Reactivity studies with gold-supported molybdenum nanoparticles

Article abstract of DOI:10.1016/j.susc.2004.10.035

The reconstructed (22 × √3)-Au(1 1 1) surface was used as a template and inert support for depositing Mo nanoparticles for reactivity studies of desulfurization and the formation of MoS_x nanoparticles. Nanoparticles of Mo were prepared on the Au(1 1 1) substrate by two methods: physical vapor deposition (PVD) of Mo and UV-assisted chemical vapor deposition (UV-CVD) through a molybdenum hexacarbonyl precursor. STM studies have shown that the Mo nanoparticles are thermodynamically unstable on the Au(1 1 1) surface, and that gold encapsulates Mo at temperatures above 300 K. Reactivity studies using Auger electron spectroscopy (AES) and temperature programmed desorption (TPD) show that bare Mo nanoparticles are very reactive and can cause complete dissociation of hydrogen sulfide, methyl mercaptan, and thiophene. The presence of gold atoms on the Mo nanoparticles modifies their reactivity. In the case of H₂S and CH₃SH, the overall activity for desufurization is unaffected by gold encapsulation; however, the selectivity to form methane from CH₃SH increased from 20% on bare Mo particles to 60% on gold-covered Mo particles. In contrast, gold-encapsulated Mo nanoparticles are relatively inert towards dissociation of thiophene. We believe that the interaction of R-SH compounds with Au-encapsulated Mo nanoparticles proceeds through intermediacy of surface gold thiolates.

Full text of DOI:10.1016/j.susc.2004.10.035

Surface Science 574 (2005) 244–258
www.elsevier.com/locate/susc
Reactivity studies with gold-supported molybdenum
nanoparticles
a
a
Denis V. Potapenko , Jillian M. Horn a,b, Robert J. Beuhler ,
a
Zhen Song , Michael G. White
a,b,
*
a
Chemistry Department, Brookhaven National Laboratory, Bldg 555, Upton, NY 11973, USA
b
Department of Chemistry, Stony Brook University, Stony Brook, NY 11790, USA
Received 11 August 2004; accepted for publication 21 October 2004
Available online 5 November 2004
Abstract
p
The reconstructed (22 · 3)-Au(111) surface was used as a template and inert support for depositing Mo nanopar-
ticles for reactivity studies of desulfurization and the formation of MoSx nanoparticles. Nanoparticles of Mo were pre-
pared on the Au(111) substrate by two methods: physical vapor deposition (PVD) of Mo and UV-assisted chemical
vapor deposition (UV-CVD) through a molybdenum hexacarbonyl precursor. STM studies have shown that the Mo
nanoparticles are thermodynamically unstable on the Au(111) surface, and that gold encapsulates Mo at temperatures
above 300K. Reactivity studies using Auger electron spectroscopy (AES) and temperature programmed desorption
(TPD) show that bare Mo nanoparticles are very reactive and can cause complete dissociation of hydrogen sulfide,
methyl mercaptan, and thiophene. The presence of gold atoms on the Mo nanoparticles modifies their reactivity. In
the case of H2S and CH3SH, the overall activity for desufurization is unaffected by gold encapsulation; however, the
selectivity to form methane from CH3SH increased from 20% on bare Mo particles to 60% on gold-covered Mo par-
ticles. In contrast, gold-encapsulated Mo nanoparticles are relatively inert towards dissociation of thiophene. We
believe that the interaction of R–SH compounds with Au-encapsulated Mo nanoparticles proceeds through intermedi-
acy of surface gold thiolates.
ꢀ 2004 Elsevier B.V. All rights reserved.
Keywords: Scanning tunneling microscopy; Thermal desorption spectroscopy; Chemical vapor deposition; Surface structure, morp-
hology, roughness, and topography; Molybdenum; Sulfides; Clusters
1. Introduction
*
Corresponding author. Tel.: +1 631 344 4345; fax: +1 631
344 5815.
E-mail address: mgwhite@bnl.gov (M.G. White).
MoS2 is an important catalyst used in hydrodesul-
furization (HDS) processes, where sulfur-containing
0039-6028/$ - see front matter ꢀ 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2004.10.035

D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
245
impurities are removed from petroleum [1]. When
present in crude oil, these impurities can cause
many undesirable affects, such as releasing envi-
ronmentally harmful sulfur oxides when the fuel
is burned [2]. With new legislation being passed
around the globe to reduce the amount of sulfur
impurities allowed in fuel [3], HDS catalysts with
better efficiency are needed. One approach is to
modify the physical (e.g., size) and chemical prop-
erties (e.g., alloy, support) of nanostructured
forms of the catalytic material in an effort to tailor
its catalytic activity and selectivity. It has been
shown that nanoparticles may have different chem-
ical properties compared to that of the bulk mate-
rial [4–6], which results from unique adsorption
sites and/or electronic properties of the nanoparti-
cle that are characteristic of its size and chemical
environment [6–8].
In this work, we report on the preparation and
reactivity of Mo nanoparticles on Au(111). Qual-
itative data on the morphology and growth of the
Mo and MoSx nanoparticles is presented through
STM images. We find that nanoparticles of bare
Mo are thermodynamically unstable on the
Au(111) surface and that gold atoms move to
the Mo surfaces at temperatures above 300K.
Upon sulfiding, the gold is expelled from the par-
ticles and moves back to the terraces. Quantitative
measurements of this effect have been completed
using Auger electron spectroscopy (AES). The
reactivities of bare and gold-covered Mo nanopar-
ticles prepared by PVD on Au(111) were studied
using simple HDS test molecules (thiophene,
methanethiol, and hydrogen sulfide). Thiophene
is a frequently used test molecule for HDS process
and is important because it is the simplest of the
cyclic hydrocarbons that are resilient toward
HDS [23,24]. Reaction products were monitored
with temperature programmed desorption (TPD)
measurements, and the surface species were char-
acterized using AES. The presence of gold atoms
on the Mo nanoparticles modifies their reactivity.
In the case of H2S and CH3SH, the overall activity
for desufurization is unaffected by gold encapsula-
tion; however, the selectivity to form methane
versus deposited carbon increases for CH3SH
decomposition. By contrast, gold-encapsulated
Mo nanoparticles exhibit little activity for the dis-
sociation of thiophene.
Recently, Besenbacher and co-workers have
shown that the Au(111) surface can be used as
a template for MoS2 nanoparticle growth by
physical vapor deposition (PVD) of Mo in the
presence of H2S [9]. The Au(111) surface is used
as a substrate due to its chemical inertness and its
p
characteristic 22 · 3 herringbone surface recon-
struction which provides a template for nanopar-
ticle growth [10]. Strain relief at the Au(111)
surface results in a network of dislocation stripes
which separate alternating fcc and hcp stacking
regions that appear as a zigzag, herringbone
structure in STM images [11]. Previous studies
have shown that metal clusters (Ni, Fe, Rh, Co,
Mo and Pd) preferentially nucleate at the elbows
of the herringbone structure when deposited by
PVD [12–19]. Mo nanoparticles are stable on
gold surfaces since these two metals are mutually
insoluble according to the phase diagram [20]. In
the case of Mo PVD on Au(111), the resulting
array of nanoclusters can be sulfided to produce
relatively monodispersed MoS2 islands [9,21]. In
more recent studies, Mo nanoparticles on
Au(111) have been prepared using chemical
vapor deposition (CVD) of a molybdenum hexa-
carbonyl (Mo(CO)6) precursor [17,22]. In con-
trast to PVD, these clusters were found to be
mobile when deposited, forming three-dimen-
sional ramified cluster islands [17] or concentrat-
ing at the step edges [22].
2. Experimental
2.1. STM experiments
The STM experiments were carried out in an
ultrahigh vacuum system equipped with a scan-
ning tunneling microscope (STM, Omicron), a
sputter gun, LEED/Auger and e-beam evaporator
(Oxford Applied Research). A preparation cham-
ber was used for sample cleaning and preparation
with a directed doser used for chemical vapor
deposition. The Au(111) surface was cleaned by
cycles of sputtering with Ne+ (1keV) at 520K, fol-
lowed by annealing to approximately 900K until
the surface was determined to be clean, with

246
D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
well-defined terraces and herringbone reconstruc-
tion observed in STM images. Molybdenum parti-
cles were deposited by physical vapor deposition
(PVD) using an e-beam evaporator and a UV-
assisted chemical vapor deposition (UV-CVD)
procedure previously demonstrated for Mo depo-
sition on Ru(001), Rh(100), Ag(111) and graph-
ite [25–28] surfaces. The UV-CVD procedure
involves dosing the Mo(CO)6 precursor on the
gold surface at less than 150K, followed by expo-
sure to UV light (k = 300nm). The latter was
generated by a monochromator-equipped He–Xe
arc-lamp (Oriel) and induces partial dissociation
of the precursor, i.e.,
was deposited on the Au(111) surface by PVD
using an e-beam evaporator similar to that em-
ployed in the STM studies. The deposition rates
of the evaporator source were calibrated using a
quartz microbalance. A directed doser was used
to expose the surface to thiophene (Aldrich), meth-
anethiol (Matheson), and hydrogen sulfide (Math-
eson). The liquid thiophene sample was held in a
glass container and degassed by multiple freeze-
thaw cycles (with liquid nitrogen) while pumping.
The surface coverages of S-containing organic
compounds are reported in ‘‘single layers’’. This
is the coverage, above which the onset of multi-
layer desorption of a given compound from the
Au(111) surface occurs. The reactivity was studied
by temperature programmed desorption (TPD)
and AES. During the TPD measurements, multi-
ple masses were monitored simultaneously with a
differentially pumped quadrupole mass spectrome-
ter at a heating rate of 2K/s. The sample tempera-
ture was measured with a K-type thermocouple in
direct contact with the gold sample.
hm
MoðCOÞ ðadsÞ ! MoðCOÞxðadsÞ þ ð6 À xÞCOðgÞ:
6
The excitation wavelength (k = 300nm) corre-
sponds to a maximum in the adsorption spectrum
of Mo(CO)6 [29], and the UV dose (10mW/cm2 for
2min) was determined by monitoring the CO
desorption kinetics as a function of UV fluence
and exposure time [25–28]. Following UV irradia-
tion, the sample was heated to 400K to remove the
remaining CO
3. Results and discussion
MoðCOÞ ðadsÞ T¼400 K MoðadsÞ þ xCOðgÞ:
ƒƒƒƒ!
x
3.1. Gold encapsulation of Mo particles on
Au(111)
The Mo(CO)6 precursor (Alfa Aesar) was held
in an independently pumped glass tube at room
temperature, which was pumped out briefly before
each dose. For sulfidation, the formed nanoparti-
cles were exposed to 50L of D2S (Cambridge Iso-
tope Laboratories) while the sample was kept at
520K, followed by annealing to 670K. All STM
images were acquired with a tungsten tip at room
temperature.
Thermal desorption studies have shown that the
chemical properties of freshly prepared Mo/
Au(111) surfaces are qualitatively different than
that of the same surfaces that have been annealed.
Specifically, the surfaces that were prepared by
PVD of Mo on a cold (ꢀ100K) Au(111) sample
caused complete dissociation of all test molecules
tried (H2S, CH3SH, and C4H4S). However, after
a first TPD run with any of the above compounds,
the surface was significantly passivated, and subse-
quent TPD experiments on the same surface exhib-
ited different reactivity patterns, as reported in the
next section. Here, we report the results of an
investigation into the origin of this passivation ef-
fect. Experiments have shown that the observed
change in surface chemical properties does not de-
pend on the amount of the substance deposited for
the first TPD run. Moreover, heating of a freshly
prepared Mo/Au(111) surface to temperatures
2.2. Reactivity experiments
Reactivity experiments were carried out on a
separate ultrahigh vacuum system equipped with
a quadrupole mass spectrometer (Extrel), Auger
spectrometer (Physical Electronics), a metal evap-
oration source (Oxford Applied Research), and a
sputter gun (LHK). The Au(111) surface was
cleaned by sputtering cycles of Ne+ (1keV), fol-
lowed by annealing to 850K. The surface was
determined to be clean with AES. Molybdenum

D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
247
above 400K without exposure to any substance
200 K
2.0
1.5
1.0
0.5
led to the same passivation effect. These observa-
tions suggest that the passivation of the Mo/
Au(111) surface is not caused by carbon or sulfur
deposition from the dissociation of adsorbed com-
pounds, but rather is a result of a temperature-in-
duced structural change on the Mo/Au(111)
surface.
Fig. 1 shows AES peak height ratios of Mo
(190eV) versus Au (74eV) for the Au(111) sur-
face, on which approximately 1.5ML of Mo was
deposited by PVD at 100K. In a set of separate
experiments, this sample was incrementally heated
to temperatures up to 800K with a heating rate of
2K/s. Cooling of the sample began immediately
upon reaching each of the target temperatures. It
is seen that the surface Mo/Au ratio decreases with
heating of the sample to temperatures above
300K. The structural transformations responsible
for the change in surface Mo/Au ratio appear to
be kinetically limited processes, since the ratio
does not saturate after successive temperature
ramps.
300 K
initial
(100 K)
400 K
500 K
600 K
700 K
800 K
1
2
3
4
5
6
7
8
Experiment number
Fig. 1. Mo (190eV) versus Au (74eV) AES peak height ratios
from Mo deposited by PVD (ꢀ1.5ML) on a Au(111) surface at
100K (the first point). Points at higher temperatures correspond
to consecutive temperature ramps (2K/s) in which the surface
was ramped to a final temperature between 200 and 800K. AES
spectra were taken after each temperature ramp when the
sample cooled back to 100K.
vapor source nucleates preferentially at the elbows
p
STM studies of the Mo/Au(111) surface mor-
phology give a deeper insight into the tempera-
ture-induced structural transformations of this
of the 22 · 3 herringbone surface reconstruction,
as can be seen in Fig. 2b. To a lesser extent, Mo
nucleates at random sites on the terraces and also
at the step edges. This distribution of Mo clusters
is qualitatively similar to that obtained previously
by Helveg et al. [9]. Cross-sectional analyses of the
surface. Fig.
2 shows STM images of the
Au(111) surface with about 0.2ML of Mo depos-
ited by the PVD method. Molybdenum from the
Fig. 2. STM image of the Au(111) surface with about 0.2ML of Mo deposited by the PVD method. Image sizes: (a) 190 · 190nm, (b)
p
50 · 50nm. The underlying (22 · 3)-Au(111) ‘‘herringbone’’ surface reconstruction pattern can be seen in (b).

248
D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
STM images shows that the average height of the
˚
Mo particles is about 4A while the interatomic dis-
the deposition, e.g., site exchange between Au and
Mo atoms.
˚
tance in bulk molybdenum is 2.7A. This suggests
We also note that deposition of Mo on Au(111)
by thermal decomposition of Mo(CO)6 (conven-
tional CVD) leads to particle growth and mor-
phology that are very different from both PVD
and UV-CVD. This is seen in a recent STM study
by Hrbek and co-workers, where Mo nanoparti-
cles were deposited by exposing the gold surface
to Mo(CO)6 at 500K [17]. These nanoparticles
demonstrate a different growth mechanism, with
the Mo clusters forming ramified cluster islands
on the terraces and step edges of gold, which were
not seen with the UV-CVD method. The complete
report on UV-assisted CVD of Mo on Au(111) is
given elsewhere [31]. Here we offer only the results
directly related to the surface chemistry studies
reported in the following sections.
Fig. 3a shows an STM topographic image of the
Au(111) surface with 0.03ML of Mo prepared by
UV-CVD, which was then annealed at 525K for
5min. As seen in the image, most of the Mo clus-
ters have nucleated at the step edges. Using the
rows of the clusters as a reference for the original
positions of the step edges on the surface, one
can argue that the ‘‘bays’’, marked with the white
arrows in Fig. 3a, were formed after the clusters
have nucleated. The formation of similar etch
that the Mo particles row in a Volmer–Weber
fashion on the Au(111) surface, with the forma-
tion of 3D clusters at higher Mo coverages [30].
On the basis of the images in Fig. 2, we assume
that at Mo coverages of 1ML (which were used
in our reactivity studies) the surface still contains
multitudes of partially coalesced Mo nanoparticles
with some fraction of the Au(111) surface remain-
ing uncovered.
A UV-assisted chemical vapor deposition (UV-
CVD) procedure was also used to deposit Mo, and
this approach proved to be more useful for study-
ing Au–Mo interactions on the Au(111) surface.
The UV-CVD method of Mo deposition has two
important advantages for STM studies over the
PVD method used for reactivity studies. First,
the UV-CVD procedure allows preferential nucle-
ation of Mo clusters on the step edges of the sur-
face, which is very helpful in tracing the flow of
step edges [22]. In contrast, Mo nucleates mostly
on terraces when deposited by PVD (see Fig. 2)
[9]. Secondly, the UV-CVD procedure can be used
to nucleate Mo particles at low temperatures. With
PVD, hot Mo atoms from the vapor may cause
structural changes in the Au(111) substrate during
Fig. 3. Topographic (a) and differential (b) representations of the STM image of the Au(111) surface on which ꢀ0.03ML of Mo was
deposited through UV-assisted CVD using Mo(CO)6 as a precursor. The sample was annealed at 520K before imaging at room
temperature. The white arrows point at etching ‘‘bays’’ arising from gold consumption by Mo clusters. Image sizes: 160 · 160nm.

D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
249
holes was observed previously [17,22] and the de-
tailed mechanism for formation of these bays is
discussed elsewhere [31]. The presence of the bays
indicates consumption of gold atoms somewhere
on the surface. Since all the surface areas imaged
by STM looked qualitatively similar to that shown
in Fig. 3a and no regions of apparent Au segrega-
tion could be found, we arrive at the conclusion
that the Mo clusters were consuming gold during
annealing.
The differential representation in Fig. 3b of the
same STM data shows that after annealing, the
terraces form an irregular herringbone reconstruc-
tion pattern. The origin of the Au(111) surface
reconstruction lies in a subtle mismatch between
the bulk crystal lattice and the topmost layer
[11]. If top layers of the Au terraces (nanoparti-
cle-free areas) contained even a small fraction of
alloyed Mo we would not expect to see surface pat-
terns similar to the herringbone reconstruction.
The fact that the Au(111) surface reconstruction
pattern is irregular, but not lifted completely, sug-
gests that very little Mo migrates to the gold ter-
races, and prefers to remain in compact clusters
after annealing. The same conclusion can be
drawn from the Au–Mo phase diagram [20] which
shows that the solubility of Mo in Au is extremely
low (<1.25 at.% even at 1337K).
From the same Au–Mo phase diagram [20],
these two metals do not form any stable bimetallic
compound and the solubility of gold in molybde-
num is extremely low (<0.025 at.%). Therefore, it
is unlikely that the gold atoms penetrate the Mo
clusters but rather segregate on their surface or
encapsulate them. This picture is qualitatively con-
sistent with the lower surface free energy of Au
(1.62J/m2) relative to Mo (2.88J/m2) which ener-
getically favors Au capping a Mo surface [32,33].
Additional support is provided by density func-
tional theory (DFT) calculations by Liu et al.,
showing that a single layer of Mo atoms on a
Au(111) surface is less stable than a sandwich
structure in which the Mo layer is covered by a
surface layer of Au atoms [34]. More sophisticated
calculations of the surface segregation energies by
Ruban et al., also suggest the tendency of gold to
bury molybdenum [35]. These authors also pre-
dicted site exchange or intermixing between Mo
nanoparticles with the Au substrate. Our STM re-
sults indicate that the Mo–Au interaction results in
encapsulation of Mo nanoparticles by Au atoms,
migrating from the step edges.
Encapsulation of the deposited metal by the
substrate material is usually (but not always) ob-
served when the two metals are immiscible and
the surface free energy of the substrate is lower
than that of the deposited metal [36]. Encapsula-
tion behavior was reported for Co on Cu(100)
[37], on Cu(111) [38], on Ag(100) [39], and on
Au(111) [14], Cu on Pb(111) [40], Rh on
Au(111) [41], and on Ag(100) [42]. It is interesting
to mention that in case of Co/Au(111) system, Co
nanoparticles become encapsulated by sinking into
the substrate, and the expelled Au proceeds to-
ward the step edges to extend the terraces [14],
i.e., the step flow dynamics is just the opposite to
what we observe in our studies. On the other hand,
encapsulation of Rh nanoparticles deposited on
Ag(100) also consumes the substrate material
from the step edges and etching ‘‘fjords’’, analo-
gous to our ‘‘bays’’, were observed in the STM
images [42].
Migration of Au onto the Mo particles is also
consistent with the observed kinetic limitations
for the encapsulation process at high Mo cover-
ages (>0.5ML), illustrated by Fig. 1. Since the
Au atoms required for encapsulation are supplied
from the step edges and not from the terraces or
from underneath the Mo clusters, a large surface
coverage of clusters would create a hindrance to
the step flow. It is possible that at high Mo cover-
ages other mechanisms of gold supply may need to
be activated to reach complete encapsulation.
The aforementioned DFT study of Liu et al.
[32] also predicts that in the presence of a sulfur
adsorbate on a Mo/Au(111) surface, Mo would
be pulled back to the surface to form sulfides,
MoSx. This phenomenon was seen in our studies.
Fig. 4 shows an STM image of a Au(111) surface
with Mo nanoparticles sulfided with D2S. The sur-
face was prepared by deposition of 0.02ML of Mo
using UV-CVD method then annealed at approxi-
mately 520K for 5min. With appropriate
Mo(CO)6 deposition and UV exposure conditions,
STM images confirmed that the as-prepared Mo/
Au(111) surface had a much lower number of

250
D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
0.36
0.34
0.32
0.30
0.28
0
2
4
6
8
10
12
14
16
S coverage, %
Fig. 5. Measured Mo (190eV) to Au (74eV) peak height ratios
obtained by AES and plotted versus sulfur content of a Mo/
Au(111) surface. The surface was prepared by the deposition of
about 0.8ML of Mo by PVD and annealing to 700K. Sulfur
was deposited in a series of TPD experiments with H2S (see text
for details).
Fig. 4. STM image of the Au(111) surface with about 0.02ML
of Mo deposited in a way similar to that of Fig. 3. The surface
then reacted with D2S at 520K and annealed at 670K. The
white arrows point at chains of Mo clusters that mark the step
positions before the reaction with D2S. Image size:
240 · 240nm.
bays in the terraces compared to that in Fig. 3. As
before, most of the Mo clusters were situated pref-
erentially on the step edges of the Au(111) surface.
The clusters were then sulfided by exposure to
D2S.
The white arrows in Fig. 4 point at chains of
Mo clusters that apparently nucleated at the step
edges during deposition. Upon D2S exposure,
however, the terraces on the gold surface were ex-
tended beyond the lines of the clusters in areas
marked with the arrows. We propose that the ex-
tended portions of the terraces were formed by
the diffusion of gold, expelled from the surface of
Mo clusters in the reaction with D2S.
AES measurements are consistent with the
above hypothesis. Sulfur, deposited on a Mo/
Au(111) surface binds preferentially to Mo
[9,34]. The deposited sulfur atoms should then
screen Mo atoms to a greater extent than Au
atoms. If no restructuring of the Mo–Au surface
is involved in the process of Mo sulfidation, the
Mo/Au Auger signal ratio would decrease with
increasing S coverage. Experimental data plotted
in Fig. 5 exhibits just the opposite trend. The data
in Fig. 5 were obtained through a series of con-
secutive TPD experiments with a Mo/Au(111)
surface (0.8ML) which was annealed to 700K
and then cooled to 90K and exposed to H2S at
saturation coverage for each TPD run. Auger
spectra were recorded from the initial surface
and after each of the TPD experiments. The ratio
of Mo (190eV) versus Au (74eV) Auger peak
heights was then plotted against sulfur coverage.
The plot clearly shows that the Mo/Au Auger sig-
nal ratio increases as more sulfur is deposited on
the surface. This behavior can be explained by
having Au atoms expelled from the surface of
the Mo nanoparticles by deposited sulfur. Since
the expelled Au atoms migrate to the Mo-free
areas of the surface, the overall gold AES signal
intensity decreases, consistent with the trend seen
in Fig. 5.
Although the experiments above provide strong
evidence for gold encapsulation of the Mo nano-
particles, the physical structure of the Au-covered
Mo surface is not discernable from the STM
images. Of particular interest is whether Au simply
wets the Mo surface to form islands, i.e., gold
caps, or forms a Au/Mo surface alloy with a
well-defined morphology and composition. Gold
is known to form surface alloys with other early
transition metals, e.g., Au/W [43–45] and Au/Ni

D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
251
[46,47], even though the two metals are immiscible
in the bulk (similar to Au/Mo) [48]. Characteristic
of such surface alloys are well-ordered c(2 · 2) sur-
face structures that appear at up to 0.5ML in
which Au atoms are substituted for the substrate
atoms in the surface layer. Similar c(2 · 2) surface
structures have been observed for Cu, Ag [49] and
Pd [50] overlayers on Mo(100) surfaces and very
recent theoretical studies [45] indicate that Cu,
Ag and Au will form stable substitutional surface
alloys on a Mo(001) surface. In the case of Mo
nanoparticles, surface alloying may be favored
since nanoparticles are expected to be rich in sur-
face vacancies and defects, e.g., kink and edge sites
[51]. The latter are often associated with metal sites
of low co-ordination number and high reactivity,
and Au substitution at these sites could lead to
the observed chemical passivation.
Molecular hydrogen (m/e = 2) was the major
desorption product observed in TPD experiments
on Mo/Au(111) surfaces exposed to H2S, CH3SH,
and C4H4S, as shown in Fig. 6. In these experi-
ments, the surfaces were prepared by PVD of
ꢀ1ML of Mo onto a Au(111) surface held at
about 85K, which ensures that the deposited Mo
particles were not encapsulated by gold. The test
molecules were then deposited at the same sample
temperature by means of a directed doser at cover-
ages slightly above a single layer. In the case of
CH3SH, methane desorption was also seen at
290K, as shown in Fig. 6. The identity of the
desorbing product as methane was confirmed by
simultaneous monitoring of mass spectrometer
We can use the Mo AES signal attenuation dur-
ing annealing of a Mo/Au surface to estimate the
gold layer thickness on Mo nanoparticles. The
largest decrease in the normalized Mo signal, ob-
served after annealing to 800K, corresponded to
attenuation value of 0.65. Taking the inelastic
mean free path for an electron in gold as 0.6nm
[52], the observed attenuation corresponds to a
0.26nm thick gold film or ꢀ1.1ML of gold. This
estimation strongly favors Au capping of Mo
nanoparticles as opposed to the formation of a
Mo–Au surface alloy. Although the above analysis
is very approximate, we will refer to this Au migra-
tion as encapsulation in the remainder of the
paper.
m/e = 2
a) H2S
b) CH3SH
c) C4H4S
d) bare surface
3.2. Reactivity of Mo nanoparticles on Au(111)
surface
m/e = 15
x 2
e) CH3SH
In this section we report the results of a com-
bined TPD/AES study of chemistry of sulfur-con-
taining compounds on Mo/Au(111) surfaces.
Quantitative analysis of the data was performed
using TPD peak areas as a measure of the amount
of desorbing products. Changes in AES peak
intensities, before and after a TPD experiment,
were used as a measure of the amounts of depos-
ited elements. Separate studies were carried out
on bare and Au-encapsulated Mo nanoparticles,
denoted Mo/Au and (Mo)Au respectively.
100
200
300
400
500
600
700
Temperature, K
Fig. 6. Thermal desorption spectra from Mo/Au(111) surfaces
that were prepared by PVD of 1ML of Mo on the cold
(<100K) Au(111) surface followed by dosing with various
compounds: (a) H2S, (b) and (e) CH3SH, and (c) thiophene.
Spectrum d) was taken after Mo deposition without dosing.
Ionic signals for Hþ2 at m/e = 2 (a–d) and CHþ3 at m/e = 15 (e)
are shown.

252
D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
signals at m/e = 15 and 16. The ratio of the two sig-
nals matched that for the electron fragmentation
of CH4 [53]. No other hydrocarbons were detected
in the desorption spectra (except for residual
amounts of the unreacted deposited compounds).
Hydrogen desorbs in a very broad temperature
range of 100–500K from all Mo/Au(111) surfaces
covered with sulfur compounds. In the case of
H2S, the hydrogen desorption peak has a maxi-
mum at 180K (Fig. 6a) and a broad shoulder
at ꢀ350K. Exposing Mo/Au(111) surfaces to
C4H4S produces H2 desorption peaks with a much
less pronounced maximum at ꢀ220K (Fig. 6c). Fi-
nally, hydrogen desorption spectra from Mo/
Au(111) surfaces exposed to CH3SH exhibit a
double peak with maxima near 220 and 300K
(Fig. 6b). The fact that the second H2 desorption
peak coincides in temperature with the CH4
desorption peak suggests a common mechanism
for the formation of these two products.
A quantitative analysis of hydrogen desorption
was not possible due to the fact that Mo/Au(111)
surfaces prepared by PVD were nearly saturated
with hydrogen even before deposition of any com-
pound. This can be seen in Fig. 6d, which shows
H2 desorption from a freshly prepared Mo/
Au(111) surface. Apparently, hydrogen was co-
deposited with Mo during PVD. Molybdenum
has a very low vapor pressure and requires high
temperatures for deposition (above the melting
point of 2883K). At such high temperatures the
evaporative source itself produces hydrogen, rais-
ing the partial pressure of hydrogen in the cham-
ber to approximately 2 · 10À8 Torr, in spite of
water cooling and preliminary outgassing of the
source.
with regard to any of the S-containing compounds.
Indeed, after single TPD runs the sulfur and car-
bon coverages of the surfaces were close to the cor-
responding saturation values (see Fig. 8). We also
conducted a number of experiments with D2S
deposited on Mo/Au(111) surfaces. A careful
analysis of the D/H ratio in the thermal desorption
products showed that only about 25% of the
desorbing molecular hydrogen (deuterium) origi-
nates from the deposited multilayer of H2S
(D2S). The remainder of the desorbing hydrogen
was co-deposited together with molybdenum.
Based on AES measurements, the sulfur coverage
after such an experiment was 20%, while the max-
imum coverage obtained after repeated exposures
to H2S was 24%. Together these observations sug-
gest that there are many more adsorption sites for
hydrogen than for sulfur on the Mo/Au(111) sur-
face. Another conclusion derived from the isotope
experiments is that the hydrogen deposited with
Mo and the hydrogen deposited from H2S (D2S)
are chemically indistinguishable on the surface.
Since the D/H ratio remained virtually constant
during the entire desorption process, surface
hydrogen from both sources must desorb through
an identical recombinative process.
The reactivity of Au-encapsulated Mo nanopar-
ticles was also studied. In these experiments, the
substrate surfaces were prepared by deposition of
about 1ML of Mo atoms from the PVD source
followed by heating of the sample to 600K. In
the case of H2S, (Mo)/Au(111) surfaces produced
molecular hydrogen upon heating (as well as intact
H2S molecules) and the corresponding desorption
peak had a temperature profile resembling that
of bare Mo particles (Fig. 6a). However, the inten-
sity of the hydrogen peak from the (Mo)/Au(111)
surface was at least five times lower. The latter is
due to the removal of adsorbed hydrogen that
was co-deposited with Mo by annealing the sample
prior to H2S exposure. Since the background levels
of hydrogen in the vacuum chamber were elevated
during desorption, these m/e = 2 TPD signals were
not suitable for quantitative analysis (e.g., see Fig.
7).
Hydrogen saturation was investigated through
experiments in which Mo was deposited in a back-
ground pressure of D2 (up to 3 · 10À7 Torr).
Although the D/H ratio on the prepared surfaces
could exceed one, the total area of m/e = 2, 3,
and 4 TPD peaks from these surfaces was close
to that of the m/e = 2 TPD peak from a Mo/
Au(111) surface that was prepared with no D2
exposure. This suggests that freshly prepared
Mo/Au(111) surfaces were initially saturated with
hydrogen. On the other hand, this surface hydro-
gen did not passivate the Mo/Au(111) surfaces
Thermal desorption products from CH3SH-
covered (Mo)/Au(111) surfaces were also similar
to those from the Mo/Au(111) surface, including

D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
253
m/e = 48 TPD signal. The broad peak between
300 and 550K, however, is not accompanied by in-
tact CH3SH desorption and can be attributed to
cracking of the methane desorption product in-
stead. In the case of thiophene-covered (Mo)/
Au(111) surfaces, only desorption of intact thio-
phene molecules could be distinguished, and we
do not show the corresponding TPD spectrum.
Analyses of the AES spectra provide a quantita-
tive measure of the extent of the surface reactions.
Fig. 8a–c summarize the results in a series of sulfur
uptake curves for H2S, CH3SH and C4H4S. Each
of the points in the figures was obtained after a
TPD experiment, with height ratios between S
(150eV) and Mo (190eV) plotted as a function
of the amount of deposited compound. Open cir-
cles represent experiments with a bare Mo/
Au(111) surface. Only one experiment on bare
Mo/Au(111) could be performed since at elevated
temperatures Au atoms have migrated to Mo sur-
faces that have not reacted with sulfur. The filled
symbols depict sulfur coverages after successive
TPD experiments on (Mo)/Au(111) surfaces that
were prepared by annealing to 600K.
m/e = 2
(1)
m/e = 15
(1)
(2)
(3)
m/e = 48
(3)
(2)
(1)
100
200
300
400
500
600
Temperature, K
Fig. 7. Three consecutive thermal desorption spectra from a
Mo/Au(111) surface that was exposed to a saturated layer of
CH3SH before each experiment. The initial surface was
prepared by PVD of 1ML of Mo on a cold (100K) Au(111)
surface followed by annealing to 600K. The order of the TPD
run is shown in the parenthesis. Ionic signals for Hþ2 ; CH3þ and
CH3SH+ (m/e = 2,15 and 48) are shown.
As seen in Fig. 8a and b, the reactivities of Mo/
Au(111) and (Mo)/Au(111) surfaces toward H2S
and CH3SH are similar. Apparently, the presence
of gold does not significantly hinder the reaction
between the sulfur from these compounds and
molybdenum. The uptake curves suggest satura-
tion of the surfaces with sulfur, yet no specific sat-
uration sulfur concentration was reached in our
experiments. Even with prolonged exposures of
the heated surfaces to S-containing compounds
the concentrations of sulfur on the surfaces kept
increasing, apparently due to the slow process of
sulfur diffusion into the bulk of the Mo
nanoparticles.
Unlike the other two compounds, thiophene
(C4H4S) reacts quite differently toward Mo/
Au(111) compared with (Mo)/Au(111) surfaces
(see Fig. 8c). While thiophene is completely disso-
ciated on bare Mo nanoparticles, only a small
fraction of deposited thiophene reacts with gold-
encapsulated particles. Additional experiments
showed that the reactivity of the (Mo)/Au(111)
surface with thiophene depends on the degree
of encapsulation of the Mo particles. In four
methane and molecular hydrogen. Fig. 7 shows
three consecutive TPD spectra from the (Mo)/
Au(111) surface, for each of which a single layer
of CH3SH was deposited. Only the first spectrum
is shown for m/e = 2, since in the other two runs
the hydrogen peak was not discernable. As can
be seen from the figure, subsequent TPD runs
resulted in less methane (m/e = 15) and a propor-
tional increase in intact CH3SH desorption
(m/e = 48), demonstrating saturation of the surface
with sulfur (and carbon). It should be noted that
mass fragments with m/e = 15 ðCHþ3 Þ also origi-
nate from cracking of the parent molecule,
CH3SH, in the ionizer of the mass spectrometer.
This can be seen in Fig. 7 as a small peak in
the m/e = 15 TPD at 160K that replicates the

254
D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
(a) H2S
(b) CH3SH
3
2
1
0
2
1
0
on Au-encapsulated Mo
on bare Mo
0
1
2
3
4
0
1
2
3
4
55
50
45
40
35
30
(d) ∆ Mo% upon anneal
1.5
1.0
0.5
0.0
(c) C4H4S
anneal
after Mo
temperature:
deposition
550 K
600 K
after anneal
700 K
750 K
0
1
2
3
4
550
600
650
700
750
cumulative exposure, layers
anneal temperature, K
Fig. 8. (a)–(c) Sulfur uptake curves for different compounds on Mo/Au(111) surfaces. Each data point was obtained by deposition of a
given compound on a surface followed by thermal desorption. The ratio of the S (152eV) to Mo (190eV) AES peak intensities
represent the sulfur content of the surfaces. Open circles in each figure correspond to sulfur uptake of bare Mo nanoparticles on
Au(111) and closed symbols represent sulfur uptake of Au-encapsulated Mo nanoparticles. Plot (c) shows the uptake curves for a
number of (Mo)/Au(111) surfaces prepared by annealing to different temperatures. The change of Mo content of the surfaces during
annealing (derived from AES spectra) is shown in plot (d).
experiments, Mo/Au(111) surfaces were prepared
in identical ways, and annealed to different tem-
peratures: 550, 600, 700, and 750K. AES spectra
were taken before and after each anneal and the
apparent Mo content of the surfaces is plotted in
Fig. 8d. The change in Mo AES signal increases
as the annealing temperature is increased. Consis-
tent with our earlier observations (see Fig. 1),
higher annealing temperatures produce a higher
average density of gold caps on the Mo nanoparti-
cles. Reactivity experiments with these four sur-
faces (Fig. 8c) showed that the uptake of sulfur
from the deposited thiophene is greater for the sur-
faces annealed to lower temperatures. In other
words, higher gold coverages make the surfaces
more resilient to the reaction with thiophene. In
our estimations, less than 1% of thiophene from
the deposited layer reacted with the (Mo)/
Au(111) surface which was annealed to 750K.
On the other hand, the anneal temperature had
no effect on sulfur deposition from H2S or CH3SH
on gold-encapsulated (Mo)/Au(111) surfaces.
To better understand the reaction mechanisms
of sulfur-containing compounds with different
Mo/Au(111) surfaces, the amounts of carbon
and sulfur deposition were analyzed following sur-
face reactions. The values of the fractional change
in the amount of carbon and sulfur on the surface
(DC/DS) are shown in Table 1. These were calcu-
lated as the ratio of increments of the carbon
(275eV) and sulfur (150eV) AES peak heights di-
Table 1
Elemental ratios of deposited carbon vs. sulfur, DC/DS, for the
reactions of sulfur-containing compounds with bare and gold-
encapsulated Mo nanoparticles on Au(111)
Surface
DC/DS ratio
H2S
CH3SH C4H4S
Mo/Au(111) (bare) 0.0 0.1 0.8 0.2 4.6 0.4
(Mo)Au(111) (encapsulated) 0.1 0.2 0.4 0.2 4.1 0.4
The DC and DS values were derived from AES spectra before
and after the reactions and are proportional to the number
densities of deposited surface carbon and sulfur (normalized to
the AES spectrum of CS2).

D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
255
vided by the same ratio that was obtained in a
calibration experiment with CS2 deposited on
Au(111) surface (divided by two to account for
the S:C ratio in CS2). Calculated in this way, the
DC/DS values reflect the actual atomic ratios of
the elements deposited in the TPD experiments.
The numbers presented in Table 1 were obtained
by averaging over a number of DC/DS values from
different experiments (from 3 to 23 for different
compounds) and the reported uncertainties are
mean square deviations of the averaged samples.
Since H2S does not contain carbon, the DC/DS
for decomposition of this molecule is expected to
be equal to zero. This is what we see in Table 1.
The first H2S column of the table shows that unde-
sirable carbon contamination of the sample (due
to reaction with the background CO, for instance)
is very small in our TPD experiments.
ratios for thermal reaction with both Mo/
Au(111) and (Mo)/Au(111) surfaces are some-
what above four. This number is consistent with
the 4:1 carbon to sulfur stoichiometry and assum-
ing thiophene undergoes complete dissociation on
Mo/Au(111) surfaces. The results of our TPD
experiments with thiophene are consistent with
complete dissociation, since we have never ob-
served any desorption product other than H2 and
intact C4H4S. The fact that the DC/DS values are
above four (and in one case it is even beyond the
uncertainty range) may be explained by imperfect
calibration. For example, if some of the molecules
in the CS2 multilayer used in calibrations are ori-
ented normally to the surface, sulfur atoms would
screen the AES signal from carbon atoms. This
would lower the observed DC/DS ratio for the
CS2 calibration runs and result in overestimated
DC/DS ratios for the other compounds.
Comparison of our results with published reac-
tivity studies of Mo and Au surfaces brings a per-
spective that allows a mechanistic interpretation of
the experimental data. Overall, the reactivity pat-
terns of S-containing compounds on bare Mo
nanoparticles on the Au(111) surface are very sim-
ilar to that of close-packed Mo surfaces. Similar
to our results for bare Mo particles, molecular
hydrogen was the only desorption product from
thiophene-exposed Mo(110) [24] and Mo(100)
surfaces [54,55]. It is interesting to mention that
reactions of thiophene analogues with higher
hydrogen content do produce gaseous hydrocar-
bons on the Mo(110) surface: butadiene in the
case of 2,5-dihydrothiophene [56] and butane and
butene in the case of tetrahydrothiophene [57]. In
our experiments, however, no gaseous hydrocar-
bons were formed with thiophene, even though
the Mo/Au(111) surfaces were pre-saturated with
hydrogen.
The C:S ratio in CH3SH is one, therefore, if
CH3SH were to dissociate completely and produce
only hydrogen, then the DC/DS ratio would be
equal to one according to the reaction,
CH3SH þ Mo=Auð111Þ
! 2H2ðgÞ þ S; C=Mo=Auð111Þ:
ð1Þ
On the other hand, if CH3SH were to react with
the surface with the deposition of S and desorption
of CH4, i.e.,
CH3SH þ Mo=Auð111Þ
! CH4ðgÞ þ S=Mo=Auð111Þ;
ð2Þ
then the expected DC/DS ratio would be equal to
zero. The DC/DS values for the reaction of CH3SH
with Mo/Au(111) and (Mo)/Au(111) surfaces lie
between zero and one, implying that both pro-
cesses (1) and (2) occur upon heating the sample.
A comparison of the TPD peak areas for molecu-
lar hydrogen and methane correlate well with the
numbers in Table 1. In the case of bare Mo nano-
particles, DC/DS = 0.8, which implies the hydrogen
producing pathway (1) is preferred, while the par-
tial dissociation pathway (2) is more prominent
in the case of gold-encapsulated molybdenum
(DC/DS = 0.4).
Methanethiol (CH3SH) deposited on the
Mo(110) surface behaves very similarly to what
is observed in our experiments with the Mo/
Au(111) surface. Specifically, about 40% of the
methanethiol deposited on Mo(110) forms meth-
ane according to reaction (2), while the rest non-
selectively decomposes to form hydrogen, as in
reaction (1) [58]. Inspection of Table 1 shows that
on the Mo/Au/(111) surface only about 20% of the
Although the overall activity of the Mo nano-
particles towards C4H4S depends strongly on the
degree of Au encapsulation (Fig. 8c), the DC/DS

256
D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
CH3SH decomposes to yield methane with the rest
undergoing complete decomposition. This result is
reasonable since we expect the Mo nanoparticles
to be more active towards complete decomposition
(reaction (1)) than the low-index Mo(110) surface
due to the higher density of reactive edge sites or
other co-ordinatively unsaturated sites. Again,
we should note that the low methane production
yields were observed in spite of the abundance of
pre-adsorbed hydrogen on the Mo/Au(111)
surface.
Thiophene, being a cyclic compound, does not
chemically interact with the Au(111) surface [59].
At the same time, dehydrogenation of organothi-
ols R–SH on the Au(111) and Au(100) surfaces
is a well-known process responsible for the for-
mation of self-assembled monolayers of thiolates
R–S–Au [60]. In the case of alkanethiols
CH3(CH2)nSH, their sticking probability for disso-
ciative chemisorption on both Au(111) and
Au(100) surfaces decreases dramatically with
decreasing chain length n [61]. In the case of
CH3SH (n = 1), only moderate dehydrogenation
was detected on the Au(111) surface under UHV
conditions and monolayer coverages [59] (similar
to the conditions in our studies). The dissociation
of H2S (n = 0) has never been observed on gold
surfaces [62,63], although theoretical calculations
show that such dissociation is slightly more favor-
able energetically than that of methanethiol [64].
The gold atoms capping the Mo nanoparticles
are likely to be coordinatively unsaturated and
more reactive than atoms on close-packed
Au(111) or Au(100) surfaces. We propose that
hydrogen sulfide and methanethiol form thiolate
intermediates (HS– and CH3S–) on the gold cov-
ered (Mo)/Au(111) surface. The formation of R–
S–Au thiolate intermediates [64] keeps the mole-
cules tethered to the (Mo)/Au(111) surface during
TPD experiments until the temperature is suffi-
ciently high to promote mobility of the Au atoms
and expose Mo sites. The latter react with the sul-
fur head group of the adsorbed thiolates to form a
S-Mo bond and reaction products which desorb or
deposit on the surface of the nanoparticles, e.g.,
On the other hand, methane production from
deposited methanethiol increases to 60% on Au
encapsulated Mo particles compared to 20% on
bare Mo/Au(111) particles and 40% on
Mo(110). In this respect, the encapsulated (Mo)/
Au(111) clusters exhibit higher selectivity for the
formation of methane (reaction (2)) versus com-
plete decomposition to surface carbon and sulfur
(reaction (1)). A decrease in carbon deposition is
usually considered a distinct advantage in most
metal-catalyzed reactions as surface carbon often
results in deactivation of the catalyst. For exam-
ple, a Au/Ni surface alloy catalyst has been shown
to have enhanced resistance to carbon poisoning
in steam reforming reactions with hydrocarbons
[47,51]. The modified behavior of the surface alloy
is attributed to Au atoms blocking high reactivity
sites on the supported Ni particles, thereby pre-
venting the formation of carbon deposits [51].
Gold may play a similar role on (Mo)/Au(111)
surfaces; however, the nearly inexhaustible supply
of gold atoms from the Au(111) surface is likely
to result in far greater coverages of Au atoms on
the surface of annealed Mo nanoparticles. In this
case, the Au capping atoms act intermediate
between a physical barrier to adsorbed molecules
and a chemical modifier of the (Mo)/Au(111)
surfaces.
H
(a), H2(g), C(a), and CH4(g). As the surface of the
Mo nanoparticle becomes sulfided, gold atoms
are expelled back to the Au(111) surface (see
Fig. 4). Thiophene, on the other hand, is only
weakly physisorbed on the gold-encapsulated sur-
face and most likely desorbs before it can react
with molybdenum. The dissociation of thiophene
would also require a larger active site (ensemble
of Mo atoms), which may be sterically inaccessible
in the presence of Au atoms on the Mo surface.
[24,54]. These two factors make it highly improba-
ble for thiophene to react with molybdenum on a
gold-encapsulated surface except when the initial
coverage of the Au atoms is low (Fig. 8c and d).
Another unusual property of the gold-encapsu-
lated molybdenum nanoparticles is that they are
virtually unreactive towards thiophene, while they
are almost as reactive to methanethiol and hydro-
gen sulfide as bare molybdenum nanoparticles (see
Fig. 8a–c). To better understand the chemical
properties of the gold encapsulated (Mo)Au(111)
surfaces, we first consider interactions of the
molecules studied here with pure gold surfaces.

D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
257
In a similar manner, the increase of the methane
yield from methanethiol caused by the gold encap-
sulation can also be explained by limited access to
active sites on the Mo surface. Even if the sulfur
atom of a methanethiol molecule has reached
Mo atoms through the gold film, further dehydro-
genation of the CH3 group may be hindered by the
presence of Au atoms in neighboring sites. In this
situation, the CH3 group is more likely to react
with surface hydrogen and desorb as CH4. The
surface hydrogen can originate from the S-H bond
which is broken in forming the gold thiolate or the
dehydrogenation of a fraction of the adsorbed
methyl groups. Due to the limited sensitivity of
the mass spectrometer, we were not able to distin-
guish between these two options.
and CH3SH, the overall activity for desufurization
is unaffected by gold encapsulation, however, the
relative yield of methane from CH3SH increased
to 60% on gold covered Mo particles. In contrast,
gold-encapsulated Mo nanoparticles are relatively
inert towards dissociation of thiophene. We pro-
pose that H2S and CH3SH form thiolate interme-
diates, HS– and CH3S–, that facilitate reaction
on gold encapsulated Mo nanoparticles.
Acknowledgment
The authors would like to thank Dr. Tanhong
Cai and Dr. Jan Hrbek for valuable discussions,
and also for use of the STM and gold crystal. This
research was carried out at Brookhaven National
Laboratory
under
contract
DE-AC02-
4. Summary
98CH10086 with the US Department of Energy
(Division of Chemical Sciences).
Mo nanoparticles were formed on Au(111),
and their reactivity towards H2S, CH3SH, and
C4H4S was studied. STM images have shown that
the nanoparticles formed by UV-CVD nucleate
preferentially at the step edges. Gold migrates to
the surfaces of the Mo nanoparticles when the
sample is annealed above 300K, which is a reflec-
tion of the surface free energy differences between
Mo and Au. Gold atoms required for the encapsu-
lation of the Mo particles are consumed from the
step edges of the Au(111) surface. This leads to
the formation of etching bays behind the lines of
the Mo nanoparticles nucleated at the step edges.
Upon exposure to H2S, Mo reacts to form MoSx
and the gold atoms are expelled from the clusters
back to the terraces. According to STM images,
the nanoparticle morphology remains unchanged
following this reaction.
Reactivity studies of bare Mo nanoparticles
prepared by PVD on Au(111) have shown reactiv-
ity patterns similar to that of the Mo(110) surface.
Each of the S-containing compounds dissociated
completely on the Mo/Au(111) surface producing
mainly hydrogen. Besides hydrogen, the only
desorption product detected was methane evolving
from CH3SH with about 20% yield. The presence
of gold atoms on the Mo nanoparticles signifi-
cantly alters their reactivity. In the case of H2S
References
[1] J.G. Speight, The Chemistry and Technology of Petro-
leum, second ed., New York, McGraw-Hill, 1991.
[2] A.C. Stern, R.W. Boubel, D.B. Turner, D.L. Fox, Funda-
mentals of Air Pollution, second ed., FL Academic Press,
Orlando, 1984.
[3] K.G. Knudsen, B.H. Cooper, H. Topsoe, Appl. Catal. A
189 (1999) 205.
[4] X. Lai, D.W. Goodman, J. Mol. Catal. A 162 (2000) 33,
and references therein.
[5] U. Heiz, W.-D. Schneider, in: K.-H. Meiwes-Broer (Ed.),
Metal Clusters at Surfaces: Structure, Quantum Properties,
Physical Chemistry, Springer, Berlin, 2000, pp. 237–274,
and references therein.
[6] V.I. Bukhtiyarov, M.G. SlinÕko, Russian Chem. Rev. 70
(2001) 147, and references therein.
[7] H.J. Freund, M. Baumer, J. Libuda, T. Risse, G. Ruppr-
echter, S. Shaikhutdinov, J. Catal. 216 (2003) 223.
[8] R. Schlogl, S.B.A. Hamid, Angew. Chem. Int. Ed. 43
(2004) 1628, and references therein.
[9] S. Helveg, J.V. Lauritsen, E. Laegsgaard, I. Stensgaard,
J.K. Norskov, B.S. Clausen, H. Topsoe, F. Besenbacher,
Phys. Rev. Lett. 84 (2000) 951.
[10] B. Hammer, Y. Morikawa, J.K. Norskov, Phys. Rev. Lett.
76 (1996) 2141.
[11] J.V. Barth, H. Brune, G. Ertl, R.J. Behm, Phys. Rev. B 42
(1990) 9307.
[12] C.J. Baddeley, R.M. Ormerod, A.W. Stephenson, R.M.
Lambert, J. Phys. Chem. 99 (1995) 5146.

258
D.V. Potapenko et al. / Surface Science 574 (2005) 244–258
[13] B. Voigtlander, G. Meyer, N.M. Amer, Phys. Rev. B 44
(1991) 10354.
[39] C.G. Zimmermann, M. Yeadon, K. Nordlund, J.M.
Gibson, R.S. Averback, Phys. Rev. Lett. 83 (1999) 1163.
[40] C. Nagl, E. Platzgummer, M. Schmid, P. Varga, S. Speller,
W. Heiland, Phys. Rev. Lett. 75 (1995) 2976.
[41] E.I. Altman, R.J. Colton, Surf. Sci. Lett. 304 (1994) L400.
[42] S.-L. Chang, J.-M. Wen, P.A. Thiel, S. Gunther, J.A.
Meyer, R.J. Behm, Phys. Rev. B 53 (1996) 13747.
[43] G.A. Attard, D.A. King, Surf. Sci. 188 (1987) 589.
[44] J. Takahashi, M. Sasaki, S. Suzuki, S. Sato, T. Abukawa,
S. Kono, J. Osterwalder, Surf. Sci. 304 (1994) 65.
[45] X.K. Shu, P. Jiang, J.G. Che, Surf. Sci. 545 (2003) 199,
and references therein.
[14] S. Padovani, F. Scheurer, J.P. Bucher, Europhys. Lett. 45
(1999) 327.
[15] D.D. Chambliss, R.J. Wilson, S. Chiang, Phys. Rev. Lett.
66 (1991) 1721.
[16] J.A. Rodriguez, J. Dvorak, T. Jirsak, J. Hrbek, Surf. Sci.
490 (2001) 315.
[17] Z. Song, T. Cai, J.A. Rodriguez, J. Hrbek, A.S.Y. Chan,
C.M. Friend, J. Phys. Chem. B107 (2003) 1036.
[18] J.A. Rodriguez, Surf. Sci. Rep. 24 (1996) 232.
[19] I. Chado, F. Scheurer, J.P. Bucher, Phys. Rev. B 64 (2001)
094410.
[46] D. Singh, H. Krakauer, Surf. Sci. 216 (1989) 303.
[47] F. Besenbacher, I. Choredorff, B.S. Clausen, B. Hammer,
A.M. Molenbroek, I. Stensgaard, Science 279 (1998) 1913.
[48] C.T. Campbell, Annu. Rev. Phys. Chem. 41 (1990) 775,
and references therein.
[20] T.B. Massalski, H. Okamoto, P.R. Subramanian, L.
Kacprzak (Eds.), Binary Alloy Phase Diagrams, Materials
Park, Ohio, ASM International, 1990.
[21] M.V. Bollinger, J.V. Lauritsen, K.W. Jacobsen, J.K.
Norskov, S. Helveg, F. Besenbacher, Phys. Rev. Lett. 87
(2001) 196803.
[49] J.R. Cerda, F. Soria, F.J. Palomares, P.L.d. Andres, Surf.
Sci. 269/270 (1992) 713.
[22] M.M. Biener, C.M. Friend, Surf. Sci. 559 (2004) L173.
[23] J.A. Rodriguez, J. Dvorak, T. Jirsak, Surf. Sci. 457 (2000)
L413.
[50] D. Wu, W.K. Lau, Z.Q. He, Y.J. Feng, M.S. Altman,
C.T. Chan, Phys. Rev. B62 (2000) 8366.
[51] A.M. Molenbroek, J.K. Norskov, B.S. Clausen, J. Phys.
Chem. B105 (2001) 5450.
[24] J.T. Roberts, C.M. Friend, Surf. Sci. 186 (1987) 201.
[25] H.H. Huang, C.S. Sreekanth, C.S. Seet, X. Jiang, G.Q.
Xu, Surf. Sci. 365 (1996) 769.
[52] W.S.M. Werner, Surf. Interface Anal. 31 (2001) 141.
[53] E. Stenhagen, S. Abrahamsson, F.W. McLafferty (Eds.),
Atlas of Mass Spectral Data, John Wiley, New York, 1969.
[54] F. Zaera, E.B. Kollin, J.L. Gland, Surf. Sci. 184 (1987) 75.
[55] D.G. Kelly, M. Salmeron, G.A. Somorjai, Surf. Sci. 175
(1986) 465.
[26] Z.C. Ying, W. Ho, J. Chem. Phys. 94 (1991) 5701.
[27] S.K. So, W. Ho, J. Chem. Phys. 95 (1991) 656.
[28] T.A. Germer, W. Ho, J. Chem. Phys. 89 (1988) 562.
[29] N.A. Beach, H.B. Gray, J. Am. Chem. Soc. 90 (1968) 5713.
[30] A. Zangwill, Physics at Surfaces, Cambridge University
Press, Cambridge, 1988.
[56] A.C. Liu, C.M. Friend, J. Am. Chem. Soc. 113 (1991) 820.
[57] J.T. Roberts, C.M. Friend, J. Am. Chem. Soc. 108 (1986)
7204.
[31] D.V. Potapenko, J.M. Horn, M.G. White, in preparation.
[32] P. Liu, J.A. Rodriguez, J.T. Muckerman, J. Hrbek, Surf.
Sci. 530 (2003) L313.
[58] B.C. Wiegand, P. Uvdal, C.M. Friend, Surf. Sci. 279
(1992) 105.
[33] L.Z. Mezey, J. Gibei, Jpn. J. Appl. Phys. 21 (1982) 1569.
[34] P. Liu, J.A. Rodriguez, J.T. Muckerman, J. Hrbek, Phys.
Rev. B67 (2003) 155416.
[59] G. Liu, J.A. Rodriguez, J. Dvorak, J. Hrbek, T. Jirsak,
Surf. Sci. 505 (2002) 295.
[60] A. Ulman, Self-Assembled Monolayer of Thiols, Academic
Press, New York, 1998.
[35] A.V. Ruban, H.L. Skriver, J.K. Norskov, Phys. Rev. B59
(1999) 15990.
[61] L.H. Dubois, B.R. Zegarski, R.G. Nuzzo, J. Chem. Phys.
98 (1993) 678.
[36] R.Q. Hwang, M.C. Bartelt, Chem. Rev. 97 (1997) 1063.
[37] A.K. Schmid, D. Altan, H. Itoh, B. Heinrich, T. Ichinok-
awa, J. Kirschner, Phys. Rev. B48 (1993) 2855.
[38] J.d.l. Figuera, J.E. Prieto, G. Kostka, S. Muller, C. Ocal,
R. Miranda, K. Heinz, Surf. Sci. 349 (1996) L139.
[62] V. Bondzie, S. Dixon-Warren, Y. Yu, J. Chem. Phys. 111
(1999) 10670.
[63] A.J. Leavitt, T.B. Beebe Jr., Surf. Sci. 314 (1994) 23.
[64] J. Gottschalck, B. Hammer, J. Chem. Phys. 116 (2002) 784.