Modification of supramolecular binding motifs induced by substrate registry: Formation of self-assembled macrocycles and chain-like patterns

Article abstract of DOI:10.1002/chem.200901502

The self-assembly properties of two Zn" porphyrin isomers on Cu(111) are studied at different coverage by means of scanning tunneling microscopy (STM). Both isomers are substituted in their wieso-positions by two voluminous 3,5-di(iert-butyl)phenyl and tw

Full text of DOI:10.1002/chem.200901502

FULL PAPER
DOI: 10.1002/chem.200901502
Modification of Supramolecular Binding Motifs Induced By Substrate
Registry: Formation of Self-Assembled Macrocycles and Chain-Like Patterns
Leslie-Anne Fendt,*[a] Meike Stçhr,*[b] Nikolai Wintjes,[b] Mihaela Enache,[b]
Thomas A. Jung,*[c] and FranÅois Diederich*[a]
Abstract: The self-assembly properties
ramers occur most frequently but ev-
erything from cyclic dimers to hexa-
ACHTUNGTRNEmNUNG ers can be observed. Upon annealing
of the samples at temperatures
>1508C, dimeric macrocyclic struc-
tures are observed, in which the two
porphyrins are bridged by Cu atoms,
originating from the surface, under for-
mation of two CN···Cu···NC coordina-
tion bonds. The trans-isomer builds up
molecules assemble in a periodic,
of two ZnII porphyrin isomers on Cu-
densely packed structure. Both cis- and
trans-bis(4’-cyanobiphenyl)-substituted
ZnII porphyrins behave very differently
(111) are studied at different coverage
by means of scanning tunneling micros-
copy (STM). Both isomers are substi-
tuted in their meso-positions by two
voluminous 3,5-di(tert-butyl)phenyl and
two rod-like 4’-cyanobiphenyl groups,
respectively. In the trans-isomer, the
two 4’-cyanobiphenyl groups are oppo-
site to each other, whereas they are lo-
cated at right angle in the cis-isomer.
For coverage up to one monolayer, the
cis-substituted porphyrins self-assemble
to form oligomeric macrocycles held
together by antiparallel CN···CN dipo-
on Cu
phyrins in literature on less reactive
surfaces such as Au(111) and Ag(111).
ACHTUNGTNER(NUNG 111) compared to similar por-
A
ACHTUNGTRENNUNG
On the latter surfaces, there is no
signal visible between molecular orien-
tation and the crystal directions of the
linear chains on Cu
N
substrate, whereas on CuACTHGNURETNNU(G 111), very
age, whereas for higher coverage the
strong adsorbate–substrate interactions
have a dominating influence on all ob-
served structures. This strong porphy-
Keywords: porphyrinoids
ning probe microscopy
·
scan-
self-
rin–substrate interaction enables
a
·
much broader variety of structures, in-
cluding also less favorable intermolecu-
lar bonding motifs and geometries.
assemby · supramolecular chemis-
try · surfaces
lar interactions and CN···H-CACHTUNGTRNEUNG
(sp2) hy-
drogen bonding. Cyclic trimers and tet-
Introduction
Organic nanomaterials are expected to be of critical impor-
tance for the construction of nanodevices to address tomor-
rowꢀs challenges in electronics, opto-electronics, photonics,
and energy and information storage.[1–4] Self-assembled
supramolecular layers on surfaces, obtained from highly pro-
grammed molecular entities, are anticipated to play a lead-
ing role in the development of these future devices. For
their fabrication, it is necessary to have a detailed under-
standing on how self-assembled structures are built up, and
even more so, on how their building blocks can be designed
in a way to generate predictable architectures and display
all features requested for future applications. Furthermore,
knowledge on substrate reactivity is essential, since differen-
ces in substrate reactivity can lead to very different behavior
of adsorbed molecules.[5,6] Scanning tunneling microscopy
(STM) has made the direct observation of molecular assem-
blies on solid surfaces possible, which caused a major break-
[a] L.-A. Fendt, Prof. Dr. F. Diederich
Laboratorium fꢁr Organische Chemie, HCI
ETH Zꢁrich, 8093 Zꢁrich (Switzerland)
Fax : (+41)44-632-1109
E-mail: fendt@org.chem.ethz.ch
diederich@org.chem.ethz.ch
[b] Dr. M. Stçhr, Dr. N. Wintjes, M. Enache
Department of Physics, University of Basel
4056 Basel (Switzerland)
Fax : (+41)61-267-3784
E-mail: meike.stoehr@unibas.ch
[c] Dr. T. A. Jung
Laboratory for Micro- and Nanotechnology
Paul Scherrer Institute, 5232 Villigen PSI (Switzerland)
Fax : (+41)56-310-2646
E-mail: thomas.jung@psi.ch
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200901502.
Chem. Eur. J. 2009, 15, 11139 – 11150
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
11139

through towards understanding the complex interactions in
an assembly, both between the adsorbed molecules as well
as between molecules and substrate.[7,8] Porphyrins proved
on many occasion to be the perfect building blocks to con-
struct such molecular devices, as they are inexpensive, easy
to modify, and possess a rigid, defined geometry. As a key
feature, porphyrins prefer lying flat on metal substrates, due
to their extended p-system that is in strong contact with the
surface. Continuing our studies on differently modified por-
phyrins and their behavior on surfaces, we started from an
already known ZnII porphyrin, that had been studied exten-
For the preparation of cis-porphyrin 1, cis-bis-
AHCTUNGTRENNUNG
sively on Cu
ACHTUNGTRENNUNG
(111)[9] and, in the free-base form, on Au-
ACHTUNGTRENNUNG
trans-oriented 4-cyanophenyl substituents of this molecule
by one phenylene unit. Additionally we synthesized the
same extended porphyrin with two 4’-cyanobiphenyl groups
in an angular cis-orientation. Here, we report the synthesis
of the cis- (1) and trans- (2) bis(4’-cyanobiphenyl) ZnII por-
phyrins (Scheme 1) and their various intermolecular interac-
For the preparation of 2, using the same triple Pd0-cata-
lyzed cross-coupling sequence, bis(boronic ester)-substituted
ZnII–porphyrin 11 was synthesized in quantitative yield from
pinacolborane and the known trans-bisACTHNUTRGNEUNG(bromo)porphyrin 10
(Scheme 1c).[17] Cross-coupling of 11 with 4-bromo-1-iodo-
benzene afforded 12 and coupling with 4-cyanophenylboron-
ic ester 9 yielded the desired trans-bis(4’-cyanobiphenyl)–
ZnII–porphyrin 2 in 30% yield.
tion motifs in supramolecular assemblies on Cu
ces.
ACHTUNGTRENUN(NG 111) surfa-
Besides dipolar interactions, metal–ligand interactions
represent another powerful tool to construct rigid supra-
molecular architectures on metal surfaces, featuring ther-
mally and mechanically stable binding motifs.[11,12] A particu-
lar focus of this investigation addresses the strong influence
Both porphyrins are highly colored, stable solids, with
melting points above 3008C, and were fully characterized
(for their UV/Vis and 13C NMR spectra, see the Supporting
Information). An unambiguous identification of their con-
stituency was possible by 13C NMR spectroscopy. In the
spectrum of C2v-symmetric cis-derivative 1, four resonances
for a-carbon atoms and four resonances for b-carbon atoms
on the four pyrrole rings were expected, and they are ob-
served in CDCl3 at 150 MHz at d 149.82, 149.99, 150.50, and
of the CuACHTUNGTRENNUNG(111) substrate on self-assembly. We show in this
study that, upon annealing to 1508C and upwards, a number
of Cu adatoms are provided from the step edges which can
undergo coordinative bonding with the cyano groups on the
porphyrins, thereby building novel structures. By comparing
our results with the behavior of very similar porphyrins on
different noble metal substrates, we illustrate the delicate
balance between intermolecular interactions and those be-
tween molecules and the substrate.
150.65 ppm (a-carbons), and
d 131.60, 131.75, 132.52,
132.69 ppm (b-carbons), respectively. In contrast, the spec-
trum of the D2h-symmetric trans-derivative 2 only features
the expected two a-resonances at d 149.92 and 150.60 ppm
and two b-resonances at 131.71 and 132.61 ppm. Gratifying-
ly, despite their high molecular weight (C74H66N6Zn,
1102.46 Da), both compounds could be sublimed undecom-
posed at 10À5 Torr and bath temperatures of 340–3708C, as
confirmed by mass spectrometry, which makes them amena-
ble to STM studies under ultrahigh vacuum (UHV-STM)
after sublimation onto the substrate.
Results and Discussion
Synthesis and characterization: First, we attempted to syn-
thesize trans-bis(4’-cyanobiphenyl)porphyrin
2
via the
widely used MacDonald-type [2+2] condensation reaction
(Scheme 1a).[13] Accordingly, 4-(4’-cyanophenyl)-benzalde-
hyde (3) and 3,5-di(tert-butyl)phenyl-substituted dipyrro-
Conformation of porphyrins on surfaces: Crystalline meso-
phenyl-substituted porphyrins usually feature
a nearly
A
planar porphyrin macrocycle with the phenyl rings adopting
an approximately orthogonal orientation towards this plane.
tetrapyrrole, which would then subsequently be oxidized to
porphyrin 2. Instead, scrambling of the substituents occurred
during the condensation step, resulting in the formation of
cis-porphyrin 1 after the oxidation step as the only product
in 13% yield. This scrambling in an acid-catalyzed dipyrro-
methane–aldehyde condensation is an ever-present possibili-
ty; in some cases, the exchange is complete and in others no
exchange occurs at all.[14] Furthermore, a mixture of trans-
and cis-isomers bearing the same substituents can be very
difficult to separate. Therefore, another synthetic approach,
eliminating the possibility of scrambling, was chosen to pro-
vide pure 1 and 2.
À
In solution, rotation about the C C bond connecting the
porphyrin and phenyl rings is strongly hindered. Out-of-
plane distortions of the meso-phenyl have been observed for
crystalline 4-cyanophenylporphyrins, with the tilt angle be-
tween the porphyrin plane and the straight line passing
[18]
ꢀ
through C1Ph-C4Ph-C N amounting up to 108. In crystal-
line tetrakis(4-cyanophenyl)–ZnII–porphyrin, the 4-cyano-
phenyl rings are orientated orthogonally to the porphyrin
ꢀ
macrocycle and undergo intermolecular, dipolar C N···H-Cb
contacts with pyrrolic hydrogens of neighboring molecules
in the crystal lattice.[19]
11140
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11139 – 11150

Supramolecular Binding Motifs
FULL PAPER
Scheme 1. Syntheses of the cis- and trans-ZnII porphyrin isomers 1 and 2. Ar=3,5-di(tert-butyl)phenyl. py=pyridine. NBS=N-bromosuccinimide.
Chem. Eur. J. 2009, 15, 11139 – 11150
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11141

F. Diederich, L.-A. Fendt, M. Stçhr, T. A. Jung et al.
Whereas meso-substituted porphyrins are rather rigid in
the bulk and in solution, with conformational preferences as
described, if deposited on metal surfaces, strong substrate
interactions induce deformations to the overall structures
that are not observed in the liquid and crystalline state. Cal-
culations from STM images of tetrakis
ACHTUNGERTN[NUNG 3,5-di(tert-butyl)-
phenyl]–CoII–porphyrins on Ag
AHCTUNGTRENNUNG
of-plane tilt angle of 258 of the meso-phenyl rings with re-
spect to the porphyrin plane, compared to the 108 tilt seen
in the crystal.[20]
Furthermore, the intermolecular interaction motifs
change on surfaces. Thus, the two-dimensional assembly of
the bis(4-cyanophenyl) analogue of 2, obtained on Au
ACHTUNGTNER(NUNG 111)
surface under UHV conditions, was shown by STM to in-
Figure 1. Conformational isomers A and B of cis-porphyrin 1 with the
“dark line” representing the two corner pyrrole rings that are twisted
downwards, whereas the other two corners are twisted upwards. This phe-
nomenon is induced through the 3,5-di(tert-butyl)phenyl substituents
twisting sideways; therefore as a second artifact, the tert-butyl groups on
top of the downwards-bent pyrrole ring appear bigger and brighter, while
the other tert-butyl group appears smaller.
ꢀ
ꢀ
volve antiparallel dipolar C N···C N interactions as well as
ꢀ
additional, hydrogen-bond-type C N···H-Cortho interactions
involving the positive polarized hydrogen ortho to the CN
group (see Figure 2 below).[10] This bonding motif requires
the meso-phenyl rings to rotate away from their preferred
perpendicular conformation and strive for closer coplanarity
with the porphyrin macrocycle, an effect which is exclusively
observed on surfaces due to strong interactions between the
porphyrin core and the substrate.[21] The rotation induces
steric repulsion between the meso-phenyl hydrogens and the
b-hydrogens of the porphyrinic core, resulting in a saddle-
shaped deformation of the macrocycle.[22] High-resolution
UHV-STM, as shown in Figure 1 for cis-isomer 1, allows ob-
servation of this phenomenon by imaging the porphyrin
core as two oblong protrusions where two opposite pyrrole
rings are pointing upwards, separated by a dark line repre-
senting the two rings that are twisted downwards. Due to
the different nature of the porphyrin substituents, their iden-
tification by STM is straightforward: bulky 3,5-di(tert-butyl)-
phenyl side groups are visible as two bright circular spots of
two sizes, whereas the rod-like 4’-cyanobiphenyl substituents
appear as short lines with distinctively less intensity
(Figure 1).[23] Since the upwards-pointing tert-butyl group of
the 3,5-di(tert-butyl)phenyl substituent is forcing the pyrrole
ring downwards, this group appears brighter than the other
tert-butyl group close to the upwards-facing pyrrole ring. On
the basis of this assignment of the orientation of the sub-
stituents, two conformational isomers of the cis-isomer can
be identified: “Type A” is characterized by the bending line
separating the 4’-cyanobiphenyl from the 3,5-di(tert-butyl)-
ꢀ
H···N C interactions featured by associating cyanophenyl
rings are substantially weaker than classic hydrogen bonds
À
À
À
involving more acidic O H, S H, or N H H-bond donors.
Therefore, they compete with other weak interactions, such
as C-H···O, p–p, halogen–halogen, and C-H···p interactions,
in solution, in the solid state, and in surface assemblies.
Whereas classic hydrogen bonds are highly directional and
therefore can organize molecules into networks with pre-
dictable architectures,[24–26] the weak Csp2-H···N C hydrogen
ꢀ
bonds can form in several competing binding alternatives, as
displayed in Figure 2.[27] This makes two- and three-dimen-
ꢀ
sional structures built upon Csp2-H···N C interactions harder
to predict, but offers the advantage that this binding motif is
able to adapt to very different geometries. Therefore, nu-
merous types of association patterns are possible, whereas
those based on classic hydrogen bonds are geometrically
very limited. On metal substrates, geometries a) and b) have
been observed for two-dimensional assemblies,[10,28] with b)
being preferred for porphyrins with meso-4-cyanophenyl
groups in trans-orientation, and a) for porphyrins with
meso-4-cyanophenyl groups in cis-orientation. Furthermore,
geometry c) has been observed for chain-like structures
where the substrate determined a rotation of 1208 between
phenyl
substituents
and
“type B” is characterized by
the bending line being on the
symmetry axis of the molecule.
In trans-porphyrin 2 due to its
point symmetry, this deforma-
tion does not lead to achiral
conformational isomers, but in-
stead to conformational enan-
tiomers (i.e., mirror images).
The C-H···N binding motif:
Figure 2. Typical C-H···N interactions between cyanophenyl moieties: a) antiparallel dipole–dipole bonding, b)
The hydrogen-bond-type Csp2
-
trimeric bonding, and c) hydrogen bonding-type association, where the angle V is variable.
11142
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11139 – 11150

Supramolecular Binding Motifs
FULL PAPER
two neighboring (meso-cyanophenyl)–ZnII–porphyrins, re-
preference visible for the formation of trimers and tetramers
(Figures 3a and 5). We associate this behavior to the above
discussed strong adsorbate–substrate interactions, inducing
all porphyrins to align with their dark lines along one of the
ꢀ
sulting in intermolecular C-H···N C interactions (Figure 2c,
with V=1808).[28]
Formation of oligomeric self-assembled macrocycles by cis-
isomer 1: Upon deposition of the cis-isomer 1 onto a Cu-
ACHTUNGTRENNUNG(111) surface at sub-monolayer coverage, the porphyrins
three principal directions of the CuACTHNGUTERNNU(G 111) substrate (Fig-
ure 3a). Trimers showed to be one of the most “straightfor-
ward” forms of assembly in that they are observed to consist
almost exclusively of all-type A conformational isomers,
with only few all-type B trimers observed. This preference
for same-type trimers is readily explained by the fact that in
order to form a triangular cluster, same-type porphyrins ar-
range in such a way that their dark lines adopt three differ-
ent orientations separated by rotations of 1208, which nicely
build up oligomeric structures. Unlike previous work from
Yokoyama et al., where a similar molecule with shorter 4-
cyanophenyl binding groups was forming exclusively tetra-
ACHTUNGTRENNUNGmers on a AuACHTUNGTRENNUNG
(111) substrate,[10] here everything from dimer-
ic to hexameric as well as chain-like structures could be ob-
served. We assign this structural diversity mainly to the fol-
lowing effects. Firstly, as introduced before, the antiparallel
CN···CN dipole–dipole bond-
corresponds to the three principal directions of the CuACHTUNGTRENNUNG(111)
ing motif was observed to be
the most preferred arrange-
ment of two 4-cyanophenyl
groups on
a gold substrate
(Figure 2),[10] but the cyano-
phenyl groups can also interact
in different geometries (Fig-
ure 2b and c), provided that
the accompanying loss in dipo-
lar-bonding energy is compen-
sated by another contribution.
Secondly, it has been shown
that the interaction strength of
large aromatic molecules [that
is,
3,4,9,10-perylenetetracar-
boxylic dianhydride (PTCDA)]
with a noble metal substrate
decreases in the order Cu
ACHTUNGTNER(NUNG 111)
> Ag(111) > Au
N
ACHTUNGTRENNUNG
a substrate change is expected
to influence the delicate bal-
ance between adsorbate–adsor-
bate and adsorbate–substrate
interactions in such a way that
the adsorption position of the
molecule together with its con-
formational state and intermo-
lecular-bonding pattern will be
altered. This corresponds well
to our former studies, where si-
multaneous imaging of similar
porphyrins and CuACTHUNRGTNE(NUG 111) sub-
strate atoms showed that the
porphyrins are oriented on the
surface with their characteristic
dark line aligning with one of
the three principal directions
of the substrate.[28]
Although macrocyclic oligo-
mers ranging from dimers to
hexamers have been observed
in our data, there was a clear
Figure 3. STM images for cis-porphyrin 1 on CuACTHNGUTERNNUG
(111). a) Detailed image (18ꢃ18 nm2) showing different oligo-
meric structures from which the orientation of the porphyrins with respect to the substrate can be identified.
b)–g) Detailed STM images with the corresponding molecular models for the different oligomeric structures
(image size given in brackets): b) dimer (5.0ꢃ5.0 nm2); c) all-type A trimer (6.3ꢃ6.3 nm2); d) mixed-type
tetramer (6.4ꢃ6.4 nm2); e) all-type B tetramer (6.2ꢃ6.2 nm2); f) irregular pentamer (7.6ꢃ7.6 nm2); g) regular
hexamer (8.2ꢃ8.2 nm2). The visualizations of the macrocycles and the underlying copper surface were generat-
ed using Spartan 08, Wavefunction Inc., Irvine, CA (USA), 2008. The molecular planes of the monomers were
left undistorted, but the surface-induced deformation is denoted by darkening/lightening the respective down-
or upwards bent pyrrole groups. The dihedral angles between the porphyrin plane and each of the appended
phenyl rings, as well as between two phenyl rings, were set to a standard of 308.[21] The molecules were then ar-
ranged taking into account their preferred orientation on the surface, their preferred intermolecular bonding
pattern, and the STM images.
Chem. Eur. J. 2009, 15, 11139 – 11150
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11143

F. Diederich, L.-A. Fendt, M. Stçhr, T. A. Jung et al.
substrate (Figure 3c). This accordance with the lattice
makes up for the fact, that despite the triangular geometry,
the CN substituents do not bend towards an angle of 608 to
adopt a perfectly antiparallel bonding motif. Instead, they
keep their preferred rectangular shape and display a mixed,
about 1508 bonding motif, between antiparallel CN···CN and
position of the porphyrins is restricted to one of the three
dimensions of the copper lattice, they enjoy some degree of
freedom. As a result of the few observed hexamers, one
rarely sees one that is a completely regular same-type as-
ꢀ
sembly with all porphyrins undergoing the same C N···H-
Csp2 H-bonding (Figure 3g). Once again, only all-type B reg-
ular assemblies are observed, since the substituents are ex-
pected to slightly bend outwards. More often than seeing
regular assemblies, it is observed that one molecule of the
circle switches type in order to undergo antiparallel
ꢀ
C N···H-C hydrogen-bonding interactions. Hence, if there is
some distortion, it is not substantial enough to be recog-
nized with STM. Yet, for the trimers the clear preference
seen for type A assemblies might be due to a small differ-
ence in energy between type A and B molecules for the case
that the 4’-cyanobiphenyl substituents are indeed slightly
bent towards each other, invisible for STM.
ꢀ
ꢀ
ꢀ
C N···C N bonding with one neighbor, and C N···H-C H-
bonding with the other. Also observed were hexamers
where one molecule folds inwards, presumably for bonding
angle optimization, or with an extra molecule occupying the
ring cavity to expand two binding sites towards a preferred
trimeric pattern.
The tetramers observed on the Cu surface can be divided
into two subgroups: on the one hand there are the same-
type tetramers (Figure 3e), and on the other hand we see
mixed tetramers showing alternating type A and B mole-
cules (Figure 3d). Furthermore, same-type tetramers look
slightly distorted in our STM data, with the 4’-cyanobiphe-
The dimer represents one of the most interesting struc-
tures observed in the course of this study. One would expect
dimer formation to occur between two same-type molecules
via a bonding pattern observed before in a trimeric arrange-
ment, where the CN group of one porphyrin points towards
both aromatic hydrogens of the neighborꢀs 4’-cyanobiphenyl
group. This arrangement would allow the two molecules to
keep the substituents in optimal perpendicular conformation
towards each other (Figure 2b). However, this is exclusively
observed if another substituent is coming from the side to
complete the trimeric motif (Figure 3b). Upon heating to
elevated temperatures >1508C, instead, the 4’-cyanobiphe-
ACHTUNGTRENNUNGnyl substituents not meeting in the optimal 1808 angle, de-
spite the geometric possibility given by the molecular shape.
Mixed-type tetramers on the contrary, are shown to build up
regular rectangles and interact via antiparallel CN···CN
dipole–dipole bonding. This difference in behavior between
same-type and mixed-type tetramers is again attributed to
the Cu
ACHTUNGTRENNUNG
porphyrinsꢀ dark lines onto the three CuACHTUNGTRENNNUG
Therefore, mixed-type tetramers (Figure 3d) contain por-
phyrins, in which dark lines are aligned parallel to each
other along the same axis. Thereby the 4’-cyanobiphenyl
legs are in an unstrained 1808 angle, which allows them to
undergo optimal antiparallel dipolar bonding. At the same
time, porphyrins in a same-type tetramer stick to the surface
with their dark lines drawing a substrate-induced angle of
1208 and for that reason cannot optimally interact with one
another. This effect accounts for the overall distorted struc-
ture, with two inwards and two outwards bent dipolar supra-
molecular synthons (Figure 3e). It has to be mentioned fur-
ther that, contrasting the trimer case, in the tetramer sub-
group of same-type assemblies, where the substituents have
to bend outwards a little, all-type B seems to be much pre-
ferred over all-type A. Therefore, we contemplate that while
porphyrins with legs that are slightly inwards bent prefer
type A, porphyrins with slightly outwards bent legs prefer
type B.
AHCTUNGTREGnNNNU yl substituents of porphyrin dimers seem to bend towards
each other until they no longer stand perpendicular, but are
now drawing an estimated angle of 608 (Figure 4d). This
means that the two 4’-cyanobiphenyl substituents are each
bent about 158 away from their preferred rectangular posi-
tion towards each other ((90–60)/2)8=158). Such a strong
distortion away from the favorite 908 angle between sub-
stituents is unprecedented in porphyrin chemistry, and no
studies on the lateral flexibility of porphyrin substituents
have been performed to date. Presumably, the two substitu-
ents can only bend towards each other in such a way, due to
two major effects. First of all, as previously pointed out, the
saddle-shaped deformation of the porphyrin ring induced by
the strong interactions with the copper substrate causes the
pyrrole ring between the 4’-cyanobiphenyl residues, which
would usually prevent such a rapprochement by steric repul-
sion, to twist up- or downwards and give way to the two legs
to come closer together. Without this surface-induced defor-
mation, lateral movements of the substituents would not be
possible. Second, and even more important, this yet un-
known lateral distortion of the meso substituents of about
158 each can be attributed to the coordination of two cyano
groups of opposite molecules towards a single Cu adatom. It
is known that at elevated temperatures the Cu step edges
become mobile and thus, the Cu adatoms required for dimer
formation are available.[12] In all likelihood, the gain in
energy with two cyano groups involved in coordination
bonding to a Cu adatom makes up for the molecular distor-
tion (Figure 4c). While at lower temperatures, no dimers of
Pentamers are very rarely observed, which meets the ex-
pectations, since the pentagonal structure neither fits the
CuACHTUNGTRENNUNG(111) lattice nor the 908 angle between binding substitu-
ents in the porphyrin. Therefore, observed pentamers are
not regular; the porphyrins are interacting with each other
in different ways, and type A and B porphyrins come togeth-
er randomly in one structure, yet with their dark lines
always sitting along one of the three Cu
tions (Figure 3 f).
ACHTUNGTRENNUNG
Hexameric structures are favored again by the CuCAHTUNGTRENNUNG
geometry, nevertheless they already show the limitations in
building up rather big regular hollow structures: even if the
11144
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11139 – 11150

Supramolecular Binding Motifs
FULL PAPER
ing antiparallel dipolar bond-
ing possible without distortion
of the substituents. No dimers
are present at this coverage;
few hexameric structures and
barely any pentamers are ob-
served. When increasing the
coverage up to 0.5 ML, trimers
and tetramers still are the
major fractions, though trimers
now are slightly more favored
than tetramers, presumably be-
cause they encircle a smaller
area. Hexameric and pentame-
ric structures are becoming
more frequent, and the first
dimers appear on the surface.
At a coverage of 0.75 ML, the
preference of trimers over tet-
ramers becomes even more ap-
parent. No hexameric and pen-
tameric structures are ob-
served, while dimers continue
to get progressively more pop-
ular (Figure 5).
Figure 4. a) STM image (50ꢃ26 nm2) for submonolayer coverage of 1 on Cu
ACTHNUTRGNE(NUG 111) annealed at 1508C. A few
dimeric structures can be observed. The red square marks a dimer that is enlarged in c). b) STM image (60ꢃ
Upon further increase of the
coverage, the available space
decreases and structures with
60 nm2) for submonolayer coverage of 1 on Cu
ACTHNUTRGNEUNG(111) after annealing at 2008C. Now, mainly dimeric structures
can be observed. c) High-resolution STM image (5.5ꢃ2.5 nm2) of the dimer marked by the red square in a). A
copper atom is visible in between two substituent legs (red arrow). d) Simplified model of the two 4’-cyanobi-
phenyl groups bend together by ca. 158 each in order to undergo coordination bonds to two Cu atoms. Ar=
3,5-di(tert-butyl)phenyl.
smaller
encircled
space
become favored. This effect to
opt for the most compact ar-
rangement reaches its maxi-
this kind could be found due to the insufficient amount of
(thermal) energy to enable the lateral movements of the
substituents, annealing at 1508C is sufficient for the forma-
tion of Cu-coordinated dimers (Figure 4a). Along this line,
even more dimers appeared upon heating to 2008C (Fig-
ure 4b).
mum at coverage >0.75 ML. As expected, there was no reg-
ular network formation observable; the molecules started to
form an irregular assembly incorporating macrocyclic oligo-
mers, irregular curvy chains, and extensive branching.
Coexistence of different porphyrin conformers have been
already observed for a tetrakis(meso-aryl)-substituted por-
phyrin on a CuACHTUNGTRENNUNG(111) substrate, which adjusts to a crystal lat-
tice mismatch by mixing different unusual conformational
isomers.[29] Additionally, the elongation of the binding sub-
stituent from 4-cyanophenyl to 4’-cyanobiphenyl offers even
more tolerance for distortion, as more bond angles can be
bent to adjust to diverse spatial arrangements. Most mole-
cules involved in this kind of dimer formation are type A,
which again confirms the hypothesis derived from the trim-
ers, same-type tetramers, and hexamers, that A is rather pre-
ferred when the cyanobiphenyl substituents are bent to-
wards each other, while B is favored when the legs are
bending apart.
Coverage dependence of the most abundant macrocyclic
structures formed by cis-isomer 1: At a coverage of 0.25 ML
(ML=monolayer), mainly trimeric and tetrameric structures
are observed. Tetramers are slightly preferred, which might
be due to the fact that this geometry is the only one render-
Figure 5. Analysis of the frequency for the appearing macrocyclic oligo-
mer structures in dependence of the total molecular coverage, with n
being the number of porphyrins involved, plotted against the percentage
of occurrences at a given coverage. ML=monolayer.
Chem. Eur. J. 2009, 15, 11139 – 11150
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11145

F. Diederich, L.-A. Fendt, M. Stçhr, T. A. Jung et al.
A broad spectrum of self-assembled structures of cis-isomer
1 influenced by molecule–substrate interactions: While in-
vestigating the arrangement of cis-bis(4’-cyanobiphenyl)-
our earlier studies as well as with the requirement of the
two downwards twisted pyrrole rings to sit on one of the
three main CuACTHNUTRGNEUNG
(111) directions.[28] The trimeric branching
substituted ZnII porphyrins on Cu
(111) and observing the
pattern is caused by antiparallel CN···CN interactions to-
gether with CN···H-Csp2 H-bonding interactions, resulting in
a cyclic arrangement of the cyano groups (Figure 6b).[10]
Both observations corresponded nicely to patterns found for
a similar trans-bis(4-cyanophenyl)-substituted molecule in
our previous studies.[28]
very strong directing force of the substrate, we could not
help but notice that this indeed did not occur in the work of
Yokoyama and co-workers, who investigated the formation
of tetramers for a very similar cis-bis(4-cyanophenyl) por-
phyrin, yet on a AuACHTUNGTRENNUNG
(111) surface.[10] In contrast to our find-
ings, Yokoyama et al. observed only tetramers (cf. Section
on the formation of oligomeric self-assembled macrocycles
by cis-isomer 1), and they consisted exclusively of type A
porphyrins.[30] Yet, these same-type tetrameric assemblies
displayed regular antiparallel CN···CN binding and porphy-
rin dark lines in a 908 angle towards each other. This alter-
On AuACTHNGUTERN(UNG 111) at low coverage, porphyrins with only one 4-
cyanophenyl substituent were observed to interact exclusive-
ly via the trimeric bonding motif, whereas porphyrins with
two trans-(4-cyanophenyl) substituents, that is, two binding
sites, formed solely chains via antiparallel dipole–dipole
bonding.[10] Instead, for a trans-bis(4-cyanophenyl)–ZnII–por-
nation of dark lines on Au
(111), not fitting the geometry of
phyrin on Cu
was observed.[28] This shows again, as seen before for cis-
porphyrins, that depositing similar molecules on Au(111)
gives rise to only one, the thermodynamically most pre-
ferred, structure, whereas Cu(111) as a substrate supports
ACHTUNGTRENUN(NG 111) both trimeric and antiparallel bonding
the gold lattice, occurs solely due to molecular conforma-
tional preferences, and the influence of the substrate is neg-
ligibly small.[10,23] This discrepancy is again explained by the
increase of adsorbate–substrate interactions; while Yokoya-
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
ma worked on a less interacting Au
molecules interact in their one preferred bonding motif, the
Cu(111) substrate demands exact alignment of the porphyr-
inꢀs dark lines onto the Cu(111) main axes. The Cu(111)
G
the molecules and allows them to interact in a variety of
ways.
In contrast to our previous work, at high surface coverage,
no hexagonal porous network was building up upon deposi-
E
G
substrate directs the porphyrin molecules and, due to this
strong registry effect, can thus give rise to a much broader
array of possible structures, also including weaker intermo-
lecular bonding patterns.
tion of trans-isomer 2 onto the CuACTHUGNETRNNU(G 111) surface. Instead, an
adlayer was observed, consisting of interlocked chains (Fig-
ure 6d). A similar type of network was so far only observed
for porphyrins held together by classic hydrogen bonding on
AuACTHNUTRGNEUNG
(111), which induces a linear geometry,[23] but not for
Assemblies of trans-isomer 2 on Cu
(111): Deposition of
flexible binding motifs such as the dipolar and weak H-
bonding-type interaction between cyanophenyl residues. It is
assumed that for a hypothetical hexaporous network based
on the trimeric motif shown in Figure 2b, upon elongation
of the substituent from 4-cyanophenyl to 4’-cyanobiphenyl
the pore-to-pore distance would be too big and the network
would collapse. In the adlayer, the molecules are once more
held together only by antiparallel dipole–dipole interactions
along one dimension, displaying linear parallel chains as
seen at lower coverage. Yet at higher coverage, these paral-
lel chains become interlocked with each other through van
der Waals interactions of the
trans-isomer 2 at low coverage leads to the formation of
linear chains, which were sometimes interconnected by a tri-
meric branching pattern (Figure 6a+b). The molecules
along a chain are connected by antiparallel dipole–dipole in-
teractions of their 4’-cyanobiphenyl substituents. Among
identical enantiomers, two molecules of 2 can bind in a
linear fashion (where the dark lines are parallel, Figure 6c,
1+2, and 3+4) or with a kink (where the dark line of the
second molecule is oriented along one of the two other Cu-
ACHTUNGTRENNUNG(111) axes, Figure 6c, 2+3). This is in correspondence with
voluminous 3,5-di(tert-butyl)-
phenyl side groups between
adjacent porphyrins of two
chains, resulting in a further
gain of stability. This assembly
type can only be constructed
from the elongated bis(4’-cya-
nobiphenyl)-porphyrin;
the
chains of the bis(4-cyanophen-
yl)-porphyrin cannot be inter-
locked, since the gaps between
ACTHNUTRGNEUNG
Figure 6. a) STM image (50ꢃ50 nm2) at low coverage of 2 on Cu(111); b) Cut-out (15ꢃ7 nm2) showing a tri-
the
3,5-di(tert-butyl)phenyl
meric branching site; c) Cut-out (10.4ꢃ5.7 nm2) showing the bonding between two identical conformational
enantiomers with their dark lines along the same direction (1+2, and 3+4), as well as between two identical
enantiomers with their dark lines along different directions (2+3); d) STM image (15ꢃ15 nm2) of trans
side groups would be too
small.
Surprisingly, there is one
more vital contribution that
isomer 2 on Cu
ACHTUNGTRENNUNG(111) at high coverage showing the 2D arrangement of a dense, non-porous network (molecules
encircled in white, dark lines assigned in red).
11146
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11139 – 11150

Supramolecular Binding Motifs
FULL PAPER
has to be added to successfully assemble this network: the
cules of 1 could form dimers via coordination bonding with
single Cu adatoms released from the surface by heating, de-
spite the presence of a substantial strain resulting from the
strong bending of the 4’-cyanobiphenyl groups. This new dis-
tortion can be attributed to the substrate-induced deforma-
tion of the porphyrin, where the pyrrole ring with its out-
wards-pointing hydrogen atoms is twisted away from in-be-
tween the two substituents, therewith releasing the steric
hindrance.
trans-Porphyrin 2 builds up very ordered chains with occa-
sional triangular branching. Two identical enantiomers of
trans-2, with their dark lines oriented parallel to each other,
were seen to form linear bonding, while the same enantio-
mers, with dark lines lying on different axes of the lattice,
show kinked bonding. At higher coverage, trans-isomer 2,
unlike similar porphyrins, was seen to form a very dense
strong support from the Cu
the bis(4’-cyanobiphenyl)-porphyrin under the same condi-
tions on Ag
(111), a less influencing substrate,[6] there was no
ACHTUNGTNER(NUNG 111) surface. When depositing
ACHTUNGTRENNUNG
sign of an adlayer of this type.
The surface influence can be observed further, when com-
paring this adlayer with the before-mentioned similar one
held together by classic hydrogen bonds on Au
this classic hydrogen-bond network, one can see that the
molecules are not directed by the Au(111) surface, since
(111).[23] In
ACHTUNGTRENNUNG
G
their dark lines are alternating—solely defined by the supra-
molecular assembly and ignoring the three-fold surface sym-
metry of the AuACHTUNGTRENNUNG(111) lattice. In contrast, the CuACHUTNGTREN(NNGU 111) sur-
face influence on the adlayer held together by non-directive
dipole–dipole bonds can be even observed by eye, since
here the dark lines of all molecules are aligned along one
lattice dimension (Figure 6d). During the deposition process
of the molecules on the hot surface, reorientation of the
molecules has taken place. In the subsequent cooling pro-
cess the adlayer forms, which leads to the segregation of the
two enantiomers. The same observation can be made in the
porphyrin chains at lower coverage: the porphyrinꢀs dark
non-porous network on a CuACTHNURGTNE(UNG 111) surface, which was held
together by antiparallel dipole–dipole interactions in one di-
mension, and van der Waals interactions in the other. The
same network geometry was reached in the past employing
a strongly directing H-bonding pattern on the weak sub-
strate AuACTHNUGRTENNUG ACHUTGNTREN(NUGN 111), we build up the
(111).[23] In our case on Cu
lines in a linear chain are alternating on Au
N
same network with a very flexible binding motif, but on a
strongly influencing substrate. This network presumably
forms for three reasons: firstly, the molecular chains of elon-
gated trans-porphyrin 2 fit snugly into one another, and sec-
ondly, a hypothetical hexagonal porous network of 2, analo-
gous to common networks of shorter trans-bis(4-cyanophen-
yl)-porphyrins,[9] would be unstable due to its critically big
pore-to-pore size. But lastly and most astonishingly, the sub-
strate holds the molecules in place, since this adlayer forms
on Cu(111) they are parallel (Figure 6c). Therefore, the
ACHTUNGTRENNUNG
same network geometry can be reached both by a strongly
directional bonding motif (such as classic H-bonding) on a
less interacting surface as well as by a geometrically flexible
bonding motif (such as antiparallel dipole–dipole interac-
tions) on a strongly interacting substrate.
Conclusions
only on Cu
surface. Again, the substrate influence is visible, since all
dark lines are aligned along the Cu(111) direction, whereas
in the classic hydrogen bond lattice on the weaker interact-
ing substrate Au(111) the dark lines alternate against the ge-
ACHUTGTNRENNUG(111), not on a slightly less supportive AgACHTUNGTRENNUNG(111)
We have presented a study on the behavior of two porphy-
rin isomers at lower and higher coverage on a CuACTHNUTRGENUG(N 111) sub-
strate, each isomer containing two rod-like 4’-cyanobiphenyl
and two voluminous 3,5-di(tert-butyl)phenyl side groups. Re-
markably detailed STM images allowed to fully understand
the formation of oligomeric macrocycles of cis-isomer 1—a
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
ometry of the gold lattice.
This reveals the complexity of supramolecular structure
formation induced by the delicate balance of the various
possible molecule–molecule and molecule–substrate interac-
tions competing with each other on a metal surface. Due to
this complexity, the prediction of self-assembled structures
via the bottom-up approach remains an immense challenge.
Conclusively, the role of the substrate in the design of such
novel self-assembled structures cannot be emphasized
enough, since it can make the difference between whether a
certain network can be formed or not.
type of nanoislands—at lower coverage on a CuACTHNUTRGENUG(N 111) sur-
face. It was shown how flexible 4’-cyanobiphenyl substitu-
ents can adapt to very different geometries if supported by
strong adsorbate–substrate interactions, due to their elongat-
ed structure as well as diverse 4’-cyanobiphenyl bonding
motifs. In all detected structures on CuACTHNUTRGNEUNG(111), substrate influ-
ence was dominating. Comparisons were drawn between our
molecules and studies with a cis-bis(4-cyanophenyl) porphy-
rin on a gold substrate. Whereas the porphyrins on Au
only formed regular tetramers via antiparallel CN···CN di-
polar bonding with the “dark lines” (Figure 1) ignoring the
(111)
Future work will evaluate the possibility to adjust the oli-
gomeric macrocycles formed by cis-1 to a uniform size
through co-evaporating a fitting template, which the mole-
cules then can encircle. Moreover the role of the substrate
in supramolecular chemistry on surfaces will be further in-
vestigated, therewith making another step towards a com-
plete insight into the requirements to build up predictable
self-assembly architectures.
directions of the gold substrate, our porphyrins on CuACHTUNTRGNEUNG(111)
formed everything from macrocyclic dimers to hexamers
using an entire set of different bonding patterns, and with
all molecules supported and aligned by the substrate.
Although there is a clear preference evident for structures
where cis-porphyrin 1 is close to its normal rectangular con-
formation, it was especially impressing to see how two mole-
Chem. Eur. J. 2009, 15, 11139 – 11150
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11147

F. Diederich, L.-A. Fendt, M. Stçhr, T. A. Jung et al.
(neat): n˜ = 3071 (w), 2962 (s), 2925 (m), 2870 (m), 2326 (w), 2165 (w),
2052 (w), 1990 (w), 1590 (s), 1477 (m), 1363 (m), 1288 (m), 1004 (s), 796
(s), 771 (s), 690 cmÀ1 (s); UV/Vis (CHCl3): lmax (e) = 560 (8400), 523
(12300), 424 nm (322700 mÀ1 cmÀ1); HR-MALDI-MS (3-HPA): m/z (%):
calcd for C48H50Br2N4Zn+: 904.1688; found: 904.1678 (32) [M+], 905.1731
(26), 906.1665 (90), 907.1703 (71), 908.1658 (100), 909.1682 (72), 910.1654
(66), 911.1678 (39), 912.1664 (23), 913.1674 (12).
Experimental Section
Materials and general methods: Reagents and solvents were purchased at
reagent grade from Acros, Aldrich, and Fluka, and used as received. All
reactions were performed under an inert atmosphere by applying a posi-
tive pressure of N2. Compounds 5,[15] 9,[16] and 10[17] were synthesized ac-
cording to literature procedures. Adsorption chromatography columns
and plug filtrations were carried out with SiO2 60 (particle size 0.040–
0.063 mm, 230–400 mesh ASTM; Fluka) or SiO2 60 (particle size 0.063–
0.200 mm, 70–230 mesh ASTM; Merck). Flash column chromatography
was run at a maximum head pressure of 0.2 bar. Thin-layer chromatogra-
phy (TLC) was conducted on Macherey–Nagel Alugram SIL G/UV234
aluminum sheets coated with SiO2 60 F254; visualization by UV light
(254 or 366 nm). Melting points (m.p.) were measured on a Bꢁchi B-540
melting-point apparatus in open capillaries and are uncorrected.
1H NMR and 13C NMR spectra were measured on a Varian Gemini 300,
Varian Mercury 300, Bruker ARX 300, Bruker DRX 400, Bruker AV
400, Bruker DRX 500, and Bruker Avance III 600 (with cryo-probe) in-
struments. Chemical shifts are reported in ppm downfield from tetra-
5,10-BisACTHNUTRGNE[NUG 3,5-di(tert-butyl)phenyl]-15,20-bis(4,4’,5,5’-tetramethyl-[1,3,2]di-
oxaborolan-2-yl)-porphyrinato–N21,N22,N23,N24zinc(II) (7): To
a 50 mL
round-bottomed flask under N2 containing 6 (66 mg, 7.3ꢃ10À2 mmol) in
1,2-dichloroethane (30 mL), pinacolborane (0.6 mL, 4.1 mmol), [PdCl2-
AHCTUNGTRENNUNG
(PPh3)2] (5 mg, 0.7ꢃ10À2 mmol), Et3N (1 mL), and two drops of H2O
were added. After heating to reflux for 1.5 h, the mixture was cooled
down and filtered over Celite. After washing with brine (1ꢃ5 mL) and
water (2ꢃ5 mL), the mixture was dried (MgSO4) and the solvent evapo-
rated. FC (SiO2; hexane/CH2Cl2 1:3; 1% v/v Et3N) yielded 7 (73 mg,
quant.) as a violet solid. M.p. >3008C; 1H NMR (300 MHz, CDCl3,
258C): d =10.06 (s, 2H), 9.88 (d, J
N
ACHTUNGTRENNUNG
4.5 Hz, 2H), 9.01 (s, 2H), 8.07 (d, J
N
ACHTUNGTRENNUNG
1.5 Hz, 2H), 1.89 (s, 24H), 1.54 ppm (s, 36H); 13C NMR (75.41 MHz,
CDCl3, 258C): d = 152.87, 152.75, 150.50, 150.23, 148.73, 141.87, 132.54,
129.87, 121.76, 120.29, 85.16, 35.20, 31.98, 25.55 ppm; IR (neat): n˜ = 2975
(s), 2389 (w), 2034 (w), 1687 (w), 1583 (m), 1516 (s), 1337 (s), 1285 (m),
1258 (s), 1167 (s), 943 (s), 854 (s), 784 (s), 713 (m), 666 cmÀ1 (s); UV/Vis
(CHCl3): lmax (e) = 588 (3900), 553 (7900), 423 nm (232000 mÀ1 cmÀ1);
HR-MALDI-MS (3-HPA): m/z (%): calcd for C60H74B2N4O4Zn+:
1000.5182; found 999.5195 (37), 1000.5182 (100) [M+], 1001.5199 (94),
1002.5171 (88), 1003.5180 (70), 1004.5163 (61), 1005.5182 (34), 1006.5218
(14).
ACHTUNGTRENNUNGmethylsilane using the residual solvent signals as an internal reference
(CHCl3, dH 7.26 ppm, dC 77.00 ppm). Coupling constants (J) are given in
Hz. The resonance multiplicity is described as s (singlet), d (doublet),
t (triplet), q (quartet), and m (multiplet). Broad resonances (br) are indi-
cated. All spectra were recorded at 298 K. Infrared spectra (IR) were re-
corded on a Perkin–Elmer BX FT-IR spectrometer; neat samples were
measured. Selected absorption bands are reported by wavenumber
(cmÀ1), and their relative intensities are described as s (strong), m
(medium), or w (weak). UV/Vis spectra were recorded on a Varian Cary-
5 spectrophotometer. The spectra were measured in CHCl3 in a quartz
cuvette (1 cm) at 298 K. The absorption maxima (lmax) are reported in
nm with the extinction coefficient (e) mÀ1 cmÀ1 in brackets; shoulders are
indicated as sh. Mass spectra were measured as follows: HR-FT-ICR-
MALDI (m/z (%)): Varian IonSpec FT-ICR; MALDI-TOF (m/z (%)):
Bruker Daltonics UltraFlex II; matrix: [(2E)-3-[4-(tert-butyl)-phenyl]-2-
methylprop-2-enylidene]-malononitrile (DCTB) or 3-hydroxypyridine-2-
carboxylic acid (3-HPA). All peaks of the molecular cluster are reported
in m/z units with M+ as the molecular ion and intensities are given in
brackets.
5,10-BisACTHNUGRTENUNG[3,5-di(tert-butyl)phenyl]-15,20-bis(4-bromophenyl)porphyrinato-
N21,N22,N23,N24zinc(II) (8): To a 20 mL Schlenk flask under nitrogen,
charged with a solution of 7 (31 mg, 3.0ꢃ10À2 mmol) and 1-bromo-4-io-
dobenzene (17 mg, 6.0ꢃ10À2 mmol) in PhMe (10 mL), Cs2CO3 (128 mg,
0.4 mmol), [PdACTHNUTRGNEUNG
(PPh3)4] (4 mg, 3.0ꢃ10À3 mmol), and one drop of H2O
were added. After heating the mixture to reflux overnight, TLC monitor-
ing showed complete conversion of the starting material. After cooling
down, the mixture was filtrated over a plug and the solvent evaporated
in vacuo. FC (SiO2; hexane/CH2Cl2 1:1; 1% v/v Et3N) afforded 8 as a
dark brown solid (32 mg, quant.). M.p.>3008C; 1H NMR (300 MHz,
Sublimation experiments: To verify the thermal stability of molecules 1
and 2, sublimation tests were performed prior to the STM experiments at
10À5 mbar in a tailor-made sublimation apparatus, including a turbomo-
lecular vacuum pump and a liquid tin heating bath.[28] In a typical experi-
ment, the compound (1–3 mg) was sublimed at bath temperatures be-
tween 340 and 3708C within 10–30 min. Mass spectrometric analysis of
the material trapped on the cooling finger confirmed that porphyrin de-
rivatives 1 and 2 sublime without fragmentation or decomposition.
CDCl3, 258C): d = 9.05 (s, 2H), 9.04 (d, J
2H), 8.95 (d, J(H,H)=4.8 Hz, 2H), 8.12 (d, J
(d, J(H,H)=1.5 Hz, 4H), 7.91 (d, J(H,H)=5.1 Hz, 4H), 7.82 (t, JACHTUNGTRENNUNG
ACHTUNGTRENNUNG
A
ACHTUNGTRENNUNG
A
U
1.5 Hz, 2H), 1.55 ppm (s, 36H); 13C NMR (75.41 MHz, CDCl3, 258C): d
= 150.48, 150.35, 149.73, 149.70, 148.61, 141.45, 141.67, 141.21, 135.89,
132.43, 129.75, 122.81, 122.01, 119.40, 115.75, 35.09, 31.95 ppm; IR (neat):
n˜ =3061 (w), 2965 (s), 2944 (m), 2922 (m), 2862 (m), 2326 (w), 1593 (m),
1484 (m), 1362 (m), 1247 (m), 1204 (m), 1071 (m), 999 (s), 794 (s),
715 cmÀ1 (s); UV/Vis (CHCl3): lmax (e)=596 (1700), 554 (5400), 425 nm
(167100 mÀ1 cmÀ1); HR-MALDI-MS (3-HPA): m/z (%): calcd for
C60H58Br2N4Zn+: 1056.2314; found 1056.2297 (27) [M+], 1057.2345 (27),
1058.2293 (79), 1059.2321 (72), 1060.2285 (100), 1061.2294 (79),
1062.2274 (71), 1063.2288 (48), 1064.2278 (30), 1065.2271 (17).
STM measurements: The STM experiments were performed with a com-
mercial low temperature scanning tunneling microscope (LT-STM) (Omi-
cron NanoTechnology GmbH) under ultra-high vacuum (UHV, base-
pressure of 10À10 mbar) conditions at 77 K. Typical scanning parameters
used were around À1.5 V sample bias and 20 pA tunneling current. As
substrate, a CuACHTUNGTRENNUNG(111) single crystal was used which was cleaned by cycles
of sputtering with Ar+ ions and annealing at 800 K. The organic com-
pounds were sublimed from a commercial Knudsen-cell-type evaporator
(Kentax UHV equipment) onto the substrate held at room temperature.
5,10-BisACHTNUTRGNEUNG[3,5-di(tert-butyl)phenyl]-15,20-bis(4’-cyanobiphenyl)porphyrina-
to-N21,N22,N23,N24zinc(II) (1): To a 20 mL Schlenk flask under nitrogen,
charged with a solution of 8 (20 mg, 1.9ꢃ10À2 mmol) and 9 (10 mg, 4.1ꢃ
10À2 mmol) in PhMe (10 mL), Cs2CO3 (80 mg, 0.3 mmol), [Pd
ACHTUNGTRENNUNG(PPh3)4]
5,10-BisACHTUNGTRENNUNG[3,5-di(tert-butyl)phenyl]-15,20-bromoporphyrinato-
(4.4 mg, 4.0ꢃ10À3 mmol), and one drop of H2O were added. After heat-
ing the mixture to reflux overnight, TLC monitoring showed complete
conversion of the starting material. After cooling down, the mixture was
filtrated over a plug and the solvent evaporated in vacuo. FC (SiO2;
hexane/CH2Cl2 1:1; 1% v/v Et3N) afforded 1 as a dark brown solid
(15 mg, 72%). M.p. >3008C; 1H NMR (300 MHz, CDCl3, 258C): d =
N21,N22,N23,N24zinc(II) (6): To a 20 mL Schlenk flask charged with a solu-
tion of 5 (100 mg, 13.3ꢃ10À2 mmol) in CH2Cl2 (15 mL) and pyridine
(0.5 mL) under nitrogen at 08C, N-bromosuccinimide (NBS, 48 mg, 26.7ꢃ
10À2 mmol) was added in one portion. After stirring at 08C for 10 min,
the reaction was quenched with acetone (1.5 mL) and the solvent evapo-
rated in vacuo. FC (SiO2; hexane/CH2Cl2 2:1; 1% v/v Et3N) afforded 6 as
a dark brown solid (66 mg, 55%). m.p.>3008C; 1H NMR (300 MHz,
9.05 (s, 2H), 9.05 (d, J
(H,H)=4.5 Hz, 2H), 8.35 (d, J
1.8 Hz, 4H), 8.02 (d, J(H,H)=5.1 Hz, 4H), 7.98 (d, J
4H), 7.88 (d, (H,H)=5.1 Hz, 4H), 7.81 (t, (H,H)=1.8 Hz, 2H),
1.54 ppm (s, 36H); 13C NMR (150 MHz, CDCl3, 258C): d
150.65,
150.50, 149.99, 149.82, 148.62, 145.43, 143.50, 141.60, 138.16, 135.09,
ACHTUNGTRENNUNG
A
R
ACHTUNGTRENNUNG
CDCl3, 258C): d = 9.53 (d, J
2H), 8.95 (d, J(H,H)=4.8 Hz, 2H), 8.10 (d, J
(t,
(H,H)=1.8 Hz, 2H), 1.57 ppm (s, 36H); 13C NMR (75.41 MHz,
ACHTUNGTRENNUNG
A
ACHTUNGTRENNUNG
A
ACHTUNGTRENNUNG
J
A
J
N
JACHTUNGTRENNUNG
CDCl3, 258C, TMS): d =150.98, 150.85, 149.52, 149.40, 148.42, 140.86,
133.66, 132.75, 129.34, 123.76, 120.93, 112.24, 104.46, 35.19, 31.89 ppm; IR
=
11148
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11139 – 11150

Supramolecular Binding Motifs
FULL PAPER
132.86, 132.69, 132.52, 131.75, 131.60, 129.73, 127.95, 125.41, 123.08,
120.92, 119.81, 119.05, 111.15, 35.07, 31.76 ppm; IR (neat): n˜ = 3067 (w),
2960 (m), 2928 (m), 2868 (m), 2228 (m), 1606 (m), 1592 (m), 1490 (m),
1362 (m), 1248 (m), 1206 (m), 1006 (s), 797 (s), 720 cmÀ1 (s); UV/Vis
(CHCl3): lmax (e)=596 (1600), 554 (6400), 427 (205500), 406 nm (sh,
23700 mÀ1 cmÀ1); HR-MALDI-MS (3-HPA): m/z (%): calcd for
C74H66N6Zn+: 1102.4635; found 1102.4612 (89) [M+], 1103.4664 (98),
1104.4647 (100), 1105.4652 (81), 1106.4620 (73), 1107.4647 (52),
1108.4651 (25).
141.59, 138.19, 135.06, 132.87, 132.61, 131.71, 129.87, 127.97, 125.41,
122.88, 120.89, 120.03, 119.07, 111.15, 35.08, 31.76 ppm; IR (neat): n˜
=
2961 (m), 2927 (m), 2859 (m), 2225 (m), 1605 (m), 1592 (m), 1487
(m),1251 (m), 998 (s), 803 (s), 720 cmÀ1 (s); UV/Vis (CHCl3): lmax (e)=
597 (2400), 554 (6200), 427 (172000), 406 nm (sh, 20200 mÀ1 cmÀ1); HR-
MALDI-MS (3-HPA): m/z (%): calcd for C74H66N6Zn+ 1102.4635; found
1102.4649 (89) [M+], 1103.4696 (96), 1104.4657 (100), 1105.4660 (79),
1106.4624 (73), 1007.4643 (51), 1008.4682 (23).
5,15-BisACHTUNGTRENNUNG[3,5-di(tert-butyl)phenyl]-10,20-bis(4,4’,5,5’-tetramethyl-[1,3,2]di-
oxaborolan-2-yl)-porphyrinato-N21,N22,N23,N24zinc(II) (11): To a 50 mL
round-bottomed flask under N2 containing 10 (700 mg, 7.7ꢃ10À1 mmol)
in 1,2-dichloroethane (250 mL), pinacolborane (6.4 mL, 43.9 mmol),
Acknowledgements
[PdCl2ACHTUNGTRENNUNG
(PPh3)2] (54 mg, 7.7ꢃ10À2 mmol), Et3N (2.2 mL), and two drops of
This research was supported by the Swiss National Science Foundation,
the NCCR “Nanoscale Science”, Basel, the European Union through the
Marie-Curie Research Training Network PRAIRIES, contract MRTN-
CT-2006-035810, and the Wolfermann-Nꢄgeli-Stiftung. We also thank the
Swiss Federal Commission for Technology and Innovation, KTI, and
Nanonis Inc. for the fruitful collaboration on the data acquisition system.
H2O were added. After heating to reflux for 1.5 h, the mixture was
cooled down and filtered over Celite. After washing with brine (1ꢃ
75 mL) and water (2ꢃ50 mL), the mixture was dried (MgSO4) and the
solvent evaporated. FC (SiO2; hexane/CH2Cl2 1:2; 1% v/v Et3N) yielded
11 (773 mg, quant.) as a violet solid. M.p. >3008C; 1H NMR (300 MHz,
CDCl3, 258C): d
=
9.93 (d, J
N
ACHTUNGTRNE(NUNG H,H)=
4.8 Hz, 4H), 8.09 (d, J
A
1.86 (s, 24H), 1.56 ppm (s, 36H); 13C NMR (75.41 MHz, CDCl3, 258C):
d = 152.89, 150.03, 148.16, 141.68, 132.45, 129.45, 121.80, 120.54, 85.15,
35.21, 31.96, 25.54 ppm; IR (neat): n˜ = 2967 (s), 2323 (w), 2050 (w), 1702
(w), 1592 (m), 1528 (s), 1344 (s), 1297 (m), 1269 (s), 1138 (s), 1000 (s),
854 (s), 794 (s), 713 (s), 660 mÀ1 (s); UV/Vis (CHCl3): lmax (e)=588
(2400), 552 (8100), 421 nm (173000 mÀ1 cmÀ1); HR-MALDI-MS (3-HPA):
m/z (%): calcd for C60H74B2N4O4Zn+: 1000.5182; found 1000.5181 (100)
[M+], 1001.5196 (87), 1002.5179 (84), 1003.5179 (69), 1004.5158 (62),
1005.5185 (32), 1006.5219 (12).
[1] A. C. Grimsdale, K. Mꢁllen, Angew. Chem. 2005, 117, 5732–5772;
Angew. Chem. Int. Ed. 2005, 44, 5592–5629.
[2] Y. S. Zhao, H. B. Fu, A. D. Peng, Y. Ma, D. B. Xiao, J. N. Yao, Adv.
Mater. 2008, 20, 2859–2876.
[3] S. Fukuzumi, T. Kojima, J. Mater. Chem. 2008, 18, 1427–1439.
[4] K. Ariga, J. P. Hill, M. V. Lee, A. Vinu, R. Charvet, S. Acharya, Sci.
Technol. Adv. Mater. 2008, 9, 014109/1–96.
[5] A. Zangwill, Physics at Surfaces, Cambridge University Press, Cam-
bridge (UK), 1990.
[6] F. S. Tautz, Prog. Surf. Sci. 2007, 82, 479–520.
[7] S. De Feyter, F. C. De Schryver, Chem. Soc. Rev. 2003, 32, 139–150.
[8] S. De Feyter, A. Gesquiere, M. M. Abdel-Mottaleb, P. C. M. Grim,
F. C. De Schryver, C. Meiners, M. Sieffert, S. Valiyaveettil, K.
Mꢁllen, Acc. Chem. Res. 2000, 33, 520–531.
[9] N. Wintjes, D. Bonifazi, F. Cheng, A. Kiebele, M. Stçhr, T. A. Jung,
H. Spillmann, F. Diederich, Angew. Chem. 2007, 119, 4167–4170;
Angew. Chem. Int. Ed. 2007, 46, 4089–4092.
[10] T. Yokoyama, S. Yokoyama, T. Kamikado, Y. Okuno, S. Mashiko,
Nature 2001, 413, 619–621.
[11] U. Schlickum, R. Decker, F. Klappenberger, G. Zoppellaro, S. Klyat-
skaya, M. Ruben, I. Silanes, A. Arnau, K. Kern, H. Brune, J. V.
Barth, Nano Lett. 2007, 7, 3813–3817.
[12] N. Lin, A. Dmitriev, J. Weckesser, J. V. Barth, K. Kern, Angew.
Chem. 2002, 114, 4973–4977; Angew. Chem. Int. Ed. 2002, 41, 4779–
4783.
[13] G. P. Arsenault, E. Bullock, S. F. MacDonald, J. Am. Chem. Soc.
1960, 82, 4384–4389.
[14] K. M. Kadish, K. M. Smith, R. Guilard, J. S. Lindsey, The Porphyrin
Handbook, Vol. 1, Academic Press, San Diego, 2000, p. 88.
[15] Y. Nakamura, I. W. Hwang, N. Aratani, T. K. Ahn, D. M. Ko, A.
Takagi, T. Kawai, T. Matsumoto, D. Kim, A. Osuka, J. Am. Chem.
Soc. 2005, 127, 236–246.
[16] T. Ishiyama, M. Murata, N. Miyaura, J. Org. Chem. 1995, 60, 7508–
7510.
[17] F. Y. Cheng, S. Zhang, A. Adronov, L. Echegoyen, F. Diederich,
Chem. Eur. J. 2006, 12, 6062–6070.
[18] A. D. Bond, N. Feeder, J. E. Redman, S. J. Teat, J. K. M. Sanders,
Cryst. Growth Des. 2002, 2, 27–39.
[19] R. Krishna Kumar, S. Balasubramanian, I. Goldberg, Mol. Cryst.
Liq. Cryst. 1998, 313, 105–114.
[20] F. Buchner, K. Comanici, N. Jux, H.-P. Steinrꢁck, H. Marbach, J.
Phys. Chem. C 2007, 111, 13531–13538.
[21] T. A. Jung, R. R. Schlittler, J. K. Gimzewski, Nature 1997, 386, 696–
698.
5,15-BisACHTUNGTRENNUNG(3,5-di(tert-butyl))-10,20-bis(3-bromophenyl)-porphyrinato-
N21,N22,N23,N24zinc(II) (12): A 20 mL Schlenk flask charged with 11
(100 mg, 0.1 mmol) and 1-bromo-4-iodobenzene (57 mg, 0.2 mmol) in tol-
uene (10 mL) was degassed for 30 min. Then [PdACHTNURTGNEUNG(PPh3)4] (12 mg, 1.0ꢃ
10À2 mmol), Cs2CO3 (423 mg, 1.3 mmol), and two drops of distilled water
were added and the mixture heated to reflux overnight. After TLC indi-
cated complete conversion, the crude product was cooled down, filtered
through a plug (SiO2), and the solvent evaporated in vacuo. FC (SiO2; cy-
clohexane/CH2Cl2 1:1; 1% v/v Et3N) yielded 12 (106 mg, quant.) as a
violet solid. M.p. >3008C; 1H NMR (300 MHz, CDCl3, 258C): d = 9.70
(d,
(H,H)=1.8 Hz, 4H), 8.13 (d, J
5.1 Hz, 4H), 7.85 (t, J
(H,H)=1.8 Hz, 2H), 1.57 ppm (s, 36H); 13C NMR
J
N
JACHTUNGTRENNUNG
J
N
N
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
(75.41 MHz, CDCl3, 258C): d = 150.44, 149.62, 148.49, 141.70, 141.45,
141.38, 135.53, 132.61, 129.69, 122.77, 122.04, 119.32, 115.74, 35.12,
31.81 ppm; IR (neat): n˜ = 3054 (w), 2978 (s), 2955 (m), 2923 (m), 2821
(m), 2301 (w), 1599 (m), 1463 (m), 1372 (m), 1236 (m), 1210 (m), 1075
(m), 1005 (s), 789 (s), 720 cmÀ1 (s); UV/Vis (CHCl3): lmax (e) = 596
(7800), 553 (16200), 425 nm (159900 mÀ1 cmÀ1); HR-MALDI-MS (3-
HPA): m/z (%): calcd for C60H58Br2N4Zn+: 1056.2314; found 1056.3308
(28) [M+], 1057.3355 (27), 1058.3300 (83), 1059.3333 (74), 1060.3290
(100), 1061.3309 (80), 1062.3275 (69), 1063.3286 (47), 1064.3265 (30),
1065.3271 (15).
5,15-Bis
N21,N22,N23,N24zinc(II) (2):
(80 mg, 7.5ꢃ10À2 mmol) and
(8 mL) was degassed for 30 min. Then [PdAHCUTNGTRENNUNG
ACHTUNGTRENNUNG
A
9
10À2 mmol), Cs2CO3 (320 mg, 98ꢃ10À2 mmol), and one drop of distilled
water were added and the mixture heated to reflux overnight. After TLC
indicated complete conversion, the crude product was cooled down and
filtered through a plug (SiO2) and the solvent evaporated in vacuo. FC
(SiO2; cyclohexane/CH2Cl2 1:9; 1% v/v Et3N) yielded 2 (17 mg, 30%) as
a purple solid. M.p. >3008C; 1H NMR (300 MHz, CDCl3, 258C): d =
9.04 (d, J
N
ACHTUNGERTN(NUNG H,H)=4.5 Hz, 4H), 8.35 (d,
J
J
N
JACHTUNGTRENNUNG
[22] T. Yokoyama, S. Yokoyama, T. Kamikado, S. Mashiko, J. Chem.
Phys. 2001, 115, 3814–3818.
G
ACHTUNGTRENNUNG
7.8 Hz, 4H), 7.81 (t, J
(150 MHz, CDCl3, 258C): d
G
[23] T. Yokoyama, T. Kamikado, S. Yokoyama, S. Mashiko, J. Chem.
Phys. 2004, 121, 11993–11997.
=
Chem. Eur. J. 2009, 15, 11139 – 11150
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11149

F. Diederich, L.-A. Fendt, M. Stçhr, T. A. Jung et al.
[24] A. Llanes-Pallas, M. Matena, T. Jung, M. Prato, M. Stçhr, D. Bonifa-
zi, Angew. Chem. 2008, 120, 7840–7844; Angew. Chem. Int. Ed.
2008, 47, 7726–7730.
[28] N. Wintjes, J. Hornung, J. Lobo-Checa, T. Voigt, T. Samuely, C. Thil-
gen, M. Stçhr, F. Diederich, T. A. Jung, Chem. Eur. J. 2008, 14,
5794–5802.
[25] J. V. Barth, J. Weckesser, C. Cai, P. Gꢁnter, L. Bꢁrgi, O. Jeandupeux,
K. Kern, Angew. Chem. 2000, 112, 1285–1288; Angew. Chem. Int.
Ed. 2000, 39, 1230–1234.
[29] J. P. Hill, Y. Wakayama, K. Ariga, Phys. Chem. Chem. Phys. 2006, 8,
5034–5037.
[30] T. Yokoyama, personal communication.
[26] G. R. Desiraju, Acc. Chem. Res. 2002, 35, 565–573.
[27] K. E. Maly, T. Maris, E. Gagnon, J. D. Wuest, Cryst. Growth Des.
2006, 6, 461–466.
Received: June 3, 2009
Published online: September 16, 2009
11150
www.chemeurj.org
ꢂ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11139 – 11150