Self-assembly of chiral molecular honeycomb networks on Au(111)

Article abstract of DOI:10.1021/ja7106542

Self-assembly of the specifically designed 3-fold symmetric 1,3,5-trikis(4′-carboxylphenyl)-2,4,6-trikis(4′-tert-butylphenyl)-benzene molecule on a Au(111) surface is investigated by means of in situ ultrahigh vacuum scanning tunneling microscopy. A variety of chiral honeycomb networks with increasing interpore distance is observed. Experiments and force field calculations reveal that the formation of the two-dimensional hexagonal porous networks is driven by a balance of molecular close-packing within half-unit cells and cyclic hydrogen bonding dimerization between half-unit cells. The honeycomb networks are shown to be chiral owing to asymmetric molecular close-packing within the half-unit cells. Copyright

Full text of DOI:10.1021/ja7106542

Published on Web 06/18/2008
Self-Assembly of Chiral Molecular Honeycomb Networks on Au(111)
Wende Xiao,† Xinliang Feng,‡ Pascal Ruffieux,† Oliver Gro¨ning,† Klaus Mu¨llen,*,‡ and
Roman Fasel*,†
Empa, Swiss Federal Laboratories for Materials Testing and Research, Feuerwerkerstrasse 39, CH-3602 Thun,
Switzerland, and Max-Planck-Institute for Polymer Research, Ackermannweg 10, 55128 Mainz, Germany
Received December 21, 2007; E-mail: muellen@mpip-mainz.mpg.de; roman.fasel@empa.ch
In recent years, the transfer and adaptation of the concepts of
crystal engineering1 to the case of two-dimensional (2D) molecular
self-assembly on solid surfaces has intensively been explored.2 The
formation of a variety of well-defined, 2D porous molecular
networks based on specific, noncovalent interactions such as
hydrogen bonding (HB), dipolar coupling, and metal co-ordination
has been reported.3 Such regular open networks offer the possibility
to host functional guest molecules in a controllable way, with
potential applications in single-molecule-based devices.
Hexaphenylbenzene (HPB) is one of the smallest rigid star-
shaped polyphenylenes, and has recently been designed to engineer
porous crystals.4 Surface patterning with HPB has only rarely been
investigated,5 presumably due to a lack of HPB derivatives
exhibiting appropriate molecular symmetry and functionality.
Herein, we report on the formation of surface-supported chiral
honeycomb networks by self-assembly of a novel, specifically
designed C3 symmetric 1,3,5-trikis(4′-carboxylphenyl)-2,4,6-trikis(4′-
tert-butylphenyl)-benzene (1, Figure 1a) on the Au(111) surface.
In-situ ultrahigh vacuum scanning tunneling microscopy (STM)
investigations and force field calculations show that 1 self-assembles
Figure 1. STM images after rt deposition of 0.3 ML 1 on Au(111) followed
by annealing to 100-180 °C: (a) ball-and-stick model of 1; (b) EHT-STM
into a series of 2D hexagonal porous networks, where molecules
of 1 within each half-unit cell are close-packed via van der Waals
(vdW) interactions, and all half-unit cells are connected to each
other via dimeric HBs between carboxyl groups. Asymmetric
molecular close-packing within the half-unit cells induces chirality
in the honeycomb networks. The coexistence of various honeycomb
orders at submonolayer coverages highlights the importance of both
specifically designed directional interactions (Desiraju-Wuest
postulate)1b,c and weaker, less specific interactions such as vdW
that favor close-packing (Kitaigorodskii’s principle)6 in determining
2D crystal structures.
simulation of 1; (c) overview image showing coexistence of two different
ordered phases A and B; (d) details of the ordered honeycomb network
structure of phase A. The unit cell of the honeycomb network is marked
with dashed lines.
of ordered phases as well as some disordered structures at a sample
temperature of ∼50 K, as demonstrated in Figure 1c. Images at
high magnification (Figure 1d) of phase A show a regular
honeycomb lattice with a pore-to-pore distance of 2.96 ( 0.05 nm.
Each unit cell consists of two molecules, and each molecule
resembles an equilateral triangle with the three brightest spots
originating from the three tert-butyl spacers. In highly resolved
images such as the one shown in Figure 1d, the six phenyl rings of
the HPB core can be identified as fainter intensity maxima. STM
image simulations based on semiempirical extended Hu¨ckel theory
(EHT) support this interpretation of the experimental STM images
of 1, as demonstrated in Figure 1b.
Amber3 force field calculations (see Supporting Information)
reveal that the formation of the phase A honeycomb network is
driven by the maximization of HBs, namely, the formation of three
dimeric HBs for each molecule with its three nearest neighbors
via carboxyl groups. Since the carboxyl groups prefer to stay in
the plane of the connecting phenyl rings, the formation of dimeric
HBs is only possible between molecules of opposite handedness.
Therefore, the honeycomb network observed in phase A is overall
racemic. We denote it as HC1, where HC stands for the honeycomb
network of 1, and the subscript number 1 indicates that the side
length of the triangular half-unit cell is one molecule. The
corresponding structural model is superposed to Figure 1d (see also
Figure 3a). Similar structures have been observed for other C3
Molecule 1 consists of a HPB core and three alternating tert-
butyl spacer and carboxyl groups. In the solid state and on surfaces,
the outer phenyl rings of the HPB core are rotated around their
σ-bonds connecting to the central benzene ring because of intra-
molecular steric hindrance, resulting in a dihedral angle of ∼65°.4
Because of the resulting propeller shape of the HPB core, surface-
adsorbed 1 has two opposite chiral conformations. We note that
the handedness of single molecules is intrinsic and its definition
requires no consideration of the molecule-surface system (see
Supporting Information). The inert Au(111) surface is chosen as
the substrate, so that molecule-surface interactions are minimized
and the self-assembly of 1 is expected to be mainly guided by the
specifically designed intermolecular interactions.
After vapor deposition of 0.3 monolayer (ML) 1 onto clean
Au(111) at room temperature (RT) and subsequent annealing to
temperatures between 100-180 °C, we observe by STM a variety
† Swiss Federal Laboratories for Materials Testing and Research.
‡ Max-Planck-Institute for Polymer Research.
9
8910 J. AM. CHEM. SOC. 2008, 130, 8910–8912
10.1021/ja7106542 CCC: $40.75
2008 American Chemical Society

C O M M U N I C A T I O N S
Figure 2. STM images of ordered chiral honeycomb networks observed
after deposition of 0.3 ML 1: (a, b) L- and R-domains of the HC2 structure;
(c) L-domain of the HC3 structure; (d) R-domain of the HC4 structure. Unit
cells are indicated by white dashed lines. Individual molecules are
schematically indicated as red and blue triangles (opposite chiral conforma-
tions of 1) with corners at tert-butyl positions.
symmetric molecules, such as trimesic acid (TMA) or 1,3,5-
benzenetribenzoic acid.7,8
Phase B also exhibits a honeycomb lattice, but with a periodicity
of 4.63 ( 0.05 nm, much larger than that of phase A. Each unit
cell again consists of two triangular subunits (half-unit cells), which,
however, are not single molecules but trimers of 1 with an
intermolecular separation of 1.7 ( 0.05 nm, as seen in the close-
ups shown in Figure 2(a,b). The orientation of the triangle
constructed by connecting the centers of mass of three molecules
inside a subunit is rotated by ( 15° with respect to the triangular
molecular configurations. We denote this honeycomb network as
HC2 because its triangular subunit has a side length of two
molecules. Force field calculations give an equilibrium inter-
molecular separation of 1.72 nm for the trimer and a molecular
arrangement in perfect agreement with the one derived from STM
images such as Figure 2(a,b). The three molecules of 1 inside a
subunit have the same handedness and are close-packed via vdW
interactions. As for the HC1 structure, the subunits are bound to
each other via dimeric HBs between carboxyl groups, implying
that molecules in the two different subunits of a unit cell have
opposite handedness. Therefore, the HC2 network is racemic from
a molecular point of view. On the other hand, asymmetric close-
packing of the three molecules within the half-unit cells leads to
chiral trimer subunits, as illustrated in Figure 2(a,b).9 The trimers
in the two half-unit cells have identical molecular arrangements,
but with an in-plane rotation of 180°. The chirality of the trimer
subunits is thus transferred to the unit cell and to the entire
honeycomb network. Extended homochiral domains of both the left-
handed (L) and the right-handed (R) honeycomb network HC2 are
observed (Figure 2(a,b)). A structural model of the L-domain of
the HC2 network is illustrated in Figure 3b.9
Figure 3. Structural models of the honeycomb networks of 1. (a) HC1
structure. (b-d) L-domains of the HC2, HC3, and HC4 structures, respectively.
molecules to each half-unit cell of the HCn structure (n ) 1, 2,
3,..., ∞). The molecules inside a half-unit cell are close-packed via
vdW interaction, forming a chiral triangle, while the half-unit cells
are bound to each other via n HBs. The pore-to-pore distance dn of
the HCn structure can be described by dn ) (n - 1)p + q, where
q ) 2.96 nm is the interpore distance of the HC1 structure, and p
) 1.7 nm is the intermolecular distance of the close-packed
monolayer (which can be regarded as the HC∞ structure with infinite
interpore distance). We indeed observe highly ordered chiral
domains of the HC3 and HC4 structures (Figure 2(c,d)), as well as
extended close-packed domains (HC∞). Structural models for the
L-domains of the HC3 and the HC4 structures are given in Figure
3(c,d). The measured interpore distances of the HC3 and HC4
structures are 6.4 ( 0.05 and 8.05 ( 0.05 nm, respectively, well
consistent with the values expected from the generalized model.
Local patches corresponding to even higher order HCn structures
with n ) 5, 6, 7, and 8 are also observed, but with relatively poor
long-range order.
The formation of a series of honeycomb networks with increasing
interpore distance is not limited to 1. During the completion of
this manuscript, a similar behavior was reported for TMA on
Au(111): Depending on the TMA coverage, honeycomb networks
with half-unit cells containing molecules in up to 8 rows were
observed.8 Comparing the self-assembly of 1 and TMA molecules,
it is obvious that in both cases the half-unit cells are strongly bound
to each other via dimeric HBs, while the molecules within the half-
unit cells are close-packed via weaker interactions (trimeric HBs
for the TMA molecules and vdW interactions for 1). Notably, the
TMA honeycomb networks are achiral, while the ones of 1 are
The HC2 structure can be viewed as an inflation of the HC1
structure by addition of one more row (with two molecules) into
its half-unit cell. Generalizing this view, higher order honeycomb
networks HCn+1 can be obtained by appending a row of n + 1
9
J. AM. CHEM. SOC. VOL. 130, NO. 28, 2008 8911

C O M M U N I C A T I O N S
In being rationalized in terms of interaction energy densities
rather than interaction energies per molecule, our results suggest
that Kitaigorodskii’s principle of closest-packing might also apply
to 2D molecular systems.6 They indicate that considering interaction
energies per molecule might not be sufficient to determine lowest-
energy configurations of 2D molecular systems: Intrinsically
stronger intermolecular association (favoring low order honey-
combs) may be counterbalanced by tighter packing (favoring higher
order honeycombs). A similar conclusion has been drawn for 3D
crystallization by Angeloni and co-workers: When determining the
three-dimensional crystal structures of a series of [PtCl4]2- and
[SbCl5]2- salts,10 they found that local HB interactions, although
important in determining the local intermolecular geometry, are not
the only or even the decisive influence on the crystal structures.
The close-packing of the complex ions also plays an important role,
and the crystal structures of these molecular salts result from the
subtle balance of directional HBs and close-packing. Similarly, our
results provide a 2D example where both, the Desiraju-Wuest
postulate and Kitaigorodskii’s principle of closest packing, play
important roles in determining molecular crystal structures.1,6
In summary, we have investigated the self-assembly of the novel
C3 symmetric star-shaped molecule 1 on a Au(111) surface at the
submolecular level by STM. We observe a variety of chiral
honeycomb structures where molecules inside half-unit cells are
close-packed via vdW interactions while half-unit cells are con-
nected to each other via dimeric HBs between carboxyl groups.
The networks are chiral due to nonsymmetrical molecular close-
packing inside the half-unit cells. The coexistence of various
honeycomb orders highlights the subtle interplay of directional HB
and less specific vdW interactions in determining 2D crystal
structures.
Figure 4. (a) Interaction energy per molecule (Enmol) vs the order (n) of
the honeycomb networks of 1. (b) Interaction energy density (EnA) vs the
order (n) of the honeycomb networks of 1. Different ratios (R) of the
interaction energy of dimeric hydrogen bonds to that of a close-packed trimer
are plotted. The red curves with R ) 1.02 correspond to the actual interaction
energy values as obtained from Amber3 force field calculations.
chiral due to the asymmetric molecular close-packing inside half-
unit cells. Generally, we propose that if a molecule with C3
symmetry can form both a honeycomb network via a strong and
directional binding and a hexagonal close-packed structure via a
weaker interaction, it will also be able to self-assemble into a series
of higher order honeycomb network structures with increasing
interpore distance.
In the case of TMA, the formation of higher order honeycomb
networks with increasing coverage is understood from simple
H-bond optimization considerations.8 In the present case of the HPB
species 1, however, the coexistence of various honeycomb orders
at submonolayer coverages calls for a more elaborate explanation.
We have investigated the energetics of the HCn family of
honeycomb networks by means of molecular force field calculations
(see Supporting Information). For a honeycomb network HCn of
Acknowledgment. Financial support by the European Com-
mission(RADSAS,NMP3-CT-2004-001561)isgratefullyacknowledged.
Supporting Information Available: Synthesis and characterization
of compounds, STM experiments, molecular modeling, chirality of
surface-adsorbed 1, and force field calculations. This material is
available free of charge via the Internet at http://pubs.acs.org
mol
order n, the total interaction energy per molecule is given by En
) (3nEHB + n(n - 1)ECP)/(n(n + 1)), where EHB and ECP refer to
the energy gains of forming a dimeric HB and a close-packed trimer,
respectively. The resulting interaction energy as a function of order
n is shown in Figure 4a. The red curve corresponds to a ratio R )
EHB/ECP of 1.02 between the dimeric hydrogen-bonding and trimeric
close-packing interaction energies as given by Amber3 force field
calculations (EHB ) -7.9 kcal/mol; ECP ) -7.75 kcal/mol; see
Supporting Information). According to this interaction energy per
molecule curve, submonolayers of 1 should obviously condense
exclusively in the strongly favored order 1 network HC1, which is
contrary to experimental observations. Kinetic limitations are not
able to explain the formation of higher order networks, since they
are expected to act in the opposite direction, that is, to hinder the
formation of higher order, more complex structures.
References
(1) (a) Schmidt, G. M. J. Pure Appl. Chem. 1971, 27, 647. (b) Desiraju, G. R.
Angew. Chem., Int. Ed. 1995, 34, 2311. (c) Su, D.; Wang, X.; Simard, M.;
Wuest, J. D. Supramol. Chem. 1995, 6, 171.
(2) Barth, J. V. Annu. ReV. Phys. Chem. 2007, 58, 375.
(3) (a) Theobald, J. A.; Oxtoby, N. S.; Phillips, M. A.; Champness, N. R.;
Beton, P. H. Nature 2003, 424, 1029. (b) Stepanow, S.; Lingenfelder, M.;
Dmitriev, Al; Spillmann, H.; Delvigne, E.; Lin, N.; Deng, X.; Cai, C.; Barth,
J. V.; Kern, K. Nat. Mater. 2004, 3, 229. (c) Pawin, G.; Wong, K. L.;
Kwon, K.-Y.; Bartels, L. Science. 2006, 313, 961.
(4) (a) Maly, K. E.; Gagnon, E.; Maris, T.; Wuest, J. D. J. Am. Chem. Soc.
2007, 129, 4306. (b) Kobayashi, K.; Sato, A.; Sakamoto, S.; Yamaguchi,
K. J. Am. Chem. Soc. 2003, 125, 3035. (c) Kobayashi, K.; Shirasaka, T.;
Sato, A.; Horn, E.; Furukawa, N. Angew. Chem., Int. Ed. 1999, 38, 3483.
(5) (a) Gross, L.; Rieder, K. H.; Moresco, F.; Stojkovic, S. M.; Gourdon, A.;
Joachim, C. Nat. Mater. 2005, 4, 892. (b) Gross, L.; Moresco, F.; Ruffieux,
P.; Stojkovic, S. M.; Gourdon, A.; Joachim, C.; Rieder, K. H. Phys. ReV.
B 2005, 71, 165428.
The situation looks different if we consider the total interaction
2
energy density EnA ) (3nEHB + n(n - 1)ECP)/ꢀ3/2dn , that is, the
interaction energy per unit area. The resulting interaction energy
density curves for different EHB/ECP ratios R are shown in Figure
4b. For the ratio R ) 1.02 corresponding to our calculated
interaction energy strengths, the interaction energy density depends
only weakly on the order n of the network, and exhibits a minimum
for n ) 2∼3. Higher order networks (n ) 4, 5,...) are only slightly
less favorable. The near degeneracy of the interaction energy density
for order 2, 3,..., 8 honeycomb networks thus rationalizes the
experimentally observed coexistence of these different honeycombs
of 1.
(6) Kitaigorodskii, A. I. Acta Crystallogr. 1965, 18, 585.
(7) (a) Griessl, S.; Lackinger, M.; Edelwirth, M.; Hietschold, M.; Heckl, W. M.
Single Mol. 2002, 3, 1. (b) Ruben, M.; Payer, D.; Landa, A.; Comisso, A.;
Gattinoni, C.; Lin, N.; Collin, J.-P.; Sauvage, J.-P.; De Vita, A.; Kern, K.
J. Am. Soc. Chem. 2006, 128, 15644.
(8) Ye, Y.; Sun, W.; Wang, Y.; Shao, X.; Xu, X.; Cheng, F.; Li, J.; Wu, K. J.
Phys. Chem. C. 2007, 111, 10138.
(9) See the Supporting Information for detailed molecular models of the L-
and R-domains of the HC2 structure.
(10) Angeloni, A.; Crawford, P. C.; Orpen, A. G.; Podesta, T. J.; Shore, B. J.
Chem. Eur. J. 2004, 10, 3783.
JA7106542
9
8912 J. AM. CHEM. SOC. VOL. 130, NO. 28, 2008