Tuning Conductance in π-σ-π Single-Molecule Wires

Article abstract of DOI:10.1021/jacs.6b04394

While the single-molecule conductance properties of π-conjugated and σ-conjugated systems have been well-studied, little is known regarding the conductance properties of mixed σ-π backbone wires and the factors that control their transport properties. Her

Full text of DOI:10.1021/jacs.6b04394

Article
pubs.acs.org/JACS
Tuning Conductance in π−σ−π Single-Molecule Wires
Timothy A. Su,†,§ Haixing Li,‡,§ Rebekka S. Klausen,†,⊥ Jonathan R. Widawsky,‡ Arunabh Batra,‡
Michael L. Steigerwald,*,† Latha Venkataraman,*,†,‡ and Colin Nuckolls*,†
†Department of Chemistry, Columbia University, New York, New York 10027, United States
‡Department of Physics and Applied Math, Columbia University, New York, New York 10027, United States
S
* Supporting Information
ABSTRACT: While the single-molecule conductance proper-
ties of π-conjugated and σ-conjugated systems have been well-
studied, little is known regarding the conductance properties
of mixed σ−π backbone wires and the factors that control their
transport properties. Here we utilize a scanning tunneling
microscope-based break-junction technique to study a series of
molecular wires with π−σ−π backbone structures, where the
π-moiety is an electrode-binding thioanisole ring and the σ-
moiety is a triatomic α−β−α chain composed of C, Si, or Ge
atoms. We find that the sequence and composition of group 14
atoms in the α−β−α chain dictates whether electronic
communication between the aryl rings is enhanced or
suppressed. Placing heavy atoms at the α-position decreases
conductance, whereas placing them at the β-position increases conductance: for example, the C−Ge−C sequence is over 20
times more conductive than the Ge−C−Ge sequence. Density functional theory calculations reveal that these conductance
trends arise from periodic trends (i.e., atomic size, polarizability, and electronegativity) that differ from C to Si to Ge. The
periodic trends that control molecular conductance here are the same ones that give rise to the α and β silicon effects from
physical organic chemistry. These findings outline a new molecular design concept for tuning conductance in single-molecule
electrical devices.
INTRODUCTION
■
In this study we systematically vary the strength of coupling
between the π and σ backbone components of a molecular wire
to raise or lower the conductance of a single-molecule electrical
device. The miniaturization of electronic devices and the related
interest in creating molecular-scale devices have motivated the
desire to understand the charge transport characteristics of
molecules. There have been many studies that have elucidated
how the strength of σ-conjugation or π-conjugation within a
molecular wire backbone can be tuned to control single-
molecule conductance.1−6 Ottosson and coworkers have
studied σ−π conjugation computationally and spectroscopi-
cally.7 However, there has yet to be an experimental description
of how the coupling between σ and π backbone components
Figure 1. Chemical structures of the π−σ−π molecular wires.
influences molecular conductance. Though organic electronic
materials commonly feature π−σ substructures,8−12 it is not
see two counterbalancing trends in the data. Conductance
decreases as the α atom is varied from C to Si to Ge; however,
conductance increases as the composition of the β atom
changes from C to Si to Ge. Mechanistic chemical analysis and
density functional theory (DFT) calculations suggest that these
opposing conductance trends arise from differences in how the
well understood how the nature of the π−σ interaction affects
charge transport.
For this study we create a test bed of molecular wires with
π−σ−π backbone structures, where the π-substructure is an
electrode-binding p-thioanisole ring and the σ-moiety is a
triatomic α−β−α sequence of catenated group 14 atoms (α and
β = C, Si, or Ge) (Figure 1). We systematically vary the
composition and sequence of atoms in the σ-chain to
understand how such alterations influence conductance. We
Received: May 6, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.6b04394
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
atomic composition and sequence within the α−β−α chain
influence the coupling between the σ-chain and the thioanisole
ring. These effects are the single-molecule conductance
equivalent of the α silicon and β silicon hyperconjugation
that is an essential component of physical organic chemistry.
Hyperconjugation provides an intellectual bridge between
physical organic chemistry and the conductance trends we
observe here.
RESULTS AND DISCUSSION
■
Synthesis. We synthesize CSiC and CGeC from the double
addition of 4-(methylthio)benzylzinc chloride13 to dimethyl-
dichlorosilane or dimethyldichlorogermane (Scheme 1). We
Scheme 1. Synthesis of the α−β−α Wires with p-Thioanisole
End Groups
Figure 2. Logarithmically-binned 1D conductance histograms of
GeCGe, SiCSi, CCC, CSiC, and CGeC. The 1D histograms compile
over 10 000 measurement traces without data selection for each
molecule. Conductance peak maxima (marked by arrows) are obtained
from Gaussian fits to the conductance data. The counts/trace
intensities of these conductance peaks are normalized by factors of
0.88, 1.15, 1.00, 0.70, and 0.82 respectively to highlight conductance
trends. The unscaled conductance histograms of CCC, SiSiSi, and
GeGeGe are shown in the inset; these molecules have approximately
the same conductance. We observe a small conductance peak at ∼10−3
G0 that we have discussed previously;19,20 it is due to an interaction
between the Au electrodes and α−β σ-bond (see SI for additional
comments).
prepared SiCSi, GeCGe, CCC, SiSiSi, and GeGeGe by
reacting 4-(methylthio)phenyllithium with the appropriate
halide-terminated σ-chain.2,14−16 The Supporting Information
(SI) contains further details for the preparation and character-
ization of these molecular wires.
Table 1. Conductance Values from STM-BJ Measurements,
DFT-Calculated Orbital Energies, and DFT-Calculated
Molecular Lengths
STM-BJ Measurement Details. We measure the con-
ductance of these molecules with a scanning tunneling
microscope-based break-junction (STM-BJ) technique.1 In
this technique, Au−molecule−Au junctions are measured by
repeatedly breaking and forming point contacts between the Au
tip and substrate in a solution of the target molecule in 1,2,4-
trichlorobenzene under ambient conditions. After the Au−Au
point contact is broken, the aurophilic thiomethyl groups on
the molecule bind the electrodes to form a Au−molecule−Au
junction. Conductance is measured across the junction as a
function of tip−substrate displacement, and the resulting
measurement traces reveal molecule-dependent plateaus
signifying junction formation with conductance values below
G0 (2e2/h), the quantum of conductance describing a single
Au−Au atomic contact.17 The junction breaks once the
distance between the electrodes becomes too large for the
molecule to span. We form and break thousands of molecular
junctions and analyze all measured traces using logarithmically-
binned one-dimensional (1D) histograms18 that provide a
distribution of measured conductance values for all traces.
Single-Molecule Conductance Data Analysis. The 1D
conductance histograms for the α−β−α wires are shown in
Figure 2. Table 1 lists the conductance peak values for the
molecules measured here. The purple rows in Table 1
demonstrate that varying the α-position from C to Si to Ge
(while fixing the β-atom as C) results in a conductance trend
that follows GeCGe < SiCSi < CCC. Conversely, the blue rows
in Table 1 show that varying the β-position from C to Si to Ge
(while fixing the α-atom as C) increases conductance such that
CCC < CSiC < CGeC. To highlight these competing trends,
the conductance of CGeC is over 20 times that of GeCGe.
These opposing trends negate each other when both the α- and
a
Conductance peak value obtained from fitting a Gaussian curve to the
b
corresponding 1D conductance histogram. Calculated at the B3LYP/
cc-pVTZ level of DFT. Difference in energy between the anti-
symmetric and symmetric combinations of the S pπ lone pairs (the
c
d
HOMO and HOMO−1 respectively). Through-space distance
between the aryl carbons that terminate the α−β−α chain in the
DFT optimized structure. The length relative to CCC is given in
parentheses.
β-positions are varied (gray rows in Table 1): the CCC, SiSiSi,
and GeGeGe π−σ−π wires give approximately the same
conductance value.
The conductance trends we observe are related to the
particular composition and sequence of group 14 atoms in the
α−β−α chain that dictate the extent to which the aryl rings
interact with one another through the bridging σ-chain.
Changing the composition of the α and β atoms modulates
three major electronic factors within the molecule (Figure 3A−
C).
B
DOI: 10.1021/jacs.6b04394
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
are the HOMO and HOMO−1 that comprise the antisym-
metric (S1 − S2) and symmetric (S1 + S2) combinations of the
two S pπ lone pairs, respectively.
Conductance should increase as the S1 − S2 orbital rises in
energy and is brought into closer alignment with the Fermi
energy (EF). Conductance should also increase as the
communication between the S pπ lone pairs becomes stronger.
We can assess the coupling between the lone pairs by
comparing the difference in energy between the S1 − S2 and
S1 + S2 molecular orbitals: a small splitting indicates that the
lone pairs are mostly independent and noninteracting, whereas
a large splitting indicates that the lone pairs are strongly
coupled.19 We can therefore estimate the conductive impact of
tuning the α and β atom composition by evaluating two
parameters from our DFT calculations: the S1 − S2 (HOMO)
energy and the energy difference between the antisymmetric
and symmetric combinations of the S pπ orbitals (HOMO/
HOMO−1 splitting). We list the values of the energies for
these two orbitals in Table 1.
The DFT trends are consistent with the chemical analysis
described above. The interaction of the α−β σ-bonds with the
thioanisole π-orbitals is visually apparent in the HOMO
surfaces in Figure 3D. From CCC to SiCSi to GeCGe, the
HOMO becomes lower in energy and the splitting magnitude
decreases because the electron density of the α−β bond is
removed further away from the thioanisole rings. The HOMO
rises in energy, and the HOMO/HOMO−1 splitting becomes
larger going from CCC to CSiC to CGeC due to the C−
Si(Ge) bond polarity and the increasing donor power of the β
atom. The HOMO value of CCC is similar to that of SiSiSi and
GeGeGe: though the C−C σ-bond is less donating than the
Si−Si or Ge−Ge bond, it still strongly destabilizes the
thioanisole π-orbitals because of the shorter Ar−C bond. The
smaller HOMO/HOMO−1 splitting calculated for CCC is
consistent with the weaker conjugation of C−C σ-bonds
compared to Si−Si and Ge−Ge σ-bonds.14
Figure 3. (A−C) Atomic size, polarizability, and electronegativity
dictate the strength of interaction between the thioanisole rings
through the σ-chain and, thereby, the conductance of the wire. (D)
The influence of three periodic factors on the nature of the π−σ
interaction is apparent in the HOMO surfaces (isovalue = 0.05) for
GeCGe, CGeC, and GeGeGe. These structures were calculated at the
B3LYP/cc-PVTZ level.
The first factor is specific to the α atom (Figure 3A): the Ar−
C bond length (1.51 Å) is substantially shorter than the Ar−Si
(1.89 Å) and Ar−Ge (1.98 Å) bond lengths.21 As a result, the
aryl ring and σ-chain interact more strongly when α = C than
when α = Si or Ge because the ring is brought into closer
proximity to the central σ-chain. The second factor (Figure
3B), atomic polarizability, follows the trend C ≪ Si < Ge.22
Atomic polarizability dictates the electron donating ability and
diffuseness of the α−β σ-bond and thereby influences the
ability of the σ-bond to electronically interact with the
thioanisole π-orbitals. The third factor (Figure 3C) arises
from the polarity of the α−β bond. The electron density of a
C−Si or a C−Ge σ-bond is directed toward the C atom rather
than the Si or Ge atom because C is much more electronegative
than Si or Ge.23 The α−β bond polarity is an important factor
because it dictates whether the σ-bond electron density is
oriented toward the aryl ring or away from it.
Based on these three factors, the CH2−GeMe2 σ-bonds in
CGeC interact more strongly with the thioanisole π-orbitals
than they do in GeCGe. The shorter Ar−C bond brings the
aryl ring and σ-chain into closer contact for CGeC than the
Ar−Ge bond does for GeCGe. Furthermore, the α−β bond
density is polarized toward the aryl rings in CGeC, but away
from the aryl rings in GeCGe. The characteristics of GeGeGe
balance the three factors described above: the Ar−Ge bond is
relatively long, but the Ge−Ge α−β bond is more polarizable
and more covalent (the electron density is spread more evenly
between the two atom centers) than in the Ge−C bonds.
Density Functional Theory Analysis. DFT calculations
provide a numerical assessment of how these three chemical
features influence the molecule’s electronic properties (see the
Supporting Information for notes and details of the
calculations).24 Conductance across a molecular junction is
facilitated by two primary molecular factors: (1) close energetic
alignment between the conducting orbital of the molecule and
the Fermi level (EF) of the electrode and (2) strong coupling
between the two distal lone pair orbitals that link the molecule
to the electrodes.25−27 As we have shown previously, it is
particularly informative to investigate the highest energy
molecular orbitals that describe the lone pair orbitals.19,27 In
the case of the molecules studied here, these molecular orbitals
Analogy to α and β Silicon Hyperconjugation Effects.
The periodic trends that control conductance here are the same
ones that form the basis of the α and β silicon effects, which are
fundamental principles of hyperconjugation from physical
organic chemistry.28,29 This is shown schematically in Figure
4A. These effects describe the electronic interaction between a
π orbital and an adjacent C−Si σ-bond based on whether the Si
atom is either one atom (α) or two atoms (β) away from the π
orbital. For example, carbocations are weakly stabilized by α-Si
atoms30 but are strongly stabilized by β-Si atoms.31 The
conductance trends we observe here are thus an electrical
manifestation of the α and β silicon effects.
A well-established metric for gauging the strength of
+
hyperconjugation is the Hammett parameter, σp , which
indexes the extent to which a positive charge on a benzene
ring is stabilized by a particular substituent.32 Negative values
+
for σp correspond to stronger electron donation from the
substituent to the phenyl ring. Previously, we demonstrated a
correlation between the logarithmic conductance ratio and
Hammett parameter for electron-donating and -withdrawing
substituents on 1,4-diaminobenzene.33 In Figure 4B we apply a
similar approach in comparing log(Gα−β−α/GCCC) against the
+
σp value for structurally similar molecules from ref 34, given
here in parentheses: GeCGe (PhGeMe3), SiCSi (PhSiMe3),
CCC (PhCH2Me), CSiC (PhCH2SiMe3), and CGeC
(PhCH2GeMe3). Such analysis reveals an approximately linear
agreement.
C
DOI: 10.1021/jacs.6b04394
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Present Address
⊥Department of Chemistry, The Johns Hopkins University,
Baltimore, MD 21218, USA.
Author Contributions
§T.A.S. and H.L. contributed equally.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We would like to thank Prof. James L. Leighton, Dr. Raul
́
Hernan
́
dez San
́
chez, and Dr. Daniel W. Paley for insightful
Figure 4. (A) Common representations of the β silicon effect in
physical organic chemistry. The top panel describes the resonance
pictures of the α- and β-stabilizing silicon effects. The bottom panel
juxtaposes the β-stabilizing effect with the β-destabilizing effect; the β-
destabilizing picture is relevant to the HOMO energy trends observed
in the π−σ−π wires. (B) The logarithm of the π−σ−π wire’s
conductance relative to that of CCC is plotted against the Hammett
discussions as well as Dr. Brandon Fowler and Dr. Yasuhiro
Itagaki for mass spectrometry characterization. T.A.S. is
supported by an NSF Graduate Research Fellowship under
Grant No. 11-44155. H.L. is supported by the Semiconductor
Research Corporation and New York CAIST program. These
studies are supported by the NSF under Grant No. CHE-
1404922.
+
σp coefficients (relative to benzene) obtained for molecules from ref
34 that are similar in structure to the ones studied here.
REFERENCES
(1) Xu, B.; Tao, N. J. Science 2003, 301, 1221.
(2) Klausen, R. S.; Widawsky, J. R.; Steigerwald, M. L.;
Venkataraman, L.; Nuckolls, C. J. Am. Chem. Soc. 2012, 134, 4541.
(3) Su, T. A.; Li, H.; Steigerwald, M. L.; Venkataraman, L.; Nuckolls,
C. Nat. Chem. 2015, 7, 215.
■
Despite the fact that the Hammett experiments do not fully
replicate the break-junction environment, this simple approach
correlates quite well with the observed conductance trends. The
+
general correlation we find between conductance and σp
(4) Mayor, M.; Weber, H. B.; Reichert, J.; Elbing, M.; von Hanisch,
̈
supports the idea that tailoring the strength of σ−π interactions
is useful not only for altering chemical reaction rates but also
for tuning charge transport in single-molecule junctions.
C.; Beckmann, D.; Fischer, M. Angew. Chem., Int. Ed. 2003, 42, 5834.
(5) He, J.; Chen, F.; Li, J.; Sankey, O. F.; Terazono, Y.; Herrero, C.;
Gust, D.; Moore, T. A.; Moore, A. L.; Lindsay, S. M. J. Am. Chem. Soc.
2005, 127, 1384.
(6) Venkataraman, L.; Klare, J. E.; Nuckolls, C.; Hybertsen, M. S.;
Steigerwald, M. L. Nature 2006, 442, 904.
CONCLUSIONS
■
The observed conductance trends, chemical analysis, and DFT
calculations reveal two new design rules for π−σ−π wires that
ultimately arise from periodic trends such as atomic size,
polarizability, and electronegativity: (1) placing heavy atoms
(Si, Ge) in the α-position of the σ-chain decreases the
conductance of the wire by diminishing the interaction between
the thioanisole rings, and (2) placing heavy atoms in the β-
position increases junction conductance by providing a stronger
electronic pathway for communication between the aryl rings.
These design principles can extend to bulk electronic materials
as well: one possible opportunity is to simultaneously solubilize
and tune the electronics of π-conjugated polymers by replacing
the β-carbon atom of an alkyl solubilizing chain with a silicon or
germanium atom. More broadly, our results demonstrate that
the principles of physical organic chemistry are an untapped
vein27 for designing molecules and tuning charge transport for
single-molecule electronics.
(7) (a) Emanuelsson, R.; Lofas, H.; Zhu, J.; Ahuja, R.; Grigoriev, A.;
̈
̊
Ottosson, H. J. Phys. Chem. C 2014, 118, 5637. (b) Emanuelsson, R.;
Denisova, A. V.; Baumgartner, J.; Ottosson, H. Organometallics 2014,
33, 2997. (c) Goransson, E.; Emanuelsson, R.; Jorner, K.; Markle, T.
̈
F.; Hammarstrom, L.; Ottosson, H. Chem. Sci. 2013, 4, 3522.
̈
(8) Miller, R. D.; Michl, J. Chem. Rev. 1989, 89, 1359.
(9) Chicart, P.; Corriu, R. J. P.; Moreau, J. J. E.; Garnier, F.; Yassar, A.
Chem. Mater. 1991, 3, 8.
(10) Usta, H.; Lu, G.; Facchetti, A.; Marks, T. J. J. Am. Chem. Soc.
2006, 128, 9034.
(11) Sun, Y.; Welch, G. C.; Leong, W. L.; Takacs, C. J.; Bazan, G. C.;
Heeger, A. J. Nat. Mater. 2011, 11, 44.
(12) Surampudi, S.; Yeh, M.-L.; Siegler, M. A.; Hardigree, J. F. M.;
Kasl, T. A.; Katz, H. E.; Klausen, R. S. Chem. Sci. 2015, 6, 1905.
(13) Metzger, A.; Piller, F. M.; Knochel, P. Chem. Commun. 2008,
5824.
(14) Su, T. A.; Li, H.; Zhang, V.; Neupane, M.; Batra, A.; Klausen, R.
S.; Kumar, B.; Steigerwald, M. L.; Venkataraman, L.; Nuckolls, C. J.
Am. Chem. Soc. 2015, 137, 12400.
(15) Barrau, J.; Hamida, N. B.; Satge, J. J. Organomet. Chem. 1985,
282, 315.
(16) Barton, T. J.; Hoekman, S. K. J. Am. Chem. Soc. 1980, 102, 1584.
(17) Agraït, N.; Rodrigo, J.; Vieira, S. Phys. Rev. B: Condens. Matter
Mater. Phys. 1993, 47, 12345.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.6b04394.
■
S
Additional comments, synthetic procedures, character-
ization of compounds, STM-BJ measurement details, and
computational details (PDF)
(18) Gonzalez, M. T.; Wu, S.; Huber, R.; van der Molen, S. J.;
Schonenberger, C.; Calame, M. Nano Lett. 2006, 6, 2238.
̈
(19) Klausen, R. S.; Widawsky, J. R.; Su, T. A.; Li, H.; Chen, Q.;
Steigerwald, M. L.; Venkataraman, L.; Nuckolls, C. Chem. Sci. 2014, 5,
1561.
AUTHOR INFORMATION
■
(20) Su, T. A.; Widawsky, J. R.; Li, H.; Klausen, R. S.; Leighton, J. L.;
Steigerwald, M. L.; Venkataraman, L.; Nuckolls, C. J. Am. Chem. Soc.
2013, 135, 18331.
(21) The calculated Ar−α bond length is essentially the same
regardless of β in the molecules studied here. DFT optimizations
Corresponding Authors
*mls2064@columbia.edu
*lv2117@columbia.edu
*cn37@columbia.edu
D
DOI: 10.1021/jacs.6b04394
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
performed with a B3LYP functional and a 6-31G** basis set for light
atoms or an LACVP** basis set for heavy atoms.
(22) Miller, T. M. In CRC Handbook of Chemistry and Physics, 96th
ed.; Haynes, W. M., Ed.; CRC Press: Boca Raton, FL, 2015; pp 187−
196.
(23) Allen, L. C. J. Am. Chem. Soc. 1989, 111, 9003.
(24) Jaguar, version 8.3; Schrodinger, Inc.: New York, NY, 2014.
(25) Quek, S. Y.; Khoo, K. H. Acc. Chem. Res. 2014, 47, 3250.
́
(26) Leary, E.; La Rosa, A.; Gonzalez, M. T.; Rubio-Bollinger, G.;
Agraït, N.; Martín, N. Chem. Soc. Rev. 2015, 44, 920.
(27) Su, T. A.; Neupane, M.; Steigerwald, M. L.; Venkataraman, L.;
Nuckolls, C. Nat. Rev. Mater. 2016, 1, 16002.
(28) Whitmore, F. C.; Sommer, L. H. J. Am. Chem. Soc. 1946, 68,
481.
(29) Sommer, L. H.; Whitmore, F. C. J. Am. Chem. Soc. 1946, 68,
485.
(30) It is worth comparing the HOMO energy (B3LYP/cc-pVTZ) of
SiCSi (−5.74 eV) with 4-methylthioanisole (−5.64 eV) because it
demonstrates that C−H bonds energetically destabilize the thioanisole
even more so than Si−C bonds do. The weaker interaction between π
orbitals and Si−C bonds relative to C−H bonds is a common feature
of the α silicon effect.
(31) Lambert, J. B. Tetrahedron 1990, 46, 2677.
(32) Hammett, L. P. J. Am. Chem. Soc. 1937, 59, 96.
(33) Venkataraman, L.; Park, Y. S.; Whalley, A. C.; Nuckolls, C.;
Hybertsen, M. S.; Steigerwald, M. L. Nano Lett. 2007, 7, 502.
(34) Egorochkin, A. N.; Razuvaev, G. A.; Lopatin, M. A. J. Organomet.
Chem. 1988, 344, 49.
E
DOI: 10.1021/jacs.6b04394
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX