Acetylene trimerization on Ag, Pd and Rh atoms deposited on MgO thin films

Article abstract of DOI:10.1039/b414399j

The acetylene trimerization on the group VIII transition metal atoms, Rh and Pd, as well as on Ag atoms supported on MgO thin films has been studied experimentally and theoretically. The three metal atoms with the atomic configurations 4d⁸5s¹ (Rh), 4d¹⁰s⁰ (Pd) and 4d¹⁰5s¹ (Ag) behave distinctly differently. The coinage metal atom silver is basically inert for this reaction, whereas Pd is active at 220 and 320 K, and Rh produces benzene in a rather broad temperature range from 350 to ca. 430 K. The origins of these differences are not only the different electronic configurations, leading to a weak interaction of acetylene with silver due to strong Pauli repulsion with the 5s electron but also the different stability and dynamics of the three atoms on the MgO surface. In particular, Rh and Pd atoms interact differently with surface defects like the oxygen vacancies (F centers) and the step edges. Pd atoms migrate already at low temperature exclusively to F centers where the cyclotrimerization is efficiently promoted. The Rh atoms on the other hand are not only trapped on F centers but also at step edges up to about 300 K. Interestingly, only Rh atoms on F centers catalyze the trimerization reaction whereas they are turned inert on the step edges due to strong steric effects. The Owner Societies 20O5.

Full text of DOI:10.1039/b414399j

View Article Online / Journal Homepage / Table of Contents for this issue
R E S E A R C H P A P E R
Acetylene trimerization on Ag, Pd and Rh atoms deposited
on MgO thin films
a
a
a
a
a
Ken Judai, Anke S. Wo
¨
Annalisa Del Vitto, Livia Giordano and Gianfranco Pacchioni
rz, Ste
´
phane Abbet, Jean-Marie Antonietti, Ueli Heiz,
b
b
b
a
Institute of Surface Chemistry and Catalysis, University of Ulm, Albert-Einstein-Allee 47,
D-89069 Ulm, Germany
Dipartimento di Scienza dei Materiali, Universita` Milano-Bicocca, via R. Cozzi 53,
I-20125 Milano, Italy
b
Received 17th September 2004, Accepted 25th October 2004
First published as an Advance Article on the web 8th November 2004
The acetylene trimerization on the group VIII transition metal atoms, Rh and Pd, as well as on Ag atoms
supported on MgO thin films has been studied experimentally and theoretically. The three metal atoms with the
8
1
10 0
10
1
atomic configurations 4d 5s (Rh), 4d s (Pd) and 4d 5s (Ag) behave distinctly differently. The coinage metal
atom silver is basically inert for this reaction, whereas Pd is active at 220 and 320 K, and Rh produces benzene
in a rather broad temperature range from 350 to ca. 430 K. The origins of these differences are not only the
different electronic configurations, leading to a weak interaction of acetylene with silver due to strong Pauli
repulsion with the 5s electron but also the different stability and dynamics of the three atoms on the MgO
surface. In particular, Rh and Pd atoms interact differently with surface defects like the oxygen vacancies
(
F centers) and the step edges. Pd atoms migrate already at low temperature exclusively to F centers where the
cyclotrimerization is efficiently promoted. The Rh atoms on the other hand are not only trapped on F centers
but also at step edges up to about 300 K. Interestingly, only Rh atoms on F centers catalyze the trimerization
reaction whereas they are turned inert on the step edges due to strong steric effects.
the surface of the oxide. The theoretical results suggest that the
oxygen vacancies at the surface of MgO (so called F centers)
are responsible for the stabilization of the Pd atoms, and for
the enhancement of their reactivity via charge transfer. An
additional proof of the active role of F centers originates from
the analysis of the adsorption of CO on Rh, Pd and Ag atoms
deposited on a MgO thin film: Fourier transform infrared
(FTIR) and temperature programmed desorption (TPD) spec-
tra in combination with ab initio calculations indicate that the
F centers are most likely the trapping sites for these metal
atoms. In addition, these studies revealed complex diffusion
patterns of both the metal atoms and the metal–CO complexes
on the MgO surface when the substrate temperature is in-
creased.
1
. Introduction
The cyclotrimerization of acetylene to form benzene, 3C H -
2
2
6 6
C H , is a well-known reaction studied on single crystal
ꢀ
12
surfaces from UHV conditions (10 –10 atm) to atmo-
ꢀ8
ꢀ
1
1–4
5,6
spheric pressure (10 –1 atm).
the reaction proceeds through the formation of the intermedi-
ate, C
bound to form benzene. The entire process can be divided into
three steps:
Pdsite þ 2C
Pdsite(C
On the Pd(111) surface
7
–10
H
4 4
,
to which an additional acetylene molecule is
1
1
H
2 2
- Pdsite(C
2
2
H )
2
- Pdsite(C
4
H
4
)
(1)
4
4
H ) þ C
2
H
2
- Pdsite(C
4
H
4
2 2
)(C H )
-
Pdsite(C H )
6
(2)
(3)
Despite the detailed picture accumulated so far, there are
still several open questions, e.g. whether F centers play a
similar role for other metal atoms or how the element-specific
stabilities of metal atoms on the oxide surface are manifested in
their chemical reactivities. A way to gain insight into these
problems is to investigate systematically a series of deposited
transition metal atoms. Here we present a comparative study of
the acetylene cyclotrimerization on Ag, Pd and Rh atoms. Such
a comparison is interesting as the electron configuration
6
Pdsite(C H ) - Pd þ C H
6
6
site
6
6
On Pd(111) the first two steps, reactions (1) and (2), occur with
very low energy barriers whereas the desorption of benzene (3)
is the rate determining step. For low benzene coverages,
formation of benzene is observed at around 500 K. This
reaction has also been studied on size-selected Pdn clusters
deposited on MgO thin films grown on a Mo(100) surface.
The cluster-assembled materials obtained in this way exhibit
peculiar activity and selectivity in the polymerization of acetyl-
ene to form benzene and aliphatic hydrocarbons. Up to Pd3
only C is formed, whereas the branching ratio for the
formation of C , C and C varies with size for larger
clusters. Theoretical calculations have shown that for iso-
lated atoms the reaction mechanism is the same as for the
Pd(111) surface and that the interaction with the substrate is
essential in activating the Pd atom for this reaction. In fact, a
free Pd atom is inert for the cyclotrimerization reaction and the
activation is connected to the existence of donor levels on
1
2
8
1
10
0
10
1
changes from 4d 5s of Rh to 4d 5s for Pd and to 4d 5s
for Ag. The interplay between s and d occupation in the three
metals is relevant for both the analysis of the interaction with
the donor levels of the substrate, and for studying the activa-
tion of the adsorbed molecules. The cyclotrimerization of
1
3
6
H
6
6
H
6
4
H
6
4 8
H
1
4
1 1 1
acetylene on Ag /MgO, Pd /MgO and Rh /MgO has been
studied experimentally by means of temperature programmed
reaction (TPR) spectroscopy in combination with DFT calcu-
lations to get information on the reaction mechanism. The
obtained results show that, in contrast to Pd and Rh atoms, Ag
is inert independently of the adsorption site. In addition, both
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 9 5 5 – 9 6 2
955

View Article Online
1
Pd and Rh atoms are only reactive when stabilized at F centers.
However, while diffusion processes do not seem to play an
important role in the activity of supported Pd atoms, they have
a major effect on the chemistry of supported Rh atoms.
the only exception of complexes formed on F
the ground state is singlet.
s
centers where
Gaussian-type atomic orbital basis sets were employed to
construct the Kohn–Sham orbitals. The 6-31G* basis set has
2
been used for O and Mg. In the F
3
1
s
s
and F cases where one
or two electrons, respectively, are trapped in the vacancy, a
diffuse 6-31þG* basis set has been employed on the Mg atoms
nearest the vacancy to describe the electron localization in the
2
. Experimental and computational details
2
a. Experiments
2
4
2
6
10
The metal atoms are produced by a high-frequency laser
cavity. A 18-electrons ECP where the 4s 4p 4d electrons are
explicitly treated in the valence, and a double-z plus polariza-
1
5
evaporation source. The positively charged ions are guided
by home-built ion optics through differentially pumped va-
cuum chambers and are size-selected from the cluster distribu-
tion by a quadrupole mass spectrometer (Extrel Merlin System;
mass limits: 1000, 4000, 9000 amu) before deposition onto thin
MgO(100) surfaces with a known density of F centers.
In these films the oxygen vacancies are generated by using a
defined preparation procedure (Mg evaporation rate: 2–5 ML
min , O
annealed to 840 K for 10 min. Auger electron spectroscopy
2
5
tion basis have been used for Rh and Ag (LanL2DZ), while C
and H have been treated with a 6-311G(p,d) basis. No basis set
superposition error (BSSE) correction has been applied to the
results. The calculations have been performed at the spin-
2
6,27
polarized DFT level with the B3LYP functional.
calculations have been performed with the GAUSSIAN98
package.
All
8
2
ꢀ1
ꢀ6
16
Torr). The films are then
2
background: 10
(
magnesium and oxygen and the absence of any impurity.
AES) measurements show a one-to-one stoichiometry for
1
3. Results
5
3
a. Experiments
When not indicated elsewhere the thicknesses are about 10
monolayers, as determined by AES peak intensities and by
X-ray photoemission (XPS), using the intensity attenuation of
To characterize the model catalysts in more detail, it is
necessary to identify experimentally the adsorption sites of
the clusters in order to relate the results to the computational
investigations. We first summarize results published else-
1
7
the Mo 3d core level with increasing film coverage. The films
8
1
have also been studied by electron energy loss spectroscopy
and exhibit characteristic losses between 1 and 4 eV, lying
within the MgO band gap. Similar loss structures have been
2
9
where, where the probe molecule CO is used to get informa-
tion on the stability and dynamics of the adsorbed atoms. As
an example, CO desorbs from the adsorbed Pd atoms at a
1
9
observed before and according to first-principle calculations
using large cluster models, they have been attributed to transi-
tions, characteristic for neutral surface F centers in various
3
0
temperature of about 250 K, which corresponds to a binding
energy, E
b
, of about 0.7 ꢂ 0.1 eV. FTIR spectra suggest that at
2
0
coordinations on flat terraces, at steps and at kinks. The
density of these oxygen vacancies is estimated to be larger than
saturation two different sites for CO adsorption exist on a
single Pd atom. The vibrational frequency of the most stable,
1
3
ꢀ2
ꢀ1
5
ꢁ 10 cm
.
In these experiments it is important to deposit 0.5–0.8% of a
singly adsorbed CO molecule is 2055 cm . Density functional
cluster model calculations have been used to model possible
defect sites at the MgO surface where the Pd atoms are likely to
be adsorbed. CO/Pd complexes located at regular or low-
coordinated O anions of the surface exhibit considerably
stronger binding energies, E
tional shifts than were observed in the experiment. CO/Pd
complexes located at oxygen vacancies (F or F centers) are
1
5
ꢀ2
monolayer atoms (1 ML ¼ 2.25 ꢁ 10 clusters cm ) only at
9
0 K and with low kinetic energy in order to land them isolated
on the surface and to prevent agglomeration on the MgO films.
Monte Carlo simulations revealed that under such experimen-
¼ 2–2.5 eV, and larger vibra-
1
b
9
tal conditions, e.g. cluster flux (ca. 10 cm ), cluster density
ꢀ1
1
3
ꢀ1
13
ꢀ1
(
MgO(100) films, less than 10% of the atoms coalesce during
ca. 10 cm ), and defect density (ca. 5 ꢁ 10 cm ) on the
characterized by much smaller binding energies, E
b
¼ 0.5 ꢂ 0.2
2
1
migration to the trapping centers.
or 0.7 ꢂ 0.2 eV, which are in agreement with the experimental
value. This comparison not only identifies adsorption sites of
the atoms on the MgO surface, but by the absence of vibra-
tional frequencies typical for bridge-bonded CO (up to about
300 K) it is concluded that the atoms stay well-isolated. The
thermal stability was also investigated for Rh atoms and it was
The chemical reactivity and stability of the deposited metal
atoms is studied by TPD and by FTIR studies. The latter is
mainly used to investigate the stability of the atoms on the
surface by using the probe molecule CO.
2
9
found that they are stable up to at least 450 K.
2
b. Calculations
For the cyclotrimerization on deposited Ag, Pd and Rh
atoms, the nanocatalysts were exposed at 90 K, using a
calibrated molecular beam doser, to about 1 Langmuir (L) of
acetylene. In a TPR study, catalytically formed benzene
For the calculations we have used the same embedded cluster
approach adopted to model reactions (1) and (2) on Pd /MgO.
1
2
c
ꢀ
The sites considered for adsorption are: terrace O
O
, step
4c , F and F centers at a terrace and they have been
represented by the following clusters: O Mg5 for terrace,
5
2ꢀ
1
(C
by a mass spectrometer as function of temperature. It is
interesting to note that the reactant (C ) is only physisorbed
6 6 4 6 4 8
H ), butadiene (C H ) and butene (C H ) were monitored
9
O
12Mg10 and O15Mg12 for step, O
8
Mg
5
for neutral F
s
and
center. The MgO
2 2
H
1
1
[
O
8
5
Mg ]
for the positively charged F
s
on MgO and desorbs at temperatures lower than 150 K, e.g.
before the reaction takes place.
Fig. 1 depicts the results for Rh atoms. Clearly, only the
cyclization of C
possible products like H
thus Rh atoms are very selective for the cyclotrimerization
reaction. The formation of C occurs in a rather broad
temperature window between 375 and 475 K. In contrast to Pd
atoms the reaction temperature is distinctly higher for Rh.
Earlier studies revealed that, depending on the thickness of the
defect-rich MgO films, benzene is formed on Pd atoms at 220 K
(Fig. 2a) or 300 K (Fig. 2b), respectively. This change in
temperature was not observed for the other two metals (Rh
and Ag) and the origin of this observation is not yet clear.
clusters have been embedded in effective core potentials (ECP)
and a large array of point charges (PC) in order to represent
2
the Madelung potential in the cluster region. The whole
2
H
2 2
is catalyzed by the Rh atoms and other
system, cluster þ ECPs þ PCs is electrically neutral, except
2
, C , C or C are not formed,
4
H
4
4
H
6
4 8
H
1
for the F
s
center which carries a net positive charge. The
positions of the adsorbed molecules, of Rh or Ag atoms and of
their first neighbors on the MgO surface have been fully
optimized with no symmetry constraints. The transition state
6 6
H
(
TS) search for reactions (1) and (2) has been performed using
the Berny algorithm and in selected cases the nature of the TS
has been verified by performing a frequency analysis. The
calculations have been performed in spin polarized mode. All
Ag and Rh surface complexes have doublet ground states with
9
56
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 9 5 5 – 9 6 2
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Article Online
Fig. 1 TPR spectra of the catalytic formation of C
C
on the thin MgO films was 0.8% ML (1 ML ¼ 2.2 ꢁ 10 ꢁ cm ) and
the model catalysts were exposed at 90 K to 1 Langmuir (L) of
acetylene. Rh atoms selectively form benzene in a rather broad
temperature range starting from around 300 K up to about 550 K.
6 6 4 4
H (a) and C H ,
4
H
6
, C and C 12 (b) on deposited Rh atoms. The atom density
8
4
H
6
H
Fig. 2 TPR spectra of the catalytic formation of C H on deposited
6
6
15
ꢀ1
Pd atoms. The atom density on the thin MgO films was 0.8% ML
15
ꢀ1
(1 ML ¼ 2.2 ꢁ 10 cm ) and the model catalysts were exposed at
90 K to 1 Langmuir (L) of acetylene. Depending on the preparation
method of the defect-rich films a reaction temperature at 220 K (a)
6 6
and 320 K (b) was observed. Note that the peaks of the C H
formation are narrow (fwhm: ca. 15 K (a); ca. 40 K (b)) in comparison
to Rh atoms. In none of the experiments could both peaks be
detected in the same spectrum. This behavior was only observed
in the case of Pd atoms. The origin of the temperature change is not
yet understood.
Another difference with respect to Rh are the much narrower
C H desorption peaks for Pd atoms, for the low-temperature
formation the width is just 25 K (Fig. 2a). As in the case of
rhodium, Pd atoms are very selective for the formation of
benzene at both temperatures. The behavior of Ag-atoms is
completely different and none of the possible product mole-
6
6
cules (H
in Fig. 3.
2
, C
4
4
H , C
4
H
6
, C
H
4 8
, C
H
6 6
) were detected as shown
Free Rh atoms. Before considering the activity of supported
Rh atoms, we discuss the cyclotrimerization reaction on a free
Rh atom (Figs. 4 and 5). Rh binds acetylene by 1.57 eV and the
ground state is doublet (Figs. 4a and 5); the adsorption is
accompanied by a substantial activation of the molecule
(elongation of the C–C bond, loss of linearity (Table 1)). The
second acetylene is adsorbed with an energy gain of 1.23 eV to
give rise to the complex shown in Fig. 4b. A very small barrier,
3
b. Calculations
Ag /MgO. Ag atoms deposited on the terrace sites do not
bind C H . The geometry optimizations show the existence of a
1
2
2
local minimum 0.44 eV above the dissociation limit; this is in
line with the results for CO adsorption on Ag /MgO, which
0.04 eV, separates this structure from the Rh(C
(Fig. 4c) a stable intermediate which is 0.86 eV lower in energy
than Rh(C . Rh(C ) can adsorb a third acetylene
molecule with a gain of 0.76 eV to form Rh(C )(C
(Fig. 4e). This is the precursor state of the final product,
Rh(C ) (Fig. 4f), from which it is separated by a barrier of
4 4
H ) complex
1
show little or no tendency of the supported Ag atoms to adsorb
molecules. When the Ag atoms are trapped at the F centers the
acetylene molecules are only slightly bound (0.07 eV); as these
results have not been corrected by the BSSE we conclude that
for none of the defect sites considered in this work Ag atoms
bind acetylene. These results are consistent with the experi-
mental observation that Ag is inert towards the formation of
benzene. The inert character of Ag must be related to the
presence of a partially occupied 5s level. This orbital is spatially
expanded and since the 5s electron cannot be promoted into
the filled 4d shell (unlike for Rh or Pd) a strong Pauli repulsion
occurs. The only way to increase the Ag activity would be the
oxidation of the atom to reduce the population of the 5s
orbital; this process is however not possible on a basic oxide
like MgO.
2
2
H )
2
4 4
H
H
4
H )
2 2
4
H
6 6
0.24 eV (Fig. 5). The last step is associated with an energy gain
of 2.10 eV (Fig. 5). Once formed, benzene is immediately
released as it is virtually unbound to the metal atom (Fig. 5).
Thus, on a free Rh atom the reaction occurs with very small
barriers. Notice that Rh is able to bind even three acetylene
molecules; Rh(C
by 0.5 eV. In this respect, Rh has quite a distinct
H
)
(Fig. 4d), is more stable than Rh(C
H
)
þ
2
2
3
2
2
2
2 2
C H
behavior compared to Pd. In fact, a gas-phase Pd atom binds
only weakly the third acetylene and is thus inert for the
1
2,14
reaction.
The fact that benzene is formed on Pd atoms
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 9 5 5 – 9 6 2
957

View Article Online
Fig. 5 Energy profile for the conversion of acetylene to benzene
promoted by a gas-phase Rh atom. In the insets are shown the
structures of the two transition states involved in the reaction mechan-
ism.
3
1
theory G E G
0
exp(ꢀE
¼ 10 Hz, one can estimate, at a deposition temperature of
00 K, a hopping frequency G E 89 GHz and a residence time
d
/kT) and a pre-exponential factor
1
3
G
0
1
ꢀ8
t ¼ 1/G E 1.1 ꢁ 10 s. As a consequence, the kinetic energy of
atoms arriving on the surface and the thermal energy of the
substrate are sufficient to induce extremely rapid diffusion on
the flat terraces even at low temperature; the diffusion process
will stop only in the presence of strong binding sites like steps
1
or oxygen vacancies (F or F centers). Thus, it is very unlikely
that Rh atoms deposited on the (100) faces will be involved in
the trimerization reaction. Still, reactions (1)–(3) have been
considered on this site for the sake of completeness. A Rh atom
bound to a MgO terrace binds the first (Fig. 6a) and second
(
0
Fig. 6b) acetylene molecule strongly, De1 ¼ 2.23 eV and De2
¼
Fig. 3 TPR spectra of the catalytic formation of C
C
6 6 4 4
H (a) and C H ,
.89 eV, respectively (Fig. 7); the formation of the C H
4 4
4
H
6
4
H
8
6
, C and C H12 (b) on deposited Ag atoms. The atom density
15
ꢀ1
intermediate (Fig. 6c) is exothermic, DE ¼ ꢀ0.83 eV and
on the thin MgO films was 0.8% ML (1 ML ¼ 2.2 ꢁ 10 cm ) and the
acetylene binds also to the Rh(C H ) complex (D ¼ 0.90
model catalysts were exposed at 90 K to 1 Langmuir (L) of acetylene.
is
4
4
e3
In contrast to Pd and Rh atoms, only a very small amount of C
formed on Ag atoms.
6
H
6
eV, Fig. 6d). However, the barrier for the first transition state,
#
1
Rh(C H ) - Rh(C H ), DE ¼ þ1.74 eV, is quite high (Fig.
2
2 2
4
4
#
7
0
). The second barrier, Rh(C
4
4
H )(C
2
2
H ) - Rh(C
H
6 6
), DE
2
¼
.40 eV, is much smaller, indicating that step (1) is rate
supported on MgO reveals the active role of the substrate in
contrast to Rh.
determining. Once formed, benzene is bound to the supported
Rh atom by 0.92 eV (Figs. 6e and 7).
2 2 3
Unlike the gas phase, the supported Rh(C H ) complex is
Rh /MgO(terrace). Previous work has shown that Rh atoms
1
unstable and tends to dissociate one of the three acetylene
molecules. The reason may be steric hindrance. In fact, in
the free Rh(C H ) complex one acetylene is normal to the
2 2 3
have a high mobility on the MgO surface. The diffusion
barriers, E , from an O site to the next one are of the order
of 0.1 eV. Using an Arrhenius expression from transition state
d
5c
plane containing the Rh atom and the other two acetylene
molecules; on the surface this arrangement is not possible
because of the repulsive interaction with the substrate. In
general, steric hindrance has an important consequence for
the whole reactivity. The propensity of the Rh atom to
promote the trimerization reaction is in fact strongly reduced
on the MgO support. The strongest manifestation of this effect
is the high barrier to transforming two adsorbed acetylene
4 4
molecules into a C H intermediate. The barrier, 1.74 eV,
exceeds the binding energy of acetylene. One reason for this
large activation barrier is the attractive interaction of the two
Table 1 Adsorption properties of C
Rh/MgO complexes
2
H
2
on free Rh atom and on
Rh on MgO
1
Free Rh
O
terrace
O
step
F
F
˚
d(C–C)/A
1.288
1.977
147.9
1.57
1.278
2.004
1.315
1.960
1.274
2.052
1.247
2.135
˚
d(C–Rh)/A
(H–C–C)/1
(C )/eV
(RhC )/eV
+
149.8
2.23
1.65
140.2
1.76
2.50
149.6
1.06
2.82
157.4
0.58
1.76
Fig. 4 Structure of gas-phase Rh–acetylene complexes: (a) Rh(C
b) Rh(C ; (c) Rh(C ); (d) Rh(C ; (e) Rh(C )(C
Rh(C
2
H
); (f)
2
);
D
D
e
2 2
H
(
2
H
H ).
6
2
)
2
4
H
4
2
H
2
)
3
4
H
4
2 2
H
e
2
H
2
—
6
9
58
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 9 5 5 – 9 6 2
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Article Online
Fig. 8 Structure of Rh–acetylene complexes supported on a MgO
(
2 2 2 2 2
001) step: (a) Rh(C H ); (b) Rh(C H ) . Selected distances are given.
considered in this work, this site corresponds to the strongest
activation of acetylene, as shown by the long C–C distance,
short C–Rh distance, and small HCC angle (Table 1). The
Rh(C H ) complex is bound to the surface by 2.5 eV, with
2
2
the Rh atom interacting simultaneously with the O4c atom of
˚
the step, r(O–Rh) ¼ 2.168 A, and with the O5c anion of the
˚
basal plane, r(O–Rh) ¼ 2.386 A (Fig. 8a). The larger activation
of the acetylene molecule reflects the interaction with two
2 2
surface basic sites. The high stability of the Rh(C H ) complex
makes it difficult to bind the second acetylene molecule:
various geometries have been considered but we find only
unbound or weakly bound Rh(C
(
2 2 2
H ) complexes (Fig. 8b)
the calculations have been repeated with the larger O15Mg12
cluster but the results do not change). Again, this is due in part
to steric effects, since Rh cannot easily accommodate two
acetylene molecules when it is bound at a step.
Based on these results and assuming that Rh atoms are
bound at the MgO step sites, one cannot explain the observed
trimerization of benzene. On the other hand, Rh(C H ) com-
2
2
Fig.
MgO(001) terrace: (a) Rh(C
6 6
Rh(C )(C ); (e) Rh(C H
6
Structure of Rh–acetylene complexes supported on
); (b) Rh(C ; (c) Rh(C ); (d)
). Selected distances are given.
a
plexes form easily at these sites. The barrier for diffusion of
Rh(C H ) from a step to a terrace site is o1.4 eV, i.e. smaller
2
H
2
2
H
2
)
2
4 4
H
2
2
4
H
4
2
H
2
that the energy required to detach acetylene. This means that a
temperature increase may result in the diffusion of the
2 2
Rh(C H ) complex before acetylene desorption occurs.
acetylene molecules with the surface (Fig. 6b). The H atoms of
the two molecules form a kind of H-bonding with the surface
oxide anions. The energy required to bring the two molecules
together is thus much higher than in the gas phase.
Rh
strongly at F centers. This surface complex binds C H by 1.06
1
/MgO(F center). As other metal atoms, also Rh binds
2
2
eV (Table 1). It is interesting to note that the properties of the
adsorbed hydrocarbon molecule on Rh/F5c are similar to those
computed for Rh on a terrace site (Table 1 and Fig. 9). The
main difference however is in the binding energy: on the F
center it is about one-half that on the terrace. Rh/F5c can bind
a second C H molecule by 0.96 eV. The formation of the C H
Rh /MgO(step). At variance with Pd, Rh atoms bind quite
1
2
9
strongly at step sites of the MgO surface (D
e
¼ 1.95 eV).
Thus, while these sites are of little interest in the case of Pd,
they are potentially involved in the reactivity of Rh. The
interaction of acetylene with a Rh atom at a step site has been
analyzed in detail. At step sites Rh binds strongly the first
acetylene, De1 ¼ 1.77 eV, see Fig. 8a. Among the cases
2
2
4
4
complex implies an energy gain, DE ¼ ꢀ0.91 eV (Fig. 10)
similar to that observed for Rh on step and terrace sites and
even in the gas phase. A third acetylene can be added to the
Rh(C H ) complex on a F center with D ¼ 0.90 eV (Fig. 9d).
4
4
e3
The formation of benzene is exothermic by 3.23 eV, and the
resulting molecule (Fig. 9e) is only weakly bound to the
supported atom (D
e
¼ 0.40 eV). The barriers for steps (1)
#
#
and (2) of the reaction are similar, DE ¼ þ1.18 eV, DE
¼
1 2
þ1.07 eV (Fig. 10), and are about 20% higher than those
computed for Pd. Frequency calculations of the saddle point
structures show the existence of only one negative frequency
value associated to the vibrational mode of the reaction
coordination chosen for the reaction path. Thus, the two
structures correspond to real TS’s. If one assumes a Redhead
1
3
equation and a 10 pre-exponential factor, the computed
barriers correspond to a desorption temperature of about 400
K. This temperature is comparable to the temperature of
maximal benzene formation of 430 K, as observed in Fig. 1.
Thus, Rh atoms trapped on F centers contribute to the
formation of benzene at ca. 400 K. The quite broad peak
observed experimentally which starts already at around 350 K,
is due to the fact that some benzene can be formed during the
diffusion of Rh-acetylene species on the terrace sites (e.g. by
Fig. 7 Energy profile for the conversion of acetylene to benzene
promoted by a Rh atom supported on a MgO(001) terrace. In the
insets are shown the structures of the two transition states involved in
the reaction mechanism.
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 9 5 5 – 9 6 2
959

View Article Online
Fig. 9 Structure of Rh–acetylene complexes supported on a neutral
oxygen vacancy (F center) on the MgO(001) surface: (a) Rh(C ); (b)
Rh(C ; (c) Rh(C ); (d) Rh(C )(C ); (e) Rh(C ).
Fig. 11 Structure of Rh–acetylene complexes supported on a charged
oxygen vacancy (F center) on the MgO(001) surface: (a) Rh(C H );
2 2
2
H
2
1
2
H
2
)
2
4
H
4
4
H
4
2
H
2
6 6
H
(
2 2 2 4 4 4 4 2 2 6 6
b) Rh(C H ) ; (c) Rh(C H ); (d) Rh(C H )(C H ); (e) Rh(C H ).
formation of small Rh clusters). The fact that the barriers for
benzene formation for Rh on F centers are about 20% higher
than those computed for Pd on the same site is qualitatively
consistent with the higher reaction temperature for the former
atom (see Figs. 1 and 2).
than Rh(C H ) . The addition of the third acetylene molecule
2
2 2
1
4
4
to Rh(C H
4
) (Fig. 11d), occurs with a gain of 1.01 eV. Thus,
all the intermediates have a sufficiently high stability for
the reaction to occur according to steps (1)–(3). From the
Rh(C H )(C H ) complex one can form benzene with a large
4
4
2
2
gain of 3.79 eV (Fig. 12). As for the neutral F center, benzene is
weakly bound to the supported Rh atom and immediately
1
1
1
Rh /MgO(F center). Rh on a F center is able to coordi-
nate and activate two acetylene molecules; the first (Fig. 11a), is
bound by 0.58 eV only, the second slightly more, 0.66 eV (Fig.
1
reflected in long C–C and Rh–C bonds, as well as in a large
desorbs (D ¼ 0.24 eV). Despite several attempts, the search for
e
the transition states of steps (1) and (2) has failed. Only rough
1
1b). The weak interaction of Rh/F5c with acetylene is
estimates have been obtained. A scan of the potential energy
surface for step (1) suggests a barrier o0.7 eV. For step (2)
even the scan of the potential energy surface has not produced
stable results, and no estimate of the barrier has been obtained.
HCC angle (Table 1), typical signs for modest activation. On a
1
F
4 4
site the Rh(C H ) complex (Fig. 11c), is 0.84 eV more stable
Fig. 10 Energy profile for the conversion of acetylene to benzene
promoted by a Rh atom supported on a neutral F center on the
MgO(001) surface. In the insets are shown the structures of the two
transition states involved in the reaction mechanism.
Fig. 12 Energy profile for the conversion of acetylene to benzene
promoted by a Rh atom supported on a charged F center on the
MgO(001) surface. In this case it has not been possible to identify the
transition states (see text).
1
9
60
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 9 5 5 – 9 6 2
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Article Online
1
Therefore, evidence about the ability of the F center to
promote the trimerization reaction is only partial, and based
exclusively on the stability of the various intermediates.
of the Rh atom supported at a step and rule out this surface
complex as a potential catalyst for the reaction. The binding of
acetylene to Rh supported on a MgO step, 1.77 eV, is such that
only at high temperature acetylene will desorb from this site.
Thus, as temperature increases it is more likely to induce
diffusion of the stable Rh(C H ) unit rather than acetylene
4
. Discussion and conclusions
2
2
In this section we formulate a plausible hypothesis for the
observed different behavior in the trimerization of acetylene
promoted by Rh, Pd and Ag atoms deposited on MgO thin
films on the basis of the findings described above. Ag is
practically inert for this reaction, as found both experimentally
and theoretically. This low reactivity is due to the Pauli
repulsion between the valence 5s electron on Ag and the
incoming molecules. This repulsive interaction does not allow
the acetylene molecules to bind to the metal atom, so that no
further reaction is possible. Interestingly, larger Ag clusters are
not catalyzing this reaction either.
dissociation. A very similar conclusion has been drawn for the
9
2
case of CO adsorption. Indeed, we evaluate that the barrier
to displace Rh(C ) from a step to a terrace site is about 1.3
2 2
H
eV. This means that around room temperature two different
processes can occur. If the Rh atoms have been directly
stabilized at the F centers (minority sites) the reaction will
take place on these sites as it has been found for Pd. In the case
where the Rh atoms are trapped at step sites, the Rh(C H )
2
2
complexes start to detach from the steps, diffuse on the surface
until they are captured by an empty F center. Rh(C H )
2
2
1
binds by 2.8 eV or 1.77 eV to F or F centers, respectively.
On these new sites the reaction can occur only if additional
acetylene molecules are readsorbed from the gas phase or by
reverse spill-over from the surface. The latter, however, is
unlikely when considering the relative low binding energy of
acetylene on pure MgO. The other possibility is that the
1
2–14
The Pd case has been studied in detail in the past.
It was
found that the reaction on Pd /MgO can be explained only by
1
assuming that the oxygen vacancies, either in neutral or
charged forms, are directly involved. The general picture is
that Pd atoms are deposited at low temperature (o100 K) on
the MgO thin films and that they possess enough residual
kinetic energy once they land on the surface to freely diffuse; in
fact, the thermal energy of the substrate only is not sufficient to
Rh(C
Rh(C
2
H
H
2
) unit diffuses until it encounters another Rh or
complex stabilized at a nucleation center (F center,
)
n
2
2
divacancy, etc.). This would lead to the nucleation of a small
cluster and benzene could form from this new species. The
existence of a complex diffusion pattern induced by the tem-
perature increase can explain the broad peak observed in the
TPD spectrum. According to the proposed explanation, ben-
zene could form in part from isolated Rh atoms at F centers
and in larger part from very small clusters generated during the
adsorbate-induced diffusion. In this respect it is important to
note that larger clusters (Rh10–Rh30) do not show any reactiv-
ity for the polymerization of acetylene.
In summary, the specific electronic configuration of silver
renders these atoms inert for the polymerization of acetylene.
Pd atoms, on the other hand, are turned into active catalysts
for the cyclotrimerization reaction only when adsorbed on
color centers as charge is donated from this defect site into
the atom. Finally, although already reactive as free atoms, Rh
atoms on a MgO surface catalyze the cyclotrimerization only
when trapped on F centers as otherwise steric effects reduce the
stability of the intermediates.
3
2
overcome the diffusion barrier, E E 0.4 eV. In the diffusion
d
process the Pd atoms are likely to be trapped at specific sites
where the interaction is stronger. Among these are point
defects like oxygen vacancies, divacancies but not extended
defects as steps. In fact, Pd atoms bind at steps with an
2
9
adsorption energy only slightly higher than on flat terraces.
As a consequence, the trapping energy for a Pd atom on a step
is relatively small and the Pd atoms most likely populate the F
centers present on the surface before acetylene is introduced in
the reaction chamber. This would explain the narrow peak
observed in the TDS spectrum of Fig. 2a assuming all the
active atoms to reside on the same site. From this point of view
the picture is relatively simple, and the only open question is
1
related to the nature of the trapping site, F, F , or other sites
not considered so far (e.g., divacancies). The best candidates
1
are the F and the F centers because (a) they are both strong
trapping sites for Pd atoms, and (b) the energy profile for the
reaction occurring on these centers are to a large extent
consistent with the measured desorption temperature (the
reaction profiles for Pd on other MgO sites are totally incon-
sistent with the experiment). In comparison to larger clusters,
single atoms are very selective for the formation of benzene. In
fact, by increasing cluster size the branching ratio of possible
product molecules is changing in favor of C H for cluster sizes
Acknowledgements
This work was supported by the ‘Deutsche Forschungs-
¨
gemeinschaft’ and the ‘Landesstiftung Baden-Wurttemberg’
4
6
with around 8 atoms and in favor of C
3
6
H
8
for sizes around
and by the European Project STRP GSOMEN. K.J. and
S.A. thank the Alexander v. Humboldt foundation, K.J. the
Japan Society for the Promotion of Science (JSPS), S.A. and
J.-M.A. the Swiss National Science foundation for financial
support, A.S.W. acknowledges support from the Graduierten-
kolleg ‘Molekulare Organisation und Dynamik an Grenz- und
Oberflachen’. The work of A.D.V., L.G. and G.P. is supported
¨
by the Italian Ministry of University and Research through a
Cofin 2003 project.
1
3
0 atoms.
The situation for Rh is far more complex. Recent combined
experimental and theoretical studies based on the use of CO as
a probe molecule have shown that Rh atoms are more mobile
2
9
than Pd atoms on MgO. The barrier for diffusion of Rh on
the flat terraces is about one third that of Pd. However, at
variance with Pd, Rh atoms bind quite strongly at the step sites
(
the binding energy on these sites is about twice that on the
9
2
1
terrace sites). The work done on CO/Rh /MgO suggests that
a significant fraction of the Rh atoms are bound at steps
already at low temperature and that only a minority of the
deposited Rh atoms have been stabilized at F centers in the
diffusion process (this can be explained with the higher prob-
ability to encounter an extended defect like a step than a point
defect like an F center in the diffusion process). Thus, Rh
atoms at steps could be involved in the cyclization reaction.
However, the calculations show that by exposure to acetylene
References
1
2
3
4
5
W. T. Tysoe, G. L. Nyberg and R. M. Lambert, J. Chem. Soc.,
Chem. Commun., 1983, 623–625.
W. S. Sesselmann, B. Woratschek, G. Ertl, J. Kuppers and
H. Haberland, Surf. Sci., 1983, 130, 245.
P. M. Holmblad, D. R. Rainer and D. W. Goodman, J. Phys.
Chem. B, 1997, 101, 8883.
I. M. Abdelrehim, K. Pelhos, T. E. Madey, J. Eng and J. G. Chen,
J. Mol. Catal. A, 1998, 131, 107.
one can form a stable Rh(C H ) complex but not the
2
2
Rh(C
2
H
2
)
2
or Rh(C
2
H
2
)
3
complexes, which are essential to
promote the reaction. Steric effects reduce the binding ability
T. M. Gentle and E. L. Muetterties, J. Phys. Chem., 1983, 87,
2469.
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 9 5 5 – 9 6 2
961

View Article Online
6
7
8
9
T. G. Rucker, M. A. Logan, T. M. Gentle, E. L. Muetterties and
G. A. Somorjai, J. Phys. Chem., 1986, 90, 2703.
C. H. Patterson and R. M. Lambert, J. Am. Chem. Soc., 1988, 110,
18 U. Heiz and W. D. Schneider, J. Phys. D, Appl. Phys., 2000, 33,
R85.
¨
19 D. Peterka, C. Tegenkamp, K. M. Schroder, K. M. Ernst and
6
871.
R. M. Ormerod and R. M. Lambert, J. Phys. Chem., 1992, 96,
111.
H. Pfnur, Surf. Sci., 1999, 431, 146.
¨
20 C. Sousa, G. Pacchioni and F. Illas, Surf. Sci., 1999, 429, 217.
21 S. Abbet, K. Judai, L. Klinger and U. Heiz, Pure Appl. Chem.,
2002, 74, 1527.
22 M. A. Nygren, L. G. M. Pettersson, Z. Barandiaran and L. Seijo,
J. Chem. Phys., 1994, 100, 2010.
8
R. M. Ormerod, R. M. Lambert, H. Hoffmann, F. Zaera, J. M.
Yao, D. K. Saldin, L. P. Wang, D. W. Bennet and W. T. Tysoe,
Surf. Sci., 1993, 295, 277.
1
1
0
1
G. Pacchioni and R. M. Lambert, Surf. Sci., 1994, 304, 208.
H. Hoffmann, F. Zaera, R. M. Ormerod, R. M. Lambert, J. M.
Yao, D. K. Saldin, L. P. Wang, D. W. Bennett and W. T. Tysoe,
Surf. Sci., 1992, 268, 1.
23 R. Ditchfield, W. Hehre and J. A. Pople, J. Chem. Phys., 1971, 54,
724.
24 R. Soave, A. M. Ferrari and G. Pacchioni, J. Phys. Chem. B, 2001,
105, 9798.
1
1
1
1
1
2
3
4
5
6
S. Abbet, A. Sanchez, U. Heiz, W.-D. Schneider, A. M. Ferrari,
25 P. J. Hay and W. R. Wadt, J. Chem. Phys., 1985, 82, 299.
26 A. D. Becke, J. Chem. Phys., 1993, 98, 5648.
27 C. Lee, W. Yang and R. G. Parr, Phys. Rev. B, 1988, 37, 785.
28 M. J. Frisch et al., GAUSSIAN98, Revision A.7, Gaussian Inc.,
Pittsburgh, PA, 1998.
G. Pacchioni and N. Ro
S. Abbet, A. Sanchez, U. Heiz and W. D. Schneider, J. Catal.,
001, 198, 122.
¨
sch, J. Am. Chem. Soc., 2000, 122, 3453.
2
A. M. Ferrari, L. Giordano, G. Pacchioni, S. Abbet and U. Heiz,
J. Phys. Chem. B, 2002, 106, 3173.
U. Heiz, F. Vanolli, L. Trento and W. D. Schneider, Rev. Sci.
Instrum., 1997, 68, 1986.
¨
29 K. Judai, S. Abbet, A. S. Worz, U. Heiz, L. Giordano and
G. Pacchioni, J. Phys. Chem. B, 2003, 107, 9377.
30 S. Abbet, E. Riedo, H. Brune, A.-M. Ferrari, L. Giordano and
G. Pacchioni, J. Am. Chem. Soc., 2001, 123, 6172.
31 V. Musolino, A. Selloni and R. Car, J. Chem. Phys., 1998, 108,
5044.
A. Sanchez, S. Abbet, U. Heiz, W. D. Schneider, H. Ha
R. N. Barnett and U. Landman, J. Phys. Chem. A, 1999, 103,
573.
M. H. Schaffner, F. Patthey and W. D. Schneider, Surf. Sci., 1998,
17, 159.
¨
kkinen,
9
17
32 L. Giordano, C. Di Valentin, J. Goniakowski and G. Pacchioni,
Phys. Rev. Lett., 2004, 92, 96105.
4
9
62
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 9 5 5 – 9 6 2
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5