Oxidation reactions over RuO2: A comparative study of the reactivity of the (110) single crystal and polycrystalline surfaces

Article abstract of DOI:10.1006/jcat.2001.3281

Oxidation reactions over RuO₂ surfaces were investigated to determine the site requirement for CO (as well as for methanol) and O₂ adsorption on the surface of a well-defined RuO₂(110) crystal. The reactivity of the (110) single crystal was compared with that of polycrystalline RuO₂ species. Both results, those in UHV over RuO₂(110) and those at 1 atm over polycrystalline RuO₂, were in good agreement, an indication that the surface chemistry of the (110) single crystal was similar to that of the polycrystalline material. Both the activity of the RuO₂(110) surface in the UHV regime and the activity of polycrystalline RuO₂ in the high-pressure regime were studied by TPD and XPS. The unusually high catalytic activity for oxidation reactions of RuO₂ was attributed to the strong bonding of the reactants over the undercoordinated Ru atoms together with the presence of weakly bound undercoordinated oxygen species, which served as the oxidizing agent.

Full text of DOI:10.1006/jcat.2001.3281

Journal of Catalysis 202, 296–307 (2001)
doi:10.1006/jcat.2001.3281, available online at http://www.idealibrary.com on
Oxidation Reactions over RuO : A Comparative Study of the Reactivity
2
of the (110) Single Crystal and Polycrystalline Surfaces
�
�,1
,1
H. Madhavaram, H. Idriss, S. Wendt,† Y. D. Kim,† M. Knapp,† H. Over,† J. Aßmann,‡
E. L o¨ ffler,‡ and M. Muhler‡
�
Centre for Surface and Materials Science, Department of Chemistry, The University of Auckland, Auckland, New Zealand; †Fritz-Haber
Institut der Max-Planck Gesellschaft, Faradayweg 4-6, 14195 D-Berlin, Germany; and ‡Laboratory of Industrial Chemistry,
Ruhr-Universit a¨ t Bochum, D-44780 Bochum, Germany
Received January 24, 2001; revised May 13, 2001; accepted May 20, 2001; published online August 9, 2001
that regard, a detailed investigation of the site requirement
The present work is devoted to the investigation of oxidation re- for CO (as well as for methanol—as a representative of
actions over RuO surfaces. Two main points are addressed. First, organic pollutants) and O adsorption on the surface of a
2
2
on the fundamental level, a detailed investigation of the site require-
well-defined RuO2 is analyzed. (ii) With regard to applica-
ment for CO (as well as for methanol) and O adsorption on the
2
tion, what is the reactivity of polycrystalline RuO toward
2
surface of a well-defined RuO (110) crystal is conducted. Second,
2
these oxidation reactions and how does it compare with
a comparison with polycrystalline RuO2 toward these oxidation
UHV experiments of RuO2 single crystal? A complemen-
tary question arises from the second one. (iii) What is the
behavior of a reduced (oxygen defected) Ru oxide surface
toward these same reactions? This latter point is central
to our understanding of oxidation reactions over oxides in
reactions is presented. Both results, those in UHV over RuO (110)
2
and those at atmospheric pressure over polycrystalline RuO , agree
2
fairly well. This indicates that the surface chemistry of the (110) sin-
gle crystal is very similar to that of the polycrystalline material. Both
the activity of the RuO (110) surface in the UHV regime and the
2
activity of polycrystalline RuO in the high-pressure regime were general, since ideally the surface is continuously oscillating
2
investigated by temperature-programmed desorption and by X-ray between the oxidized and reduced states. To achieve this
photoelectron spectroscopy. The reasons for the unusually high objective, a comparative study is conducted to investigate
catalytic activity for oxidation reactions of RuO2 are traced
back to the strong bonding of the reactants over the under-
coordinated Ru atoms together with the presence of weakly
bound undercoordinated oxygen species, serving as the oxidizing
This work shows that both results agree fairly well.
agent. �c 2001 Academic Press
the reactivity of rutile RuO (110) single crystal, under ultra
high vacuum conditions, as well as that of polycrystalline
RuO2 at 1 atmosphere, toward these oxidation reactions.
2
Rutile RuO2(110) surface (like rutile TiO2(110) and
SnO2(110)) possesses alternating rows of fivefold and six-
fold coordinated Ru cations (the fivefold coordinated Ru
cations are designated cus-Ru, where cus stands for co-
I. INTRODUCTION
ordinatively unsaturated sites. The energies of the relaxed
RuO2 is now receiving considerable attention for its ex-
(
110) surfaces of these three oxides are relatively compa-
traordinarily high catalytic activity (below room tempera-
ture) toward CO oxidation. This reaction was investigated
at the atomic scale by applying a whole battery of surface
science techniques under ultra high vacuum (UHV) condi-
tions (1–3). To test the potential of this surface for other
reactions, the present work is devoted to the investigation
of the oxidation reaction of methanol over RuO2 surfaces
and compares it to CO. Considering both oxidation reac-
tions, CO to CO2 and methanol to formaldehyde and CO2,
as a case study, the work mainly addresses the following
questions. (i) On a fundamental level, what is (are) the site
requirement(s) of this very efficient oxidation reaction? In
rable (1.78 (4), 1.78 (5), and 2.06 (6) eV per unit cell for
TiO2(110), RuO2(110), and SnO2(110), respectively).
Only very recently, has detailed work been conducted
under UHV conditions of the rutile RuO2 (110) single
crystal surface (3). The adsorption sites of RuO2 for sev-
eral molecules such as CO, N2 (7), and methanol (8) are
shown to be on the cus-Ru atoms. Two main differences are
clearly observed between rutile RuO2(110) and the corre-
sponding surfaces of TiO2 or SnO2: (i) Thermal desorp-
tion spectroscopy (TDS) (1), as well as first principles den-
sity functional theory (DFT) calculations (7), has shown
that unlike CO adsorption over cus-Ti of TiO2(110) or cus-
Sn of SnO2(110), CO adsorption on cus-Ru is very strong
(1.2 eV (7)). This is to be compared with CO-TiO2(110) =
1
To whom correspondence should be addressed.
2
96
0021-9517/01 $35.00
Copyright �c 2001 by Academic Press
All rights of reproduction in any form reserved.

OXIDATION REDUCTIONS OVER RuO2
297
0
.3 eV (4) and CO-SnO2(110) = 0.25 eV (6). (ii) Unlike temperature. It was cleaned by freeze–pump–thaw cycles.
TiO2 and SnO2, terminal bonded atomic oxygen can be Dosing of the reactant on the oxide was conducted by allow-
stabilized on RuO2(110) with a binding energy of 3.1 eV ing the vapor pressure (ca. 100 torr) to equilibrate between
(
9); for comparison the bridging oxygen on RuO2(110) is the reactor and methanol, typically for 3 min. The reactor
bonded by 4.6 eV (values are given with respect to neu- was then pumped by a roughing pump for ca. 1 h at room
tral atomic oxygen in the gas phase). These two essential temperature to remove traces of the reactant in the TPD
differences, the strong CO adsorption and the presence of line as well as weakly adsorbed molecules on the surface
dissociative adsorption of O2 molecules (with high sticking of the catalyst. The gas flow was introduced into the vac-
probability) (10), at moderate temperatures, explain natu- uum chamber through the interface (50). The ramping rate
rally the high activity of RuO2(110) for CO oxidation (11). during TPD was kept fixed at 15 K/min. The fragmenta-
This high activity is unusual when compared not only to tion pattern of each product was checked to identify unam-
TiO2(110) and SnO2(110) but also, in fact, to any other biguously the reaction products by following the method
known oxide single crystal to date.
described in Refs. (17, 18).
The reactions of methanol and ethanol on the surfaces
XPS analyses were performed using a KRATOS XSAM-
�
9
of oxide materials are of technological and fundamental 800 model with a base pressure of ca. 10 torr. AlK�(h� =
importance (12, 13). Moreover, besides its use as a model 1486.6 eV) radiation was used at 180 W. Ar-ion sputtering
for fundamental research, methanol is an atmospheric pol- experiments were performed, at room temperature, using
lutant, with a tropospheric concentration reaching up to a direct-beam KRATOS ion gun at a pressure of ca. 5 �
�
7
5
ppbv and a relatively long lifetime (ca. 1 week) (14). It 10 Torr. An accelerating voltage of 2.7 kV was used, and
is used as a fuel additive and as such its concentration in the emission current was 20 mA. Unreduced RuO2 was
the atmosphere is expected to rise worldwide. In addition, loaded into the system without further treatment. Narrow
methanol is a good candidate, in the next decade or so, scans were obtained with a pass energy of 20 eV, which
as a source for H2 production for fuel cells. Since present translates to an energy resolution of 1.1 eV. In the case
fuel cell technology does not tolerate the presence of CO of H2-reduced RuO2 the reactor valves were closed be-
(
formed by partial oxidation of methanol), it is thus impor- fore transferring the sample to the XPS instrument. There,
tant to rely on chemical processes completely selective for the powder was transferred onto the stainless steel sample
the oxidation of carbon-containing molecules to CO2. stub with indium foil, under an oxygen-free nitrogen at-
A detailed study of the reaction of alcohols over oxide mosphere. The sample stub was then transferred into the
surfaces can be found in Ref. (12). It is, however, worth vacuum chamber. The whole process of transferring the
mentioning a few points. Over TiO2(110) methanol is unre- sample took 30–40 s.
active (15). For example, DFT calculations have shown that
dissociative adsorption is almost not favored on TiO2(110) erator and a Phillips 1050 goniometer. Radiation by X
nor on SnO2(110)) surfaces. The adsorption energies of rays was achieved using a Cu tube (broad focus) (K� : � =
methanol over TiO2(110), at � = 1 ML, is 70.4 and 74.3 kJ 1.514 A˚ ) at 44 kV and 20 mA. Diffraction patterns were
mol for nondissociative and dissociative modes, respec- collected from 10 to 90 (2�).
tively (15). Those on SnO2(110) for the same modes were The temperature-programmed reduction (TPR) experi-
10.4 and 114.5 kJ mol (16). As will be shown in this ments were performed in a quartz cell built according to
XRD spectra were collected using a Phillips 1130 gen-
(
�
1
�
�
�
1
1
work there is a dramatic difference between the reactivity the design by Monti and Baiker (19) using a heating rate of
of rutile RuO2(110) and that of rutile TiO2(110) toward this 2 K/ min. Prior to the TPR measurement, RuO2 was heated
organic compound.
to 573 K using a heating rate of 10 K/min in synthetic air
20.5% O2 in N2) with a volumetric flow rate of 100 ml/min.
(
Under these flow conditions, RuO2 was calcined at 573 K
for 2 h followed by cooling in synthetic air to room temper-
ature. After flushing in diluted H2 (4.2 vol% H2 in Ar, pu-
rities 99.999% , supplied by Messer Griesheim) for 30 min,
the heating ramp up to 573 K was started using a flow rate of
II. EXPERIMENTAL DETAILS
II.A. Polycrystalline RuO2
Polycrystalline RuO2 (purity >99.9% ) was obtained
8
4.1 ml/min. The effluent gas mixture was passed through a
from Merck-Schuchardt Corp. BET analysis has shown that
cold trap at 195 K to remove water and analyzed quantita-
tivelybya calibrated thermalconductivitydetector (Hydros
2
it has a surface area of 9.8 m /g. TPD was performed using
a fixed-bed reactor containing 30 mg of RuO2. More details
about the TPD apparatus can be found in Ref. 50.
RuO2 is heated under O2 for 2 h or under hydrogen for
1
00, 0–5% H2 in Ar, supplied by Rosemount).
II.B. RuO2 Single Crystal
1
8 h at 573 K (P = 760 Torr). After cooling to room tem-
perature (under O2 or H2) the reactor was pumped down
The UHV measurements were conducted in a stainless
to 10 Torr. Methanol was placed in a saturator at room steel chamber as described elsewhere (20). RuO2 (110)
�
3

2
98
MADHAVARAM ET AL.
RuO , and high-temperature (573 K) H -reduced RuO
2
2
2
are presented in Fig. 2. The corresponding O(1s) to Ru(3d)
ratios are given in Table 1 (also shown is the effect of
+
Ar sputtering on this ratio). The XPS Ru(3d5/2) is shown
at 281.2 � 0.1eV with a spin orbit splitting 3d5/2–3d3/2 of
4
.1 � 0.1 eV. This spectrum is very similar to that reported in
4
+
Refs. (22, 23). These line positions are those of Ru cations
22, 23). A component at ca. 282.8 eV is also seen. This con-
(
4+
tribution is assigned in Ref. (23) to unscreened Ru cations
by comparison with the binding energy of nonmetallic Ru4
+
oxides. In other words the line at 281.2 eV (in Fig. 2) is due
to screened Ru4 cations since RuO2 is metallic (24). The
full width at half maximum (FWHM) of the Ru(3d5/2) line
of 2.0 eV is quite large, in particular if this value is com-
pared to a line width of about 0.6 eV for a well-defined
RuO2(110) surface (see Section III.B.1). We interpret the
large value for the FWHM of polycrystalline RuO2(110) as
originating from a inhomogeneous broadening due to the
presence of several orientations of RuO2 surfaces and the
+
presence of defects. A clear shift is seen upon H2 reduction
+
(
(
Fig. 2c) or Ar sputtering (Fig. 2b). The shift of 0.7 � 0.1 eV
XPS Ru(3d5/2) = 280.5 eV) is identical whether the sur-
+
face has been reduced with H2 or Ar -sputtered for 40 min.
+
We should note that Ar -sputtering preferentially removes
FIG. 1. Successive O2-TPD after 0.1 L exposure over RuO2(110) sur-
face at 200 K. The inset shows the noncalibrated peak area of desorbing
oxygen from the RuO2 surface. The line at 280.5 eV is prob-
O2 molecules as a function of dosing pressure. The y axis in the figure ably showing a large contribution from Ru metal, since Ru
shows the noncorrected m/e 32 mass spectrometer signal.
metal has a binding energy of 280.0–280.3 eV (Table 1) and
no other stable Ru cations with oxidation state lower than
+
4 are known to exist in the solid state. The O(1s) emis-
single crystal surfaces were prepared as described in de-
tail elsewhere (3, 21). Owing to the extreme reactivity of
the surface, its stoichiometry must be checked prior to
every run (or at least on a daily bases). Restoration of the
surface can be achieved by successive dosing of oxygen at
room temperature followed by TPD. The saturation of a
TD peak at 420–430 K (previously assigned as O� (9) and
shown as the precursor for filling in the removed bridging
oxygen) indicates that surface defects have been healed.
Figure 1 shows a typical sequence of desorption spectra
of O2 (m/e 32); the inset shows the peak area as a func-
tion of the number of TPD runs i.e., until saturation). Ul-
tra pure methanol (from Aldrich) was carefully degassed
by cycles of cooling with liquid nitrogen and annealing
the vessel to room temperature, while continuously pump-
ing the vapor above the frozen methanol. This procedure
was iterated until no bubbles appeared during methanol
melting.
sion occurs at 529.7 eV, indicative of O anions. The tail
of the O(1s) peak is most likely a contribution from sur-
face hydroxyls at ca. 532 eV (25) (Fig. 3). Also shown in
the figure is the O(1s) obtained after hydrogen reduction
as well as after Ar-ion sputtering. Table 1 shows that one
indeed is removing O atoms by H2 reduction, further con-
firming the XPS Ru(3d) results. However, the remaining
presence of these XPS O(1s) lines indicates the presence of
chemisorbed oxygen on and/or within the reduced material.
This interpretation is supported by the observed O(1s) peak
position at 530.5 eV, which is assigned to chemisorbed oxy-
gen (Fig. 3, see Section III.B.1 for other details). We should
emphasize that the bond energy of Ru–O in chemisorbed
O on Ru metal is stronger by more than 1 eV than the cor-
responding bond of RuO2 (9). In fact, this is why Ru metal
is unusually nonreactive compared to other noble metals;
once the surface is covered with O atoms, it becomes very
difficult to remove them (1, 2, 26). One can measure the
extent of reduction by the decrease of the O(1s) to Ru(3d)
corrected peak area intensity ratios. This change can also be
correlated to the FWHM of the Ru(3d) peaks since this lat-
ter decreased with increasing sputtering time. As indicated
in Figs. 4 and 5 sputtering as a function of time (as well
as H2 reduction) does decrease the FWHM of Ru(3d5/2)
III. RESULTS
III.A. Polycrystalline RuO2
III.A.1. X-Ray Photo Emission Data
X-ray photoelectron spectra of Ru3d for O2-annealed peaks with concomitant lowering of the binding energy.
+
polycrystalline RuO2, room temperature Ar -sputtered Also shown is the decreasing of the O(1s)/Ru(3d) lines with

OXIDATION REDUCTIONS OVER RuO2
299
TABLE 1
XPS Binding Energy of Ru(3d5/2), O(1s), and Corrected Peak Area Ratios
XPS Ru (3d5/2)
FWHM
(eV)
XPS O (1s)
(eV)
1 = (O(1s)–Ru(3d5/2))
XPS
O(1s)/Ru(3d)
Material
(eV)
(eV)
Ref.
Ru/O
Thin film
Ru/O
Powder
RuO2(110)
Single crystal
281.0
281.9
280.8
530.0
249.0
22
28
0.5�
529.48
529.0
248.88
248.5
This work
27
2
80.5
RuO2(100)
Single crystal
RuO2
280.7
529.6
248.9
27
Polycrystalline
281.2
2.0
529.7
529.4
529.2
248.5
248.7
248.5
1.86
This work
2
80.7
24
25
RuO2/Ti Electrode
280.7
+
Ar -sputtered RuO2
Polycrystalline
5
1
4
min
0 min
0 min
280.9
280.7
280.6
1.43
1.30
1.26
530.7
530.6
530.5
249.8
249.9
249.9
0.75
0.40
0.40
This work
This work
This work
RuO2(110)
Single crystala
H2-reduced RuO2
Polycrystalline
Ru
279.6
280.5
280.0
0.37
0.31
27
1.13
530.4
249.9
This work
Polycrystalline
24
25
22
2
2
80.2
80.3
Ru(0001)
b
Single crystal
Ru(0001)–(1 � 1)O
279.8
0.20
36
280.12c
0.21
b
530.07
530.07
249.95
249.02
This work
This work
d
b
2
81.05
0.39
a
2
0
.5 kV, 15 �m/cm , 30 min [27].
b
c
d
High-resolution core level shift—see experimental part for more details.
Bulk Ru metal (Fig. 10).
Ru three-fold coordinated to O atoms (Fig. 10).
+
increasing sputtering (due to preferential oxygen removal). as well as the 40-min. Ar -ion-sputtered samples might be
Because of the presence of chemisorbed oxygen (together larger than those expected from pure Ru metal.
with an energy resolution of >1.1 eV), the FWHM and the
position of the XPS Ru (3d5/2) line of the hydrogen-reduced not more efficient in reducing RuO2 than Ar sputtering
It does appear that H2 reduction is at least as good as if
+
+
FIG. 2. XPS of the Ru(3d) region for polycrystalline RuO2 (a), 40 min
Ar -sputtered RuO2 (b), and H2-reduced RuO2 (c).
FIG. 3. XPS O(1s) of O2-annealed RuO2 (a), 40 min Ar - sputtered
RuO2 (b), and H2-reduced RuO2 (c).
+

3
00
MADHAVARAM ET AL.
FIG. 4. XPS of the Ru(3d5/2) region of polycrystalline RuO2 as a func-
tion of sputtering time with Ar ions; all other conditions are kept constant.
FIG. 6. XRD of (a) O2-annealed (573 K, 1 atm) and (b) H2-reduced
(573 K, 1 atm) polycrystalline RuO2.
+
+
(
a) 5 min Ar -sputtered RuO2, (b) 10 min Ar -sputtered RuO2, (c) 40 min
+
Ar -sputtered RuO2, and (d) H2-reduced RuO2.
with XPS Ru(3d) lines and may strongly indicate that the
(
see the values of both 1 and O(1s)/Ru(3d) in Table 1). amount of oxygen (about 20% ) observed by XPS can be
This is by no means universal, as H2 reduction usually has a assigned to chemisorbed or dissolved oxygen on and/or in
+
marginal effect when compared to Ar sputtering for most metallic Ru (Fig. 6).
oxide materials. For example, H2 is thought to adsorb only
on defect sites of TiO2(110) surface, with an isosteric heat III.A.3. Temperature-Programmed Reduction of RuO2
�
1
of adsorption of 83 kJmol (29) while no evidence for its
Figure 7displaystwo TPR profilesobtained byusingsam-
ple weights of 102.4 mg and 26.6 mg. Calculating the con-
stant P defined as P = �S0/Fc0 according to Malet and
Caballero (36) yields values of 9.8 and 2.5 K, respectively,
where � is the heating rate, S0 is the amount of reducible
species, and Fc0 is the molar H2 flow rate into the reactor.
Values of P lower than 20 K were found to be appropriate
adsorption is found on the fully oxidized Ti foil (30). How-
ever, Ar-ion sputtering may reduce up to 75% of the Ti
cations into Ti (x < 4) of single crystal or polycrystalline
4+
x+
TiO2 surfaces (31, 32).
III.A.2. X-Ray Diffraction Data
The effect of the extent of reduction was also monitored to obtain good peak resolution (36). Accordingly, a narrow
by XRD. XRD did indicate that indeed a bulk transfor- TPR peak with a full width at half maximum of 20 K (solid
mation occurs because RuO2 has been transformed to Ru curve) was obtained. The sharp onset of reduction below
metal by H2 reduction. The spectra compare well with those 390 K indicates an autocatalytic process triggered by the
found in the literature (33–35). They are also consistent formation of metallic Ru nuclei providing atomic hydrogen.
Both the peak position at 391 K and the low FWHM are
in good agreement with the TPR results obtained by Hurst
FIG. 5. Corrected peak area ratios of XPS O(1s) to XPS Ru(3d) of
+
+
polycrystalline RuO2 (a), 5 min Ar -sputtered RuO2 (b), 10 min Ar -
+
sputtered RuO2 (c), 40 min Ar -sputtered RuO2 (d), and H2-reduced
RuO2 (e) as a function of the FWHM of the corresponding XPS Ru(3d5/2)
(
x-axis). Also shown is the corresponding binding energy position of
FIG. 7. TPR of polycrystalline RuO . (Solid line) Sample weight,
2
Ru(3d5/2) (right hand side of y-axis).
102.4 mg; (dotted line) sample weight, 26.6 mg. Heating rate, 2 K/s.

OXIDATION REDUCTIONS OVER RuO2
301
et al. (37). Decreasing S0 by a factor of 4 yields an even
TABLE 2
Carbon Yield and Peak Temperature of Products Resulting
higher resolution (dotted curve). This TPR profile consists
of a very narrow peak at 388 K and a broader peak at 398 K.
The latter may originate from an inhibiting influence of wa-
ter on the reduction kinetics.
The degree of reduction was calculated based on the fol-
lowing equation:
from TPD after Methanol Adsorption over Polycrystalline RuO at
2
3
00 K
Reactant/product
Temperature (K)
Carbon yield (% )
Methanol (m/e 31)
Formaldehyde (m/e 30)
Methyl formate (m/e 60)
CO2 (m/e 44)
CO (m/e 28)
H2 (m/e 2)
360
360 and 480
360
410 and 480
410 and 480
370, 480, >520
>520
1.72
0.44
0.58
87.66
9.59
—
RuO2 + 2H2 → Ru + 2H2O.
[1]
Averaging the amount of H2 consumed in six TPR experi-
ments yields a value of 99.7% , which indicates full reduction
to metallic Ru within the experimental accuracy. The pres-
ence of polycrystalline metallic Ru at the end of the TPR
was also confirmed by XRD.
H2O (m/e 18)
—
for. Products desorbed in three temperature domains. The
first (below 390 K) consisted of methanol, formaldehyde,
methyl formate, H2O, and H2. As summarized in Table 2
these desorptionscombined contributed 2.7% ofthe overall
carbon products. In Fig. 8 H2 is shown to desorb at a slightly
higher temperature than formaldehyde (about 10 K).
Several runs were conducted addressing this particular
observation. H2 desorption temperature was seen either
exactly at the formaldehyde peak temperature or shifted
by up to 10 K higher (as seen in Fig. 8). The shift was not
consistent from one run to the other. It is most likely that
cooling the reactor under O2 had an effect on the state of
III.A.4. Temperature-Programmed Reactions over RuO2
Figure 8 shows TPD after methanol adsorption at room
temperature over polycrystalline RuO2. Formaldehyde
(
(
m/e 29, 30), methyl formate (m/e 60), CO (m/e 28), CO2
m/e 44), H2 (m/e 2), and water (m/e 18) were observed,
in addition to unreacted methanol (m/e 31) (Table 2). All
primary alcohols have their strongest signal at 31 m/e due
+
to CH2OH species. It is more accurate to consider the
m/e 31 signal for quantitative analysis rather than the par-
ent molecule because of its high intensity. This is partic-
ularly important in the case of methanol when TPD is
�
8
the RuO surface. A stabilization effect due to surface oxy-
conducted at pressure above ca. 10 Torr because itsparent
molecule (m/e 32) interferes with O2 from the background.
Quantitative analyses were computed following the meth-
ods described in Refs. (17, 18), where the major fragmen-
tation patterns (up to 5) of each product were accounted
2
gen on the recombinative desorption of H2 where surface
O atoms have indeed delayed D2 desorption over U metal
has been observed by other workers (38). Since the exper-
iments were conducted over RuO2 that has been cleaned
with O2 at 573 K and cooled under O2 to 300 K before evac-
uation, it is not unreasonable to consider that some weakly
held oxygen left on the surface retards the recombination
of H atoms.
The second desorption domain (380–520 K) mainly con-
sisted of CO and CO2 with some formaldehyde and H2. This
latter desorption contributed over 97% of the total carbon
products. The third desorption domain (510–750 K) con-
tained H2O with some H2. No carbon-containing products
are observed in this temperature region with the probable
exception of some formaldehyde.
Three main points are worth noting. (i) Almost all ad-
sorbed methanol (saturation coverage) has reacted, as evi-
denced by a desorption of only 1.7% of the initial amount
3
(
the surface was exposed to 18 � 10 torr of methanol prior
to TPD). (ii) The total decomposition mainly gave CO2
a CO2/CO of ca. 9), indicating that oxygen anions are
easily removed from the surface. (iii) Some formaldehyde
m/e 30) is left (0.44% ) and some has reacted via
(
(
Tishchenko reaction (ester formation) to give methyl for-
mate (0.58% ) (see Refs. 12 and 13 for more details on this
reaction).
FIG. 8. TPD after methanol adsorption at 300 K over polycrystalline
RuO2. RuO2 has been O2-annealed at 573 K (1 atm) and cooled under O2
to 300 K.

3
02
MADHAVARAM ET AL.
due to the high concentration of formaldehyde (Scheme B)
because 2 mol of formaldehyde may react, giving 1 mol of
methyl formate (Tishchenko reaction).
Scheme A
CH3OH + O (s) → CH3O (a) + OH (a) OH dissociation
[2]
CH3O (a) → CH2O (g) + H(a)
H(a) + OH (a) → H2O (g)
CH2O (a) + 2O (s) → CO2 (g) + H2O (g)
CH2O + O (s) → CO (g) + H2O (g)
Dehydrogenation [3]
Water removal
Decomposition
Decomposition
[4]
[5]
[6]
Scheme B
2CH2O (a) → HC(O)OCH3 (g)
[7]
Yields in Tables 2 and 3 are carbon-based and not molar-
based. In other words about 1/3 of the formaldehyde left
is forming methyl formate (assuming no other reaction of
the latter).
FIG. 9. TPD after methanol adsorption at 300 K over H2-reduced
RuO2 (573 K, 1 atm).
III.A.5. Temperature-Programmed Reactions
over Reduced RuO2
III.B. UHV Experiments
III.B.1. X-Ray Photo Emission Data of
Ru(0001)–(1 � 1)O and RuO2(110) Single Crystals
TPD measurements of methanol over H2-reduced RuO2
were also conducted. Considerable differences in the prod-
uct distribution are observed (Fig. 9 and Table 3). First,
High-resolution core level shift (HRCLS) experiments
all products desorbed at lower temperatures. Second, an of a well-defined RuO (110) surface in comparison with
2
early CO2 desorption is clearly seen with a peak at 360 K. the Ru(0001)–(1 � 1)O were performed at the beamline
Both the 360 and 450 K desorption contributed 50% of I311 at MAXII in Lund, Sweden (39). The photon energies
the desorption products. CO desorption accompanied the used were h� = 380 eV and 680 eV for Ru(3d) and O(1s),
low-temperature CO2 peak and contributed ca. 11% of respectively. The energy resolutions of the Ru(3d5 ) and
/2
the carbon yield. Formaldehyde, which was observed in O(1s) spectra were set to 80 MeV and 180 MeV, respec-
trace amounts on the stoichiometric surface, has increased tively (to be compared with the 1.1 eV in the polycrys-
severalfold, accounting for 20% of the overall products. talline experiments). The HRCL spectra of Ru(3d5/2) and
Similarly, methyl formate desorption has increased ca. 20 O(1s) were taken at 100 K and are depicted in Fig. 10. For
times (11.5% ). Because formaldehyde is most likely a re- the Ru(0001)–(1 � 1)O surface clearly two well-resolved
action intermediate to CO2 formation, it is not unexpected Ru(3d5 ) peaks are visible; these are assigned to bulk Ru
/2
to observe an incomplete oxidation on the reduced RuO2 (280.12 eV, FWHM: 0.21 eV) and Ru coordinated to three
surface (see Scheme A [2–4] below, where (a) is adsorbed, oxygen atoms in the surface layer (281.05 eV, FWHM:
(
s) is surface, VO is oxygen vacancy, (g) is gas). Moreover, 0.39 eV). The corresponding O(1s) photoemission line is
the increase of methyl formate may simply be explained as at 530.07 eV (FWHM: 0.55 eV) (Fig. 10a; see also Table 1).
1
1 = (O(1s)–Ru(3d5/2)) (where Ru(3d5/2) is at 280.12 eV)
isequalto 249.95eV. Thisvalue isveryclose to that obtained
for polycrystalline oxide that has been Ar-ion sputtered or
H2-reduced (249.8–249.9).
TABLE 3
Carbon Yield and Peak Temperature of Products Resulting from
TPD after Methanol Adsorption at 300 K over H2-Reduced (at
Upon oxidation of the Ru(0001)–(1� 1)O surface, as de-
tailed in Section II, as well as in Refs. (3, 21, 39), RuO2(110)
is formed. Its surface was also analyzed (Fig. 10b). There is
a major change in both the Ru and O regions. The Ru(3d5/2
peak is now at about 280.8 eV, but a component at 280.2 eV
can still be present (it is worth recalling that LEED (21) and
STM (3) have shown the coexistence of Ru(0001)–O(1� 1)
with RuO2(110)). The small component at 280.2 eV is thus
due to the contribution from Ru metal atoms in Ru(0001)–
O(1 � 1) phase. The shift to lower binding energy of the
5
73 K, 1 atm, Cooled under H2 to 300 K) Polycrystalline RuO2
Reactant/product
Temperature (K)
Carbon yield (% )
Methanol (m/e 31)
Formaldehyde (m/e 30)
Methyl formate (m/e 60)
CO2 (m/e 44)
CO (m/e 28)
H2 (m/e 2)
360
360
360
6.82
20.63
11.25
49.97
11.32
—
360 and 450
360 and >400
360
H2O (m/e 18)
370
—

OXIDATION REDUCTIONS OVER RuO2
303
FIG. 11. LEED spot intensity (1, 1) of RuO2(110) single crystal that
has been subjected to increasing CO exposure followed by O2 treatment.
ii. Temperature-programmed reaction of CO over
RuO2(110). Successive CO-TPD experiments (after CO
exposure at 170 K) were conducted over stoichiomet-
ric RuO2(110)—as the starting material, but without
regeneration with O2 molecules in between runs. The
amount of CO2 produced decreased with successive runs
upon a cumulative exposure of 2.5 L, down to below the
detection limits of the QMS (Fig. 12a). This is due to
removal of available surface oxygen (most likely the two-
fold bridging oxygen atoms; see Section IV). This be-
havior is in fact observed only if the ramping stops just
after CO and CO2 desorptions have taken place (450 K).
In contrast, if CO-TPD runs are conducted up to elevated
FIG. 10. HRCLS spectra of Ru3d5/2 and O(1s) for (a) Ru(0001)–
1 � 1) O single crystal and (b) RuO2(110) single crystal.
(
O(1s) (from 530.07 to 529.48 eV) together with that of
the main Ru atoms to higher binding energy (from 280.12
to 280.8 eV) contributing to 11 = (O(1s)–Ru(3d5/2)) =
2
48.88 eV is very close to that observed on polycrystalline
materials (Table 1).
III.B.2. CO and O2 Reactions over RuO2(110)
Single Crystal
i. LEED spot intensity measurements. Oxygen exposure
to the clean RuO2(110) surface leads to the stabilization of
a weakly held oxygen species that desorbs with a peak at
4
20 K (see Fig. 1). This weakly held oxygen has been shown
to adsorb over cus-Ru atoms (9). The weakly held oxygen
serves as a precursor state to heal the missing bridging oxy-
gen atoms of the RuO2(110) surface. This behavior is clar-
ified in Fig. 11, in which the spot intensity of a RuO2(110)
related LEED beam is shown as a function of the CO ex-
posure (as well as O2 exposure) at a sample temperature
of 520 K. With increasing CO exposure the LEED inten-
sity decreased; a change in slope is observed at about 90 L.
Oxygen postexposure was able to recover a large part of
the original LEED intensity signal. A saturation is reached
after an O2 exposure of 80 L. The main point to make from
this figure is that the oxidation–reduction cycle of this ox-
ide single crystal surface can be conducted and monitored
under mild conditions (low temperatures and relatively low
exposures). To completely restore the LEED signal a more
prolonged exposure to O2 or severalO2-TPD runs, asshown
in Fig. 1, are required.
FIG. 12. TPD after methanol adsorption (2 L) over RuO2(110) single
crystal surface at 200 K.

3
04
MADHAVARAM ET AL.
temperatures, a different TPD profile is observed (Fig. 12b).
In this latter situation postdosing of CO (at 293 K) results
in a gradual increase of peak areas of desorbed CO2. Sat-
uration is achieved by ca. 570 K. This latter observation
provides clear evidence of self-regeneration of the active
sites due to a temperature effect. Comparison with STM
pictures, collected under similar conditions (3), indicates
that this behavior may be associated with oxygen migration
from the threefold sites to the twofold (bridging) sites (see
Section IV). In other words, this behavior is not necessarily
due to bulk to surface diffusion of lattice oxygen.
TABLE 4
Carbon Yield and Peak Temperature of Products Resulting
from TPD after Methanol Adsorption (2 L) over Stoichiometric
RuO (110) Single Crystal at 200 K
2
Reactant/product
Temperature (K)
Carbon yield (% )
CH3OH (m/e 31)
HCHO (m/e 30)
CO2 (m/e 44)
390
390
385
13.0
24.2
13.0
27.2
15.4
7.2
4
85
385
85
CO (m/e 28)
4
CO2/CO
CO2 + HCHO
1.78
64.4
iii. Temperature-programmed reaction of methanol over
RuO2(110). In Fig. 13 we present the thermal desorp-
tion data after the clean RuO2(110) surface has been
exposed to 2.4 L methanol at 200 K. The main product
ca. 70% ) leaving the surface after a rise in the sample tem-
(
perature is CO2. The main peak appears at ca. 480 K and
is accompanied by a small peak at 380 K. These peak po-
sitions compare favorably with those of the high-pressure
experiments (Fig. 8). CO desorption took place at the same
temperatures (380 and 480 K) but with a different distri-
bution. The low-temperature desorption was about twice
as large as the high-temperature desorption (see Table 4
for quantitative analysis). The similarity with Fig. 8 is easily
noticed. Methanol desorption takes place at 380 K. Most
of the formaldehyde leaves the surface at 380 K (a large
peak) although a small desorption at 500 K is also observed.
Both methanol and formaldehyde desorptions appear at
the same position as those observed in the high-pressure
work (compare Figs. 8 and 13).
IV. DISCUSSION
Several main points resulting from the above study are
worth mentioning.
1
. Ru cations in polycrystalline RuO2 can be easily re-
+
duced (either by H2 or upon Ar sputtering) as evidenced
by a shift of the XPS Ru(3d5/2) line by 0.7 eV (from 281.2 to
2
80.5 eV) (Fig. 2). This shift is consistent with that observed
by HRCLS from model surfaces of Ru(0001)–(1 � 1)O
and RuO2(110) single crystals (from 281.05 to 280.05 eV)
(
Fig. 10).
2
. One can measure the shift as a 1, where 1 = XPS
O(1s) � XPS Ru(3d5/2). 1 was equal to 248.88 and 248.5 eV
for RuO2 single crystal and polycrystalline materials, re-
+
spectively. On the reduced (either by H2 or by Ar bom-
bardment) RuO2 polycrystalline surface 1 = 249.9 eV,
which is identical to that of Ru(0001)–(1� 1)O single crystal
(
1 = 249.95 eV) (Table 1).
3
. XRD (Fig. 6) as well as TPR (Fig. 7) has indicated
that H2-reduced polycrystalline RuO2 has been bulk trans-
formed to Ru metal. Although the O(1s) to Ru(3d) peak-
area ratios decreased dramatically (Figs. 4 and 5), the
FIG. 13. Successive TPD after CO adsorption over RuO2(110) single
crystal surface without (a) and with (b) surface regeneration; regeneration
was conducted by heating the surface to the indicated temperature.

OXIDATION REDUCTIONS OVER RuO2
305
TABLE 5
Formaldehyde Yield during Methanol-TPD over Several Oxide Single Crystals and Computed
Activation Energies of Formaldehyde Desorption
Peak temperature
(K)
Activation energy f
(kcal/mol)
Oxide single crystals
Rutile TiO2(110)a
Yield (% )
References
0
0
0
—
—
—
520
640
560–580
450
450
45
43
47
40
41
44
42
b
Cubic ZrO2(100)
Rocksalt MgO(100)c
d
32 (0.03–0.06)g
40 (0.1)
Wurtzite ZnO(0001)-Zn
Rutile TiO2(001)-{011}
Fluorite CeO2(111) thin film
Rutile SnO2(110)e
1–2
4
h
—
35–36
10–20
15
24
28 (0.36–0.72)
28 (0.54)
24 (1)
Cubic ZrO2(110)
Rutile RuO2(11)
43
390
This work
a
Molecular adsorption at room temperature, UPS data; no other products were observed.
Other products included CH4 and CO.
Dissociative adsorption, XPS data; no other products were observed.
Peak areas and yield were computed from Fig. 1 in Ref. 40.
b
c
d
e
f
Given value = 11% . Variations from run to run put a yield value between 10 and 20% (42).
13 � 1
s ) (49).
Estimated by Redhead analysis, assuming first-order pre-exponential factor (10
g
h
Yield (% )/activation energy (kcal/mol).
Quantitative analyses were not provided; formaldehyde formation was highly sensitive to prior an-
nealing conditions (44).
residual presence of O atoms at binding energy higher than to formaldehyde can be compared to that of other oxides
that of O(1s) in RuO2, together with the absence of the single crystals such as ZnO (0001)–Zn (40), TiO2(001)-
RuO2 phase in XRD, suggests the formation of some dis- {011} faceted (41), SnO2(110) (42), ZrO2(110) (43), and
solved or chemisorbed oxygen on metallic Ru. The shift CeO2(111) (44) single crystals. TiO2(110) (45), MgO(100)
in the XPS O(1s) peak position to higher binding energy (46), and ZrO2(100) (43) single crystals showed no activ-
(
Fig. 3 and Table 1) is also indicative of chemisorbed oxy- ity toward methanol oxidation to formaldehyde. Interest-
gen atoms. ingly all these oxide single crystals have surface cations with
. The surface of RuO2(110) could be regenerated by at least one coordinative unsaturation. Table 5 shows the
dosing O2 molecules (<1 L, direct dosing) at room temper- TPD yield, the peak temperature, and a computed acti-
ature (Fig. 1). vation energy of formaldehyde desorption for these oxides
. The self-regeneration of the active centers for CO ox- together with those of RuO2(110) single crystal. RuO2 is in-
idation to CO2 was clearly shown to occur when one heats deed more active than the other oxides. One may estimate
the RuO2(110) to 570 K and above. the efficiency of the reaction by dividing the reaction yield
. Both polycrystalline RuO2 and RuO2(110) surfaces (% ) by the activation energy (or peak temperature). The
4
5
6
behaved similarly toward methanol oxidation to formalde- higher the number, the more active the surface. These num-
hyde, as well as toward its total decomposition to CO and bers were found as low as 0.03 mol/kcal for ZnO(0001)–Zn
CO2 (Figs. 8 and 13).
and rose to 1 mol/kcal for RuO2(110) single crystal. Equally
important, the other two rutile oxides with a (110) surface
In addition to the above six points, results from previous orientation are not active (see also Table 5). In other words,
and parallel works (1–3, 7, 8) have indicated that CO ad- the coordinative unsaturation is not the sole requirement
sorption occurs preferentially on cus-Ru (DFT calculation for making the oxidation. Other surface properties enter
(
7, 8), STM (3)). Similarly, LEED-IV analyses have sug- into consideration and these are discussed below.
gested that methanol adsorption occurs also on these same
sites (8).
Points 1 to 4 indicate one single conclusion. It is a sim-
We have asked three questions in the introduction and we
attempt to answer them with the help of this study below.
Since RuO2(110) is very active toward adsorption as well
ple task to make reduction-oxidation cycles on RuO2 single as oxidation reactions, the difference from the other oxides
crystal and polycrystalline surfaces using H2 or CO as re- must lie somewhere else. It has been demonstrated that CO
ducing agents and O2 as the oxidizing agent.
oxidation obeys the Mars–van Krevelen mechanism (3),
RuO2(110) single crystal is unique for its activity toward where the undercoordinated surface oxygen anions (bridg-
CO to CO2 and as such no comparison can be made to other ingoxygen anions) are consumed asa result oftheir reaction
oxides. However, its reactivity toward methanol oxidation with adsorbed CO making CO2. The surface defects created

3
06
MADHAVARAM ET AL.
that way are regenerated by gas-phase oxygen adsorption.
The binding energies of bridging O and on-top O were
In other words, there are at least three requirements for the computed to be 4.6and 3.1eV, respectively(valuesare given
reaction to occur.
a) An active lattice oxygen must be available to react
with adsorbed CO (such as the undercoordinated bridging
and terminal bonded oxygen atoms).
with respect to neutral atomic oxygen in the gas phase (9).
Therefore one would expect that the CO oxidation reaction
would start by consuming the on-top O species first and
then the bridging O species. This has been shown in recent
HREELS experiments (48). It was observed that the inten-
sity of on-top O species has decreased with CO exposure
(
(
b) Strong adsorption sites for the reactant must be
present.
at room temperature (i.e., reacted with CO to give CO ).
2
(
c) There must be an efficient dissociative adsorption of
Once all on-top O atoms have been removed, the intensity
of the signal from bridging O atoms decreased. When all the
available bridging oxygen atoms were consumed (complete
disappearance of their signal), the CO oxidation reaction
stopped. This result can be compared to those of Fig. 13,
where CO oxidation stops when the available surface oxy-
gen atoms are consumed.
In summary, this work has shown that detailed analysis of
a well-defined surface can shed light on a complex reaction
such as CO or methanol oxidation, over a polycrystalline
oxide. The fact that RuO2 is metallic, and thus allows ex-
tensive LEED and electron spectroscopy studies, together
with its unusually high activity toward oxygen and CO ad-
sorptions and reactions has been of tremendous help in
comparing the activity of a well-defined surface of
RuO2(110) to that of the polycrystalline material.
O2 molecules to regenerate lattice oxygen ions.
Requirement (a) has been clearly shown in this study by
the following:
(
i) Adsorption of CO on RuO2(110) single crystal gives
large amounts of CO2. This amount decreases with suc-
cessive runs (on the nonregenerated surface) until its dis-
appearance. Yet, it can be easily switched back by simply
flashing the surface to 570 K (lattice oxygen migration to
the active sites) (Fig. 13).
(
ii) Comparing the methanol TPD runs over the sto-
ichiometric and the H2-reduced polycrystalline RuO2, we
observe that the amount of CO2 decreased by almost 50%
and that of formaldehyde (the partial oxidation product)
increased to 20% of the product yield (Table 3).
Requirement (b) is indeed fulfilled. As mentioned in the
introduction, the adsorption energy of CO over cus-Ru in
the rutile RuO2(110) structure (1.2 eV (3)) is ca. 4 and 5
times stronger than the corresponding adsorption over cus-
Ti of TiO2(110) or cus-Sn of SnO2(110), respectively. A
weak adsorption will result in an easy displacement of the
reactant by other reactant/products (such as water) and the
surface becomes nonreactive, particularly if the oxidation
process is to occur at low temperatures. Requirement (c) is
also met on RuO2. Dosing O2 molecules followed by flash-
ing the surface to ca. 600 K (to remove on-top oxygen) over
RuO2(110) single crystal (Fig. 1) fully restored the surface.
This is also evidenced by LEED spot intensity measure-
ments (Fig. 11).
REFERENCES
1
2
3
4
5
. B o¨ ttcher, A., Niehus, H., Schwegmann, S., Over, H., and Ertl, G.,
J. Phys. Chem. B 101, 11185 (1997).
. B o¨ ttcher, A., Rogozia, M., Niehus, H., Over, H., and Ertl, G., J. Phys.
Chem. B 103, 6267 (1999).
. Over, H., Kim, Y. D., Seitsonen, A. P., Wendt, S., Lundgren, E., Schmid,
M., Varga, P., Morgante, A., and Ertl, G., Science 287, 1474 (2000).
. Casarin, M., Maccato, C., and Vittadini, A., J. Phys. Chem. B 102,
10745 (1998).
. Kim, Y. D., Schwegmann, S., Seitsonen, A. P., and Over, H., J. Phys.
Chem. B 105, 2205 (2001).
6
7
. Melle-Franco, M., and Pacchioni, G., Surf. Sci. 461, 54 (2000).
. Kim, Y. D., Seitsonen, A. P., and Over, H., Phys. Rev. B 6311(11), 5419
(
2001).
8. Kim, Y. D., and Over, H., unpublished LEED analysis.
. Kim, Y. D., Seitsonen, A. P., Wendt, S., Wang, J., Fan, C., Jacobi, K.,
Over, H., and Ertl, G., J. Phys. Chem. B 105, 3752 (2001).
0. B o¨ ttcher, A., and Niehus, H., Phys. Rev. B 60, 14396 (1999).
9
One of the intriguing results of this present study is
that temperature-programmed reactions under UHV (sin-
gle crystal) and high-pressure conditions (polycrystalline
samples) are virtually identical. This also means that the
1
1
1. Kim, Y. D., Over, H., Krabbes, G., and Ertl, G., Top. Catal. 14, 95
(
2001).
reaction pathways are very much alike and therefore the 12. Idriss, H., and Barteau, M. A., Adv. Catal. 45, 261 (2000).
1
1
3. Idriss, H., and Seebauer, E. G., Catal. Lett. 66, 139 (2000).
4. Goldan, P. D., Kuster, W. C., and Fehenfeld, F. C., J. Geophy. Res. 102,
active centers on RuO2 are the same. Moreover, we also
know that the efficiency for CO oxidation does not depend
very much on the orientation of RuO2 (47), at least not
when we compare the (110), (100), and (101) orientations—
all three surfaces contain cus-Ru. We conclude from this
observation that the presence of cus-Ru atoms is an impor-
tant requirement for the high activity of RuO2. The (110),
6315 (1997).
1
1
5. Bates, S. P., Kresse, G., and Gillan, M., Surf. Sci. 409, 336 (1998).
6. Calatayud, M., Andr e´ s, J., and Beltr a´ n, A., Surf. Sci. 430, 213
(1999).
1
1
1
2
7. Ko, E. I., Benziger, J. B., and Madix, R. J., J. Catal. 62, 264 (1980).
8. Idriss, H., Kim, K. S., and Barteau, M. A., J. Catal. 139, 119 (1993).
9. Monti D. A. M., and Baiker, A., J. Catal. 83, 323 (1983).
0. Over, H., Bludau, H., Skottke-Klein, M., Moritz, W., and Ertl, G., Phys.
Rev. B 46, 4360 (1992).
(
100), and (101) surfaces of RuO2 are low-energy surfaces
that should equally be the prevailing orientations found on
polycrystalline RuO2.
21. Kim, Y. D., Seitsonen, A. P., and Over, H., Surf. Sci. 465, 1 (2000).

OXIDATION REDUCTIONS OVER RuO2
307
2
2
2. Chan, H. Y. H., Takoudis, C. G., and Weaver, M. J., J. Catal. 172, 336
1997).
3. Cox, P. A., Goodenough Cox, J. B., Tavener, P. J., Telles D., and Edgell,
R. G., J. Solid State. Chem. 62, 360 (1986).
36. Malet, P., and Caballero, A., J. Chem. Soc., Faraday Trans. 1 84 , 2369
(1988).
37. Hurst, N. W., Gentry, S. J., Jones, A., and McNicol, B. D., Catal. Rev.—
Sci. Eng. 24, 233 (1982).
(
24. Kim, K. S., and Winograd, N., J. Catal. 35, 66 (1974).
25. K o¨ tz, R., Lewerenz, H. J., and Stucki, S. 130, 825 (1983).
26. Zhang, C. J., Hu, P., and Alavi, A., J. Am. Chem. Soc. 1212, 7931
38. Lloyd, J. A., Manner, W. L., and Paffett, M. T., manuscript in review.
39. Over, H., Seitsonen, A. P., Lundgren, E., Wiklund, M., and Andersen,
J. N., Phys. Rev. Lett., submitted.
(
1999).
40. Vohs, J. M., and Barteau, M. A., Surf. Sci. 176, 91 (1986).
41. Kim, K. S., and Barteau, M. A., Surf. Sci. 223, 13 (1989).
42. Gercher, V. A., Cox, D. F., and Themlin, J-M., Surf. Sci. 306, 279
(1994).
2
2
2
7. Atanasoska, L. J., O’Grady, W. E., Atanasoski, R. T., and Pollak,
F. H., Surf. Sci. 202, 142 (1988).
8. Shen, J. Y., Adnot, A., and Kaliagine, S., Appl. Surf. Sci. 51, 47
(
1991).
9. G o¨ pel, W., Rocker, G., and Feierabend, R., Phys. Rev. B. 28, 3427
1983).
43. Dilara P. A., and Vohs, J. M, Surf. Sci. 321, 8 (1994).
44. Siokou, A., and Nix, R. M., J. Phys. Chem. B 103, 6984 (1999).
45. Onishi, H., Aruga, T., Egawa, C., and Iwasawa, Y., Surf. Sci. 193, 33
(1988).
(
3
3
3
0. Raup G. B., and Dumesic, J. A., J. Phys. Chem. 89, 5240 (1985).
1. Idriss, H., and Barteau, M. A., Catal. Lett. 26, 123 (1994).
2. Reztova, T., Chang, C.-H., Koresh, J., and Idriss, H., J. Catal. 185, 223 47. Over, H., Gierer, M., Bludau, H., Ertl, G., and Tong, S. Y., Surf. Sci.
46. Peng, X. D., and Barteau, M. A., Langmuir 7 1426 (1991).
(
1999).
314, 243 (1994).
3
3
3. Cotton F. A., and Mague, J. T., Inorg. Chem. 5, 317 (1966).
4. Rogers, D. B., Shannon, R. D., Sleight, A. W., and Gillson, J. L., Inorg.
48. Fan, C., Wang, J., Jacobi, K., and Ertl, G., J. Chem. Phys. 114, 10058
(2001).
´
Chem. 8, 841 (1969).
5. Lele, S., and Anantharaman, T. R., Acta Cryst. A24, 654 (1968).
49. Redhead, P. A., Vacuum 12, 203 (1962).
50. Madhavaram, H., and Idriss, H., J. Catal. 184, 533 (1999).
3