Content uploaded by Dong Xiang
Author content
All content in this area was uploaded by Dong Xiang on Jun 14, 2021
Content may be subject to copyright.
Three-Terminal Single-Molecule Junctions Formed by Mechanically
Controllable Break Junctions with Side Gating
Dong Xiang,
†,‡
Hyunhak Jeong,
†
Dongku Kim,
†
Takhee Lee,*
,†
Yongjin Cheng,
‡
Qingling Wang,
§
and Dirk Mayer*
,∥
†
Department of Physics and Astronomy, Seoul National University, Seoul 151-747, Korea
‡
School of Mathematics and Physics, China University of Geosciences, Wuhan 430074, China
§
Faculty of Mechanical and Electronic Information, China University of Geosciences, Wuhan 430074, China
∥
Peter-Grünberg-Institute, PGI-8, Research Center Jülich and JARA Fundamentals of Future Information Technology, Jülich 52425,
Germany
*
SSupporting Information
ABSTRACT: Molecules are promising candidates for elec-
tronic device components because of their small size, chemical
tunability, and ability to self-assemble. A major challenge when
building molecule-based electronic devices is forming reliable
molecular junctions and controlling the electrical current
through the junctions. Here, we report a three-terminal
junction that combines both the ability to form a stable
single-molecule junction via the mechanically controllable
break junction (MCBJ) technique and the ability to shift the
energy levels of the molecule by gating. Using a noncontact side-gate electrode located a few nanometers away from the
molecular junction, the conductance of the molecule could be dramatically modulated because the electrical field applied to the
molecular junction from the side gate changed the molecular electronic structure, as confirmed by the ab initio calculations. Our
study will provide a new design for mechanically stable single-molecule transistor junctions fabricated by the MCBJ method.
KEYWORDS: Molecular electronics, electron transport, molecular transistors, single molecule junction,
mechanically controllable break junction, nanogap electrodes
Producing molecular electronic devices in which individual
molecules or molecular monolayers are utilized as active
electronic components is a promising approach for the ultimate
miniaturization and integration of electronic devices.
1−5
In
particular, building a single-molecule field-effect transistor
(FET) is considered a critical step in molecular electronics.
Although previous theoretical calculations have demonstrated
that single molecules can be used as an active channel in
FETs,
6−12
an experimental demonstration of a true three-
terminal device, one that depends on external modulation of
molecular orbitals, has been a tremendous challenge because it
requires one to (1) find a reliable method to bridge a single
molecule to the source and drain electrodes and (2) place the
gate electrode a few nanometers away from the molecule to
achieve the required gate field.
13
Previous efforts to overcome these challenges have coupled
an electrochemical gate electrode
13−16
or a bottom-gate
electrode into a molecular junction.
17−28
The electrochemical
gate can effectively control the current through the
molecules,
13−15
but it can only be operated in electrolytes.
The bottom gate can directly modulate the molecular orbital
electrostatically in a molecular FET contact configuration.
However, it is not trivial to obtain stable and high-yield single-
molecule junctions with the bottom-gate configuration.
23
Motivated by the bottom-gate design, we adopted a three-
terminal device that combines both the ability to form a
mechanically stable single-molecule junction and the ability to
shift the energy levels of the molecule using a side-gate
electrode in a noncontact configuration.
29,30
Our experiments,
supported by first-principle calculations, demonstrate that the
properties of electronic transport through the single-molecule
junction can be directly modulated by the external electrical
field of the side gate and will enhance the prospects for three-
terminal molecular electronic devices.
The scheme for combining electrostatic gating and
mechanical adjustability is to add a gate electrode to a
mechanically controlled break junction (MCBJ), as shown in
Figure 1a. The MCBJ technique has been widely used to study
electronic transport in molecular electronics, and many
significant results have been obtained.
31−43
Spring steel
substrates were used for the fabrication of the MCBJ chips.
Electrode structures containing a thin bridge with a constriction
of 30 nm and a nanoscale side-gate electrode were defined by e-
beam lithography (Figure 1b). After the lift-offof a 40 nm thick
Received: March 23, 2013
Revised: May 8, 2013
Published: May 23, 2013
Letter
pubs.acs.org/NanoLett
© 2013 American Chemical Society 2809 dx.doi.org/10.1021/nl401067x |Nano Lett. 2013, 13, 2809−2813
Au film, the isolating layer was locally removed by reactive ion
etching to generate a suspended metal bridge
34,44,45
(see Figure
S1 in the Supporting Information for detailed information
about the chip fabrication).
The fabricated chip was mounted into a homemade three-
point bending apparatus. The two outer posts of the three-
point bending apparatus were fixed, while the third one worked
as a push rod in the Zdirection, as shown in Figure 1c. When
the push rod exerts a bending force on the substrate, the
movement in the Zdirection causes an elongation of the
constriction until the bridge breaks, resulting in two separated
electrodes that act as source and drain electrodes. The distance
between the source and drain electrodes for both the opening
and the closing operation modes was tuned by bending or
relaxing the substrate, respectively (see Figure S3 in the
Supporting Information). The precision of the gap between
two electrodes is determined by the attenuation factor. For our
setup, the gap can be controlled, in principle, with sub-
Angstrom accuracy
44−46
(see Figures S4 and S5 in the
Supporting Information).
The 1,4-benzenedithiol (denoted as BDT), a conjugated
molecule with two thiol termini as binding groups, was
integrated between the two electrodes to form a metal/
molecule/metal junction. For this purpose, a 1 mM ethanol
solution of BDT was prepared in a protective atmosphere in
which the oxygen level was less than 1 ppm. After a self-
assembly period of 10 min on the Au surface, the sample was
thoroughly rinsed with ethanol and dried in a nitrogen stream.
Subsequently, the samples were mounted into the MCBJ. The
gate-effected conductance measurement was performed at
room temperature.
The breaking process of the metal bridge with assembled
molecules on the surface was investigated. A state can be
identified in which two electrodes become bridged by a single
molecule by monitoring the conductance change during the
junction-breaking process (Figure 2). First, the breaking
process of the metal wire is observed, which causes discrete
conductance values of multiples of Go(Go, the conductance
quantum, 7.75 ×10−5Ω−1). The molecule can bind covalently
to both the electrodes of the MCBJ by the two thiol termini of
BDT, and a metal/molecule/metal junction is formed. In
particular, typical plateaus, which correspond to a lock-in state,
were always observed in the molecular junction after the
breaking of the metal bridge. During such a lock-in state, the
conductance of the molecular junction was almost independent
of the displacement of the push rod. Normally, the last plateau
of the current before the junction completely collapses is
attributed to a single metal/molecule/metal junction. More
than 200 metal/molecule/metal junctions were analyzed, and
the single-molecule conductance was determined statistically to
be 1.1 ×10−2Go, which is consistent with previous
reports.
47−49
Compared with the molecular junctions, no
pronounced plateau was observed at values below 1 Goin the
junctions without molecules, as shown in Figure 2.
The conductance of molecular junction at different gate
voltages was subsequently analyzed by a statistical method.
Figure 3 shows the corresponding histograms at different gate
voltages. It can be seen from this figure that the typical peaks
did not obviously shift at different gate voltage. However the
peaks were broadened to a higher conductance value as the gate
voltage increased negatively. The broad peaks can be attributed
to the variation of effective gate coupling (i.e., depending on
the exact molecule position with respect to the side gate
electrode).
Once a metal/molecule/metal junction was constructed, the
source and drain electrodes were frozen, and the current
through two electrodes in the molecular junction (Isd) was
recorded with a fixed bias voltage (12 mV) while sweeping the
gate voltage VG.
The gate-controlled current as a function of gate voltage is
illustrated in Figure 4a. The tunnel current passing through the
molecular junction was weakly dependent on the gate voltage
Figure 1. (a) Schematic illustrating a molecular transistor junction. An
external electric field generated by a side-gate electrode is applied to
the single-molecule junction formed by the MCBJ. Note that the
distance (∼5 nm) from gate to molecular junction for working
molecular transistor junctions is not scaled proportionally in this
schematic. (b) Top view of a scanning electron microscopy (SEM)
image of a microfabricated MCBJ chip consisting of a freestanding
metal bridge with a gate electrode (left SEM image). The push rod
exerts a bending force to bend the substrate and breaks the metal
bridge at the smallest constriction, resulting in two separated
electrodes (right SEM image). (c) Schematic of the MCBJ setup.
The distance between the electrodes for both opening and closing
operation modes can be tuned by bending or relaxing the substrate,
respectively.
Figure 2. Tunneling currents as a function of the push rod
displacement in the open gap period. In the case of the BDT
molecular junction (blue curves), the current jumped to a plateau
(lock-in state) during the open period. In the lock-in state, the current
was almost independent of the gap size, indicating a metal/molecule/
metal junction was formed (cartoon). In contrast, the lock-in state was
absent in the molecule-free junction (gray curve). A bias voltage of 12
mV was applied between the source and drain electrodes. The current
histogram extracted from 200 curves indicates that there is a peak for
the lock-in state in the molecular junction, whereas no such peak exists
in molecule-free junctions.
Nano Letters Letter
dx.doi.org/10.1021/nl401067x |Nano Lett. 2013, 13, 2809−28132810
for bias voltages within ±3 V. However, the current and
corresponding conductance value increased rapidly as the gate
voltage became more negative. The effective gate effect is
attributed to the special geometry of the electrodes. Since a pair
of taped source−drain electrodes will be formed naturally
during the break of metal bridge in our experiments. An
independent simulation demonstrated that nonuniform tapered
electrodes can yield 3 orders of magnitude improvement in gate
coupling compared to the case of the uniform (rectangular)
bulk electrodes.
50
It should be noted that the molecular
junctions became unstable, even collapsing, when the applied
gate voltage increased beyond ±10 V. Therefore, we could only
record the current at either high positive or high negative bias
voltages in one measurement. In cases in which the metal/
molecule/metal junction is broken, the molecular junctions can
be rebuild (with ∼50% yield) by decreasing the gap size
between two electrodes by 0.2−0.4 nm followed by with-
drawing the electrodes to the same gap size as before. Then, the
gate-modulated current can be observed again after the single-
molecule junction is reformed. In this sense, there are certainly
advantages in MCBJ-type molecular transistor junctions
compared to other kind of junctions, such as electromigration
fabricated nanogap junctions.
Three types of chips were investigated, and the gate electrode
was located in various positions relative to the axis between the
source and drain electrodes (see Figure S2 in the Supporting
Information). More than 100 individual molecular junctions
within the three types of chips were examined. We found that
only those chips for which the gate electrode was close enough
to the molecular junction (the perpendicular distance from the
gate electrode to the tip axis was less than 10 nm) displayed
typical FET behavior. For these chips, 12 out of 67 junctions
showed typical FET behavior (∼18% yield), as shown in Figure
4a and Table S1. Please note that the gate efficiency is sensitive
to the exact position of molecules with respect to the side gate.
For those molecules located far away from the side facing the
gate, the screening of the gold electrodes will be dominant
which results in the failure of FET behavior. Even for those
junctions that exhibited FET behavior, the gated tunneling
current spread over a wide range especially under the negative
gate voltage, which agrees with the observation shown in Figure
3. The variation of the gate coupling from junction to junction
becomes clearer when we plot the modulation ratio of each
junction,asshowninFigure S10 in the Supporting
Information. Interestingly, for all the FET behavior junctions,
the increase of the current at negative gate voltages is larger
than for positive gate voltages, resulting in an asymmetric ISD
versus VG.
In contrast to the molecular junction, no gate effect current
was observed when the molecules were absent (called as
molecule-free junction) even at different gap sizes (Figure 3b).
Four different types of molecule-free junctions were tested, in
which the relative positions of the source and drain electrodes
were dramatically changed: the first configuration involved
direct point-to-point contact of bare Au atoms between the
source and drain electrode in which the corresponding
conductance was observed to be 1 Go, and the remaining
configurations involved changing the relative distance between
the source and drain electrode to be smaller than, equal to, and
larger than 1 nm (approximate BDT molecule length). Note
that the current was almost independent of the gate voltage in
all four types of molecule-free junctions (see Figure S6 in the
Supporting Information). By comparing the results of Figures 4
and Figure S6, it can be concluded that the observed variation
of the current was caused by the modulation of the electronic
structure of the molecules by the applied gate voltage. It should
be noted that leakage current level is important for molecular
transistors. We measured the gate−source current (IGS) and
gate−drain (IGD) current at different gate voltages. We found
that the leakage current was almost 3 orders of magnitude
smaller than the current passing through the molecular junction
and it was almost close to the limitation of the measurement
setup (see Figure S7 in the Supporting Information).
Figure 3. Conductance histograms at different gate voltages. The
bottom histogram is the reference histogram from molecule-free
junctions and other histograms are those for molecular junctions at
different gate voltages of 0, −4, and −8 V. Each histogram was built
from 200 conductance curves obtained during the open cycle process.
The arrow in each histogram indicates the conductance peak. The
source−drain voltage was fixed at 12 mV.
Figure 4. Source−drain current (ISD) versus gate voltage (VG) curves
for the Au-BDT-Au molecular and molecule free three-terminal
junction. (a) The current was modulated by the gate voltage in 12
different molecular junctions. (b) No gate-controlled current was
observed in the molecule-free junction (red and gray curves). When
the gap size between the source and drain electrodes was reduced by
∼0.1 nm, the tunneling current increased slightly, but no gate-effected
current was observed (green and blue curves). The inset schematics
illustrate the three-terminal junction (a) with or (b) without BDT
bridged between the source and drain electrodes.
Nano Letters Letter
dx.doi.org/10.1021/nl401067x |Nano Lett. 2013, 13, 2809−28132811
To further clarify how the electronic transport was affected
by the gate electrode, first-principle calculations based on
density functional theory (DFT) with the Dmol3 simulation
package were performed to explore the electronic transport
properties.
6−9,51
Before this step, determination of the value of
the electric field strength and potential around the molecular
junction is necessary. Here, the simulation of the electrical field
and the potential distribution was carried out with COMSOL
Multiphysics using an electrostatic physics model. The
geometry of the three electrodes was assumed to be a cone-
like body with a sphere-like tip. The gap size between the
source and drain electrodes was 1.1 nm, the molecule length.
The perpendicular distance between the gate electrode and the
source−drain axis was set to 5 nm, while different voltages were
applied to the gate electrode. From the simulation, we found
that the electrical field strength between the source and drain
electrodes with a gate bias is an order of magnitude larger than
that for the case when the gate electrode was absent (see Figure
S8 in the Supporting Information). Our simulation also showed
that the field distribution as well as the potential distribution
around the gate electrode was inhomogeneous and that it
strongly depended on the location of the gate electrode with
respect to the source and drain electrodes, as shown in Figure
5a and Figure S9. We fixed the position of the gate electrode in
the center between the source and drain electrodes and
extracted the average value of the electric field from the
simulation. This value was used as input for the following DFT
calculation.
The local density approximation (LDA) with PWC
(Perdew−Wang approaches for the correlation functional)
was used to describe the exchange-correction energy functional.
The numerical basis set of DNP (double numerical plus
polarization) was adopted. In our calculations, the cutoff
energy, which determines the number of plane-wave basis
functions, is chosen to be 380 eV. In our model for DFT
calculation, the molecule contacted the source and drain
electrodes through S−Au bonding with two different
configurations (in-plane and out-of-plane configurations) to
form an Au/1,4-BDT/Au junction. The DFT calculations
showed that the wave function density of the molecular orbital
varied tremendously at different gate voltages in both
configurations. Figure 5b shows the in-plane configuration
case in which the bonding Au atom is located parallel to the
benzene ring plane. When the electric field was gradually
increased, the delocalization of the HOMO (highest occupied
molecular orbital) increased as well. The HOMO for VG=−9
V became more delocalized than that for VG= 0 V. For the out-
of-plane configuration in which the bonding Au atom deviates
from the benzene ring plane, a significant variation of the
molecular orbitals was predicted when an electrical field was
applied (see Figure S11 in the Supporting Information). The
DFT calculation results indicated that the electronic structure
of the molecular junction was dramatically changed at a higher
electric field, which is qualitatively consistent with our
experimental data.
In summary, we studied three-terminal transistor single-
molecular junctions formed by the mechanically controlled
break junction (MCBJ) technique with a noncontact side gate.
The distance between the source and drain electrodes was
precisely controlled to achieve a single-molecule junction, while
an in-plane gate-electrode was coupled into the molecular
junction to shift the molecular energy levels. Our experimental
results combined with the DFT calculations demonstrated that
direct modulation of the electronic transport characteristics
through a single-molecule junction with side gating was
possible, yielding a new design platform for mechanically stable
single-molecule transistor junctions.
■ASSOCIATED CONTENT
*
SSupporting Information
Chip fabrication, experimental setup, calculation of the
attention factor, characterization of molecule-free junctions,
simulation of the electrical field, and three terminal molecular
transistor junctions. This material is available free of charge via
the Internet at http://pubs.acs.org.
■AUTHOR INFORMATION
Corresponding Author
*E-mail: tlee@snu.ac.kr (T.L.); dirk.mayer@fz-juelich.de
(D.M.).
Notes
The authors declare no competing financial interest.
■ACKNOWLEDGMENTS
The authors acknowledge the financial support from the
National Creative Research Laboratory program (Grant No.
Figure 5. (a) Simulation results of the electrical field distribution with
the gate voltage of −8 V and the source−drain voltage of 12 mV. The
source and drain electrodes were separated by 1.1 nm, and the
perpendicular distance between the gate and the source−drain axis was
set to 5 nm. The arrows indicate the electrical field vector directions.
Two color bars were used: one (left color bar) shows the electrical
field strength distribution on three cross sections, and the other (right
gray bar) presents the potential distribution on the electrode surface.
(b) The gate voltage dependence of the spatial distribution of the
HOMO and LUMO for an in-plane configuration in which the
bonding Au atom is located parallel to the benzene ring plane. Blue
and yellow regions indicate regions with positive and negative wave
function amplitudes, respectively. With increasing gate voltages, both
the HOMO and LUMO tend to become delocalized.
Nano Letters Letter
dx.doi.org/10.1021/nl401067x |Nano Lett. 2013, 13, 2809−28132812
2012026372) and the National Core Research Center (Grant
No. R15-2008-006-03002-0) through the National Research
Foundation of Korea (NRF) funded by the Korean Ministry of
Science, ICT & Future Planning. D.X. and Y.C. thank the
fundamental Research Funds for National University and the
National Basic Research Program (Grant No. 2011CB710606)
of China.
■ABBREVIATIONS
MCBJ, mechanically controllable break junction; FET, field-
effect transistor; BDT, benzenedithiol; HOMO, highest
occupied molecular orbital; LUMO, lowest unoccupied
molecular orbital; DFT, density functional theory
■REFERENCES
(1) Cuevas, J. C; Scheer, E. Molecular Electronics: An Introduction to
Theory and Experiment; World Scientific Publishing Co. Pte. Ltd.:
Singapore, 2010; pp 12−45.
(2) Yoon, H. J.; et al. Angew. Chem., Int. Ed. 2012,124, 4736−439.
(3) Reus, W. F.; Thuo, M. M.; Shapiro, N. D.; Nijhuis, C. A.;
Whitesides, G. M. ACS Nano 2012,6, 4806−4822.
(4) Kasper, M. P.; Bjørnholm, T. Nat. Nanotechnol. 2009,4, 551−
556.
(5) Joachim, C.; Gimzewski, J. K.; Aviram, A. Nature 2000,408,
541−548.
(6) Liang, Y. Y.; Chen, H.; Mizuseki, H.; Kawazoe, Y. J. Chem. Phys.
2011,134, 144113.
(7) Shimazaki, T.; Yamashita, K. Nanotechnology 2007,18, 424012.
(8) Damle, P.; Rakshit, T.; Paulsson, M.; Datta, S. Nanotechnology
2002,1, 145−153.
(9) Di Ventra, M.; Pantelides, S. T.; Lang, N. D. Appl. Phys. Lett.
2000,76, 3448.
(10) Osorio, E. A.; Bjørnholm, T.; Lehn, J.-M.; Ruben, M.; van der
Zant, H. S. J. Phys.: Condens. Matter 2008,20, 374121.
(11) Xu, Y.; et al. Appl. Phys. Lett. 2011,99, 043304.
(12) Ke, S. H.; Baranger, H. U.; Yang, W. Phys. Rev. B 2005,71,
113401.
(13) Xu, B. Q.; Xiao, X. Y.; Yang, X. M.; Zhang, L.; Tao, N. J. Am.
Chem. Soc. 2005,127, 2386−2387.
(14) Tao, N. J. J. Mater. Chem. 2005,15, 3260−3263.
(15) Grunder, S.; Huber, R.; Wu, S.; Schönenberger, C.; Calame, M.;
Mayor, M. Chimia 2010,64, 140−144.
(16) White, H. S.; Kittlesen, G. P.; Wrighton, M. S. J. Am. Chem. Soc.
1984,106, 5375−5377.
(17) Keane, Z. K.; Ciszek, J. W.; Tour, J. M.; Natelson, D. Nano Lett.
2006,6, 1518−1521.
(18) Ghosh, A. W.; Rakshit, T.; Datta, S. Nano Lett. 2004,4, 565−
568.
(19) Park, H.; Park, J.; Lim, A. K. L.; Anderson, E. H.; Alivisatos, A.
P.; McEuen, P. L. Nature 2000,407,57−60.
(20) Martin, C. A.; van Ruitenbeek, J. M.; van der Zant, H. S. J.
Nanotechnology 2010,21, 265201.
(21) Champagne, A. R.; Pasupathy, A. N.; Ralph, D. C. Nano Lett.
2005,5, 305−308.
(22) Liang, W.; Shores, M. P.; Bockrath, M.; Long, J. R.; Park, H.
Nature 2002,417, 725−729.
(23) Song, H.; et al. Nature 2009,462, 1039−1043.
(24) Piva, P. G.; et al. Nature 2005,435, 658−661.
(25) Kubatkln, S.; et al. Nature 2003,425, 698−701.
(26) Visconti, P.; Torre, A. D.; Maruccio, G.; D’Amone, E.; Bramanti,
A.; Cingolani, R.; Rinaldi, R. Nanotechnology 2004,14, 807−811.
(27) Osorio, E. A.; et al. Nano Lett. 2007,7, 3336−3342.
(28) Osorio, E. A.; Ruben, M.; Seldenthuis, J. S.; Lehn, J.; van der
Zant, H. S. J. Small 2010,6, 174−178.
(29) Mangin, A.; Anthore, A.; Della Rocca, M. L.; Boulat, E.; Lafarge.,
P. J. Appl. Phys. 2009,105, 014313.
(30) Martin, C. A.; Smit, R. H. M.; van der Zant, H. S. J.; van
Ruitenbeek, J. M. Nano Lett. 2009,9, 2940−2945.
(31) Lörtscher, E.; Gotsmann, B.; Lee, Y.; Yu, L.; Rettner, C.; Riel, H.
ACS Nano 2012,6, 4931−4939.
(32) Wheeler, P. J.; Russom, J. N.; Evans, K.; King, N. S.; Natelson,
D. Nano Lett. 2010,10, 1287−1292.
(33) Kim, Y.; et al. Nano Lett. 2012,12, 3736−3742.
(34) Wu, S.; et al. Nat. Nanotechnol. 2008,3, 569−574.
(35) Lörtscher, E.; Weber, H. B.; Riel, H. Phys. Rev. Lett. 2007,98,
176807.
(36) Parks, J. J.; et al. Science 2010,328, 1370−1373.
(37) Diez-Perez, I.; et al. Nat. Nanotechnol. 2011,6, 226−231.
(38) Reed, M. A.; et al. Science 1997,278, 252−254.
(39) Smit, R. H. M.; et al. Nature 2002,419, 906−909.
(40) Kim, Y.; Song, H.; Strigl, F.; Pernau, H.-F.; Lee, T.; Scheer, E.
Phys. Rev. Lett. 2011,106, 196804.
(41) Ballmann, S.; et al. Phys. Rev. Lett. 2012,109, 056801.
(42) Tsutsui, M.; Taniguchi, M.; Kawai, T. Nano Lett. 2009,9, 1659−
1662.
(43) Tsutsui, M.; Taniguchi, M. J. Appl. Phys. 2013,113, 084301.
(44) Xiang, D.; et al. Chem.Eur. J. 2011,47, 13166.
(45) Xiang, D.; Zhang, Y.; Pyatkov, F.; Offenhäusser, A.; Mayer, D.
Chem. Commun. 2011,47, 4760−4762.
(46) Zhou, C.; Muller, C. J.; Deshpande, M. R.; Sleight, J. W.; Reed,
M. A. Appl. Phys. Lett. 1995,67, 1160−1162.
(47) Tian, J.; et al. Nanotechnology 2010,21, 274012.
(48) Xiao, X.; Xu, B.; Tao, N. J. Nano Lett. 2004,4, 267−271.
(49) Kim, Y.; Pietsch, T.; Erbe, A.; Belzig, W.; Scheer, E. Nano Lett.
2011,11, 3734−3738.
(50) Datta, S. S.; Strachan, D. R.; Charlie Johnson, A. T. Phys. Rev. B
2009,79, 205404.
(51) Mishchenko, A.; et al. J. Am. Chem. Soc. 2011,133, 184−187.
Nano Letters Letter
dx.doi.org/10.1021/nl401067x |Nano Lett. 2013, 13, 2809−28132813