Structure sensitivity in the CO oxidation on rhodium: Effect of adsorbate coverages on oxidation kinetics on Rh(100) and Rh(111)

Article abstract of DOI:10.1063/1.1289764

Temperature-programmed reaction spectroscopy was used to study the surface reaction between carbon monoxide (CO) and oxygen (O) atoms on rhodium (Rh) at a range of different adsorbate coverages. The kinetic was described by Langmuir-Hinshelwood mechanism and is structure sensitive. The result obtained was a greatly enhanced rate of reaction at higher coverages.

Full text of DOI:10.1063/1.1289764

Structure sensitivity in the CO oxidation on rhodium: Effect of adsorbate coverages on
oxidation kinetics on Rh(100) and Rh(111)
M. J. P. Hopstaken and J. W. Niemantsverdriet
Citation: The Journal of Chemical Physics 113, 5457 (2000); doi: 10.1063/1.1289764
View online: http://dx.doi.org/10.1063/1.1289764
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/113/13?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Simulation of the effect of surface-oxide formation on bistability in CO oxidation on Pt-group metals
J. Chem. Phys. 126, 074706 (2007); 10.1063/1.2483966
Kinetic oscillations in the catalytic CO oxidation on Pt(100) with adsorbed impurities
J. Chem. Phys. 113, 10353 (2000); 10.1063/1.1322656
High density adsorbed oxygen on Rh(111) and enhanced routes to metallic oxidation using atomic oxygen
J. Chem. Phys. 110, 2757 (1999); 10.1063/1.477877
Effect of inert sites on the kinetic oscillations in the catalytic CO oxidation on Pt(100)
J. Chem. Phys. 109, 8617 (1998); 10.1063/1.477528
Kinetic phase diagram for CO oxidation on Pt(210): Pattern formation in the hysteresis and oscillation regions
J. Chem. Phys. 106, 4291 (1997); 10.1063/1.473131
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
141.100.75.156 On: Tue, 25 Nov 2014 09:54:17

JOURNAL OF CHEMICAL PHYSICS
VOLUME 113, NUMBER 13
1 OCTOBER 2000
Structure sensitivity in the CO oxidation on rhodium: Effect of adsorbate
coverages on oxidation kinetics on Rh„100… and Rh„111…
a)
M. J. P. Hopstaken and J. W. Niemantsverdriet
Schuit Institute of Catalysis, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven,
The Netherlands
͑
Received 14 April 2000; accepted 3 July 2000͒
Temperature-programmed reaction spectroscopy has been used to study the surface reaction
between CO and O-atoms on Rh͑100͒ and Rh͑111͒ at a range of different adsorbate coverages.
Comparison of the reaction on both surfaces in the low coverage regime, where the kinetics can be
described by a straightforward Langmuir–Hinshelwood mechanism reveals that the CO oxidation is
structure sensitive, with the rate constant being an order of magnitude higher on the Rh͑100͒ than
on the Rh͑111͒ surface. As a consequence, the selectivity of the COϩO reaction to CO is about
2
100% on Rh͑100͒, whereas on Rh͑111͒ the oxidation reaction competes with CO desorption. At low
CO coverage, CO oxidation is an elementary step on Rh͑100͒ for a broad range of oxygen
1
2.7Ϯ0.7
coverages. We report kinetic parameters E ϭ103Ϯ5 kJ/mol and ␯ϭ10
for ␪ ϭ␪ →0 on
a
O
CO
Rh͑100͒. The activation energy for CO oxidation on Rh͑100͒ decreases continuously with
increasing O-coverage. At low coverage (␪ Ͻ0.25 ML͒ we attribute this to destabilization of CO,
O
leading to an increase in the CO formation rate. At higher coverage (␪ Ͼ0.25 ML͒ O-atoms
2
O
become destabilized as well, as lateral interactions between O-atoms come into play at these
coverages. The interactions result in a greatly enhanced rate of reaction at higher
coverages. © 2000 American Institute of Physics. ͓S0021-9606͑00͒70237-6͔
I. INTRODUCTION
der into islands and reactions may even occur in the oscilla-
tory regime, has received considerable interest from
3
25
The oxidation of CO on noble metals such as
experimentalists and theoreticians in the past decade. Re-
cent scanning tunneling microscopy ͑STM͒ work by Wintter-
lin et al. points to the importance of reactive CO–O pairs at
the boundaries between islands of adsorbed CO and O on
1
–3
4–20
21–23
platinum,
rhodium,
palladium,
and
6
,21,24
ruthenium
is one of the most studied catalytic reactions
in surface science. In the context of automotive exhaust ca-
talysis the CO oxidation is important in two respects: first, in
the reaction between O and CO, which is most efficiently
Pt͑111͒.26 On rhodium, the reacting species O and COads
ads
4
2
may either segregate into separate phases, or form mixed
structures,2
7,28
depending on coverages and on temperature.
catalyzed by platinum and palladium, and second, in the re-
moval of atomic oxygen originating from NO decomposi-
tion, for which rhodium and to a lesser extent palladium are
the best catalysts. CO oxidation has been studied on different
Several studies demonstrate that different oxygen struc-
tures on rhodium surfaces possess different reactivity toward
adsorbed CO,1
5,16,29–31
whereas reconstruction has also been
rhodium surfaces, e.g., Rh͑111͒,4
–13
Rh͑110͒,
polycrystalline Rh, and supported-Rh
as well.
11,14,15
15
16
shown to affect the reactivity on Rh͑110͒. Baraldi et al.,
6
,7,10,16–18
19
Rh͑100͒,
particles
investigating the reaction between CO and preadsorbed oxy-
gen on Rh͑100͒, observed a higher reactivity for the recon-
8
,20
8
10
Kinetic studies by Oh et al. and Peden et al. revealed
that turnover frequencies of the CO oxidation at high pres-
sures on rhodium single crystals and supported particles are
very similar, indicating this reaction to be structure insensi-
tive. However, under these conditions the surface is largely
covered by CO and the rate is limited by the dissociative
adsorption of oxygen. Kinetic studies at conditions where
oxygen is the majority reactive intermediate, however, do
reveal rates that depend on structure, as shown by Bowker
et al. for Rh͑111͒ and Rh͑110͒.11 Several authors report ki-
netic parameters for the CO oxidation on Rh͑111͒, however,
these values differ widely, e.g., from activation energies of
structed (2ϫ2)p4g O-structure (␪ ϭ0.50 ML͒ than for the
O
3
2
p(2ϫ2)-O structure (␪ ϭ0.25 ML͒. The lower reactivity
O
of the latter was explained by reduced mobility of CO, pre-
venting it to overcome the barrier imposed by CO–O repul-
sion.
The strong dependence of mechanism and kinetics on
adsorbate coverages is an important factor behind the large
differences in activation energies and preexponential factors
as reported in the literature. In fact, such parameters are only
meaningful for elementary surface reactions in the limit of
low coverages of at least one of the adsorbates.3
3,34
The purpose of this paper is to determine the kinetic
parameters of the reaction between preadsorbed CO and O
on Rh͑100͒ and Rh͑111͒ by means of temperature-
programmed reaction spectrometry. The results confirm the
structure sensitivity of the CO oxidation reaction, and dem-
onstrate in a systematic way the effect of adsorbate cover-
ages on the rate of the reaction.
13
7
0 kJ/mol to 190 kJ/mol.4
The kinetics of surface reactions under high coverage
conditions, where lateral interactions cause adsorbates to or-
a͒Corresponding author: Phone: ϩ31 40 247 3067, Fax: ϩ31 40 245 5054,
E-mail: J.W.Niemantsverdriet@tue.nl
0021-9606/2000/113(13)/5457/9/$17.00
5457
© 2000 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
41.100.75.156 On: Tue, 25 Nov 2014 09:54:17
1

5458
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
M. J. P. Hopstaken and J. W. Niemantsverdriet
II. EXPERIMENT
The experiments were done in two stainless steel ultra-
high vacuum systems with a base pressure of 10Ϫ10 mbar.
The experiments with Rh͑100͒ were done in a Leybold TPD/
SIMS spectrometer equipped with a SSM 200 mass spec-
trometer system as previously described by Borg et al.35 The
experiments on Rh͑111͒ were done in a home built UHV
system, equipped with a Baltzers Prisma quadrupole mass
spectrometer ͑QMA 200͒ and an Omicron LEED/Auger sys-
3
6
tem, described in detail in a previous publication. Some of
the experiments with Rh͑100͒ were repeated in the second
system to verify that the temperature calibration in both sys-
tems was the same.
In both systems, the rhodium crystals of ͑100͒ and
Rh͑111͒ orientation with a thickness of about 1.2 mm were
spotwelded on two tantalum wires of 0.3 mm diameter,
pressed into small grooves on the side of the crystal. This
construction allows for resistive heating up to 1450 K. The
samples could be cooled to 100 K by flowing liquid nitrogen
through the manipulator. Temperatures were measured with
a chromel-alumel thermocouple spotwelded to the back of
the crystal.
FIG. 1. Kinetic plots obtained from coverage-corrected leading edge analy-
sis for CO oxidation on Rh͑100͒ and Rh͑111͒ at equal reactant coverages.
Kinetic parameters can be derived by plotting the natural logarithm of the
reaction rate constant (rox /(␪COϫ␪O)) vs reciprocal temperature. The acti-
vation energy Ea and preexponential factor ␯ can be determined from the
slope (ϪEa /R) and the intercept ͑ln͑␯͒͒, respectively. The inset shows the
The Rh͑100͒ crystal surface was cleaned by cycles of
argon sputtering and annealing under oxygen and UHV, as
described elsewhere.37 The Rh͑111͒ crystal surface was
corresponding CO ϪTPD spectra for Rh͑100͒ and Rh͑111͒.
2
cleaned in a similar fashion at slightly higher temperatures,
as described previously.13 After the cleaning procedure, re-
sidual oxygen was removed and surface order was restored
by annealing the crystals to 1400 K. Surface cleanliness and
ordering were checked by Auger spectroscopy, which did not
show any contamination and LEED, which showed sharp
p(1ϫ1) patterns for both crystals. Also, the absence of CO2
during TPD of CO confirms the complete removal of re-
sidual oxygen on both crystals.
least at low coverages, first-order in adsorbed O and CO,
while CO desorbs instantaneously. Provided both reactants
2
are homogeneously distributed over the surface, the rate of
CO oxidation rox is given by the following Arrhenius-
equation:
Ea
roxϭvϫ␪COϫ␪ ϫexp
ͩ
Ϫ
ͪ
,
͑1͒
O
RT
Oxidation of CO to CO was studied by depositing a
2
fixed amount of atomic O at 273 K after which the sample
was cooled to 175 K or below to adsorb CO. Carbon mon-
which can be rewritten
rox
Ea
oxide ͑Hoek Loos, 99.995% pure͒ and oxygen ͑20% O /80%
2
ln
ͩ
ͪ
ϭln ␯Ϫ
.
͑2͒
Ar͒ were used without further purification. The initial cover-
ages of CO were determined from the amount of desorbing
␪COϫ␪O
RT
The adsorbate coverages ␪ and ␪ can be calculated at any
temperature from the CO and CO TPD spectra. A plot of
CO
O
carbon atoms, by integrating the CO and CO desorption
2
2
peaks, dividing the CO peak area by the relative difference in
ln(r /(␪ ϫ␪ ) versus 1/T yields the activation energy Ea
ox
CO
O
mass spectrometer sensitivity for CO and CO ͑ϭ1.1͒, and
2
and the preexponential factor ␯ from the slope and the inter-
cept, respectively.
adding these contributions. For the saturation coverage of
CO on Rh͑100͒ we used the value of 0.75 ML for the
p(4
Rh͑111͒ the value of 0.75 ML corresponding to the ͑2ϫ2͒-
ͱ2ϫͱ
2)R45°ϪCO/Rh͑100) structure36 and for CO on
III. RESULTS
8,39
3
CO/Rh͑111͒ structure.3
The CO TPD spectra were cor-
To illustrate the experimental approach and the kinetic
analysis we compare the results of CO oxidation experiments
at similar coverages on the two rhodium surfaces. Approxi-
mately 0.16 ML of O-atoms and 0.07 ML of CO were coad-
sorbed on each surface. The inset of Fig. 1 shows the TPD
rected for cracking of CO in the mass spectrometer. The
2
amount of preadsorbed oxygen on Rh͑100͒ was determined
from the amount of CO formed in TPD, since the removal
2
of Oads appeared to be complete in the presence of an equal
or higher amount of CO. On Rh͑111͒ removal of Oads is not
always complete, so the amount of oxygen was determined
spectra of CO . The structure sensitivity of the reaction is
2
evident from the significantly faster rate of reaction on
Rh͑100͒ than on Rh͑111͒. A second difference between the
two systems is that on Rh͑100͒ all CO is oxidized to CO2,
whereas on Rh͑111͒ part of the CO desorbs. The main part
of Fig. 1 shows the Arrhenius plots according to Eqs. ͑1͒ and
͑2͒ over the entire temperature range of the TPD peaks. The
from O -TPD, using the saturation coverage of 0.50 ML for
2
2
7,40,41
O on Rh͑111͒.
rate ␤ϭ5 K/s.
All TPD spectra were taken at a heating
Kinetic parameters for the reaction between CO and O
were determined on the assumption that the reaction is, at
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
141.100.75.156 On: Tue, 25 Nov 2014 09:54:17

J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Oxidation kinetics of CO on rhodium
5459
FIG. 2. CO formation from different amounts of CO on oxygen-precovered Rh͑100͒. Oxygen atoms were deposited by adsorbing a O /Ar gas mixture ͑20%
2
2
O /80% Ar͒ at 273 K; a heating rate of 5 K/s was applied. Note that the desorption spectra at different O-precoverages have not been scaled with respect to
2
each other .
fact that these lines are straight for a large temperature inter-
val confirms that at these low coverages the reaction exhibits
first-order kinetics with respect to both ␪ and ␪ . Hence,
initial CO coverage; see Fig. 3. Essentially, as long as
O-atoms are available all CO is oxided on Rh͑100͒. Only at
higher oxygen coverages (␪ ϭ0.28 and 0.40 ML͒, some CO
O
CO
O
the kinetic parameters can be derived with good accuracy.
We will discuss the values of these parameters later on, when
experiments at a large range of coverages have been de-
scribed.
desorbs in parallel to CO , but also in this case the selectivity
towards oxidation is high.
2
Kinetic analysis as explained in the experimental part
was applied to those spectra which exhibit a single desorp-
tion state. Figure 4 gives an example of activation energies
and preexponential factors obtained for the series of spectra
at an oxygen precoverage of 0.09 ML. The peak shifts in the
TPD spectra to lower temperatures with increasing CO cov-
erage are reflected in a decrease of the activation energy and
the preexponential factor. Also note that the change in slope
around ␪COϭ0.1 ML corresponds with the downward shift
A. Oxidation of CO on Rh„100…
Figure 2 gives an overview of the temperature-
programmed reactions between fixed amounts of pread-
sorbed oxygen (␪ ϭ0.09 ML, 0.16 ML, 0.28 ML, and 0.40
O
ML͒, and different amounts of CO from almost zero up to
saturation coverage. At low coverages, CO forms in a single
2
peak. As the total coverage increases, several additional
channels become available for oxidation. We discuss these
later on.
of the leading edge in the CO TPD spectra ͑Fig. 2͒.
2
At ␪ ϭ0.40 ML and low CO coverages, the oxygen
O
coverage is much higher than that of CO, and the reaction
becomes pseudo first order in CO. In this case we can also
apply Chan-Aris-Weinberg ͑CAW1/2͒ analysis.4 This yields
Integration of the CO and CO TPD spectra ͑not shown͒
2
2
yield the fractions of CO reacted to CO as a function of
2
FIG. 4. Activation energy Ea ͑left axis͒ and preexponential factor ␯ ͑right
FIG. 3. Amount of COads reacted with Oads to CO on Rh͑100͒ ͑left͒ and
axis͒ for CO oxidation on Rh͑100͒ at ␪ ϭ0.09 ML as a function of initial
2
O
Rh͑100͒ ͑right͒, derived by integration of TPD spectra. The dotted lines are
representative for the ideal case of stoichiometric reaction between COads
CO-coverage. Kinetic parameters were derived using coverage-corrected
leading edge analysis as in Fig. 1. Dotted lines between experimental data
points are drawn as a guide for the eye.
and Oads
.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
41.100.75.156 On: Tue, 25 Nov 2014 09:54:17
1

5460
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
M. J. P. Hopstaken and J. W. Niemantsverdriet
for cracking of CO . At low coverages, CO is formed be-
2
2
tween 400 and 500 K. At higher coverages, a faster CO2
formation channel around 350 K becomes available, in good
agreement with earlier observations.4 This faster reaction
,27
channel is possibly associated with CO in a different binding
geometry.2
7,28
Also note the small upward temperature shift
of the CO desorption with increasing ␪CO as more O-atoms
are removed, decreasing the repulsion between COads and
Oads. The amount of CO reacted to CO has been included in
2
Fig. 3 and the activation energies and preexponential factors
for oxidation, derived in a similar way as for Rh͑100͒, in
Fig. 5.
Analyzing the spectra in terms of Eq. ͑2͒ was not suc-
cessful for the spectra at high oxygen coverage of 0.35 ML,
as the kinetic plots could not be fitted with straight lines. As
for low initial CO coverages the coverage of oxygen de-
creases only slightly during CO formation, the reaction fol-
2
lows pseudo first-order kinetics in ␪CO . This allows for the
application of Chan-Aris-Weinberg ͑CAW1/2͒ analysis,
which yields an activation energy E ϭ70Ϯ6 kJ/mol and a
a
Ϫ1
preexponential factor ␪ϭ108
.4Ϯ0.5
s
for CO oxidation in
the limit of zero CO coverage. These values are confirmed
by Redhead analysis, which yields E ϭ72 kJ/mol, assuming
a
ϭ108
.4
s
Ϫ1
(Tmaxϭ425 K, ␤ϭ5 K/s͒. This value is in-
␯
cluded in Fig. 5.
We also considered CO formation for ␪O close to the
saturation value of 0.50 ML, where O-atoms are reported to
FIG. 5. Activation energy Ea ͑upper panel͒ and preexponential factor ␯
lower panel͒ for CO oxidation on Rh͑100͒ and Rh͑111͒ in the limit of zero
CO coverage, as a function of the amount of preadsorbed O. Datapoints,
represented by filled symbols, were derived using coverage-corrected lead-
ing edge analysis and the open symbols using Chan–Aris–Weinberg analy-
sis assuming zeroth order kinetics in ␪O and first-order kinetics in ␪CO . The
2
͑
form a 3-domain ͑2ϫ1͒-structure with O-atoms in rows sepa-
rated by one Rh-lattice constant.2
7,40,41
The uptake of CO is
limited to only 0.03 ML of CO at saturation. Interestingly,
recent high-resolution core-electron spectroscopy measure-
ments by Jaworowski et al.28 point to the existence of an
dotted line at ␪ ϭ0.25 ML indicates the completion of the p(2ϫ2)-O
O
structure. Solid lines between experimental data points are drawn as a guide
for the eye.
ordered 2OϩCO phase with ␪ ϭ0.50 ML and ␪COϭ0.25
O
ML, which we have not been able to realize. At these very
high O/CO ratios, oxidation is complete as CO desorption is
an activation energy E ϭ74Ϯ8 kJ/mol and a preexponential
a
1
1Ϯ1.3 Ϫ1
factor ␯ϭ10
s
in the zero CO coverage limit. As a
hardly observed. Formation of CO occurs in a single state
2
4
3
double check, we applied the Redhead formula for ␯
around 410 K, not shifting with ␪CO . Analysis by Eq. ͑2͒
1
1Ϯ1 Ϫ1
ϭ10
s
, taking Tmaxϭ350 K and ␤ϭ5 K/s, which
yields an activation energy of 80Ϯ6 kJ/mol and a preexpo-
nential factor of 101
0Ϯ0.5 Ϫ1
s
for CO oxidation in the zero
yields an activation barrier E ϭ76Ϯ5 kJ/mol, consistent
a
with the CAW1/2 analysis.
CO coverage limit. As the CO oxidation becomes first-order
in ␪CO under these conditions, Chan–Aris–Weinberg analy-
sis becomes applicable again, which yields the lower values
The kinetic parameters obtained as in Fig. 4 were ex-
trapolated to zero CO coverage for all four initial
O-coverages; these extrapolated values are shown in Fig. 5.
Both the activation energy and the preexponential factor for
CO oxidation decrease with increasing O-coverage on
Rh͑100͒. The preexponential factors are all in accordance
with the reaction COadsϩOads being an elementary step.3
8
.5Ϯ0.5 Ϫ1
of E ϭ69Ϯ5 kJ/mol and ␯ϭ10
s .
a
As the spectra in Fig. 6 and the quantification in Fig. 3
show, oxidation of CO is accompanied by desorption of CO
at all coverages. This desorption is at lower temperatures
than observed for CO on oxygen-free Rh͑111͒. At compa-
rable coverages of O and CO, the peak temperatures for oxi-
dation on Rh͑111͒ are at least 35 K higher than on Rh͑100͒.
At high coverages, the difference becomes even larger. As a
result, the oxidation reaction on Rh͑111͒ falls largely in the
temperature range where CO desorption occurs as well, and
3,34
B. Oxidation of CO on Rh„111…
The oxidation of CO on Rh͑111͒ has been described
extensively in the literature.4 However, as the kinetic pa-
rameters reported vary considerably between different au-
thors, we prefer to determine some kinetic data on this sur-
face in the same manner as we did for Rh͑100͒, to allow
meaningful comparison.
–13
the selectivity for CO formation is significantly lower on
2
Rh͑111͒ than on Rh͑100͒.
The kinetic parameters for CO oxidation on Rh͑111͒ are
completely different than those for comparable CO- and
O-coverages on Rh͑100͒. The preexponential factors for CO
Figure 6 shows the TPD spectra of CO and CO for fixed
amounts of 0.17 and 0.35 ML O coadsorbed with varying
amounts of CO. The CO TPD spectra have been corrected
2
7
9
Ϫ1
oxidation on Rh͑111͒ are in the range of 10 –10
s ,
which is lower than normally observed for an elementary
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
141.100.75.156 On: Tue, 25 Nov 2014 09:54:17

J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Oxidation kinetics of CO on rhodium
5461
FIG. 6. TPRS spectra of CO2 ͑left͒
and CO ͑right͒, for a fixed amount of
0.16 ML O ͑top͒ and 0.35 ML ͑bot-
tom͒ coadsorbed with varying
amounts of CO (TadsϽ175 K͒. Oxy-
gen atoms were deposited by adsorb-
ing a O /N gas mixture ͑20%/80%͒ at
2
2
293 K and a heating rate of 5 K/s was
applied. CO-TPD spectra have been
corrected for cracking of CO2.
step. This may indicate that CO oxidation on Rh͑111͒ cannot
be described by a simple mean-field Langmuir–Hinshelwood
approach, as we discuss further on.
Finally, Fig. 7 shows the uptake of CO on O-precovered
Rh͑100͒ and Rh͑111͒ surfaces. On Rh͑100͒ O-atoms and CO
molecules are mutually exclusive with respect to adsorption
sites, despite the fact that CO and O are adsorbed in different
sites, as the sum of ␪O and ␪CO is constant ͑Ϸ0.82 ML͒.
However, on Rh͑111͒ the sum of the coverages decreases
significantly when the surface fills up. When the rhodium
surfaces are close to saturation with O-atoms, Rh͑100͒ can
still accommodate a stoichiometric amount of CO, whereas
on Rh͑111͒ hardly any CO can be chemisorbed.
IV. DISCUSSION
A. Structure sensitivity of the CO oxidation on
rhodium
The direct comparison of the rate constant for CO oxi-
dation over Rh͑100͒ and Rh͑100͒ in Fig. 1, as well as the
difference in kinetic parameters at different oxygen cover-
ages collected in Fig. 5, clearly illustrate the structure sensi-
tivity of the CO oxidation reaction in the regime where re-
actant coverages are low. As stated in the Introduction, the
reaction becomes apparently insensitive for catalyst structure
under conditions where CO is the majority reacting interme-
FIG. 7. Effect of O-precoverage on the total saturation coverage (␪O
diate, i.e., under steady state conditions for CO and O par-
2
ϩ␪CO)sat and the CO saturation coverage ␪CO,sat on Rh͑100͒ and Rh͑111͒.
tial pressures ranging from 1 to ϳ100 mbar at relatively low
Whereas on Rh͑100͒ the total coverage is hardly affected over the whole
coverage range, less CO can be accommodated within the O-structure at
higher O-coverage on Rh͑111͒.
temperatures ͑ϳ500 K͒. Under these conditions, the CO for-
2
mation exhibits first-order in the partial pressure of O and
2
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
41.100.75.156 On: Tue, 25 Nov 2014 09:54:17
1

5462
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
M. J. P. Hopstaken and J. W. Niemantsverdriet
negative first-order in CO.10 Here the reaction is rate-limited
by the availability of dissociatively adsorbed oxygen. Hence,
under these conditions desorption of CO is required to liber-
ate empty sites for dissociation of O . Comparison of CO
2
4
4
11,45
16,36
desorption from Rh͑111͒, Rh͑110͒,
Rh͑100͒,
and
supported Rh-particles46 shows that the rates of CO desorp-
tion are quite similar on these surfaces. So it appears that,
instead of CO oxidation, desorption of CO is structure insen-
sitive. Only weak dependence of the adsorption energy on
surface structure of a molecular adsorbate like CO is in line
with theory predictions.3
3,47
This probably explains the very
similar reaction rates, observed under high-pressure steady
state experiments. This is the regime investigation by
6
8
10
Goodman, Oh, and Peden and co-workers. Bowker et al.
studied the reaction under steady state conditions as a func-
tion of temperature on both Rh͑111͒ and Rh͑110͒.11 He ob-
served that the rates were similar at low temperature, but
different at high temperature, in full agreement with the no-
tion that COads dominates the surfaces at lower and Oads at
higher temperatures. A similar conclusion can be drawn from
the experiments by Schwartz et al. for CO oxidation on
7
Rh͑111͒ and Rh͑100͒. For temperatures higher than 600 K,
the CO formation rate on Rh͑100͒ was found to be higher
2
by one order of magnitude than on Rh͑111͒, at comparable
reactant pressures.
The major differences between the two rhodium surfaces
with respect to CO oxidation investigated here are:
•
At relatively low O-coverage the rate on Rh͑100͒ is
roughly an order of magnitude faster than on Rh͑111͒. Struc-
ture dependence becomes even more pronounced at higher
O-coverage, as the rate on Rh͑100͒ becomes about 10 faster
FIG. 8. Selected CO and CO desorption traces from Figs. 2 and 6 for equal
2
2
reactant coverages on Rh͑100͒ and Rh͑111͒. Clearly CO formation goes to
2
than on Rh͑111͒.
completion on Rh͑100͒ as only a small excess of CO is seen to desorb,
whereas on Rh͑111͒ desorption and oxidation of CO are competing pro-
cesses.
•
As long as O-atoms are available, all CO oxidizes to
CO on Rh͑100͒; on Rh͑111͒ CO desorption competes with
2
oxidation, as the rate of CO oxidation is comparable to CO
desorption on Rh͑111͒. This is demonstrated in Fig. 8. We
exclude ͑partial͒ migration of Oads into subsurface sites on
7,10
p(O )/p(CO) ratio.
The fact that CO oxidation is faster
2
on Rh͑100͒ than on Rh͑111͒ is somewhat surprising. In view
of the higher heat of adsorption of atoms on more open sur-
faces, while the heat of adsorption of CO on different sur-
faces of rhodium is not too different, one might have ex-
pected the rate of reaction to be slower on the ͑100͒ surface.
The fact that this is not so many have to do with differences
Rh͑111͒ as the cause for this lowered selectivity to CO . A
2
recent X-ray Photoemission Diffraction ͑XPD͒ study by
4
8
Wider et al. shows that subsurface sites only become oc-
cupied at elevated temperature and high O exposures, i.e.,
2
when ␪ exceeds 0.50 ML.
O
•
CO oxidation on Rh͑100͒ exhibits kinetic parameters
in the activated complex or transition state for CO forma-
1
2.7Ϯ0.7 Ϫ1
2
(
E ϭ103Ϯ5 kJ/mol, ␯ϭ10
representative of a true elementary step, whereas the CO
s
at ␪ ϭ␪ →0)
a
O
CO
tion, on which information for the case on rhodium is not
available, however. In the transition state for CO oxidation
on the closest-packed surfaces of Pt and Ru it appears that
partial loosening of the oxygen–metal bond gives the highest
contribution to the reaction energy barrier.
oxidation on Rh͑111͒ is characterized by kinetic parameters
7
.5Ϯ1 Ϫ1
(
E ϭ65Ϯ5 kJ/mol; ␯ϭ10
a
s
at ␪ ϭ0.16 ML; ␪
O CO
→
0) indicative for a more complicated reaction sequence.
The total coverage of O and CO, achieved when CO is
49–51
It is well pos-
sible that this step is energetically more favorable on the
100͒ surface, owing to the different geometry on the more
adsorbed to saturation on O-precovered surfaces is signifi-
cantly larger on Rh͑100͒ than on Rh͑111͒, as shown in Fig.
͑
open surface.
7
. Whereas the sum ␪ ϩ␪ is approximately constant and
O CO
equal to 0.82 ML on Rh͑100͒, the total coverage decreases in
monotonous fashion from 0.75 ML for fully saturated CO/
Rh͑111͒ to 0.5 ML for fully saturated O/Rh͑111͒. Appar-
ently, no CO molecules can be accommodated within the
B. Different CO formation channels on Rh„110…
2
The overview of temperature-programmed CO forma-
2
tion from COads and Oads at different initial coverages in Fig.
2 reveals several reaction channels. At low coverages of both
͑2ϫ1͒-O structure on Rh͑111͒. This is consistent with the
observation of inhibition in the CO formation on Rh͑111͒
COads and Oads, CO forms in a single peak around 450 K.
2
2
for high oxygen coverage, at higher temperature and high
Analyzing this in terms of the straightforward rate equation
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
141.100.75.156 On: Tue, 25 Nov 2014 09:54:17

J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Oxidation kinetics of CO on rhodium
5463
͑
2͒, which expresses first-order kinetics in each reactant,
1
2.7Ϯ0.7 Ϫ1
yields E ϭ103Ϯ7 kJ/mol and ␯ϭ10
s . In particu-
a
lar the preexponential has a value characteristic for an el-
3
3,34
ementary step
and hence, it appears reasonable to at-
tribute the CO reaction channel at 450 K to a reaction
2
between species that are randomly distributed across the sur-
face and encounter each other because at least one the reac-
tants, in this case CO, diffuses freely over the surface. A
previous Monte Carlo simulation of ordering phenomena of
NO on Rh͑111͒ indicated that diffusion of molecules such as
NO and CO started to become detectable at much lower tem-
peratures of 200 K already, and that diffusion of these mol-
ecules was by far the fastest surface process at temperatures
FIG. 9. Structural model for the CO oxidation on Rh͑100͒ for initial
O-coverages below 0.25 ML ͑left͒ and above ͑right͒ and low ␪CO . For ␪O
Ͻ0.25 ML, O-atoms are arranged in local p(2ϫ2)-islands and the first CO
molecules are probably linearly adsorbed on the bare metal patches, giving
rise to CO2 formation between 425 and 450 K. Upon completion of the
5
2
ordered p(2ϫ2)-O (␪ Ͼ0.25 ML͒ structure on Rh͑100͒, excess O-atoms
O
above 250 K. Adsorbed atoms diffuse much more slowly,
although mobility of O-atoms on ruthenium at room tem-
are arranged in local c(2ϫ2)-islands and the first CO molecules are ad-
sorbed in a higher bonding geometry in close proximity to Oads. This gives
5
3
perature has been demonstrated.
rise to CO formation around 350 K.
2
Considering reaction around 450 K as the reference case
for the limit of low reactant coverage, we see that increasing
the precoverage of O-atoms but keeping the CO coverage
low ͑below about 0.10 ML͒ leads to an intrinsically faster
reaction, characterized by a lower activation energy, and a
preexponential factor that changes only insignificantly ͑Fig.
that such surfaces can still accommodate about 0.25 ML of
CO.
These considerations suggest we have the following re-
action channels for Oads and COads on Rh͑100͒:
• Low coverage of Oads and COads: Reaction between
5͒. The reason that the reaction goes faster with increasing
isolated species, characterized by a CO formation peak
around 450 K.
number of oxygen atoms in the low coverage regime (␪O
Ͻ0.25 ML͒ is probably that the O-atoms enhance the elec-
trostatic potential of the surface with a long range effect that
is similar but opposite in sign as small amounts of alkali
do. As a result the CO becomes destabilized on the entire
surface. This leads to enhanced mobility and, probably also,
reactivity toward O.
2
•
␪OϽ0.25 ML: O present locally in p(2ϫ2) islands,
ads
reaction of CO adsorbed initially adsorbed on free rhodium
5
4
with O-atoms at the perimeter of these islands, CO forma-
2
tion between 400 and 450 K. At higher CO coverages, reac-
tion within the islands, between 300 and 400 K.
•
␪OϾ0.25 ML: Oads present in a complete p(2ϫ2)
In the high coverage regime (␪ Ͼ0.25 ML͒ the reaction
O
structure, with locally a c(2ϫ2) structure. At low CO cov-
rate is probably increased due to repulsive interactions be-
5
5
erage all COads is inside the p(2ϫ2) and reacts to CO be-
2
tween O-atoms, as evidenced both from TPD, single crystal
5
6
tween 300 and 400 K. At higher coverages of COads reaction
adsorption calorimetry ͑SCAC͒, and density functional
theory ͑DFT͒ calculations.5
7,58
occurs in the c(2ϫ2), with a sharp CO formation peak at
2
This leads to a decrease in the
2
80 K. We wonder if this peak is associated with the lifting
oxygen binding energy and hence a lower reaction barrier for
CO oxidation. DFT calculations for CO oxidation on noble
metal surfaces reveal that the most important part of the en-
ergy barrier for CO oxidation is caused by partially loosen-
of the well-documented clockwise–anticlockwise recon-
struction of the Rh͑100͒ surface upon saturation with
3
2,60–62
O.
The splitting up of the 350 K CO formation into a lower
4
9–51
2
ing the O-atom in the transition state for CO formation.
2
and a higher temperature CO formation channel with in-
2
This also explains the observed trend in increasing activity
for CO oxidation going from left ͑Ru͒ to right ͑Pd͒ in the
periodic table, concomitant with the decrease in the oxygen
binding energy.
creasing CO coverage as seen in the right part of Fig. 2 is
due to depletion of oxygen atoms. As ␪ decreases below the
O
critical coverage of 0.25 ML, CO molecules will segregate to
the bare metal patches, in-between the ͑2ϫ2͒-O islands and
hence reaction proceeds in the lower reactivity channels.
Oxygen atoms are known to order on Rh͑100͒ and this
strongly affects the COadsϩOads reaction at higher coverages,
as evidenced by Fig. 2. We propose the structural model of
Fig. 9 to explain some of the observed shifts in the CO2
formation. For oxygen coverages below 0.25 ML, the
Oads-atoms arrange in islands of p(2ϫ2) geometry.59 The
first CO molecules will adsorb on the free metal patches, as
CO prefers to bind linearly, as has been observed
C. Kinetic parameters CO oxidation
Kinetic parameters for CO oxidation on different
rhodium surfaces as measured by various researchers are pre-
sented in Table I. As we have stated before, there is large
scatter in the kinetic data for Rh͑111͒. This can partly be
explained by different reaction conditions. Under high pres-
sure conditions, when the surface is almost completely cov-
ered by CO and the rate is limited by the availability of O,
3
6
experimentally. When the free metal is filled, additional
CO has to penetrate the O-islands. According to Fig. 7, this
is well possible on the Rh͑100͒ surface. However, for Oads
coverages above 0.25 ML, the atoms in excess of 0.25 ML
6
–8,10
the activation energy amounts to ca. 110 kJ/mol.
Impor-
in all cases, sug-
gesting that this is indeed an elementary step. These kinetic
tantly, the preexponential factor is 1012
s
Ϫ1
occupy centered positions in the ͑2ϫ2͒ cell and at ␪ ϭ0.5
O
ML the c(2ϫ2) structure is complete.59 Figure 9 indicates
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
141.100.75.156 On: Tue, 25 Nov 2014 09:54:17

5464
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
M. J. P. Hopstaken and J. W. Niemantsverdriet
TABLE I. Kinetic parameters for CO oxidation on Rh-surfaces.
Ea
͑kJ/mol͒
␯
Ϫ1
Surface
Method
(s
)
Conditions
Ref.
7
.5Ϯ1
Rh͑111͒
UHV conditions,
TPD ͑leading edge͒
65Ϯ5
80Ϯ6
10
10
Low ␪O ; low ␪CO
High ␪O ; low ␪CO
This
work
10Ϯ0.5
UHV conditions
Molecular beam reactive
scattering
102Ϯ2
1012.3Ϯ0.3
High ␪O ; low ␪CO
9
1
0–300 mbar, steady
106
1012
Low ␪O ; high ␪CO
TϽ425 K
6, 10
7
state CO production
2
10–300 mbar, steady
84Ϯ8
-
state CO productions
2
UHV conditions,
TPD ͑isostere analysis͒
188
117
-
-
Low ␪O ; low ␪CO
High ␪O ; low ␪CO
4
2
Rh͑111͒
10–300 mbar, steady
121
126
101
10
Low ␪O ; high ␪CO
Low ␪O ; high ␪CO
8
12
Rh/Al O
state CO production
2
3
2
5
.6
Rh-wire
Rh͑110͒
Rh͑100͒
Kinetic modeling
60
10
10
High ␪O ; low ␪CO
Low ␪O ; high ␪CO
19
14
10.9
105
Molecular beam
Kinetic modeling
71
107
Low ␪O ; high ␪CO
UHV conditions,
TPD ͑leading edge͒
103Ϯ5
75Ϯ8
101
2.7Ϯ0.7
1Ϯ1.2
Low ␪O ; high ␪CO
High ␪O ; low ␪CO
This
work
1
10
parameters most probably reflect the kinetics of CO desorp-
tion from a CO-saturated surface as this frees up empty sites,
necessary for the adsorption of oxygen. Steady-state experi-
ments consider the overall reaction mechanism CO ϩO
might also provide a good explanation for why the activation
energy is apparently lower on Rh͑111͒ than on Rh͑100͒.
g
2,g
→
CO2,g , and thus they are not very informative with respect
V. CONCLUSIONS
to surface reaction step COadsϩOads→CO2,gϩ2*. TPD ex-
periments only reflect the latter step, as the reactants are
preadsorbed. Our experiments indicate that oxidation of CO
in the low coverage limit on Rh͑100͒ is an elementary step,
whereas on Rh͑111͒ it is not. This is most probably due to a
nonhomogeneous distribution of the adsorbates on Rh͑111͒.
We have used Temperature-Programmed Reaction Spec-
troscopy to monitor the reaction between COads and Oads on
two different Rh-surfaces, ͑100͒ and ͑111͒. We find that ap-
plication of coverage-corrected leading-edge analysis, i.e.,
plotting the natural logarithm of the desorption rate divided
by the actual coverages of the reactants versus 1/T, works
remarkably well to analyze the TPD spectra and to determine
the kinetic parameters. The main results from this work are:
• Oxidation of CO on rhodium surfaces is a structure
Mixed structures of CO and O on Rh͑111͒ are only observed
when the O-atoms are well-ordered.2
7,28
However, in the ab-
sence of a well ordered O-overlayer, segregation of the reac-
tants in islands is reported. This implies that CO oxidation
4
on Rh͑111͒ cannot be described by the simple Arrhenius-
equation ͑1͒, as reaction only occurs along the island perim-
eters. Surprisingly, this does not seem to occur on the more
open Rh͑100͒ surface, although formation of O-islands has
been reported. Surface structure might play a decisive role
in this issue, as the absolute atom-density on Rh͑100͒ is
lower. This would imply that the oxygen structures on
Rh͑100͒ are more open than on Rh͑111͒ at equal coverage.
Possibly, this enables CO to penetrate more easily within the
oxygen structures on Rh͑100͒, whereas restricted mobility of
CO on Rh͑111͒ prevents CO to mix within O-islands. This is
in agreement with Fig. 7.
sensitive
reaction.
The
surface
reaction
step
COadsϩOads→CO2,g is intrinsically faster on Rh͑100͒ than on
Rh͑111͒.
• As a consequence, selectivity to CO2 is generally
higher on Rh͑100͒. On Rh͑100͒ reaction between COads and
Oads goes to completion, whereas on Rh͑111͒ oxidation and
desorption of CO are competing channels.
• At low CO coverage, CO oxidation is an elementary
step on Rh͑100͒ for a broad range of oxygen coverages. We
5
9
report kinetic parameters E ϭ103Ϯ5 kJ/mol and ␯
a
ϭ101
2.7Ϯ0.7
for ␪ ϭ␪ →0.
O
CO
• The activation energy for CO oxidation on Rh͑100͒
We cannot, however, rule out that the CO oxidation on
Rh͑111͒ is dominated by steps as recently was found to be
continuously decreases with increasing O-coverage. At low
coverage (␪ Ͻ0.25 ML͒ we attribute this to destabilization
O
6
3
the case for the dissociation of N on Ru͑0001͒. This would
of CO, leading to an increase in the CO formation rate. At
2
2
also lead to a low value of the preexponential factor, and it
higher coverage (␪ Ͼ0.25 ML͒ O-atoms become destabi-
O
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
141.100.75.156 On: Tue, 25 Nov 2014 09:54:17

J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Oxidation kinetics of CO on rhodium
5465
2
2
8
9
lized as well, as lateral interactions between O-atoms come
into play at these coverages.
A. J. Jaworowski, A. Beutler, F. Strisland, R. Nyholm, B. Setlik, D. Hes-
kett, and J. N. Andersen, Surf. Sci. 431, 33 ͑1999͒.
G. Comelli, V. R. Dhanak, M. Kiskinova, N. Pangher, G. Paolucci, K. C.
Prince, and R. Rosei, Surf. Sci. 260, 7 ͑1992͒.
G. Comelli, V. R. Dhanak, M. Kiskinova, G. Paolucci, K. C. Prince, and
R. Rosei, Surf. Sci. 269Õ270, 360 ͑1992͒.
P. W. Murray, F. M. Leibsle, Y. Li, Q. Guo, M. Bowker, G. Thornton, V.
R. Dhanak, K. C. Prince, and R. Rosei, Phys. Rev. B 47, 12976 ͑1993͒.
W. Oed, B. D o¨ tsch, L. Hammer, K. Heinz, and K. Muller, Surf. Sci. 207,
•
On Rh͑100͒ the coverages of O-atoms and CO mol-
ecules add up to a constant coverage of 0.82 ML at all
O-precoverages. On Rh͑111͒, however, O-atoms progres-
sively exclude the adsorption of CO, such that the saturation
coverages of CO decreases from 0.75 ML in the absence of
30
3
3
3
3
1
2
3
4
Oads to zero at an O-saturated surface (␪ ϭ0.5 ML͒.
O
55 ͑1988͒.
R. A. van Santen and J. W. Niemantsverdriet, Chemical Kinetics and
Catalysis ͑Plenum, New York, 1995͒.
V. P. Zhdanov, Surf. Sci. Rep. 12, 183 ͑1991͒.
H. J. Borg, J. P. C.-J. M. Reijerse, R. A. van Santen, and J. W. Niemants-
verdriet, J. Phys. Chem. 101, 10052 ͑1994͒.
A. M. de Jong and J. W. Niemantsverdriet, J. Phys. Chem. 101, 10126
ACKNOWLEDGMENTS
This work was financially supported by the Netherlands
Organization for Scientific Research–Chemical Sciences
35
3
3
6
7
͑NWO–CW͒ and performed under the auspices of the Neth-
͑
1994͒.
M. J. P. Hopstaken and J. M. Niemantsverdriet, J. Phys. Chem. B 104,
058 ͑2000͒.
erlands Institute of Catalysis Research ͑NIOK͒.
3
1
J. L. Gland and E. B. Kollin, J. Chem. Phys. 78, 963 ͑1983͒.
C. T. Campbell, G. Ertl, H. Kuipers, and J. Segner, J. Chem. Phys. 73,
38
39
40
A. Beutler, E. Lundgren, R. Nyholm, J. N. Andersen, B. Setlik, and D.
Heskett, Surf. Sci. 371, 381 ͑1997͒.
M. Gierer, A. Barbieri, M. A. van Hove, and G. A. Somorjai, Surf. Sci.
2
5
862 ͑1980͒.
3
4
G. Ertl, Adv. Catal. 37, 213 ͑1990͒.
T. Matsushima, T. Matsui, and M. Hashimoto, J. Chem. Phys. 81, 5151
391, 176 ͑1997͒.
D. G. Castner and G. A. Somorjai, Appl. Surf. Sci. 6, 29 ͑1980͒.
K. C. Wong, W. Liu, and K. A. R. Mitchell, Surf. Sci. 360, 137 ͑1996͒.
C.-M. Chan, R. Aris, and W. H. Weinberg, Appl. Surf. Sci. 1, 360 ͑1978͒.
P. A. Redhead, Vacuum 12, 203 ͑1962͒.
P. A. Thiel, E. D. Williams, J. T. Yates, Jr., and W. H. Weinberg, Surf.
Sci. 84, 54 ͑1979͒.
͑
1984͒.
4
4
4
4
1
2
3
4
5
6
7
T. W. Root, L. D. Schmidt, and G. B. Fisher, Surf. Sci. 150, 173 ͑1985͒.
D. W. Goodman and C. H. F. Peden, J. Phys. Chem. 90, 4839 ͑1986͒.
S. B. Schwartz, L. D. Schmidt, and G. B. Fisher, J. Phys. Chem. 90, 6194
͑
1986͒.
8
9
0
S. H. Oh, G. B. Fisher, J. E. Carpenter, and D. W. Goodman, J. Catal. 100,
60 ͑1986͒.
L. S. Brown and S. J. Sibener, J. Chem. Phys. 89, 1163 ͑1988͒; 90, 2807
1989͒.
4
4
5
6
M. Bowker, Q. Guo, and R. W. Joyner, Surf. Sci. 253, 33 ͑1991͒.
V. Nehasil, S. Zafeiratos, S. Ladas, and V. Matol ´ı n, Surf. Sci. 433-435,
3
215 ͑1999͒.
͑
1
47
48
C. H. F. Peden, D. W. Goodman, D. S. Blair, P. J. Berlowitz, G. B. Fisher,
and S. H. Oh, J. Phys. Chem. 92, 1563 ͑1988͒.
M. Bowker, Q. Guo, Y. Li, and R. W. Joyner, Catal. Lett. 18, 119 ͑1993͒.
J. I. Colonell, K. D. Gibson, and S. J. Sibener, J. Chem. Phys. 103, 6677
B. Hammer, L. B. Hansen, and J. K. No”rskov, Phys. Rev. B 59, 7413
͑1999͒.
1
1
1
2
J. Wider, T. Greber, E. Wetli, T. J. Kreutz, P. Schwaller, and J. Oster-
walder, Surf. Sci. 417, 301 ͑1998͒.
A. Eichler and J. Hafner, Surf. Sci. 433-435, 58 ͑1999͒.
C. Stampfl and M. Scheffler, Surf. Sci. 433-435, 119 ͑1999͒.
A. Alavi, P. Hu, T. Deutsch, P. L. Silvestrelli, and J. H u¨ tter, Phys. Rev.
4
9
͑1995͒.
13
50
M. J. P. Hopstaken, W. J. H. van Gennip, and J. W. Niemantsverdriet,
Surf. Sci. 433-435, 69 ͑1999͒.
M. Bowker, Q. Guo, and R. W. Joyner, Surf. Sci. 280, 50 ͑1993͒.
F. M. Leibsle, P. W. Murray, S. M. Francis, G. Thornton, and M. Bowker,
Nature ͑London͒ 363, 706 ͑1993͒.
A. Baraldi, L. Gregoratti, G. Comelli, V. R. Dhanak, M. Kiskinova, and R.
Rosei, Appl. Surf. Sci. 99, 1 ͑1996͒.
L. W. H. Leung and D. W. Goodman, Catal. Lett. 5, 353 ͑1990͒.
B. A. Gurney, L. J. Richter, J. S. Villarubia, and W. Ho, J. Chem. Phys.
5
1
1
1
4
5
Lett. 80, 3650 ͑1998͒.
52
R. M. van Hardeveld, M. J. P. Hopstaken, J. J. Lukkien, P. A. J. Hilbers,
A. P. J. Jansen, R. A. van Santen, and J. W. Niemantsverdriet, Chem.
Phys. Lett. 302, 98 ͑1999͒.
J. Wintterlin, J. Trost, S. Renisch, R. Schuster, T. Zambelli, and G. Ertl,
Surf. Sci. 394, 159 ͑1997͒.
1
6
53
54
55
1
1
7
8
T. V. W. Janssens, G. R. Castro, K. Wandelt, and J. W. Niemantsverdriet,
Phys. Rev. B 49, 14599 ͑1994͒.
8
7, 6710 ͑1987͒.
19
C. T. Campbell, S.-K. Shi, and J. M. White, J. Phys. Chem. 83, 2255
1979͒.
T. Bunluesin, H. Cordatos, and R. J. Gorte, J. Catal. 157, 222 ͑1995͒.
G. Ertl, Adv. Catal. 28, 2 ͑1979͒.
T. Matsushima and H. Asada, J. Chem. Phys. 85, 1658 ͑1986͒.
I. Z. Jones, R. A. Bennett, and M. Bowker, Surf. Sci. 439, 235 ͑1999͒.
A. B o¨ ttcher, H. Niehus, S. Schwegmann, H. Over, and G. Ertl, J. Phys.
G. B. Fisher and S. J. Schmieg, J. Vac. Sci. Technol. A 1, 1064 ͑1983͒.
R. Kose, W. A. Brown, and D. A. King, Surf. Sci. 402-404, 856 ͑1998͒.
D. Loffreda, D. Simon, and P. Sautet, J. Chem. Phys. 108, 6447 ͑1998͒.
E. Hansen and M. Neurock, Surf. Sci. 441, 410 ͑1999͒.
A. Baraldi, V. R. Dhanak, G. Comelli, K. Prince, and R. Rosei, Phys. Rev.
B 56, 10511 ͑1997͒.
D. Alf e´ , S. de Gironcoli, and S. Baroni, Surf. Sci. 410, 151 ͑1998͒.
A. Baraldi, J. Cerd a´ , J. A. Martin Gago, G. Comelli, S. Lizzit, G. Paolucci,
and R. Rosei, Phys. Rev. Lett. 82, 4874 ͑1999͒.
J. R. Mercer, P. Finetti, F. M. Leibsle, R. McGrath, V. R. Dhanak, A.
Baraldi, K. C. Prince, and R. Rosei, Surf. Sci. 352-354, 173 ͑1996͒.
S. Dahl, A. Logadottir, R. C. Egeberg, J. H. Larsen, I. Chorkendorff, E.
T o¨ rnqvist, and J. K. No”rskov, Phys. Rev. Lett. 83, 1814 ͑1999͒.
͑
5
5
5
5
6
7
8
9
2
2
2
2
2
0
1
2
3
4
6
6
0
1
Chem. B 101, 11185 ͑1997͒.
2
2
2
5
6
7
R. J. Gelten, A. P. J. Jansen, R. A. van Santen, J. J. Lukkien, J. P. L.
Segers, and P. A. J. Hilbers, J. Chem. Phys. 108, 5921 ͑1998͒.
J. Wintterlin, S. V o¨ lkening, T. V. W. Janssens, T. Zambelli, and G. Ertl,
Science 278, 1931 ͑1997͒.
S. Schwegmann, H. Over, V. DeRenzi, and G. Ertl, Surf. Sci. 375, 91
62
63
͑1997͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
41.100.75.156 On: Tue, 25 Nov 2014 09:54:17
1