Two-dimensional porous molecular networks of dehydrobenzo[12]annulene derivatives via alkyl chain interdigitation

Article abstract of DOI:10.1021/ja0655441

The self-assembly of a series of hexadehydrotribenzo[12]annulene (DBA) derivatives has been scrutinized by scanning tunneling microscopy (STM) at the liquid-solid interface. First, the influence of core symmetry on the network structure was investigated b

Full text of DOI:10.1021/ja0655441

Published on Web 12/08/2006
Two-Dimensional Porous Molecular Networks of
Dehydrobenzo[12]annulene Derivatives via Alkyl Chain
Interdigitation
Kazukuni Tahara,† Shuhei Furukawa,‡ Hiroshi Uji-i,‡ Tsutomu Uchino,†
Tomoyuki Ichikawa,† Jian Zhang,‡ Wael Mamdouh,‡ Motohiro Sonoda,†
Frans C. De Schryver,‡ Steven De Feyter,*,‡ and Yoshito Tobe*,†
Contribution from the DiVision of Frontier Materials Science, Graduate School of Engineering
Science, Osaka UniVersity, Toyonaka, Osaka 560-8531, Japan, and Department of Chemistry,
DiVision of Molecular and Nanomaterials, Laboratory of Photochemistry and Spectroscopy, and
Institute of Nanoscale Physics and Chemistry, Katholieke UniVersiteit LeuVen (KULeuVen),
Celestijnenlaan 200 F, B-3001 LeuVen, Belgium
Received August 1, 2006; E-mail: Steven.DeFeyter@chem.kuleuven.be; tobe@chem.es.osaka-u.ac.jp.
Abstract: The self-assembly of a series of hexadehydrotribenzo[12]annulene (DBA) derivatives has been
scrutinized by scanning tunneling microscopy (STM) at the liquid-solid interface. First, the influence of
core symmetry on the network structure was investigated by comparing the two-dimensional (2D) ordering
of rhombic bisDBA 1a and triangular DBA 2a (Figure 1). BisDBA 1a forms a Kagome´ network upon
physisorption from 1,2,4-trichlorobenzene (TCB) onto highly oriented pyrolytic graphite (HOPG). Under
similar experimental conditions, DBA 2a shows the formation of a honeycomb network. The core symmetry
and location of alkyl substituents determine the network structure. The most remarkable feature of the
DBA networks is the interdigitation of the nonpolar alkyl chains: they connect the π-conjugated cores and
direct their orientation. As a result, 2D open networks with voids are formed. Second, the effect of alkyl
chain length on the structure of DBA patterns was investigated. Upon increasing the length of the alkyl
chains (DBAs 3c-e) a transition from honeycomb networks to linear networks was observed in TCB, an
observation attributed to stronger molecule-substrate interactions. Third, the effect of solvent on the structure
of the nonpolar DBA networks was investigated in four different solvents: TCB as a polar aromatic solvent,
1-phenyloctane as a solvent having both aromatic and aliphatic moieties, n-tetradecane as an aliphatic
solvent, and octanoic acid as a polar alkylated solvent. The solvent dramatically changes the structure of
the DBA networks. The solvent effects are discussed in terms of factors that influence the mobility of
molecules at the liquid-solid interface such as solvation.
Introduction
copy (STM) is a powerful technique for the investigation of
two-dimensional (2D) crystal structures of π-electron-conjugated
The control of the spatial organization of functional π-electron-
conjugated systems1 has led to major advances in the field of
organic materials, such as liquid crystal materials, molecular-
scale electronics,2 and molecular devices.3 Recently, the struc-
tural control of organic monolayers has been presented as a
further important step to achieve highly ordered organic
nanostructures and also as an opportunity to fabricate molecular
scale devices on flat substrates.1b,4 Scanning tunneling micros-
systems with atomic scale resolution. STM not only operates
under ultrahigh vacuum conditions but also at the interface
between two condensed media; one is an atomically flat
conducting solid and the other is a gas, a liquid, a liquid
crystalline material, or a gel.
One of the useful approaches is to decorate the surface with
molecules that form regular 2D periodic patterns by self-
assembly at the liquid-solid interface.5 In such systems, it is
essential to control the competition between the following four
modes of interaction to construct the desired molecular patterns;
molecule-molecule, molecule-substrate, molecule-solvent, and
solvent-substrate interactions. Especially, the control of mol-
† Osaka University.
‡ Katholieke Universiteit Leuven.
(1) (a) Hoeben, F. J. M.; Jonkheijm, P.; Meijer, E. W.; Schenning, A. P. H. J.
Chem. ReV. 2005, 105, 1491-1546. (b) Grimsdale, A. C.; Mu¨llen, K.
Angew. Chem., Int. Ed. 2005, 44, 5592-5629.
(2) (a) Joachim, C.; Gimzewski, J. K.; Aviram, A. Nature 2000, 408, 541-
548. (b) Nitzan, A.; Ratner, M. A. Science 2003, 300, 1384-1389.
(3) Collier, C. P.; Mattersteig, G.; Wong, E. W.; Luo, Y.; Beverly, K.; Sampaio,
J.; Raymo, F. M.; Stoddart, J. F.; Heath, J. R. Science 2000, 289, 1172-
1175.
(4) (a) Samor´ı, P.; Severin, N.; Simpson, C. D.; Mu¨llen, K.; Rabe, J. P. J. Am.
Chem. Soc. 2002, 124, 9454-9457. (b) Piot, L.; Marchenko, A.; Wu, J.;
Mu¨llen, K.; Fichou, D. J. Am. Chem. Soc. 2005, 127, 16245-16250. (c)
Okawa, Y.; Aono, M. Nature 2001, 409, 683-684.
(5) For recent reviews on STM measurements at the liquid-solid interface,
see, for example: (a) De Feyter, S.; De Schryver, F. C. Chem. Soc. ReV.
2003, 32, 139-150. (b) De Feyter, S.; De Schryver, F. C. J. Phys. Chem.
B 2005, 109, 4290-4302. (c) Samor´ı, P.; Rabe, J. P. J. Phys.: Condens.
Matter 2002, 14, 9955-9973. (d) Giancarlo, L. C.; Flynn, G. W. Acc. Chem.
Res. 2000, 33, 491-501. (e) Wan, L.-J. Acc. Chem. Res. 2006, 39, 334-
342. (f) Hermann, B. A.; Scherer, L. J; Housecroft, C. E.; Constable, E. C.
AdV. Funct. Mater. 2006, 16, 221-235.
9
10.1021/ja0655441 CCC: $33.50 © 2006 American Chemical Society
J. AM. CHEM. SOC. 2006, 128, 16613-16625
16613

A R T I C L E S
Tahara et al.
Figure 1. Structures of (a) decadehydrotetrabenzo[12]annuleno[12]annulene (bisDBAs, 1a,b), (b) hexadehydrotribenzo[12]annulene (DBA, 2a-b and 3a-
f), and (c) triphenylene (4a,b) derivatives.
ecule-molecule interactions is crucial for the formation of
molecular networks: typically, discrete molecular building-
blocks are ‘connected’ to each other by virtue of directional
intermolecular interactions such as hydrogen bonding or metal
coordination. Being able to correlate molecular features such
as shape, position of interacting sites, as well as electronic
properties with the resulting topology of the molecular archi-
tectures would enable us to design and construct the network
structures and their functions. This strategy is already well-
known as “crystal engineering” in three-dimensional (3D) crystal
systems.6 Of special interest are 2D molecular porous net-
works.7,8
In 2D systems, the molecule-substrate interaction is an
important issue to regulate network topologies in terms of
symmetry matching of the molecules with the substrate. The
strength of the molecule-substrate interactions can be controlled
by adjusting the van der Waals interaction between molecules
and substrate. For example, the molecule-substrate interaction
on graphite linearly increases with the length of the alkyl
chains.9,10
hydrogen bonding, van der Waals interactions) between the
solvent and the other molecules.13 The solvent also affects the
mobility of the molecules14 at the liquid-solid interface, e.g.,
by affecting the adsorption-desorption dynamics. In this
dynamic process, the mobility of the molecules is affected by
the solvation energy (molecule-solvent interaction) and possibly
also by solvent viscosity.15 Hence, for a successful rational
construction of 2D molecular networks, an accurate molecular
design taking into account these four interactions is essential.
In this contribution, we systematically study the formation
of 2D molecular networks of dehydrobenzo[12]annulene (DBA)
derivatives on highly oriented pyrolytic graphite (HOPG). Our
strategy to form 2D regular networks is based on choosing rigid
π-electron-conjugated frameworks with long alkyl chains as
molecular building-blocks. In this respect, DBA derivatives are
good candidates because of their planar π-electron-conjugated
framework, unique molecular shape, suitable core size to
favor directional alkyl chain interdigitation, and core symmetry
which fits the graphite lattice (Figure 1).16 Furthermore, their
synthetic versatility allows chemical modification of their
periphery.17
At the liquid-solid interface, solvent molecules play a
significant role in the network formation and have therefore a
strong effect on the network structures. Sometimes solvent
molecules are coadsorbed in the molecular network (solvent-
substrate interaction).11 Coadsorption of solvent molecules
critically depends among other factors on the size and shape of
the solvent molecules12 as well as the mode of interaction (e.g.,
Instead of using hydrogen bonds,5,7,18 we focus on directional
alkyl chain interdigitation and tune the strength of molecule-
molecule and molecule-substrate interactions by stepwise
elongation of the alkyl chains. For that purpose, in addition to
the previously synthesized decyl-substituted rhombic-shaped
bisDBA 1a and triangular-shaped DBA 2a, a series of DBA
derivatives with alkoxy chains of variable length 3a-e were
synthesized (Figure 1).
(6) (a) Schmidt, G. M. J. Pure Appl. Chem. 1971, 27, 647-678. (b) Crystal
Engeering: The Design of Organic Solid; Desiraju, G. R., Ed.; Elsevier:
New York, 1989. (c) Desiraju, G. R. Angew. Chem., Int. Ed. Engl. 1995,
34, 2311-2327. (d) Blagden, N.; Davey, R. J. Cryst. Growth Des. 2003,
3, 873-885.
(7) (a) Theobald, J. A.; Oxtoby, N. S.; Phillips, M. A.; Champness, N. R.;
Beton, P. H. Nature 2003, 424, 1029-1031. (b) Stepanow, S.; Lingenfelder,
M.; Dmitriev, A.; Spillmann, H.; Delvigne, E.; Lin, N.; Deng, X.; Cai, C.;
Barth, J. V.; Kern, K. Nature Mat. 2004, 3, 229-233. (c) Li, Z.; Han, B.;
Wan, L. J.; Wandlowski, Th. Langmuir 2005, 21, 6915-6928. (d) Ishikawa,
Y.; Ohira, A.; Sakata, M.; Hirayama, C.; Kunitake, M. Chem. Commun.
2002, 2652-2653. (e) Lu, J.; Lei, S. B.; Zeng, Q. D.; Kang, S. Z., Wang,
C.; Wan, L. J.; Bai, C. L. J. Phys. Chem. B 2004, 108, 5161-5165. (f)
Griessl, S. J. H.; Lackinger, M.; Jamitzky, F.; Markert, T.; Hietschold, M.;
Heckl, W. M. J. Phys. Chem. B 2004, 108, 11556-11560. (g) Griessl, S.
J. H.; Lackinger, M.; Jamitzky, F.; Markert, T.; Hietschold, M.; Heckl, W.
M. Langmuir 2004, 20, 9403-9407.
(8) (a) Liu, Y.; Lei, S.; Yin, S.; Xu, S.; Zheng, Q.; Zeng, Q.; Wang, C.; Wan,
L.; Bai, C. J. Phys. Chem. B 2002, 106, 12569-12574. (b) Gyarfas, B. J.;
Wiggins, B.; Zosel, M.; Hipps, K. W. Langmuir 2005, 21, 919-923. (c)
Jeong, K. S.; Kim, S. Y.; Shin, U. S.; Kogej, M; Hai, N. T. M.; Broekmann,
P.; Jeong, N; Kirchner, B.; Reiher, M.; Schalley, C. A. J. Am. Chem. Soc.
2005, 127, 17672-17685. (d) Schull, G.; Douillard, L.; Fiorini-Debuiss-
chert, C.; Charra, F.; Mathevet, F.; Kreher, D.; Attias, A.-J. Nano Lett.
2006, 6, 1360-1363.
(11) (a) Yablon, D. G.; Wintgens, D.; Flynn, G. W. J. Phys. Chem. B 2002,
106, 5470-5475. (b) Vanoppen, P.; Grim, P. C. M.; Ru¨cker, M.; De Feyter,
S.; Moessner, G.; Valiyaveettil, S.; Mu¨llen, K.; De Schryver, F. C. J. Phys.
Chem. 1996, 100, 19636-19641.
(12) Mamdouh, W.; Uji-i, H.; Ladislaw, J. S.; Dulcey, A. E.; Percec, V.; De
Schryver, F. C.; De Feyter, S. J. Am. Chem. Soc. 2006, 128, 317-325.
(13) Nath, K. G.; Ivasenko, O.; Miwa, J. A.; Dang, H.; Wuest, J. D. Nanci, A.;
Perepichka, D. F.; Rosei, F. J. Am. Chem. Soc. 2006, 128, 4212-4213.
(14) (a) Kim. K.; Plass, K. E.; Matzger, A. J. Langmuir 2003, 19, 7149-7152.
(b) Kim, K.; Plass, K. E.; Matzger, A. J. J. Am. Chem. Soc. 2005, 127,
4879-4887.
(15) (a) Venkataraman, B.; Breen, J. J.; Flynn, G. W. J. Phys. Chem. 1995, 99,
6608-6619. (b) Li, C.-J.; Zeng, Q.-D.; Wang, C.; Wan, L.-J.; Xu, S.-L.;
Wang, C.-R.; Bai, C.-L. J. Phys. Chem. B 2003, 107, 747-750. (c)
Lackinger, M.; Griessl, S.; Heckl, W. M.; Hietschold, M.; Flynn, G. W.
Langmuir 2005, 21, 4984-4988. (d) Kampschulte, L.; Lackinger, M.;
Maier, A.-K.; Kishore, R. S. K.; Griessl, S.; Schmittel, M.; Heckl, W. M.
J. Phys. Chem. B 2006, 110. 10829-10836. (e) Shao, X.; Luo, X.; Hu, X.;
Wu. K. J. Phys. Chem. B 2006, 110, 1288-1293.
(16) Furukawa, S.; Uji-i, H.; Tahara, K.; Ichikawa, T.; Sonoda, M.; De Schryver,
F. C.; Tobe, Y.; De Feyter, S. J. Am. Chem. Soc. 2006, 128, 3502-3503.
(17) Youngs, W. J.; Tessier, C. A.; Bradshaw, J. D. Chem. ReV. 1999, 99, 3153-
3180.
(9) (a) Smith, D. P. E.; Ho¨rber, H.; Gerber, C.; Binnig, G. Science 1989, 245,
43-45. (b) Smith, D. P. E.; Ho¨rber, J. K. H.; Binning, G.; Nejoh, H. Nature
1990, 344, 641-644. (c) Hara, M.; Iwakabe, Y.; Tochigi, K.; Sasabe, H.;
Garito, A. F.; Yamada, A. Nature 1990, 344, 228-230.
(10) (a) Paserba, K. R.; Gellman, A. J. J. Chem. Phys. 2001, 115, 6737-6751.
(b) Paserba, K. R.; Gellman, A. J. Phys. ReV. Lett. 2001, 86, 4338-4341.
(c) Mu¨ller, T.; Flynn, G. W.; Mathauser, A. T.; Teplyakov, A. V. Langmuir
2003, 19, 2812-2821.
(18) There are a number of papers about 2D molecular network through the
hydrogen bonding as connectivity. See, for example: (a) Griessl, S.;
Lackinger, M.; Edelwirth, M.; Hietschold, M.; Heckl, W. M. Single Mol.
2002, 3, 25-31. (b) Lei, S. B.; Wang, C.; Yin, S. X.; Wang, H. N.; Xi, F.;
Liu, H. W.; Xu, B.; Wan, L. J.; Bai, C. L. J. Phys. Chem. B 2001, 105,
10838-10841. (c) Otsuki, J.; Nagamine, E.; Kondo, T.; Iwasaki, K.;
Asakawa, M.; Miyake, K. J. Am. Chem. Soc. 2005, 127, 10400-10405.
9
16614 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006

Two-Dimensional Porous Molecular Networks
A R T I C L E S
Scheme 1. Synthesis of Hexaalkoxy DBA 3a-e
Deprotection of the trimethylsilyl group of the coupling products
was performed by stirring with potassium carbonate in methanol/
THF (1:1), giving 8a-e. For the construction of the annulene
framework, the copper-catalyzed cyclotrimerization of 8a-e was
employed.22 The reaction of 8a-e with copper iodide and
triphenylphosphine in DMF proceeded smoothly at 160 °C to
afford hexasubstituted DBAs 3a-e in modest yields.23
2. STM Observation of DBA Networks at the Liquid-
Solid Interface. First, the formation of DBA networks at the
1,2,4-trichlorobenzene (TCB)-graphite interface was probed for
1a, 2a, and 3a-e in order to evaluate the effect of the π-electron
core symmetry as well as the length of the alkyl chains on the
geometry of the 2D networks. In a next step, the spontaneous
monolayer formation was repeated in several solvents to
elucidate how the choice of solvent affects the structure of the
typical DBA networks formed at the TCB-HOPG interface,
i.e., the Kagome´ lattice of 1a, the honeycomb structure of 2a
and 3a, and the linear B array of 3d (see below). The choice of
these solvents was motivated by their different characteristics:
TCB as a polar aromatic solvent, 1-phenyloctane as a solvent
having both aromatic and aliphatic moieties, n-tetradecane as
an aliphatic solvent, and 1-octanoic acid as a polar alkylated
solvent. All experiments were performed at 20-22 °C.
2.1. Molecular Networks of DBA Derivatives 1a, 2a, and
3a-e in TCB. (a) Structure and Dynamic Behavior of 1a.
Figures 2a and 2c display STM images of a physisorbed
monolayer obtained from a solution of bisDBA 1a in TCB on
HOPG. In the STM images, aromatic moieties are often
observed to show a high tunneling efficiency allowing a
straightforward interpretation of the bright-dark contrast in the
images.24 Therefore, bright rhombic features correspond to the
π-electron-conjugated cores,25 which also agrees with the
molecular size. The striped features between the π-systems
correspond to alkyl chains. Surprisingly, a dynamic change of
the molecular alignment of 1a was observed; early on in a
measuring session, a linear network (linear A) was observed
(Figure 2a)26 which was then gradually replaced by a Kagome´
domain (Figure 2c). This observation suggests that the linear
A structure is a kinetically favored molecular pattern while the
Kagome´ network is the thermodynamically more stable one.14a
A model reflecting the molecular ordering of 1a according
to the linear A type packing is shown in Figure 2b. All molecules
are oriented in the same way; extended alkyl chains bridge the
gap between adjacent cores. This leads to alkyl chain interdigi-
tation at two sides (2 times two alkyl chains per core) while at
the other two sides, only one alkyl chain per core extends to
the adjacent core. All alkyl chains are aligned along two of the
main symmetry axes of graphite. A unit cell contains one
molecule of 1a, and the cell parameters are R ) 2.4 ( 0.1 nm,
â ) 2.6 ( 0.1 nm, and γ ) 101 ( 4°.
The solvent turned out to be an important factor as the 2D
patterns were dramatically affected by the nature of the solvent.
Previous studies of solvent effects were mainly focused on 2D
molecular networks held together by hydrogen bonding15a-d or
lamella-type structures of alkylated molecules stabilized by van
der Waals interactions.14,15e The present investigation highlights
the important role of solvent on the structure and stabilization
of molecular networks which are formed, not via hydrogen
bonding, but via the interaction between nonpolar groups based
upon van der Waals interactions, i.e., directional alkyl chain
interdigitation.
Results
1. Synthesis. Two sets of DBA derivatives with different core
shapes, rhombus and triangle, are selected for the STM
investigations. The synthesis of bisDBA 1a and DBA 2a having
six decyl chains was reported previously.19 To modify the alkyl
chain length of the triangle-shaped DBAs, hexaalkoxy DBAs
(3a-e) (R ) C10H21; 3a, C12H25; 3b, C14H29; 3c, C16H33; 3d,
C18H37; 3e) were synthesized from catechol. The synthesis of
DBAs 3a-e is outlined in Scheme 1. Catechol was alkylated
by treatment with the respective alkyl bromide to yield 1,2-
dialkoxybenzenes 5a-e. Iodination of 5a-e with periodic acid
and iodine in a mixture of acetic acid, water, and H2SO4 gave
1,2-dialkoxy-4,5-diiodobenzenes 6a-e.20,21 Diiodides 6a-e
were treated with (trimethylsilyl)acetylene (TMSA) in the
presence of copper iodide and Pd(PPh3)4 as catalysts to give
1,2-dialkoxy-3-iodo-4-(trimethylsilyl)ethynylbenzenes 7a-e.
A packing model of the Kagome´ network of 1a is presented
in Figure 2d. Again, the interaction between adjacent molecules
is based on alkyl chain interdigitation, now at all sides, involving
(22) Iyoda, M.; Sirinintasak, S.; Nishiyama, Y.; Vorasingha, A.; Sultana, F.;
Nakao, K.; Kuwatani, Y.; Matsuyama, H.; Yoshida, M.; Miyake, Y.
Synthesis 2004, 1527-1531.
(19) Sonoda, M.; Sakai, Y.; Yoshimura, T.; Tobe, Y.; Kamada, K. Chem. Lett.
2004, 972-973.
(20) Kovalenko, S. V.; Peabody, S.; Manoharan, M.; Clark, R. J.; Alabugin, I.
V. Org. Lett. 2004, 6, 2457-2460.
(23) Hexaalkoxy[12]DBAs 3a and 3c were reported in, ref. 21a.
(24) Lazzaroni, R.; Calderone, A.; Lambin, G.; Rabe, J. P.; Bre´das, J. L. Synth.
Met. 1991, 41, 525-528.
(21) Compounds 6a-6c were previously known. (a) Zhang, D.; Tessier, C. A.;
Youngs, W. J. Chem. Mater. 1999, 11, 3050-3057. (b) Zhou, Q.; Carroll,
P. J.; Swager, T. M. J. Org. Chem. 1994, 59, 1294-1301. (c) Ong, C. W.;
Liao, S.-C.; Chang, T. H.; Hsu, H.-F. J. Org. Chem. 2004, 69, 3181-
3185.
(25) The bright features also match the shape of the frontier orbitals of model
compound 1b (see Supporting Information).
(26) The yellow line in Figure 2a indicates the domain boundary that is originated
form the difference of the alkyl chains interdigitation pattern.
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16615

A R T I C L E S
Tahara et al.
Figure 2. STM images of bisDBA 1a physisorbed from TCB on HOPG and their model structures. (a) Linear A network of bisDBA 1a (Iscl ) 0.50 nA,
Vbias ) -1.04 V). The yellow line indicates a domain boundary. (b) Molecular model of the linear A network of 1a. (c) The Kagome´ network of bisDBA
1a (Iscl ) 0.50 nA, Vbias ) -1.04 V). Inset: Expansion of the central void of the Kagome´ structure. The yellow circle in the inset indicates the positions of
some bright spots, attributed to TCB molecules. The yellow arrow indicates a trapped bis-annulene molecule in the center of a void. (d) Molecular model
of the Kagome´ network of 1a. The main symmetry axes of graphite are indicated in the lower left corner of the STM images.
in total three alkyl chains per side (2 + 1). The alkyl chains
run parallel to the main graphite axes underneath. A unit cell
contains three molecules of 1a. The cell parameters of the
Kagome´ network are R ) 5.0 ( 0.1 nm, â ) 4.9 ( 0.1 nm,
and γ ) 120 ( 1°.
In the Kagome´ network, the cyclic arrangement of six
molecules leads to the formation of a void. The average diameter
of the void estimated from the distance between the π-electron
moieties of 1a having the same orientation across the void is l
) 3.2 ( 0.2 nm (Figure 2d). By taking into account the size of
the void, it is likely to be filled by six TCB solvent molecules
(inset in Figure 2c).16 In addition, during the measurements,
occasionally a fuzzy but bright spot appeared in the center of
such a void (yellow arrow in Figure 2c), which we attribute to
a trapped bis-annulene molecule. These observations suggest
that there is a competition between coadsorption of solvent
molecules and 1a itself in the voids.
(b) Structure of Molecular Network of 2a. Figure 3a
presents a typical STM image of a monolayer formed by
triangle-shaped compound 2a upon applying a drop of a solution
of 2a dissolved in TCB onto a freshly cleaved HOPG surface.
The bright triangles correspond to the π-conjugated cores of
2a. A honeycomb network is formed; each apex position of the
hexagon is occupied by a molecule of 2a and the sides of
neighboring triangles face each other and are linked by alkyl
chain interdigitation (Figure 3b) leading to appreciable inter-
molecular interactions. Furthermore, the alkyl chains are
orientated along the three main symmetry axes of graphite. The
cell parameters of the honeycomb network of 2a are R ) 3.9
( 0.1 nm, â ) 3.8 ( 0.1 nm, and γ ) 119 ( 1°. There are two
molecules in a unit cell. Again a regular pattern of voids is
observed, now as the result of the formation of cyclic hexamers
of 2a as part of the honeycomb network. The average diameter
of the void estimated from the distance between π-electron
systems of 2a having the opposite orientation across the void
is l ) 3.1 ( 0.1 nm (Figure 3b).
(c) Effect of Elongation of the Alkyl Chains of Triangular
DBA on the Network Structure. As the length of the alkoxy
chains of DBAs 3a-e increases (O-C (terminal) length varies
from 1.25 to 2.24 nm), the 2D ordering changes accordingly.
DBA 3a having decyloxy side chains forms a honeycomb
network (Figure 4a). The detailed structural features of the
honeycomb network of 3a are similar to those of 2a. There is
no clear difference between the physisorption behavior of
compounds 2a and 3a in TCB, except for a slight difference in
the unit cell parameters: R ) 4.2 ( 0.1 nm, â ) 3.9 ( 0.1 nm,
and γ ) 122 ( 1° for 3a; R ) 3.9 ( 0.1 nm, â ) 3.8 ( 0.2
nm, and γ ) 119 ( 1° for 2a.
Figure 4b shows a typical STM image of 3b at the TCB-
graphite interface. The honeycomb structure was dominantly
9
16616 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006

Two-Dimensional Porous Molecular Networks
A R T I C L E S
Figure 3. (a) An STM image of 2a physisorbed at the TCB-HOPG interface. (Iscl ) 0.50 nA, Vbias ) -0.99 V). (b) Molecular model of the honeycomb
domain of 2a. The main symmetry axes of graphite are indicated in the lower left corner of the STM image.
Figure 4. STM images (50 nm × 50 nm) of 3a-e at the TCB-HOPG interface. (a) Honeycomb network of 3a (Iscl ) 1.0 nA, Vbias ) -0.42 V). (b)
Honeycomb (bottom) and linear B type (top) networks of 3b (Iscl ) 0.60 nA, Vbias ) -0.28 V). The yellow line separates a honeycomb and a linear B type
domain. (c) Linear B type network of 3c (Iscl ) 0.65 nA, Vbias ) -0.18 V). Red circles indicate honeycomb structures observed at domain boundaries. (d)
Linear B type network of 3d (Iscl ) 0.65 nA, Vbias ) -0.65 V). A red circle indicates a honeycomb structure. (e) Linear B type network of 3e (Iscl ) 0.70
nA, Vbias ) -0.53 V).
observed. In this system, we observed a dynamic change from
a linear structure, called “linear B”, which corresponds to the
upper domain in Figure 4b to the honeycomb network covering
the largest area in this image (See also Figure 5b for a linear B
model and Supporting Information). However, a complete
conversion to the honeycomb network was within 4 h after
dropping a solution onto the HOPG surface never observed. In
most cases, a small area of the linear B type packing coexisted
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16617

A R T I C L E S
Tahara et al.
Figure 5. An STM image of a monolayer of 3c and molecular model of the network at the TCB-graphite interface. (a) Linear B type network (Iscl ) 0.65
nA, Vbias ) -0.25 V). Top view (b) and side view (c) of a tentative model representing the linear B type network of 3c. The main symmetry axes of graphite
are indicated in the lower left corner of the STM image.
Table 1. Molecular Networks and Unit Cell Parameters of DBA Derivatives 1a, 2a, and 3a-e Physisorbed at the TCB-HOPG Interface
unit cell parametersa
compound
predominant network
R
(nm)
â
(nm)
γ
(deg)
average APMb (nm2)
NACc on surface
1a
Linear A
Kagome´
2.4 ( 0.1
2.6 ( 0.1
4.9 ( 0.1
3.8 ( 0.2
3.9 ( 0.1
4.5 ( 0.1
5.4 ( 0.1
5.4 ( 0.1
101 ( 4
120 ( 1
119 ( 1
122 ( 1
119 ( 1
112 ( 1
105 ( 2
-
6.1
7.1
6.5
6.9
8.9
6.0
6.3
-
6
6
6
6
6
4
4
-
5.0 ( 0.1
3.9 ( 0.1
4.2 ( 0.1
4.5 ( 0.1
2.4 ( 0.1
2.4 ( 0.1
2a
3a
3b
3c
3d
3e
Honeycomb
Honeycomb
Honeycombd
Linear Bd
Linear B
Linear Be
-
-
a For each compound, unit cells are indicated in the figures in the main text or in the Supporting Information. b Area per molecule. c Number of alkyl
chains. d The linear B and the honeycomb structures occasionally coexist. e The cell parameters are not determined due to the low resolution of the STM
images.
with the predominant honeycomb network, which is attributed
to the elongation of the alkyl chains. Comparison of the unit
cell parameters of the honeycomb networks of 2a, 3a, and 3b
revealed that the angle (γ) is constant (ca. 120°) while the length
of the unit cell vectors R and â increases (from 3.8 to 4.5 nm)
upon going from alkyl to alkoxy and increasing the alkoxy chain
length (Table 1). Hence, the diameter (l) and area of central
void in such a hexagon is controlled by the alkyl chain length
(l ) 3.1 ( 0.1 nm for 2a, 3.4 ( 0.2 nm for 3a, and 3.7 ( 0.2
nm for 3b).
Molecule 3c which has six tetradecyloxy chains forms both
honeycomb and linear B type domains (Figure 4c). Though large
scale images show honeycomb structures near domain bound-
aries, the linear B network was predominantly observed. No
change to the honeycomb structure was observed within 5 h
after sample preparation. In linear B domains, molecules have
the same orientation in a given row while this orientation
alternates from row to row leading to a zigzag alignment. A
high-resolution image of 3c emphasizes the orientation of the
alkyl chains (Figure 5a). Four interdigitated alkyl chains between
cores are clearly observed. However, the other alkyl chains are
not observed presumably because they are not adsorbed on the
graphite surface but are exposed to the liquid phase (Figures
5b and 5c).
observed (Figures 4d and 4e). The length of the unit cell vectors
R and â are similar for 3c and 3d (2.4 and 5.4 nm, respectively);
however, the angle (γ) of 3d (105°) is smaller than that of 3c
(112°) (Table 1). In case of 3e, more randomly ordered areas
are observed (Figure 4e).
2.2. Solvent Effect on Molecular Networks of DBAs. (a)
Molecular Network of Rhombus-Shaped 1a. The solvent
effect on the structure of molecular DBA networks has been
examined by comparing their self-assembly in three different
organic solvents (1-phenyloctane, n-tetradecane, and 1-octanoic
acid) in addition to TCB described above. The DBA networks
are strongly solvent dependent. As mentioned above, in TCB
the linear A network is kinetically favored while the Kagome´
network is the thermodynamic stable one. Interestingly, another
type of linear structure (linear A′) appeared by changing the
solvent. The difference between the linear A and A′ structure
is the number of adsorbed alkyl chains on the surface and the
density of the molecular networks as described below.
In phenyloctane, the coexistence of Kagome´ and linear A′
structures was observed (Figure 6a). The structural details of
the linear A′ packing are discussed later.
Figure 6b shows an STM image of a monolayer of 1a
physisorbed from tetradecane on HOPG. In this solvent, small
domains of the linear A structure were observed. The network
type was confirmed by comparison of the unit cell parameters
(R ) 2.4 ( 0.1 nm, â ) 2.6 ( 0.1 nm, γ ) 97 ( 3°) with
Upon increasing the alkoxy chain length further (3d and 3e),
the same type of linear networks are also predominantly
9
16618 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006

Two-Dimensional Porous Molecular Networks
A R T I C L E S
Figure 6. (a) Coexistence of Kagome´ and the linear A′ domains of 1a physisorbed from phenyloctane on HOPG (Iscl ) 0.70 nA, Vbias ) -0.54 V). The
yellow line separates a Kagome´ and a linear A′ domain. (b) Linear A domains of 1a physisorbed from tetradecane on HOPG (Iscl ) 0.70 nA, Vbias ) -0.65
V). Unit cell parameters: R ) 2.4 ( 0.1 nm, â ) 2.6 ( 0.1 nm, γ ) 97 ( 3°. The yellow line indicates a domain boundary. (c) A high-resolution STM
image of the linear A′ type pattern formed upon physisorption of 1a from octanoic acid (Iscl ) 0.60 nA, Vbias ) -0.34 V). All molecules have the same
orientation. Unit cell parameters; R ) 2.5 ( 0.1 nm, â ) 1.9 ( 0.2 nm, γ ) 100 ( 2°. (d) Molecular model of the linear A′ pattern based on the STM image
in Figure 6c. (e) Another Linear A′ domain as the result of physisorption of 1a from octanoic acid on HOPG (Iscl ) 0.60 nA, Vbias ) -0.88 V). The
difference in the orientation of the molecules in adjacent rows is highlighted. This leads to slightly different unit cell parameters for nonequivalent rows. (f)
Molecular model of the linear A′ structure based on the STM image in Figure 6e. The main symmetry axes of graphite are also indicated in the STM images.
those observed for the linear A type in TCB (R ) 2.4 ( 0.1
nm, â ) 2.6 ( 0.1 nm, γ ) 101 ( 4°). In tetradecane, the
Kagome´ domain was not observed.
(b) Molecular Network of Triangular-Shaped DBA 2a. For
the decyl-substituted triangular compound 2a, the monolayers
show a more complex solvent dependence. Upon physisorption
from phenyloctane, dynamic changes in the monolayer structure
are observed. Soon after applying a drop on the substrate the
trimer structure, which is identical to the network observed in
octanoic acid (Figure 7e), appeared. The detailed structure of
the trimer network is discussed further in the text. This trimer
molecular network gradually transformed into a zigzag align-
ment (Figures 7a and 7b) and honeycomb structure. Most of
trimer domains disappeared within an hour. A zigzag row is
indicated in Figure 7a by the yellow line. Triangle-shaped bright
features correspond to the π-conjugated cores of DBA. At two
sides a core is linked to the adjacent ones by interdigitating
alkyl chains. At the third side, two alkyl chains are located
between the zigzag rows. Surprisingly, one phenyloctane
molecule is coadsorbed between these two non-interdigitated
alkyl chains (Figure 7b). This was confirmed by the observation
of small bright features that matched the size of the phenyl ring
of phenyloctane (see also model in Figure 7c). Unit cell
parameters are R ) 3.6 ( 0.2 nm, â ) 5.0 ( 0.2 nm, and γ )
135 ( 1°. There are two DBA molecules and two phenyloctane
molecules in a unit cell. Moreover, this structure gradually
changed to the honeycomb network, resulting in the coexistence
of honeycomb and zigzag networks. The formation of the
In octanoic acid, bisDBA 1a dominantly forms the linear A′
structure (Figures 6c and 6e) with sometimes a small Kagome´
domain (see Supporting Information). Normally, the conjugated
cores are oriented identically throughout a domain (Figure 6c).
The corresponding molecular model is shown in Figure 6d. The
unit cell parameters are R ) 2.5 ( 0.1 nm, â ) 1.9 ( 0.2 nm,
and γ ) 100 ( 2°. Sometimes, however, the orientation of the
conjugated cores in adjacent rows is different as shown in Figure
6e and the molecular model in Figure 6f. The same structural
variation is also observed in the linear A′ type domains for
monolayers physisorbed from phenyloctane (Figure 6a). How-
ever, the overall effect on the monolayer structure is minor
(slight difference in the unit cell parameters for nonequivalent
rows): the main aspect of the linear A′ pattern is that four alkyl
chains per molecule are adsorbed and the monolayer is stabilized
by alkyl chain interdigitation.
In both linear A and A′ networks, adjacent cores are
connected via four interdigitating alkyl chains. In the linear
A structure, the two remaining alkyl chains are adsorbed
on the graphite surface, parallel to one of its symmetry axes,
while those two alkyl chains in the linear A′ structure are
desorbed.
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16619

A R T I C L E S
Tahara et al.
Figure 7. (a) An STM image of a monolayer of 2a physisorbed from phenyloctane (Iscl ) 0.85 nA, Vbias ) -0.133 V). Unit cell parameters; R ) 3.6 (
0.2 nm, â ) 5.0 ( 0.2 nm, γ ) 135 ( 1°. The yellow line indicates a zigzag row. (b) An STM image of a 2a pattern formed upon physisorption from
phenyloctane (Iscl ) 1.0 nA, Vbias ) -0.59 V). Yellow “boxes” indicate coadsorbed phenyloctane molecules. The yellow line marks a zigzag row. (c)
Molecular model structure of the zigzag network observed in phenyloctane. (d) An STM image of 2a physisorbed from tetradecane (Iscl ) 1.0 nA, Vbias
-0.78 V). Unit cell parameters; R ) 4.0 ( 0.2 nm, â ) 4.0 ( 0.1 nm, γ ) 122 ( 2°. (e) An STM image of 2a in octanoic acid (Iscl ) 0.90 nA, Vbias
)
)
-0.66 V). Unit cell parameters; R ) 5.9 ( 0.2 nm, â ) 5.7 ( 0.2 nm, γ ) 122 ( 1°. (f) Tentative model of the trimer network of 2a in octanoic acid. The
yellow arrow indicates a molecule in the center of the hexagon. The main symmetry axes of graphite are also indicated in the lower right part of the STM
images.
thermodynamically favored honeycomb network requires the
expulsion of coadsorbed phenyloctane molecules and interdigi-
tation of all alkyl chains. Thus, coadsorption of the solvent
molecules first led to the formation of the zigzag network from
the kinetically favorable trimer structure. Consequently, de-
sorption of the coadsorbed solvent molecules led to the
honeycomb network. In spite of the similar alkyl chain length
of phenyloctane and octanoic acid, the latter one was not
coadsorbed presumably because of homodimer formation in-
volving hydrogen bonding.27
Figure 7d shows the honeycomb network of 2a formed upon
self-assembly at the tetradecane-HOPG interface. There is no
clear difference between the honeycomb networks observed in
TCB and those formed in tetradecane as indicated by their
respective unit cell parameters (R ) 4.0 ( 0.2 nm, â ) 4.0 (
0.1 nm, γ ) 122 ( 2° for tetradecane and R ) 3.9 ( 0.1 nm,
â ) 3.8 ( 0.2 nm, γ ) 120 ( 1° for TCB).
the other two alkyl chains facing the center of the trimer could
not be resolved, suggesting they are too mobile or directed to
the liquid phase. Due to the limited degree of alkyl chain
interdigitation, molecule-molecule interactions in the trimer
networks are expected to be weaker compared to those stabiliz-
ing the honeycomb network. Fuzzy spots were observed at the
center of the hexagons (Figure 7e, yellow arrow), indicating
the adsorption of another 2a molecule. This structure did not
change to other structures even after 5 h at 20-22 °C.
(c) Molecular Networks of Decyloxy-Substituted Trian-
gular DBA 3a. In contrast to decyl-substituted 2a, compound
3a forms a hexagonal-like structure in three solvents (pheny-
loctane, tetradecane, and octanoic acid). Figure 8b represents a
typical STM image of 3a physisorbed from tetradecane on
HOPG, revealing the formation of a hexagonal packing.
Basically, this hexagonal packing reflects the structure of a
honeycomb network and one molecule of 3a in each central
void of the honeycomb. These structural features were confirmed
by the observation of the ordering of the alkyl chains. The
central molecules in the hexagons appear fuzzy most probably
due to their dynamics. In a large hexagonal domain, the
adsorption-desorption process of the central molecules was
indeed observed in a series of STM of the same area images
recorded sequentially (see Supporting Information). The hex-
agonal structures filled with fuzzy molecular images in the center
Figure 7e is a typical STM image of an adlayer of 2a on
graphite physisorbed from octanoic acid. Three triangular bright
features, the π-conjugated cores, form triangular-shaped trimers,
as indicated in yellow in Figure 7e. Figure 7f shows a tentative
model. Though four alkyl chains per molecule are clearly visible,
(27) (a) Loveluck, G. J. Phys. Chem. 1960, 64, 385-387. (b) Czarnecki, M. A.
Chem. Phys. Lett. 2003, 368, 115-120. (c) Eliason, T. L.; Havey, D. K.;
Vaida, V. Chem. Phys. Lett. 2005, 402, 239-244.
9
16620 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006

Two-Dimensional Porous Molecular Networks
A R T I C L E S
Figure 8. STM images of 3a physisorbed at the liquid-HOPG interface. The yellow arrows points to molecules at the center of a hexagon. (a) A hexagonal
domain of 3a physisorbed from phenyloctane (Iscl ) 0.65 nA, Vbias ) -0.55 V). Unit cell parameters; R ) 4.2 ( 0.1 nm, â ) 4.1 ( 0.1 nm, γ ) 114 (
1°. (b) A hexagonal domain of 3a physisorbed from tetradecane (Iscl ) 0.60 nA, Vbias ) -0.93 V). Unit cell parameters; R ) 4.1 ( 0.2 nm, â ) 4.1 ( 0.1
nm, γ ) 121 ( 2°. (c) A hexagonal domain of 3a physisorbed from octanoic acid (Iscl ) 0.75 nA, Vbias ) -0.60 V). The main symmetry axes of graphite
are also indicated in the STM images.
Figure 9. STM images and a tentative packing model of 3d at the liquid-HOPG interface. (a) Coexistence of the linear B and linear B′ structures upon
physisorption of 3d from phenyloctane (50 nm × 50 nm, Iscl ) 0.80 nA, Vbias ) -0.81 V). (b) A random structure in tetradecane (Iscl ) 0.90 nA, Vbias
)
-0.55 V). (c) A linear B′ domain of 3d at the octanoic acid-graphite interface (Iscl ) 0.60 nA, Vbias ) -0.52 V). Molecular model of the linear B′ domain
as observed in tetradecane is superimposed. The main symmetry axes of graphite are also indicated in the lower part of the STM images.
of the voids also appear in phenyloctane (Figure 8a) and octanoic
acid (Figure 8c).28 The unit cell parameters are comparable for
the honeycomb and hexagonal structures (R ) 4.2 ( 0.1 nm, â
) 3.9 ( 0.1 nm, γ ) 122 ( 1° for TCB (honeycomb), R ) 4.1
( 0.2 nm, â ) 4.1 ( 0.1 nm, γ ) 121 ( 2° for tetradecane
(hexagonal), and R ) 4.2 ( 0.1 nm, â ) 4.1 ( 0.1 nm, γ )
114 ( 1° for phenyloctane (hexagonal)).
(d) Molecular Networks of Hexadecyloxy-Substituted
Triangular DBA 3d. Solvent-induced changes of the network
structure were also observed for 3d. As described above, at the
TCB-HOPG interface, the linear B structure was observed
(Figure 4d). After dropping a phenyloctane solution of 3d on
HOPG, the coexistence of the linear B network and a different
linear structure (linear B′) was observed (Figure 9a, see
Supporting Information). The linear B′ structure was dominantly
observed in octanoic acid. The structural details of the linear
B′ structure are discussed later. In tetradecane, monolayers
become more complex: random structures together with small
linear B′ domains are observed (Figure 9b).
interface. The linear B′ structure consists of molecular rows with
an alternating orientation of the molecules in a given row. The
intermolecular distance between adjacent molecules in a row
is not very well defined but varies slightly. Four alkyl chains
per molecule take part in the alkyl chain interdigitation between
adjacent rows. The position of the other two alkyl chains could
not be determined presumably due to the free movement of the
alkyl chains in the solution or because of their mobility on the
graphite surface (Figure 9c). All visualized alkyl chains in a
given domain lie parallel to one of the three main symmetry
axes of graphite.
Discussion
Figure 10 summarizes the different types of molecular
networks of DBAs 1a, 2a, 3a, and 3d and their model structures.
Rhombic bisDBA 1a shows three different structures (Figure
10b): Kagome´, linear A, and linear A′. Also decyl-substituted
triangular DBA 2a forms three different networks (Figure
10c): honeycomb, zigzag, and trimer. The alkoxy-substituted
triangular DBAs 3a-e show four different structures (Figures
10c, 10d, and 10e): the honeycomb, hexagonal (honeycomb
void filled (not shown)), linear B, and linear B′ structures. Also
the solubility of the compounds in the different solvents is
Figure 9c shows a typical linear B′ structure with a small
random domain of 3d formed at the octanoic acid-graphite
(28) In phenyloctane, small part of the zigzag structure of 3a was also observed.
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16621

A R T I C L E S
Tahara et al.
Figure 10. (a) Schematic representation of the different types of interfacial monolayer structures of DBAs 1a, 2a, 3a, and 3d along with the properties of
the solvents used. The major network type is colored in red. The arrows indicate the dynamic conversions. The index m represents the number of alkyl
chains adsorbed per molecule on the graphite surface. The index n refers to the number of alkyl chain interdigitation sites per molecule. ‘m’ and ‘NAC’
(Table 1) are equivalent. Large values for m and n indicate a highly ordered molecular alignment. Solubilities of the compounds in each solvent (g/L) are
indicated in blue (refers to the amount of compound dissolved after 12 h at room temperature). (b) Models of the three network structures of 1a (Kagome´,
linear A, and linear A′). (c) Models of the three network structures of DBA 2a (honeycomb, zigzag, and trimer). (d) A model of the linear B structure of
3d. (e) A model of the linear B′ structure of 3d. The brighter rods indicate alkyl chains which are not adsorbed on the surface.
rhombi and triangles are shared with their neighbors. It should
be pointed out that in the networks formed from TCB (the
Kagome´ network of 1a and the honeycomb network of 2a), all
sides of the rhombic or triangular π-electron systems in the
Kagome´ and honeycomb networks are shared with their
neighbors via directional alkyl chain interdigitation (Figures 2d
and 3b). The linear type alignment of rhombi (Figure 11b), i.e.,
the linear A structure, is not the thermodynamically most stable
Figure 11. Close packing of rhombi (green) (a, b) and triangles (blue) (c).
Connecting the gravity centers of adjacent rhombi or triangles (shown as
yellow lines) gives rise to the formation of the Kagome´ network in case of
rhombi (a) and the honeycomb network in case of triangles (c).
structure as the degree of directional alkyl chain interdigitation
is lower compared to the Kagome´ structure (The details will
be discussed in the next section). The shape of the π-electron
system thus determines the directional interaction based upon
full alkyl chain interdigitation, giving rise to two different
structures, the Kagome´ (for the rhombic π-electron systems)
and the honeycomb (for the triangular π-electron systems).
There are some reports, mainly involving alkoxylated aro-
matic compounds, on 2D patterns formed as the result of alkyl
chain interdigitation as main intermolecular interaction.30 How-
ever, the question arises why both DBAs form regular patterns
via directional alkyl chain interdigitation. This question can be
answered by comparison of molecular networks of triangle-
indicated (see also Supporting Information). Here we also
introduce indices m and n which refer to molecule-substrate
and molecule-molecule interactions, respectively. The index
m represents the number of alkyl chains adsorbed per molecule
on the graphite surface. The index n refers to the number of
alkyl chain interdigitation sites per molecule.
1. Core Shape-Directed Kagome´ and Honeycomb Network
Formation Involving Directional Alkyl Chain Interdigitation.
The shape of the π-electron-conjugated systems is one of the
important factors which determines the alignment of the DBA
molecules on the surface.16 Interestingly, the observed direction
of each π-electron-conjugated system in 2D DBA networks
corresponds very well to the ideal dense packing model of
rhombi and triangles. Figure 11 shows the “ideal” dense packing
mode of rhombi (Figures 11a and 11b) and triangles (Figure
11c).29 The dense packing models show that all sides of the
(29) The dense packing models of the rhombi and triangle shown in Figure 11
highlight the difference of the positions of the directional interacting sites.
These models are not related with the surface coverage of the whole
molecular network, i.e., the principle of closest packing, as discussed later.
(30) (a) Qiu, X.; Wang, C.; Zeng, Q.; Xu, B.; Yin, S.; Wang, H.; Xu, S.; Bai,
C. L. J. Am. Chem. Soc. 2000, 122, 5550-5556. (b) Merz, L.; Gu¨ntherodt,
H.-J.; Scherer, L. J.; Constable, E. C.; Housecroft, C. E.; Neuburger, M.;
Hermann, B. A. Chem. Eur. J. 2005, 11, 2307-2318. (c) Xu, S.; Zeng,
Q.; Lu, J.; Wang, C.; Wan, L.; Bai, C. L. Surf. Sci. 2003, 538, L451-
L459.
9
16622 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006

Two-Dimensional Porous Molecular Networks
A R T I C L E S
average area per molecule on the surface to compare the
substrate coverage of the molecules. For different networks
consisting of the same molecules, the density of the molecular
network can be estimated by comparison of the average area
per molecule. The smaller the value of the average area per
molecules, the higher the density of the molecular network is
and vice versa.
The careful comparison between the linear A structure (Figure
2a) and the Kagome´ structure (Figure 2c) disclosed that the
dynamic process most likely originates from the difference in
molecule-molecule interaction. Because the average area per
molecule in the Kagome´ structure (7.1 nm2) is larger than in
the linear A structure (6.1 nm2), we can rule out that in this
particular case the surface coverage plays a critical role in
defining the network structure. In addition, molecular modeling
confirmed that the orientation and conformation of the alkyl
chains of 1a are similar in both the linear A (Figure 2b) and
the Kagome´ (Figure 2d) networks. Moreover, molecule-
substrate and molecule-solvent interactions are similar for both
networks. On the other hand, there is a clear difference between
both network structures in terms of molecule-molecule interac-
tions (index n in Figure 10). The molecule-molecule interac-
tions are believed to be stronger in the Kagome´ structure than
in the linear A structure as only in the Kagome´ structure the
alkyl chains are fully interdigitated (n ) 4 for the Kagome´
network and n ) 2 for the Linear A network). For this reason,
in combination with the solvent effect of TCB described later,
the linear A structure gradually changed into the Kagome´
network which appeared as the thermodynamically favored
structure.
3. Effect of Enhancement of Molecule-Substrate Interac-
tions as a Result of Elongation of Alkoxy Chains of
Triangular DBAs. Network structures of aromatic compounds
with alkyl substituents on graphite are reported to change by
changing the alkyl chain length.30c,31,34 For the 2D networks of
triangle-shaped DBA derivatives 3a-e in TCB, the linear B
structure gradually becomes predominant with increasing alkyl
chain length (from C14 on) due to the increase of the van der
Waals interactions in terms of molecule-molecule and molecule-
substrate interactions as represented schematically in Figure 13.
The dynamic behavior and coexistence of the honeycomb and
linear B structures observed for 3b (C12) indicate that both
structures have a comparable free energy. For molecules with
longer chains than 3b, more stable linear B type domains were
observed. By comparison with the triphenylene derivatives
having six alkoxy chains, it appears that the transition from the
hexagonal to the linear structures happens at the same chain
length (C14).31a A further increase in the length of the alkyl
chains leads to a gradual misalignment of these alkyl chains
with respect to the main symmetry axes of graphite due to an
increased interaction. In the case of 3e, more randomly oriented
small domains of the linear type network were observed than
for 3c and 3d as shown in Figure 4e. Also triphenylenes with
longer alkyl chains show the formation of distorted networks.31a
This is the result of strong molecule-substrate interactions,
which restricts the in-plane mobility and the desorption-
adsorption dynamic behavior of the molecules.
Figure 12. Schematic drawings of the effect of core size on alkyl chain
interactions and interdigitation. (a) Triangles with a suitable core size for
directional alkyl chain interdigitation with a partner. (b) The core size is
too small for directional alkyl chain interdigitation. A side-by-side alignment
of alkyl chains is observed.
shaped triphenylene derivatives with those of DBAs 3a-e
having the same number and length of alkoxy chains.31 For
example, it has been reported by Bai et al. that triphenylene
derivative 4a with decyloxy side chains shows a hexagonal
packing pattern.31 However, there is a clear difference between
the packing patterns of the corresponding DBA 3a (Figure 3b)
and the triphenylene derivative 4a. Even though the hexagonal
network of triphenylene 4a involves alkyl chain interactions
between neighboring molecules, no interdigitation of the alkyl
chains is observed because of the small core size. In fact, the
closest interatomic distance between two alkoxy oxygen atoms
estimated from the optimized geometries of the two model
compounds (3f and 4b, R ) OMe: obtained by DFT calcula-
tions at the B3LYP/6-31g* level of theory) is 0.98 and 0.73
nm, for the annulene and triphenylene, respectively (Figure 1).32
Moreover, this interatomic distance (0.98 nm) for DBA 3a
relates well with the typical distance between closely packed
alkyl chains (0.40-0.45 nm).33 Therefore, DBA 2a can interact
with adjacent molecules via directional alkyl chain interdigitation
(Figure 12a), while triphenylene 4a cannot accommodate alkyl
chains of neighboring molecules between its own alkyl chains
(Figure 12b). Alkyl chain interdigitation leads to enhanced
molecule-molecule interactions.
Hence, core symmetry, size, and location and orientation of
alkyl or alkoxy groups at the periphery of π-electron-conjugated
systems are key factors to determine network structure.
2. Dynamic Changes Driven by the Molecule-Molecule
Interactions in the Network Formation of 1a in TCB.
Molecules generally prefer to pack as densely as possible to
maximize the substrate coverage, which is favored for enthalpic
reasons (i.e., the principle of closest packing).15d We used the
(31) (a) Wu, P.; Zeng, Q.; Xu, S.; Wang, C.; Yin, S.; Bai, C. L. ChemPhysChem
2001, 12, 750-754. (b) Charra, F.; Cousty, J. Phys. ReV. Lett. 1998, 80,
1682-1685.
(32) Optimization of model compounds were performed under D2h symmetrical
constrain for 1b and D3h symmetrical constrain for 2b, 3f and 4b.
(33) (a) McGonigal, G. C.; Bernhardt, R. H.; Thomson, D. J. Appl. Phys. Lett.
1990, 57, 28-30. (b) Rabe, J. P.; Buchholz, S. Science 1991, 253, 424-
427. (c) Herwig, K. W.; Matthies, B.; Taub, H. Phys. ReV. Lett. 1995, 75,
3154-3157. (d) Tao, F.; Bernasek, S. L. J. Am. Chem. Soc. 2005, 127,
12750-12751.
(34) (a) Askadskaya, L.; Boeffel, C.; Rabe, J. P. Bunsenges. Phys. Chem. 1993,
97, 517-521. (b) Ito, S.; Wehmeier, M.; Brand, J. D.; Ku¨bel, C.; Epsch,
R.; Rabe, J. P.; Mu¨llen, K. Chem. Eur. J. 2000, 6, 4327-4342. (c) Mena-
Osteriz, E. AdV. Mater. 2002, 14, 609-616.
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16623

A R T I C L E S
Tahara et al.
disfavor or prevent the rearrangement from a kinetically favored
pattern to a thermodynamically stable structure. Also it is well
documented that the solubility is closely related to the molecule-
solvent interaction (solvophilic effect).38 In addition, the viscos-
ity of the solvent may also affect the out-of-plane (adsorption-
desorption dynamics) mobility of the molecules.15 Higher
solvent viscosity would make the rearrangement process slower.
(b) Solvent Effects for the Network Formation of Decyl-
Substituted Rhombic 1a and Triangular 2a. Here we discuss
the solvent dependence observed for decyl-substituted bisDBA
1a and DBA 2a. In aromatic solvents (TCB and phenyloctane),
the networks formed by both compounds mainly consist of
Kagome´ and honeycomb structures, respectively. As a result
of strong intermolecular interaction, highly ordered 2D networks
are formed. In the aromatic solvents, the thermodynamically
stable structures are observed because molecular dynamics
should be favored due to the solvophilic effects. Indeed, DBAs
1a and 2a are very soluble in TCB and phenyloctane as listed
in Figure 10a (1a: 7.3 g/L in TCB and 0.81 g/L in phenyloctane,
2a: 350 g/L in TCB and 130 g/L in phenyloctane). The observed
dynamic behavior of 1a in TCB (transition from the linear A
network to the Kagome´ network) and 2a in phenyloctane (from
the trimer network via the zigzag network to the honeycomb
network) is consistent with this view. In tetradecane, the
intermediate case in terms of the solvophilic effect, the properties
of the molecular network are somewhat similar to those observed
in the aromatic solvents. The linear A structure (m ) 6, n ) 2)
of 1a and the honeycomb structure (m ) 6, n ) 3) of 2a were
observed. The persistence of the linear A structure in tetradecane
may be ascribed to the low mobility of 1a in this solvent. The
intermediate solvophillic effect of tetradecane is also supported
by the solubility measurements of 1a in each solvent as shown
in Figure 10a (0.15 g/L in tetradecane). In contrast, in octanoic
acid remarkably different network structures, the linear A′
structure (m ) 4, n ) 2) of 1a and the trimer structure (m ) 4,
n ) 2) of 2a, were observed. Both networks do not show the
maximum possible alkyl chain interdigitation interactions (n )
2) and two alkyl chains per molecule do not interact strongly
with graphite or are directed to solvent (m ) 4), leading to a
limited stability. As a result, the network density is higher than
in aromatic solvents. Indeed, the average area per molecule of
1a is 7.1 nm2 for the Kagome´ network in TCB and 4.7 nm2 for
the linear A′ network in octanoic acid.39 For the molecular
network of 2a, the average area per molecule is 6.5 nm2 for the
honeycomb network in TCB and 4.8 nm2 for the trimer network
in octanoic acid. We assume that the rearrangement process of
the molecular networks is restricted in octanoic acid because
of the low solubility of the DBAs (1a: 0.13 g/L, 2a: 19 g/L).
However, there is no significant difference in the solubilizing
properties between tetradecane and octanoic acid for both
compounds. Though the reason for this significant solvent effect
is unclear, one of the possibilities that may lead to the observed
differences is the solvent viscosity.40
Figure 13. Schematic view of alkyl chain effects in the resulting network
structures. The brighter rods indicate alkyl chains which are not adsorbed
on the surface.
4. Solvent Effects on the Network Formation of DBA. (a)
The Role of the Solvent. Solvent-molecule interactions play
an important role in the self-assembly of DBAs. All solvents
used (TCB, phenyloctane, octanoic acid,35 and tetradecane4b,33a,36
)
are known not to form stable monolayers on graphite at room
temperature (i.e., weak solvent-substrate interactions). There-
fore, the presence of a solvent monolayer can be safely excluded
in these solvents. The fact that Kagome´ patterns of 1a and
honeycomb domains of 2a are observed in several structurally
noncorrelated solvents indicates that these patterns are not
stabilized by the coadsorption of solvent molecules in the
network voids. Therefore, we assume that these nanoporous 2D
networks are the thermodynamically favored structures.
At the liquid-solid interface, there is an equilibrium between
the molecules which stay on the surface (adsorb) and those
which go back into the solution (desorb).15 The overall mobility
(in-plane and out-of-plane) of the molecules is important for
the transition of the kinetically favored to the thermodynamically
favored structure(s). For the in-plane as well as out-of-plane
dynamics, the solvents play a significant role. Since the
molecules have a solvent-dependent solvation energy, not only
the thermodynamic stability of the patterns but also kinetic
factors such as the desorption rate and in-plane diffusion rate
should be solvent-dependent. In general, for π-electron-
conjugated compounds with peripheral alkyl chains, the solvent-
molecule interaction can be described in terms of solvophobic
and solvophilic effects. These effects have frequently been used
to interpret the formation of aggregates of such molecules in
solution.37 Therefore, it seems reasonable to assume that
solvophobic effects between DBAs and poor solvents would
(35) The desorption energy of the octanoic acid dimer from the graphite surface
is lower than that of tetradecane, see: Ref. 10c.
(36) (a) Katsonis, N.; Marchenko, A.; Taillemite, S.; Fichuo, D.; Chouraqui,
G.; Aubert, C.; Malacria, M. Chem. Eur. J. 2003, 9, 2574-2581. (b) Bucher,
J. P.; Roeder, H.; Kern, K.; Surf. Sci. 1993, 289, 370-380. (c) Perronet,
K.; Charra, F. Surf. Sci. 2004, 551, 213-218.
(37) (a) Smithrud, D. B.; Diederich, F. J. Am. Chem. Soc. 1990, 112, 339-343.
(b) Cram, D. J.; Choi, H.-J.; Bryant, J. A.; Knobler, C. B. J. Am. Chem.
Soc. 1992, 114, 7748-7765. (c) Lahiri, S.; Thompson, J. L.; Moore, J. S.
J. Am. Chem. Soc. 2000, 122, 11315-11319. (d) Ho¨ger, S.; Bonrad, K.;
Mourran, A.; Beginn, U.; Mo¨ller, M. J. Am. Chem. Soc. 2001, 123, 5651-
5659. (e) Tobe, Y.; Utsumi, N.; Kawabata, K.; Nagano, A.; Adachi, K.;
Araki, S.; Sonoda, M.; Hirose, K.; Naemura, K. J. Am. Chem. Soc. 2002,
124, 5350-5364. (f) Kastler, M.; Pisula, W.; Wasserfallen, D.; Pakula, T.;
Mu¨llen, K. J. Am. Chem. Soc. 2005, 127, 4286-4296. (g) Terech, P.; Weiss,
R. G. Chem. ReV. 1997, 97, 3133-3160.
In conclusion, the solvent affects the speed of conversion from
a kinetically favored structure to a thermodynamically preferred
pattern. The rate of such a rearrangement process is most likely
(38) (a) Anslyn, E. V.; Dougherty, D. A. Modern Physical Organic Chemistry;
University Science Books: Sausalito, CA, 2004; pp 153-157. (b) Shinoda,
K. Principles of Solution and Solubility; Marcel Dekker: New York, 1978.
(39) The addition of the area occupied by two alkyl chain (<1.2 nm2) to the
average area per molecules of the linear A′ increases the area to ca. 5.9
nm2, which is also smaller than that of the Kagome´ network.
9
16624 J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006

Two-Dimensional Porous Molecular Networks
A R T I C L E S
dominated by the solubilizing properties of the solvent. Ad-
ditionally, the viscosity of the solvent can also affect the
rearrangement process in the poor solvents. DBAs tend to form
a kinetically favored, less ordered structure initially. If the
mobility of the DBAs is large enough, the DBA networks will
rearrange to more ordered structures despite the fact that the
surface coverage is reduced.
The core symmetry and position of the alkyl or alkoxy chains
determine the molecular ordering. In combination with the core
size, these structural properties favor the formation of nonpolar
porous networks thanks to the directional interdigitation of alkyl
chains as well as the registry of the alkyl chains with the main
symmetry axes of the substrate. Because of the strong directional
alkyl chain interdigitation, 2D open networks with voids are
formed. Tuning molecule-substrate interactions, realized for
instance by changing the length of the alkyl chains, leads to
different molecular networks. Therefore, molecule-molecule
and molecule-substrate interactions can be controlled by
rational molecular design. The synthetic versatility of DBA
derivatives22,41 will allow us to make various molecular networks
with specific topologies. Moreover, the formation of regularly
spaced voids in the Kagome´ network of bisDBA 1a and the
honeycomb structures of DBAs (2a, 3a, and 3b) suggests that
the central free space can be used as the host space for guest
molecules which otherwise would not adsorb on graphite.7,8
The solvent plays a significant role in the formation of DBA
networks, basically by affecting the number of adsorbed alkyl
chains and the degree of alkyl chain interdigitation. In addition
to the well-established co-adsorption effects of solvents, the
present study highlights other aspects. The solvent-induced
polymorphism discloses that perfect molecular networks (maxi-
mal directional alkyl chain interdigitation) can be achieved in
good aromatic solvents. On the other hand, the molecular
network partially loses this stabilizing interaction in poor
solvents, resulting in a decrease of network order. These solvent
effects on the 2D networks can be attributed to solvophobic/
solvophilic interactions between molecules and solvents during
the formation of molecular networks. We propose here that the
solubilizing properties of the solvents critically change the rate
of the rearrangement processes which transform the initially
formed kinetically favored structure into the thermodynamically
stable final pattern. These insights are useful in the area of “2D
crystal engineering.”
(c) Effects of Ether Functionalities. A notable difference
in the solvent effect between the molecular ordering of alkyl-
substituted 2a and that of alkoxy-substituted 3a was observed,
though both have the same π-conjugated core and identical alkyl
chain lengths. In contrast to 2a which forms the honeycomb
network in TCB and tetradecane, the trimer f zigzag f
honeycomb networks in phenyloctane, and the trimer structure
in octanoic acid, the alkoxy derivative 3a forms hexagonal-
like networks in all solvents; the honeycomb network in TCB
and the hexagonal network in the other solvents. Both networks,
so also the hexagonal network, are characterized by the
honeycomb alignment as the basic motif. The differences with
the alkyl derivatives most likely originate from the conforma-
tional flexibility of the ether group located between the aromatic
core and the alkyl chains. Indeed, the solubility of 3a is higher
than that of 2a. Therefore, the network transformations of 3a
might become easier than those of 2a, and accordingly 3a forms
optimal alkyl chain interdigitation in all solvents. These results
suggest that the mobility of the alkyl chains may also affect
the molecular mobility thereby affecting the 2D network
formation and the outcome of the self-assembly process.
(d) Solvent Effects for the Network Formation of Hexa-
decyloxy-Substituted Triangular 3d. Hexadecyl-substituted 3d
shows complex patterns upon changing the solvents. In aromatic
solvents, the linear B structure is formed. Linear B′ and
amorphous like random patterns are observed in tetradecane and
octanoic acid. It is expected that the linear B network (m ) 4,
n ) 2) of 3d is thermodynamically more preferred than the
linear B′ (m ) 4, n ) 2) or the random patterns. Even though
both networks are characterized by the same index numbers (m
) 4, n ) 2), the linear B′ pattern is considered to be the less
ordered and less stable one as this packing is strictly spoken
not 2D crystalline (slight positional differences of the triangular
cores) and the domain sizes are small.
Acknowledgment. This work was supported by CREST of
JST (Japan Science and Technology Agency) and a Grant-in-
Aid for Scientific Research from the Ministry of Education,
Culture, Sports, Science, and Technology, Japan. S.F. is grateful
to the JSPS Postdoctoral Fellowships for Research Abroad.
S.D.F. and F.D.S. thank the Federal Science policy through
IUAP-V-03, the Institute for the promotion of innovation by
Science and Technology in Flanders (IWT), and the Fund for
Scientific Research-Flanders (FWO). This paper is dedicated
to Professor Philip E. Eaton on the occasion of his 70th birthday.
As shown in Figure 10a, the solubility of 3d in tetradecane
and octanoic acid is much lower compared to the solvents
containing aromatic groups (5.0 g/L for tetradecane and 18 g/L
for octanoic acid). Therefore, lack of favorable solvation
interactions and the presence of strong molecule-substrate
interactions, due to the long alkyl chains, restrict the dynamic
rearrangement process. As a result, kinetically favored though
less ordered patterns are observed in poor solvents.
Supporting Information Available: Synthetic procedures for
3a-e, experimental details, STM images of 1a, 2a, and 3a-e,
optimized structures of model compounds 1a, 2b, 3f, and 4b.
This material is available free of charge via the Internet at
http://pubs.acs.org.
Summary
In summary, we have accomplished STM measurements of
a series of DBA derivatives at the liquid-graphite interface.
JA0655441
(40) Viscosities of the solvents at 20 °C are 1.41 cP for TCB, 2.58 cP for
phenyloctane, 2.33 cP for tetradecane, and 5.83 cP for octanoic acid. (a)
Ducoulombier, D.; Zhou, H.; Boned, C. Peyrelasse, J.; Saint-Guirons, H.;
Xans, P. J. Phys. Chem. 1986, 90, 1692-1700. (b) Beilstein, Handbuch
der Organishen Chemie, 4 Aufl., E IV 5, 1978; p 664. (c) Techniques of
Chemistry, Organic SolVent; Riddick, J. A., Bunger, W. B., Eds.; Wiley-
Interscience: Toronto, 1970; p 261.
(41) (a) Tobe, Y.; Sonoda, M. In Modern Cyclophane Chemistry; Gleiter, R.,
Hopf, H., Eds.; Wiley-VCH: Weinheim, 2004; pp 1-40. (b) Jones, C.
S.; O’Connor, M. J.; Haley, M. M. In Acetylene Chemistry: Chemistry,
Biology and Material Science; Diederich, F., Stang, P. J., Tykwinski, R.
R., Eds.; Wiley-VCH: Weinheim, 2005; pp 303-385.
9
J. AM. CHEM. SOC. VOL. 128, NO. 51, 2006 16625