On the nature of the active state of supported ruthenium catalysts used for the oxidation of carbon monoxide: Steady-state and transient kinetics combined with in situ infrared spectroscopy

Article abstract of DOI:10.1021/jp0401675

The oxidation of CO over Ru/MgO and Ru/SiO₂ catalysts was used as a simple model reaction to derive turnover frequencies at atmospheric pressure, which were observed to agree with kinetic data obtained under high-vacuum conditions with supported ruthenium catalysts and the RuO ₂(110) single-crystal surface. Thus, it was possible to bridge both the pressure and the materials gap. However, a partial deactivation was observed initially, which was identified as an activated process, both under net reducing and net oxidizing conditions. Temperature-programmed reduction (TPR) experiments were performed subsequently in the same reactor, to monitor the degree of oxidation, as a function of the reaction temperature and the CO/O ₂ reactant feed ratio. Using in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) measurements, the structural changes of the ruthenium catalysts during the oxidation of CO were confirmed, under relevant reaction conditions. Under net reducing conditions, only domains of RuO₂ seem to exist on the metallic ruthenium particles, whereas, under net oxidizing conditions, the ruthenium particles were fully oxidized to bulk RuO₂ particles, which may expose less-active facets, such as the RuO₂(100)-c(2 ?? 2) surface.

Full text of DOI:10.1021/jp0401675

14634
J. Phys. Chem. B 2004, 108, 14634-14642
On the Nature of the Active State of Supported Ruthenium Catalysts Used for the
Oxidation of Carbon Monoxide: Steady-State and Transient Kinetics Combined with in
Situ Infrared Spectroscopy†
Jens Assmann,‡ Vijay Narkhede,‡ Lamma Khodeir,‡ Elke Lo1ffler,‡ Olaf Hinrichsen,‡
Alexander Birkner,§ Herbert Over, and Martin Muhler*,‡
Lehrstuhl fu¨r Technische Chemie, Ruhr-UniVersita¨t Bochum, D-44780 Bochum, Germany, Lehrstuhl fu¨r
Physikalische Chemie I, Ruhr-UniVersita¨t Bochum, D-44780 Bochum, Germany, and Physikalisch-Chemisches
Institut, Justus-Liebig-UniVersita¨t, D-35392 Giessen, Germany
ReceiVed: February 26, 2004; In Final Form: May 11, 2004
The oxidation of CO over Ru/MgO and Ru/SiO2 catalysts was used as a simple model reaction to derive
turnover frequencies at atmospheric pressure, which were observed to agree with kinetic data obtained under
high-vacuum conditions with supported ruthenium catalysts and the RuO2(110) single-crystal surface. Thus,
it was possible to bridge both the pressure and the materials gap. However, a partial deactivation was observed
initially, which was identified as an activated process, both under net reducing and net oxidizing conditions.
Temperature-programmed reduction (TPR) experiments were performed subsequently in the same reactor, to
monitor the degree of oxidation, as a function of the reaction temperature and the CO/O2 reactant feed ratio.
Using in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) measurements, the structural
changes of the ruthenium catalysts during the oxidation of CO were confirmed, under relevant reaction
conditions. Under net reducing conditions, only domains of RuO2 seem to exist on the metallic ruthenium
particles, whereas, under net oxidizing conditions, the ruthenium particles were fully oxidized to bulk RuO2
particles, which may expose less-active facets, such as the RuO2(100)-c(2 × 2) surface.
1. Introduction
ruthenium catalysts. Thus, an extensive investigation of the
deactivation phenomena is of major importance to establish a
suitable model of the active state of the catalyst, taking results
from recently published RuO2 single-crystal studies into account.
Supported ruthenium catalysts,1-6 and especially ruthenium
single crystals,7-13 have received much attention recently,
because of their high activity in oxidation reactions. Originally,
ruthenium single-crystal surfaces had been classified to be very
poor catalysts for the oxidation of CO under ultrahigh vacuum
(UHV) conditions,14,15 until it was shown that the RuO2(110)
single-crystal film formed autocatalytically on Ru(0001) under
strongly oxidizing conditions is highly active.13 By means of
kinetic investigations under well-defined high-vacuum condi-
tions in the temporal-analysis-of-products (TAP) reactor, the
active phase of supported ruthenium catalysts was identified as
RuO2.1
The deactivation of the supported ruthenium catalysts during
the oxidation of CO is not well understood so far, which was
observed at atmospheric pressure16,17 as well as under high-
vacuum conditions.1 Cant et al.16 concluded that portions of the
metallic ruthenium surface are covered by an inactive oxide
layer, whereas Kiss and Gonzalez suggested that the partial
deactivation of Ru/SiO2 is a result of the formation of lattice
oxygen.17 However, the duration of the deactivation period, the
influence of the support and of the experimental conditions such
as temperature and the CO/O2 reactant feed ratio has not yet
been studied in detail. Furthermore, no attempt has yet been
made to quantify the degree of oxidation of the supported
The present work is focused on the deactivation phenomena
of two differently supported ruthenium catalysts during the
oxidation of CO. The Ru/MgO and Ru/SiO2 catalysts were
prepared by the molecular organic chemical vapor deposition
(MOCVD) method, resulting in a mean ruthenium particle size
of 2 nm.1 The oxidation of CO was monitored during continuous
temperature cycles. Thus, it was possible to determine the
duration of the deactivation and to correlate it with the
experimental conditions. To the best of our knowledge, turnover
frequencies for the oxidation of CO at atmospheric pressure
were derived for the first time using Ru/MgO. In addition, the
transient behavior of preoxidized catalysts was investigated. In
a series of temperature-programmed reduction (TPR) measure-
ments, the degree of oxidation was determined as a function of
the reaction temperature and the pretreatment conditions, using
the same experimental setup. Thus, it was possible to derive
the degree of oxidation immediately after the catalytic oxidation
of CO without any artifacts, because of a lengthy sample
transfer. A shrinking metallic core model is proposed, which
attributes the lower catalytic activity of fully oxidized RuO2
particles to the presence of inactive faces such as the RuO2(100)-
c(2 × 2) surface. Finally, this model was confirmed by an in
situ diffuse reflectance infrared Fourier transform spectroscopy
(DRIFTS) study of the oxidation of CO under relevant reaction
conditions.
† Part of the special issue “Gerhard Ertl Festschrift”.
* Author to whom correspondence should be addressed. Telephone: +49
234 32 28754. Fax: +49 234 32 14115. E-mail address: muhler@techem.
rub.de. URL: http://www.techem.rub.de.
‡ Lehrstuhl fu¨r Technische Chemie.
§ Lehrstuhl fu¨r Physikalische Chemie I.
Justus-Liebig-Universita¨t.
10.1021/jp0401675 CCC: $27.50 © 2004 American Chemical Society
Published on Web 08/19/2004

Supported Ru Catalysts Used for Oxidation of CO
J. Phys. Chem. B, Vol. 108, No. 38, 2004 14635
out
CO2
in
CO2
2. Experimental Section
C
- C
yield of CO2 (%) )
× 100
(2)
in
2.1. Catalyst Preparation and Characterization. The
catalysts Ru/MgO and Ru/SiO2 were prepared by the MOCVD
method.1,18 The Ru3(CO)12 precursor (supplied by Strem) was
deposited on MgO (ALFA, purity of 99.9955%) and on SiO2
(Aerosil 200, from Degussa). The adsorbed precursor was
thermally decomposed in a quartz ampule on a vacuum line
under high-vacuum conditions by executing a multistage tem-
perature program up to 723 K. Ruthenium loadings of 3.0 and
3.2 wt% were determined for Ru/SiO2 and Ru/MgO, respec-
tively. A sieve fraction of 250-355 µm was used for the TPR
and catalytic activity measurements. Using transmission electron
microscopy (TEM), X-ray diffraction (XRD), and H2 chemi-
sorption, the size of the reduced metallic ruthenium particles
was determined to be approximately equal to or smaller than 2
nm.1 For the determination of the ruthenium particle size after
reaction, a Hitachi model H-8100 TEM system, operated at 200
kV, was used. Prior to the TEM investigations, the oxidation
of CO was performed in the fixed-bed microreactor described
below. The subsequent transfer of the ruthenium samples to the
TEM equipment was performed in inert gas without contact
with air, using a glovebox and a special TEM vacuum transfer
holder.
C
CO
in
where C
is the inlet concentration of CO2 and Cout is the
CO2
CO2
outlet concentration of CO2. Turnover frequencies were derived
for low conversions of CO after steady state had been reached.
2.3. Temperature-Programmed Reduction. The tempera-
ture-programmed reduction (TPR) experiments were conducted
in the same experimental setup as that already described
previously. According to the suggested criteria for highly
resolved TPR experiments,19,20 the concentration of H2, the flow
rate, and the heating ramp were correlated with the amount of
catalyst (55-215 mg). The measurements were performed using
the 5.2% H2/Ar mixture at a flow rate of 20 NmL/min. The
temperature was first held for 30 min at 300 K, and then was
linearly increased up to 773 K with heating ramps of 5-10
K/min. The consumption of H2, the reaction products such as
water and methane, and desorbing molecules such as CO and
CO2 were monitored using a calibrated quadrupole mass
spectrometer (Thermostar, from Pfeiffer Vacuum). The con-
sumption of hydrogen was additionally analyzed by a thermal
conductivity detector (Binos, from Fisher-Rosemount). The
TPR measurements were conducted after the supported ruthe-
nium catalysts had been oxidized in pure O2 at different
temperatures, as well as after the oxidation of CO had been
monitored for over 12 h under different reaction conditions.
Based on the consumption of H2 and the formation of CH4, the
degree of oxidation of the ruthenium particles was calculated.
H2, which was consumed for the methanation of adsorbed COads
(eq 3) or CO2 (eq 4) formed by the decomposition of carbonates,
was subtracted from the total consumption of hydrogen:
2.2. Catalytic Activity Measurements. The oxidation of CO
was conducted at atmospheric pressure and at temperatures in
the range of 298-423 K, using a fixed-bed microreactor under
plug-flow conditions. Different amounts of catalysts (12.5-50
mg) of the 250-355 µm sieve fraction were placed in the quartz
reactor, and these samples were diluted with quartz particles of
the same sieve fraction. The temperature was measured directly
below the fixed bed, using a thermocouple that was placed in
the hollow shaft of the reactor. The following gases and gas
mixtures (supplied by Messer Griesheim) were used: argon
(>99.999% purity), 5.2% H2 in argon (hydrogen purity,
>99.9%), 10.0% CO in neon (CO purity, >99.997%; neon
purity, >99.99%), and 10% O2 in neon (oxygen purity,
>99.995%). The total flow rates maintained by three mass flow
controllers were in the range of 50-150 NmL/min. (Note: NmL
represents milliliters normalized to 273.15 K and 1 atm.) The
CO and CO2 gas concentrations were measured using an Uras
14 infrared analyzer module, and the O2 concentration was
measured using a Magnos 16 analyzer (both supplied by
Hartmann & Braun).
COads + 3H2 h CH4 + H2O
CO2 + 4H2 h CH4 + 2H2O
(3)
(4)
The corrected amount of the hydrogen consumption was then
correlated with the reduction of RuOx (eq 5):
RuOx + xH2 h Ru + xH2O
(5)
2.4. In Situ Diffuse Reflectance Infrared Fourier Trans-
form Spectroscopy (DRIFTS). DRIFT spectra were recorded
using a Nicolet Protege´ 460 spectrometer that was equipped
with a diffuse reflectance accessory, a Harrick low-temperature
reaction chamber, and an MCT detector. The sample cup was
covered by a water-cooled stainless-steel dome with KBr
windows. The coolant temperature was maintained at 303 K.
The sample temperature was controlled by a thermocouple that
was in direct contact with the catalyst. The heating element and
a second thermocouple were connected to a temperature
controller. The inlet and outlet of the DRIFTS cell were
connected to the same setup that was used for the investigation
of the CO oxidation kinetics. For the in situ DRIFTS measure-
ments, all catalysts were used as powders. Prior to the oxidation
of CO, both ruthenium catalysts were pretreated as follows:
(1) Heating in a 5.2% H2/Ar mixture at a rate of 10 K/min
up to 773 K and holding at this temperature for 30 min.
(2) Purging with argon at 773 K for 30 min.
Prior to the steady-state oxidation of CO, the supported
ruthenium catalysts were heated to 773 K in a 5.2% H2/Ar gas
mixture, using a heating ramp of 5 K/min, and kept at this
temperature for 1 h. The reactor with the reduced samples was
then purged in argon and cooled to the desired starting
temperature. These conditions are known to result in a fully
reduced and adsorbate-free state of the ruthenium catalysts.18
The CO/O2 reactant feed ratios were changed in the range
between 4 and 0.5, using partial pressures of CO and O2 in the
range of 5-35 mbar.
The conversion of CO was calculated as follows:
in
CO
out
CO
C
- C
conversion of CO (%) )
× 100
(1)
in
C
CO
(3) Cooling in argon down to 323 K.
This procedure ensured a complete reduction of the ruthenium
catalysts. Subsequently, the oxidation of CO was started with
different CO/O2 feed ratios, using a total flow rate of 150 mL/
min at 323 K. After holding this temperature for 30 min, it was
in
CO
out
CO
where C is the inlet concentration of CO and C is the
outlet concentration of CO. The calculation of the yield of CO2
was based on the inlet concentration of CO (C ):
in
CO

14636 J. Phys. Chem. B, Vol. 108, No. 38, 2004
Assmann et al.
Figure 1. Conversion of CO over 50 mg of the Ru/MgO catalyst
diluted with 150 mg of quartz, as a function of the temperature using
heating and cooling ramps as indicated. The total flow rate was 50
NmL/min with a CO/O2 reactant feed ratio of 2.1 (1.8% CO/0.85%
O2).
successively increased, using a heating ramp of 2 K/min up to
343, 363, and 423 K. At 423 K, feed-switching experiments
that used an excess of oxygen and an excess of CO were
performed without additional purging in argon. The DRIFT
spectra were recorded as single-beam spectra in the series mode
of the spectrometer, using a resolution of 4 cm-1 and an average
of 150 scans. To obtain the spectrum of the catalyst, the single
beam spectrum of the sample was divided by the background
spectrum (KBr). The spectra are presented in apparent absorption
units (-log R).
3. Results
3.1. Catalytic Activity and Deactivation. The conversion
measurements were performed with both prereduced supported
ruthenium catalysts, changing between decreasing and increasing
temperatures in a cyclic way. To achieve a steady-state
conversion of CO, each temperature was held constant for 2 h.
This period of time was not sufficient when reducing the
temperature, because of the typical ignition/extinction behavior21
that is observed for highly exothermic heterogeneous oxidation
reactions. Figure 1 illustrates the ignition/extinction phenomenon
using 50 mg Ru/MgO and a CO/O2 reactant feed ratio of 2.1.
When heating from 320 K up to 425 K (heating ramp of 0.5
K/min), the ignition of the reaction was observed at ∼370 K.
The maximum theoretical conversion of CO was reached at 375
K. Cooling to the initial temperature, using a cooling ramp of
0.5 K/min, no decrease of the conversion was observed at
temperatures of >355 K. This temperature difference in the
hysteresis is caused by the heat-transfer limitations at full
conversion. The presence of a hot spot was indicated by an
increase of the reactor temperature, relative to the furnace
temperature. Calculating the heat-transfer coefficient and con-
sidering the diagnostic criteria for heat-transport limitations in
fixed-bed reactors (Reynolds number of Re ) 0.13, Prandtl
number of Pr ) 0.47, and Nusselt number of Nu ) 0.43),22,23
the heat-transfer resistance in the boundary layers adjacent to
the particle must be taken into account at high conversion.
However, at low conversion (<10%), an influence of heat-
transfer limitations can be excluded.
Figure 2. Conversion of CO for the oxidation of CO over 12.5 mg of
Ru/MgO diluted with 75 mg of quartz, as a function 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 (a) 2.0 (1.8% CO/0.9% O2), (b) 1.0 (1.8% CO/1.8% O2), (c) 0.5
(1.8% CO/3.6% O2), and (d) 4.0 (1.8% CO/0.45% O2).
In the following, the results of catalytic activity measurements
that were performed with the diluted Ru/MgO catalyst, are
shown. In Figure 2a, the conversion of CO is shown for Ru/
MgO as a function of time and temperature, using a CO/O2
reactant feed ratio of 2/1. The starting temperature was 373 K,
which was increased stepwise to 423 K and subsequently
reduced to the initial temperature. This procedure was repeated
in a second cycle. During the first increase in temperature, the

Supported Ru Catalysts Used for Oxidation of CO
J. Phys. Chem. B, Vol. 108, No. 38, 2004 14637
conversion decreased slowly after every increase in temperature,
and steady state was not reached within 2 h. The ignition of
the reaction was observed at >393 K, which led to full
conversion of CO. During the cooling period, a delayed decrease
of the conversion was monitored, which can be explained by
the extinction phenomena already mentioned previously. At the
initial temperature of 373 K, the conversion of CO was lower
in the second cycle, compared to the beginning of the reaction,
and no further deactivation was observed during the second
heating ramp, i.e., after every increase in temperature, steady
state was reached immediately. The ignition temperature was
slightly shifted to higher temperatures, in comparison to the first
cycle, and after cooling again to room temperature, it was
observed that the conversion of CO was similar to that observed
at the beginning of the second cycle. Thus, the deactivation
observed under the chosen reaction conditions is restricted to
the initial period of the time-on-stream (first cycle).
Figure 3. Conversion of CO over 12.5 mg of Ru/MgO preoxidized at
room temperature and diluted with 75 mg of quartz, as a function 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 same temperature program was chosen for measurements
that were conducted in an excess of oxygen with CO/O2 reactant
feed ratios of 2/2 and 2/4. The corresponding conversion of
CO is shown in Figure 2b and 2c. During the first heating ramp,
a decrease in the activity can be observed again after every
increase in temperature. This deactivation process seems to be
accelerated by the higher concentration of O2. Moreover, the
conversion of CO decreases as the concentration of oxygen
increases. After the ignition of the reaction, full conversion was
not reached at 423 K. Maximum CO conversions of ∼85% and
55% were obtained for CO/O2 reactant feed ratios of 2/2 and
2/4, respectively. In contrast to the measurements with the
stoichiometric CO/O2 reactant feed ratio of 2/1, where full
conversion was reached below 423 K, it was possible to observe
the ongoing deactivation process also after reaching the highest
temperature step. The extent of the deactivation at this temper-
ature was much higher, in comparison to the lower temperatures
with lower degrees of conversion. In the second cycle, no further
deactivation was observed during the heating ramp for the first
two temperature steps (see Figure 2b and 2c). After reaching a
temperature of 423 K, the decrease of the conversion of CO is
lower, in comparison to the first cycle, and the degrees of
conversion of the following cooling steps are similar to those
measured during the heating ramp in the second cycle. These
observations imply that, under net oxidizing conditions, the
deactivation is restricted to the first cycle.
Figure 4. Conversion of CO over 12.5 mg of Ru/MgO diluted with
75 mg of quartz at 423 K, using a total flow rate of 50 NmL/min as a
function of time. The CO/O2 reactant feed ratio was changed systemati-
cally: (a) CO/O2 ratio ) 2 after oxidation of the catalyst at room
temperature, (b) CO/O2 ratio ) 1 after that described in trace a, (c)
CO/O2 ratio ) 0.5 after that described in trace b, and (d) CO/O2 ratio
) 2 after that described in trace c.
(see Figure 2a). The result of this measurement, which is shown
in Figure 3, is hardly distinguishable from the results obtained
during the second cycle, using the reduced Ru/MgO catalyst
(see Figure 2a); i.e., no deactivation was observed during the
first heating period after reaching the next temperature step,
and the degrees of conversion agreed with those of the
deactivated catalyst within 2%. Obviously, the oxidative pre-
treatment did result in the same less-active state of the Ru/MgO
catalyst. The degrees of conversion obtained during the initial
period were strongly dependent on the pretreatment condi-
tions: starting the oxidation of CO with a reduced catalyst leads
to a higher initial activity but includes a deactivation process.
On the other hand, the initial activity of the preoxidized catalyst
is stable, but rather low. Control experiments were also
performed, using temperature programs starting at 423 K, which
showed again that the degree of deactivation is determined by
the highest reaction temperature and not by the sequence of
increasing and decreasing temperatures.
Because the catalytic activity was observed to be strongly
influenced by the CO/O2 reactant feed ratio under net oxidizing
conditions (see Figure 2b and 2c), flow-switching experiments
were performed with both supported ruthenium catalysts at 423
K to study the dynamics of this inhibition. Figure 4 shows the
results obtained with the Ru/MgO catalyst. Using a stoichio-
metric CO/O2 reactant feed ratio of 2/1, full conversion of CO
was observed, even after an oxidizing pretreatment at room
The oxidation of CO was also performed using the same
temperature program in an excess of CO (CO/O2 reactant feed
ratio of 4). As a consequence of the net reducing conditions,
the maximum theoretical conversion of CO was restricted to
50%. Figure 2d shows the conversion of CO as a function of
time and temperature. An enhanced deactivation during the first
increase in temperature is observed. However, this process is
again restricted to the initial period of the time-on-stream (first
cycle), because it did not appear during the second cycle, during
which steady state was reached immediately after every increase
in temperature. To restore the high initial activity, the reduction
in H2/argon at 773 K had to be repeated, independent of the
CO/O2 reactant feed ratio, which had been applied during the
reaction. The subsequently observed deactivation was again
controlled by the reactant gas mixture and the temperature.
To study the role of the degree of oxidation in more detail,
the reduced Ru/MgO sample was oxidized at room temperature
and subsequently purged in argon. The oxidation of CO then
was monitored using the same temperature program and the
stoichiometric CO/O2 reactant feed ratio of 2, which were each
applied for the measurements with the initially reduced catalyst

14638 J. Phys. Chem. B, Vol. 108, No. 38, 2004
Assmann et al.
Figure 6. TPR profiles obtained with Ru/SiO2 oxidized at (a) room
temperature and (b) 473 K. The K-factor19 and P-factor20 were 83 s
and 13.8 K, respectively.
Figure 5. Arrhenius plot of the rates of CO2 formation (TOF denotes
the turnover frequency, defined as the number of CO2 molecules formed
per metal surface site per second) over the supported ruthenium catalysts
(Ru/MgO and Ru/SiO2) for different CO/O2 reactant feed ratios and in
dependence on total pressure. Data obtained on Ru(0001) at high
pressure,15 on a silica-supported ruthenium catalyst prepared by
impregnation,16 and TOFs over supported rhodium catalysts16,17 also
are included.
temperature (see trace a in Figure 4). Switching to a CO/O2
reactant feed ratio of 2/2 led to a decrease of the conversion of
CO to ∼70% (see trace b in Figure 4). After increasing the O2
concentration further (a CO/O2 reactant feed ratio of 2/4 was
used), the conversion of CO decreased to ∼30% (see trace c in
Figure 4). Switching back to the stoichiometric CO/O2 reactant
feed ratio restored the initial activity after a period of ∼4 h
(see trace d in Figure 4), indicating a reversible change of the
state of the catalyst under net oxidizing conditions.
Figure 7. TPR profiles obtained after CO oxidation at 423 K under
net oxidizing conditions (CO/O2 reactant feed ratio of <2): (a) Ru/
SiO2 (213 mg), (b) Ru/SiO2 (99 mg), (c) Ru/MgO (201 mg), and (d)
Ru/MgO (103 mg). The gray trace displays the formation of methane,
multiplied by a factor of 4.
Turnover frequencies (TOFs), which represent the number
of CO2 molecules formed per metal surface site per second,
were calculated for low and steady-state conversion for the
deactivated state of the Ru/MgO catalyst. The TOFs obtained
from measurements with a CO/O2 reactant feed ratio of 2 are
plotted for Ru/MgO and Ru/SiO2 in an Arrhenius diagram
(shown in Figure 5). Also, TOFs obtained in a high-pressure
single-crystal study15 and TOFs found in a recent high-vacuum
TAP reactor study that used the same supported ruthenium
catalysts (Ru/MgO and Ru/SiO2)1 are included in this figure.
The results show that an influence of the support on the activity
at atmospheric pressure can be neglected. In addition, the
activation energy of ∼82 kJ/mol for Ru/MgO is in very good
agreement with both the single-crystal data and the TAP results,
indicating the absence of a pressure gap. A comparison with
supported rhodium catalysts (also included in Figure 5) reveals
a much higher activity of the ruthenium catalysts.
3.2. Temperature-Programmed Reduction (TPR) and
Transmission Electron Microscopy (TEM) Measurements
after Reaction. The degrees of oxidation of the supported
ruthenium catalysts were determined using the TPR method.
The TPR profiles obtained after oxidation of the Ru/SiO2 sample
at room temperature and at 473 K in flowing O2 are presented
in Figure 6. After oxidation at room temperature (Figure 6a),
the TPR profile consists of only one peak at ∼360 K, whereas
after oxidation at 473 K, a sharper peak was observed at 413
K, in addition to the peak at lower temperatures. The total
consumption of H2 yielded O/Ru ratios of 1.3 and 1.5,
respectively. For Ru/MgO, the same O/Ru ratios were deter-
mined. However, no consumption of hydrogen was observed
at >300 K for the Ru/MgO sample that was oxidized at room
temperature; i.e., it was fully reduced already at room temper-
ature when switching to H2/Ar. The TPR profile of the Ru/
MgO sample oxidized at 473 K (not shown) was observed to
consist of the same two peaks as those of the Ru/SiO2 catalyst
(Figure 6b) but with lower intensity. In addition to the
consumption of H2 at 360 and 413 K, H2 was already consumed
at the initial temperature of 300 K during the TPR experiment,
resulting in the same O/Ru ratio of 1.5 calculated for both
catalysts that were oxidized at 473 K.
The results of the TPR measurements, after reaction at 423
K under net oxidizing conditions (CO/O2 reactant feed ratio of
<2), are shown in Figure 7 for both supported catalysts. Under
the conditions applied for the oxidation of CO, full conversion
was observed, inducing strong heat production. The shape of
the TPR profiles of the Ru/SiO2 sample (see Figure 7a and 7b)
is similar to those obtained after oxidation in pure O2; however,
the height of both peaks at 383 and 415 K is significantly
increased. The TPR profiles of the Ru/MgO sample after
reaction display an additional feature (see Figure 7c and 7d).
In addition to the consumption of H2 at 300 K and at ∼360 and
415 K, a broad shoulder appeared at ∼490 K. The desorption
of methane produced during the reduction process was detected
by mass spectroscopy at the same temperature of 490 K (see
gray trace in Figure 7d). Assuming the methanation of CO2, a
H2/CH4 ratio of 4 must be considered. Multiplying the amount
of methane by a factor of 4 indicates that the methanation of
CO2 fully accounts for the additional consumption of H2 at 490
K. After subtracting the amount of H2 consumed for the latter
reaction, O/Ru ratios were calculated. Independent of the
support, O/Ru ratios equal to 2 were observed, indicating the
presence of fully oxidized RuO2 particles under net oxidizing
reaction conditions.

Supported Ru Catalysts Used for Oxidation of CO
J. Phys. Chem. B, Vol. 108, No. 38, 2004 14639
Figure 8. TPR profiles obtained after CO oxidation at 423 K under
net reducing conditions (CO/O2 reactant feed ratio of >2): (a) Ru/
MgO (103 mg) and (b) Ru/SiO2 (100 mg). The gray traces display the
formation of methane, multiplied by a factor of 3.
The TPR profiles of the Ru/MgO and Ru/SiO2 catalysts after
reaction under net reducing conditions (CO/O2 reactant feed ratio
of 2.2) are shown in Figure 8. In contrast to the TPR profiles
presented in Figure 7, only one TPR peak was observed, at 500
and 453 K for Ru/MgO and Ru/SiO2, respectively. At the same
temperatures, the desorption of methane was monitored. As-
suming the methanation of adsorbed CO (eq 3), a H2/CH4 ratio
of 3 must be considered. Multiplying the amount of methane
by a factor of 3 led to almost-coincident traces, as shown in
Figure 8 (gray traces in Figure 8a and 8b). Thus, no significant
oxidation of the ruthenium particles was observed under net
reducing conditions at high conversion. Therefore, it is reason-
able to assume that only a few RuO2 domains are present on
the surface of the metallic ruthenium particles under these
conditions.
Figure 9. (a) TEM micrograph and (b) corresponding particle size
distribution of the Ru/SiO2 catalyst obtained after CO oxidation at 423
K using a CO/O2 reactant feed ratio of 0.5.
The TEM images of the ruthenium catalysts were recorded
after the oxidation of CO at 423 K in an excess of O2. Under
these reaction conditions, fully oxidized RuO2 particles are
present, as shown by the TPR results. Figure 9a and 9b
respectively show the TEM micrograph and the corresponding
particle size distribution of the RuO2/SiO2 sample. The latter
displays two maxima, at ∼2.5 and 4.5 nm. The comparison with
the particle size distribution of the reduced Ru/SiO2 sample
obtained prior to reaction1 reveals minor sintering of the
ruthenium particles in the oxidized state, as a consequence of
the net oxidizing conditions and the high temperatures applied
during the oxidation of CO.
3.3. In Situ DRIFTS Measurements. The oxidation of CO
was also investigated using the in situ DRIFTS cell as a reactor.
The deactivation phenomena were observed to be similar, in
comparison to the results obtained under plug-flow conditions
in the fixed-bed reactor, whereas the degrees of conversion were
consistently lower. The IR spectra in the carbonyl region
obtained with the Ru/MgO and Ru/SiO2 catalysts recorded at
423 K, using a CO/O2 reactant feed ratio of 1/2, are shown as
dotted traces in Figure 10. The spectra are characterized by two
bands, at ∼2130 and ∼2070 cm-1, and a weak shoulder at
∼2000 cm-1. As demonstrated in a recent DRIFTS study,1 these
bands can be assigned to C-O vibrations of CO adsorbed on a
RuO2 surface layer, which is comparable to the RuO2(110)
single-crystal surface. Both high-frequency bands are assigned
to a multicarbonyl species formed by the binding of two or three
CO ligands to a Ru site, which may be similar to the 1-fold
coordinatively unsaturated Ru site present on the RuO2(110)
single-crystal surface.12,24 The weak shoulder at ∼2000 cm-1
is assigned to C-O vibrations of CO adsorbed in an asymmetric
bridge position on oxygen vacancies in the RuO2 film.1,24 The
Figure 10. DRIFT spectra obtained during CO oxidation at 423 K
using a total flow rate of 150 NmL/min with CO/O2 ratios of 0.5 (1.7%
CO/3.4% O2, dotted traces) and 4.0 (1.7 CO/0.43% O2, solid traces)
over (a) Ru/MgO and (b) Ru/SiO2.
low intensity of this band is obviously due to the low CO/O2
reactant feed ratio. The assignment of the bands is summarized
in Table 1.
In contrast to these results, the DRIFT spectra recorded at
423 K in an excess of CO revealed different bands in the
carbonyl region, as shown in Figure 10 (solid traces). These
spectra were normalized using the second high-frequency band
observed in an excess of oxygen. Independent of the support,
an additional band at ∼2050 cm-1 is observed. This band is
assigned to CO adsorbed linearly on reduced metallic Ru sites,
in agreement with the C-O vibration frequency of CO adsorbed
on the Ru(0001) single-crystal surface.25 In addition to the
observed presence of metallic Ru sites, evidence is also provided

14640 J. Phys. Chem. B, Vol. 108, No. 38, 2004
Assmann et al.
TABLE 1: Assignment of the Bands of Adsorbed CO
Observed during CO Oxidation over the Ru/MgO and
Ru/SiO2 Catalysts at 423 K, Using a Total Flow Rate of 150
NmL/min with CO/O2 Ratios of 0.5 (1.7% CO/3.4% O2) and
4.0 (1.7% CO/0.43% O2)
analysis of the TPR measurements for both ruthenium catalysts
that were preoxidized at room temperature, which is in agree-
ment with O2 chemisorption measurements.1 This ratio is
equivalent to a RuO2 surface layer when the dispersion of
ruthenium (∼65%) is taken into account. After the oxidation
of CO under net reducing conditions and at high conversion,
no indication for oxidized ruthenium particles was observed by
the TPR experiments within the experimental accuracy, because
the total consumption of H2 was due to the methanation of
adsorbed CO. Note that the amount and the peak temperature
of the methane desorption were dependent on the support. For
the Ru/MgO sample, a larger amount of methane was observed,
compared with the Ru/SiO2 sample. Moreover, the methanation
occurs at higher temperatures for Ru/MgO. This observation
suggests that CO is slightly stronger bound onto the surface of
Ru/MgO and, therefore, less reactive for the methanation
reaction, which is in agreement with recent infrared studies.1
RuO2 Assignment (cm-1
CO/O2 ratio multicarbonyl asymmetric bridge on-top (cm-1
)
metallic Ru,
)
Ru/MgO Catalyst
0.5
4.0
2125/2067
2125/2067
1990
1990
2043
2051
Ru/SiO2 Catalyst
0.5
4.0
2127/2078
2010
2010
2134/2078
for the presence of oxidized Run+ sites by the two high-
frequency bands at ∼2130 and ∼2070 cm-1. Thus, RuO2
domains are present, leading again to an assignment of the band
at 1990 cm-1 in the spectra of Ru/MgO to CO adsorbed on
oxygen vacancies of the RuO2 surface.1
In contrast, even in a small excess of O2, the ruthenium
particles were completely oxidized to RuO2. As shown by the
TEM measurements, the fully oxidized RuO2 particles were still
rather small (<5 nm), even after reaction at 423 K and under
strong oxidizing conditions. Sintering of the RuO2 particles
during the oxidation of CO had indeed occurred to a minor
extent, as judged by the bimodal particle size distribution with
maxima at 2.5 and 4.5 nm in the case of Ru/SiO2. Thus, the
two TPR signals observed at ∼370 and 415 K can be assigned
to the different rates of reduction of smaller and larger RuO2
particles, respectively. The observation of only one TPR peak
at ∼360 K after oxidation of the Ru/SiO2 sample at room
temperature is in agreement with this interpretation, because
sintering of the particles is only favored at higher temperatures.
By means of a series of O2 and H2 chemisorption measurements,
it was shown that oxidation at room temperature causes no
agglomeration of the ruthenium particles.1 For significantly
larger oxidized ruthenium particles, TPR maxima at >473 K
were observed.2
According to the DRIFTS results, the active state of the
ruthenium catalysts in an excess of O2 is characterized by the
presence of RuO2 surfaces that are similar to the RuO2(110)
single-crystal surface, whereas, under net reducing conditions,
mildly reduced RuO2 domains and metallic ruthenium surfaces
coexist. In the case of Ru/SiO2, the metallic ruthenium surfaces
seem to be dominant, as shown by the higher intensity of the
band at 2051 cm-1 in Figure 10b (solid trace).
4. Discussion
The results of the catalytic activity measurements presented
for the Ru/MgO sample indicate the influence of the CO/O2
reactant feed ratio, the temperature, and the pretreatment of the
sample on the rates of CO oxidation. The results obtained with
the Ru/SiO2 sample were determined to be in good agreement
with those of the Ru/MgO catalyst, as shown by the good
agreement between the turnover frequencies plotted in Figure
5. Thus, a significant influence of the support on the deactivation
behavior and the CO conversion can be ruled out. The turnover
frequencies are also independent of the total pressure, which is
illustrated by the good agreement with the TOFs obtained in a
recent high-vacuum TAP reactor study.1
The TOFs were determined after the deactivation process had
been finished. The repetition of heating and cooling cycles (up
to four cycles) provided clear evidence that the deactivation of
the pre-reduced supported ruthenium catalysts is restricted to
the initial period of the time-on-stream. This is valid for all
CO/O2 reactant feed ratios, which determined the extent and
rate of deactivation. The preoxidized sample had a lower initial
activity and did not deactivate during the first increase in
temperature; instead, a rather slow activation process was
observed. No difference in the conversion of CO was observed
during the second cycle of the catalytic activity measurements
that were performed with a stoichiometric CO/O2 reactant feed
ratio. Obviously, the active state of the catalyst after the
termination of the initial period is independent of the pretreat-
ment. Therefore, we can assume a partial reduction of the
oxidized ruthenium surface and a partial oxidation of the reduced
ruthenium catalyst during the first cycle when a CO/O2 reactant
feed ratio of g2 is used. The state of the catalyst with the highest
activity seems to be metallic bulk ruthenium that was covered
by a partially reduced RuO2 surface layer. Correspondingly, a
re-reduction of the used catalysts in H2 at 773 K was determined
to be necessary to restore the initial high activity, in good
agreement with the observations by Kiss and Gonzalez.17
This hypothesis is supported by the results of TPR measure-
ments. An O/Ru ratio of 1.3 was derived from the quantitative
The consumption of H2 at room temperature observed for
the Ru/MgO sample either in the preoxidized state or after CO
oxidation under net oxidizing conditions can be rationalized by
recent single-crystal studies on the interaction between H2 and
the RuO2(110) single-crystal surface.9 Dissociatively adsorbed
hydrogen was observed to desorb completely only at temper-
atures of >350 K. Up to this temperature, dihydride species
formed from H2 and a coordinatively unsaturated O surface atom
transform to monohydride before water molecules desorb at
∼400 K. Based on this single-crystal study, the consumption
of H2 at <300 K in the case of Ru/MgO is assigned to the
dissociative adsorption of H2 on the supported RuO2 particles,
probably leading to the formation of dihydride or monohydride
species.
The results of the in situ DRIFTS experiments correspond
well with the conclusions drawn from the kinetic investigations.
The observed CO bands clearly indicate that the active surface
is RuO2 under net oxidizing conditions. Furthermore, the
coexistence of metallic ruthenium islands and RuO2 domains
was observed for CO/O2 reactant feed ratios of g2 at high
temperatures and high conversion. The influence of the CO/O2
reactant feed ratio on the nature of the Ru surface sites and the
total degree of oxidation found in the DRIFTS and TPR
measurements is in agreement with theoretical calculations,10
which demonstrated the influence of the CO/O2 ratio on both
the surface structures and the stability of bulk RuO2. Both the
theoretical predictions and our results reveal the importance of
phase coexistence conditions that lead to enhanced dynamics

Supported Ru Catalysts Used for Oxidation of CO
J. Phys. Chem. B, Vol. 108, No. 38, 2004 14641
Figure 11. Schematic of the shrinking metallic core model. The degree of oxidation is determined by the CO/O2 reactant feed ratio and the
temperature. The most active state is represented by a partially reduced RuO2 surface layer supported on a metallic ruthenium core. The formation
of inactive RuO2 surface facets is favored under net oxidizing conditions, leading to bulk RuO2 particles, whereas hardly active metallic ruthenium
surfaces are exposed under net reducing conditions.
at the catalyst surface. These changes are presumably more
pronounced at higher temperatures and are very sensitive to the
CO/O2 reactant feed ratio. Drastic surface morphology changes
have been observed for the mildly reduced RuO2(110) single-
crystal surface after annealing to 600 K.26 As a consequence of
this annealing process, a restored oxide with holes and small
ruthenium clusters was formed.
An illustration of the processes that occur and the parameters
that control the changes of the supported ruthenium catalysts is
presented in Figure 11. The pre-reduced ruthenium particles are
initially very active for the oxidation of CO. Even under net
reducing conditions, the ruthenium particles are covered by a
RuO2 surface layer. The rate of its formation and its thickness
are controlled by the CO/O2 reactant feed ratio. Increasing the
temperature under net reducing conditions leads to a partial
reduction of the RuO2 layer. Thus, hardly active metallic Ru
sites are exposed to the gas phase, as confirmed by the in situ
DRIFTS measurements. Increasing the temperature under net
oxidizing conditions leads to an ongoing oxidation of the
ruthenium particles, which are finally fully oxidized to bulk
RuO2.
Based on this shrinking-core model, the restriction of the
deactivation to the initial period during the time-on-stream (first
cycle) is comprehensible. Under net oxidizing conditions, the
highest temperature applied for the oxidation of CO determines
the maximum degree of oxidation. Cooling and then heating
again has no more influence on the degree of oxidation of the
ruthenium particles. In consequence, steady-state conversion is
immediately observed. Under net reducing conditions, the degree
of oxidation is also changed in a continuous way, which is
controlled by the temperature. At low conversion, the active
catalyst is covered by a highly active RuO2 surface layer, which
is partially reduced. Only at high conversion and under net
reducing conditions, the reduction of the RuO2 surface layer is
faster than its reoxidation, a process that was also observed for
the oxidation of methane over Ru/Al2O3.2
sensitivity found on RuO2 single-crystal surfaces, the shrinking
metallic core model is extended by taking the different possible
surface structures of the RuO2 layer into account. The higher
the degree of oxidation, the more inactive RuO2 planes are
exposed, leading to an overall lower catalytic activity.
Further studies that apply in situ extended X-ray absorption
fine structure (EXAFS) and DRIFTS combined with O2 TPD
and Oads + CO temperature-programmed surface reaction
(TPSR) measurements are underway, to provide additional
evidence for the model of the active state and its different
methods of deactivation (shown in Figure 11).
5. Conclusions
(1) By optimizing the experimental conditions in a fixed-
bed reactor operated at atmospheric pressure, it was possible to
derive turnover frequencies (TOFs) for the oxidation of CO over
Ru/MgO at low conversion. The good agreement with kinetic
data obtained with the RuO2(110) single-crystal surface and with
Ru/SiO2 under high vacuum conditions demonstrates that both
the materials and the pressure gap were successfully bridged.
The support is necessary to maintain a high dispersion of the
ruthenium particles but has no direct influence on the catalytic
activity.
(2) The systematic investigation of the deactivation process
revealed that it is affected by the CO/O2 ratio and the reaction
temperature, and, in particular, that it is limited to the initial
period of the time-on-stream.
(3) Based on temperature-programmed reduction (TPR)
measurements that were performed in the same reactor after
the oxidation of CO, the degree of oxidation of the ruthenium
particles was derived. When starting in the metallic state under
net reducing conditions, the formation of a partially reduced
highly active RuO2 surface layer was detected. Under net
oxidizing conditions, the ruthenium particles were determined
to be completely oxidized to bulk RuO2.
Finally, the question must be addressed as to why the fully
oxidized RuO2 particles are less active catalytically. RuO2
single-crystal studies showed that two different phases coexist
on the RuO2(100) surface.26,27 This surface is characterized by
an (1 × 1) phase that is as active as the RuO2(110) surface,
and a c(2 × 2) phase that is inactive for the adsorption of CO.
Furthermore, the formation of the inactive c(2 × 2) phase is
favored under oxidizing conditions and higher temperatures.
Thus, also inactive surface planes are assumed to be present on
the supported RuO2 particles to an increasing extent with
decreasing CO/O2 reactant feed ratio. Based on the structure
(4) The dynamic changes of the surface structure of the
ruthenium catalysts, as a function of the CO/O2 reactant feed
ratio, were demonstrated by in situ diffuse reflectance infrared
Fourier spectroscopy (DRIFTS) experiments conducted under
reaction conditions. Obviously, the highest catalytic activity is
correlated with the presence of a partially reduced RuO2 surface
layer supported on metallic bulk ruthenium. Further reduction
of this RuO2 surface layer was observed under net reducing
conditions at higher temperatures, leading to a deactivated state
of the catalysts due to exposed metallic ruthenium surfaces.
Based on the structure sensitivity of the CO oxidation over RuO2

14642 J. Phys. Chem. B, Vol. 108, No. 38, 2004
Assmann et al.
(10) Reuter, K.; Scheffler, M. Phys. ReV. B 2003, 68, 045407.
surfaces, the decrease of the activity under net oxidizing
conditions is rationalized by assuming that bulk RuO2 particles
expose less-active facets such as the RuO2(100)-c(2 × 2)
surface.
(11) Reuter, K.; Scheffler, M. Phys. ReV. Lett. 2003, 90, 046103.
(12) Fan, C. Y.; Wang, J.; Jacobi, K.; Ertl, G. J. Chem. Phys. 2001,
114, 10058.
(13) Over, H.; Kim, D. Y.; Seitsonen, P. A.; Wendt, S.; Lundgren, E.;
Schmid, M.; Varga, P.; Morgante, A.; Ertl, G. Science 2000, 287, 1474.
(14) Lee, H. I.; White, J. M. J. Catal. 1980, 63, 261.
(15) Peden, C. H. F.; Goodman, D. W. J. Phys. Chem. 1986, 90, 1360.
(16) Cant, N. W.; Hicks, P. C.; Lennon, B. S. J. Catal. 1978, 54, 372.
(17) Kiss, J. T.; Gonzalez, R. D. J. Phys. Chem. 1984, 88, 892.
(18) Bielawa, H.; Hinrichsen, O.; Birkner, A.; Muhler, M. Angew. Chem.
2001, 113, 1093; Angew. Chem., Int. Ed. 2001, 40, 1061.
(19) Monti, D. A. M.; Baiker, A. J. Catal. 1983, 83, 323.
(20) Malet, P.; Caballero, A. J. Chem. Soc., Faraday Trans. 1988, 84,
2369.
(21) Assovskii, I. G. Combust., Explos. Shock WaVes 2003, 34, 163.
(22) Mears, D. E. Ind. Eng. Chem. Process Des. DeV. 1971, 10, 541.
(23) Mears, D. E. J. Catal. 1971, 20, 127.
(24) Seitsonen, A. P.; Kim, Y. D.; Knapp, M.; Wendt, S.; Over, H. Phys.
ReV. B 2002, 65, 035413.
Acknowledgment. The authors gratefully acknowledge
financial support by the Deutsche Forschungsgemeinschaft
(DFG, SPP 1091).
References and Notes
(1) Assmann, J.; Lo¨ffler, E.; Birkner, A.; Muhler, M. Catal. Today
2003, 85, 235.
(2) Balint, I.; Miyazaki, A.; Aika, K. J. Catal. 2003, 220, 74.
(3) Balint, I.; Miyazaki, A.; Aika, K. React. Kinet. Catal. Lett. 2003,
80, 81.
(4) Snytnikov, P. V.; Sobyanin, A. V.; Belyaev, D. V.; Tsyrulnikov,
G. P.; Shitova, B. N.; Shlyapin, A. D. Appl. Catal., A 2003, 239, 149.
(5) Han, Y.-F.; Kahlich, M. J.; Kinne, M.; Behm, R. J. Phys. Chem.
Chem. Phys. 2002, 4, 389.
(6) Zang, L.; Kisch, H. Angew. Chem. 2000, 112, 4075; Angew. Chem.,
Int. Ed. 2000, 39, 3921.
(25) Peden, C. H. F.; Goodman, D. W.; Weisel, M. D.; Hoffmann, M.
Surf. Sci. 1991, 253, 44.
(26) Over, H.; Knapp, M.; Lundgren, E.; Seitsonen, A. P.; Schmid,
M.; Varga, P. Chem. Phys. Chem. 2004, 5, 10.
(27) Kim, Y. D.; Schwegmann, S.; Seitsonen, A. P.; Over, H. J. Phys.
Chem. B 2001, 105, 2205.
(7) Over, H.; Muhler, M. Prog. Surf. Sci. 2003, 72, 3.
(8) Wendt, S.; Seitsonen, A. P.; Over, H. Catal. Today 2003, 85, 167.
(9) Wang, J.; Fan, C. Y.; Sun, Q.; Reuter, K.; Jacobi, K.; Scheffler,
M.; Ertl, G. Angew. Chem. 2003, 115, 2201.