Catalytic oxidation of ammonia on RuO2(110) surfaces: Mechanism and selectivity

Article abstract of DOI:10.1021/jp045735v

The selective oxidation of ammonia to either N₂ or NO on RuO₂(110) single-crystal surfaces was investigated by a combination of vibrational spectroscopy (HREELS), thermal desorption spectroscopy (TDS) and steady-state rate measurements under continuous flow conditions. The stoichiometric RuO₂(110) surface exposes coordinatively unsaturated (cus) Ru atoms onto which adsorption of NH₃ (NH₃-cus) or dissociative adsorption of oxygen (O-cus) may occur. In the absence of O-cus, ammonia desorbs completely thermally without any reaction. However, interaction between NH₃-cus and O-cus starts already at 90 K by hydrogen abstraction and hydrogenation to OH-cus, leading eventually to N-cus and H ₂O. The N-cus species recombine either with each other to N ₂ or with neighboring O-cus leading to strongly held NO-cus which desorbs around 500 K. The latter reaction is favored by higher concentrations of O-cus. Under steady-state flow condition with constant NH₃ partial pressure and varying O₂ pressure, the rate for N₂ formation takes off first, passes through a maximum and then decreases again, whereas that for NO production exhibits an S-shape and rises continuously. In this way at 530 K almost 100% selectivity for NO formation (with fairly high reaction probability for NH₃) is reached.

Full text of DOI:10.1021/jp045735v

J. Phys. Chem. B 2005, 109, 7883-7893
7883
Catalytic Oxidation of Ammonia on RuO2(110) Surfaces: Mechanism and Selectivity
Y. Wang, K. Jacobi,* W.-D. Scho1ne, and G. Ertl
Department of Physical Chemistry, Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6,
D-14195 Berlin, Germany
ReceiVed: September 20, 2004; In Final Form: February 24, 2005
The selective oxidation of ammonia to either N2 or NO on RuO2(110) single-crystal surfaces was investigated
by a combination of vibrational spectroscopy (HREELS), thermal desorption spectroscopy (TDS) and steady-
state rate measurements under continuous flow conditions. The stoichiometric RuO2(110) surface exposes
coordinatively unsaturated (cus) Ru atoms onto which adsorption of NH3 (NH3-cus) or dissociative adsorption
of oxygen (O-cus) may occur. In the absence of O-cus, ammonia desorbs completely thermally without any
reaction. However, interaction between NH3-cus and O-cus starts already at 90 K by hydrogen abstraction
and hydrogenation to OH-cus, leading eventually to N-cus and H2O. The N-cus species recombine either
with each other to N2 or with neighboring O-cus leading to strongly held NO-cus which desorbs around 500
K. The latter reaction is favored by higher concentrations of O-cus. Under steady-state flow condition with
constant NH3 partial pressure and varying O2 pressure, the rate for N2 formation takes off first, passes through
a maximum and then decreases again, whereas that for NO production exhibits an S-shape and rises
continuously. In this way at 530 K almost 100% selectivity for NO formation (with fairly high reaction
probability for NH3) is reached.
1. Introduction
The catalytic oxidation of ammonia to NO, the so-called
Ostwald process, is an important industrial process for the
production of nitric acid.1,2 On the other hand, an alternate route
of ammonia oxidation leading to nitrogen and water has become
of increasing interest in connection with the removal of ammonia
from waste streams (see ref 3 and references therein). This is a
particularly obvious example for the relevance of selectivity in
heterogeneous catalysis. This term denotes the ratio of the rate
for production of the desired product over the sums of the rates
for all reaction channels. As emphasized recently by Somorjai:4
“The focus of catalysis research in the 21st century should be
on achieVing 100% selectiVity for the desired product”. In
contrast to catalytic activity (i.e., the productivity per site and
second), our knowledge about the factors affecting selectivity
is still rather limited. According to the just given definition,
research on this phenomena should concentrate on the rates
(under steady-state flow conditions) and their control by the
various external parameters. For the example to be presented
here this latter factor is revealed to be particularly important:
Whereas with low relative O2 concentrations N2 is the main
reaction product, the selectivity toward NO formation increases
continuously with the proportion of oxygen and reaches
eventually nearly 100%. The study was performed with a well-
defined RuO2(110) single-crystal surface as model system, and
both the reaction mechanism and the question of selectivity
could be clarified on an atomic scale.
Figure 1. Ball-and-stick model of the stoichiometric RuO2(110) surface
in perspective view together with adsorbed NH3 and O atoms (O-cus).
species to be discussed below. It is characterized by two
different, coordinatively unsaturated surface atoms organized
in rows along [001]: (i) 2-fold coordinated oxygen atoms (O-
bridge) and (ii) 5-fold coordinated Ru atoms (Ru-cus). Ad-
ditional oxygen atoms may be adsorbed by further exposure to
O2 on top of Ru-cus, called O-cus in the following. A maximum
coverage of about 80% of the Ru-cus atoms may thus be
achieved.9,10 The O-cus species is weakly bound on the surface
and desorbs at temperatures as low as 400-500 K. Therefore,
this species is expected to be very reactive as verified, e.g., by
CO oxidation,6,8 carbonate formation,10,11 and ethylene oxida-
tion.12
In the present work the interaction of ammonia with the
stoichiometric and oxygen-rich RuO2(110) surfaces is studied
in detail using thermal desorption spectroscopy (TDS) and high-
resolution electron energy-loss spectroscopy (HREELS). At the
stoichiometric surface NH3 chemisorbs molecularly on Ru-cus,
as also sketched in Figure 1, and desorbs molecularly again. In
the presence of O-cus, however, high catalytic activity for NH3
oxidation is observed, leading either to NO or N2, whereby the
product selectivity is controlled by the O-cus coverage. Iden-
tification of the reaction intermediates enables formulation of
the reaction mechanism which is verified in steady-state rate
measurements under flow conditions.
Recently, the RuO2(110) surface, prepared by exposing Ru-
(0001) to high doses of O2 at elevated sample temperatures,5-7
has been found to exhibit high catalytic activity for CO
oxidation.5,6,8 The stoichiometric RuO2(110) surface is schemati-
cally shown in Figure 1 together with the relevant adsorbed
* To whom the correspondence should be addressed. Phone: +49-30-
8413-5201. Fax: +49-30-8413-5106. E-mail: jacobi@fhi-berlin.mpg.de.
10.1021/jp045735v CCC: $30.25 © 2005 American Chemical Society
Published on Web 03/24/2005

7884 J. Phys. Chem. B, Vol. 109, No. 16, 2005
Wang et al.
2. Experimental Section
The experiments were performed in an ultrahigh vacuum
(UHV) apparatus consisting of two chambers separated by a
valve. The base pressure was 2 × 10-11 mbar. The upper
chamber was used for sample preparation and contained a
quadrupole mass spectrometer for TDS and for reaction rate
measurements, as well as facilities for low-energy electron
diffraction (LEED), gas dosing, and surface cleaning by Ar ion
sputtering. The lower chamber housed a HREEL spectrometer
(Delta 0.5, SPECS). Spectra were taken in specular geometry
at an angle of incidence of 55° with respect to the surface
normal. The primary electron energy was set to 3 eV and the
energy resolution was better than 2.5 meV.
Figure 2. Thermal desorption spectra of NH3 (mass 17) for various
NH3 exposures at 90 K on the stoichiometric RuO2(110) surface. The
heating rate was 3 K s-1. The exposure is given in units of langmuir.
The substrate, a Ru(0001) single crystal, was clamped
between two Ta wires. A NiCr-Ni thermocouple was spot-
welded to the back of the Ru crystal. The sample temperature
could be varied between 85 and ∼1300 K by combining cooling
with liquid nitrogen and heating by radiation or by combined
radiation and electron bombardment from the backside of the
sample. The RuO2(110) surface was prepared in-situ following
a procedure described in preceding publications.5,6 In brief, the
Ru crystal was cleaned by applying repeated sputtering and
annealing cycles. The oxide film was then grown epitaxially
by exposing the clean Ru(0001) surface to about 107 langmuir
of O2 (1 langmuir ) 1.33 × 10-6 mbar.s) at 700 K using a gas
shower. The preparation could be repeated after restoring the
original Ru(0001) surface by sputtering and annealing cycles.
transformed into turnover frequencies, i.e., number of reactive
events per active site and second.
3. Results and Discussion
3.1. Adsorption of NH3 on the Stoichiometric RuO2(110)
Surface. Figure 2 shows the NH3 TD spectra recorded after
exposing the stoichiometric RuO2(110) surface to varying
amounts of NH3 at 90 K. At low exposures only one desorption
peak is observed at around 460 K (denoted as R1-NH3). As the
coverage is increased, the R1 state shifts to lower temperatures
and is saturated at an exposure of 0.2 langmuir, whereas a broad
feature centered at about 355 K evolves (denoted as R2-NH3).
After further exposure new desorption peaks appear at lower
temperatures: â-NH3 at 170 K and γ-NH3 at 130 K. The latter
is not saturated with increasing NH3 exposure, indicating its
origin from multilayer desorption. In analogy to the data for
NH3 on Ru single-crystal surfaces,13-15 R-NH3 and â-NH3 are
attributed to the desorption from the first and second monolayer,
respectively.
Prior to each experiment, impurities such as H2O or CO were
desorbed and the surface ordering was improved by heating the
sample to 700 K for 1 min. The chemical cleanliness of the
surface and the surface order were controlled by standard
techniques. The following gases were admitted to the surface:
NH3 (purity 99.98%, Linde), ND3 (minimum 99 atom % for D,
Campro), 16O2 (purity 99.998%, Messer Griesheim), 18O2
(minimum 99 atom % 18O2, Isotec) and C16O (purity 99.997%,
Messer Griesheim).
Interestingly, compared with the data for Ru single-crystal
surfaces,13-15 both R-NH3 states are shifted upward by about
100 K, which indicates that the ammonia monolayer is more
strongly bound on the RuO2(110) than on the Ru(0001) surface.
This finding can be rationalized by the existence of electro-
negative O-bridge atoms on RuO2(110), causing some electron
deficiency at the Ru-cus atoms, which then interact more
strongly with the N lone pair of ammonia. By assuming a
preexponential of 10-13 s, activation energies of 32 and 43 kJ
mol-1 for the desorption of multilayer and second-layer am-
monia result, respectively. The latter value is again higher than
with elemental Ru and has presumably to be attributed to
additional hydrogen bonding between the second-layer NH3 and
O-bridge, as further suggested below from the HREEL spectra.
To check for ammonia oxidation, we have performed TDS
measurements at mass 18 (H2O), 28 (N2), and 30 (NO). No
desorption of NO is detected. The TD spectrum of N2 exhibits
only a weak peak at 160 K which is associated with the
adsorption of molecular nitrogen obtained most likely due to
decomposition of NH3 in the dosing line as reported in the
literature.14,16 These results clearly demonstrate that ammonia
undergoes only molecular desorption and oxidation does not
take place on the stoichiometric RuO2(110) surface.
Steady-state rate measurements were performed by operating
the upper chamber as a continuous-flow reactor. A tube of a
diameter of 5 mm attached to our mass spectrometer, which
reached about 0.5 mm to one side of our sample and about 3
mm to the other side. By this arrangement we basically collected
reaction products coming directly from the sample surface and
also provided, by the tilt of the sample with respect to the axis
of the mass spectrometer, access of ambient molecules to the
sample surface. By this arrangement, reactions other than from
the front surface of the sample did not play a role. The later
conclusion was also tested by additional experiments.
Reactants (NH3 and O2) were introduced through separate
leak valves and were continuously pumped off together with
the reaction products by a turbomolecular pump. The partial
pressures of the products pi were transformed into the respective
rates of formation through
ri ) piNoV/τARTg (molecules/s)
(1)
where No is Avogadro’s number, V is the volume of the reactor,
τA is the residence time in the reactor as evaluated from the
effective pumping speed, R is the gas constant, and Tg is the
gas temperature. As will be shown, catalytic reaction under the
chosen conditions proceeds predominantly on the Ru-cus sites.
The RuO2(110) single-crystal surface exhibits 5 × 1014 Ru-cus
sites/cm2, so that with the known surface area (0.3 cm2) the
absolute rate values evaluated according to eq 1 can be
Spurious amounts of H2O are observed in the temperature
range 500-700 K, indicative for the recombination of OH-
bridge.17 The formation of the OH-bridge is not due to
interaction with NH3 but rather to the interaction of hydrogen
from the residual gas with the O-bridge, because the peak

Oxidation of Ammonia on RuO2(110)
J. Phys. Chem. B, Vol. 109, No. 16, 2005 7885
Figure 3. HREEL spectra for various NH3 exposures on the stoichiometric RuO2(110) surface at 90 K. All spectra were recorded at 90 K in
specular geometry with an incidence angle of 55° with respect to the surface normal and with a primary energy of 3 eV. The factors ×100 and
×400 are relative to the elastic intensity. The exposure is given in units of langmuir.
intensity is independent of NH3 exposure. From our recent
measurements10-12,18 it is known that adsorption of hydrogen
from the residual gas cannot be prevented for the reactive RuO2-
(110) surface even at a background pressure of 2 × 10-11 mbar.
In addition, there is a weak H2O desorption peak at 420 K that
is attributed to adsorption on Ru-cus from the residual gas
atmosphere.19
Figure 3 shows a series of HREEL spectra recorded after
NH3 exposure at 90 K. The stoichiometric RuO2(110) surface
is characterized by an intense band at 69 meV from the O-bridge
stretching mode against the surface,9 and phonon contributions
at 12 and 45 meV. After an exposure of 0.02 langmuir of NH3
the spectrum exhibits a new peak at 149 meV, which becomes
more intense and shifts slightly downward in energy with
increasing NH3 exposure. This feature is attributed to the NH3
umbrella mode δs(NH3) and indicative for molecular adsorption.
After an exposure of 0.1 langmuir, additional NH3-related modes
appear at 199, 403, and 418 meV arising from the asymmetric
deformation mode δa(NH3), the symmetric stretching mode δa-
(NH3), and the asymmetric stretching mode δa(NH3), respec-
tively. Beyond 0.2 langmuir exposure the surface is saturated
with ammonia in the first monolayer.
The HREEL spectra exhibit additional features at 25, 57, 76,
and 83 meV, which cannot be attributed to the frustrated
translational and rotational modes of chemisorbed ammonia.
This conclusion is supported by the isotope substitution experi-
ment with ND3. New peaks are observed at 112.5, 145, 295,
and 312.5 meV, which display the expected isotope shifts with
respect to NH3. However, the peaks at 25, 57, 76, 83, and 433
meV remain unchanged, which means that they originate not
from NH3. Actually, they are indicative for a water-like species
(H2O-bridge) formed through interaction of hydrogen with
O-bridge.17 The lack of the D2O-related modes in the HREEL
spectrum of ND3 demonstrate that hydrogen originates from the
background instead from surface reaction of NH3, in line with
the TDS results.
Increasing the NH3 exposure to 0.3 langmuir leads to the
buildup of the second layer, characterized by a new peak at
150 meV for the umbrella mode and an intense loss at 54 meV
for a librational mode. Exposure to 2 langmuirs of NH3 causes
the formation of multilayers. By comparison with literature data,
the observed features can be assigned to the frustrated translation
perpendicular to the surface (30 meV), librations (51, 57 meV),
symmetrical and asymmetrical deformations (153, 203 meV),
as well as the symmetrical and asymmetrical N-H stretching
modes (405, 422 meV). Hydrogen bonds are formed between
NH3 molecules in the multilayer as reflected by the broadening
of the symmetrical stretching mode at 405 meV.
To determine the adsorption site of ammonia, we have
performed the following experiment: The Ru-cus atoms were
first capped by CO-cus through exposure to 1 langmuir of CO
at 90 K, giving rise to the ν(Ru-CO) mode at 38 and the ν-
(C-O) mode at 262 meV (Figure 4).5 When this surface was
further exposed to 0.2 langmuir of NH3 at 90 K, no monolayer
ammonia but only second-layer type ammonia was detected with
vibrational losses at 47, 53, 133, 201 and 420 meV. This
conclusion is supported by the corresponding TD spectrum
shown in the insert exhibiting merely the â-state. This demon-
strates that NH3 chemisorbs at Ru-cus and also suggests that
the second-layer ammonia can interact with the coordinatively
unsaturated O-bridge atoms presumably via NH‚‚‚O hydrogen
bonds.
To further confirm the presented picture with the stoichio-
metric RuO2(110) surface, the sample was exposed to 2
langmuirs of NH3 at 90 K, and subsequently annealed to
different temperatures. The corresponding HREEL spectra are
shown in Figure 5. After annealing at 140 K, ammonia adsorbed
in multilayers beyond the second layer has been desorbed. The
two umbrella modes δs(NH3) at 145 and 151 meV are indicative
for bilayer ammonia, with two intense losses at 30 and 54 meV
for the frustrated translational and librational modes, respec-
tively. Annealing to 210 K leads to desorption of the second-
layer ammonia, and the spectrum exhibits now the vibrational
features of (monolayer) ammonia chemisorbed on Ru-cus. The
asymmetrical deformation mode δas(NH3) shifts to 196 meV.
Chemisorbed NH3 remains unaffected in its molecular state
during further annealing and desorbs completely upon heating
to 470 K.
In this experimentsapart from the spectral features charac-
teristic for adsorbed ammoniasno other surface species could
be identified. Thus, the appearance of two desorption states (R1
and R2) in TDS has to be attributed to the repulsive interaction
between chemisorbed ammonia molecules. Finally, we note that
the H2O-bridge related losses were detected again at 27, 58,
76, 84, and 434 meV due to spurious adsorption of hydrogen
from the residual gas. After heating to 470 K, the decomposition

7886 J. Phys. Chem. B, Vol. 109, No. 16, 2005
Wang et al.
Figure 4. HREEL spectra recorded after the stoichiometric RuO2(110) surface was exposed to (a) 1 langmuir of CO at 90 K and then (b) 0.2
langmuir of NH3 at 90 K. The corresponding TD spectrum of NH3 (mass 17) is shown in the inset. Parameters are as for Figure 3.
Figure 5. HREEL spectra recorded after exposing the stoichiometric RuO2(110) surface to 2 langmuirs of NH3 at 90 K and subsequently annealed
to the indicated temperatures. Parameters as for Figure 3.
of the H2O-bridge species has taken place, as indicated by the
appearance of OH related peaks at 54 and 446 meV.
an exposure of 1 langmuir, which corresponds to 80% of Ru-
cus sites being occupied by O atoms.9
Afterward the sample
was cooled to 90 K and then exposed to 0.2 langmuir of NH3.
The resulting TD spectra for NH3, N2, NO, and H2O are
reproduced in Figures 6a-d, respectively. No other N-containing
products such as N2O and NO2 were detected for all O-
coverages.
In summary, on the stoichiometric RuO2(110) surface NH3
chemisorbs molecularly on Ru-cus and desorbs completely
below 500 K. Compared to the Ru surfaces, ammonia is more
strongly bound to the RuO2(110) surface due to the enhanced
electron donation from the N lone pair of NH3 to the Ru-cus
atoms of the substrate. The relatively high desorption barrier
for the second-layer ammonia is attributed to additional
hydrogen bond interaction with O-bridge atoms. There is no
indication for dissociation followed by further reaction.
Desorbing NH3 shows up in two desorption peaks at about
430 and 190 K corresponding to chemisorbed molecules in the
first (R) and second (â) layer in agreement with the findings
for the stoichiometric surface as discussed above. The amount
of chemisorbed NH3 is significantly decreased compared with
the stoichiometric RuO2(110) surface. This may have two
reasons: On the O-rich surface fewer unoccupied Ru-cus sites
are available for NH3 chemisorption. Even more important,
3.2. Interaction of Ammonia with the Oxygen-Rich RuO2-
(110) Surface. Oxygen-rich RuO2(110) surfaces were prepared
by exposing the stoichiometric RuO2(110) surface to O2 at room
temperature. The maximum coverage of O-cus is achieved for

Oxidation of Ammonia on RuO2(110)
J. Phys. Chem. B, Vol. 109, No. 16, 2005 7887
Figure 6. TD spectra recorded after exposing 0.2 langmuir of NH3 to RuO2(110) surfaces with various preexposures to O2 at 90 K. The data are
for the following species: (a) NH3 (mass 16), (b) N2 (mass 28), (c) NO (mass 30), and (d) H2O (mass 18). The heating rate was 3 K s-1. The
exposures are given in units of langmuir.
however, adsorbed NH3 will react with neighboring O-cus
atoms, as will become evident later. No desorption of first-layer
NH3 is observed after an exposure of 0.5 langmuir of O2 at 300
K, although even ∼35% Ru-cus sites are still unoccupied. (Note
that the weak feature at 400 K is due to the cracking of H2O-
cus, as discussed below.) Instead NH3 molecules come increas-
ingly off from second-layer type sites (â). If compared with
the stoichiometric RuO2(110) surface, the desorption peak of
the second-layer ammonia shifts upward to 190 K, indicating
stronger interaction with the underlying layer, most likely due
to additional NH‚‚‚O hydrogen-bond attraction.
The observation of N2 (Figure 6b) and NO (Figure 6c)
desorption together with the absence of the O2 from O-cus
recombination confirms that surface reaction of adsorbed
ammonia has taken place. The high temperature of the N2 peak
at about 420 K suggests that it originates from N atom
recombination whereas the weak broad peak in the temperature
range 140-240 K is attributed to spurious effects from
coadsorbed molecular species. On the other hand, NO is
desorbed at 505 K as a result of recombination between adsorbed
N and O. The peak temperature is consistent with the data for
NO adsorption on Ru-cus.18 This means that the NO release
into the gas phase is desorption- instead of reaction-limited. NO
formation proceeds indeed already at lower temperatures, as
will be confirmed by the HREELS data below.
much higher intensitysas compared with the TD spectra from
the stoichiometric surfacesit is concluded that this H2O-cus is
produced through ammonia oxidation even if it might contain
some contribution from the interaction of O-cus with the
background hydrogen. At lower O-cus coverage a weak H2O
peak in the temperature range between 530 and 700 K is
detected caused by recombination of OH-bridge formed via
reaction of O-bridge with background H2.17 However, at
saturation only the desorption of H2O-cus is observed which
clearly confirms that O-cus instead of O-bridge serve as the
active species for ammonia oxidation.
Detailed inspection of the TDS data indicate that for a given
NH3 exposure the amounts of N2 and NO desorbing depend on
the O-cus coverage: At low O-cus coverage the formation of
N2 is observed, whereas with increasing O-cus concentration
the N2 production is suppressed and instead the NO production
increases. To quantitatively analyze the dependence of reaction
products on the O-cus coverage, the amounts of N2 and NO
released were determined by normalizing the desorption peaks
with those obtained after saturation of N2 at 90 K and NO at
300 K.18 The results are shown in Figure 7 and will be reflected
also by the steady-state kinetic data to be presented below.
Obviously, low O-cus coverage favors N2 production, which
takes place via the following overall reaction:
2NH3 + 3O f N2 + 3H2O
(2)
The occurrence of surface reactions is further manifested by
the observation of an H2O peak at 400 K (Figure 6d), which is
characteristic for H2O adsorbed on the Ru-cus sites.19 From the
whereas for high O-cus coverage NO is preferentially formed

7888 J. Phys. Chem. B, Vol. 109, No. 16, 2005
Wang et al.
was prepared by exposing the stoichiometric surface to 1
langmuir of O2 at room temperature, corresponding to an O-cus
coverage of about 80%. The sample was then exposed to 0.2
langmuir of NH3 at 90 K, and subsequently annealed to the
temperatures indicated in Figure 8. The respective HREEL
spectra were measured at 90 K. To unambiguously identify the
reaction intermediates and products, in addition to HREEL
spectra, obtained after exposing the 18O-saturated RuO2(110)
surface to 0.2 langmuir of ND3 at 90 K, were recorded and are
presented in Figure 9. The observed vibrational energies, mode
assignments, and isotope shifts are summarized in Table 1.
The O-saturated surface is characterized in Figure 8 by the
peak at 103 meV (99 meV for 18O-cus in Figure 9), which is
assigned to the stretching mode of O-cus against Ru. After
exposure of 0.2 langmuir of NH3 at 90 K, additional modes
occur that are characteristic for bilayers of adsorbed ammonia
in which the second-layer NH3 is weakly bonded to NH3
molecules underneath and/or to oxygen species (O-bridge and
O-cus) with hydrogen bonds. On the basis of isotope substitution
experiments (see Figure 9) these spectral features are assigned
to the frustrated translation mode (28 meV), the frustrated
rotation modes (40, 58 meV), the umbrella mode (151 meV),
the asymmetric deformation modes (199 meV, first-layer NH3;
207 meV, second-layer NH3), and the symmetric (404 meV)
and asymmetric (420 meV) stretching modes, respectively.
Figure 7. Iintegrated amounts of N2 and NO in units of monolayers
released in the experiments underlying Figure 6 as a function of the
O-cus coverage.
according to
2NH3 + 5O f 2NO + 3H2O
(3)
If one assumes optimum occupation of the cus sites, we can
estimate the maximum amount of N2 formed according to route
(2), taking into account the overall stoichiometry. It is reached
at an O-cus coverage of 0.6 monolayer, leading to formation
of 0.2 monolayer of N2. From our experiments a maximum N2
amount of 0.15 monolayer is achieved at an O-cus coverage of
0.66 monolayer, accompanied by 0.05 monolayer of NO. From
these numbers we can derive the total amount of O-cus reacted
with NH3 according to both routes under these conditions as
0.52 monolayer. The remaining amount of 0.14 monolayer of
O-cus is most likely consumed by reaction with hydrogen from
the residual gas to H2O, because no O2 desorption is observed
in the TD spectra.
At O-cus saturation the amount of N2 is decreased to only
0.03 monolayer, whereas NO becomes the main product
according to route (3) with a maximum of 0.21 monolayer. At
saturation of O-cus about 20% Ru-cus sites are still available
for NH3 adsorption, which would allow a maximum NO
production of 0.2 monolayer in good agreement with the
experimental value.
According to our previous investigation of NH and NH2
species on Ru single-crystal surfaces,14,20 the peaks observed
at 172 and 186 meV are characteristic for NH2 as reaction
intermediate and are assigned to the rocking and scissoring mode
of this intermediate, respectively. The observation of NH2
together with the finding that the O-cus peak shifts slightly to
101 meV and loses significantly in intensity reveals that
ammonia oxidation starts to occur at temperatures as low as 90
K. This conclusion is further supported by the detection of a
small amount of H2O-cus, monitored by the O-H stretching
mode at 436 meV and the libration mode at 113 meV.19
By warming the sample to 200 K (see Figure 8), we observe
new losses at 61, 110, 127, and 446 meV, which are indicative
for H2O-cus,19 accompanied by a further decrease of the O-cus
peak intensity at 101 meV. In addition, the peaks attributed to
HREELS experiments provided further insight into the
reaction mechanism. First, the O-saturated RuO2(110) surface
Figure 8. HREEL spectra recorded after exposing the O-cus saturated RuO2(110) surface (prepared by dosing 1 langmuir of O2 at room temperature)
to 0.2 langmuir of NH3 at 90 K and subsequent annealing to the indicated temperatures. Parameters as for Figure 3.

Oxidation of Ammonia on RuO2(110)
J. Phys. Chem. B, Vol. 109, No. 16, 2005 7889
Figure 9. HREEL spectra recorded after exposing the 18O-saturated RuO2(110) surface (prepared by dosing 1 langmuir of 18O2 at room temperature)
to 0.2 langmuir of ND3 at 90 K and subsequent annealing to the indicated temperatures. Parameters as for Figure 3.
TABLE 1: Mode Assignments, Vibrational Energies in Units
of meV, and Isotope Shifts for Different Species during NH3
(ND3) Oxidation at the 16O (18O)-Saturated RuO2(110)
Surfacea
isotope-substitution experiments (see Figure 9) these are at-
tributed to the frustrated translations parallel to the surface (14
and 28 meV), the frustrated translation perpendicular to the
surface (61 meV), the frustrated rotation modes (110 and 127
meV), the scissoring mode (192 meV), and the symmetric (436
meV) and asymmetric (446 meV) stretching modes, respectively.
(iii) A weak peak at 225 meV is attributed to the internal N-O
stretching mode indicative for the formation of NO on Ru-cus18
and supported by the expected isotope shift for N18O to 221
meV.
species
T|
T
R
δs
δa
νs
νa
NH3
ND3
154
117
199
144
404
292
420
311
(1.32) (1.38) (1.38) (1.35)
NH2
ND2
172
127
186
140
(1.35) (1.33)
H216O 15
28
61
59
98
192
436
312
446
327
Furthermore, a weak feature at 52 meV is resolved, which
exhibits an isotope shift of 1.04 for ND3. By comparison with
the Ru-OHbridge stretching mode, observed at 64 meV,17 it is
suggested that this peak arises from an OH group at Ru-cus as
a reaction intermediate of NH3 oxidation. On Ru single-crystal
surfaces14,15,20,22 the Ru-N stretching mode has been detected
in the energy region between 50 and 80 meV. With the present
system unfortunately no clear assignment can be made due to
overlap with various other spectral features. The NH group,
which might be formed during NH3 oxidation and exhibitss
according to our reassignment14,20sa characteristic bending
mode in the 80-90 meV energy region, can also not be
identified.
110
127
73
D218O 14
143
(1.07) (1.03) (1.34) (1.34)
(1.40) (1.36)
27
(1.04)
82
(1.34)
94
(1.35)
16OH
18OD
52
50
(1.04)
N16O
N18O
231
226
(1.02)
a T, translational mode; | and to the surface plane; R, librational
mode; δs and δa symmetric and asymmetric deformation modes; νs and
νa, symmetric and asymmetric stretching modes.
Upon further annealing to 320 K, the following changes are
observed: (i) The NH3-related peaks disappear nearly com-
pletely, indicating that chemisorbed NH3 has been reacted off.
NH3 desorption from the (chemisorbed) R-state can, however,
be ruled out on the basis of the TDS results (see Figure 6a). (ii)
The H2O-cus related peaks grow. An additional loss is resolved
at 98 meV and assigned to a libration mode of H2O-cus, as
supported by the isotope shift of 1.34. (iii) The NO stretching
band shifts upward to 231 meV and gains in intensity. Its blue
shift by 6 meVswith respect to the value found at 250 Ksis
associated with the disappearance of the NH3 species. At lower
temperature the electron density of the substrate is increased
due to electron donation from the N lone pair of the chemisorbed
ammonia to the surface. As a consequence, electron back-
donation from the surface to the antibonding 2π* orbital of
coadsorbed NO is enhanced, resulting in weakening of the
internal N-O bond.
the NH3 bilayer are no longer observed in accordance with the
desorption of the second-layer NH3 (Figure 6a); the spectrum
displays now the δs(NH3) mode at 154 meV and the δa(NH3)
mode at 199 meV for NH3 chemisorbed on Ru-cus. Interestingly,
compared with the stoichiometric RuO2(110) surface, the δs-
(NH3) mode shifts upward by 10 meV on the O-rich surface.
This finding is attributed to a change of the charge at the surface:
21 A more positive charge at the surface leads to a higher mode
frequency. On the O-rich RuO2(110} surface the electron density
of the substrate is reduced by the electronegative O-cus adatoms
leading to a more positive charge at the unoccupied Ru-cus sites.
After heating to 250 K, the HREEL spectrum changes
again: (i) The NH2 related modes vanish, which reveals that
NH2 is not stable at higher temperatures. (ii) H2O-cus related
losses gain in intensity and are well resolved. On the basis of

7890 J. Phys. Chem. B, Vol. 109, No. 16, 2005
Wang et al.
Figure 10. HREEL spectra recorded after exposing the RuO2(110) surface partly covered by O-cus (prepared by dosing 0.3 langmuir of O2 at
room temperature) to 0.2 langmuir of NH3 at 90 K and subsequent annealing to the indicated temperatures. Parameters as for Figure 3.
Heating to 400 K leads to the disappearance of the H2O-cus
related modes, whereas the NO-cus mode at 231 meV becomes
the leading one (note the change in the sensitivity factor); i.e.,
the main amount of NO-cus is produced only when the H2O-
cus species desorbs completely at 400 K. Besides NO, the
spectrum does not exhibit other N-containing products, in good
agreement with the TDS results. By annealing to 530 K, NO
disappears due to the desorption of NO at 510 K (Figure 6c),
and the spectrum (not shown) displays again the dominant peak
at 69 meV characteristic for the stoichiometric RuO2(110)
surface.
The preferential reaction of ammonia to N2 and H2O at low
O-cus coverage becomes evident from the HREEL data
reproduced in Figure 10. A surface partially (i.e., to about 40%)
covered by O-cus was prepared by dosing 0.3 langmuir of O2
at room temperature.10 Subsequent exposure to 0.2 langmuir of
NH3 at 90 K causes the appearance of new peaks at 152, 199,
406, and 420 meV due to NH3-cus. Simultaneously, the O-cus
related feature at 103 meV is strongly reduced in intensity,
indicating that NH3 oxidation starts already at these low
temperatures. This conclusion is further supported by the
observation of the intermediate species NH2 (186 meV) and
OH (52 meV) as well as of the product H2O-cus (112, 433,
and 445 meV). (In addition, a small amount of H2O-bridge is
also formed as indicated by the scissoring mode at 221 meV
originating from interaction with hydrogen from residual gas.)
Upon annealing to 320 K, O-cus is transformed into H2O-
cus (61, 98, 107, 128, 433, and 446), whereas there are NH3
species left on the surface (152 and 199 meV), in line with the
observation of NH3 desorption in TDS (Figure 6a). At this
temperature, the weak feature observed at 225 meV cannot be
attributed to the H2O-bridge any more, because the H2O-bridge
is not stable and dissociated completely to OH-bridge at 320.17
We attribute therefore this peak to minor amounts of NO-cus
formed.
a red shift by 5 meV, reflecting the change of the electron
density at the surface when H2O-cus is desorbed. Most
importantly, the NO stretching mode at 225 meV remains
unchanged with very small intensity during heating to 400 K,
which reveals that at low O-cus coverage NO formation is
almost negligible. This points indirectly to N2 being the main
reaction product formed by recombination of the two neighbor-
ing N-cus species, as supported by the TDS results. The release
of N2 into the gas phase is limited by this recombination step,
because adsorbed N2 leaves the surface below 200 K.18
3.3. Steady-State Reaction Kinetics. The results presented
so far demonstrated that with the stoichiometric RuO2(110)
surface NH3 undergoes simply adsorption and desorption,
whereas in the presence of excess oxygen (occupying Ru-cus
sites) reaction to both N2 and NO may take place. From these
N2 comes off the surface around 420 K (by recombination of
N-cus), whereas NO is released (desorption-limited) at about
500 K. Therefore steady-state rate measurements were only
performed above the latter temperature.
The reaction rate r ()turnover frequency) will depend on
three parameters, the temperature T and the partial pressures of
NH3 and O2, p(NH3) and p(O2), respectively. Figure 11a shows
the results from experiments with fixed T ) 530 K and p(NH3)
) 1 × 10-7 mbar as a function of p(O2). For p(O2) ) 0; i.e.,
without occupation of the Ru-cus sites by O-atoms, rNO as well
as rN vanish, confirming the previous conclusions whereafter
2
both, NH3 and O, have to be adsorbed on Ru-cus to undergo
catalytic reaction. With increasing p(O2) first rN rises, passes
2
through a maximum and then decreases again, whereas rNO has
a sigmoid shape and rises continuously. With p(NH3) ) 10-7
mbar the impact frequency of an incoming NH3 molecule onto
a Ru-cus site will be about 0.1 s-1, and from Figure 11a one
can read the value of the turnover frequency for NO formation
to be 0.025 s-1 at p(O2) ) 2 × 10-6 mbar, so that we may
conclude that for the highest O2 pressure applied in this
experiment (2 × 10-6 mbar) an ammonia molecule has a
probability of about 25% to be transformed into NO.
Further annealing to 400 K leads to several changes: (i) The
H2O-cus related losses vanish completely due to the desorption
of H2O-cus. (ii) The new losses at 54 and 446 meV are assigned
to OH-bridge formed via the decomposition of the H2O-bridge.17
(iii) The umbrella mode of NH3 is observed at 147 meV with
The switching of the dominant reaction product from N2 to
NO with increasing O2 pressure becomes even more evident
when the kinetic data of Figure 11a are transformed into

Oxidation of Ammonia on RuO2(110)
J. Phys. Chem. B, Vol. 109, No. 16, 2005 7891
rates are lower and the selectivity for N2 formation is favored,
whereas that for NO production reaches a maximum value of
only about 0.8 (Figure 12b).
No full exploration of the full parameter space (T, pO , pNH
)
2
3
could be performed. Qualitatively the following conclusions
were drawn: At fixed partial pressures of NH3 and O2, both
the rates and selectivities remain relatively independent of
temperature between about 530 and 630 K. Below 530 K the
rate for NO formation decreases markedly (because of limiting
thermal desorption of NO), whereas that for N2 production
remains still fairly constant. Above 630 K both rates decrease
continuously, presumably mainly because the steady-state
surface concentrations of the reacting species decrease due to
thermal desorption.
However, at higher temperatures another reaction step has
to be taken into account: Reaction between NH3 and the bridge
(or even bulk) O atoms of the RuO2(110) surface is associated
with a higher activation energy than that with O-cus, so that
these processes will no longer be negligible at higher temper-
atures. HREELS data demonstrated indeed that the RuO2(110)
surface may even become partly reduced to Ru(0001) (covered
by chemisorbed O atoms), which phase is catalytically com-
pletely inactive.
4. Discussion
Figure 11. (a) Steady-state rates of N2 and NO formation (expressed
as turnover frequencies) for a constant NH3 partial pressure, p(NH3) )
1 × 10-7 mbar, as a function of O2 partial pressure p(O2) at T ) 530
K. (b) Steady-state selectivities for N2 and NO formation resulting from
the data of the Figure 11a.
Because the properties of the various surface species have
been analyzed already in detail in the preceding section, the
following discussion will concentrate on the mechanism and
kinetics of the catalytic reactions leading to N2 or NO, whereby
particular emphasis will be put on the phenomenon of catalytic
selectivity.
Under technical conditions, oxidation of ammonia by atmo-
spheric oxygen is typically carried out with Pt/Rh gauzes
operated at 1100-1200 K providing NO yields of up to 98%.1,2
The product selectivity was found to depend mainly on
temperature: At low temperatures N2 is the main product,
whereas above 850 K NO becomes increasingly dominant. In
view of the extremely short contact times (∼10-3 s) possible
surface intermediates were hard to detect and discussions about
the reaction mechanism remained largely speculative.23 This
situation remained essentially unchanged in later studies with
various metal24,25 or oxide24-29 surfaces. More recently, surface
science experiments provided more insight.30-35 For Pt it was
found that the selectivity was not only governed by temperature
but also by the composition of the gas phase, in that excess
oxygen favors the production of NO (as with the present
system).33-35 Bradley et al.34 proposed a reaction mechanism
consisting of a complex series of steps that contains competition
between NO desorption and NO dissociation as essential
ingredients. However, so far no reports about the steady-state
kinetics and its modeling are available for this type of study
with metal surfaces.
With the RuO2(110) surface the Ru-cus atoms can unequivo-
cally be identified as the active sites for catalytic reaction for
the following reasons: (i) NH3 chemisorbed in the first layer
will be attached to Ru-cus. (ii) Reaction will only take place if
also O atoms are coadsorbed on Ru-cus. (iii) Only H2O
molecules adsorbed on Ru-cus are detected as reaction products.
If H2O-bridge or OH-bridge groups would have been formed
through interaction between NH3-cus and O-bridge, the corre-
sponding spectral features would be thermally stable up to 600
K,17 which is not the case. (iv) The NO molecules formed are
also adsorbed on Ru-cus.18 Their formation via O-bridge can
be excluded on the basis of the isotope substitution experiments.
Figure 12. Data as with Figure 11, but for T ) 500 K.
selectivity, namely the rates for N2 and NO divided by the total
rates, respectively, viz. SN ) (rN )/(rN + rNO) and SNO ) (rNO)/
2
2
2
(rN + rNO). As can be seen from Figure 11b the selectivity for
2
NO production rises from zero to almost 100%, whereas that
for N2 formation, on the other hand, decreases continuously.
Analogous kinetic data for a somewhat lower temperature
(500 K) are reproduced in Figure 12a. The qualitative behavior
is similar to that of the data of Figure 11. However, the absolute

7892 J. Phys. Chem. B, Vol. 109, No. 16, 2005
Wang et al.
the probability for recombination to NO will continuously rise
while that for N2 formation drops. This model provides a
straightforward rationalization for the observed variations of
reactivity and selectivity.
To further substantiate this concept an attempt was made for
mathematical modeling. For this purpose the reaction scheme
depicted in Figure 14 is approximated by the following steps:
k1
Figure 13. Sketch of the RuO2(110) surface along the Ru-cus atoms
(open circles) in [001] direction with O-cus and NH3-cus adsorbed on
neighboring sites, illustrating H-bridge interaction with O-cus as primary
reaction step.
O2 + 2*
98 2Oad
(4)
(5)
k2
NH3 + * 8 NH3,ad
k-2
k3
NH3,ad + Oad
9
8 Nad + H2O v + *
(6)
(7)
(8)
k4
Nad + Oad
Nad + Nad
98 NO v + 2*
k5
9
8 N2 v + 2*
* denotes a free Ru-cus site, and the overall surface composition
is determined by
Figure 14. Overall scheme for the various surface processes and their
characteristic reaction temperatures.
[Oad] + [NH3,ad] + [Nad] + [*] ) 1
(9)
The primary reaction step is illustrated in Figure 13, which
shows a cut through the surface in [001] direction along the
Ru-cus atoms. NH3 adsorbed on Ru-cus will desorb unless it
interacts with a neighboring O-cus species. That is why no
reaction at all is observed with the stoichiometric (i.e., O-cus
free) RuO2(110) surface. However, there exists obviously a fairly
strong interaction between an H-atom from NH3 and a neigh-
boring O-cus atom as sketched by the dotted line. Hydrogen
abstraction and formation of OH-cus takes place at temperatures
as low as 90 K, as demonstrated by the HREELS data. The
further progress of the surface reactions is depicted schematically
by Figure 14 whereby no account of the exact stoichiometry is
taken for the sake of simplicity.
Although NHad as a possible further intermediate cannot be
ruled out and Nad cannot be directly identified spectroscopically,
there is very strong evidence that this species plays the key
role in the formation of the reaction products. Depending on
the configuration of neighboring adspecies, either recombination
with another Nad to N2 (which then immediately desorbs) or
with Oad to adsorbed NO will take place. Thermal activation of
these two competing processes is presumably not very different
and also affected by desorption of H2O because these steps all
take place in the same temperature range. NOad can clearly be
identified by HRELSS and desorbs only around 500 K, so that
at lower temperatures this process will be limiting the overall
rate. There is also no indication for dissociation of adsorbed
NO at elevated temperatures,18 which was considered to be the
source for Nad in the reaction over Pt surfaces.34,35
The rates for NO and N2 production are then given by
d[NO]
dt
τNO
)
) k4[Nad][Oad]
(10)
(11)
d[N2]
τN
)
) k5[Nad]2
2
dt
Under steady-state flow conditions the relevant surface con-
centrations are determined within the Langmuir-Hinshelwood
(mean field) approximation by
d[Oad]
) k1po2(1 - [Oad] - [Nad] - [NH3,ad])2 -
dt
k3[NH3,ad][Oad] - k4[Nad][Oad] ) 0 (12)
d[Nad]
) k3[NH3,ad][Oad] - k4[Nad][Oad] - k5[Nad]2 ) 0
dt
(13)
d[NH3,ad
]
) k2pPH (1 - [Oad] - [Nad] - [NH3,ad]) -
3
dt
k-2[NH3,ad] - k3[NH3,ad][Oad] ) 0 (14)
This is admittedly a very crude approximation in which all
additional surface intermediates are neglected, the stoichiometry
is wrong, and the kinetics is described by the Langmuir concept
ignoring any interactions between the varying adsorbed species.
In addition, the rate constants k1 to k5 are not known, so that at
best reproduction of the qualitative trends can be expected. On
the other hand, solution of the set of equations (10)-(14) is by
no means trivial.
As an example for the kind of agreement between experi-
mental results and modeling that can be achieved in this way,
Figure 15 shows a fit for the data at 500 K as reproduced in
Figure 12. This result was obtained with plausible values for
the rate constants k1 to k5 (e.g., the initial sticking coefficients
for O2 and NH3scorresponding to k1 and k2sbeing close to
unity) but is, of course, still associated with ambiguities. The
Although not all details of this reaction scheme are yet known,
it provides a safe framework also for qualitative rationalization
of the steady-state rate data. Inspection of Figures 11a and 12a
reveals that the reactions to formation of both products take
off only if there is continuous admission of O2 forming O-cus,
so that ammonia may become dehydrogenated and Nad is
formed. As long as most O-cus is consumed for this process,
recombination of Nad with Oad will be suppressed and the
formation of N2 will be dominant. Only with increasing O2
pressure will NO formation take off. A further increase of the
Oad concentration (by further increase of pO ) will reduce the
2
probability that an Nad species is next to another Nad. Instead,

Oxidation of Ammonia on RuO2(110)
J. Phys. Chem. B, Vol. 109, No. 16, 2005 7893
by the relative partial pressures of the reactants. The results also
clearly demonstrate the dependence of the selectivity on the
external parameters as was also suggested by King et al.34,35
for the reaction on Pt surfaces. Almost 100% selectivity for
NO formation is reached at 530 K, much lower than the
temperatures applied in the technical Ostwald process with
platinum-based catalysts (>1100 K). This offers interesting
prospects for possible applications.
References and Notes
(1) Satterfield, C. N. Heterogeneous catalysis in Industrial Practice;
McGraw-Hill Inc.: New York, 1991; p 312.
(2) Wagner, E.; Fetzer, T. In Handbook of heterogeneous catalysis;
Ertl, G., Kno¨zinger, H., Weitkamp, J., Eds.; Wiley-VCH: New York, 1997;
Vol. 4, p 1748.
(3) Gang, L.; Anderson, B. G.; van Grondelle, J.; van Santen, R. A.;
van Gennip, W. J. H.; Niemantsverdriet, J. W.; Kooyman, P. J.; Koester,
A.; Brongersma, H. H. J. Catal. 2002, 206, 60.
(4) Somorjai, G. A.; Borotko, Y. G. Catal. Lett. 2001, 76, 1.
(5) Wang, J.; Fan, C. Y.; Jacobi, K.; Ertl, G. Surf. Sci. 2001, 481, 113.
(6) Fan, C. Y.; Wang, J.; Jacobi, K.; Ertl, G. J. Chem. Phys. 2001,
114, 10058.
(7) Over, H.; Kim, Y. D.; Seitsonen, A. P.; Wendt, S.; Lundgren, E.;
Schmid, M.; Varga, P.; Morgante, A.; Ertl, G. Science 2000, 287, 1474.
(8) Wang, J.; Fan, C. Y.; Jacobi, K.; Ertl, G, J. Phys. Chem. B 2002,
106, 3422.
(9) Kim, Y. D.; Seitsonen, A. P.; Wendt, S.; Wang, J.; Fan, C.; Jacobi,
K.; Over, H.; Ertl, G. J. Phys. Chem. B 2001, 105, 3752.
(10) Lafosse, A.; Wang, Y.; Jacobi, K. J. Chem. Phys. 2002, 117, 2823.
(11) Wang, Y.; Lafosse, A.; Jacobi, K. J. Phys. Chem. B 2002, 106,
5476.
(12) Paulus, U. A.; Wang, Y.; Bonzel, H. P.; Jacobi, K.; Ertl, G. J. Phys.
Chem. B in press.
Figure 15. Experimental data (marked by stars) of Figure 12 in
comparison with a fit to the theoretical model with suitable param-
eters: k1 ) 1.0, k2 ) 1.2, k-2 ) 2.4, k3 ) 20, k4 ) 2.0, k5 ) 1.6.
(13) Benndorf, C.; Madey, T. Surf. Sci. 1983, 135, 164.
(14) Jacobi, K.; Wang, Y.; Fan, C. Y.; Dietrich, H. J. Chem. Phys. 2001,
115, 4306.
main purpose was to demonstrate that reasonable modeling can
be achieved with the proposed reaction mechanism, which
thereby receives further support.
(15) Wang, Y.; Jacobi, K. J. Phys. Chem. B, in press.
(16) Wang, Y.; Lafosse, A.; Jacobi, K. Surf. Sci. 2002, 507, 773.
(17) Wang, J.; Fan, C. Y.; Sun, Q.; Reuter, K.; Jacobi, K.; Scheffler,
M.; Ertl, G. Angew. Chem. Int. Ed 2003, 42, 2151.
(18) Wang, Y.; Jacobi, K.; Ertl, G. J. Phys. Chem. B 2003, 107, 13918.
(19) Lobo, A.; Conrad, H. Surf. Sci 2003, 523, 279.
(20) Wang, Y.; Jacobi. K. Surf. Sci. 2002, 513, 83.
(21) Parmeter, J. E.; Wang, Y.; Mullins, C. B.; Weinberg, W. H. J. Chem.
Phys. 1988, 88, 5225.
(22) Shi, H.; Jacobi, K.; Ertl, G. J. Chem. Phys. 1955, 102, 1432.
(23) Ya. M.; Fogel, B. T.; Nadykto, V. F.; Rybalko, V. I.; Shvachko,
V. I.; Korobchanskaya, I. E. Kinet. Catal. 1964, 5, 496.
(24) Carabineiro, S. A.; Nieuwenhuys, B. E. Surf. Sci. 2003, 532, 87.
(25) Van den Broek, A. C. M.; van Grondelle, J.; van Santen, R. A. J.
Catal. 1999, 185, 297.
(26) Long, R. Q.; Yang, R. T. J. Catal. 2002, 207, 158.
(27) Sil′chenkova, O. N.; Korchak, V. N.; Matyshak, V. A. Kinet Catal.
2002, 43, 363.
(28) Amblard, M.; Burch, R.; Southward, B. W. L. Appl. Catal. B 1999,
22, L159.
(29) Borello, E. E.; Coluccia, S.; Zecchina, A. J. Catal. 1994, 149, 270.
(30) Gland, J. L.; Korchak, V. N. J. Catal. 1978, 53, 9.
(31) Gland, J. L.; Woodward, G. C. J. Catal. 1980, 61, 543.
(32) Asscher, M.; Guthrie, W. L.; Lin, T. H.; Somorjai, G. A. J. Phys.
Chem. 1984, 88, 3233.
(33) Mieher, W. D.; Ho, W. Surf. Sci. 1995, 322, 151.
(34) Bradley, J. M.; Hopkinson, A.; King, D. A. J. Phys. Chem. 1995,
99, 17032.
5. Conclusions
The formation of a single reaction product in heterogeneous
catalysis is more the exception than the rule, as for example
verified with ammonia synthesis or carbon monoxide oxidation.
Hence selectivity will be frequently of even higher significance
than the overall reactivity. The present work represents, in our
opinion, the first example for which this phenomenon could be
explored on an atomic level for the resulting kinetics under
steady-state flow conditions.
The coordinatively unsaturated (cus) Ru atoms in the RuO2-
(110) surface were identified as catalytically active sites onto
which either ammonia or oxygen are adsorbed from the gas
phase. The concentration of O-atoms adsorbed on these cus-
sites determines both the reactivity and selectivity. Without the
presence of this species adsorbed NH3 simply desorbs. However,
a neighboring O-cus abstracts readily one of its H-atoms
initiating complete dissociation. The resulting adsorbed N-atoms
either recombine with each other to N2 or with O-cus to NO.
The selectivity under steady-state conditions is thus determined
by the local surface configurations of the adsorbates and thereby
(35) Kim, M.; Pratt, S. J.; King, D. A. J. Am. Chem. Soc. 2000, 122,
2409.