Stabilizing gold adatoms by thiophenyl derivatives: A possible route toward metal redispersion

Article abstract of DOI:10.1021/ja300304s

Tris(phenylthio)benzene molecules have been synthesized in order to explore their ability to trap single Au adatoms on an Au(111) surface. The resulting metal-organic complexes have been characterized with low-temperature scanning tunneling microscopy and infrared reflection absorption spectroscopy; possible structure models have been derived from density functional calculations. Upon room temperature deposition, the thiophenyl derivatives form dimer structures, comprising two molecules and six Au adatoms. Below 100 K, isolated molecules are found as well that have trapped up to six Au atoms. On the basis of the experimental results and calculated formation energies of the complexes, we discuss potential applications of the thioethers for the redispersion of metals on a catalyst surface. First experiments performed on Au particle ensembles prepared on alumina thin films suggest that the molecular ligands are indeed able to change the distribution of gold on the oxide surface.

Full text of DOI:10.1021/ja300304s

Article
pubs.acs.org/JACS
Stabilizing Gold Adatoms by Thiophenyl Derivatives: A Possible
Route toward Metal Redispersion
†
†
†
,†
†
‡
‡
Bing Yang, Yi Pan, Xiao Lin, Niklas Nilius,* Hans-Joachim Freund, Catherine Hulot, Anne Giraud,
,‡
§
,§
Siegfried Blechert,* Sergio Tosoni, and Joachim Sauer*
†
Fritz-Haber-Institut der MPG, Faradayweg 4-6, D-14195 Berlin, Germany
‡
Institut fu
̈
r Chemie, Technische Universitat
Institut fur Chemie, Humboldt Universitat
S Supporting Information
̈
Berlin, Straße des 17. Juni 135, D-10623 Berlin, Germany
§
̈
̈
zu Berlin, Brook-Taylor-Straße 2, D-12489 Berlin, Germany
*
ABSTRACT: Tris(phenylthio)benzene molecules have been synthe-
sized in order to explore their ability to trap single Au adatoms on an
Au(111) surface. The resulting metal−organic complexes have been
characterized with low-temperature scanning tunneling microscopy and
infrared reflection absorption spectroscopy; possible structure models
have been derived from density functional calculations. Upon room
temperature deposition, the thiophenyl derivatives form dimer
structures, comprising two molecules and six Au adatoms. Below 100
K, isolated molecules are found as well that have trapped up to six Au atoms. On the basis of the experimental results and
calculated formation energies of the complexes, we discuss potential applications of the thioethers for the redispersion of metals
on a catalyst surface. First experiments performed on Au particle ensembles prepared on alumina thin films suggest that the
molecular ligands are indeed able to change the distribution of gold on the oxide surface.
1
. INTRODUCTION
reactivation of the catalyst has been reported after such
treatments, no atomic understanding of the associated
processes could be achieved so far. In this study, we have
explored an alternative approach to alter ripening processes at
oxide surfaces. The concept relies on the stabilization of mobile
surface species by means of strongly interacting ligand
molecules. Those ligands need to be removed in a chemical
or photochemical step later, which triggers the release of the
trapped atoms and their recombination into small aggregates.
Regular applications of the ligands might therefore maintain the
high metal dispersion on the catalyst surface.
The performance of supported metal catalysts is directly
connected to the dispersion of the active metal species on the
oxide surface. The catalytic activity is found to scale with the
inverse particle diameter, rendering smaller particles more
reactive than their bulk-like counterparts.
large fraction of under-coordinated, hence chemically active
surface atoms in ultrasmall particles as well as their defect-rich
or even amorphous structure. Also, the electronic properties of
small aggregates are known to deviate from the respective bulk
behavior, and may be governed by charge transfer processes
from the oxide support in addition.
1
2
−5
One reason is the
6
7,8
Given the tremendous interest in nanosized gold as
2
,3,5,16
catalytically active material,
we have developed and tested
The particle-size distribution on supported metal catalysts is
however not fixed during operation but subject to constant
changes. The most relevant process is a diffusion-mediated
different thiophenyl derivatives, which are known to form
strong Au−S bonds. We note, however, that our design concept
can be extended to other catalytically relevant metals, such as
Pt. We have investigated, from both experimental and
theoretical points of view, the ability of the molecules to trap
single Au atoms on an Au(111) surface. The binding
configuration of the Au−ligand complexes was determined
via low-temperature scanning tunneling microscopy (STM),
while structure models were derived from density functional
(Ostwald) ripening, where small aggregates continuously lose
9
,10
atoms to larger ones.
the high surface-free-energy of the smallest metal deposits.
The ripening process gets enhanced by the elevated temper-
ature and the interaction with gas-phase molecules during
reaction conditions. As a result, the particle-size distribution
shifts toward larger values, a process that gives rise to a gradual
deactivation of the catalyst. To keep the catalyst functional, the
initial high dispersion of the metal needs therefore to be re-
established by reversing the ripening process at a certain point.
Several proposals have been made in the literature, aiming at
Driving force for this material flow is
11
12
17
theory (DFT) calculations. On the basis of computed Au−
ligand binding energies, we discuss the ability of the ligands to
stabilize metal adatoms on the surface and to break apart small
gold clusters. We note that metal coordination by suitable
1
the redispersion of metal participles on oxide supports. Most
of them are based on the addition of halides, in particular of
Received: January 11, 2012
Published: March 13, 2012
13−15
chlorine, to the catalyst feed-gas.
Although successful
©
2012 American Chemical Society
11161
dx.doi.org/10.1021/ja300304s | J. Am. Chem. Soc. 2012, 134, 11161−11167

Journal of the American Chemical Society
Article
molecular species has already been explored in great detail in
order to produce metal−organic networks with potential
applications in molecular electronics and surface pattern-
1
8−21
ing.
In our approach, the ligand-mediated Au complexation
involves the lattice-gas atoms that are intrinsically present on
22
Au(111) at room temperature. In order to mimic the real
redispersion problem in catalysis, the single crystalline gold
surface has been replaced with an ensemble of oxide-supported
Au particles. Also in this case, the ligands are expected to
stabilize single Au atoms, being exchanged among the deposits
at elevated temperature. Indeed, new, ultrasmall Au aggregates
were observed upon treating the particle ensemble with the
thiophenyl derivatives at elevated pressure and temperature.
We take this observation as first indication for the practicability
of our redisperison approach.
Figure 1. Synthesis of 1,3,5-tris(phenylthio)benzene via mixing
tribromobenzene with thiophenol at 500 K for 16 h in the presence
31
of Cu O. The final product concentration was as high as 79%.
2
outer phenyls also impede strong intermolecular interactions,
preferably without blocking the access of Au atoms to the
molecular body. The ligands were synthesized using the
Reifschneider method and purified with multiple chromato-
31
graphic columns including preparative HPLC. A second
cleaning step was performed by outgassing the molecules at 350
K in vacuum.
2. EXPERIMENTS AND CALCULATIONS
The structural integrity of the ligands has been probed with
IRAS performed in transmission and reflection geometry for
TPB on glass and Au(111), respectively (Figure 2). The spectra
The STM experiments have been carried out with a custom-built,
23
ultra-high-vacuum instrument operated at 5 K. The molecules were
vapor-deposited from a Knudsen cell onto the sputtered/annealed
Au(111) surface. To modify the density of mobile Au species on the
surface, the sample temperature was varied between 100 and 300 K
during exposure. Moreover, extra Au atoms were deposited from a hot
tungsten filament onto the support. The development of Au−ligand
adsorption complexes was deduced from constant-current STM
images. The binding behavior of the molecules was further analyzed
with infrared reflection absorption spectroscopy (IRAS) performed
with an IFS 66 spectrometer from Bruker.
The DFT calculations were performed with the VASP code, using
24
PBE as density functional. The basis set has been defined with the
2
5
projector-augmented-wave method using an energy cutoff of 300 eV.
Dispersion interactions were taken into account by a semiempirical C6
term parametrized by Grimme and implemented for periodic
26
boundary conditions (e.g., VASP) by Kerber. The parameters for
27
Figure 2. Infrared reflection−absorption spectra taken on (a) TPB on
glass and (b) 0.5 ML TPB on Au(111). The bar graph in (a) visualizes
the normal modes of the molecule, as determined with a Hessian
harmonic frequency calculation at the DFT level.
Au were adopted from a work of Tonigold and Gross. We refer to
this combined DFT/semiempirical approach as PBE+D (dispersion)
and provide details in ref 17.
2
8
STM images were simulated with the Tersoff−Hamann model,
treating the tip as single s-orbital and assuming equal tunneling
probability into all states inside a preselected energy window. This
range was set from −0.3 eV to the Fermi level in agreement with the
experimental conditions. Larger intervals did however not change the
character of the images. The constant current mode was simulated by
displaying a contrast map along a fixed electron-density isosurface.
exhibit an identical peak structure, suggesting the identity of the
molecules on both substrates. Dissociation of the TPB on the
Au(111) is therefore ruled out. The intensity variations of the
IR bands are governed by the IRAS selection rules, which
render only bands with large out-of-plane dipole components
detectable on the gold surface. Moreover, the molecule has a
defined binding geometry only at sub-monolayer coverage on
Au(111), whereas thick TPB layers on glass are amorphous.
The nature of the different IR bands has been elucidated with a
normal-mode analysis performed on the DFT level (Figure 2a).
Measured and calculated spectra are in reasonable agreement,
which supports our conclusion that the TPB remains intact
upon adsorption on the Au(111). Three absorption regions can
be identified, as detailed in Table 1 of the Supporting
Information. IR bands between 1500 and 1580 cm− are
3. RESULTS AND DISCUSSION
3.1. Synthesis of the Ligands and IR Fingerprint. The
ligands used in this work are based on sulfur-containing aryl
groups that can be tailor-made by changing the distance
between the S centers and the electron densities at the
molecular body. The interaction with gold is primarily
governed by the formation of strong, covalent Au−S bonds,
but further enhanced by the intermixing with the π-states of the
2
9,30
aryls.
Main requirements for the synthesis were a high
1
purity, a sufficient thermal stability and an adequate volatility of
the ligands to enable direct sublimation into the gas phase. The
first substance tested in this regard was 1,3,5-triazine that
turned out to be unstable against desorption of the thiophenyl
units. We then focused on different benzene derivatives, among
which the 1,3,5-tris(phenylthio)benzene (TPB) was the most
promising candidate. The molecule contains a central 1,3,5-
tristhiobenzene unit and three outer phenyl rings attached via S
atoms (Figure 1). It gains structural flexibility thanks to the
peripheral rings that can easily rotate about their S−C axis. The
32,33
assigned to the C−C stretching modes of the phenyl rings.
The in-plane C−H bending motions of the phenyls give rise to
−1
absorption bands between 800 and 1450 cm , while the out-
of-plane C−H bending occurs in a frequency range from 680 to
−1
−1
7
50 cm . Additional IR intensity at ∼3050 cm reflects the
C−H stretching motions of the rings (not shown). On the
metal substrate, the bands are only detectable if the respective
dipole moment has a substantial out-of-plane component. A
detailed assignment of the IR bands to the different subunits of
1
1162
dx.doi.org/10.1021/ja300304s | J. Am. Chem. Soc. 2012, 134, 11161−11167

Journal of the American Chemical Society
Article
the TPB and a correlation to the molecular binding geometry
on the Au(111) can be found in the Supporting Information.
have a slightly smaller spacing of 16 Å (Figure 4c). The
connecting axes between bright and dim maxima always enclose
an angle of 80−90°. Due to the six-fold symmetry of the
substrate, both isomers occur with three orientations rotated by
3.2. STM Fingerprints of TPB−Au Complexes on
Au(111). Figure 3 shows three STM images of Au(111)
6
0° against each other, as shown for the short configuration in
Figure 4b.
With increasing TPB exposure, irregular aggregates develop
on the surface, being composed of the different building blocks
discussed before (Figure 4d). A reliable structural analysis on
the basis of the STM images is difficult in this case, because of
the large variety of different configurations. Finally, flat
molecular islands, characterized by a regular sequence of
̅
darker and brighter lines running along an Au ⟨110⟩ direction,
Figure 3. STM topographic images of Au(111) exposed to increasing
amounts of TPB ligands at room temperature (450 × 450 Å , 0.3 V).
develop at high and rapid TPB exposure (Figure 4e). The
islands are composed of rhomboidal cells with 5.8 and 10.4 Å
edge length and an enclosing angle of 75°. Its crystallographic
⎛2 0⎞
2
exposed to an increasing amount of TPB at room temperature.
Individual ligand complexes are easily identified as protrusions
that preferentially bind to the elbows of the Au(111)
herringbone reconstruction and to the up-side of substrate
step edges. With increasing coverage, the adsorbates also appear
in the fcc domains of the herringbone reconstruction. In all
cases, the individual entities remain well separated from each
other, indicating a mainly repulsive interaction of TPB. Only at
high and rapid exposure, the formation of irregularly shaped
molecular islands is revealed on the Au(111) surface (Figure
relationship with the Au(111) lattice is described by a ⎜
⎟
⎝
1 4⎠
matrix. The cell would be large enough to accommodate two
TPB molecules in a slightly skewed configuration, in which two
S atoms of each TPB give rise to the tiny depressions in the
STM micrographs, while the third one is invisible due to a
slight upward tilt of the central benzene ring (see sketch in
Figure 4e). We note that this structure model is only tentative
and cannot be confirmed in the context of this paper. All
complexes described so far form spontaneously on the Au(111)
upon TPB deposition at 300 K.
3
c).
Better-resolved STM images taken at low TPB exposure
A second set of ad-structures can be produced by co-
depositing small numbers of Au atoms onto a TPB-covered
surface at 5 K, followed by gentle annealing to 100 K to
promote diffusion. During heating, the adatoms disappear from
the surface, as they attach either to the Au step edges or to the
TPB ligands. The occurrence of new ad-features in the STM
images suggests the second mechanism to be active. Some
examples of those low-temperature complexes are displayed in
Figure 5. They have a maximum diameter of 10−12 Å and are
provide insight into the internal structure of the ligand
complexes (Figure 4). The smallest unit that occurs only at
Figure 4. (a,d) STM images and superimposed structure models of
different metal−organic complexes formed by TPB and Au adatoms at
2
room temperature (30 × 30 Å , 0.5 V). The upright phenyl rings have
been omitted in the models for the sake of clarity. (b) Short and (c)
long dimer complex in different rotational configurations. (e)
Molecular island formed upon rapid TPB dosing at 300 K (40 × 40
2
Å , 0.5 V). At this condition, the ligands are unable to trap enough Au
atoms and aggregate into Au-free assemblies.
very low TPB coverage comprises a bright and a faint
protrusion with 2.4 and 1.6 Å apparent height, respectively,
separated by 7.5 Å (Figure 4a). The next bigger and by far more
abundant complex is composed of two bright (2.6 Å height)
and two dim protrusions (1.6 Å) in a cross-like configuration
Figure 5. STM images and superimposed structure models for
different metal−organic complexes, prepared by dosing Au atoms onto
2
a TPB-covered gold surface below 100 K (30 × 30 Å , 0.5 V). The
maxima in the images reflect the positions on the trapped Au atoms.
typically 1.5 Å high. This size is compatible with a single TPB
molecule, which is why we refer to them as monomers in the
following. The different ad-features can be classified by the
number of lobes that surround the center protrusion. While
structures with two and three lobes are displayed in image a,
their number increases to four, five and six in images b−d.
Apparently, the Au atoms dosed onto the surface are able to
bind to the ligands in various configurations, all of them being
stable only below room temperature.
(Figure 4b,c). It seems to consist of two of the elementary
building blocks described before and will therefore be referred
to as a dimer in the following. Two isomers can be identified
for the dimers that mainly differ in the spacing between the
bright and dim maxima. In a short configuration, the bright
protrusions are separated by 6−7 Å along an Au⟨1
direction, while the dim ones are spaced by almost 18 Å
Figure 4b). In a long configuration, the distance between the
bright maxima increases to 8−9 Å, while the dim protrusions
̅
10⟩
(
1
1163
dx.doi.org/10.1021/ja300304s | J. Am. Chem. Soc. 2012, 134, 11161−11167

Journal of the American Chemical Society
Article
A first model for the binding of the TPB molecules to the
Au(111) may be obtained directly from the experiment. Not
even the smallest, room-temperature units as shown in Figure
gold surface and has a number of Au atoms attached. When
optimizing this starting geometry, the central benzene ring
remains flat with an interface distance of about 2 Å, while the
peripheral rings turn upright (Figure 7). The central ring is
4
a can be assigned to a single molecule, as they do not display
the three-fold symmetry of TPB. This becomes even clearer for
the dimer complexes in Figure 4b−d, which clearly exhibit a
two-fold symmetry in the STM. Moreover, the co-adsorption
experiments indicate that single Au atoms can be attached to
the ligands, a process that was in fact envisioned when
designing the molecules. The process of metal coordination
would be possible even without dosing extra adatoms, as a
dense lattice gas originating from the steady detachment of Au
atoms from the surface steps is inherently present on the
Au(111). The STM data therefore suggest the formation of
metal−organic complexes by attaching mobile Au atoms to the
TPB ligands. Even the number of adatoms per complex can be
determined experimentally. With a 5 V bias pulse, a TPB dimer
can be compartmentalized into six elementary constituents, all
of them having the size of a single Au atom (Figure 6a,b). This
Figure 7. (a) Optimized structure from PBE+D calculations of a single
TPB molecule on Au(111) after attaching two, three, five, and six Au
adatoms (top and side view). (b) Simulated STM images with the ball
models overlaid. The upright phenyl rings have been replaced with
hydrogen atoms in this case. The images were obtained at an
3
isodensity level of 0.001 |e|/Å . The resulting configurations
correspond to the low-temperature monomer complexes observed in
the experiment.
stabilized by three S−Au bonds plus the dispersive forces
surf
between its π-electronic system and the highly polarizable Au
35
surface. The computed binding energy of 182 kJ/mol is
remarkably high. A dissociation of the TPB complex was found
to be unfavorable on the Au(111), as discussed in the
Supporting Information.
On the basis of the starting geometry, we have analyzed the
interaction of the TPB with individual Au adatoms. The first
species attached to the ligand hardly affects its binding
configuration. The atom occupies a three-fold hollow site on
the Au(111) that is adjacent to an S center in the TPB. The
Figure 6. (a,b) TPB dimer structures on Au(111) before and after
2
applying a +5 V pulse to the right entity (140 × 70 Å , −0.5 V). The
species appearing after manipulation are the Au atoms that had been
incorporated in the complex before. (c) TPB molecules deposited
2
onto a bilayer FeO film grown on Pt(111) (60 × 60 Å , −0.8 V). As
resulting S−Au distance amounts to 3 Å (Figure 7), which is
no Au-mediated adsorption scheme is available in this case, the three-
fold symmetry of the TPB becomes visible.
ad
considerably longer than the 2.5 Å bond length in a gas-phase
TPB−Au complex. The larger binding distance reflects the
competition between the S−Auad and Auad−surface interaction
that holds the adatom in place. Two other Au atoms can be
coordinated in a similar way by the remaining S atoms of the
TPB. However, the ligand is able to bind even three extra
adatoms, which occupy Au(111) hollow sites below the upright
standing phenyl rings. In this configuration, the bond length to
the next S center increases to 4.0 Å and stabilization mainly
arises from the interaction with a peripheral phenyl ring that
slightly bends toward the adatom. A single flat-lying TPB ligand
is therefore able to coordinate up to six Au adatoms. This
surprisingly high capacity of thioethers to coordinate Au atoms
is in agreement with earlier studies on short alkanethiolate
manipulation experiment therefore indicates the number of
incorporated Au atoms in the dimer to be six. Further evidence
for the TPB−Au complexation reaction on Au(111) comes
from a control experiment, where the ligands have been dosed
34
onto an FeO thin film grown on Pt(111) (Figure 6c).
Evidently, this surface contains neither Au nor other metal
adatoms that might interact with the ligand molecules. Indeed,
the expected three-fold symmetry of isolated TPB species
becomes immediately visible in this case and cross-shaped or
multilobed features, being typical for Au(111), are absent.
Moreover, the apparent height of the adsorbates has decreased
from 2.6 Å on the Au(111) to 0.5 Å on the FeO/Pt surface.
This height reduction is incompatible with the presence of
metal atoms inside the ad-features and suggests that only the
molecular electronic states governed by the large HOMO−
LUMO gap are responsible for the image contrast on the oxide.
The different appearance of TPB molecules on oxide and metal
surfaces therefore supports our idea that metal−organic
complexes develop on the Au(111) via TPB−Au complexation.
22,36−38
chains.
To connect the above-described binding geometries to ad-
features observed in the experiment, we have simulated STM
28
images with the Tersoff−Hamann approach. The results did
not reproduce the experimental data because of the over-
whelming contrast induced by the upright standing phenyl
rings (Figure 8a). Due to their large topographic height they
imprint a pronounced three-fold symmetry onto the images
that is independent of the actual number of adatoms attached
to the ligands. This effect is an artifact of the Tersoff−Hamann
model, which considers only the availability of electronic states
in a certain energy window but not their suitability for electron
3
.3. Theoretical Characterization of TPB−Au Com-
plexes on Au(111). To confirm the binding model derived
from the STM, we have performed an extensive computational
1
7
search for thermodynamically stable TPB−Au complexes.
A
natural starting point is a TPB molecule that lies flat on the
1
1164
dx.doi.org/10.1021/ja300304s | J. Am. Chem. Soc. 2012, 134, 11161−11167

Journal of the American Chemical Society
Article
Figure 8. (a) Simulated STM images of a complete TPB molecule
with four Au atoms attached. The simulation is artificially governed by
the three upright-standing phenyl rings that are not observed in STM.
Only at extremely short tip−sample distances, those units appear as
fuzzy regions in the image, here shown for a dimer complex in the
2
topographic (b) and the conductance channel (c) (30 × 30 Å , 25
mV). To remove such artifacts in the simulations, we have replaced the
upright phenyls with hydrogen atoms.
transport. In contrast, STM is a transport experiment that
measures the flux of electrons between tip and sample.
Although the upright standing phenyl rings possess orbitals
that might act as intermediate states for tunneling, they are
unsuitable for electron transport due to a negligible overlap
with the substrate electronic states. Consequently, the upright
phenyl rings do not show up in experimental images taken at
usual set-point parameters. Only at extremely short tip−sample
distances, they leave observable traces in the STM images, as
frequent collisions between tip and protruding molecular units
Figure 9. (a) Optimized structure of a TPB dimer complex in the
short and long configuration on Au(111). (b) Simulated STM images
with the respective ball models overlaid. The upright phenyl rings have
been replaced with hydrogen. The images were obtained at an iso-
density level of 0.0001 |e|/Å . The resulting configurations correspond
to ad-features observed upon room-temperature deposition of TPB
3
(Figure 4).
is able to trap two additional adatoms with its terminal S
centers. Interestingly, dimerization always requires two Au-
atom pairs in order to mediate the interaction between the two
molecules. This can be explained with the steric repulsion
exerted by the upright phenyl rings that inhibits the
interconnection of two ligands via a single Au adatom. In
contrast, the S-Au -Au -S units are flexible enough to mediate
39
induce heavy fluctuations on the tunnel current (Figure 8b,c).
Experimental and simulated STM data can however be
reconciled if the upright phenyl rings are replaced by H
atoms in the calculations (see Supporting Information for
details). The resulting simulations perfectly reproduce the
experimental data that have been obtained at low-temperature
Au deposition onto a TPB-covered surface (compare Figures 5
and 7b). Apparently, a binding geometry in which the flat-lying
central benzene is surrounded by up to six Au atoms is only
realized at preparation temperatures below 100 K. The room-
temperature complexes with their distinct dimer structure, on
the other hand, seem to follow another building principle.
The main difference between low- and high-temperature
TPB complexes is their apparent height in the STM, which
amounts to 1.5 and 2.6 Å in the two cases. This suggests a
larger interface distance between the room-temperature TPB
and the gold surface. Moreover the spatial extension of the
dimers (25 Å) is not compatible with a single TPB molecule
anymore, which is only 10 Å in size. However, aggregation of
two TPB monomers plus a few Au atoms could explain the
observed features. Based on these considerations, we have
developed thermodynamically stable dimer complexes at the
PBE+D level. Two structures were identified, both of them
being composed of two ligands connected via several bridging
Au atoms (Figure 9). In the first conformer, the intermolecular
interaction is mediated by four atoms occupying neighboring
fcc and hcp binding sites in the Au(111) surface. The length of
the bridging Au chain amounts to 7 Å, which is in good
agreement with the short axis of the smaller dimer shown in
Figure 4b. The long axis is defined by two additional adatoms
that are attached to the outer S centers of the dimer and 18 Å
apart. The second conformer deviates from the first one by the
length of the central atom chain, which now comprises four Au
atoms in fcc sites and is 9 Å long. This configuration can be
reconciled with the long dimer structure in the experiment
ad
ad
the bonding between two TPB molecules in a slightly staggered
configuration.
The dimer models are in good agreement with the room-
temperature complexes observed in STM. Their characteristic
cross-like shape reflects the spatial arrangement of the Au
atoms inside the metal−organic complexes. Whereas the Au
pairs bridging the two TPBs generate the two bright
protrusions in the center, the single atoms at the terminal
sulfurs appear darker (Figure 4). We note that single Au
adatoms and dimers on bare Au(111) display the same height
difference, suggesting that the STM contrast is dominated by
the gold and not by the molecular electronic states, which
indeed feature a HOMO−LUMO gap of around 4 eV. Also the
manipulation experiments confirm our structure model (Figure
6
). Applying a controlled bias pulse breaks the compound into
six identical protrusions, reflecting the six adatoms that have
been trapped in the dimer. We suspect that the ligands
themselves are removed from the surface during manipulation.
The results presented so far unambiguously demonstrate the
ability of the TPB molecules to trap Au adatoms, which either
originate from the Au(111) lattice gas or have been co-
deposited onto the surface. This conclusion is in perfect
agreement with the widely accepted picture that thiolates bind
2
9,30,36
to the Au(111) by means of single adatoms.
The ligands
therefore fulfill a first requirement to serve as potential
redispersion agent on a catalyst surface.
3.4. Energetics of Au Complexation by the TPB
Ligands. To get a better idea on the interaction strength
between the ligand molecules and the Au atoms, typical Au
complexation processes have been simulated on the basis of the
17
(Figure 4c). Similar to the short TPB dimer, also the long one
most stable monomer and dimer structures. DFT calculations
1
1165
dx.doi.org/10.1021/ja300304s | J. Am. Chem. Soc. 2012, 134, 11161−11167

Journal of the American Chemical Society
Article
revealed that a single TPB monomer is able to coordinate up to
six Au atoms on the Au(111) surface, as the following reaction
is exothermic for N ≤ 6:
TPB/Au(111) + N[Au /Au(111)]
1
→
TPB−Au /Au(111) + NAu(111)
N
The reaction energy varies between −90 and −18 kJ/mol for N
=
1−6 Au atoms attached to the ligand. The decline reflects the
more and more unfavorable binding configuration of the
adatoms at higher load. Also the energy for dissolving an Aun
cluster by a TPB monomer is negative for N ≤ 4:
Figure 10. STM topographic images of Au particle ensemble grown
on a thin alumina/NiAl(110) film (70 × 70 Å , −1.5 V). Whereas
image a shows the pristine surface, image b was obtained after dosing
2
TPB/Au(111) + Au /Au(111) → TPB−Au /Au(111)
N
N
This implies that one ligand may disrupt an Au dimer, trimer,
or tetramer cluster. Finally, extracting a single atom from an
1
00 L of TPB at 400 K. Note the large number of small ad-features
between the Au deposits, assigned to TPB−Au complexes.
Au cluster via attachment to a TPB molecule has found to be
N
favorable, as suggested by the exothermic character of the
following reaction:
annealing it to 1050 K. The resulting film is crystalline and
atomically flat and consists of two Al−O bilayers. Details of its
atomic structure and electronic nature can be found in the
TPB/Au(111) + Au /Au(111)
40,41
N
literature.
The Au ad-particles were produced by dosing
∼
1.5 ML Au from a Mo crucible onto the surface at room
temperature. The particles have been stabilized and homogen-
ized with respect to their geometry by annealing the sample to
→
TPB−Au /Au(111) + Au
/Au(111)
1
N−1
For N > 2, i.e., beyond the Au dimer, the associated energy is of
the order of −50 kJ/mol.
Also the attachment of six adatoms to a TPB dimer has
found to be exothermic, yielding an excess energy of −33 and
39 kJ/mol for the long and short conformer, respectively.
Values for lower loads (N = 1−5) cannot be determined as the
dimer configuration is stable only in the presence of all six
atoms. Moreover, a TPB dimer is able to break apart an Au6
cluster, as the corresponding energy is negative for the short
17
6
00 K for 10 min. The final ensemble properties are given by a
12
−1
particle density of ∼3 × 10 cm and a mean particle diameter
of 40 ± 10 Å.
To stimulate redispersion of the gold, we have exposed the
sample to a flux of TPB molecules at 400 K for ∼10 min (local
TPB pressure 1 × 10 mbar). Although the initial particle
−
−7
geometry did not change considerably, numerous new
aggregates were detected in the STM images after exposure
(
−17 kJ/mol) and long conformer (−50 kJ/mol). In general,
(
Figure 10b). The new species are 15 Å in size and assigned to
the average attachment energy per ligand is higher for the flat-
lying monomer compared to the up-lifted dimer structure.
However, this advantage is lost as two monomers sponta-
neously transform into a dimer at ∼100 K, a process that is
described by the following reaction:
TPB molecules that have trapped mobile Au atoms diffusing
between the larger deposits. We are unable to provide insight
into the internal structure of the complexes at this point, due to
difficulties to scan the insulating oxide at low bias. However, in
control experiments performed without TPB exposure, hardly
any new ad-features were detected, and only the density of the
original Au deposits decreased slightly due to Ostwald ripening.
On the other hand, the new ad-features were found to
completely fill up the oxide surface upon exposure to a 10 000
times higher TPB flux, indicating a substantial material
transport from the Au deposits to the bare oxide. An even
more drastic effect on the particle size distribution is expected
when dosing the molecules at catalytically relevant pressures
and temperatures; however the realization of such experiments
is beyond the scope of this work.
2
[TPB−Au /Au(111)]
3
→
TPB −Au /Au(111) + Au(111)
2 6
In agreement with experiment, dimerization is exothermic for
two monomers being loaded with three adatoms and yields an
energy gain of −11 and −44 kJ/mol for the short and long
conformer, respectively. The reaction becomes even more
exothermic for TPB monomers loaded with six Au atoms, as
the excess atoms freed upon dimerization can reaggregate into
an Au cluster.
6
All exchange reactions discussed here indicate that disruption
of small Au clusters and binding of the released atoms to a TPB
ligand is energetically favorable at certain conditions. However,
our considerations provide information only on the energy
balance of such virtual reactions, but not on possible reaction
pathways or kinetic barriers. Nonetheless, we suspect that metal
redisperison via TPB molecules might be feasible.
4
. CONCLUSIONS
High purity thiophenyl derivatives have been successfully
synthesized and investigated with respect to their adsorption
behavior on Au(111) with STM and DFT. Our study revealed
an effective stabilization of mobile atoms from the Au(111)
lattice gas by incorporating them into various metal−organic
complexes. The most stable isomer was found to comprise two
3.5. TPB Interactions with Au Particle Ensembles
Grown on Thin Alumina Films. In order to explore the
impact of the TPB molecules on a system that is closer to a real
catalyst than Au(111), we have prepared an Au particle
ensemble on an oxide thin film and exposed it to the ligand
molecules (Figure 10a). For this purpose, an ultrathin alumina
film was grown by oxidizing a NiAl(110) single crystal and
TPB molecules tied together by an Au chain, which gives rise
4
to characteristic cross-like features in the STM. At low
temperature, even single ligand molecules were found to trap
up to six metal adatoms. The stabilization of mobile Au species
via complexation with the TPB was observed even on an oxide
1
1166
dx.doi.org/10.1021/ja300304s | J. Am. Chem. Soc. 2012, 134, 11161−11167

Journal of the American Chemical Society
Article
surface covered with Au deposits, where new ultrasmall
aggregates have been detected after ligand exposure.
(18) (a) Barth, J. V.; Costantini, G.; Kern, K. Nature 2005, 437, 671.
b) Dmitriev, A.; Spillmann, H.; Lin, N.; Barth, J. V.; Kern, K. Angew.
Chem., Int. Ed. 2003, 42, 2670.
(
Our results provide evidence that TPB molecules are able to
influence diffusion processes on oxide surfaces and therewith a
decisive requirement for Ostwald ripening to occur. Moreover,
the formation of stable metal−organic complexes may be
considered as a first step to trigger the formation of new
aggregates, causing the particle-size distribution to shift to
lower values. In future experiments, we need to explore how the
ligand molecules can be removed from the surface again in
order to avoid poisoning effect by carbon residues. Possible
methods include photochemical bleaching or ozone etching.
Only after complete ligand removal, the original chemical state
of the catalyst is recovered and the redispersion cycle is closed.
(
(
19) James, S. L. Chem. Soc. Rev. 2003, 32, 276−288.
20) Stepanow, S.; Lin, N.; Payer, D.; Schlickum, U.; Klappenberger,
F.; Zoppellaro, G.; Ruben, M.; Brune, H.; Barth, J. V.; Kern, K. Angew.
Chem., Int. Ed. 2007, 46, 710−713.
(
21) Zacher, D.; Shekhah, O.; Wo
̈
ll, C.; Fischer, R. A. Chem. Soc. Rev.
2
009, 38, 1418−1429.
(22) Poirier, G. E.; Pylant, E. D. Science 1996, 272, 1145−1148.
(23) Rust, H.-P.; Buisset, J.; Schweizer, E. K.; Cramer, L. Rev. Sci.
Instrum. 1997, 68, 129.
(24) (a) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K.A..;
Pederson, M. R.; Singh, D. J.; Fiolhais, C. Phys. Rev. B 1992, 46, 6671.
(
(
b) Kresse, G.; Furthmu
̈
ller, J. Comput. Mater. Sci. 1996, 6, 15.
(
25) (a) Blochl, P. E. Phys. Rev. B 1994, 50, 17953−17959.
̈
b) Kresse, G.; Joubert, J. Phys. Rev. B 1999, 59, 1758−1775.
ASSOCIATED CONTENT
Supporting Information
■
(26) (a) Grimme, S. J. Chem. Phys. 2006, 124, 034108. (b) Kerber,
*
S
T.; Sierka, M.; Sauer, J. J. Comput. Chem. 2008, 29, 2088.
(27) Tonigold, K.; Gross, A. J. Chem. Phys. 2010, 132, 224701.
(28) Tersoff, J.; Hamann, D. R. Phys.Rev. B 1985, 31, 805.
Infrared absorption−reflection data of TPB on Au(111),
dissociation energies for the optimized molecular structures,
and details of the STM imaging process. This material is
available free of charge via the Internet at http://pubs.acs.org.
(
29) Molina, L. M.; Hammer, B. Chem. Phys. Lett. 2002, 360, 264−
71.
30) Pyykko
31) (a) Reifschneider, W. US Patent, 3102916, 19630903, 1963.
b) Adams, R.; Reifschneider, W.; Ferreti, A. Org. Synth. 1973, 5, 107.
32) Wan, L. J.; Terashima, M.; Noda, H.; Osawa, M. J. Phys. Chem. B
000, 104, 3563−3569.
(33) Syomin, D.; Kim, J.; Koel, B. E.; Ellison, G. B. J. Chem. Phys. B
001, 105, 8387−8394.
34) Rienks, E. D. L.; Nilius, N.; Rust, H.-P.; Freund, H.-J. Phys. Rev.
B 2005, 71, 241404.
35) (a) Baber, A. E.; Tierney, H. L.; Sykes, E. C. H. ACS Nano 2008,
2
(
(
̈
, P. Chem. Soc. Rev. 2008, 37, 1967−1997.
AUTHOR INFORMATION
Corresponding Author
nilius@fhi-berlin.mpg.de; blechert@chem.tu-berlin.de; js@
chemie.hu-berlin.de
■
(
(
2
2
(
Notes
The authors declare no competing financial interest.
(
ACKNOWLEDGMENTS
This work has been supported by the DFG through the
Cluster of Excellence UNICAT”. We acknowledge support
■
“
2, 2385−2391. (b) Tierney, H. L.; Baber, A. E.; Sykes, E. C. H.;
Akimov, A; Kolomeisky, A. B. J. Phys. Chem. C 2009, 113, 10913−
10920.
from the Norddeutscher Verbund fu
tungsrechnen.
̈
r Hoch- und Ho
̈
chstleis-
(36) (a) Maksymovych, P.; Sorescu, D. C.; Yates, J. T. Phys. Rev. Lett.
2
2
006, 97, 146103. (b) Maksymovych, P.; Yates, J. T. J. Am. Chem. Soc.
008, 130, 7518.
(
37) (a) Gro
̈ ̈
nbeck, H.; Hakkinen, H.; Whetten, R. L. J. Phys. Chem. C
REFERENCES
■
2
008, 112, 15940. (b) Mazzarello, R.; Cossaro, A.; Verdini, A.;
(
1) Ertl, G.; Kno
heterogeneous catalysis, 2nd ed.; Wiley-VCH: Weinheim, 2008.
2) (a) Haruta, M.; Yamada, N.; Kobayashi, T.; Iijima, S. J. Catal.
989, 115, 301. (b) Haruta, M. Cat. Tech 2002, 6, 102.
̈
zinger, H.; Schueth, F.; Weitkamp, J. Handbook of
Rousseau, R.; Casalis, L.; Danisman, M. F.; Floreano, L.; Scandolo, S.;
Morgante, A.; Scoles, G. Phys. Rev. Lett. 2007, 98, 016102.
(
1
(
(
(
38) Weidner, T.; Ballav, N.; Siemeling, U.; Troegel, D.; Walter, T.;
Tacke, R.; Castner, D. G.; Zharnikov, M. J. Phys. Chem. C 2009, 113,
9609−19617.
39) Another reason for the invisibility of upright phenyl rings might
3) Todd, P.; St. Clair, D.; Goodman, D. W. Top. Catal 2000, 13, 5.
4) Hayden, B. E.; Pletcher, D.; Rendall, M. E.; Suchsland, J. P. J.
1
(
Phys. Chem. C 2007, 111, 17044.
5) Risse, T.; Shaikhutdinov, S.; Nilius, N.; Sterrer, M.; Freund, H.-J.
Acc. Chem. Res. 2008, 41, 949.
6) Lin, X.; Nilius, N.; Freund, H.-J.; Walter, M.; Frondelius, P.;
Honkala, K.; Hakkinen, H. Phys. Rev. Lett. 2009, 102, 206801.
7) Molina, L. M.; Hammer, B. Appl. Catal., A 2005, 291, 21−31.
8) Yoon, B.; Hakkinen, H.; Landman, U.; Worz, A. S.; Antonietti, J.;
be their quick motion below the tip. Using Hessian harmonic
frequency calculations, we have identified the torsion of those groups
to be the lowest vibrational excitation (1−7 meV) that can be excited
by inelastic electron tunneling even at 5 K. This motion leads to a
blurred topographic signature of the peripheral rings and renders their
identification difficult
(
(
̈
(
(
̈
̈
(
40) (a) Jag
Hoffmann, W.; Franchy, R.; Ibach, H. Surf. Sci. 1991, 259, 235−252.
b) Kulawik, M.; Nilius, N.; Rust, H.-P.; Freund, H.-J. Phys. Rev. Lett.
003, 91, 256101.
41) Kresse, G.; Schmid, M.; Napetschnig, E.; Shishkin, M.; Ko
L.; Varga, P. Science 2005, 308, 1440−1442.
̈
er, R. M.; Kuhlenbeck, H.; Freund, H. J.; Wuttig, M.;
Abbet, S.; Judai, J.; Heiz, U. Science 2005, 307, 403.
(
(
8
(
9) Campbell, C. T. Surf. Sci. Rep. 1997, 27, 1.
10) Narayanan, R.; El-Sayed, M.-A. J. Am. Chem. Soc. 2003, 125,
340.
11) Voorhees, P. W. J. Stat. Phys. 1985, 38, 231.
12) Starr, D. E.; Shaikhutdinov, S. K.; Freund, H. J. Top. Catal.
(
2
(
̈
hler,
(
2
(
2
(
005, 36, 33.
13) Monzon, A.; Garetto, T. F.; Borgna, A. Appl. Catal., A 2003,
48, 279−289.
14) Bernal, S.; Botana, F. J.; Calvino, J. J.; Cifredo, G. A.; Perezomil,
J. A. Catal. Today 1995, 23, 219−250.
(
(
15) Fung, S. C. Catal. Today 1999, 53, 325−338.
16) Chen, M. S.; Goodman, D. W. Chem. Soc. Rev. 2008, 37, 1860−
1870.
(
17) Tosoni, S.; Boese, D. A.; Sauer, J J. Phys. Chem. C 2011, 115,
24871−24879.
1
1167
dx.doi.org/10.1021/ja300304s | J. Am. Chem. Soc. 2012, 134, 11161−11167