Pd nanoparticles with highly defined structure on MgO as model catalysts: An FTIR study of the interaction with CO, O2, and H2 under ambient conditions

Article abstract of DOI:10.1016/j.jcat.2007.02.002

Model catalytic studies under ambient conditions require materials with well-defined structures for the establishment of clear relationships between structural and catalytic properties. In the present work, such a system was prepared by controlled decomposition of organometallic Pd precursors on a MgO support. The resulting Pd nanoparticles of highly defined morphology were characterized by high-resolution transmission electron microscopy and investigated with respect to their interaction with CO, O₂, and H₂ by IR spectroscopy (DRIFTS). Temperature-dependent CO adsorption experiments revealed the available adsorption sites on the Pd particles and indicated CO disproportionation on the Pd particles. We report in particular on the interplay between metal and support in these processes, which was analyzed by a thorough comparison of Pd/MgO with the pure MgO support. Investigations of the interaction with CO and H₂ suggest that Pd facilitates the formation of formates on MgO via dissociation of hydrogen on the Pd particles. Furthermore, CO oxidation was studied as a function of the CO/O₂ ratio in the temperature range 100-150 °C.

Full text of DOI:10.1016/j.jcat.2007.02.002

Journal of Catalysis 247 (2007) 145–154
www.elsevier.com/locate/jcat
Pd nanoparticles with highly defined structure on MgO as model catalysts:
An FTIR study of the interaction with CO, O2, and H2
under ambient conditions
H. Borchert a, B. Jürgens a, V. Zielasek a, G. Rupprechter b, S. Giorgio c, C.R. Henry c, M. Bäumer a,∗
a
University of Bremen, Institute of Applied and Physical Chemistry, 28359 Bremen, Germany
Vienna University of Technology, Institute of Materials Chemistry, 1210 Vienna, Austria
b
c
1
CRMCN-CNRS, Campus de Luminy, Case 913, 13288 Marseille Cédex 9, France
Received 15 September 2006; revised 26 January 2007; accepted 1 February 2007
Available online 6 March 2007
Abstract
Model catalytic studies under ambient conditions require materials with well-defined structures for the establishment of clear relationships
between structural and catalytic properties. In the present work, such a system was prepared by controlled decomposition of organometallic Pd
precursors on a MgO support. The resulting Pd nanoparticles of highly defined morphology were characterized by high-resolution transmission
electron microscopy and investigated with respect to their interaction with CO, O , and H by IR spectroscopy (DRIFTS). Temperature-dependent
2
2
CO adsorption experiments revealed the available adsorption sites on the Pd particles and indicated CO disproportionation on the Pd particles.
We report in particular on the interplay between metal and support in these processes, which was analyzed by a thorough comparison of Pd/MgO
with the pure MgO support. Investigations of the interaction with CO and H suggest that Pd facilitates the formation of formates on MgO via
2
dissociation of hydrogen on the Pd particles. Furthermore, CO oxidation was studied as a function of the CO/O ratio in the temperature range
2
◦
100–150 C.
© 2007 Elsevier Inc. All rights reserved.
Keywords: Palladium; Magnesium oxide; Carbon monoxide; IR spectroscopy; Transmission electron microscopy
1. Introduction
ever, the extent to which results obtained on model catalysts
under UHV conditions remain meaningful for real catalysts un-
der reaction conditions remains unclear [17–19]. For example,
investigations of CO adsorption on Pd surfaces revealed that
CO did not dissociate on Pd single-crystal surfaces [9], even
when the surfaces were roughened by an Ar-sputtering tech-
nique and when the experiments were carried out at mbar pres-
sures [20–22]. On real Pd catalysts, however, evidence of CO
dissociation was found in some cases [9,20,23,24]. Particle size
effects and metal–support interactions have been proposed to
affect the ability of Pd catalysts to dissociate CO [9,21,23,24],
but this topic remains under debate.
Elucidating clear relationships between structural and cat-
alytic properties is much more complicated for real catalysts
than in UHV studies, due to the usually more complex and less
defined structure of real catalysts and the complexity of real re-
action conditions. These difficulties promoted the preparation
Supported palladium particles have a wide range of appli-
cations as catalysts for technological processes, such as hy-
drogenation [1,2] and dehydrogenation reactions [3], metha-
nation [4], CO oxidation [5,6] and NO reduction [7,8], and
considerable efforts have been undertaken to gain more insight
into the reaction mechanisms. In some cases, detailed informa-
tion on relationships between structural and catalytic properties
could be obtained in model catalytic studies under ultra-high-
vacuum (UHV) conditions, in which single-crystal based sam-
ples with well-defined structures were exposed to small doses
of gas molecules in a highly controlled manner [9–16]. How-
*
Corresponding author.
E-mail address: mbaeumer@uni-bremen.de (M. Bäumer).
The CRMCN is associated to the Universities of Aix-Marseille II and III.
1
0021-9517/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2007.02.002

146
H. Borchert et al. / Journal of Catalysis 247 (2007) 145–154
of model catalysts by wet-chemical methods with well-defined
structural properties and their investigation under ambient con-
ditions; for example, Lear et al. recently studied CO adsorp-
tion on alumina-supported Pd particles prepared from different
palladium precursors [25,26]. Different surface structures ob-
tained from the various preparations were correlated with fre-
quency shifts and intensity changes in the IR spectra [25,26].
Bertarione et al. investigated the adsorption of CO [27] and
methanol [28] on alumina- and magnesia-supported Pd parti-
cles by IR spectroscopy and found that the adsorption behavior
of CO depended on the support material as well as the na-
ture of the Pd precursor. When using Pd/MgO prepared from
a [Pd(NH3)4](NO3)2 precursor, IR bands due to carbonates
and bicarbonates on the support were observed in addition to
bands corresponding to CO adsorption on Pd [27]; however,
when using alumina as support or preparing Pd/MgO from a
palladium chloride precursor, bands corresponding to carbon-
ates and bicarbonates were absent [27]. Moreover, a detailed IR
spectroscopic study of methanol adsorption allowed the authors
to propose a mechanism of methanol decomposition, based on
the observation of adsorbed CO forming on Pd after methanol
exposure and annealing to ∼520 K [28].
Samples were pressed into pellets and studied in a reaction
cell equipped with a controlled gas supply system, heating
unit, and photometric detector (Hartmann & Braun URAS 10E)
for CO and CO2 gas analysis at the exit side of the cell. Be-
fore measurements, the samples were pretreated in the reaction
cell under continuous gas flow of ∼50 %vol H2 in Ar for 1
h at 400 ◦C, then cooled under continuous Ar flow to the re-
quired temperature. Spectra were recorded with a resolution
of ꢀ8 cm−1. All spectra shown are referenced to background
spectra recorded under continuous Ar or N2 flow after pretreat-
ment.
3. Results and discussion
3.1. Structural characterization by transmission electron
microscopy
Structural characterization of the samples was done by TEM.
Fig. 1a shows an overview TEM image of the Pd/MgO sample,
clearly showing individual Pd nanoparticles on cubic MgO sup-
port particles. The average diameter of the Pd nanoparticles is
about 6 nm, as indicated by the size histogram in Fig. 1b. High-
resolution images (Figs. 2a and 2b) show the particles growing
Thorough investigations of chemically prepared Pd model
catalysts remain rare; detailed studies of such systems are
needed. Based on the controlled decomposition of organometal-
lic precursors on magnesium oxide, we were able to prepare Pd
nanoparticles on MgO that have well-defined properties with
respect to crystallite size and morphology. In the present work,
we present a systematic investigation of the interaction of CO,
O2, and H2 with this well-defined Pd/MgO system and consider
in particular the metal–support interplay. For this purpose, we
combined high-resolution TEM measurements with IR spec-
troscopic investigations by DRIFTS (diffuse reflection infrared
Fourier transform spectroscopy).
2. Experimental
Palladium nanoparticles on magnesium oxide were prepared
by wet impregnation of MgO cubes (Fluka, specific surface
30 m2/g) with palladium acetylacetonate [Pd(acac)2] dissolved
in acetonitrile. The solution was stirred, the solvent was evap-
orated, and the sample was dried at 80 ◦C for 3 h. Then the
impregnated MgO powder was annealed in vacuum to 400 ◦C
and reduced at the same temperature in H2 at 400 mbar for
3 h. Using this technique, particles of different size and with
different number densities were prepared depending on the pre-
cursor concentration [29]. For particle sizes <8 nm, the number
density is equal to ∼103–104 clusters/µm2 and the size dis-
tribution is narrow. Previous work demonstrated that particles
>10 nm obtained by the same technique are less well defined
with respect to size and morphology [29]. In the present case,
the preparation was conducted so as to obtain ∼6-nm parti-
cles.
TEM investigations were performed with a Jeol 3010 trans-
mission electron microscope operated at 300 kV. IR spectro-
scopic experiments were performed in diffuse reflection geom-
etry (DRIFTS) with a Biorad FTIR spectrometer (FTS 60A).
Fig. 1. (a) Overview TEM image of the Pd nanoparticles on MgO. (b) His-
togram of the particle size distribution.

H. Borchert et al. / Journal of Catalysis 247 (2007) 145–154
147
2
Fig. 3. IR spectra of (a) a reference MgO sample (Acros Organics, 16 m /g)
2
and (b) the MgO used as support in this work (Fluka, 30 m /g) after exposure
◦
to a continuous gas flow of 0.1 % CO in N at 150 C as a function of expo-
vol
2
sure time. The inset in (b) shows the O–H stretching region of the background
spectrum recorded before CO exposure.
3.2. CO–MgO interaction
Fig. 2. (a) High-resolution TEM image of an individual Pd nanoparticle on
MgO. The particle is oriented [001] in the electron beam. The particles posses
a well-defined shape, limited by (001) and (111) faces which is illustrated by the
schematic drawing in part (c) of the figure. (b) High-resolution TEM image of a
Pd nanoparticle on MgO. The particle and the substrate are seen along the [110]
direction. The accommodation with the substrate is clearly seen at the edges.
The schematic drawing of the projection is shown in (d). (c) Schematic draw-
ing of a truncated octahedron projected along a [001] direction. (d) Schematic
drawing of a truncated octahedron projected along a [110] direction.
As a first step of the investigation of magnesia-supported
Pd nanoparticles, we studied the interaction of CO with pure
MgO. Apart from the MgO used as a support in the present
work, an additional reference compound (Acros Organics, p.a.
quality, 16 m2/g) was also studied. Fig. 3 shows IR spectra
for the two MgO samples. Spectra were recorded in a continu-
ous gas flow of ∼0.1 %vol CO in N2 (total pressure: 1 bar) at
150 ◦C at exposure times of up to 3 h. In both cases, absorp-
tion bands developed in the O–H stretching region, the C–H
stretching region, and the region below 1700 cm−1, that is, well
below the carbonyl-stretching region. The absence of bands in
the carbonyl-stretching region is in agreement with studies of
CO adsorption on MgO(001) revealing that molecular CO has
a very low enthalpy of adsorption on MgO and can be seen
only below ∼100 K under UHV conditions [30]. The contin-
uous development of bands in the O–H and C–H stretching
regions implies that there must be a source of hydrogen. Al-
though the residual water content of the gases used was on the
order of only a few ppm, this is apparently sufficient to supply
water to the samples on the time scale of 3 h (details discussed
below).
in defined epitaxial relations on the support; the (001) face of
the Pd particles lies parallel to the (001) face of the MgO sup-
port, with the [100] axes of the Pd parallel to the [100] axes of
the MgO lattice. Furthermore, the Pd nanoparticles have well-
defined facets. The shape can be described as an octahedron
with (111) facets with the corners truncated by (100) facets (cf.
Figs. 2c and 2d). The high-resolution images and their power
spectrum show perfect accommodation between the metal and
oxide lattices. Thus, the Pd nanoparticles prepared on MgO
in this work exhibit numerous well-defined structural features,
making this system suitable for model catalytic studies under
ambient conditions.

148
H. Borchert et al. / Journal of Catalysis 247 (2007) 145–154
Table 1
◦
Assignment of IR bands observed after CO adsorption on MgO at 150 C
−1
IR band, wavenumber [cm ],
this work
Literature
Assignment
(lit.)
References
1215
1293
1220–1270
1275, 1307
1273, 1329
1220–1270
1250–1270
1348, 1385, 1390
1340, 1390
1360
1370
1370
1300–1370
1472
1535
1470–1530
1605
1598
1599
1585, 1620
1670
1626, 1659
1620–1670
1530–1620
2728, 2820, 2931
2820
Bidentate carbonate, bonded to 2 metal sites
Bidentate carbonate
Bidentate carbonate
Bidentate carbonate, bonded to 2 metal sites
Bidentate carbonate, bonded to 1 metal site
Formate
Formate
Formate
Formate
Monodentate carbonate
Monodentate carbonate
Carbonite
Monodentate carbonate
Monodentate carbonate
Formate
Formate
Formate
Formate
Bidentate carbonate
Bidentate carbonate
Bidentate carbonate, bonded to 2 metal sites
Bidentate carbonate, bonded to 1 metal site
Formate
Formate
Formate
Formate
[34]
[37]
[38]
[34]
[34]
[32]
[34]
[33]
[31]
[37]
[34]
[31]
[37]
[34]
[31]
[32]
[33]
[34]
[37]
[38]
[34]
[34]
[32]
[31]
[33]
[34]
1356, 1382
∼1510
1603, 1618
1650
2720, 2801, 2829, 2950
2730, 2830, 2957
2760, 2840
∼3400
H O, multiple bonded
Mg(OH)
2
2
3670
The frequencies of the bands, as well as their assignments,
are summarized in Table 1. Both magnesia samples showed
bands at ∼1356, ∼1382, ∼1603, ∼1618, ∼2720, ∼2801,
∼2829, and ∼2950 cm−1. All of these bands can clearly be
assigned to formate species [31–34]. The splitting of the band
at around ∼1600 cm−1 indicates the presence of two slightly
different types of formate species, probably due to the exis-
tence of two types of hydroxyl groups on MgO, as discussed by
Davidov [34]. After exposure of more than 1 h, an additional
broad band appeared at around ∼1510 cm−1 that could be due
to carbonite ions (CO22−) resulting from the interaction of CO
with coordinatively unsaturated O2− ions at the MgO surface
[31,35,36], or to monodentate carbonate species [35,37]. The
band at ∼3670 cm−1 and the broad band at ∼3400 cm−1 are
probably due to Mg(OH)2 configurations and adsorbed water
molecules, respectively. Note that isolated hydroxyl groups on
MgO were reported to give rise to a rather sharp band at around
3750 cm−1 [33,34]. Such bands were visible in the background
spectra recorded before exposure to CO (Fig. 3b, inset), but ad-
ditional isolated hydroxyl groups do not seem to develop on
exposure to CO, as indicated by the absence of a band at 3750
cm−1 in the spectra referenced to the background (see Fig. 3).
In addition to the bands discussed so far, two resonances
at 1650 and ∼1293 cm−1 were more pronounced in the MgO
sample used as the support material in this work. These bands
indicate the formation of bidentate carbonate species, presum-
ably bonded to two Mg2+ cations [34,37,38]. The formation
of carbonate species probably is related to residual oxygen in
the system. Smart et al. studied CO adsorption on MgO at
different pressures and observed carbonate formation even at
1.3 × 10−5 mbar [37]. With increased exposure time, the rela-
tive intensity of the bands corresponding to bidentate carbon-
ates decreased, and mainly bands corresponding to formates
developed at the considered temperature of 150 ◦C.
Another noteworthy temporal evolution in the IR spectra was
seen. In the first ∼30 min, formate species rapidly appeared
in the spectra. Afterwards, the band possibly corresponding to
carbonite ions began to develop while the formate formation
slowed. In a study of CO adsorption on CaO and MgO, Babaeva
et al. found that the development of formates on CaO proceeded
at 300 K via the formation of carbonite ions as an intermediate
species [31]. Carbonite ions have been found to react to for-
mates in the presence of water vapor [31]. The authors found
many similarities between CaO and MgO and observed the for-

H. Borchert et al. / Journal of Catalysis 247 (2007) 145–154
149
◦
Fig. 4. Possible, schematic mechanism of formate formation on MgO at 150 C.
mation of carbonite ions also on MgO [31]. Based on these
investigations, a reaction mechanism can be suggested to ex-
plain the temporal behavior observed in the present study (see
Fig. 4). In a first step, CO adsorbs on a coordinatively unsatu-
rated O2− site on the MgO surface and forms a carbonite ion.
Next, a water molecule adsorbs on MgO and dissociates, result-
ing in the formation of two OH groups at the surface [39]. In a
third step, the carbonite ion then reacts with a neighboring OH
group to formate. In principle, step 3 may occur with the OH
groups generated during the experiment by residual water in
the gas feed, as well as with the OH groups already present on
the MgO surface after H2 pretreatment at 400 ◦C. If the car-
bonite ions were highly reactive with respect to neighboring
OH groups, this would explain why the (rapid) formation of
formates is observed only initially. At a later stage of the exper-
iment, suitable sites with carbonite ions having an OH group
in their vicinity would become increasingly rare, thereby slow-
ing this reaction pathway and leading to a build-up of carbonite
ions.
Finally, we should mention that the reactive adsorption of
CO on MgO is also temperature-dependent. At 30 ◦C, for-
mates were not generated under these conditions (data not
shown). Exposure to a continuous gas flow of CO in Ar for
several minutes led to IR bands at ∼1300, ∼1380, ∼1530, and
∼1650 cm−1, indicating the formation of mono- and bidentate
carbonate species [37,38]. Davidov also reported that temper-
atures above 100 ◦C are required for formate production on
MgO [34].
Fig. 5. IR spectra of Pd nanoparticles on MgO after exposure to CO. Initially,
the sample was exposed to a continuous gas flow of 0.1 % CO in Ar for
vol
10 min at room temperature (bottom spectrum). The sample was then heated
under pure Ar gas flow, kept for 10 min at the indicated temperature and cooled
down again to RT before the next spectrum was measured.
on Al2O3 and MgO, Bertarione et al. assigned bands at
∼2025–2076 cm−1 to CO linearly bonded to Pd atoms on
edges, steps, and terraces of (100) and (111) facets [27]. A more
detailed assignment based on UHV studies of CO adsorption on
Pd particles grown on a thin alumina film also has been reported
[18,19,41–43]. Pd particles of various sizes and morphologies
could be prepared on this support, depending on the growth
temperature. At 300 K, large (7–10 nm), well-ordered Pd crys-
tallites exposing mainly (111) facets were obtained, whereas at
90 K, small (mean size ∼2 nm), disordered Pd particles resulted
from physical vapor deposition. For the well-faceted particles,
a sharp band at ∼2100 cm−1 was observed at saturation cov-
erage and 90 K and was ascribed to terminally bonded CO on
(111) terraces. For the small, disordered particles, however, ad-
sorption at 90 K led to a rather broad band at 2050–2130 cm−1
.
This was attributed to terminal CO adsorption on edge and cor-
ner atoms, because well-developed facets were basically absent
on the small particles.
3.3. Interaction of CO and Pd/MgO: CO adsorption sites
The adsorption of CO on Pd nanoparticles on MgO was stud-
ied at 30 ◦C by exposing samples to a continuous gas flow of
∼0.1 %vol CO in Ar for 10 min. A corresponding IR spectrum
is shown in Fig. 5 (bottom spectrum). Three absorption bands
are visible at ∼1904, ∼1983, and ∼2063 cm−1 that can be
assigned, from low to high wavenumbers, to CO on threefold-
hollow sites of Pd(111) facets, to bridge-bonded CO on various
Pd sites, and to CO linearly bonded to Pd atoms, respectively
[18,19,25–27,40–43].
Apart from surface roughness, the population of linear sites
(on-top CO) is also dependent on temperature and CO pressure.
Dellwig et al. studied the CO pressure dependence systemati-
cally at 200 K using sum-frequency generation spectroscopy
[17]. Whereas at 10−7 mbar only bridge-bonded CO was ob-
served for well-faceted 6-nm Pd particles, at higher pressures
(10−3–200 mbar), bands corresponding to linearly bonded
CO on (111) surfaces also occurred [17]. Up to ∼1 mbar,
those bands were comparably weak, but at 200 mbar they
were even more intense than those corresponding to bridge-
In particular, the band corresponding to linearly bonded
CO merits discussion. In a study of Pd particles supported

150
H. Borchert et al. / Journal of Catalysis 247 (2007) 145–154
bonded CO [17]. A similar pressure/coverage dependence was
also reported for Pd(111) single-crystal surfaces [22,44,45].
In contrast, for the disordered particles, the band centered at
2100 cm−1 was observed over the whole pressure regime [17].
Lear et al. took a further step and investigated the morpho-
logy-dependent occupation of different sites on alumina-sup-
ported Pd particles prepared from various precursors [25].
In the case of tetraaminepalladium(II) tetraazidopalladate(II)
[Pd(azide)], Pd particles with an average diameter of about
∼15 nm and a well-defined crystallite morphology were ob-
tained [26]. The crystallites exposed mainly (111) and (100)
planes and were assumed to be rather defect-free [25,26]. In
this case, the IR bands of terminally bonded CO were extremely
weak [25]. Because the experiments were done by pulsing CO
onto the sample, not by applying a static CO pressure, it was
concluded that CO does not populate linear sites on the terraces
at low partial pressure, in agreement with the findings of Dell-
wig et al. [17]. In addition, the absence of surface roughness
limits the occurrence of on-top CO.
Using Pd(acac)2 as precursor (i.e., the same precursor as in
the present study), Lear et al. obtained particles much smaller
than those obtained from the Pd(azide) precursor and those used
in the present study [25]. These particles could not be analyzed
in detail by TEM. For these catalysts, an intense and sharp
band was observed at ∼2080 cm−1 that disappeared on heat-
ing the sample to about 150 ◦C. Because of the small size of the
particles (∼1.5 nm, determined indirectly from chemisorption
measurements), the authors assigned this peak to CO terminally
bonded to corner atoms [25].
In contrast, other precursors also yielding small disor-
dered Pd particles typically resulted in two IR bands with
different desorption behaviors [25]. Bands in the range of
∼2077–2090 cm−1, disappearing at 150 ◦C, were assigned
to CO terminally bonded to corner atoms, in analogy to the
Pd(acac)2 system [25]. The other band, detected in the range
of ∼2050–2064 cm−1 and exhibiting a much higher desorp-
tion temperature of ∼300◦C, was assigned to on top CO on Pd
atoms at (111)/(111) and (111)/(100) edges [25].
However, according to theoretical calculations, the chemi-
sorption energy of CO linearly bonded to different Pd sites is
expected to increase with decreasing coordination of the ad-
sorption site [9,46]. Thus, CO bonded to Pd edge atoms would
be expected to desorb at lower temperature than CO bonded
to corner atoms, which is in disagreement with the assignment
proposed by Lear et al.
shape, see [47].) Therefore, it is reasonable to assign the on-top
species observed in the present study to CO linearly bonded to
Pd edge atoms. Note that we cannot fully exclude on-top ad-
sorption on (111) facets, but these species typically have higher
frequencies [44], and are more likely to be responsible for the
very small shoulder which is discernible around ∼2100 cm−1
in the IR spectra after CO adsorption at room temperature (bot-
tom spectrum in Fig. 5).
To determine the desorption temperature of the on-top
species, we performed annealing experiments. After initial CO
adsorption at room temperature (bottom spectrum in Fig. 5), the
sample was heated under pure Ar gas flow, kept for 10 min at
the respective temperature, and then cooled to room tempera-
ture for IR spectroscopy (the other spectra shown in Fig. 5). The
band assigned to CO linearly bonded to Pd edge atoms nearly
fully disappeared after annealing to ∼150◦C, whereas the other
bands remained present. In analogy to the findings of Lear et al.,
the annealing experiments led to shifts of all adsorption bands
of CO on Pd sites to lower wavenumbers, with a more pro-
nounced effect for the bridge-bonded and threefold-bonded CO
species than for the linearly bonded species. These shifts can
be explained by reduced dipole coupling at decreased CO cov-
erage. The desorption temperature of the band clearly assigned
to CO on Pd edges in our work is very similar to the desorp-
tion temperature of the band observed at ∼2077–2090 cm−1 by
Lear et al. Therefore, we propose that the peak at higher fre-
quency in the study of Lear et al. is also caused by edge atoms,
whereas the second band at lower frequency arises from defect
sites with even lower coordination. This assignment is also in
agreement with the lower frequency of the latter species, be-
cause the lower-coordinated Pd atoms are expected to have a
higher capability for electron back donation to the CO 2π∗ or-
bitals, resulting in more red-shifted stretching frequencies [9].
3.4. Interaction of CO and Pd/MgO: Reaction
In addition to the bands discussed in the previous section, the
spectra of the Pd/MgO system also contain bands in the region
below 1700 cm−1, which indicate reactive adsorption of CO on
MgO, as in the case of the pure MgO support. At room tem-
perature, bands were observed at ∼1300, ∼1375, ∼1500, and
1657 cm−1 (see Fig. 5). As discussed in Section 3.2, bands at
∼1300 and 1657 cm−1 can be assigned to bidentate carbonate
species [34,37,38]. The bands at ∼1375 and ∼1500 cm−1, on
the other hand, indicate monodentate carbonate species [34].
Bands indicative of formate formation were not observed on
CO adsorption on the supported Pd particles at room tempera-
ture, in agreement with the absence of formates on pure MgO
below 100 ◦C (see Section 3.2).
In addition to the bands discussed so far, a prominent peak
at 1681 cm−1 was discernible in the spectra, accompanied by
a band at 1215 cm−1. These bands are definitely absent for
the unloaded MgO support and can be attributed to carbonate
[32], probably to a bidentate species bonded to two metal sites
[34,37].
Our present results help clarify the situation. Because the CO
partial pressure was around 1 mbar in our experiments, linear
CO adsorption on Pd atoms of (111) surfaces presumably plays
only a minor role, as reported by Dellwig et al. [17] and Lear
et al. [25]. The high-resolution TEM images provide evidence
that our palladium particles are well-shaped and have mainly
(111) and (100) facets (cf. Fig. 2); therefore, considerable sur-
face roughness that could induce CO bonding to on-top sites is
absent. Due to the average particle size of ∼6 nm, a consider-
able fraction (15%) of Pd atoms is located at the (111)/(111)
and (111)/(100) edges. (For a general discussion of the num-
ber of various surface sites as a function of crystallite size and
Insight into the origin of this species is obtained by consid-
ering the temperature-dependent experiments shown in Fig. 6.

H. Borchert et al. / Journal of Catalysis 247 (2007) 145–154
151
dissociation on step sites observed [48]. For supported Pd cat-
alysts under ambient conditions, some authors have reported
CO dissociation, whereas others have not. Bertarione et al. con-
ducted an IR spectroscopy study of Pd particles on MgO pre-
pared from palladium chloride (particle size ∼15 3.5 nm)
and from [Pd(NH3)4](NO3)2 precursors (mean particle size
∼11 4.5 nm) [27]. In the case of [Pd(NH3)4](NO3)2, they ob-
served carbonate and bicarbonate species after CO adsorption at
room temperature, and took the appearance of the correspond-
ing IR bands as an indication of disproportionation of CO on
Pd [27]. In the case of particles prepared from the chloride pre-
cursor, they found no indications for CO disproportionation,
but could not determine whether this was due to the larger Pd
particle size or to the surface basicity of MgO, which was re-
duced by the presence of Cl− ions [27]. Prinetto et al. studied
CO adsorption at room temperature on dispersed Pd on MgO,
Al2O3, and Mg(Al)O mixed oxides [24]. For highly dispersed
systems (particle size <2 nm), the authors observed CO dis-
proportionation only for MgO-supported Pd, demonstrating the
role of metal–support interactions [24]. Furthermore, the dis-
persion was shown to be of importance. Prinetto et al. [24]
did not observe CO disproportionation for larger Pd particles
(∼3–4 nm) on MgO. This is in contrast to Bertarione et al.’s
observation of CO disproportionation on much larger Pd parti-
cles on MgO (see above).
Fig. 6. IR spectra of Pd nanoparticles on MgO after exposure to a continu-
ous gas flow of 0.1 % CO in Ar at different temperatures. All spectra were
vol
recorded after 5 min of exposure.
Our work, which was also extended to higher temperatures,
can shine some more light on the phenomenon of CO dispro-
portionation on magnesia-supported Pd particles. As shown by
the temperature-dependent measurements, disproportionation
of CO and formation of carbonate species on MgO become
less pronounced at higher temperature. At room temperature,
large amounts of carbonate develop rapidly compared with pure
MgO. At 200 ◦C, almost no carbonates are observed, in agree-
ment with the shift of the equilibrium of the Boudouard reaction
to the CO side. This observation shows that CO2 resulting from
the disproportionation reaction is bound reversibly on MgO,
presumably in the neighborhood of the particles, because oth-
erwise, intense carbonate bands would also be expected in the
high-temperature regime. We note that at 200 ◦C, the formation
of carbonates with subsequent decomposition can be excluded.
Carbonate decomposition on MgO would require temperatures
above ∼400◦C [38,49].
The IR spectra were recorded at room temperature and 100,
150, and 200 ◦C for Pd/MgO under a continuous gas flow of
0.1 %vol CO in Ar, each after 5 min of exposure. With increas-
ing temperature, the bands corresponding to CO adsorption
on Pd shift to lower wavenumbers. Furthermore, the relative
intensities of the bands corresponding to the different adsorp-
tion sites of CO on Pd change in favor of the threefold-hollow
sites, in agreement with the order of the enthalpy of adsorption.
Along with the temperature-dependent adsorption of CO on dif-
ferent Pd sites, a clear temperature dependence of the peaks
at 1681 and 1215 cm−1 is evident. These carbonate bands are
much more pronounced at room temperature than at elevated
temperatures.
This finding clearly indicates that Pd gives rise to a tempe-
rature-dependent formation of a new carbonate species on the
MgO support on exposure to CO. A possible explanation for
this is the disproportionation of CO on Pd,
Moreover, we want to mention that in some cases, prolonged
CO exposure (∼30 min) led to slight temporal changes in the
spectra (not shown). A small increase of the carbonate band was
accompanied by changes in the relative intensities of the bands
corresponding to CO adsorption on the different Pd sites. In
agreement with other investigations [50,51], these correlations
would be compatible with a mechanism in which elemental car-
bon formed on CO disproportionation blocks threefold-hollow
sites on the Pd surface. Nevertheless, further experiments are
needed to support this interpretation.
2CO → CO2 + C,
(1)
which can be considered the net reaction of CO dissociation
and subsequent reaction of the oxygen atom with another CO
molecule to CO2 [20]. Whereas CO2 can react with coordina-
tively unsaturated O2− ions to carbonate species on MgO, the
elemental C atoms probably remain on the Pd particles. Be-
cause the overall reaction is exothermic, these processes should
become less pronounced at higher temperatures, as indeed was
observed.
As mentioned in Section 1, CO dissociation (disproportion-
ation) on Pd is a topic that remains under debate. There is
agreement that CO does usually not dissociate on Pd single-
crystal surfaces [9,20–22]. Only in the case of Pd(112) was CO
Another noteworthy result of the temperature-dependent
measurements concerns the formation of formate species. Al-
though no indications for formates were found up to 100
◦C,
a small shoulder at ∼1604 cm−1 indicative of formates ap-
peared in the spectrum at 150 ◦C (see Fig. 6). This finding

152
H. Borchert et al. / Journal of Catalysis 247 (2007) 145–154
was confirmed by the appearance of small bands in the C–H
stretching region that were not observed up to 100 ◦C. For-
mate is probably generated independently, as on pure MgO at
150 ◦C. However, compared with pure MgO, the bands corre-
sponding to formates were weak in the case of the supported
Pd particles and remained weak even after prolonged CO ex-
posure (∼30 min). This shows that formate formation on MgO
was largely suppressed by the presence of the Pd nanoparticles.
A possible reason for this could be that Pd blocks sites favorable
for the formation of formates or catalyzes their decomposition.
Blocking phenomena were also observed for Cu deposited on
ZnO [52].
3.5. Interaction of CO and O2 with Pd/MgO
Next, we used the Pd/MgO system as a model catalyst for
CO oxidation. Fig. 7 depicts an experiment in which the sample
was exposed to a 1:1 vol% CO/O2 mixture in the reaction cell
of the IR spectrometer. The first spectrum (designated “a”) was
recorded at 150 ◦C on exposure to CO only and exhibited the
characteristics discussed earlier. When oxygen was added to the
gas feed, the bands corresponding to CO adsorption on Pd sites
disappeared almost immediately (spectrum “b”). Furthermore,
an immediate increase of the bands corresponding to carbon-
ate species on the MgO support is visible. Fig. 7b shows the
CO and CO2 concentrations measured at the exit side of the gas
cell. Obviously, CO is oxidized to CO2 at 150 ◦C. We note that
absolute conversion rates cannot be determined in this experi-
ment, because the construction of the gas cell is not suitable for
this purpose. (A considerable fraction of the feed gas can pass
the cell without coming in contact with the catalyst.) Gaseous
CO2 is also visible in the IR spectra (the band at ∼2348 cm−1.)
Whereas the Pd particles were definitely metallic after the
hydrogen pretreatment and during CO adsorption (reducing
conditions), the state of Pd during the CO oxidation experi-
ments is not a priori clear. In general, the oxidation state can
have an important influence on the activity and selectivity in
oxidation reactions [53–55], and recent studies have shown that
different phases of palladium oxide can form under oxidative
reaction conditions [54–56]. In studies of Pd-based model cat-
alysts prepared under UHV conditions, dissociative chemisorp-
tion of oxygen was found up to ∼450 K, whereas the formation
of palladium oxide phases required temperatures above ∼500 K
[56,57]. The activity for CO oxidation was found to decrease
with increasing coverage of the Pd surface by oxides for such
model systems [54,55].
Fig. 7. (a) IR spectra of Pd nanoparticles on MgO after exposure to a continuous
gas flow of CO/O mixtures in Ar at different temperatures. Spectra plotted as
2
solid lines correspond to exposure to ∼2 % CO and ∼2 % O . Spectra
vol
vol
2
plotted as dotted lines correspond to exposure to CO only (O switched off).
2
(b) Plot of the CO and CO concentration measured by the photometric detector
2
as a function of time. The applied temperatures are also plotted, and the labels
“a”–“g” indicate in which period the IR spectra in part (a) of the figure were
measured. The spectrum “h” was measured at the end of the experiment, i.e.,
Spectrum “c” in Fig. 7a, recorded directly after switch-
ing off O2, exhibits only minor differences from the spectrum
recorded before exposure to oxygen (spectrum “a”). The fre-
quencies of the bands corresponding to CO adsorbed on Pd
are shifted slightly to higher wavenumbers, and the intensity
of the bands corresponding to linearly and bridge-bonded CO
are increased slightly at the expense of the band corresponding
to threefold-bonded CO. The integrated intensity of all three
bands together decreased by ∼6%. These changes provide an
indication that the surface might be partly oxygen-covered after
exposure to the CO/O2 atmosphere at 150 ◦C. But because the
after switching off O . (c) IR spectra of Pd nanoparticles on MgO after exposure
2
◦
to a continuous gas flow of different CO/O mixtures in Ar at 150 C.
2
loss of intensity of the bands corresponding to CO on metallic
Pd sites is only a few percent, and because bands corresponding
to CO adsorbed on Pd2+ cations remain absent, the degree of
oxidation is small. Thus, the postreaction IR data show that the
Pd model catalyst is almost entirely in the metallic state after

H. Borchert et al. / Journal of Catalysis 247 (2007) 145–154
153
exposure to CO/O2 at 150 ◦C, which is in agreement with the
findings that palladium oxide phases do not form on Pd model
catalysts prepared under UHV conditions in this temperature
regime [54–57]. Because the formation of oxide phases may
also depend on pressure and structural properties, we cannot
fully exclude the possibility that palladium oxide phases are re-
versibly formed in the CO/O2 atmosphere during reaction at
150 ◦C in our case.
After switching O2 on again, we lowered the temperature
to find the lowest temperature at which CO oxidation can be
observed for this system. Down to 120 ◦C we see no differ-
ences by either IR spectroscopy or analysis of the gas phase.
At 100 ◦C, however, the conversion drops, as revealed by pho-
tometric detection. Simultaneously, the IR bands corresponding
to CO adsorption on Pd reappear, but do not show the same in-
tensity as in the absence of oxygen (compare spectra “g” and
“h”). However, there do not seem to be large differences in the
relative intensities of the bands corresponding to different ad-
sorption sites compared with spectra recorded on exposure to
CO only.
Experiments with different gas-phase compositions were
also performed. Fig. 7c shows IR spectra recorded at 150 ◦C on
exposure to CO/O2 gas mixtures with compositions between
1:2 and 8:1. In the case of CO excess (ratio > 2:1), adsorbed
CO species are observed during the ongoing CO oxidation re-
action, whereas no adsorbed CO is detectable for lower ratios.
This can be explained by the usually sharp transition from a
oxygen-saturated system to a CO-saturated system typical for
the CO oxidation reaction also showing kinetically two distinct
reaction regimes (oxygen-rich and CO-rich) [58].
Fig. 8. IR spectra of (a) MgO and (b) Pd nanoparticles on MgO after exposure
to a continuous gas flow of CO/H mixtures in Ar at different temperatures. All
2
spectra were recorded after relatively short exposure times of a few minutes.
3.6. Interaction of CO and H2 with MgO and Pd/MgO
formates compared with the situation without H2 in the feed
(cf. Fig. 6). This finding suggests that above ∼200◦C, atomic
hydrogen formed by dissociation on Pd can serve as a source of
atomic hydrogen for formate generation on the MgO support.
Because hydrogen is known to dissociate on Pd even at tem-
peratures as low as 20 K, the rate-limiting step is probably a
spillover process of atomic hydrogen from the Pd particles to
the MgO support.
Finally, we studied the Pd particles and pristine magnesia
support with respect to their interaction with carbon monox-
ide and hydrogen. Vibrational studies of CO/H on Pd(111) and
Pd particles on alumina under both UHV and ambient pres-
sure conditions, which may serve as references were reported
in [59,60]. Fig. 8a shows IR spectra for pure MgO exposed to
CO and H2 in Ar. As discussed in Section 3.2, in the absence of
H2 mainly formate species develop at 150 ◦C. At this tempera-
ture, the addition of hydrogen to the feed gas does not result in
a significant increase of the corresponding IR bands. Thus, and
not unexpectedly, molecular hydrogen obviously cannot act as
a hydrogen source for formate formation on MgO at 150 ◦C.
Raising the temperature to 250 ◦C leads to a slightly increased
intensity of the IR bands and to some changes in the relative
intensities of the bands in the C–H stretching region.
4. Conclusion
In the present work, we prepared Pd nanoparticles on MgO
by controlled decomposition of organometallic precursors.
High-resolution TEM measurements revealed that these par-
ticles have well-defined structural properties that make them
suitable as a model system for fundamental catalytic studies un-
der ambient conditions. We investigated their interaction with
CO, O2, and H2 and performed IR spectroscopic experiments.
In the regime of terminally bonded CO on the Pd particles,
a peak at ∼2060–2070 cm−1 could be assigned to adsorption
at crystallite edge sites. CO adsorption was accompanied by
disproportionation of CO on the Pd particles. Whereas ele-
mental carbon seems to remain on the Pd particles, possibly
blocking threefold-hollow sites, CO2 can react further with
coordinatively unsaturated O2− ions of the MgO support to car-
For Pd particles on MgO at 150 ◦C, addition of hydrogen to
the feed gas had no pronounced effect (see Fig. 8b), in agree-
ment with a study by Morkel et al. [60]. Spectra before and after
addition of H2 revealed only a small band around ∼1605 cm−1
indicative of formates, because disproportionation of CO and
carbonate formation seem to be the predominant processes,
as discussed before. Obviously, the hydrogen supply does not
favor generation of formates at 150 ◦C. However, higher tem-
peratures increase the intensity of the bands corresponding to

154
H. Borchert et al. / Journal of Catalysis 247 (2007) 145–154
bonate species. Systematic studies at temperatures up to 200 ◦C
revealed that this phenomenon is less pronounced at higher tem-
perature. Studies of the interaction with CO and O2 showed that
the model catalysts were active for CO oxidation at tempera-
tures above ∼100◦C, and investigations of the interaction with
CO and H2 suggested that Pd facilitated the formation of for-
mates on MgO at temperatures above ∼200◦C via dissociation
of hydrogen on the Pd particles.
[27] S. Bertarione, D. Scarano, A. Zecchina, V. Johanek, J. Hoffmann,
S. Schauermann, M.M. Frank, J. Libuda, G. Rupprechter, H.-J. Freund,
J. Phys. Chem. B 108 (2004) 3603.
[28] S. Bertarione, D. Scarano, A. Zecchina, V. Johanek, J. Hoffmann, S.
Schauermann, J. Libuda, G. Rupprechter, H.-J. Freund, J. Catal. 223
(2004) 64.
[29] S. Giorgio, C. Chapon, C.R. Henry, Langmuir 13 (1997) 2279.
[30] M. Sterrer, T. Risse, H.-J. Freund, Surf. Sci. 596 (2005) 222.
[31] M.A. Babaeva, D.S. Bystrov, A.Yu. Kovalgin, A.A. Tsyganenko, J. Catal.
123 (1990) 396.
[32] D.G. Rethwisch, J.A. Dumesic, Langmuir 2 (1986) 73.
[33] K. Teramura, T. Tanaka, H. Ishikawa, Y. Kohno, T. Funabiki, J. Phys.
Chem. B 108 (2004) 346.
Acknowledgments
[34] A.A. Davydov, Infrared Spectroscopy of Adsorbed Species on the Surface
of Transition Metal Oxides, Wiley, Chichester, 1990.
[35] K.I. Hadjiivanov, G.N. Vayssilov, Adv. Catal. 47 (2002) 307.
[36] M. Bailes, F.S. Stone, Mater. Chem. Phys. 29 (1991) 489.
[37] R.St.C. Smart, T.L. Slager, L.H. Little, R.G. Greenler, J. Phys. Chem. 77
(1973) 1019.
[38] Y. Yanagisawa, K. Takaoka, S. Yamabe, T. Ito, J. Phys. Chem. 99 (1995)
3704.
[39] X.D. Peng, M.A. Barteau, Surf. Sci. 233 (1990) 283.
[40] C. Goyhenex, M. Croci, C. Claeys, C.R. Henry, Surf. Sci. 352–354 (1996)
475.
The authors thank U. Melville and K. Koblitz for tech-
nical assistance with the IR spectroscopic measurements and
Y. Borchert for useful discussions. This work was supported by
the German Science Foundation within the framework of SPP
1091.
References
[1] R.L. Burwell Jr., Langmuir 2 (1986) 2.
[2] A. Borodzinski, A. Golebiowski, Langmuir 13 (1997) 883.
[3] P.D. Tien, T. Satoh, M. Miura, M. Nomura, Energy Fuels 19 (2005) 731.
[4] M.A. Vannice, R.L. Garten, Ind. Eng. Chem. Prod. Res. Dev. 18 (1979)
186.
[5] S.-H. Oh, G.B. Hoflund, J. Phys. Chem. A 110 (2006) 7609.
[6] A.M. Venezia, L.F. Liotta, G. Pantaleo, V. La Parola, G. Deganello,
A. Beck, Z. Koppany, K. Frey, D. Horvath, L. Guczi, Appl. Catal. A 251
(2003) 359.
[41] K. Wolter, O. Seiferth, H. Kuhlenbeck, M. Bäumer, H.-J. Freund, Surf.
Sci. 399 (1998) 190.
[42] M. Bäumer, H.-J. Freund, Prog. Surf. Sci. 61 (1999) 127.
[43] I.V. Yudanov, R. Sahnoun, K.M. Neyman, N. Rösch, J. Hoffmann, S.
Schauermann, V. Johánek, H. Unterhalt, G. Rupprechter, J. Libuda, H.-J.
Freund, J. Phys. Chem. B 107 (2003) 255.
[44] G. Rupprechter, H. Unterhalt, M. Morkel, P. Galletto, L. Hu, H.-J. Freund,
Surf. Sci. 502–503 (2002) 109.
[7] H.G. Stenger Jr., J.S. Hepburn, Energy Fuels 1 (1987) 412.
[8] J.S. Hepburn, H.G. Stenger Jr., Energy Fuels 2 (1988) 289.
[9] C.R. Henry, Surf. Sci. Rep. 31 (1998) 231.
[10] M. Frank, M. Bäumer, Phys. Chem. Chem. Phys. 2 (2000) 3723.
[11] H.-J. Freund, M. Bäumer, H. Kuhlenbeck, Adv. Catal. 45 (2000) 412.
[12] T.P.St. Clair, D.W. Goodman, Top. Catal. 13 (2000) 5.
[13] K.H. Hansen, Z. Sljivancanin, E. Lægsgaard, F. Besenbacher, I. Stens-
gaard, Surf. Sci. 505 (2002) 25.
[45] M. Morkel, H. Unterhalt, M. Salmeron, G. Rupprechter, H.-J. Freund,
Surf. Sci. 532–535 (2003) 103.
[46] B. Hammer, J.K. Norskov, in: R.M. Lambert, G. Pacchioni (Eds.),
Chemisorption and Reactivity on Supported Clusters and Thin Films, in:
NATO ASI Series E, Kluwer, Dordrecht, 1997.
[47] R. van Hardeveld, F. Hartog, Surf. Sci. 15 (1969) 189.
[48] A. Wille, P. Nickut, K. Al-Shamery, J. Mol. Struct. 695–696 (2004) 345.
[49] V.G. Kulebakin, V.N. Koptelov, L.D. Bocharov, Yu.A. Dmitrienko, P.S.
Polovinkina, Refract. Ind. Ceram. 34 (1993) 148.
[14] D.W. Goodman, J. Catal. 216 (2003) 213.
[15] H.-J. Freund, M. Bäumer, J. Libuda, T. Risse, G. Rupprechter, S. Shaikhut-
dinov, J. Catal. 216 (2003) 223.
[16] S.L. Tait, Z. Dohnalek, C.T. Campbell, B.D. Kay, Surf. Sci. 591 (2005)
90.
[50] Sh.K. Shaikhutdinov, M. Frank, M. Bäumer, S.D. Jackson, R.J. Oldman,
J.C. Hemminger, H.-J. Freund, Catal. Lett. 80 (2002) 115.
[51] M. Morkel, V.V. Kaichev, G. Rupprechter, H.-J. Freund, I.P. Prosvirin, V.I.
Bukhtiyarov, J. Phys. Chem. B 108 (2004) 12955.
[17] T. Dellwig, J. Hartmann, J. Libuda, I. Meusel, G. Rupprechter, H. Unter-
halt, H.-J. Freund, J. Mol. Catal. A 162 (2000) 51.
[52] A. Ludviksson, R. Zhang, C.T. Campbell, K. Griffiths, Surf. Sci. 313
(1994) 64.
[18] T. Dellwig, G. Rupprechter, H. Unterhalt, H.-J. Freund, Phys. Rev. Lett. 85
(2000) 776.
[53] O. Demoulin, G. Rupprechter, I. Seunier, B. Le Clef, M. Navez, P. Ruiz,
J. Phys. Chem. B 109 (2005) 20454.
[19] H. Unterhalt, G. Rupprechter, H.-J. Freund, J. Phys. Chem. B 106 (2002)
356.
[54] T. Schalow, B. Brandt, M. Laurin, S. Schauermann, J. Libuda, H.-J. Fre-
und, J. Catal. 242 (2006) 58.
[20] G. Rupprechter, V.V. Kaichev, H. Unterhalt, M. Morkel, V.I. Bukhtiyarov,
Appl. Surf. Sci. 235 (2004) 26.
[55] H. Gabasch, A. Knop-Gericke, R. Schlögl, M. Borasio, C. Weilach, G.
Rupprechter, S. Penner, B. Jenewein, K. Hayek, B. Klötzer, Phys. Chem.
Chem. Phys. 9 (2007) 533.
[56] T. Schalow, B. Brandt, D.E. Starr, M. Laurin, S. Schauermann, S.K.
Shaikhutdinov, J. Libuda, H.-J. Freund, Catal. Lett. 107 (2006) 189.
[57] H. Gabasch, W. Unterberger, K. Hayek, B. Klötzer, E. Kleimenov, D.
Teschner, S. Zafeiratos, M. Hävecker, A. Knop-Gericke, R. Schlögl, J.
Han, F.H. Ribeiro, B. Aszalos-Kiss, T. Curtin, D. Zemlyanov, Surf. Sci.
600 (2006) 2980.
[21] V.V. Kaichev, M. Morkel, H. Unterhalt, I.P. Prosvirin, V.I. Bukhtiyarov,
G. Rupprechter, H.J. Freund, Surf. Sci. 566–568 (2004) 1024.
[22] V.V. Kaichev, I.P. Prosvirin, V.I. Bukhtiyarov, H. Unterhalt, G. Rup-
prechter, H.-J. Freund, J. Phys. Chem. B 107 (2003) 3522.
[23] V. Matolin, E. Gillet, Surf. Sci. 238 (1990) 75.
[24] F. Prinetto, M. Manzoli, G. Ghiotti, M. de Jesus Martinez Ortiz, D. Tichit,
B. Coq, J. Catal. 222 (2004) 238.
[25] T. Lear, R. Marshall, J.A. Lopez-Sanchez, S.D. Jackson, T.M. Klapötke,
M. Bäumer, G. Rupprechter, H.-J. Freund, D. Lennon, J. Chem. Phys. 123
(2005) 174706.
[58] T. Engel, G. Ertl, in: D.A. King, D.P. Woodruff (Eds.), The Chemical
Physics of Solid Surfaces and Heterogeneous Catalysis, vol. 4, Elsevier,
Amsterdam, 1982, p. 73.
[26] T. Lear, R. Marshall, E.K. Gibson, T. Schütt, T.M. Klapötke, G. Rup-
prechter, H.-J. Freund, J.M. Winfield, D. Lennon, Phys. Chem. Chem.
Phys. 7 (2005) 565.
[59] G. Rupprechter, M. Morkel, H.-J. Freund, R. Hirschl, Surf. Sci. 554 (2004)
43.
[60] M. Morkel, G. Rupprechter, H.-J. Freund, Surf. Sci. Lett. 588 (2005) L209.