Catalytically active states of Ru(0001) catalyst in CO oxidation reaction

Article abstract of DOI:10.1016/j.jcat.2006.02.019

Identifying the composition of the catalytically active state of metal catalysts under dynamic operating conditions is of particular importance for oxidation catalysis. Here we report new insights into the chemical identity of different catalytically active states formed on a Ru(0001) catalyst during CO oxidation at various reaction temperatures. The changes in the surface composition of the Ru catalyst and the CO₂ yield under varying reaction conditions in the 10^-4 - 10^-1 mbar pressure range were followed in situ by synchrotron-based high-pressure X-ray photoelectron spectroscopy and mass spectroscopy. The results reveal that the catalytic activity of a few layers thick surface oxide without well-defined stoichiometry and structure is comparable with that of the stoichiometric RuO₂(110) phase. This surface oxide forms under reaction conditions when RuO₂ formation is kinetically hindered and can coexist with RuO₂ in wide temperature and pressure ranges.

Full text of DOI:10.1016/j.jcat.2006.02.019

Journal of Catalysis 239 (2006) 354–361
www.elsevier.com/locate/jcat
Catalytically active states of Ru(0001) catalyst in CO oxidation reaction
a
a
a
a
a
a
R. Blume , M. Hävecker , S. Zafeiratos , D. Teschner , E. Kleimenov , A. Knop-Gericke ,
a
b
b
b,∗
R. Schlögl , A. Barinov , P. Dudin , M. Kiskinova
a
Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, 14195 Berlin, Germany
b
Sincrotrone Trieste, AREA Science Park-Basovizza, Trieste 34012, Italy
Received 24 January 2006; revised 11 February 2006; accepted 15 February 2006
Available online 20 March 2006
Abstract
Identifying the composition of the catalytically active state of metal catalysts under dynamic operating conditions is of particular importance
for oxidation catalysis. Here we report new insights into the chemical identity of different catalytically active states formed on a Ru(0001) cat-
alyst during CO oxidation at various reaction temperatures. The changes in the surface composition of the Ru catalyst and the CO yield under
2
−
4
−1
varying reaction conditions in the 10 –10 mbar pressure range were followed in situ by synchrotron-based high-pressure X-ray photoelectron
spectroscopy and mass spectroscopy. The results reveal that the catalytic activity of a few layers thick surface oxide without well-defined stoi-
chiometry and structure is comparable with that of the stoichiometric RuO (110) phase. This surface oxide forms under reaction conditions when
2
RuO formation is kinetically hindered and can coexist with RuO in wide temperature and pressure ranges.
2
2

2006 Elsevier Inc. All rights reserved.
Keywords: CO oxidation; Ru; High pressure XPS; Catalytically active states
1
. Introduction
oxidation states at various partial pressures and temperatures
10–16]. After completion of the O-(1 × 1) adsorption phase
[
Following the pioneering works demonstrating very high ac-
[17], the suggested oxidation pathway of the Ru(0001) surface
involves as an important intermediate step the incorporation of
O atoms between the first and second Ru layers and the for-
mation of an Oad–Ru–Osub trilayer hosting 2 ML of oxygen
[18,19], with 1 ML equalling the number of Ru atoms on the
(0001) surface. The conversion into a RuO2(110) structure is
supposed to occur above the critical thickness of two O–Ru–O
trilayers (with an oxygen content of 4 ML). Spectroscopic evi-
dence for subsurface oxygen species has been provided by pho-
toelectron diffraction and combined photoemission and thermal
desorption spectroscopy [20,21]. According to another model,
based on investigations of Ru oxidation carried out at tempera-
tivity of the so-called O-rich Ru(0001) surface in CO oxidation
1,2], the rutile RuO2(110), formed on the Ru(0001) surface
[
under realistic oxidation conditions, is considered the catalyt-
ically active state [3,4]. The rutile RuO2(110) surface consists
of alternating rows of six-fold and unsaturated five-fold oxygen
coordinated Ru atoms, with the latter, called cus-Ru, playing
a prominent role as active sites in CO oxidation. The inspired
UHV surface science and theoretical studies of CO oxidation
reaction on a model RuO2(110) single-crystal surface reached
the consensus that the elementary reaction steps involve CO
and O adsorption on cus-Ru, followed by a reaction between
cus-CO and cus- or bridge O atoms [5–9]. But the RuO2(110)
surface is an idealised case of an active Ru catalyst, which has
been prompted by the parallel studies focused on the oxidation
mechanism of the Ru(0001) surface and the stability of the Ru
tures above 600 K, the RuO nucleus is formed as long as the
2
oxygen exceeds 1 ML, the coverage of the most dense (1 × 1)
adsorption phase, and the RuO2(110) film grows progressively
in an autocatalytic manner [10,22].
Recent XPS microscopy and TDS studies demonstrated that,
starting from an atomically clean Ru(0001) surface, forma-
*
Corresponding author. Fax: +39 040 3758565.
E-mail addresses: kiskinov@fhi-berlin.mpg.de,
tion of a stoichiometric RuO phase is kinetically hindered at
2
kiskinova@elettra.trieste.it (M. Kiskinova).
temperatures below 500 K and occurs readily at higher tem-
0021-9517/$ – see front matter  2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.02.019

R. Blume et al. / Journal of Catalysis 239 (2006) 354–361
355
Fig. 1. Ball models of the oxidation states of Ru(0001), where O atoms are the small balls. (A) (1 × 1) saturated oxygen adsorption phase with ∼1 ML oxygen, the
onset of subsurface O incorporation at T ꢀ 400 K; (B) “surface oxide” with ∼1 ML incorporated oxygen, the dominant phase formed at T < 500 K; (C) “surface
oxide” with ꢀ 2 ML incorporated oxygen; (D) rutile RuO (110) phase.
2
peratures [2,13,14,21]. The main reason for such temperature
dependence of the Ru oxidation state is that the incorporation
up to 3 ML of oxygen below 500 K is limited to the top two
or three Ru layers. These O-rich thin films with poorly defined
RuxOy structure are called “surface oxides” and are identified
as the precursor where the RuO2 nucleates and grows at tem-
peratures above 550 K.
in situ using a setup combining differential pumping and elec-
trostatic focusing of the emitted photoelectrons [24]. Briefly,
the sample was mounted inside the reaction cell, 2 mm away
from an aperture (1 mm diameter), which provided the entrance
for the emitted photoelectrons and reaction products to the dif-
ferentially pumped stages of the electrostatic lens system of the
hemispherical analyser (Phoibos 150, SPECS GmbH). The gas-
phase products, used to measure catalytic activity, were moni-
tored on-line with a Hiden mass spectrometer located in the first
differentially pumped lens stage. The sample was heated by a
Fig. 1 shows schematically the “oxidation” states of the
Ru(0001) that have been proven experimentally [2,4,13,14,21].
In this paper we call these states the adsorption phase, referring
to the state in which oxygen in present only on the surface with
maximum coverage 1 ML (A); surface oxide, with ∼1–3 ML
oxygen incorporated within the top few Ru layers (B and C);
and rutile RuO2 phase (D). An important finding of XPS mi-
croscopy is that the surface oxide and RuO2 phases coexist in a
wide temperature–pressure range, even when formed in a pure
O2 ambient [13,21]. Undoubtedly under reaction conditions,
the CO will drive the oxidation state away from the equilibrium
achieved in O2 ambient, implying that the temperature–pressure
space of coexistence of the two phases may be expanded.
The temperature dependence of the actual “oxidation” state
and the morphology of the Ru surface reopen the disputable is-
sue about the active state of Ru catalysts during CO oxidation.
Here we use a specially built reaction chamber for simultane-
ous monitoring of the chemical state of the catalyst surface and
the reaction product released in the gas phase at pressures up
to a few mbar. We verified the catalytic activity of the differ-
ent oxidation states of Ru(0001) catalyst, including the final
stoichiometric RuO2, starting from a metallic Ru surface and
following in situ the temperature evolution of the catalyst sur-
face composition in CO + O2 environment close to the realistic
oxidation, reduction, and steady-state reaction conditions.
laser from the back side. The CO and O gas flows into the re-
2
action cell were regulated using leak valves.
The Ru(0001) sample was cleaned before each reaction cy-
cle using well-established procedures [14,25]. The base pres-
−8
sure in the chamber was ∼2 × 10 mbar, which reduced the
lifetime of the atomically clean surface at room temperature.
The temperature ramp was 2 K/min, and acquisition of a set of
Ru 3d and O 1s spectra took a maximum of 4 min. Thus, the
maximum difference between the temperatures at which the Ru
3
d and O 1s spectra were taken was less than 8 K. All spec-
tra were normalized to the incident photon flux, monitored by
a photodiode with known quantum efficiency. The Ru 3d and
O 1s spectra were measured with photon energies of 450 and
6
50 eV, respectively. Using the universal curve for the electron
mean free path [26] the effective escape depths for the O 1s and
Ru 3d photoelectrons was ∼5 Å, limiting the probing depth to
the top few layers (∼10 Å). We also used higher photon ener-
gies to increase the probing depth when necessary.
3
. Results and discussion
The experiments were carried out at different partial pres-
−
4
2
. Experimental
sure ratios of CO and O in the range of 10 –0.1 mbar. By
2
varying the CO/O2 pressure ratio, we could reproduce oxida-
tion, reduction, and steady-state conditions of the working Ru
catalyst. Simultaneous monitoring of the dynamic response of
the O 1s and Ru 3d5/2 core-level spectra and the CO2 produc-
tion allowed us to correlate the catalytic activity to the actual
oxidation state of the Ru catalyst.
The experiments were performed in a high-pressure XPS sta-
tion designed and constructed in FHI-MPG [23], attached to the
beamline U49/2-PGM2 at the BESSY synchrotron radiation fa-
cility in Berlin. The overall energy resolution of the beamline
was 0.1 eV at 500 eV. The photoelectron spectra were measured

356
R. Blume et al. / Journal of Catalysis 239 (2006) 354–361
Table 1
Energy positions of Ru 3d
oxidation states. The shift of the Ru 3d
energy reference (the position of the Ru 3d
given in the brackets
atoms at the RuO2 surface (528.7 eV) and bulk RuO2 oxygen
and O 1s components measured for the different
5
/2
(529.5 eV) [4,13]. The bridge oxygen component is clearly vis-
ible only when the RuO2(110) surface is very well ordered and
without oxygen vacancies. Because the bridge oxygen partici-
pates in the oxidation reactions [9,10], the corresponding O 1s
component should be strongly reduced in ambient CO.
We evaluated the presence of adsorbed CO from the O 1s
spectra, because the C 1s peak overlaps with the Ru 3d3/2 core
level. The O 1s peaks of CO have binding energies between
components with respect to the zero-
5/2
bulk component at 280.1 eV) is
5/2
State
Binding energy (eV)
280.1 (0)
Component
Ru
Ru(I)
Ru(II)
Ref.
Clean Ru
[25]
[25]
[25]
bulk
2
2
79.75 (−0.35)
80.25 (0.15)
Adsorbed phase
280.08
Ru(I)–1O
Ru(I)–2O
Ru(I)–3O
[25]
[4,25]
[4,25]
ad
ad
ad
5
30.8 and 531.8 eV and can be easily resolved from those of
2
2
80.48 (0.4)
81.03 (0.93)
adsorbed oxygen and oxide [28].
RuO phase
280.74 (0.64)
RuO -bulk
2
Ru-cus
Satellite
[4,13]
[27]
2
280.45 (0.35)
83 (2.92)
3
.1. Reduction of the surface oxide and RuO2
2
“
Surface oxide”
280.5 ± 0.05 (∼0.4)
RuxOy
Ru(I)–2O O
Ru(I)–3O
Ru(II)–O
[21]
[21]
[21]
[21]
The surface oxide stage (Fig. 2) and stoichiometric RuO2
2
2
2
80.6 ± 0.05 (∼0.8)
81.4 ± 0.05 (∼1.3)
80.9 ± 0.05 (∼0.5)
ad sub
O
ad sub
phase (Fig. 3) were formed by exposing Ru(0001) to 5 ×
−
2
10 mbar O2 at 450 and 620 K, respectively. The surface ox-
ide contained ∼1.5–2.0 ML oxygen located on the surface and
between the top and second Ru layers. The experiments were
sub
O 1s
530.0
O
ad
& O
[13]
[4,13]
[2]
sub
5
5
29.5
28.7
O in RuO -bulk
“Bridge” O
2
−
4
carried out in the 10 mbar range at CO/O partial pressure
2
O 1s–CO
530.8–531.8
[28]
ratio of 4, that is, in excess of CO. The reduction rate was
compatible with simultaneous monitoring of the XPS and mass
spectra, while slowly increasing the sample temperature, start-
ing at 370 K. The results provide the necessary basis for under-
standing and interpretating the in situ CO oxidation data.
Fig. 2a shows representative Ru 3d5/2 spectra obtained dur-
ing reduction of surface oxide. The bottom Ru 3d5/2 spectrum is
The necessary basis for identifying the adsorption, surface
oxide, and RuO2 states and verifying their actual role in the
CO oxidation reaction was provided by the already available
Ru 3d5/2 and O 1s core level and TD spectroscopy data. The
established binding energies of the Ru 3d5/2 and O 1s compo-
nents corresponding to the adsorption state, surface oxide state,
and rutile RuO2 phase are summarized in Table 1.
The Ru(I) and Ru(II) components account for emission from
the Ru atoms in the first and second layers, respectively. They
undergo distinct chemical shifts when binding to O, determined
by the coordination number of O atoms and their bonding
−
2
of a “clean” Ru surface, before exposure to 1×10 mbar O2 at
4
50 K. Although some CO from residual gas adsorbs on the sur-
−
8
face at 2 × 10 mbar base pressure, its coverage is relatively
low at 450 K, and the surface component Ru(I) is still present.
For deconvolution, we used a broader component to account
for the bulk and Ru(II) contributions. The Ru 3d5/2 spectrum
of the surface oxide before the reaction contains components at
configuration [4,25]. The incorporation of O subsurface, Osub
leads to a distinct shift of Ru(II), resulting in the component
Ru(II)–Osub at ∼0.5 eV, well separated from the Ru po-
,
∼
0.5, 0.8, and at 1.3 eV, which appear only in the presence of
about 1 ML of incorporated oxygen [21]. Our previous XPS mi-
croscopy results did not show significant lateral inhomogeneity
for such low-temperature oxidation states [21]. The broad O 1s
spectrum in the bottom Fig. 2b peaks at 530.0 eV, correspond-
ing to surface oxide.
bulk
sition [21]. The Ru(I) energy shift induced by the subsurface
oxygen accounts for the Ru(I)–2O O and Ru(I)–3O O
ad sub
ad sub
components at ∼0.8 and 1.3 eV, assigned to the Ru(I) atom
coordinated with two and three O adatoms, respectively, and
subsurface oxygen below. When the total amount of adsorbed
and incorporated oxygen exceeds ∼3 ML, a new broad compo-
nent at ∼0.4 eV, RuxOy, grows. This corresponds to the most
advanced oxidation state below 500 K, with a poorly defined
structure and thickness of about two rutile layers (5–6 Å) [21].
Although the energy position of the RuxOy component is prac-
tically identical to that of the Ru(I)–2Oad component, the Ru
The onset of reduction was at ∼390 K with a maximum CO
2
yield at ∼420 K, followed by a rapid decay to zero at ∼450–
4
70 K. This leads to a decrease of the O 1s signal (Fig. 2c) and
significant changes in the Ru 3d5/2 spectrum; the Rubulk grows,
accompanied by rapid attenuation of the Ru(I)–2OadOsub at
0
.8 eV, whereas the Ru(II)–Osub at 0.5 eV broadens and gradu-
ally shifts to 0.4 eV, the position of the Ru(I)–2Oad component.
These results indicate that once the oxygen from the surface
is consumed, the limited amount of oxygen incorporated be-
tween the first and second layers is thermodynamically driven
to segregate to the surface [18]. Thus the system converts into
an oxygen adsorption phase. A peculiar feature in the evolution
of the Ru 3d5/2 spectra is that the Ru(I)–3Oad component at
0.9 eV does not gain sensible intensity, indicating direct reduc-
tion of the surface oxide into an adsorption state with moderate
oxygen coverage of ∼0.7 ML.
3
d spectra are very different, because of the strong attenua-
tion of Rubulk component due to screening by the surface oxide
film. The RuO2(110) phase has two components, corresponding
to the cus-Ru and six-fold coordinated Ru atoms and a broad
satellite at ≈283 eV [4,13,27]. The O 1s spectra from the ad-
sorption and surface oxide states appear at 530.0 eV, with the
latter being only a bit broader, indicative of multiple bonding
configurations [13]. The O 1s spectrum of the RuO (110) phase
has two components, reflecting the emission from the bridge O
2

R. Blume et al. / Journal of Catalysis 239 (2006) 354–361
357
Fig. 2. (a) From bottom to top: Ru 3d
spectra taken before exposure to O , after oxidation at 450 K and cooling the “surface oxide” to 370 K and following
2
5/2
reduction with increasing the temperature. The “ad & sub” annotation indicates the three components, Ru(I)–3O
O
ad sub
, Ru(I)–2O
O
, and Ru(II)–O , which
ad sub sub
are fingerprints of the O-rich state with incorporated oxygen. (b) O 1s spectra taken after oxidation at 450 K (bottom) and after exposure of this surface to CO at
20 K (top). (c) O 1s intensity as a function of reduction temperature. The O 1s signal is normalised assuming that the intensity of the adsorption phase corresponds
3
◦
−4
−4
to 0.7 ML. dT/dt = 2 /min. Reduction conditions: p = 2 × 10 mbar, p = 0.5 × 10 mbar.
CO
O
2
Simple calculations, considering the electron escape depth
for our experimental setup, predict that for surface oxide with
an initial load of 2 ML, the reduction to 0.7 ML adsorption
phase should cause a ∼50% decrease in O 1s intensity (1 ML
subsurface oxygen contributes to ∼30% of the initial signal).
The experimentally measured decrease in O 1s intensity of
the probing depth. This indicates that the stoichiometric RuO2
film formed is thicker than ∼15 Å. The bulk oxide compo-
nent dominates the spectra, but the Ru-cus component is also
present, suggesting a uniform and structured RuO (110) sur-
2
face. The O 1s spectrum in Fig. 3b peaks at 529.5 eV, as ex-
pected for the RuO2 phase, with broadening on the side of the
bridge O surface component at 528.7 eV [4,13].
∼
35% indicates that the total initial load of oxygen within the
top two Ru layers is somewhat less than 2 ML. In fact, the
CO2 formation starts at ∼420 K and is marked by the ap-
pearance and growth of the Rubulk component with increasing
reaction temperature and decreasing O 1s intensity. As illus-
trated by the plot in Fig. 3c, the most rapid loss of O 1s intensity
occurs at ∼450–470 K, where the maximum CO2 yield was
monitored. The evolution of the Ru 3d5/2 spectra in Fig. 3a il-
lustrates how the ongoing reaction continuously consumes the
oxide. Above ∼520 K, the Ru(I)–2Oad component begins to
grow as well, accompanied by an accelerated increase in the
Rubulk and attenuation of the oxide component. This indicates
a patchy morphology of the surface, consisting of diminishing
oxide islands and inactive adsorption phase. Above ∼580 K, the
Ru(I)–3O O component at ∼1.3 eV is already rather weak
ad sub
in the Ru 3d5/2 spectra from the surface oxide before the reduc-
tion, compatible with the presence of vacancies on the surface.
This finding also is consistent with the evolution of the Ru 3d5/2
spectra during the reduction, which does not pass through a sat-
urated adsorption layer of 1 ML. One possible explanation is
that once the oxygen starts to incorporate below the surface, it
naturally leaves vacancies on the surface [18,29] where CO can
stick and react. Indeed, when the freshly formed surface oxide
was cooled to 370 K and exposed to ambient CO, a shoulder
at about 531.7 eV grows in the O 1s spectrum due to adsorp-
tion of CO molecules (Fig. 2b). By varying the CO/O pressure
ratio, the initial surface oxide can be restored under oxidizing
conditions at 450 K, whereas under reducing conditions, the
adsorbed O can be further depleted only at temperatures above
Ru 3d spectrum becomes almost identical to that of the inac-
5/2
tive adsorption state with ∼0.7 ML of oxygen, achieved during
reduction of the surface oxide (see Fig. 2a). The apparent di-
rect conversion from RuO2 to an inactive adsorption phase is
supported by the evolution of the RuO2 satellite (not shown),
distinguishable in the Ru 3d spectra up to ∼550 K. This is con-
sistent with the results in Fig. 2 showing that the surface oxide
is unstable at temperature above 500 K under reduction condi-
6
00 K, which leaves a diluted adsorption state with mostly sin-
gle coordinated Ru atoms.
The Ru 3d5/2 spectrum after oxidation in 10 2 mbar O at
−
2
6
20 K (Fig. 3a) did not contain a Rubulk component, even when
measured using photon energies of 1000 eV, which increased

358
R. Blume et al. / Journal of Catalysis 239 (2006) 354–361
Fig. 3. (a) From bottom to top: Ru 3d
spectra taken after oxidation at 620 K and cooling to 350 K, and following reduction with increasing the temperature.
5/2
(b) O 1s spectra taken before introducing CO (top) and after reduction to adsorption phase (bottom). (c) O 1s intensity changes with increasing the reduction
temperature. The O 1s signal is normalised against the intensity of the adsorption phase, corresponding to 0.7 ML. dT/dt = 2 K/min. Reduction conditions:
−
4
−4
p
= 2 × 10 mbar, p = 0.5 × 10 mbar.
CO
O
2
tions, as well as Fig. 3b showing the shift in the O 1s spectrum
to 530.0 eV.
The mass spectrometry and XPS data allowed us to correlate
the CO2 yield and oxidation state of the catalyst at different re-
action temperatures, ranging from 370 to 600 K. Fig. 4a shows
the evolution of the CO2 yield with increasing reaction temper-
ature. The plot has three distinct regions: a very weak increase
up to ∼420 K, a sharp onset of the oxidation reaction at ∼420 K
with a continuously increasing CO2 yield up to ∼550 K, fol-
lowed by a flat maximum and slow decline above ∼580 K.
The evolution of the Ru oxidation state with increasing re-
action temperature is illustrated by the selected O 1s and Ru
3d5/2 spectra in Figs. 5a and b. The surface composition of
Ru catalyst in the CO + O2 ambient at temperatures before the
sharp onset of the CO2 production is best represented by the
O 1s spectra Fig. 5a, taken at temperatures below 420 K. They
clearly show the presence of CO on the surface (component at
In brief, the stabilities of the surface oxide and the stoi-
chiometric RuO2 under reduction condition appear comparable;
both can be easily reduced in excess of CO in the gas phase mix-
ture at temperature above 400 K and the final reduced state at
temperature below 600 K is the same inactive adsorption phase
with ∼0.7 ML of oxygen. It should be noted that in surface
science experiments, carried out far from realistic temperature
and pressure conditions, no reduction of the RuO (110) was
2
observed even in extreme excess of CO (i.e., CO/O2 pressure
ratios of 10) [7].
3
.2. Catalytically active transient states of Ru surface during
CO oxidation at different temperatures
∼
531.7 eV) up to ∼400 K. CO removal at 400–420 K is ac-
companied by some loss in O 1s intensity at 530.0 eV, assigned
to surface and subsurface oxygen, and by a weak increase in
The experiments were carried out at 0.1 mbar and a CO/O2
partial pressure ratio of 1. We started with a clean Ru(0001) sur-
face and slowly increased the temperature after introducing the
reactants in the gas phase. The excess of oxygen with respect
to the reaction stoichiometry provided slightly oxidizing con-
ditions to ensure the formation of stoichiometric RuO2 above
the CO yield. This suggests that the adsorbed CO reacts with
2
surface O and leaves the surface as CO . This scenario is in ac-
2
cordance with the Ru 3d5/2 spectrum in Fig. 5b taken at 395 K,
where the absence of Ru(I)–2O O component indicates that
ad sub
the O surface coverage is substantially reduced after the re-
5
00 K, which did not readily occur when using a stoichiometric
moval of CO. We did not consider here the Ru 3d5/2 spectrum
before CO removal, because of the unknown effect of the CO
on the surface core level position. The O 1s peak at 530.0 eV
gains intensity above 400 K, indicating further accumulation
of oxygen; this also leads to the increase of Ru(II)–Osub and
CO/O2 ratio of 2. Another reason for choosing this CO/O2 ra-
tio is that the highest turnover rates over working Ru catalysts
under realistic conditions were measured at about equal partial
pressures of the reactants [30].

R. Blume et al. / Journal of Catalysis 239 (2006) 354–361
359
Fig. 4. (a) CO yield as a function of reaction temperature. (b) O 1s intensity after subtraction of the CO contribution as a function of reaction temperature. The O 1s
2
signal is normalised as in the reduction experiments (see Figs. 3b and 2b). The O 1s intensity up to ∼500 K reflects only the surface and subsurface content, whereas
above 500 K the increase is dominated by the formation of an RuO phase. (c) Plot of the CO yield versus O content at the surface and near surface region. The
2
2
dashed line in (a)–(c) indicates the onset of the RuO growth.
2
Fig. 5. (a) Ru 3d
(a) and O 1s (b) spectra illustrating the catalyst composition developed during CO oxidation with increasing of the reaction temperature from
/2
5
−
1
−1
370 to 600 K. dT/dt = 2 K/min. Reaction conditions: p = 0.5 × 10 mbar, p = 0.5 × 10 mbar. The insert in (a) is a Ru 3d map illustrating the surface
CO
O
2
6
morphology developed after exposure to 10 L O at 670 K with RuO (dark) and RuxOy (bright) islands.
2
2
Ru(I)–2OadOsub components in the Ru 3d5/2 spectra. The Ru
d5/2 and O 1s spectra undergo negligible line shape changes
Because the CO2 formation reaction cannot disturb the sur-
face composition to any significant degree, the observed tem-
perature dependence of the reaction rate is most likely related
to further accumulation of oxygen, facilitated at higher tem-
peratures. This accounts for the slow gradual increase in O 1s
intensity below 500 K, as illustrated in Fig. 4b. In this con-
text, we note that the lattice stress induced by incorporation of
O produces local changes in both the geometric and electronic
structures of the top Ru layer [18,31], thereby affecting the Ru–
O and Ru–CO adsorption bond and/or modifying the adsorbate
bonding configurations. As a result, the barrier to achieve the
transition CO–O state may be reduced, increasing surface ac-
tivity and CO2 yield.
3
at 420–500 K, despite the continuous increase in CO2 yield.
The deconvoluted Ru 3d5/2 spectra have a dominant Ru(II)–
Osub component at 0.5 eV and a rather weak Ru(I)–2OadOsub
component at 0.8 eV. This means that under the actual opera-
tional conditions, the catalyst contains a significant amount of
subsurface oxygen, whereas the O species on the surface are
effectively consumed by the ongoing reaction. The rapid dy-
namics at the surface is confirmed by the practical absence of
CO-related feature in the O 1s spectra, indicating a short life-
time of the CO species on the surface before being reacted off.

360
R. Blume et al. / Journal of Catalysis 239 (2006) 354–361
A natural consequence of the progressive incorporation of
oxygen with further increases in reaction temperature is the nu-
cleation and growth of stoichiometric RuO2. Indeed, the Ru
d5/2 spectra in Fig. 5b undergo significant changes above
00 K, due to formation of RuO2 until a steady-state com-
position is reached and maintained in the 550–600 K range.
A weak Rubulk component can still be distinguished in the
Ru 3d5/2 spectrum of this steady-state, compared with much
thicker RuO2 films grown in pure O2 ambient (see Fig. 3a).
This indicates that RuO2 growth is slower during CO oxidation,
reflecting the kinetic limitations imposed by the presence of
CO. Based on our XPS microscopy findings [13], the hindered
oxide growth should result in a patchy structure consisting of
RuO2 islands and RuxOy areas, with the latter accounting for
the presence of the Rubulk component. A typical morphology
of such a surface is illustrated by the Ru 3d5/2 image in the O
yield with a single phase by running the reaction at a con-
stant temperature above 500 K, where RuO2 nucleation and
growth occur. However, the catalytic activity of the surface ox-
ide was confirmed by recent experiments performed using a
stoichiometric CO/O2 partial pressure ratio of 2 instead of 1,
maintaining the same total pressure of 0.1 mbar. We increased
the temperature from 400 to 600 K in 20–25 K steps, waiting at
each new temperature until the CO2 signal reached a constant
value. In this case the higher CO/O2 gas pressure ratio hin-
dered formation of the RuO2 state, as indicated by the evolution
of Ru 3d and O 1s spectra, which terminated at the formation
of surface oxide. The CO2 yield grew steadily, showing sta-
ble activity at constant temperature, until levelling off at around
500 K and declining slowly above 550 K. This result is in qual-
itative agreement with the results with powdered Ru catalyst,
which demonstrate that an ultra-thin oxide layer covering the
metallic Ru core is the active state in CO oxidation at tempera-
tures below 500 K [32].
3
5
1
s panel (Fig. 5a). The RuO2 islands appear dark, because the
RuO2 phase contains less Ru atoms per unit volume than the
adsorption and O-rich intermediate states (bright) and also, be-
ing thicker, it more effectively screens the emission from the
metallic Ru below. Similar coexistence of the two phases was
observed in very wide temperature range (600–775 K) and ex-
posure range [13,21]. Because the RuxOy component appears
close to the cus-Ru component of the RuO2(110) surface (see
Table 1), and it is speculative to fit the Ru 3d5/2 spectra using
three components with unknown weight, we allowed a broaden-
ing of the dominant oxide component at ∼0.6 eV to account for
all contributions. The coexistence of the both oxidation states is
confirmed by the corresponding O 1s spectrum, which contains
the surface oxide and RuO2 components. Judging from the rel-
ative weight of the surface oxide O 1s component, about 80%
of the surface should be covered with RuO2 islands at 590 K.
Note that the RuO2 bridge O component at 528.7 eV is very
weak, probably due to its continuous consumption during the
reaction, in accordance with the mechanism suggested in pre-
vious work [7,9]. The reaction rate can be reverted back and
forth by decreasing and increasing the temperature in the range
of 550–600 K, which does not visibly affect the catalyst surface
composition.
The most striking result is that the growth of the RuO2
phase above 500 K does not affect the monotonous increase
of the CO2 yield (Fig. 4a). This suggests that the nucleation
and growth of a stoichiometric oxide phase barely affects the
reaction barrier. The plot of the CO2 yield versus O content
in Fig. 4c provides the best illustration that the high catalytic
activity of the Ru catalyst is not strongly correlated with the for-
mation of RuO2 with a well-defined surface structure. It clearly
shows that the surface oxide formed via progressive incorpo-
ration of oxygen already exhibits high catalytic activity and
that there is no significant increase with the formation of sto-
ichiometric RuO2. Here it should be noted that because RuO2
formation occurs above 500 K, when comparing the catalytic
activity of the surface oxide and RuO2, the positive tempera-
ture effect on the reaction rate should be taken into account as
well.
4. Conclusion
The in situ capability of high-pressure photoelectron spec-
troscopy has provided real-time information about the evolution
of the chemical state of Ru(0001) catalyst during CO oxidation
with increasing reaction temperature. The results confirm that
the metallic state of Ru is inactive. The most important finding
is that the CO2 production is not phase-selective; that is, there
is no distinct difference in catalytic activity between the stoi-
chiometric RuO2(110) and a few layers thick, poorly ordered
surface oxide.
According to density functional theory predictions, the
higher catalytic activity of an oxide surface compared with the
corresponding metal surface not only results from the weaker
CO and O bonding on the oxide; an important and in some cases
even decisive role is the reorganisation required to achieve the
configuration of the transition state [16]. In the frame of this
concept, the activation role of the subsurface oxygen should be
ascribed to the induced deformation of the lattice, which af-
fects the O and CO adsorption configurations. The growth of
stoichiometric RuO2 does not substantially change the reaction
barrier. This result is not surprising, because, as noted above,
the activity is determined by the O and CO bonding configura-
tions on the catalyst surface. Apparently, the amorphous surface
oxide formed below 500 K has attained a favourable structure
with activity comparable to that exhibited by the well-defined
RuO2 phase formed at higher temperatures.
In CO excess, the limited amount of subsurface oxygen is
energetically driven to segregate to the surface, and the amor-
phous surface oxide can easily lose the incorporated oxygen and
convert into an inactive adsorption phase. However, the RuO2
also is unstable in CO excess and, similar to the surface oxide,
can be reduced into an inactive adsorption phase; however, re-
oxidising the reduced catalyst back to the initial stoichiometric
RuO2 is not necessary to regain catalytic activity.
The surface-phase diagram of the RuO2(110) surface, re-
ported previously [6], considers that the catalytically active re-
gion under realistic dynamic reaction conditions can often lie
The fact of the coexistence of surface oxide and RuO2 states
does not produce unambiguous results correlating the CO2

R. Blume et al. / Journal of Catalysis 239 (2006) 354–361
361
at the boundary between two phases. The present study pro-
vides experimental evidence that the highest CO oxidation rate
is monitored in the temperature range of 500–600 K, at which
two Ru oxidation states coexist.
[2] A. Böttcher, H. Niehus, J. Chem. Phys. 110 (1999) 3186, Phys. Rev. B 60
1999) 14396.
(
[3] H. Over, Y.D. Kim, A.P. Seitsonen, E. Lundgren, M. Schmid, P. Varga, A.
Morgante, G. Ertl, Science 287 (2000) 1474.
[4] H. Over, A.P. Seitsonen, E. Lundgren, M. Wiklund, J.N. Andersen, Chem.
Phys. Lett. 342 (2001) 467.
The results of this work apply well to real Ru catalyst sys-
tems, which are nanoparticles forming amorphous oxide with
poorly defined stoichiometry [33,34]. Such “oxidized” states of
the Ru nanoparticles, often described as RuxOy, are comparable
with surface oxide with subsurface oxygen, not with the well-
structured RuO2(110) surface. This study conforms that RuO2
formed at temperatures above 500 K is also active but repre-
sents a limiting case of a well-ordered model structure. The
long-range ordering of the RuO2(110) surface is not a prerequi-
site for catalytic function, but is instrumental in unravelling and
providing insight into the mechanisms of CO oxidation at the
atomic level. As noted above, a recent study of the CO oxida-
tion on polycrystalline powdered Ru catalyst in the temperature
range of 363–453 K showed that under dynamic catalytic con-
ditions, the active state is an ultra-thin Ru oxide film, whereas
fully oxidized RuO2 particles, formed at higher temperature,
exhibit lower activity [32]. This deactivation is tentatively at-
tributed to roughening and formation of inactive RuO2 facets.
In general, the long-range ordered oxide structures avail-
able in macroscopic systems cannot be the ones working under
conditions of high chemical potential and enabled structural
dynamics (real world). However, they are excellent model sys-
tems for fundamental experimental and theoretical studies of
catalytic reactions, helping to identify the general reactivity
trend on metallic and O-rich states of catalysts used in redox
processes.
[5] Z.-P. Liu, P. Hu, A. Alavi, J. Chem. Phys. 114 (2001) 5956.
[6] K. Reuter, M. Scheffler, Phys. Rev. Lett. 90 (2003) 46103, Phys. Rev. B 60
(2003) 45407.
[
[
[
7] J. Wang, V.Y. Fan, K. Jacobi, G. Ertl, J. Phys. Chem. B 106 (2002) 3422.
8] S.H. Kim, J. Wintterlin, J. Chem. Phys. B 108 (2004) 14565.
9] K. Reuter, D. Frenken, M. Scheffler, Phys. Rev. Lett. 93 (2004) 116105.
[10] H. Over, M. Muhler, Prog. Surf. Sci. 72 (2003) 3, and references therein.
11] H. Over, M. Knapp, E. Lundgren, A.P. Seitsonen, M. Schmid, P. Varga,
[
Chem. Phys. Chem. 5 (2004) 167, and references therein.
[
[
12] Y.D. Kim, H. Over, G. Krabbes, G. Ertl, Top. Catal. 14 (2001) 95.
13] A. Böttcher, U. Starke, H. Conrad, R. Blume, L. Gregoriatti, B. Kaulich,
A. Barinov, M. Kiskinova, J. Chem. Phys. 117 (2002) 8104.
[14] R. Blume, H. Niehus, H. Conrad, A. Böttcher, J. Phys. Chem. 108 (2004)
4332.
1
[
[
[
15] A. Böttcher, B. Krenzer, H. Conrad, H. Niehus, Surf. Sci. 504 (2002) 42.
16] X.-Q. Gong, Z.-P. Liu, R. Raval, P. Hu, J. Amer. Chem. Soc. 126 (2004) 8.
17] One monolayer equals the number of Ru atoms on the (0001) surface. We
use in further as an unit to express the total O load
[18] M. Todorova, W.X. Li, M.V. Ganduglia-Pirovano, C. Stampfl, K. Reuter,
M. Scheffler, Phys. Rev. Lett. 89 (2002) 96103.
19] K. Reuter, C. Stampfl, M.V. Ganduglia-Pirovano, M. Scheffler, Chem.
Phys. Lett. 352 (2002) 311.
20] P. Quinn, D. Brown, D.P. Woodruff, T.C.Q. Noakes, P. Bailey, Surf.
Sci. 491 (2001) 208.
[21] R. Blume, H. Niehus, H. Conrad, A. Böttcher, L. Aballe, L. Gregoriatti,
A. Barinov, M. Kiskinova, J. Phys. Chem. B 109 (2005) 14052.
22] H. Over, A.P. Seitsonen, Science 297 (2002) 2003.
23] H. Bluhm, M. Hävecker, A. Knop-Gericke, E. Kleimenov, R. Schlögl, D.
Teschner, V.I. Bukhtiyarov, D.F. Ogletree, M. Salmeron, J. Phys. Chem.
B 108 (2004) 14340.
[
[
[
[
[24] D.F. Ogletree, H. Bluhm, C. Lebedev, C.S. Fadley, Z. Hussain, M.
Salmeron, Rev. Sci. Instrum. 73 (2002) 3872.
Acknowledgments
[
25] S. Lizzit, A. Baraldi, A. Groso, K. Reuter, M.V. Ganduglia-Pirovano, C.
Stampfl, M. Scheffler, M. Stichler, C. Keller, W. Würth, D. Menzel, Phys.
Rev. B 63 (2001) 205419.
The authors are indebted to Dr. W. Ranke for providing
a critical reading of the manuscript and engaging in illumi-
nating discussions. M. Kiskinova thanks the AvH Foundation
for an award to pursue research in FHI-Berlin in 2004–2005.
P. Dudin acknowledges financial support under contract NMP3-
CT-2003-505670 (NANO2). The BESSY staff is acknowledged
for their continuous support in performing the present measure-
ments.
[26] M.P. Seah, Surf. Interface Anal. 9 (1986) 85.
[27] H. Over, A.P. Seitsonen, E. Lundgren, M. Smedh, J.N. Anderesen, Surf.
Sci. 504 (2002) L196.
[28] A. Schiffer, P. Jacob, D. Menzel, Surf. Sci. 389 (1997) 116.
[29] W.X. Li, C. Stampfl, M. Scheffler, Phys. Rev. B 67 (2003) 45408.
[30] C.H. F Peden, D.W. Goodman, J. Phys. Chem. 90 (1986) 1360.
[31] M. Mavrikakis, B. Hammer, J.K. Norskov, Phys. Rev. Lett. 81 (1998)
2819.
[
32] V. Narkhede, J. Assmann, M. Muhler, Z. Phys. Chem. 219 (2005) 979.
References
[33] W. Vogel, N. Alonso-Vante, J. Catal. 232 (2005) 395.
[
34] B.-Z. Zhan, M.A. White, T.-K. Sham, J.A. Pincock, R.J. Doucet, K.V.
Ramana Rao, K.N. Robertson, T.S. Cameron, J. Amer. Chem. Soc. 125
(2003) 2195.
[1] A. Böttcher, H. Niehus, S. Schwegmann, H. Over, G. Ertl, J. Phys. Chem.
B 101 (1997) 11185.