Zinc porphyrin-driven assembly of gold nanofingers.

Abstract

Nanofingers of gold covered by porphyrins are prepared by a combination of atomic manipulation and surface self-organization. A submonolayer of zinc(II) 5,10,15,20-tetrakis(4-tert-butylphenyl)-porphyrin (ZnTBPP) axially ligated to a self-assembled monolayer of 4-aminothiophenol (4-ATP) on Au(111) is prepared and studied using a combination of ultrahigh vacuum techniques. Under the electric field produced by the STM tip, the relatively weakly bound Au surface atoms along the discommensuration lines become mobile due to the strong bond to 4-ATP, while the tendency of the porphyrins towards self-assembly result in a collective motion of gold clusters. The clusters diffuse onto the surface following well-defined pathways along the [112] direction and then reach the step edges where they assembled, thus forming nanofingers. First-principles density functional theory calculations demonstrate the reduction of the binding energies between the surface gold clusters and the substrate induced by adsorption of thiols. Scanning tunneling microscopy images show assemblies across three adjacent discommensuration lines of the Au(111)-(22 x square root 3) reconstruction, which collectively diffuse along these lines to form islands nucleated at step edges.

Full text

Page 1

1
full papers
((Subject Heading))
DOI: 10.1002/smll.200((……))
Zinc porphyrin - driven assembly of gold
nanofingers
Valentina Arima*, Robert I. R. Blyth, Francesca Matino, Letizia Chiodo, Fabio Della
Sala, Julie Thompson, Tom Regier, Roberta Del Sole, Giuseppe Mele, Giuseppe
Vasapollo, Roberto Cingolani, Ross Rinaldi
Nanofingers of gold covered by porphyrins were prepared by a
combination of atomic manipulation and surface self-organization. A sub-
monolayer of Zinc(II) 5,10,15,20-tetrakis(4-tert-butylphenyl)-porphyrin
(ZnTBPP) was axially ligated to a self-assembled monolayer of 4-
aminothiophenol (4-ATP) on Au(111). Under the electric field produced
by the STM tip, the relatively weakly bound Au surface atoms along the
discommensuration lines became mobile due to the strong bond to 4-ATP,
while the tendency of the porphyrins towards self-assembly resulted in a
collective motion of gold clusters. The clusters diffused onto the surface
following well-defined pathways along the
] 2[11
direction and then
reached the step edges where they assembled, thus forming the
nanofingers. We performed first principles density functional theory
calculations to demonstrate the reduction of the binding energies between
Keywords:
• porphyrins
• nanofingers
• scanning tunnelling microscopy

Page 2

2
the surface gold clusters and the substrate induced by adsorption of thiols.
The selective transport of gold clusters linked to 4-ATP/ZnTBPP and the
assembling process were characterized by ultra high vacuum scanning
tunneling microscopy and spectroscopy (UHV STM and STS), synchrotron
radiation excited photoemission spectroscopy (PES) and near edge X-ray
absorption fine structure (NEXAFS). The STM images showed assemblies
across three adjacent discommensuration lines of the Au(111)-(22×v3)
reconstruction, which collectively diffused along these lines, to form
islands nucleated at step edges. The tunneling conditions used, the
acquired STS spectra and the apparent heights observed suggest that gold
clusters, rather than organic molecules, are being directly imaged.
Photoemission and NEXAFS measurements confirmed the presence of the
porphyrins at sub-monolayer coverage and showed ZnTBPP was
chemisorbed on the surface without a preferred orientation of the
porphyrin ring.



1. Introduction
Porphyrins and gold nanowires are both of great interest
in optoelectronics for different reasons. The unique optical,
photoelectrochemical and chemical properties of such
molecules together with their tendency to form ordered
structures in the solid state,1 have been already exploited in
a number of devices2-10. Gold nanowires are nowadays
emerging as interconnections or active components11 in
nanodevices because of their capability to increase the
spectroscopic response from chromophores adsorbed on
them12. The combination of porphyrins with gold nanowires
[*] Dr. V. Arima, Dr. R.I.R. Blyth, Dr. F. Matino, Dr. L. Chiodo,
Dr. F. Della Sala, Dr. J. Thompson, Prof. R. Cingolani, Prof. R.
Rinaldi
National Nanotechnology Laboratory (CNR-INFM)- Distretto
Tecnologico ISUFI – Università degli studi di Lecce
via Arnesano, 73100 Lecce (ITALY)
Fax: +39 0832 298180
Email: valentina.arima@unile.it
Dr. R.I.R. Blyth, Dr. T. Regier
Canadian Light Source, 101 Perimeter Road, University of
Saskatchewan, Saskatoon, S7N 0X4 (CANADA)
Dr. J. Thompson
Dept. of Chemistry, University of Western Ontario,
London, Ontario, N6A 5B7 (CANADA)
Dr. R. Del Sole, Dr. G. Mele, Prof. G. Vasapollo
Dipartimento di Ingegneria dell'Innovazione, Universita' degli Studi
di Lecce
via Arnesano, 73100 Lecce (ITALY)

Page 3

3
seems to be an attractive formula for efficient optoelectronic
devices. An important issue is how to fabricate gold
nanowires covered by porphyrins.
Although gold microwires covered by laser dye are
easily fabricated and appear to form efficient devices12, the
preparation of gold nanowires covered by chromophores is
rather difficult because of their low stability at room
temperature11. 1-dimensional nanowires have, however,
already been produced by depositing 0.15 ML of gold on
Si(111), as investigated by Scanning Tunneling Microscopy
(STM) and photoemission spectroscopy11a.
Recently, Yin et al.13a have reported the possibility of
producing “gold fingers” on the (111) gold surface by
structuring the Au(111) surface at the nanometer scale with
an STM tip. It seems that local atomic diffusion can be
induced on the Au surface since the activation energy is
reduced by applying a voltage pulse to the tip. The second-
layer atoms are exposed by physically extracting top-layer
atoms with an STM tip. After the removal of the top-layer
gold atoms from a step edge, the step does not recede
uniformly and stripes of a few nanometers width are formed.
The mobility of gold atoms is also increased when
aromatic thiols14 are immobilized on the surface because the
strong Au–S bond weakens the adhesion between the surface
Au atoms and the underlying layers. The mobile Au atoms
aggregate and form islands. The strong intermolecular
interactions favour aggregations of the adsorbed aromatic
molecules, each with an Au atom attached, into islands.
In this work, we immobilized a sub-monolayer of the
chromophore Zinc(II) 5,10,15,20-tetrakis(4-tert-
butylphenyl)-porphyrin (ZnTBPP), by means of an aromatic
thiol monolayer, the 4-aminothiophenol (4-ATP) SAM, on
the Au(111) surface, and then gold clusters bound to the
molecules were moved under the electric field produced by
the STM tip. The molecular formulas and the scheme of the
4-ATP/ZnTBPP complex assembled on the gold surface are
shown in figure 1. Since the bond of thiols to gold seems to
weaken the interaction between the surface Au atoms and
the underlying layers14 and 4-ATP/ZnTBPP is a very stable
pyramidal square complex15, which can be easily polarized
by the tip field16a, we suggest that the gold clusters linked to
the molecules could be moved by a tip field weaker than that
required for non-functionalized gold.
In order to verify that the thiol-cluster bond reduced in
some way the interaction between the cluster itself and the
substrate, inducing therefore a higher mobility of the cluster,
we performed first principles density functional theory
(DFT) calculations.
In some theoretical studies17-19b diffusion of small
clusters on different (111) and (001) surfaces has been
investigated and described by embedded atom models and
molecular dynamics, thus determining energy barriers along
different diffusion paths and studying dynamic features such
as collective motion of groups of atoms in the cluster17-19b,
rolling for spherical clusters18 and internal motions inside
the clusters themselves19. However, these models could not
be easily applied to our case because they don’t take into
consideration that molecules absorbed on clusters strongly
affect their mobility. As a consequence, a ‘static’ approach,
reporting some structural, energetic and electronic properties
of bare clusters or thiol-covered clusters adsorbed on gold,
has been used in this work to simulate in a simple way the
complex experimental structure. Our model relies on the
observations that i) adsorbed clusters are mainly monolayers
and ii) the thiol bond effect is localized around the region of
the S -Au bond. We used a monoatomic small gold cluster,
adsorbed on an Au(111) surface, and a small thiol molecule,
methanethiol, to mimic the whole (4-ATP)
SAM/cluster/Au(111) and we found that the increased

Page 4

4
mobility of the gold clusters functionalized by the molecules
is in fact consistent with a reduction of the interaction
between the cluster itself and the substrate.
The selective transport of gold clusters linked to 4 -
ATP/ZnTBPP and the assembly process of functionalized
gold nanowires were experimentally followed and
characterized by scanning tunneling microscopy (STM) and
spectroscopy (STS),20
synchrotron radiation excited
photoemission spectroscopy (PES) and Near Edge X -ray
absorption fine structure (NEXAFS)21. Finally, gold
nanowires covered by ZnTBPPs were successfully isolated
at the step edges.
The bare gold (111) substrate has been firstly
characterized without molecules and with a 4-ATP SAM as
reported in our previous paper16b,c. The interaction of the
thiol SAM with Cobalt (II) 5,10,15,20-tetrakis(4-tert-
butylphenyl)-porphyrin (CoTBPP) has also been
investigated16a,c. In the case of ZnTBPP, the STM images
showed assemblies across three adjacent discommensuration
lines of the Au(111)-(22×v3) reconstruction, which
collectively diffused along these lines, to form islands
nucleated at step edges. The tunnelling conditions used, the
acquired STS spectra and the apparent heights observed
suggest that gold clusters, rather than organic molecules, are
being directly imaged. The noisy structures near the step
edges at higher voltages and the corresponding STS spectra
suggest that the clusters, mobile on the surface, are covered
by molecules. PES and NEXAFS measurements showed that
ZnTBPPs are chemisorbed on the surface without a preferred
orientation of the porphyrin ring and induce a significant
distortion of the gold local electronic structure. We suggest
that the relatively weakly bound Au surface atoms linked to
the 4-ATP/ZnTBPP complex become strongly mobile at very
low (tip) electric fields, while the tendency of the molecules
towards self-assembly4 results in the observed collective
motion.
2. Results
2.1. Experimental results
In the STM images here reported and acquired on an
Au(111) sample functionalized by 4-ATP and ZnTBPP
molecules, small clusters moving along the surface and
modifications occurring at the step edges can be observed.
The movement is induced by the tip field as already shown
by Guo et al.13b We believe that Au islands covered by
molecules move onto the surface and then reach the step
edges as suggested by STM and STS measurements, thus
creating new structures near the steps. This is supported by
several observations, which we report in the following.
The apparent height of the islands shown in figure 2 is
about 2.3±0.4 Å, which corresponds to the height of a single
Au atomic layer. The clusters have lateral dimensions of
4.86±0.53nm and 7.89±0.80nm. Some of them appear
elongated thus forming stripes of variable dimensions. The
shape of the clusters doesn’t appear related to the features of
the substrate.
We studied the formation of the islands in real time STM
by acquiring consecutive images on the same area. Each
scan took about 6 min. If we observe the structures in figure
2 it seems that the sizes of the clusters don’t change too
much, but the number of the islands and stripes decreases
gradually in this area. In another area of the sample more
regular clusters were found and they appear essentially
4.07±0.62nm wide, 10.84±1.89nm long and 2.2±0.2Å high.
The tip effect and diffusion phenomena can influence the
shape and the lateral dimensions of the clusters, thus giving
apparent different features to the islands.

Page 5

5
From the images of figure 2 is not easy to understand
why the islands disappeared and where they moved. A
possible explanation comes from other images acquired in
sequence near the step edges (see figure 3). Each scan took
about 5 min. Au surface step edges expand when the nearby
islands dissolve because the Au atoms in the islands are
being incorporated into the steps. Some islands of size
similar to those previously described, move along the
surface and, finally, reach the step edges where they are
dissolved, thus inducing an expansion of the step edges. The
surface is continuously modified and holes of different size
but with a depth of about 1.6±0.3Å can continuously open
and close in several positions of the sample.
All the above images were acquired at 0.2V, a voltage
for which only gold and not molecules can be reasonably
visualized. In looking for the molecules, the voltage was
varied until noise in some regions appeared. In the two
insets of figure 4, two images acquired in sequence at
different voltages are shown. The area shown in the inset of
figure 4a was imaged at 0.2V and 0.8nA, while that in 4b
was acquired at same set point current but at a voltage of
1V. On increasing the voltage, in the region of the image
indicated by the blue rectangular box a noisy pattern
appeared, while in the other parts the scanning was
undoubtedly more stable and quite similar to the image
recorded at low voltages.
The STS characteristics extrapolated from a grid of
points collected in the same area reflect this phenomenon
(see figure 4). In the two curves of figure 4a, averages of
spectroscopic points acquired respectively in the regions
indicated by the rectangular boxes (see image in the inset), a
behaviour typical of metals and due to the conduction of
gold can be recognized. The two curves of figure 4b were
extrapolated respectively from the noisy region indicated by
the blue rectangular box (blue line) and from the region
indicated by the black rectangular box (black line) of the
image shown in the inset and acquired at 1V, 0.8nA. No
zero-current region, characteristic of semiconductor
behaviour, can be recognized in the black curve. The curve
is featureless and essentially due to the conduction of gold.
The STS features of the noisy region of the image (blue line)
are quite different since the negative and positive currents
start increasing at ±0.3V, approximately where the
corresponding Normalized Conductance curve (the blue line
in the inset) shows some peaks. We suggest that the blue
curve is a combination of the gold and ZnTBPP features and
therefore the zero-current region cannot be directly related
to the HOMO-LUMO gap of ZnTBPP.29 It is possible also to
hypothesize that in the region of the image (inset of figure
4b) indicated by the blue rectangular box, where STM
images show phenomena suggesting high cluster mobility,
the noise indicates molecules acting on the surface. In the
other part of the image, where the mobility is reduced, few
molecules are present and the acquired STS characteristic is
essentially due to the gold surface.
On a more crystalline Au(111) sample we clearly
observed the formation of gold stripes of a few nanometers
width. Figure 5a shows that, after the deposition of ZnTBPP
on the 4-ATP-functionalized Au(111) surface, well-defined
assemblies of regular shape and dimensions appear on a
regular 22 ×v3 surface reconstruction. The assemblies were
observed to move on the thiol-functionalized surface,
following well-defined pathways along the
] 2 [11
direction.
We suggest that the "streaking" of the images in the lower
part of figure 5a is due to the motion of these assemblies
during the timescale of the scan, rather than accumulation of
multiple assemblies. The assemblies are 3.87±0.71nm wide,
12.71±0.60nm long and 1.9±0.3Å high. Though they appear
different, these clusters can be associated with those

Page 6

6
previously described. More detailed information about the
features of these clusters and their movements can be
extrapolated from these images. The observed structures
seem to consist of three basic units, with a separation
between the first and third units of 8.09±0.42nm and
between the first and second units of 2.87±0.47nm.
Considering that unlike true single-crystal gold, the thin
Au(111) films here used can show reconstruction line
spacing ranging from 6.3 to about 9.0 nm,30 the distances
between the units of the clusters appear closely related to the
periodicity of the underlying surface reconstruction and to
the distance between two nearest neighbour parallel
discommensuration lines. Moreover the molecular spots
appear (figure 5a) more defined near the step edge of the
gold terraces and more elongated along the diffusion
direction far from the step edge, thus suggesting that the
“speed” of their motion is related to their position with
respect to the edge.
In the vicinity of a step-edge the density of molecular
units, figure 5b, is much higher and there is clearly two
dimensional island formation, and a well-defined diffusion
direction. A line scan along across neighbouring 2-
dimensional islands is also shown in figure 5c. In order to
cross a large number of 2-dimensional islands this line scan
is at an angle of ~30º with respect to the [] 101
direction,
and the distance scale has been corrected to take this into
account. From this the periodicity of the adjacent minima
can be determined to be ~9.17nm, which, given the
uncertainties in this measurement, is again consistent with
the periodicity of the underlying Au(111)-(22xv3)
reconstruction.
Since it was possible to clearly image the fingers on a
less crystalline sample and simultaneously resolve the
herringbone reconstruction, the possible directions of the
fingers were identified.
The long axis of some fingers shown in Figure 6 makes a
145° angle to the discommensuration lines on the terrace. It
has been verified, by changing the fast-scan direction by 45°,
that this parameter doesn’t affect the direction of the fingers.
It seems that the fingers prefer to align in this way as
already shown by Guo et al13b. Fingers perpendicular to the
discommensuration lines have also been observed as
illustrated in Figure 6. The directions of bare gold fingers on
Au(111) 13b were found to be in good agreement with those
here described, thus confirming that the observed
phenomena have a common origin.
Clear evidence of the presence of ZnTBPP, and of its
influence on the gold electronic structure are shown by the
photoemission and N K NEXAFS spectra of figs. 7-8.
Fig. 7a shows the Au NNV Auger peak and the Zn 2p
photoemission peaks. The 2p is the highest intensity
photoemission peak for Zn, but sits on an intense secondary
electron background due to the Au N -shell photoelectrons,
and thus observing the Zn peaks is non-trivial. We were
unable to resolve these peaks on the (bending magnet)
BEAR beamline. The spectrum of fig. 7a was taken using a
wide exit-slit setting on the SGM undulator beamline, with a
correspondingly high photon flux, of ~1012 photons/sec. This
is almost certainly sufficient to damage the film, but does
serve to conclusively demonstrate the presence of the
porphyrin on the surface. It is clear from the quality of the
data of fig. 7a why attempts at higher resolution
spectroscopy, with correspondingly lower flux, did not result
in useful data.
The sulphur 2p photoemission spectrum is shown in fig.
7b, together with the results of a curve fit, using similar
parameters to that of Au/4-ATP16b. The spectrum is

Page 7

7
analogous to that observed for Au/4-ATP16b, showing a
number of different components as discussed elsewhere16b,
but here it is better resolved, allowing a more reliable
determination of peak widths. The width of the component
bonded directly to Au, at a 2p3/2 binding energy of 162.1
eV16b, is found to be only 0.6 eV and appears well defined.
This is in contrast to the value of 1.8 eV found for Au/4-
ATP16b, and suggests the presence of a single S -Au species,
rather than the multiple sites speculated in ref. 16b.
Although the resolution in the present experiment is rather
higher, this improvement alone cannot account for the
observed difference in width.
The N 1s photoemission spectrum is shown in fig. 8a,
together with the results of a curve fit. Here the presence of
two components is indicated, as it is also evident from a
visual inspection of the raw data. The lower energy of the
two components, at 399.6 eV, corresponds to the value for
4-ATP16b, while the higher energy component corresponds to
the porphyrin16a. The presence of only one porphyrin-related
N 1s peak is indicative of a metal-porphyrin, since metal-
free porphyrin displays a double N 1s peak due to protonated
and unprotonated nitrogen species31. The presence of a
significant 4-ATP contribution to the N 1s peak is in
contrast with the observations for Au/4-ATP/CoTBPP16a,
where a full monolayer was implied, and is consistent with
the apparent sub-monolayer coverage seen in the STM data.
Nitrogen K-edge NEXAFS spectra of Au/4-
ATP/ZnTBPP (black curves) are shown in figure 8b. The
NEXAFS spectra of metal-porphyrins display a double peak
around 400eV due to the unoccupied π* resonances32. These
are labelled as A and B in figure 8b. In the event of a
preferred orientation of the porphyrin ring, the intensities of
these peaks should show considerable variation with the
incidence angle of the photon beam. Figure 8b shows that
there is indeed angular variation in the NEXAFS spectra.
However, the spectra are a superposition of both ZnTBPP
and 4-ATP signals. The NEXAFS spectra of Au/4-ATP (red
lines) also show significant angular variation16b, as is
indicated by the two spectra of Au/4-ATP included in figure
8b. From a comparison of these spectra with those of Au/4-
ATP/ZnTBPP it appears that the angular variation seen in
the latter spectra may be due entirely to the 4-ATP
molecules. Further, the intensity of peak A does not show
any angular variation. It therefore implies that there is no
preferred orientation of the ZnTBPP molecules.
2.2. Theoretical results
Since the thiol-gold interaction has a local character, i.e.
it is localized around the region of the S-Au bond, we chose
both a small adsorbed molecule and a small gold cluster to
describe large organic molecules on large area clusters,
always preserving the energetic bond description.
We computed the binding energies for the different
analyzed systems using the formula
totadssubb
EEEE
−+=
)(

where
b E is the binding (or adsorption) energy,
sub
E
is
the energy of the relaxed substrate system, that is or the gold
slab (in the following denoted by S) or the nanocluster (N)
or the cluster/slab system (NS), and
ads
E
is the energy of
the relaxed adsorbate, that is or the isolated thiolate
molecule CH3S (denoted by M) or the nanoclusters (N) or
the whole thiolate/cluster adsorption unit (MN). Finally,
tot
E
is the energy of the relaxed full system.
The relaxed configurations for the gold nanoclusters on
gold surface (NS) and for the thiol plus gold nanoclusters on
gold surface (MNS) are shown in Figures 9a and 9b
respectively.

Page 8

8
The computed Au-S bond (MN-S) is 2.45 Å (and 2.48 Å
for the thiol on gold surface, see also the value 2.50 Å in
ref.27), and the S-C bond is 1.84 Å. The angle between the
molecular axis and the normal to the surface, which is
58.62° in the thiol/surface case34, is tilted to 59.85° for the
thiol/cluster/substrate system.
The calculated binding energies for different subsystems
are reported in Table 1. We obtain an adsorption energy of
thiolate on bare gold (M-S) of 1.87 eV, in good agreement
with available experimental data, 1.7 ±0.2 eV35, and
comparable with previous theoretical data, 1.73 eV27 and 2.3
eV33. This binding energy increases to 2.76 eV for the
molecule adsorbed on the isolated cluster (M-N) (3.3 eV in
ref.33), and to 2.06 eV for the molecule adsorbed on
cluster/surface system (M-NS). Therefore, the
molecule/cluster (M-N) bond, even if lowered in energy by
the substrate presence, is stronger than the pure
molecule/surface (M-S) interaction by 9.7%. The calculated
adsorption energy for the bare cluster on gold surface (N-S)
is 4.11 eV, and this energy decreases to 3.41 eV when a thiol
is adsorbed on the cluster (MN-S), with a variation of 19%
in bond strength.
In Fig. 10 we report the charge density variations for NS
and MNS along the direction z normal to the surface,
averaged on the surface plane: the three-dimensional charge
density difference for NS (above) and MNS (below) are also
plotted. Fig. 10 clearly shows that the presence of the
adsorbed thiol modifies the charge distribution. In fact, part
of the electron density accumulation, initially localized on
the adsorbed cluster, is shifted toward the cluster-slab bond,
while part is shifted on the molecule.
3. Discussion
Since the STM voltages used are too low to visualize the
orbitals of either 4-ATP or ZnTBPP, and the apparent height
of the observed structures is very similar to the step height
of Au(111), it appears that the part of the structures that is
being imaged consists of gold atoms or clusters that have
been pulled out of the topmost surface layer. This may be
supported by the S 2p photoemission spectra, which appear
to show a considerably more well-defined S-Au bond than it
was the case for Au/4-ATP, where multiple absorption sites
on the herringbone reconstruction are a possibility. All the
STM images acquired clearly show clusters of gold atoms
and surface modifications at the step edges. It seems that
these phenomena are correlated. Despite small differences in
the shape of the clusters, their origin and their destination
seem unequivocal: Au islands covered by molecules move
onto the surface following precise pathways and then reach
the step edges, thus creating new structures near the steps.
These structures show at atomic level the typical
herringbone reconstruction of Au(111). The photoemission
and nitrogen K-edge NEXAFS spectra confirm that 4-ATP
and ZnTBPP molecules are, in fact, present on the surface.
At the same time, STS and voltage dependent STM images
suggest that molecules are located mainly at the step edge,
where the phenomenon of cluster mobility appears more
evident, and determines the growth of the fingers.
The apparent diffusion seen in the figures 2-5a could be
caused by a combination of different effects. Theoretical
calculations show that methanethiolate through the strong
Au–S bond reduces the adhesion between the gold adatoms
and the substrate. Due to the local character of the Au-S
bond, these theoretical results can be assumed to be valid for
all sulphur-anchoring organic molecules such as the
aromatic thiols (4-ATP) used in the experiments. In addition,
in our case, the surface Au atoms become much more mobile
than the atoms on bare Au surfaces also due to the

Page 9

9
interaction between adjacent molecules, which aggregate
and form islands. A comparison of the 2-dimensional islands
in the upper part of figure 5a with the "streaking" in the
lower area of the same figure further suggests that what is
seen in the latter images is in fact diffusion. Molecular
diffusion on the surface can be understood if we consider
figure 5a as the initial phase and figure 5b as the final phase
of the molecular migration. This is shown schematically in
figure 5d.
Collective movements of gold clusters covered by
aromatic thiols (4HP)14b and their incorporation at the step
edge have been already followed by STM in a liquid
environment. In that case the phenomenon of coalescence of
islands of similar size into bigger ones, and the tip effect
was excluded. In our images, and particularly in figure 5a,
the assemblies don’t change their shape during the
experiment but only diffuse along the surface.
If thiols are responsible for the high mobility of the gold
atoms, the porphyrins, which can strongly interact with each
other by p-p interactions, 36 could influence the assembly of
the units. It is not clear why the self-assemblies consist of
three units on neighbouring discommensuration lines, and
not four or more. It is possible that the actual size of the
assemblies is slightly larger than the periodicity of the
underlying reconstruction, as discussed earlier, in which
case there would not be space for a fourth unit.
However, despite these difficulties in the explanation of
how random oriented ZnTBPPs interact on the top of the
clusters - which is not clear also in the case of gold clusters
covered by 4HP - there is a significant difference between
our assemblies a nd those described by Guo et al.14b: our
assemblies don’t modify their shape during the STM
timescale and move very fast compared with the islands
shown in ref. 14b. Our islands disappeared in about 5min
(see figure 2), while the 4HP islands slowly move or modify
their shape, on a timescale of 30min. It seems that our
assemblies haven’t the time to change their shape because
the relatively fast movement towards the step, facilitated by
the discommensuration lines, limits interactions between
them before reaching the steps.
We suggest that another effect, not only diffusion, may
help to drive the nanofinger formation process: Guo et al.13b
performed nanoscale surface modification of the Au(111)
surface by scanning the STM tip. Keeping the tunnel voltage
unchanged (about 1.5 V), but increasing the tunnel current
from less than 0.1 nA to 30 nA, they succeeded in modifying
the surface step. At the initial stage the step was changed by
extracting atoms by the STM tip. This process required a
strong electric field below the STM tip: reduction of the bias
to 0.1 V while keeping the tunneling current constant at 30
nA, sufficed to prevent atom extraction. After formation of a
finger, its length was increased with repeated scanning.
They suggested that the increase in finger length comes from
two contributions: 1) the extraction of atoms from the step
edge that connects the base of two neighbouring fingers, and
2) the atom attachment to the fingertips. They verified that
atom extraction from a step edge requires a high electric
field below the STM tip (1.0 V, 30 nA), while atom
attachment to the fingertips can occur under normal
scanning conditions (1.0 V, 0.1 nA) because gold atoms on
the surface diffuse. The initial atom-extraction process also
transferred a large number of gold atoms to the tip, and
subsequently these atoms were redeposited to the substrate
thus contributing to the growth of the fingers.
The construction of ZnTBPP functionalized gold fingers
in our case did not require the initial step of extraction of
gold atoms from the surface at high electric field, because
adhesion between the Au atoms within the top layer and
between the top and second layers was already weakened by

Page 10

10
the molecules and a very weak electric field was needed to
trigger the clusters’ mobility.
Figure 1. Molecular structures of a) 4 -aminothiophenol (4-ATP) and b)
Zinc (II) 5,10,15,20-tetrakis(4-tert-butylphenyl)-porphyrin (ZnTBPP); c)
scheme of the 4-ATP/ZnTBPP complex on the gold surface.

Figure 2. Time ev olution of the islands at a) 0min, b) 6 min, c) 12 min, d)
18min. STM images acquired at 0.2 V, 0.8nA.


Figure 3. Evolution processes near the step edges at a) 0min, b) 5 min, c)
10 min. STM images acquired at 0.2V, 0.5nA.

Figure 4. STS curves acquired at a) 0.2V, 0.8nA and b) 1V, 0.8nA. The
blue and black curves were acquired in the areas indicated by the
rectangular boxes. STM images shown in the insets were acquired at a)
0.2V, 0.8nA and b) 1V, 0.8nA. In the second inset of figure 4b there are
the normalized conductance curves corresponding to the curves shown in
figure 4b.













a.
b. c.
a.
b. c.
100nm
100nm
100nm
a.
b.
c.
d
15nm
15nm
15nm
15nm
-0,6 -0,3
Voltage (V)
0,00,30,6
-1,0
-0,5
0,0
0,5
1,0


Current (nA)
a.
50nm
0.2 V
-0,6 -0,3 0,00,30,6
-1,0
-0,5
0,0
0,5
1,0


Current (nA)
Voltage (V)
b.
-0,6-0,3 0,0 0,3 0,6
Voltage (V)
0
1
2
3
4
5



50nm
1 V

Page 11

11
Figure 5. Typical unfiltered STM images of Au(111) functionalized by 4-
ATP and ZnTBPP a) on a terrace acquired at 0.2 V, 0.5nA; b) at the step
edge acquired at 0.2 V, 0.5nA; c) line scan along the line shown in figure
5b; d) schematic diagram of the observed molecular diffusion.

Figure 6. Typical unfiltered STM image of the Au(111) reconstruction.



Figure 7. a) Spectrum of the Au NNV Auger and Zn 2p photoemission
peaks of Au/4-ATP/ZnTBPP, excited using high intensity undulator
synchrotron radiation; b) high resolution S 2p photoemission spectrum of
Au/4-ATP/ZnTBPP.












176174172170
Binding Energy (eV)
168 166164162160
Intensity (arb. units)
S
2p
hν=260
1080107010601050 1040 103010201010
Intensity (arb. units)
Binding Energy (eV)
hν=1150
Au NNV
Zn
2p
Zn
a.
b.

Diffusion direction
d.
c.
50nm
a.
100nm
b.
-20 020 40 60 80 100120 140160
Distance (nm)
0,55
0,60
0,65
0,70
0,75
0,80


Height (nm)
50nm

Page 12

12
Figure 8. a) High resolution N 1s photoemission spectrum of Au/4-
ATP/ZnTBPP; b) Nitrogen K edge NEXAFS spectra of Au/4-ATP/ZnTBPP
(black curves) as a function of the incidence angle (relative to the surface
normal). Also shown are spectra of Au/4-ATP (red curves) at grazing and
normal incidence, from ref. 16b.

Figure 9. a) 4-Au atoms cluster adsorbed on Au(111) and b) the same
system with methanethiolate molecules adsorbed in the bridge site for
each cluster.




Figure 10. Central panel: charge density difference along the direction
normal to the surface, for the cluster/slab system (solid black line) and
with a CH3S molecule adsorbed (dashed red line). Top and bottom panels:
the charge density difference spatial distribution for cluster/slab (NS), and
thiolate/cluster/slab (MNS), respectively, with red clouds denoting regions
of electron accumulation (negative charge) and blue ones denoting
electron depletion (positive charge).

Table 1 . Binding energies in eV for various subsystems (see text for
details)
This work Others
M-S 1.87 1.73 [27], 2.3 eV [33]
M-N 2.76 3.3 eV [33]
M-NS 2.06 -
N-S 4.11 -
MN-S 3.41 -


408406404402400398396394
Intensity (arb. units)
Binding Energy (eV)
hν=500
N
400405410415
B
A


0o
80
o
70
50
30
0
o
o
o
o
Intensity (arb. units)
Photon Energy (eV)

a.
b.
a. b.


charge density difference (e/Å)
z ( Å )
D r N S
D r M N S

Page 13

13
3. Conclusions
In conclusion, randomly oriented porphyrins and thiols
can assist in the movement of assemblies of gold clusters on
the surface under the effect of the electric field induced by
the tip. Density functional theory calculations showed that
organic molecules anchoring through sulphur on Au surfaces
weaken the interaction between the first two gold layers.
At the step edge the assemblies form nanofingers of gold,
covered by ZnTBPP molecules, whose optical properties are
known. In principle, this method can be applied to prepare
devices where the spectroscopic response of chromophores
is increased by gold 1-D structures.

4. Experimental Section
Chemicals and solvents. ZnTBPP was synthesized by following the
standard procedure22 for the metal-free porphyrin, using zinc(II)
acetate to insert Zn inside the ring. 4-ATP was purchased from Aldrich
(figure 1). Analytical grade ethanol and chloroform were used as
solvents. Chloroform was purified by filtration through a column of
basic alumina.
Formation of a SAM of 4-aminothiophenol. The gold (111) substrate,
on mica, was purchased from Molecular Imaging, annealed by ethanol
flame and cooled under a nitrogen stream. The substrates were
immersed into an ethanol solution of 4-ATP (10-3 M) for 2h. The
prepared SAM on gold was rinsed with the solvent and dried with
nitrogen.
Preparation of porphyrin thin films by axial coordination. Ligation
with the amino group of 4-ATP SAM on gold was achieved by
immersing the 4-ATP SAM on gold into a chloroform solution of
ZnTBPP (10-4 M) for ca.72h. The resulting film was rinsed with
chloroform and dried with nitrogen.
STM and STS measurements. The STM measurements were
performed using an Omicron VT-STM. The samples were inserted into
the vacuum chamber via a fast-entry lock, and were therefore not
subjected to baking. STM was carried out in ultra high vacuum (~10-10
mbar) at room temperature, using Pt-Ir tips. Tip quality was tested by
acquiring atomically-resolved images of graphite. STM images were
acquired in constant current mode using tip-biases ranging from ±0.2
to ±1V and set points of 0.8 and 0.5 nA.
During spectroscopy measurements, the feedback loop was switched
off and the set-point current, which sets the tip-sample distance, was
fixed during the bias scan. The spectroscopy measurements ( I-V)
were performed by collecting grids of points spaced by 5nm over scan
areas of 200x200nm. The STS spectra shown in this work are
averaged over a set of several I-V curves in the grid over the scanning
area.
Photoemission and NEXAFS measurements. Photoemission
measurements were taken, at an emission angle of 30º, on the SGM
undulator beamline at the Canadian Light Source, Saskatoon, Canada,
using a Scienta 100mm hemispherical analyser. The combined
resolution of beamline and analyzer was 160 meV for S 2p and N 1s
measurements, measured using the Fermi level of Au, and around 1
eV for the Zn 2p measurements. The photon flux was measured using
a calibrated photodiode. Samples were introduced via a load lock.
NEXAFS spectra, at room temperature, were taken using the BEAR
beamline23 at the ELETTRA synchrotron, Trieste, Italy. Samples were
introduced into the vacuum system via a load lock. The sample
mounting geometry used is reported in ref.16. A cylindrical mirror
analyser was used to acquire the NEXAFS spectra, employing Auger
yield, with a photon energy resolution of 0.3eV.
Computational Details. We performed ab initio DFT calculations
using the plane-waves code PWSCF24. UltraSoft Pseudopotentials25
were chosen, with wave function and charge density cutoff of 40 Ryd
and 350 Ryd, respectively. The PW91 parametrization26 for the
exchange-correlation potential was used. We modelled nanoclusters
by a small cluster of 4 Au atoms, adsorbed on a slab of 4 layers of

Page 14

14
gold, with a surface unit cell of 12 Au atoms (surface area of 90 Å2),
which ensure a minimum lateral distance between clusters of 5.2 Å. In
order to describe the 4-aminothiophenol adsorption on gold we used a
short chain alkanethiol, the methanethiol, adsorbed in the preferential
bridge site27-28. We extracted the initial gold slab from a fully relaxed 7
layers slab, and fixed all the positions, while we allowed the geometric
relaxation of both adsorbed cluster and molecule. The systems were
placed in a large supercell, with vertical distance between different
units of more then 30 Å, to avoid spurious interactions between slabs.
Acknowledgements
We thank the MIUR FIRB project "Molecular Devices" for
funding and the BEAR staff, particularly Bryan Doyle, for
assistance at ELETTRA. Part of the research described in this
paper was performed at the Canadian Light Source, which is
supported by NSERC, NRC, CIHR, and the University of
Saskatchewan. Theoretical calculations have been performed
at the ISUFI-CACT (Lecce). We thank the SPACI consortium
for providing computational facilities. The theoretical work was
supported by the European project SA-NANO (contract
number STRP 013698).
[1] C. A. Hunter, J. K. M. Sanders, J. Am. Chem. Soc. 1990, 112, 5525.
[2] T. Malinski, Z. Taha, Nature 1992, 358 , 676.
[3] C. H. M. Maree, S. J. Roosendaal, T. J. Savenije, R. E. I. Schropp, T. J.
Schaafsma, F. H. P. M. Habraken, J. Appl. Phys. 1996, 80, 3381.
[4] Y. Harima, H. Okazaki, Y. Kunugi, K. Yamashita, H. Ishii, K. Seki, Appl.
Phys. Lett. 1996, 69, 1059.
[5] P. E. Burrows, S. R. Forrest, S. P. Sibley, M. E. Thompson, Appl. Phys.
Lett. 1996, 69, 2959.
[6] Y. Hamada, IEEE Trans. Electron Devices 1997, 44, 1208.
[7] M. A. Baldo, D. F. O’Brien, Y. You, A. Shoustikov, S. Sibley, M. E.
Thompson, S. R. Forrest, Nature 1998, 151, 395.
[8] C.-Y. Liu, H.-I. Pan, M. A. Fox, A. J. Bard, Science 1993, 261, 897.
[9] J. M. Lupton, Appl. Phys. Lett. 2002, 81, 2478.
[10] J. R. Reimers, T. X. Lui, M. J. Crossley, N. S. Hush, Nanotechnology
1996, 7 , 424.
[11] a) H. S. Yoon, S. J. Park, J. E. Lee, C. N. Whang, I.-W. Lyo, Phys.
Rev. Lett. 2004, 92, 096801; b) I. Matsuda, M. Hengsberger, F.
Baumberger, T. Greber, H. W. Yeom, J. Osterwalder, Phys. Rev. B
2003, 68, 195319; c) R. Losio, K. N. Altmann, F. J. Himpsel, Phys.
Rev. Lett. 2002, 85, 808; d) M. Shibata, I. Sumita, M. Nakajima,
Phys. Rev. B 1998, 57, 1626.
[12] A. G. Brolo, C. J. Addison, J. Raman Spectrosc. 2005, 36, 629.
[13] a) F. Yin, R. E. Palmer, Q. Guo. Surf. Sci. 2006, 600, 1504; b) Q. Guo,
F. Yin, R. E. Palmer, Small 2005, 1, 76.
[14] a) J. M. Keel, J. Yin, Q. Guo, R. E. Palmer, J. Chem. Phys. 2002, 116,
7151; b) Q. Jin, J. A. Rodriguez, C. Z. Li, Y. Darici, N. J. Tao, Surf.
Sci. 1999, 425, 101.
[15] F. A. Cotton, G. Wilkinson Advanced Inorganic Chemistry, Fifth
Edition; John Wiley & Sons, Inc.: New York, 1988.
[16] a) V. Arima, E. Fabiano, R. I. R. Blyth, F. Della Sala, F. Matino, J.
Thompson, R. Cingolani, R. Rinaldi, J. Am.Chem.Soc. 2004, 126,
16951; b) V. Arima, F. Matino, J. Thompson, R. Cingolani, R.
Rinaldi, R. I. R. Blyth, Surf. Sci. 2005, 580, 63; c) V. Arima, R. I. R.
Blyth, F. Della Sala, R. Del Sole, F. Matino, G. Mele, G. Vasapollo,
R. Cingolani, R. Rinaldi, Mat. Sci. Eng. C 2004, 24, 567.
[17] Z.-P. Shi, Z. Zhang, A. K. Swan, J. F. Wendelken, Phys. Rev. Lett.
1996, 76, 4927.
[18] W. Fan, X. G. Gong, W. M. Lau, Phys. Rev. B 1999, 60, 10727.
[19] a) O. S. Trushin, P. Salo, T. Ala-Nissila, Phys. Rev. B 2000, 62, 1611;
b) C. M. Chang, C. M. Wie, S.P. Chen, Phys. Rev. Lett. 2000, 85
1044
[20] a) F. Rosei, M. Schunack, P. Jiang, A. Gourdon, E. Laegsgaard, I.
Stensgaard, C. Joachim, F. Besenbacher, Science 2002, 296, 328;
b) J. K. Gimzewski, C. Joachim, R. R. Schlittler, V. Langlais, H.
Tang, I. Johannsen, Science 1998, 281, 531; c) M. Bohringer; W. D.
Schneider, R. Berndt, K. Glockler, M. Sokolowski, E. Umbach, Phys.
Rev. B 1998, 57, 4081.
[21] T. Okajima, Y. Yamamoto, Y. Ouchi, K. Seki, J. Elect. Spectr. Relat.
Phenom. 2001, 114, 849.
[22] A. Adler, F. R. Longo, W. Shergalis, J.Am.Chem.Soc. 1964, 86, 3145.
[23] a) S. Nannarone, F. Borgatti, A. DeLuisa, B. P. Doyle, G. C. Gazzadi,
A. Giglia, P. Finetti, N. Mahne, L. Pasquali, M. Pedio et al., AIP
Conference Proceedings 2004, 705, 450; b) I. Pasquali, M. Pedio,

Page 15

15
G. Selvaggi, S. Nannarone, G. Naletto et al. ELETTRA News 2003,
47 (September).
[24] S. Baroni, A. D. Corso, S. de Gironzoli, P. Giannozzi, available at
http://www.pwscf.org/
[25] D. Vanderbilt Phys. Rev. B 1990, 41, 7892.
[26] J. P. Perdew, Physica B 1991, 172, 1.
[27] Y. Yourdshahyan, A. M. Rappe, J. Chem. Phys. 2002, 117, 825.
[28] D. Fragouli, T.Kitsopoulos, L. Chiodo, F. Della Sala, R. Cingolani, S.
Ray, R. Naaman, to appear in Langmuir.
[29] L. Scudiero, D. E. Barlow, U. Mazur, K. W. Hipps, J. Am. Chem. Soc.
2001, 123, 4073.
[30] W. Deng, K. W. Hipps, J. Phys. Chem. B 2003, 107, 10736.
[31] G. Polzonetti, A. Ferri, M. V. Russo, G. Iucci, S. Licoccia, R.
Paolesse, J. Vac. Sci. Technol. A 1999, 17, 832.
[32] T. Okajima, Y. Yamamoto, Y. Ouchi, K. Seki, J. Elect. Spectr. Relat.
Phenom. 2001, 114, 849.
[33] M. Konôpka, R. Rousseau, I. Štich, D. Marx, J. Am. Chem. Soc. 2004,
126, 12103.
[34] M. C. Vargas, P. Giannozzi, A. Selloni, G. Scoles, J. Chem. Phys.
2001, 105, 9509.
[35] H. Kondoh, C. Kodama, H. Sumida, H. Nozoye, J. Chem. Phys. 1999,
111, 1175.
[36] W. Deng, D. Fujita, T. Ohgi, S. Yokoyama, K. Kamikado, S. Mashiko,
J. Chem. Phys. 2002, 117, 4995.

Received: ((will be filled in by the editorial staff))
Published online on ((will be filled in by the editorial staff))