Structure-activity correlations for the oxidation of CO over polycrystalline RuO2 powder derived from steady-state and transient kinetic experiments

Article abstract of DOI:10.1524/zpch.219.7.979.67092

The oxidation of carbon monoxide was studied at atmospheric pressure in a plug-flow reactor over polycrystalline ruthenium dioxide powder in the temperature range from 363 to 453 K as a function of the pretreatment. Calcining RuO₂ in flowing oxygen resulted in purified bulk RuO₂, whereas reduction in hydrogen led to bulk Ru metal, which was partially oxidized again in flowing oxygen at increasing temperatures (T_ox) up to 573 K to obtain RuO₂/Ru shell-core particles with increasing RuO ₂ shell thickness. Using the TPR technique subsequent to steady-state CO oxidation to monitor the degree of oxidation, the most active and stable state of the unsupported ruthenium catalysts was identified as an ultra-thin RuO₂ layer covering a metallic Ru core in agreement with the shell-core model established for supported Ru catalysts. Steady-state turnover frequencies (TOFs) obtained with the ultra-thin RuO₂ films are in good agreement with TOFs reported for studies on Ru single crystal surfaces and with supported Ru catalysts. Only for RuO₂ films thicker than 1 nm (T_ox ≥ 473 K) and for fully oxidized RuO₂ deactivation was observed, presumably due to the formation of inactive RuO₂ surfaces such as the RuO₂(100)-c(2 × 2) facet. Moreover, it was demonstrated that the presence of moisture in the reactant feed inhibits the oxidation of CO completely. by Oldenbourg Wissenschaftsverlag, Muenchen.

Full text of DOI:10.1524/zpch.219.7.979.67092

Z. Phys. Chem. 219 (2005) 979–995
by Oldenbourg Wissenschaftsverlag, München
Structure-Activity Correlations for the
Oxidation of CO over Polycrystalline RuO2
Powder Derived from Steady-State and
Transient Kinetic Experiments
∗
By Vijay Narkhede, Jens Aßmann, and Martin Muhler
Lehrstuhl für Technische Chemie, Ruhr-Universität Bochum, D-44780 Bochum,
Germany
Dedicated to Prof. Dr. Helmut Knözinger on the occasion
of his 70th birthday
(Received April 26, 2005; accepted April 28, 2005)
Ruthenium Dioxide (RuO2) / CO Oxidation / Kinetics /
Temperature-Programmed Reduction (TPR) / Shell-Core Model
The oxidation of carbon monoxide was studied at atmospheric pressure in a plug-flow
reactor over polycrystalline ruthenium dioxide powder in the temperature range from 363
to 453 K as a function of the pretreatment. Calcining RuO2 in flowing oxygen resulted
in purified bulk RuO2, whereas reduction in hydrogen led to bulk Ru metal, which was
partially oxidized again in flowing oxygen at increasing temperatures (Tox) up to 573 K to
obtain RuO2/Ru shell-core particles with increasing RuO2 shell thickness. Using the TPR
technique subsequent to steady-state CO oxidation to monitor the degree of oxidation,
the most active and stable state of the unsupported ruthenium catalysts was identified as
an ultra-thin RuO2 layer covering a metallic Ru core in agreement with the shell-core
model established for supported Ru catalysts. Steady-state turnover frequencies (TOFs)
obtained with the ultra-thin RuO2 films are in good agreement with TOFs reported
for studies on Ru single crystal surfaces and with supported Ru catalysts. Only for
RuO2 films thicker than 1 nm (Tox ≥ 473 K) and for fully oxidized RuO2 deactivation
was observed, presumably due to the formation of inactive RuO2 surfaces such as the
RuO2(100)-c(2×2) facet. Moreover, it was demonstrated that the presence of moisture in
the reactant feed inhibits the oxidation of CO completely.
1. Introduction
In the past two decades remarkable progress has been made in heterogeneous
catalysis towards bridging the pressure and material gaps between porous cat-
*
Corresponding author. E-mail: muhler@techem.rub.de

980
V. Narkhede et al.
alysts operated at high pressures in plug-flow reactors and highly ordered
two-dimensional model systems studied in ultra-high vacuum (UHV). Oxidic
surfaces are more demanding for bridging these gaps than metallic surfaces,
since the activity of the oxide catalysts is frequently controlled by the pres-
ence of low coordinated defect sites. Recently, ruthenium dioxide (RuO2) has
received considerable attention due to its remarkable high activity towards CO
and methanol oxidation [1–6].
Ruthenium metal crystallizes in the hexagonal close-packed structure pref-
¯
erentially exposing (0001) and (1010) surfaces. In contrast to transition metals
such as platinum, palladium and rhodium, metallic ruthenium is a poor cat-
alyst for the oxidation of CO under UHV conditions [13]. Thin RuO2 films
with rutile structure can be produced from ruthenium metal by exposing it to
molecular oxygen at higher temperatures. Over et al. [6, 7] showed that thin
RuO2(110) and (100) surfaces can be produced at temperatures above 650 K
¯
by excessive oxygen exposure of well-defined Ru(0001) and (1010) surfaces,
respectively. Unlike the (100) surface where all rows are characterized by five-
fold coordinated Ru cations, the (110) surface of RuO2 possesses alternating
rows of fivefold and sixfold coordinated Ru cations. Both RuO2(110) and (100)
surfaces expose coordinatively unsaturated (cus) sites, which are mandatory for
high catalytic activity. It was also reported that both surfaces are similarly ac-
tive in the oxidation of CO [6]. Recently, the phase diagram of the RuO2(110)
surface as a function of temperature and the partial pressures of O2 and CO was
derived using density functional theory [8, 9].
Deactivation is an important issue for the oxidation of CO both under
UHV [6] and high-pressure conditions [10]. In a single crystal study under ox-
idizing conditions and at high temperatures (above 800 K), the coexistence of
the (1×1) and c(2×2) phases was observed on the RuO2(100) surface [6].
The c(2×2) structure was also observed on the RuO2(110) surface in case of
severe roughening under oxidizing conditions at temperatures above 800 K. It
was found that the RuO2(100)-c(2×2) surface structure is completely inactive
towards CO oxidation, since CO cannot adsorb on this surface [6]. Another ori-
gin of deactivation was revealed by Paulus et al. [11] who observed that water
strongly inhibits the oxidation of CO over the RuO2(110) surface under UHV
conditions by blocking the Ru cus sites. Thus, the dissociative adsorption of
oxygen is hindered, and the catalytic cycle is reversibly poisoned. In contrast,
Zhang and Kisch [12] reported earlier that hydrated RuO2 is a powerful, sus-
tainable and efficient catalyst for room temperature oxidation of CO by humid
air, although typical transition metal oxides do not tolerate humidity.
Recently, the high catalytic activity of supported Ru catalysts for the ox-
idation of CO was elucidated [10, 14]. It was shown that a thin RuO2 shell
covering a metallic Ru core is the most active state of supported Ru nanopar-
ticles. By comparing the kinetic behavior of the supported RuO2 particles both
under atmospheric pressure as well as under high-vacuum conditions in the
TAP reactor (temporal analysis of products) with the RuO2(110) model cata-

Structure-Activity Correlations for the Oxidation of CO . . .
981
lyst at 10−7 mbar, the absence of a pressure gap was confirmed. However, the
materials gap has not yet been bridged completely. Polycrystalline RuO2 pow-
der represents a natural link between the two-dimensional model catalysts and
the supported catalysts to investigate the role of the Ru particle size in the
oxidation of CO.
In the present study, the oxidation of CO over ruthenium powder at ambi-
ent pressure and at temperatures between 363 and 453 K was investigated as
a function of the redox pretreatment. Calcining RuO2 in flowing O2 resulted
in purified bulk RuO2, whereas reduction in H2 led to bulk Ru metal, which
was partially oxidized again in flowing O2 at increasing temperatures (Tox) up
to 573 K to obtain RuO2/Ru shell-core particles with increasing RuO2 shell
thickness. In addition, the catalytic performance of the catalyst was investi-
gated in the presence of water vapor in the reactant feed. To the best of our
knowledge, turnover frequencies (TOFs) for unsupported Ru catalysts were de-
rived for the first time and compared with data reported for the oxidation of CO
at atmospheric pressure and high vacuum conditions over supported Ru cata-
lysts and over single crystal surfaces. Furthermore, to monitor the thickness of
the RuO2 shell, temperature-programmed reduction (TPR) measurements were
performed using H2 and CO as reducing agents. By combining these steady-
state and transient kinetic experiments, the active state of Ru-based catalysts is
clearly identified as an ultra-thin RuO2 layer supported on a metallic Ru core,
whereas RuO2 films thicker than 1 nm deactivate strongly.
2. Experimental
2.1 Characterization
Polycrystalline RuO2 (purity > 99.9%, supplied by Merck LTD) was used in
this study. Both oxidized and reduced RuO2 samples were characterized by
static physisorption using an Autosorb 1-C setup (Quantachrome). Because of
the small specific surface area, Krypton gas was used for the BET measure-
ments. In addition, H2 chemisorption measurements were carried out with the
RuO2 sample reduced at 773 K in a continuous hydrogen flow. X-ray powder
diffraction patterns of the polycrystalline RuO2 powder in the pristine state and
after hydrogen reduction were recorded with a Siemens D500 diffractometer
using Cu Kα radiation in the range from 5◦ to 80◦ 2θ. Scanning electron micro-
scope (SEM) images of both pristine and reduced RuO2 samples were obtained
with a Leo-1530 scanning electron microscope.
2.2 Catalytic activity measurements
The oxidation of CO was carried out at ambient pressure and in a tempera-
ture range from 363 to 453 K using a plug-flow quartz reactor filled with
12.0–14.0 mg polycrystalline RuO2 powder diluted in 200 mg silica (sieve

982
V. Narkhede et al.
fraction 255–355 µm). The temperature was measured directly below the fixed
bed using a thermocouple placed in a hollow passage of the reactor. The
gases used in this study were supplied by Messer Griesheim: Argon pu-
rity > 99.999%, Ne purity > 99.99% in the mixtures with 10% CO (purity
> 99.997%) and 10% O2 (purity > 99.995%), 5.2% H2 in Ar with H2 purity
> 99.9% and Ar purity > 99.999%. The flow rates of the reactants were ad-
justed by three mass flow controllers. In order to carry out measurements in the
presence of moisture, the reactant feed consisting of CO/Ne, O2/Ne, and Ar
was passed through a saturator filled with deionized water.
Prior to the oxidation of CO, the catalyst was pretreated in three different
ways. At first, the pristine RuO2 powder was oxidized from room temperature
up to 573 K in the O2/Ne mixture with a flow rate of 50 Nml/min. In the sec-
ond pretreatment, the RuO2 catalyst was reduced from room temperature up
to 773 K in the H2/Ar mixture with a flow rate of 50 Nml min−1. This treat-
ment is known to yield oxygen-free metallic Ru particles. The third method
consisted of a reoxidation of the reduced catalyst at 323, 373, 473, or 573 K
using the O2/Ne gas mixture. After the pretreatment each catalyst was purged
in pure Ar for 15 min, and then the starting temperature of CO oxidation was
adjusted. The CO/O2 reactant feed ratios were varied in the range of 0.5–3.0.
The concentration of CO (1.8%) for all the ratios was kept constant, and the
concentration of oxygen was varied according to the ratios. The concentrations
of the reactants and products (CO, CO2 and O2) were monitored using non-
dispersive IR analyzers (Uras 14 and Magnos) supplied by Hartmann & Braun.
The conversion of CO was calculated as follows:
in
CO
out
CO
C
−C
conversion of CO (%) =
×100
(1)
in
CO
C
in
CO
out
CO
where C is the inlet and C is the outlet concentration of CO.
The calculation of the yield of CO2 was based on the inlet concentration
of CO:
out
CO
in
CO
C
−C
2
2
yield of CO2 (%) =
×100 .
(2)
in
CO
C
2.3 Temperature-programmed reduction (TPR)
The temperature-programmed reduction (TPR) measurements were conducted
in the same quartz reactor using a heating rate of 10 K min−1 and a flow rate of
20 Nml min−1 of the 5.2% H2/Ar mixture. Prior to the H2 TPR measurements,
the oxidation of CO over the oxidized, reduced and reoxidized ruthenium cat-
alysts was carried out using various CO/O2 reactant feed ratios. Subsequently,
the catalyst was cooled down in Ar to room temperature. Thereafter, the meas-
urement was started by flushing the catalyst in the H2/Ar gas mixture for

Structure-Activity Correlations for the Oxidation of CO . . .
983
30 min at room temperature, followed by heating the reactor to 773 K (heat-
ing ramp 10 K min−1). The final temperature was held constant for 20 min. The
consumption of H2 was detected by a calibrated thermal conductivity detector
supplied by Rosemount.
The TPR experiments with CO as reducing agent were also conducted
in the same experimental set-up with a heating ramp of 10 K min−1. Using
12.5 mg pristine RuO2 powder, the experiment was carried out with the CO/Ne
gas mixture (CO concentration of 9.86%). The consumption of CO and the
formation of CO2 were monitored by the IR analyzers described in Sect. 2.2.
To avoid the generation of exceedingly large H2 concentration gradients
across the catalyst bed, the reduction was conducted under differential con-
ditions as suggested by Monti and Baiker [16]. To obtain optimum reduction
profiles, they proposed that the characteristic number K should be between 55
and 140 s for heating rates of 0.1 ≤ β ≤ 0.3 K s−1. The K value is defined as
n0
K =
(4)
F ×c0
where n0 is the initial amount of reducible species, F is the flow rate of the
reducing gas mixture and c0 is the concentration of the reducing agent in the
carrier gas. The K values used for these measurements were 122 s and 65 s for
the H2 and CO TPR profiles, respectively. The degree of oxidation was quan-
tified based on the amount of consumed H2 or CO and the following reduction
stoichiometries:
RuO2 +2H2 → Ru+2H2O
RuO2 +2CO → Ru +2CO2 .
(5)
(6)
3. Results
3.1 Characterization
The sharp XRD peaks of the polycrystalline RuO2 powder (Fig. 1a) indicate
a high degree of crystallinity. The diffraction pattern also shows that no other
crystallographic phase than RuO2 (PDF-2 database [40-1290]) is present in
the powder. The XRD pattern obtained after the reduction of RuO2 in flowing
hydrogen at 773 K is shown in Fig. 1b. This XRD pattern clearly shows that
RuO2 was fully reduced to metallic Ru (PDF-2 database [6-663]). The XRD
peaks of metallic Ru are obviously slightly broader than those of pristine RuO2.
Thus, the Scherrer equation was used to estimate the crystallite size from the
widths of the XRD reflections resulting in mean values of about 0.2–0.3 µm
and 0.1–0.2 µm for the RuO2 and the Ru metal powder, respectively.
The BET measurements provided specific surface areas of 1.3 and
1.8 m2 g−1 for the oxidized RuO2 sample and for metallic Ru powder ob-
tained by reduction of RuO2, respectively. H2 chemisorption measurements

984
V. Narkhede et al.
Fig. 1. X-ray diffraction patterns of (A) polycrystalline RuO2 powder and (B) metallic Ru
powder obtained by reducing RuO2 in flowing H2 at 773 K.
were carried out with Ru metal which revealed a specific Ru surface area of
0.7 m2 g−1 in reasonable agreement with the BET area. A surface stoichiome-
try of Hads/Rusurf = 1 and an average area of 8.17×10−20 m2 per Ru atom were
assumed to derive the Ru metal area [14]. Based on the BET measurements,
particle sizes for RuO2 amounting to 0.7 µm and for metallic Ru amounting
to 0.3 µm can be estimated assuming spherical particles and the corresponding
bulk densities in good agreement with the XRD results.
The SEM image shown in Fig. 2a reveals that RuO2 powder consists of
crystallites of different sizes which expose regularly shaped facets. The small-
est detectable crystallite size amounts to about 0.15–0.25 µm, and these crys-
tallites adhere to each other forming chain-like structures. There are larger
agglomerations of these crystallites in the range from 0.8 µm to few microns.
In addition, large crystals with sharp columnar growth are seen. Fig. 2b shows
a SEM image of Ru metal powder obtained by reducing RuO2 in flowing hy-
drogen at 773 K. In contrast to the well-defined morphology of the pristine
RuO2 crystallites, the rounded Ru metal particles are substantially less regu-
larly shaped exposing rough and ill-defined surfaces.
3.2 CO oxidation
3.2.1 CO oxidation over preoxidized RuO2
The result of the catalytic activity measurements using a reactant feed ratio of
2.0 over preoxidized polycrystalline RuO2 powder calcined at 573 K is shown

Structure-Activity Correlations for the Oxidation of CO . . .
985
Fig. 2. Scanning electron microscopy images of a) pristine polycrystalline RuO2 powder;
b) metallic Ru powder obtained by reducing RuO2 in flowing H2 at 773 K.
in Fig. 3a. An initial reaction temperature of 363 K was used, which was then
stepwise (30 K each, holding time 30 min) raised up to 453 K followed by
stepwise decreasing the temperature to the initial temperature, continued by
a second cycle. This temperature program was applied in all the CO oxidation
measurements. Every increase in temperature resulted in an increase in the CO
conversion, which, however, was followed by a decay during the isothermal
period at each step of the first temperature cycle. In contrast to the deteriora-
tion of the catalytic activity during the first cycle, no decrease in conversion
was observable during the second cycle, implying that the deactivation of the
catalyst is restricted to the first cycle. The degrees of CO conversion observed
in the second cycle were substantially lower compared to those of the first
cycle.

986
V. Narkhede et al.
Fig. 3. Conversion of CO over 12.5 mg of a) RuO2 oxidized at 573 K, b) metallic Ru
obtained by H2 reduction of RuO2 at 773 K, each diluted with 200 mg of quartz, as a func-
tion of time and temperature. The temperature program is displayed by the gray trace.
The total flow rate was 50 Nml/min with a CO/O2 reactant feed ratio of 2.0 (1.8% CO/
0.9% O2).
The performance of the preoxidized RuO2 catalyst was also monitored
using an excess of CO (ratio CO/O2 equal to 3.0) in the reactant feed. The
resulting degrees of conversion are shown in Fig. 4a. Obviously, the deactiva-
tion phenomena are similar to those observed with the stoichiometric ratio (cf.
Fig. 3a). During every step of the first temperature cycle, the conversion of CO
was found to decrease, resulting in a stable, but considerable less active state
during the second temperature cycle. The highest theoretical conversion of CO
for the reactant feed ratio of 3.0 is 66%, whereas the maximum conversion in
this experiment was observed to be just 45%.
Finally, to test the performance of the preoxidized catalyst in an oxygen-
rich feed, catalytic measurements were performed using a CO/O2 reactant feed
ratio of 0.5. The results (not shown) reproduce again the already made obser-
vations, i.e., a decreasing degree of conversion of CO during each step of the
temperature program, again limited to the first cycle. Moreover, the degree of
deactivation in the net oxidizing feed was much higher as compared to the other
reactant feed ratios. The maximum conversion of CO was found to be 32%,
which is much lower than 45% observed for the net reducing feed (Fig. 4a) and
65% for the stoichiometric feed (Fig. 3a).
3.2.2 CO oxidation over metallic Ru obtained by reducing RuO2
To measure the catalytic activity for the oxidation of CO over polycrystalline
Ru metal, the RuO2 powder was reduced in the H2/Ar gas mixture at 773 K

Structure-Activity Correlations for the Oxidation of CO . . .
987
Fig. 4. Conversion of CO over 12.5 mg of a) RuO2 oxidized at 573 K, b) metallic Ru
obtained by H2 reduction of RuO2 at 773 K, each diluted with 200 mg of quartz, as a func-
tion of time and temperature. The temperature program is displayed by the gray trace.
The total flow rate was 50 Nml/min with a CO/O2 reactant feed ratio of 3.0 (1.8% CO/
0.6% O2).
for 1 h. The catalytic activity measurements were conducted with three differ-
ent reactant ratios using the same temperature program which was applied for
the measurements with the preoxidized catalysts. Fig. 3b shows the conversion
of CO for a CO/O2 reactant feed ratio of 2.0 as a function of time and tem-
perature. Every increase in temperature resulted in an increase in conversion.
However, an inspection of the time dependence shows that the catalytic activity
is stable at constant temperatures. As a consequence, the degrees of conversion
are essentially the same in both temperature cycles. The highest conversion of
CO was found to be about 55%.
Another catalytic activity measurement over metallic Ru powder was car-
ried out using a CO/O2 ratio of 3.0. The results are shown in Fig. 4b. The
maximum conversion of CO was found to be about 60%, which is close to the
maximum theoretical conversion of 66%. It is interesting to note that the con-
version of CO at each step of the temperature program was found to be stable,
in contrast to the preoxidized catalyst.
It was pointed out in our earlier investigations [10] that the extent of deacti-
vation over prereduced supported Ru nanoparticles is a function of temperature
and of the CO/O2 ratio, with the highest degree of deactivation for a reactant
feed ratio of 0.5. However, monitoring the oxidation of CO over the metallic
Ru sample in an excess of oxygen (CO/O2 ratio equal to 0.5, not shown), no

988
V. Narkhede et al.
Fig. 5. Conversion of CO as function of time and temperature over 14 mg RuO2/Ru
shell-core catalysts obtained by reoxidation of metallic Ru samples at (A) 232, (B) 373,
(C) 473, and (D) 573 K. The total flow rate was 50 Nml/min with a CO/O2 reactant feed
ratio of 0.5 (1.8% CO/3.6% O2).
loss in activity was observed, and the maximum conversion of CO was found
to amount to 55%.
3.2.3 CO oxidation over partially reoxidized metallic Ru
In order to study the influence of the thickness of the RuO2 layer, the oxidation
of CO in an excess of O2 (CO/O2 ratio equal to 0.5) was studied over RuO2/Ru
shell-core catalysts which were obtained by reoxidation of metallic Ru cat-
alysts at increasing temperatures. The results shown in Fig. 5 demonstrate
a strong dependence of the catalytic activity on the reoxidation temperature: the
higher the calcination temperature, the lower is the resulting catalytic activity.
The metallic Ru catalysts reoxidized at 323 and 373 K displayed an increasing
activity during time on stream resulting in higher and rather stable degrees of
conversion during the second temperature cycle. For the metallic Ru catalysts
reoxidized at 473 and 573 K, very low degrees of conversion were observed
indicating a strong negative influence of high oxidation temperatures on the
catalytic performance.
3.2.4 Influence of water
As it was reported that the hydrous ruthenium dioxide is a powerful catalyst
for room temperature oxidation of CO by humid air [12], it was interesting to
monitor the catalytic activity of RuO2 powder in the presence of water in the re-

Structure-Activity Correlations for the Oxidation of CO . . .
989
actant feed stream during the oxidation of CO. During these measurements the
reactant feed was passed through a water bubbler operated at room temperature
to obtain a water concentration amounting to 2.4%. The reaction was carried
out using a stoichiometric CO/O2 ratio and the same temperature program up
to 453 K as used in the previous experiments. In the whole temperature range,
no conversion of CO was observed (not shown). This result implies that wa-
ter strongly inhibits the CO oxidation over RuO2 in agreement with the results
reported in Ref. [11].
3.2.5 Turnover frequencies
To allow a detailed comparison of the activity of the unsupported RuO2 pow-
der with data obtained in single crystal studies [4, 13] as well as for supported
Ru nanoparticles [10, 15], turnover frequencies (TOFs) were calculated as fol-
lows:
number of molecules of CO2 formed
TOF =
.
(3)
(surface area)×(time)
TOFs obtained for the oxidation of CO in an oxygen-rich stream both over the
oxidized and reduced ruthenium catalysts are plotted in the Arrhenius diagram
shown in Fig. 6. The comparison between the differently pretreated samples
demonstrates the higher catalytic activity of metallic Ru obtained by reducing
RuO2. It is interesting to note that the calculated TOFs for the prereduced RuO2
sample correlate well with TOFs observed for the oxidation of CO over the
Ru(0001) single crystal which was covered by a thin RuO2(110) layer [4]. On
the other hand, TOFs obtained in earlier studies for supported and fully oxi-
dized RuO2 nanoparticles [15] are in good agreement with those derived for
RuO2 powder. Obviously, the catalysts with fully oxidized RuO2 are in a less
active state with a somewhat higher apparent activation energy.
3.3 Temperature-programmed reduction (TPR)
TPR experiments were conducted from room temperature up to 773 K to in-
vestigate the reducibility of polycrystalline RuO2 using H2 and CO as reducing
agents, and to derive the degree of oxidation of the ruthenium catalysts after
reaction as a function of the pretreatment. The H2 TPR profile obtained sub-
sequent to CO oxidation in an oxygen-rich reactant feed with the preoxidized
RuO2 catalysts is shown in Fig. 7a consisting of a single peak with the max-
imum at 451 K. Based on the consumption of H2 an O/Ru ratio of 2.0 was
derived for the polycrystalline RuO2 catalyst implying that the degree of oxida-
tion of the catalyst was not changed by the CO oxidation reaction. In contrast,
no consumption of H2 was observed during the TPR measurements carried
out subsequent to the oxidation of CO (CO/O2 ratio equal to 0.5; T = 456 K)
over the metallic Ru sample, i.e., the reduction of the ultra-thin RuO2 layer

990
V. Narkhede et al.
Fig. 6. Arrhenius plot of turnover frequencies (CO2 molecules formed per m2 and s) ob-
served for the oxidation of CO over oxidized RuO2 and metallic Ru obtained by H2 reduc-
tion of RuO2 at 773 K. TOFs obtained for Ru(0001) under high pressure conditions [13],
for RuO2(110)/Ru(0001) [4] and for MgO- and SiO2-supported RuO2 nanoparticles [15]
are also included.
Fig. 7. TPR profiles A) obtained after the oxidation of CO was carried out over oxidized
RuO2 (12.0 mg) at 456 K in an oxygen-enriched stream (CO/O2 reactant feed ratio of
0.5), B) obtained for pristine RuO2 (12.0 mg) using CO as reducing agent.

Structure-Activity Correlations for the Oxidation of CO . . .
991
Fig. 8. TPR profiles obtained after the oxidation of CO was carried out at 456 K in an
oxygen-enriched stream (CO/O2 reactant feed ratio of 0.5) over metallic Ru samples re-
oxidized at (A) 473 K, and (B) 573 K.
formed on the metallic Ru core is below the detection limit of the TPR
experiment.
For the metallic Ru samples reoxidized at higher temperatures, it was pos-
sible to detect the consumption of H2 as shown in Fig. 8. TPR peak maxima
are observed at 353 K for the sample reoxidized at 473 K (Fig. 8, trace A)
and at 390 K for the sample reoxidized at 573 K (Fig. 8, trace B). A quan-
titative analysis yielded rather low degrees of oxidation below 2%. This ob-
servation is fairly consistent with the data reported by Prudenziati et al. [17]
who first reduced polycrystalline RuO2 powder and reoxidized it subsequently
in air at 973 K for several hours without achieving full reoxidation. How-
ever, it was possible to obtain fully oxidized RuO2 at temperatures above
1073 K within 90 min [17]. Assuming spherical Ru particles, and using the
bulk densities and a mean diameter of roughly 0.5 µm based on the BET and
static H2 chemisorption measurements, it is possible to estimate the RuO2
layer thickness. For the reoxidation at 473 K, it amounts to about 1 nm,
whereas the reoxidation at 573 K resulted in a thickness of somewhat more
than 3 nm.
TPR measurements were also carried out using CO as reducing agent. The
corresponding TPR profile obtained for the pristine RuO2 powder is shown in
Fig. 7b. It consists of one signal with the maximum at 564 K. An O/Ru ratio
of 2.0 was derived based on the consumption of CO. Comparing the CO TPR

992
V. Narkhede et al.
profile (Fig. 7b) with the H2 TPR profile (Fig. 7a) obtained for the preoxidized
RuO2 catalyst, it is seen that the single TPR peak is shifted to much higher tem-
peratures using CO as reducing agent. Thus, it is reasonable to assume that bulk
reduction of the RuO2 catalyst may occur during CO oxidation in an excess of
CO and at temperatures above 550 K.
4. Discussion
The catalytic activity measurements clearly revealed a strong influence of the
pretreatment procedure on the catalytic activity. In case of the preoxidized
RuO2 samples, strong deactivation occurred during the first temperature cycle,
and the degree of CO conversion was stable, but much lower during the second
cycle. Furthermore, a decreasing CO/O2 reactant feed ratio enhanced the de-
activation during the first temperature cycle. Similar phenomena were reported
for fully oxidized supported RuO2 nanoparticles [10], which were attributed to
the formation of inactive RuO2 facets at higher partial pressures of O2 and tem-
peratures. This explanation is obviously also valid for the preoxidized RuO2
powder, which remains fully oxidized during the reaction as demonstrated by
the subsequent TPR measurements. In contrast, no deactivation was observed
during the oxidation of CO over metallic Ru for all CO/O2 reactant feed ratios.
Thus, the conversion of CO during the second cycle remained as high as dur-
ing the first cycle and was much higher in comparison to fully oxidized RuO2.
Similar results were obtained for the metallic Ru samples reoxidized at 323
and 373 K, i.e. no deactivation occurred and the conversion of CO was as high
as over fully reduced metallic Ru without reoxidation pretreatment. However,
higher reoxidation temperatures were found to decrease the catalytic activity
drastically.
The results of the TPR measurements and the derived TOFs provide more
insight into the active state of the differently pretreated ruthenium catalysts.
On the one hand, the quantitative analysis of the TPR profiles proves that
the degree of oxidation of the preoxidized RuO2 sample remains unchanged
during the oxidation of CO. On the other hand, the corresponding TOFs can
be extrapolated to data obtained with supported RuO2 nanoparticles [15].
This agreement indicates that fully oxidized RuO2 particles are less active
for the oxidation of CO. Studies with RuO2/Ru(0001) single crystals [6, 18]
showed that an increase of the thickness of the very active RuO2 layer to
more than 1 nm leads to roughening and restructuring. Finally, thermody-
namically more stable and catalytically inactive RuO2(100)-c(2×2) facets are
formed, which are assumed to be responsible for the deactivation of fully
oxidized RuO2 particles. The restructuring of the RuO2(110) and (100) sur-
faces during CO oxidation is another example for catalytic etching [19] de-
scribed for instance in detail for silver surfaces under methanol oxidation
conditions [20].

Structure-Activity Correlations for the Oxidation of CO . . .
993
For the stable and much more active metallic Ru samples, an ultra-thin
RuO2 layer is assumed to be present as active phase based on the TPR results
and the calculated TOFs: first, the consumption of H2 was below the detec-
tion limit during the TPR measurement carried out subsequent to CO oxidation.
Second, the TOFs are in good agreement with data obtained in Ru(0001) single
crystal studies where it was shown that an ultra-thin RuO2 layer is responsible
for the high catalytic activity [1, 4].
Thus, it is possible to bridge the materials gap between Ru single crystals
and supported Ru nanoparticles by polycrystalline RuO2. However, one major
difference remains: the absence of any deactivation for metallic Ru particles
obtained by reducing polycrystalline RuO2 powder, whereas it was not possible
to retain the high initial catalytic activity of the reduced Ru nanoparticles sup-
ported on MgO or SiO2 [10]. Considering the dimensions of the particles and
the specific surface areas, the high stability of unsupported bulk Ru can be re-
lated to the particle size in the µm range which prevents bulk oxidation under
reaction conditions. In case of the supported Ru nanoparticles, the prereduc-
tion was not successful to prevent deactivation, because the particles were fully
oxidized in the oxygen-rich reactant feed gas already at moderate tempera-
tures [10]. This conclusion is supported by the TPR measurements carried out
with the metallic Ru samples reoxidized at 323 and 373 K. The results docu-
ment that temperatures ≤ 373 K are not sufficient to reoxidize large metallic
Ru particles significantly, i.e., the RuO2 layer is still thinner than about 1 nm.
Consequently, the conversion of CO is not affected again supporting the iden-
tification of the ultra-thin RuO2 layer as active phase. Only higher calcination
temperatures ≥ 473 K reduce the catalytic activity strongly due to the increas-
ing degree of oxidation with increasing calcination temperatures [17]. The TPR
experiments allowed to estimate the resulting RuO2 film thicknesses to be more
than 1 nm resulting in structural deactivation [6, 18].
In addition to the dynamic restructuring of the RuO2 surface, which de-
pends on the degree of oxidation and therefore on the pretreatment, on the
reaction conditions and the particle size of the Ru catalyst, blocking of active
sites remains another important point for deactivation. In our study, the add-
ition of water to the reactant feed for the catalytic CO oxidation prevented any
conversion of CO over a wide temperature range up to 453 K. These results
are in contradiction to reports that hydrous RuO2 is quite active for room tem-
perature CO oxidation [12]. However, a recent UHV study of the RuO2(110)
single crystal surface showed that dosing CO and O2 at room temperature re-
peatedly resulted in accumulating unreactive features observed by STM [11].
Using techniques such as HREELS and TDS, it was shown that the coordina-
tively unsaturated Ru sites are blocked by strongly adsorbed H2O molecules.
Therefore, the dissociation of O2 was inhibited on the surface, and as a conse-
quence, the oxidation of CO was found to be suppressed in agreement with our
results. Further studies employing in situ DRIFT spectroscopy are in progress
to elucidate the degree of hydroxylation of the RuO2 surface in the presence of

994
V. Narkhede et al.
water vapor up to 453 K and the reversibility of water poisoning at atmospheric
pressure.
5. Conclusions
The redox pretreatment of polycrystalline RuO2 powder strongly influenced
the catalytic activity for the oxidation of CO: the reduction in H2 resulted in the
most active state of the catalyst, whereas the catalytic activity of the reoxidized
metallic Ru samples decreased with increasing oxidation temperature. Deacti-
vation occurred for fully oxidized RuO2 catalysts and for metallic Ru samples
reoxidized at temperatures ≥ 473 K, and its extent was more pronounced in an
excess of O2 in the reactant feed.
The comparison of turnover frequencies derived in this study with those
obtained on Ru single crystal studies and supported Ru catalysts showed that
the reduced Ru catalyst is as active as RuO2 single crystal surfaces and sup-
ported Ru catalysts, i.e. the active state is kinetically identified as an ultra-thin
RuO2 layer covering a metallic Ru core in agreement with the shell-core
model reported for supported Ru nanoparticles. In contrast to the supported Ru
nanoparticles, this RuO2/Ru catalyst did not deactivate due to the significantly
higher stability against bulk oxidation.
TPR measurements carried out subsequent to CO oxidation yielded an
estimated thickness of the stable ultra-thin RuO2 layer of less than 1 nm.
The deactivation observed for RuO2 films thicker than 1 nm and for bulk
RuO2 is ascribed to the formation of RuO2(100)-c(2×2) facets which are
inactive for the oxidation of CO. In the presence of water in the reactant
feed the catalytic activity was found to be completely suppressed. The inhi-
bition by water is assigned to the blocking of the coordinatively unsaturated
Ru sites.
Acknowledgement
Fruitful discussions with Herbert Over and financial support by Deutsche
Forschungsgemeinschaft (DFG, SPP 1091) and Fonds der Chemischen Indus-
trie (FCI) are gratefully acknowledged.
References
1. H. Over, Y. D. Kim, A. P. Seitsonen, S. Wendt, E. Lundgren, M. Schmid, P. Varga,
A. Morgante, and G. Ertl, Science 287 (2000) 1474.
2. H. Over and M. Muhler, Prog. Surf. Sci. 72 (2003) 3.
3. S. Wendt, A. P. Seitsonen, and H. Over, Catal. Today 85 (2003) 167.
4. J. Wang, C. Y. Fan, K. Jacobi, and G. Ertl, J. Phys. Chem. B 106 (2002) 3422.

Structure-Activity Correlations for the Oxidation of CO . . .
995
5. H. Madhavaram, H. Idriss, S. Wendt, Y. D. Kim, M. Knapp, H. Over, J. Aßmann,
E. Löffler, and M. Muhler, J. Catal. 202 (2001) 296.
6. H. Over, M. Knapp, E. Lundgren, A. P. Seitsonen, M. Schmid, and P. Varga, Chem.
Phys. Chem. 5 (2004) 167.
7. M. Knapp, A. P. Seitsonen, Y. D. Kim, and H. Over, J. Phys. Chem. B 108 (2004)
14392.
8. K. Reuter and M. Scheffler, Phys. Rev. B: Condens. Matter Mater. Phys. 68 (2003)
045407/1.
9. K. Reuter and M. Scheffler, Phys. Rev. Lett. 90 (2003) 046103/1.
10. J. Assmann, V. Narkhede, L. Khodeir, E. Löffler, O. Hinrichsen, A. Birkner, O. Over,
and M. Muhler, J. Phys. Chem. B 108 (2004) 14634.
11. U. A. Paulus, Y. Wang, S. H. Kim, P. Geng, J. Wintterlin, K. Jacobi, and G. Ertl,
J. Chem. Phys. 121 (2004) 11301.
12. L. Zang and H. Kisch, Angew. Chem. Int. Ed. 41 (2002) 4538.
13. C. H. F. Peden and D. W. Goodman, J. Phys. Chem. 90 (1986) 1360.
14. J. Assmann, E. Löffler, A. Birkner, and M. Muhler, Catal. Today 85 (2003) 235.
15. J. Assmann, Ruthenium als Oxidationskatalysator – IR-spektroskopische und kine-
tische Untersuchung der Oxidation von CO. Dissertation Ruhr-Universität-Bochum
(2004).
16. D. A. M. Monti and A. Baiker, J. Catal. 83 (1983) 323.
17. M. Prudenziati, B. Morten, and E. Travan, Mater. Sci. Eng. B 98 (2003) 167.
18. J. Assmann, D. Crihan, M. Knapp, E. Lundgren, E. Löffler, M. Muhler, V. Narkhede,
O. Over, M. Schmid, A. Seitsonen, and P. Varga, Angew. Chem. Int. Ed. 44 (2005)
917.
19. T. Wei and J. Phillips, Adv. Catal. 41 (1996) 359.
20. X. Bao, G. Lehmpfuhl, G. Weinberg, R. Schlögl, and G. Ertl, J. Chem. Soc. Faraday
Trans. 88 (1992) 865.