ EHOMO, and (3) the fact that
fields in molecular junctions range from 108 to 109 V/m.
However, the field emission interpretation has been called into
question. Huisman et al. pointed out that within the simple
offset. Figure 12B shows the dependence of Vtrans on EF ꢀ EHOMO
,
which is the most striking result; the correlation is excellent
(R2 = 0.92) and reveals that Vtrans increases linearly with EF ꢀ
EHOMO as we and others have reported before for similar
systems.14,27,28,45,64,66,67,68 In addition, for a given set of electrodes,
Vtrans is lower for dithiol junctions compared to the corresponding
monothiol junctions (see Figure 11). Thus, Vtrans depends on all
the critical variables in this study: molecular length, type of contact,
and metal work function.
19874
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
Figure 12. (A) Vtrans as a function of electrode work function for junctions composed of oligoacene monothiols. (B) Vtrans as a function of EF ꢀ
EHOMO,onset for homometal junctions composed of oligoacene monothiols. Symbols for metals: Ag (square), Au (circle), Pt (triangle), Ag/Au
(diamond), Ag/Pt (pentagon), and Au/Pt (inverted triangle). Colors for molecules: PhSH (black), NaphSH (green), AnthSH (blue), and TetSH (red).
Simmons barrier model for the junction that Vtrans should scale
increased. This behavior was self-consistently explained in terms
of the contact bond dipoles and the electronic structure of the
junction; namely β is independent of contact type because of
weak Fermi level pinning (EF ꢀ EHOMO varies only weakly
with Φ), but R0 varies strongly with contact type because of the
metalꢀS bond dipole that is responsible for the Fermi level
pinning. A previously published triple barrier model for molec-
ular junctions was proposed to rationalize these results in which
contact barriers, proportional to the size of the interfacial bond
64
with (EF ꢀ EHOMO
)
1/2, not EF ꢀ EHOMO
.
Furthermore,
modeling the junction IꢀV characteristics with the Landauer
equation and a broadened molecular state also yielded the
minimum in a FowlerꢀNordheim plot; the concept of field
emission was not necessary.
The question remains as to what occurs at Vtrans, for example
whether at this voltage some of the tail states in the junction
electronic density of states become accessible for resonant
tunneling.71 One would expect that in such a situation, β would
dipoles, determine R0, and the bridge barrier, EF ꢀ EHOMO
,
depend more strongly on voltage than what we observe. The
determines β. Finally, for larger voltage sweeps, a characteristic
transition voltage Vtrans associated with steeper dependence of
current on voltage was identified. Vtrans values were well corre-
lated with EF ꢀ EHOMO. Overall, the results presented here
provide a comprehensive data set for tunneling through the
oligoacene systems. We believe the findings on the oligoacene
model system will be broadly representative of the behavior of
other junctions based on π-conjugated molecules sandwiched
between metal contacts. In particular, we expect that other
systems will show that R0 and β are sensitive to different parts
of the junction tunneling barrier.
68,69
contacts also have an important effect on the value of Vtrans
,
and it may be difficult to develop a model without knowing how
potential varies across a junction (i.e., how the voltage drops at
the contacts versus across the molecular backbone). Precise under-
standing of Vtrans and its relation to the contact barriers and the
electronic states in the junction will require further experimental
and theoretical work. However, the data in Figures 11 and 12 add to
the growing body of experimental results in the literature27,28,66ꢀ73
that show Vtrans is a reproducible quantity that is directly related to
the HOMO (or LUMO) position in the junction.
’ ASSOCIATED CONTENT
’ SUMMARY
S
We investigated the transport properties of tunnel junctions
based on rigid, conjugated oligoacene monothiols and dithiols as
a function of molecular length and contact work function and
correlated the results with electronic structure determined by
UPS. UPS revealed weak Fermi level pinning in this system
which was associated with the formation of strong metalꢀS bond
dipoles. Length-dependent transport measurements showed the
junction resistance was strongly affected by the surface linker
types (chemical/physical vs bichemical contact). The tunneling
attenuation factor, β, was 1.0 per ring or 0.5/Å for the monothiol
series and 0.5 per ring or 0.2/Å for the dithiol series, and contact
resistance, R0, was 10ꢀ100 times lower for dithiol junctions,
indicating that chemical contacts reduce both the tunneling
barrier height and contact resistance significantly. When combi-
nations of metals (Ag, Au, and Pt) were employed as electrodes
in the junctions, β values were independent of metal work
function, within error, while R0 exhibited strong work function
dependence and was significantly reduced as work function
Supporting Information. Monolayer characterization
b
details, UVꢀvisible absorption spectra, transmission analysis,
correlation between Vtrans and the metal work function for
oligoacene dithiol junctions. This material is available free of
charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
frisbie@umn.edu
’ ACKNOWLEDGMENT
This article is dedicated to Dr. Seong Ho Choi, colleague,
co-author, and friend, whose untimely death in December 2010
was a great loss for us. C.D.F. and X.Y.Z. thank the National
Science Foundation (CHE-0616427) for financial support.
C.D.F. also thanks NSF for a Special Creativity Award. B.S.K.
19875
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
thanks Professor Thomas R. Hoye for help and discussion on
molecular synthesis. Parts of this work were carried out in the
College of Science and Engineering Characterization Facility,
University of Minnesota, which receives partial support from
NSF through the NNIN program.
(29) Xue, Y. Q.; Datta, S.; Ratner, M. A. J. Chem. Phys. 2001, 115,
4292.
(30) Xue, Y. Q.; Ratner, M. A. Phys. Rev. B 2004, 69, 085403.
(31) Reddy, P.; Jang, S. Y.; Segalman, R. A.; Majumdar, A. Science
2007, 315, 1568.
(32) Armarego, W. L. F.; Chai, C. L. L. Purification of Laboratory
Chemicals, 5th ed.; Butterworth Heinemann: Boston, 2003.
(33) Shaporenko, A.; Elbing, M.; Baszczyk, A.; von Hanisch, C.;
Mayor, M.; Zharnikov, M. J. Phys. Chem. B 2006, 110, 4307.
(34) Ulman, A. Chem. Rev. 1996, 96, 1533.
(35) Li, Z. Y.; Chang, S. C.; Williams, R. S. Langmuir 2003, 19, 6744.
(36) Engelkes, V. B.; Beebe, J. M.; Frisbie, C. D. J. Phys. Chem. B
2005, 109, 16801.
(37) Salomon, A.; Cahen, D.; Lindsay, S.; Tomfohr, J.; Engelkes,
V. B.; Frisbie, C. D. Adv. Mater. 2003, 15, 1881.
(38) Zangmeister, C. D.; Picraux, L. B.; van Zee, R. D.; Yao, Y. X.;
Tour, J. M. Chem. Phys. Lett. 2007, 442, 390.
(39) Zangmeister, C. D.; Robey, S. W.; van Zee, R. D.; Yao, Y.; Tour,
J. M. J. Phys. Chem. B 2004, 108, 16187.
(40) Alloway, D. M.; Hofmann, M.; Smith, D. L.; Gruhn, N. E.;
Graham, A. L.; Colorado, R.; Wysocki, V. H.; Lee, T. R.; Lee, P. A.;
Armstrong, N. R. J. Phys. Chem. B 2003, 107, 11690.
(41) Campbell, I. H.; Rubin, S.; Zawodzinski, T. A.; Kress, J. D.;
Martin, R. L.; Smith, D. L.; Barashkov, N. N.; Ferraris, J. P. Phys. Rev. B
1996, 54, R14321.
’ REFERENCES
(1) Lindsay, S. M.; Ratner, M. A. Adv. Mater. 2007, 19, 23.
(2) Kushmerick, J. G. Mater. Today 2005, 8, 26.
(3) Chen, F.; Hihath, J.; Huang, Z. F.; Li, X. L.; Tao, N. J. Annu. Rev.
Phys. Chem. 2007, 58, 535.
(4) McCreery, R. L.; Bergren, A. J. Adv. Mater. 2009, 21, 4303.
(5) Park, Y. S.; Whalley, A. C.; Kamenetska, M.; Steigerwald, M. L.;
Hybertsen, M. S.; Nuckolls, C.; Venkataraman, L. J. Am. Chem. Soc.
2007, 129, 15768.
(6) Ko, C.-H.; Huang, M.-J.; Fu, M.-D.; Chen, C.-H. J. Am. Chem.
Soc. 2009, 132, 756.
(7) Engelkes, V. B.; Beebe, J. M.; Frisbie, C. D. J. Am. Chem. Soc.
2004, 126, 14287.
(8) Kim, B.-S.; Beebe, J. M.; Jun, Y.; Zhu, X.-Y.; Frisbie, C. D. J. Am.
Chem. Soc. 2006, 128, 4970.
(9) Wu, S. M.; Gonzalez, M. T.; Huber, R.; Grunder, S.; Mayor, M.;
Schonenberger, C.; Calame, M. Nat. Nanotechnol. 2008, 3, 569.
(10) Malen, J. A.; Doak, P.; Baheti, K.; Tilley, T. D.; Segalman, R. A.;
Majumdar, A. Nano Lett. 2009, 9, 1164.
(42) Bilic, A.; Reimers, J. R.; Hush, N. S. J. Chem. Phys. 2005, 122,
094708.
(11) Kamenetska, M.; Quek, S. Y.; Whalley, A. C.; Steigerwald,
M. L.; Choi, H. J.; Louie, S. G.; Nuckolls, C.; Hybertsen, M. S.; Neaton,
J. B.; Venkataraman, L. J. Am. Chem. Soc. 2010, 132, 6817.
(12) Park, Y. S.; Widawsky, J. R.; Kamenetska, M.; Steigerwald,
M. L.; Hybertsen, M. S.; Nuckolls, C.; Venkataraman, L. J. Am. Chem.
Soc. 2009, 131, 10820.
(13) Beebe, J. M.; Engelkes, V. B.; Miller, L. L.; Frisbie, C. D. J. Am.
Chem. Soc. 2002, 124, 11268.
(14) Choi, S. H.; Kim, B.; Frisbie, C. D. Science 2008, 320, 1482.
(15) Lu, Q.; Liu, K.; Zhang, H.; Du, Z.; Wang, X.; Wang, F. ACS
Nano 2009, 3, 3861 .
(16) Wang, G.; Kim, T. W.; Jang, Y. H.; Lee, T. J. Phys. Chem. C 2008,
112, 13010.
(17) Wang, G.; Kim, T. W.; Lee, H.; Lee, T. Phys. Rev. B 2007, 76,
205320.
(18) Liu, K.; Li, G. R.; Wang, X. H.; Wang, F. S. J. Phys. Chem. C
(43) Romaner, L.; Heimel, G.; Zojer, E. Phys. Rev. B 2008, 77,
045113.
(44) Zangmeister, C. D.; Robey, S. W.; van Zee, R. D.; Kushmerick,
J. G.; Naciri, J.; Yao, Y.; Tour, J. M.; Varughese, B.; Xu, B.; Reutt-Robey,
J. E. J. Phys. Chem. B 2006, 110, 17138.
(45) Zangmeister, C. D.; Beebe, J. M.; Naciri, J.; Kushmerick, J. G.;
van Zee, R. D. Small 2008, 4, 1143.
(46) Zangmeister, C. D.; Robey, S. W.; vanZee, R. D.; Yao, Y.; Tour,
J. M. J. Am. Chem. Soc. 2004, 126, 3420.
(47) Heimel, G.; Romaner, L.; Bredas, J. L.; Zojer, E. Surf. Sci. 2006,
600, 4548.
(48) Heimel, G.; Romaner, L.; Bredas, J. L.; Zojer, E. Langmuir 2008,
24, 474.
(49) Kubatkin, S.; Danilov, A.; Hjort, M.; Cornil, J.; Bredas, J. L.;
Stuhr-Hansen, N.; Hedegard, P.; Bjornholm, T. Nature 2003, 425, 698.
(50) Hedegard, P.; Bjornholm, T. Chem. Phys. 2005, 319, 350.
(51) Kaasbjerg, K.; Flensberg, K. Nano Lett. 2008, 8, 3809.
(52) Thygesen, K. S.; Rubio, A. Phys. Rev. Lett. 2009, 102, 046802.
(53) Garcia-Lastra, J. M.; Rostgaard, C.; Rubio, A.; Thygesen, K. S.
Phys. Rev. B 2009, 80, 245427.
2008, 112, 4342.
(19) Note that according to eq 1, R is proportional to R0 at any
molecular length s, which is very different from the situation in
macroscopic transport measurements on semiconductors. In those
cases, R0 is normally an additive factor to the total R and becomes less
important as the contacts become further apart. Equation 1 shows that
because R0 is a multiplicative coefficient it is important at any length s.
See also refs 7 and 57.
(54) Neaton, J. B.; Hybertsen, M. S.; Louie, S. G. Phys. Rev. Lett.
2006, 97, 216405.
(55) Chen, J. Z.; Markussen, T.; Thygesen, K. S. Phys. Rev. B 2010,
82, 121412.
(56) Simmons, J. G. J. Appl. Phys. 1963, 34, 1973.
(57) Nitzan, A. Annu. Rev. Phys. Chem. 2001, 52, 681.
(58) Beebe, J. M.; Kushmerick, J. G. Appl. Phys. Lett. 2007, 90,
083117.
(59) Terabe, K.; Hasegawa, T.; Nakayama, T.; Aono, M. Nature
2005, 433, 47.
(60) Yamada, R.; Kumazawa, H.; Noutoshi, T.; Tanaka, S.; Tada, H.
Nano Lett. 2008, 8, 1237.
(61) Tada, T.; Nozaki, D.; Kondo, M.; Hamayama, S.; Yoshizawa, K.
J. Am. Chem. Soc. 2004, 126, 14182.
(62) Quinn, J. R.; Foss, F. W.; Venkataraman, L.; Hybertsen, M. S.;
Breslow, R. J. Am. Chem. Soc. 2007, 129, 6714.
(63) Unpublished result: We found that below a tip load of 10 nN in
monothiol junctions, the β value is almost constant while above 10 nN,
the β value is gradually decreased. Importantly, we have to increase the
load to ∼20 nN to get the β value (0.5/ring) of dithiol junctions. From
our experience, 20 nN is a very high load. Though the dithiol molecules
(20) Dameron, A. A.; Ciszek, J. W.; Tour, J. M.; Weiss, P. S. J. Phys.
Chem. B 2004, 108, 16761.
(21) Quinn, J. R.; Foss, F. W.; Venkataraman, L.; Breslow, R. J. Am.
Chem. Soc. 2007, 129, 12376.
(22) McCreery, R. Electrochem. Soc. Interface 2004, 13, 46.
(23) Kiguchi, M.; Miura, S.; Hara, K.; Sawamura, M.; Murakoshi, K.
Appl. Phys. Lett. 2007, 91, 053110.
(24) Yaliraki, S. N.; Kemp, M.; Ratner, M. A. J. Am. Chem. Soc. 1999,
121, 3428.
(25) Heimel, G.; Romaner, L.; Zojer, E.; Bredas, J. L. Nano Lett.
2007, 7, 932.
(26) Heimel, G.; Romaner, L.; Bredas, J. L.; Zojer, E. Phys. Rev. Lett.
2006, 96, 196806.
(27) Beebe, J. M.; Kim, B.-S.; Frisbie, C. D.; Kushmerick, J. G. ACS
Nano 2008, 2, 827.
(28) Beebe, J. M.; Kim, B.-S.; Gadzuk, J. W.; Frisbie, C. D.;
Kushmerick, J. G. Phys. Rev. Lett. 2006, 97, 026801.
19876
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877

Journal of the American Chemical Society
ARTICLE
have tip adhesive forces stronger than those of monothiols, we do not
believe that this can account for the lower β value.
(64) Huisman, E. H.; Guedon, C. M.; van Wees, B. J.; van der Molen,
S. J. Nano Lett. 2009, 9, 3909.
(65) It is also interesting to consider the impact of different AuꢀSꢀ
molecule bond angles on R0. The bond angles may change systematically
with electrode type, Ag, Au, Pt. We have not systematically measured
bond angles here, and the degree to which bond angles versus work
function play a role will need to be determined.
(66) Song, H.; Kim, Y.; Jang, Y. H.; Jeong, H.; Reed, M. A.; Lee, T.
Nature 2009, 462, 1039.
(67) Wang, G.; Kim, T. W.; Jo, G.; Lee, T. J. Am. Chem. Soc. 2009,
131, 5980.
(68) Pakoulev, A. V.; Burtman, V. J. Phys. Chem. C 2009, 113, 21413.
(69) Yu, L. H.; Gergel-Hackett, N.; Zangmeister, C. D.; Hacker,
C. A.; Richter, C. A.; Kushmerick, J. G. J. Phys.: Condens. Matter 2008,
20, 374114.
(70) Malen, J. A.; Doak, P.; Baheti, K.; Tilley, T. D.; Majumdar, A.;
Segalman, R. A. Nano Lett. 2009, 9, 3406.
(71) Araidai, M.; Tsukada, M. Phys. Rev. B 2010, 81, No. 235114.
(72) Bennett, N.; Xu, G.; Esdaile, L. J.; Anderson, H. L.; Macdonald,
J. E.; Elliott, M. Small 2010, 6, 2604.
(73) Baldea, I. Chem. Phys. 2010, 377, 15.
19877
dx.doi.org/10.1021/ja207751w |J. Am. Chem. Soc. 2011, 133, 19864–19877