Atomic scale investigation of the oxidation of CO on RuO2(110) by scanning tunneling microscopy

Article abstract of DOI:10.1021/jp047600v

The reaction of CO molecules with O-bridge atoms on the stoichiometric surface was monitored on the atomic scale and was found to proceed according to the established reaction model. The CO molecules first adsorbed on Ru-cus atoms, and then reacted with neighboring O-bridge atoms, creating vacancies in the O-bridge rows. The reaction largely occurred at random positions, defects did not play a role, and there were no significant contributions of islands of the reactants. The reaction of CO-cus with O-cus was investigated by O₂ titration on the CO-cus covered surface. On the saturated surface, where CO occupies every other Ru-cus atom, the reaction started at CO-cus vacancies, because the O₂ molecules need two neighboring sites for dissociation. In the later stage, the reaction was mainly random, because the CO-cus molecules displayed some mobility on the partially reacted surface, allowing O₂ molecules to adsorb.

Full text of DOI:10.1021/jp047600v

J. Phys. Chem. B 2004, 108, 14565-14569
14565
Atomic Scale Investigation of the Oxidation of CO on RuO2(110) by Scanning Tunneling
Microscopy†
Sang Hoon Kim‡,§ and J. Wintterlin*,‡,|
Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, D-14195 Berlin, Germany and
Ludwig-Maximilians-UniVersita¨t Mu¨nchen, Department Chemie, Butenandtstr. 11, D-81377 Mu¨nchen, Germany
ReceiVed: June 3, 2004
The reaction between adsorbed CO molecules and O atoms on epitaxially grown RuO2(110) was investigated
by scanning tunneling microscopy. Two types of titration experiments were performed at 300 K, in which
either CO was dosed on the stoichiometric oxide surface, or O2 on the CO saturated surface. In both cases,
reaction steps could be atomically resolved, and the surface mobility of adsorbed CO molecules and O atoms
could be studied. CO molecules adsorbing on the stoichiometric surface react with bridge bonded O atoms
at largely random positions. O2 molecules adsorbing on the CO saturated surface react with cus-bonded CO
molecules first at CO vacancies, and then also in a mainly random fashion. The statistical nature of the
reaction is explained by the relatively high surface diffusion barriers, which are of the same size as the
activation energies for the reactions.
1. Introduction
The oxidation of CO on the RuO2(110) surface has developed
into a new model system for catalytic reactions on oxide
surfaces. This breakthrough in understanding came with papers
by Over et al. in 20001 and Fan et al. in 2001,2 following
pioneering work by Bo¨ttcher et al. in the late 1990s.3-6 Initially,
the interest in this system was stimulated by the unique catalytic
properties of Ru, and it turned out that these are caused by the
formation of an RuO2 film at high oxygen pressures. However,
Figure 1. Structure model of the stoichiometric RuO2(110) sur-
RuO2 has attractive features, such as metallic conductivity and
the simple way in which single crystal, (110) oriented films
can be prepared by surface oxidation of Ru(0001), that make it
suitable for studying catalysis on oxides in general. In this paper,
we present scanning tunneling microscopy (STM) experiments
that were performed during reaction between CO molecules and
O atoms on RuO2(110) and that aimed at an atomic scale
analysis of the reaction mechanism. Previously published STM
data of adsorbed CO and O mainly focused on the identification
of the various species and on their adsorption sites.1,7-9
The essential structure elements of the stoichiometric RuO2-
(110) surface (Figure 1) are [001]-oriented rows of O atoms
(“O-bridge”), which reside on bridge positions with respect to
the Ru atoms of the next layer, and rows of “coordinatively
unsaturated Ru sites” (“Ru-cus”) between.1,10-12 The equivalent
Ru atoms in the bulk have one O atom on top, whereas at the
surface they carry one perpendicular dangling bond each,
rendering them a prominent role in the surface chemistry of
RuO2(110).
face. Indicated are the rows of topmost O atoms on bridge positions
(“O-bridge”) and the coordinatively unsaturated Ru sites (“Ru-cus”).
not stable, the CO-cus reacts with a neighboring O-bridge atom
to give CO2, which desorbs instantly. The created O-bridge
vacancy can then be filled by another CO molecule, to give
“CO-bridge”. In this way, all O-bridge atoms can be reacted
off and replaced by CO-bridge molecules. An O2 molecule
adsorbing on the surface that has before reacted with CO and
is covered by CO-bridge first dissociates at the dangling bonds
of a pair of Ru-cus atoms to form two “O-cus”. The O-cus atoms
can then react with neighboring CO-bridge molecules to give
CO2, the vacancies on the bridge sites can be occupied by further
O atoms, which restores the original structure of the stoichio-
metric surface. There are also conditions, where O-cus and CO-
cus can exist simultaneously, in which case these two species
directly react with each other.
This reaction scheme was mainly derived from titration
experiments, in which CO was dosed on an O covered surface,
or vice versa. For steady-state conditions, there are indica-
tions19,20 that two reaction channels play a prominent role,
namely the reaction between CO-cus molecules and O-bridge
atoms and between CO-cus and O-cus.
The reaction between CO and O on RuO2(110) is currently
understood as follows.1,2,7,9,13-18 A CO molecule that adsorbs
on the stoichiometric surface first binds to the dangling bond
of an Ru-cus atom to form “CO-cus”. At 300 K, this species is
In the present study, we have investigated these two processes
by means of STM. The experiments were performed as titrations,
by dosing CO on the stoichiometric surface, and by dosing O2
on the CO-cus covered surface. We focus on microscopic
questions that have remained open: What is the spatial
distribution of the reaction? How mobile are the adsorbed CO
† Part of the special issue “Gerhard Ertl Festschrift”.
* To whom correspondence should be addressed.
‡ Fritz-Haber-Institut der Max-Planck-Gesellschaft.
§ Present address: Materials Sciences Division, Lawrence Berkeley
National Laboratory, University of California, Berkeley, CA 94720.
| Ludwig-Maximilians-Universita¨t Mu¨nchen, Department Chemie.
10.1021/jp047600v CCC: $27.50 © 2004 American Chemical Society
Published on Web 07/30/2004

14566 J. Phys. Chem. B, Vol. 108, No. 38, 2004
Kim and Wintterlin
molecules and O atoms during the reaction? These spatial effects
are essential ingredients for a complete description of the
reaction mechanism. STM investigations of other surface
reactions, e.g., of the oxidation of CO on Pt(111),21 have shown
that the adsorbed reactants are not necessarily randomly
distributed, but can cluster in islands because of adsorbate-
adsorbate interactions, and that such effects can dominate the
reaction kinetics. For adsorbed CO and O on RuO2(110) DFT
calculations revealed adsorbate-adsorbate interactions of some
0.1 eV,15,20 i.e., an order of magnitude smaller values than the
respective adsorption energies. However, such energies are still
large compared to kT at 300 K and hence possibly relevant for
the surface distribution of the reactants. Nonrandomness may
also be caused by reaction at defects.22 For rutile TiO2(110),
which is structurally identical to RuO2(110), an important role
of defects has indeed been observed.23
Figure 2. High-resolution image of the stoichiometric RuO2(110)
surface. Tunneling voltage V ) -0.6 V, tunneling current I ) 2.2 nA,
area 27 Å × 40 Å. (V is the sample potential with respect to the potential
of the tip; same for the other STM data.) The rectangles in the image
and in the top view model mark the unit cell of the stoichiometric RuO2-
(110) surface.
2. Experimental Section
The STM experiments were performed in an ultrahigh
vacuum (UHV) system that was additionally equipped with
Auger electron spectroscopy, optics for low energy electron
diffraction, and an ion gun for sputtering. The STM was of the
so-called “pocket size” type and was described in an earlier
publication.24 The base pressure in the UHV system was below
1 × 10-10 Torr. The Ru(0001) surface was prepared by
sputtering/oxidation/annealing cycles as described elsewhere.25
RuO2(110) films were prepared by dosing of 2 × 105 L or 1 ×
106 L O2 (1 L ) 1 × 10-6 Torr × s) on the Ru(0001) surface,
at sample temperatures between 700 and 850 K.1,2 The O2 dosing
was performed by means of a glass capillary gas shower that
allowed to locally enhance the pressure in front of the sample
to an estimated value of 10-3 Torr, whereas the pressure in the
chamber could be kept at 1.2 to 2 × 10-5 Torr. After preparation
of the oxide film the sample was transferred into the STM.
In the STM experiments, first a suitable area of the clean
RuO2(110) surface was chosen. Then either small amounts of
the respective gases were dosed in a step-by-step fashion, while
the STM tip was withdrawn during each dosing step, or constant
pressures of CO or O2 were adjusted with the tip in tunneling
distance. Data recorded under the latter conditions were included
only when shadow effects of the tip could be ruled out, which
was achieved by control experiments, in which the tip was
retracted. STM images were recorded at room temperature.
Figure 3. Series of STM images, recorded directly after dosing 0.3 L
CO on the stoichiometric RuO2(110) surface. All frames are from the
same surface area. (a) Stoichiometric RuO2(110) surface, (b) after 30
s, (c) after 90 s, (d) after 210 s. Bright rows are O-bridge, very bright
dots occurring in (b) are interpreted as CO-cus molecules, dark sites
as O-bridge vacancies created by the reaction (see arrows). V ) -0.6
V, I ) 2.2 nA, 35 Å × 35 Å.
3. Results
3.1. Reaction of CO-cus with O-Bridge on the Stoichio-
metric RuO2(110) Surface. STM images of the stoichiometric
RuO2(110) surface showed bright rows, which, for high-
resolution conditions, displayed an internal structure of close
packed, bright oval spots (Figure 2). Similar data were obtained
in previous STM investigations.1,7 The period along the rows
is approximately 3 Å, the distance between two rows about 6.5
Å, in good agreement with the dimensions of the unit cell of
the RuO2(110) surface (3.11 Å × 6.38 Å). The rows in the STM
data therefore reflect the row structure of the RuO2(110) surface
(Figure 1 and model in Figure 2). Although it is not possible to
directly allocate the bright oval spots to certain atoms or bonds
on the surface, calculations of the EF-LDOS have indicated that
they represent the O-bridge atoms.8 In this respect, the RuO2
surface is different from the TiO2 surface, where the O atoms
appear dark.26 The sample bias did not affect the contrast of
the images. Sometimes, a contrast inversion was observed,
which was mostly correlated with an unstable state of the tip,
or occurred when the tip state changed during an adsorption
experiment.
First, we present data that allow to verify the reaction scheme
between the stoichiometric RuO2(110) surface and CO. The
images of Figure 3 were recorded directly after dosing a small
amount (0.3 L) of CO on the stoichiometric surface and are all
from the same area of the surface. Figure 3a shows the row
structure of the stoichiometric surface, the bright spot in the
lower left corner is a defect that remained unchanged in the
series and can serve as a marker of the scanning area. In the
image recorded 30 s after the exposure (Figure 3b) one dark
site appeared that is interpreted as an O-bridge vacancy created
by the reaction (see arrow). In addition, three very bright spots
appeared, located on sites between the bright rows. These are
the positions of the Ru-cus atoms, suggesting that the features
are three CO molecules, still bonded to the dangling bonds of
the Ru-cus atoms between the unreacted O-bridge rows. The

Oxidation of CO on RuO2(110) by STM
J. Phys. Chem. B, Vol. 108, No. 38, 2004 14567
Figure 5. STM images recorded during adsorption of O2 on the CO
saturated surface. Bright dots are CO-cus molecules in a 2-fold periodic
arrangement along [001], dark sites in (a) are vacancies. The circle
marks the development of one vacancy. PO2 ) 2 × 10-8 Torr; the time
elapsed after the start of the exposure is indicated. V ) 0.6 V, I ) 2.2
nA, 160 Å × 180 Å.
(Figure 4b, recorded after 1.4 L CO). This observation suggests
the operation of some correlation effects for the reaction
perpendicular to the O-bridge rows, e.g., some destabilization
of O-bridge atoms on the rows parallel to a reacted row, but
this trend is weak and the general appearance of the distribution
is still largely a random one. There was also no correlation of
the reaction sites with steps or other defects.
To draw conclusions about reaction sites the mobility of
O-bridge sites has to be known, because subsequent hopping
events after the reaction could smear out the original distribution.
However, successively recorded images of partially reacted
surfaces show an almost static distribution; the occasional
hopping events correspond to a hopping rate of the order of
magnitude of 10-5 s-1. Assuming a pre-exponential factor of
1013 s-1 this value corresponds to an activation barrier for the
diffusion of O-bridge atoms of E/diff,O-bridge ) 1.0 eV, which
is not unreasonable considering the calculated value of
E/diff,O-bridge ) 1.2 eV.7 We conclude that the O-bridge atoms
are immobile on the time scale of the reaction and that the
distribution of vacancies evidences a statistical reaction prob-
ability.
3.2 Reaction of O-cus with CO-cus on the CO-Bridge
Covered RuO2(110) Surface. The second reaction, between
O-cus and CO-cus, was investigated by adsorption of O2 on
the CO covered surface. On the stoichiometric surface, a CO-
cus layer cannot be prepared at 300 K, because CO-cus would
react with O-bridge. However, as was shown previously, CO-
cus can be stabilized (with respect to desorption)9 if the surface
has before reacted with CO and is saturated with CO-bridge
molecules. At 300 K the CO-cus molecules occupy every other
Ru-cus atom (ΘCO-cus at saturation is 0.5), which creates a
characteristic chain structure with a 2-fold period in [001]
direction in the STM images (Figure 5a). The bright dots have
been shown to be due to the CO-cus molecules,9 the dark sites
can thus be interpreted as CO-cus vacancies.
Figure 4. STM images, recorded after exposures of 0.8 and 1.4 L of
CO. (a) and (b) are from different surface areas. The data show the
distribution of O-bridge vacancies (dark sites), created by reaction with
CO, which is completely random in (a) and shows some preferential
direction in (b) (arrows). (a) V ) 0.6 V, I ) 2.2 nA, 115 Å × 65 Å.
(b) V ) -0.6 V, I ) 2.2 nA, 195 Å × 120 Å.
interpretation as CO-cus species is supported by the fact that
they appear very bright in the STM images, similar to the well
characterized, more stable CO-cus species that exists on the
completely reacted surface, between rows of CO-bridge mol-
ecules.9 90 s after the exposure (Figure 3c) one of the bright
spots disappeared, and a further dark site appeared at one of
the neighboring bright rows (arrow). After 210 s, a second bright
spot disappeared, and again a dark site appeared at a neighboring
bright row (arrow). These observations are fully consistent with
the above-described reaction mechanism. Obviously, the series
shows the reaction between CO-cus and O-bridge to give CO2,
which desorbed and left O-bridge vacancies behind, which were
imaged as dark sites. Because the reaction was directly
monitored, ambiguities about the interpretation of the various
features in the STM data can be ruled out here.8
The distribution of reacted sites on larger areas is shown in
Figure 4. In Figure 4a, recorded after adsorption of 0.8 L of
CO and showing a surface area with terraces, the dark sites do
not show any order, such as a regular distribution along the
rows, or any clustering. The distribution indicates that the
reaction between CO-cus molecules and O-bridge atoms occurs
at random positions along the O-bridge rows. Sometimes some
row formation of dark sites in [1h10] direction was observed
In the experiment (Figure 5), the surface was first saturated
with CO and then exposed to a constant O2 pressure; the series
shows the same surface area. It is seen that on the whole area,

14568 J. Phys. Chem. B, Vol. 108, No. 38, 2004
Kim and Wintterlin
Figure 7. Successive STM images of the partially CO-cus covered
surface (ΘCO-cus ≈ 0.25), the bright dots represent CO-cus molecules.
Some locations are indicated where CO molecules have hopped. V )
-0.6 V, I ) 2.2 nA. 63 Å × 63 Å.
is 4.5 ( 0.8 × 10-3 s-1, which, with a pre-exponential factor
of 1013 s-1, corresponds to an activation energy of E/react ) 0.9
eV. This value agrees well with calculated activation energies,
for which values of 0.62 eV27 and 0.89 eV20 have been reported.
Also in this case, the mobility of the preadsorbed reactant
(CO-cus) was determined (the bridge sites were completely
covered by CO). Successively recorded STM images (Figure
7, with ΘCO-cus ) 0.22) showed several hopping events per
frame. The hopping rates depended on ΘCO-cus and varied
between 90 and 400 × 10-5 s-1 (the maximum occurred at
approximately ΘCO-cus ) 0.25), but on the average CO-cus is
2 orders of magnitude more mobile than O-bridge at 300 K.
By again assuming Arrhenius behavior and a preexponen-
tial factor of 1013 s-1 an activation energy of approximately
E/diff,CO-cus ) 0.9 eV is obtained, in agreement with the
published calculated value of 0.9 eV.15 The CO-cus hopping
rate and the activation energy are hence similar to the above
obtained values for the CO-cus + O-cus reaction, and it is
possible that the CO-cus distribution can change during the
reaction. Closer inspection of the reaction series in Figure 5
indeed shows that several changes in the CO-cus configuration
occurred. Such rearrangement processes probably play some role
for the observed kinetics after the nucleation period. A reaction
order of unity means that each CO-cus molecule has the same
reaction probability, which, because of the 2-fold period of the
molecules, requires fluctuations along the chains to create two
neighboring empty Ru-cus atoms for the dissociative adsorption
of O2.
Figure 6. Analysis of the reaction between CO-cus and O-cus. (a)
Time evolution of ΘCO-cus. (b) Double logarithmic plot of the data from
(a).
the bright dots decreased and black and gray sites appeared.
The circle shows the evolution of the area around one vacancy.
After 76 s (Figure 5b), additional black sites had formed next
to the vacancy, indicating that CO-cus molecules at these sites
were preferentially reacted off, so that new vacancies were
created. After 136 s (Figure 5c) some of the black sites that
first appeared around the vacancy turned gray.
The initial reaction at vacancies can be easily explained by
the established reaction model according to which an O2
molecule first dissociates before it reacts. The entire process
therefore requires sites with at least two neighboring empty Ru-
cus atoms, which in the 2-fold periodic CO-cus chains exist at
the vacancies (three empty Ru-cus atoms), but not in the chains
themselves (one empty Ru-cus atom). The gray features can be
explained by O atoms adsorbed on empty Ru-cus atoms that
were created by the reaction.
After the nucleation period the reaction does no longer show
spatial correlation effects. In Figure 5c, the surface has started
to form black and gray sites also on other parts, but there were
also many defects before that may have acted as nucleation
centers. At this stage, the reaction is mainly random.
By quantitative analysis of the entire time series, the reaction
kinetics could be extracted. The plot of the coverage of CO-
cus molecules vs exposure time (Figure 6a) shows a kink at
approximately 70 s, after which ΘCO-cus decreases more quickly.
This change of the reaction rate is apparently connected with
the transition from the nucleation stage at defects to the random
reaction on the entire surface. A double logarithmic plot of the
rate after the kink vs ΘCO-cus shows a straight line with slope
1.0 ( 0.3 (Figure 6b), indicating that the reaction follows a
first-order kinetics with respect to ΘCO-cus. For a Langmuir-
Hinshelwood reaction mechanism between O-cus and CO-cus
4. Discussion
The STM data show that both investigated reactions on the
RuO2(110) surface, the reaction between O-bridge and CO-cus
and the reaction between O-cus and CO-cus, are largely random
processes, i.e., each particle has more or less the same reaction
probability. Only in the initial stage of the latter reaction, when
the surface is saturated with CO-cus molecules, do vacancies
play a role, which can be simply explained by the site pairs
required for O2 dissociation. With this exception, the RuO2-
(110) surface is thus a surprisingly good example of the
Langmuirian checker board model, according to which catalyst
surfaces consist of equal sites that are statistically occupied by
the reactants. By contrast, reactions on metal surfaces often show
considerable deviations from this model, in particular because
the adsorbed particles form islands, and reactions take place at
the perimeters of the islands.21 Such effects appear largely absent
here, although calculations revealed appreciable interactions
between adsorbed O atoms and CO molecules.15,20 What is the
reason for this different behavior? An obvious difference
between the RuO2(110) surface and the metal surfaces is the
one expects a second-order rate law according to -dΘCO-cus
/
dt ) k‚ΘCO-cus‚ΘO-cus (which is based on the assumption that
the two reacting species are randomly distributed). However,
at the O2 pressure of the experiment the impingement rate of
oxygen was much greater than the reaction rate (the time to
build up one monolayer of O-cus atoms was 25 s, given a
sticking coefficient of unity), so that the equation may be
simplified to a pseudo-first-order kinetics with respect to
Θ
CO-cus. This result is in agreement with the experimental
finding. The kinetics is thus consistent with the impression of
a random reaction probability in the second stage of the reaction.
The pseudo-first-order rate constant k, obtained from these data,

Oxidation of CO on RuO2(110) by STM
J. Phys. Chem. B, Vol. 108, No. 38, 2004 14569
E/diff,CO-cus ) 0.9 eV was estimated, for the reaction between
CO-cus and O-cus E/react ) 0.9 eV).
much more directed chemical bonds on the oxide. Diffusion
barriers are therefore of the same size as the activation energies
for the reaction. Our estimated diffusion barrier of 1.0 eV for
the O-bridge atoms can be related to calculated barriers for the
reaction between O-bridge and CO-cus, for which values
between 0.7 and 1.3 eV have been reported,17,18,20 and the
estimated diffusion barrier of 0.9 eV for the CO-cus molecules
has the same value as the barrier for the reaction between O-cus
and CO-cus. Hence, the adsorption layer on RuO2(110) is almost
static during the reaction (apart from the local fluctuations within
the CO-cus chains), so that lateral interactions do not lead to
the large scale rearrangement effects occurring in the very
mobile adsorbate layers on metal surfaces.
Acknowledgment. S. H. K. thanks the German Academic
Exchange Service (DAAD) for financial support.
References and Notes
(1) Over, H.; Kim, Y. D.; Seitsonen, A. P.; Wendt, S.; Lundgren, E.;
Schmid, M.; Varga, P.; Morgante, A.; Ertl, G. Science 2000, 287, 1474.
(2) Fan, C. Y.; Wang, J.; Jacobi, K.; Ertl, G. J. Chem. Phys. 2001,
114, 10 058.
(3) Bo¨ttcher, A.; Niehus, H.; Schwegmann, S.; Over, H.; Ertl, G. J.
Phys. Chem. B 1997, 101, 11 185.
(4) Bo¨ttcher, A.; Niehus, H. J. Chem. Phys. 1999, 110, 3186.
(5) Bo¨ttcher, A.; Niehus, H. Phys. ReV. B 1999, 60, 14396.
(6) Bo¨ttcher, A.; Rogozia, M.; Niehus, H.; Over, H.; Ertl, G. J. Phys.
Chem. B 1999, 103, 6267.
(7) Over, H.; Seitsonen, A. P.; Lundgren, E.; Schmid, M.; Varga, P. J.
Am. Chem. Soc. 2001, 123, 11 807.
(8) Over, H.; Seitsonen, A. P.; Lundgren, E.; Schmid, M.; Varga, P.
Surf. Sci. 2002, 515, 143.
(9) Kim, S. H.; Paulus, U. A.; Wang, Y.; Wintterlin, J.; Jacobi, K.;
Ertl, G. J. Chem. Phys. 2003, 119, 9729.
(10) Kim, Y. D.; Seitsonen, A. P.; Over, H. Surf. Sci. 2000, 465, 1.
(11) Kim, Y. D.; Seitsonen, A. P.; Wendt, S.; Wang, J.; Fan, C.; Jacobi,
K.; Over, H.; Ertl, G. J. Phys. Chem. B 2001, 105, 3752.
(12) Reuter, K.; Scheffler, M. Phys. ReV. B 2001, 65, 035 406.
(13) Wang, J.; Fan, C. Y.; Jacobi, K.; Ertl, G. Surf. Sci. 2001, 481, 113.
(14) Kim, Y. D.; Seitsonen, A. P.; Over, H. Phys. ReV. B 2001, 63,
115 419.
5. Conclusions
Two reactions that are expected to play an important role in
the oxidation of CO on the RuO2(110) surface were investigated
by time-resolved STM experiments at 300 K.
The reaction of CO molecules with O-bridge atoms on the
stoichiometric surface was monitored on the atomic scale and
was found to proceed according to the established reaction
model. The CO molecules first adsorb on Ru-cus atoms, and
then react with neighboring O-bridge atoms, creating vacancies
in the O-bridge rows. The reaction largely occurs at random
positions, defects do not play a role, and there are no significant
contributions of islands of the reactants. The statistical nature
of the reaction was explained by the large diffusion barrier of
the O-bridge atoms (a value of E/diff,O-bridge ) 1.0 eV was
estimated), which is similar to calculated activation barriers for
the reaction, so that the configuration is largely static during
the reaction.
(15) Seitsonen, A. P.; Kim, Y. D.; Knapp, M.; Wendt, S.; Over, H. Phys.
ReV. B 2001, 65, 035 413.
(16) Kim, Y. D.; Over, H.; Krabbes, G.; Ertl, G. Top. Catal. 2001, 14,
95.
(17) Liu, Z.-P.; Hu, P.; Alavi, A. J. Chem. Phys. 2001, 114, 5956.
(18) Wendt, S.; Seitsonen, A. P.; Kim, Y. D.; Knapp, M.; Idriss, H.;
Over, H. Surf. Sci. 2002, 505, 137.
(19) Wang, J.; Fan, C. Y.; Jacobi, K.; Ertl, G. J. Phys. Chem. B 2002,
106, 3422.
The reaction of CO-cus with O-cus was investigated by O2
titration on the CO-cus covered surface. On the saturated surface,
where CO occupies every other Ru-cus atom, the reaction starts
at CO-cus vacancies, because the O2 molecules need two
neighboring sites for dissociation. In the later stage, the reaction
is mainly random, because the CO-cus molecules display some
mobility on the partially reacted surface, allowing O2 molecules
to adsorb. Consistently with the largely random reaction
probability in this stage, a reaction order of 1 with respect to
(20) Reuter, K.; Scheffler, M. Phys. ReV. B 2003, 68, 045 407.
(21) Wintterlin, J.; Vo¨lkening, S.; Janssens, T. V. W.; Zambelli, T.; Ertl,
G. Science 1997, 278, 1931.
(22) Henrich, V. E.; Cox, P. A. The Surface Science of Metal Oxides;
Cambridge University Press: Cambridge, 1996.
(23) Schaub, R.; Thostrup, P.; Lopez, N.; Lægsgaard, E.; Steensgaard,
I.; Nørskov, J. K.; Besenbacher, F. Phys. ReV. Lett. 2001, 87, 266 104-1.
(24) Brune, H.; Wintterlin, J.; Trost, J.; Ertl, G.; Wiechers, J.; Behm,
R. J. J. Chem. Phys. 1993, 99, 2128.
(25) Trost, J.; Zambelli, T.; Wintterlin, J.; Ertl, G. Phys. ReV. B 1996,
54, 17 850.
(26) Diebold, U.; Anderson, J. F.; Ng, K.-O.; Vanderbilt, D. Phys. ReV.
Lett. 1996, 77, 1322.
(27) Seitsonen, A. P.; Over, H. High Performance Computing in Science
and Engineering in Munich 2002; Springer: New York, 2003.
ΘCO-cus was obtained. Island effects do not play a role, which
is again explained by a relatively large diffusion barrier
compared to the reaction barrier (for CO-cus a value of