Covalent Assembly and Characterization of Nonsymmetrical Single-Molecule Nodes

Article abstract of DOI:10.1002/anie.201605421

The covalent linking of molecular building blocks on surfaces enables the construction of specific molecular nanostructures of well-defined shape. Molecular nodes linked to various entities play a key role in such networks, but represent a particular chal

Full text of DOI:10.1002/anie.201605421

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201605421
German Edition: DOI: 10.1002/ange.201605421
Molecular Electronics
Covalent Assembly and Characterization of Nonsymmetrical Single-
Molecule Nodes
Christophe Nacci, Andreas Viertel, Stefan Hecht, and Leonhard Grill*
Abstract: The covalent linking of molecular building blocks
on surfaces enables the construction of specific molecular
nanostructures of well-defined shape. Molecular nodes linked
to various entities play a key role in such networks, but
represent a particular challenge because they require a well-
defined arrangement of different building blocks. Herein, we
describe the construction of a chemically and geometrically
well defined covalent architecture made of one central node
and three molecular wires arranged in a nonsymmetrical way
and thus encoding different conjugation pathways. Very differ-
ent architectures of either very limited or rather extended size
were obtained depending on the building blocks used for the
covalent linking process on the Au(111) surface. Electrical
measurements were carried out by pulling individual molec-
ular nodes with the tip of a scanning tunneling microscope. The
results of this challenging procedure indicate subtle differences
if the nodes are contacted at inequivalent termini.
ical linking procedure.[12] Typically, the building blocks can
neither be deposited by sublimation nor from solution, owing
to thermal decomposition and lack of solubility, respectively.
Different covalently linked molecular nanostructures have
been obtained on the basis of a variety of chemical reactions
and molecular building blocks. They can be classified on one
hand as one-dimensional chains, such as polyporphyrins,[13]
polyfluorene,[14,15] polyphenylene,[16] graphene nanorib-
bons,[17,18] and polyethylene strings,[19] which have mainly
been studied by scanning tunneling microscopy (STM), but
also by more chemical analytical techniques, such as mass
spectrometry.[20] On the other hand, two-dimensional net-
works, either with an approximately rectangular[13,21,22] or with
a hexagonal lattice[23–25] can be created if each molecular
building block exhibits at least three potential (cross-)linking
functionalities.
The synthesis of molecular nodes is very challenging as it
requires specific chemical structures, which must be precisely
incorporated between the connecting chains. To the best of
our knowledge, only one case has been reported thus far, and
the synthesis resulted in symmetrical nodes.[17] The ex situ
synthesis of functional intramolecular circuits (containing at
least one node) is unlikely to be a suitable approach, first
because solvent molecules can hardly be avoided, thus not
permitting deposition onto a surface under clean conditions,
and second because of a lack of compatibility with the future
target of a planar assembly of many functional units.
Herein, we present the assembly of three-terminal
molecular nodes with various chemical structures of both
the node molecule itself and the molecular connectors.
Importantly, these nodes connect molecular wires in a non-
symmetrical fashion with chemically different transport
channels (in contrast to symmetrical nodes with only one
type of channel), a point of key interest for molecular
electronics. Although electronic transport in molecular nodes
has been described theoretically,[26] no conductance measure-
ments through single nodes exist in which the inequivalent
contact geometries, possible with three node connections
(and two electrodes), can be controlled.
For the molecular node, we tested hexabenzocoronene-
based Br3HBC (Figure 1a) and hexaphenylbenzene-based
Br3HPB (Figure 1c). These molecules are equipped with Br
substituents, which define the linking positions to other
molecules[13] and are clearly visible in STM images of intact
molecules (Figure 1b,d). The difference between HBC and
HPB nodes is that the former are planar aromatic systems,
whereas the latter are twisted and exhibit a larger HOMO–
LUMO gap as compared to HBC. These trifunctional node
molecules were then connected by on-surface synthesis with
short oligofluorene monomers (DBTF; Figure 1e) in a non-
O
ne of the main motivations for preparing covalently
linked molecular nanostructures is their possible use as
interconnects in molecular electronics. The idea that individ-
ual molecules could exhibit an electronic function goes back
to Aviram and Ratner, who proposed single-molecule rec-
tifiers.[1] The concept of monomolecular electronics[2–4] is
based on the functionalities of each individual molecule,
which interacts specifically with another, in contrast to large
ensembles in solution.[5] Key components in molecule-based
electronic circuits are molecular nodes: junctions where
several molecular chains merge.
Covalent linking of molecular building blocks on atomi-
cally defined single-crystal surfaces, so-called on-surface
synthesis,[6] is a rapidly growing field.[7–11] Such reactions
result in stable oligomers/polymers of well-defined shape and
composition, which are given by the chemical structure of the
initial building blocks as, for example, realized in a hierarch-
[*] Dr. C. Nacci, Prof. Dr. L. Grill
Department of Physical Chemistry
Fritz Haber Institute of the Max Planck Society
Faradayweg 4–6, 14195 Berlin (Germany)
and
Department of Physical Chemistry, University of Graz
Heinrichstrasse 28, 8010 Graz (Austria)
E-mail: leonhard.grill@uni-graz.at
Homepage: http://chemie.uni-graz.at/de/nano
Dr. A. Viertel, Prof. Dr. S. Hecht
Department of Chemistry and IRIS Adlershof
Humboldt-Universitꢀt zu Berlin
Brook-Taylor-Strasse 2, 12489 Berlin (Germany)
Supporting information for this article can be found under:
http://dx.doi.org/10.1002/anie.201605421.
Angew. Chem. Int. Ed. 2016, 55, 1 – 6
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

Angewandte
Communications
Chemie
monomer diffusion for more efficient
covalent connection, we employed
hexaphenylbenzene (HPB) molecu-
lar nodes on Au(111) (Figure 1c,d).
To limit the activation temperature
for the covalent coupling and avoid
undesired desorption of the molec-
ular nodes, we used iodine substitu-
ents instead of bromine on the fluo-
rene molecules (Figure 1e) owing to
their lower activation temperature.[12]
After covalent linking (sample heat-
ing at 390 K for 5 min), long polymer
chains consisting of both HPB and
terfluorene units were found on the
À
surface (see Figure S3). The C C
bond-formation process is very effi-
cient as evident from the extended
networks, probably as a result of the
increased mobility of the molecular
HPB nodes (in contrast to HBC
nodes; Figure 2). However, these
structures exhibit “connectors” with
a broad length distribution, that is,
Figure 1. Molecular-node building blocks. a–d) Chemical structures of the hexabenzocoronene
(HBC)- and chexaphenylbenzene (HPB)-based molecules studied, each equipped with three Br
substituents for on-surface synthesis, and corresponding STM images of single molecules on
Au(111). Imaging parameters: b) À50 mV, 100 pA; d) 50 mV, 250 pA. e,f) Chemical structures of
dibromoterfluorene (DBTF) and bromoquaterfluorene (BQF). g) Sketch of a molecular node with
three linkers attached, resulting in characteristic para, meta, and ortho channels for charge
transport.
symmetrical arrangement with para, meta, or ortho p-
conjugation paths (Figure 1g). Experiments were carried
out under ultrahigh vacuum with STM at a temperature of
10 K (see the Supporting Information for details).
By depositing Br3HBC and DBTF molecules onto Au-
(111) and subsequently heating the sample to 523 K, we
succeeded in forming nanostructures from these molecules
(Figure 2). New heterostructures were identified, thus prov-
ing the successful covalent connection of the different
molecular building blocks. However, in comparison with
other on-surface synthesis reactions,[12,27] the resulting struc-
tures were rather small (see Figure S1 in the Supporting
Information for details). Many HBC nodes were connected
with only one terfluorene unit (see arrow in Figure 2a),
whereas two-terminal nodes (Figure 2b–d) were found on
a few occasions only (about 5% of 638 structures), and three-
terminal molecular nodes were extremely rare (0.3%).
Besides these few oligomers, individual HBC molecules
Figure 2. Two-terminal molecular nodes. a) STM topography
were present at the surface to a large extent (about 54%),
lacking the three Br substituents. This result indicates that
although the dehalogenation process was successful, the
linking step was inefficient, presumably owing to limited
HBC diffusion as a result of the strong interaction of these
molecules with the metal substrate.[28] This interaction
became evident in pulling experiments, which turned out to
be very inefficient, in contrast to previous studies.[15,29] They
failed preferentially at a tip height of about 20 ꢀ (see
Figure S2), which precisely matches the distance of the HBC
node from the string terminus, and thus the moment at which
the node is lifted off the surface. Consequently, the difficulty
in the pulling experiments can be ascribed to the rather strong
adsorption of HBC on Au(111).[28]
(343ꢁ343 ꢂ2, À0.5 V, 50 pA) of heterostructures comprised of Br3HBC
and DBTF on Au(111) by on-surface synthesis. The arrow indicates
a node connected to a single wire. b–d) STM topographies (86ꢁ86 ꢂ2,
À0.5 V, 50 pA) showing the three different possible channels compris-
ing a HBC node and two terfluorene wires on Au(111): ortho (b), para
(c), and meta (d).
long polyfluorene chains, and thus lack the desired control
over the final structure.
The growth of well-defined molecular nodes with attached
chains therefore requires efficient covalent linking reactions
as well as a limited length of the molecular wires. These
conditions, somewhat contradictory, led us to a new molecular
building block: bromoquaterfluorene (BQF; Figure 1 f),
which exhibits only one terminal Br substituent, thus guar-
To reduce the interaction of the molecular node with the
Au(111) surface, and consequently to enhance the node
2
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 6
These are not the final page numbers!

Angewandte
Communications
Chemie
anteeing that the on-surface reaction is limited to the addition
the molecular structures is evident from the rather gentle
of one quaterfluorene (QF) unit only and prohibiting the
formation of extended polyfluorene structures.
manipulation conditions, with a tunneling resistance of about
1.25 MW (20 nA at a bias voltage of 25 mV) being sufficient to
overcome the potential-energy barrier for separation. These
manipulation experiments[30] resulted in individual node
structures (inset of Figure 3a), which is not only important
for the unambiguous identification of the cluster structures,
but also for pulling experiments[15] with single-node struc-
tures. In a few instances, we observed the formation of
molecular nodes with more than three quaterfluorene legs/
arms attached (as in the center of Figure 3a). Owing to the
very low abundance (ca. 2%) of these species on the surface
(see the Supporting Information for details), no structural
insight into the nature of the attachment and the underlying
chemical mechanism could be obtained thus far.
After the deposition of trifunctional Br3HPB (Fig-
ure 1c,d) and BQF (Figure 1 f) onto Au(111) and subsequent
sample heating, no extended networks were identified (as
with diiodoterfluorene molecules), but rather small structures
of defined length, according to the presence of only one
halogen substituent, that is, one potential reactive site, in each
quaterfluorene molecule. Three-terminal molecular nodes—
consisting of the node building block and precisely three short
molecular wires, one at each linking site—were clearly visible
as individual units or in weakly coupled clusters (Figure 3b).
The central question in conductance experiments by
pulling with the STM tip was whether any difference between
the para, meta, and ortho transport channels (Figure 1g)
could be distinguished in single molecular nodes. Accordingly,
three configurations can be defined in a pulling geometry
(Figure 4a–c) with two superimposed tunneling transport
Figure 3. Three-terminal molecular nodes. a) Constant-current STM
image (686ꢁ686 ꢂ2, 1 V, 50 pA) taken after dehalogenating the species
on the surface. Individual molecular nodes and node clusters were
also identified (see arrows and the chemical structures). Inset: STM
image (1 V, 50 pA) of an individual three-terminal molecular node. b–
d) Untangling of molecular-node clusters by STM-based lateral manip-
ulation. Sequence of constant-current STM images (192ꢁ189 ꢂ2, 1 V,
50 pA) showing molecular nodes untied by constant-current lateral
manipulation (set point: 25 mV, 20 nA). The arrows represent the
direction along which the manipulation took place. The molecular
node manipulated in (c; left) is missing in (d), probably because it
was picked up by the STM tip.
Figure 4. Single-molecular-node conductance measurements.
a–c) Schematic illustration of the STM pulling process: a) Vertical
approach of the STM tip towards the surface; b) contact between the
STM tip and the molecular-wire terminus; c) tip retraction: the wire is
pulled off the surface. d,e) I(z) pulling curves related to configurations
A and B. The current is plotted at tip heights above 45 ꢂ; from
geometric considerations of the QF–HPB length, only above a tip
height of 45 ꢂ is it clear that the molecular node itself is lifted from
the surface and thus contributes to the current in the junction.
Other structures were also present on the surface (see
Figure S4), but a substantial fraction of the molecular
structures (29% of 604) were in the predefined three-
terminal-node configuration (Figure 1g), in agreement with
a simple statistical model (see Figure S6). These nodes were
highly defined in terms of composition, dimensions, and
symmetry. In contrast to the covalent HPB–fluorene bonds,
three-terminal molecular nodes in clusters were held together
by weak interactions and could be separated readily by STM
manipulation (Figure 3b–d). The weak interaction between
channels: configuration A with charge transport along para
and meta channels (Figure 4d), configuration B with charge
transport along para and ortho channels (Figure 4e), and
ortho and meta channels in the case of configuration C (not
shown). During the pulling process, we chose an individual
defect-free molecular node in the STM image and then
selectively contacted the STM tip with one of the three,
inequivalent, oligofluorene strands. The current dropped
Angew. Chem. Int. Ed. 2016, 55, 1 – 6
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

Angewandte
Communications
Chemie
when the contact between the tip and the molecular structure
was lost, typically at a vertical tip height of around 70 ꢀ
(Figure 4d,e), which corresponds approximately to the max-
imum length of the node structure (i.e. along the para
configuration) in the junction. Hence, the current drop is
a signature for complete removal of the molecule from the
surface.
than for single polyfluorene wires[15]—probably because here
two wires are arranged in parallel. Hence, more efficient
charge transport seems to be present in configuration B,
which combines both p-conjugated ortho and para channels,
as compared to configuration A, which involves a p-con-
jugated para channel in combination with a cross-conjugated
meta channel. Owing to the limited number of successful
pulling experiments, our results, however, should be regarded
with some caution, and future measurements, involving for
example a flat geometry and multiple tips, should be carried
out to validate this finding.
In conclusion, we have shown how molecular three-
terminal wire–node architectures—highly defined in terms of
dimensions, composition, symmetry, and geometrical arrange-
ment—can be formed. Rather small changes in the chemical
structures of the compounds involved had a profound effect
on the products of the on-surface synthesis, owing to
characteristic diffusion properties and reactivities. Different
electrical currents were measured through single molecular
nodes, depending on the precise location of the electrode
contacts with respect to the three connectors, however, in only
very few successful cases. First indications of different trans-
port behavior depending on the contributing conjugation
pathways through the node molecule have been revealed.
Although the node component (HPB) adsorbs less
strongly than the HBC molecule, the total molecule–surface
interaction remained strong enough to render the pulling
experiments very difficult with a success rate (over all
configurations) of about only 3% (over 844 attempts).
Interestingly, the success rates differed strongly, with the
best efficiency for configuration A (about 10%) and a much
reduced yield for configuration B (about 1%), whereas
configuration C could not be lifted successfully and measured
at all. We attribute these differences, which can not be caused
by the HPB node, primarily to the lateral bending angle of the
two quaterfluorene linkers that remain in contact with the
surface when the third is pulled. For example, in the simple
case of only two linkers in a para configuration (attached at
opposite sites to the node), this lateral bending angle with
respect to the node would be zero (note that bending off the
surface of course takes place). When going now from
configuration A to C, the bending angles of the surface-
attached linkers increase, which raises the energy barrier for
the process, thus rendering pulling more difficult. Considering
the prerequisites for a pulling experiment, that is, to find
a molecular node structure in a defect-free area and to
disentangle it from other node structures by lateral manip-
ulation (Figure 3b–d), a successful pulling experiment is an
extremely challenging and therefore very rare event.
Acknowledgements
Important discussions with Christian Joachim and Francisco
Ample are gratefully acknowledged. We thank the European
Union for financial support through the AtMol project.
Conductance I(z) curves were measured for configura-
tions A and B (all successful events are plotted in Fig-
ure 4d,e), and an approximately exponential decay behavior
was found for both. Hence, the conductance G of individual
configurations was determined on the basis of the inverse
decay length b by G = Go eÀbz (Go is the contact conductance
and z the electrode distance).[15,29] The obtained I(z) curves
are superimposed by oscillations that are probably caused by
the mechanical properties of the molecular structure, the
intramolecular degrees of freedom, during the pulling pro-
cess, as found in previous studies.[15,31] These intramolecular
rotary and bending motions can also affect the conjugation in
the molecules and consequently the charge transport through
them as visible in the current curves.
Keywords: conjugation · molecular electronics ·
molecular nodes · scanning probe microscopy ·
single-molecule manipulation
[1] A. Aviram, M. Ratner, Chem. Phys. Lett. 1974, 29, 277.
[2] J. M. Tour, Acc. Chem. Res. 2000, 33, 791.
[3] C. Joachim, J. K. Gimzewski, A. Aviram, Nature 2000, 408, 541.
[4] J. R. Heath, M. A. Ratner, Phys. Today 2003, 56, 43.
[5] A. P. de Silva, S. Uchiyama, Nat. Nanotechnol. 2007, 2, 399.
[6] D. F. Perepichka, F. Rosei, Science 2009, 323, 216.
[7] A. Gourdon, Angew. Chem. Int. Ed. 2008, 47, 6950; Angew.
Chem. 2008, 120, 7056.
[8] L. Bartels, Nat. Chem. 2010, 2, 87.
These electrical measurements through individual molec-
ular nodes with full control over the contacting geometry
(configuration A or B) proved very difficult, and therefore,
despite substantial experimental effort in time-consuming
measurements, only very few conductance curves I(z) could
be obtained. On the basis of the acquired data (Figure 4d,e),
it appears that the current curves point to a different behavior
in the two configurations, whereby configuration B seems to
exhibit better conductance (i.e. smaller decay) than config-
uration A. This observation was quantified by a statistical
analysis, from which we extracted very small b values of bA =
(0.053 Æ 0.008) ꢀÀ1 and bB = (0.020 Æ 0.010) ꢀÀ1 (bias vol-
tages between À1.8 and À1.9 V), which are clearly smaller
[9] M. Lackinger, W. M. Heckl, J. Phys. D 2011, 44, 464011.
[10] J. Mꢁndez, M. F. Lꢂpez, J. A. Martꢃn-Gago, Chem. Sov. Rev.
2011, 40, 4578.
[11] Q. Fan, C. Wang, Y. Han, J. Zhu, W. Hieringer, J. Kuttner, G.
Hilt, J. M. Gottfried, Angew. Chem. Int. Ed. 2013, 52, 4668;
Angew. Chem. 2013, 125, 4766.
[12] L. Lafferentz, V. Eberhardt, C. Dri, C. Africh, G. Comelli, F.
Esch, S. Hecht, L. Grill, Nat. Chem. 2012, 4, 215.
[13] L. Grill, M. Dyer, L. Lafferentz, M. Persson, M. V. Peters, S.
Hecht, Nat. Nanotechnol. 2007, 2, 687.
[14] A. Saywell, J. Schwarz, S. Hecht, L. Grill, Angew. Chem. Int. Ed.
2012, 51, 5096; Angew. Chem. 2012, 124, 5186.
[15] L. Lafferentz, F. Ample, H. Yu, S. Hecht, C. Joachim, L. Grill,
Science 2009, 323, 1193.
4
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 6
These are not the final page numbers!

Angewandte
Communications
Chemie
[16] J. A. Lipton-Duffin, O. Ivasenko, D. F. Perepichka, F. Rosei,
[25] M. Bieri, M. Treier, J. Cai, K. Ait-Mansour, P. Ruffieux, O.
Small 2009, 5, 592.
Grçning, P. Grçning, M. Kastler, R. Rieger, X. Feng, K. Mꢄllen,
R. Fasel, Chem. Commun. 2009, 6919.
[26] S. Ami, C. Joachim, Phys. Rev. B 2002, 65, 155419.
[27] J. Adisoejoso, T. Lin, X. S. Shang, K. J. Shi, A. Gupta, P. N. Liu,
N. Lin, Chem. Eur. J. 2014, 20, 4111.
[28] P. Ruffieux, O. Grçning, M. Bielmann, C. Simpson, K. Mꢄllen, L.
Schlapbach, P. Grçning, Phys. Rev. B 2002, 66, 073409.
[29] M. Koch, F. Ample, C. Joachim, L. Grill, Nat. Nanotechnol. 2012,
7, 713.
[30] G. Meyer, J. Repp, S. Zçphel, K.-F. Braun, S. W. Hla, S. Fçlsch,
L. Bartels, F. Moresco, K.-H. Rieder, Single Mol. 2000, 1, 79.
[31] C. Nacci, F. Ample, D. Bleger, S. Hecht, C. Joachim, L. Grill, Nat.
Commun. 2015, 6, 7397.
[17] J. Cai, P. Ruffieux, R. Jaafar, M. Bieri, T. Braun, S. Blankenburg,
M. Muoth, A. P. Seitsonen, M. Saleh, X. Feng, K. Mꢄllen, R.
Fasel, Nature 2010, 466, 470.
[18] S. Linden, D. Zhong, A. Timmer, N. Aghdassi, J. H. Franke, H.
Zhang, X. Feng, K. Mꢄllen, H. Fuchs, L. Chi, H. Zacharias, Phys.
Rev. Lett. 2012, 108, 216801.
[19] D. Zhong, J.-H. Franke, S. K. Podiyanachari, T. Blçmker, H.
Zhang, G. Kehr, G. Erker, H. Fuchs, L. Chi, Science 2011, 334,
213.
[20] S. Boz, M. Stçhr, U. Soydaner, M. Mayor, Angew. Chem. Int. Ed.
2009, 48, 3179; Angew. Chem. 2009, 121, 3225.
[21] M. Inꢅt Veld, P. Iavicoli, S. Haq, D. B. Amabilino, R. Raval,
Chem. Commun. 2008, 1536.
[22] M. Abel, S. Clair, O. Ourdjini, M. Mossoyan, L. Porte, J. Am.
Chem. Soc. 2011, 133, 1203.
[23] R. Gutzler, H. Walch, G. Eder, S. Kloft, W. M. Heckl, M.
Lackinger, Chem. Commun. 2009, 4456.
[24] J. F. Dienstmaier, A. Gigler, A. J. Goetz, P. Knochel, T. Bein, A.
Lyapin, S. Reichlmaier, W. M. Heckl, M. Lackinger, ACS Nano
2011, 5, 9737.
Received: June 3, 2016
Revised: August 1, 2016
Published online: && &&, &&&&
Angew. Chem. Int. Ed. 2016, 55, 1 – 6
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!

Angewandte
Communications
Chemie
Communications
Molecular Electronics
The path matters! Hexaphenylbenzene-
based molecular nodes and polyfluorene
wires are the constituents used for the on-
surface synthesis of asymmetric node
structures with a well-defined geometry,
size, and composition (see structure).
Measurements of electronic transport
through individual molecular nodes when
lifted from the surface by an STM tip
revealed different transport behavior
depending on the p-conjugation pathway.
C. Nacci, A. Viertel, S. Hecht,
L. Grill*
&&&& — &&&&
Covalent Assembly and Characterization
of Nonsymmetrical Single-Molecule
Nodes
6
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 6
These are not the final page numbers!