Kinetics of ammonia oxidation on stepped platinum surfaces. I. Experimental results

Article abstract of DOI:10.1016/j.susc.2004.12.027

The kinetics of ammonia oxidation with oxygen have been investigated in the 10^-5 and 10^-4 mbar range on Pt(5 3 3) and Pt(4 4 3). Only N₂ and NO but no N₂O were detected as reaction products. The dependence of product formation on temperature and on the partial pressures of the reactants has been studied under steady state conditions. The reactive sticking coefficients were determined under reaction conditions with s _reac of ammonia reaching nearly 0.2. The Pt(5 3 3) surface was found to be catalytically more active than Pt(4 4 3) by a factor of 2-4.

Full text of DOI:10.1016/j.susc.2004.12.027

Surface Science 577 (2005) 1–14
www.elsevier.com/locate/susc
Kinetics of ammonia oxidation on stepped platinum surfaces.
I. Experimental results
A. Scheibe, U. Lins, R. Imbihl *
Institut f u¨ r Physikalische Chemie und Elektrochemie, Universit a¨ t Hannover, Callinstr. 3-3a, 30167 Hannover, Germany
Received 12 October 2004; accepted for publication 14 December 2004
Available online 13 January 2005
Abstract
ꢀ5
ꢀ4
The kinetics of ammonia oxidation with oxygen have been investigated in the 10 and 10 mbar range on Pt(533)
and Pt(443). Only N and NO but no N O were detected as reaction products. The dependence of product formation
2
2
on temperature and on the partial pressures of the reactants has been studied under steady state conditions. The reactive
sticking coefficients were determined under reaction conditions with sreac of ammonia reaching nearly 0.2. The Pt(533)
surface was found to be catalytically more active than Pt(443) by a factor of 2–4.
ꢀ
2005 Elsevier B.V. All rights reserved.
Keywords: Platinum; Stepped platinum surface; Ammonia oxidation; Kinetics
1
. Introduction
4NH
3
+ 5O
2
! 4NO + 6H
2
O, DH = –226 kJ/mol
ðR1Þ
At lower temperature N is the preferred reaction
Catalytic ammonia oxidation with oxygen over
platinum catalysts is an important process in the
industrial production of nitric acid and in environ-
mental catalysis [1]. Nitric acid is produced via
formation of NO at T > 800 K in the so-called
Ostwald process
2
product
4NH + 3O ! 2N + 6H O, DH = –317 kJ/mol
3
2
2
2
ðR2Þ
In view of its industrial importance, compara-
tively few kinetic and mechanistic studies of cata-
lytic ammonia oxidation have been made and
only a relatively small number of single crystal
investigations have been conducted [2–15]. A prin-
ciple problem exists in extrapolating these single
*
Corresponding author. Tel.: +49 5117622349; fax: +49
5
117624009.
E-mail address: imbihl@pci.uni-hannover.de (R. Imbihl).
0
039-6028/$ - see front matter ꢀ 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2004.12.027

2
A. Scheibe et al. / Surface Science 577 (2005) 1–14
crystal results to industrial conditions since the Pt
catalysts undergo severe morphological changes
due to the strong exothermicity of the reaction.
A Pt/Rh gauze used in the Ostwald process be-
comes already visibly roughened during the first
minutes of operation [16–18]. The restructuring
of a catalyst under reaction conditions is a quite
general phenomenon in heterogeneous catalysis of-
ten associated with an activation or deactivation of
the catalyst [5,16–20]. The reaction-induced sub-
strate changes are part of the well known pressure
and material gap problem in catalytic research and
catalytic ammonia oxidation is certainly an excel-
lent example for this gap. A pressure gap also
shows up in the product formation because in all
ity of oxygen controls the selectivity towards NO
formation in the competition N vs NO formation
2
[9]. In a molecular beam study Somorjai et al.
found that NO production on Pt(111) occurs via
a fast and a slow pathway associated with activa-
tion energies 121 ± 13 kJ/mol and 59 ± 13 kJ/mol,
respectively [10]. With vibrational spectroscopy
Michev and Ho identified on Pt(111) the interme-
diates OH, NH, and NH2 [12]. The group of
King studied the reaction on Pt(100) with a
molecular beam and with vibrational spectroscopy
[13,14]. The main conclusion from these studies
was that direct ammonia dissociation according
to NH3,ad ! N + 3Had plays only a minor role
ad
in the overall activity. The main contribution
comes from ammonia being activated through di-
rect interaction with atomic oxygen or OH strip-
ping one or several H atoms from the molecule.
This conclusion is also supported by theoretical
calculations [27]. A mechanistic scheme originally
suggested by Fogel involved NO formation as pri-
mary step but this scheme found no support by
experiments [3,14,15].
studies N O formation according to
2
4
NH
3
+ 4O
2
! 2N
2
O + 6H
2
O
ðR3Þ
was only detected at high pressure (p > 1 mbar).
In order to perform a systematic study of the ef-
fect of reaction-induced restructuring and its effect
on the kinetics we chose two stepped Pt(111) sur-
faces, Pt(533) and Pt(443), as model systems for
UHV studies under low pressure conditions
Here, we study in detail the kinetics of the reac-
tion on the two surfaces under low pressure condi-
tions, where mass transport limitations play no
role and where the reaction is truly isothermal.
As reported before, already under low pressure
conditions the surface undergoes restructuring
under certain reaction conditions associated with
a change in selectivity [15]. Qualitatively, the results
can be understood in terms of a simple mechanistic
model. The interpretations are bolstered by simula-
tions with a realistic mathematical model, which
reproduces the main experimental results [28].
ꢀ3
(
p < 10 mbar). The Pt(533) surface written in
microfacet notation as 4(111) · 1(100) exhibits
mono-atomic steps with (100) orientation while
Pt(443) decomposed into 7(111) · 1(111) con-
tains mono-atomic steps with (111) orientation.
Oxygen adsorption and NO decomposition, which
both are mechanistic steps of the overall process,
are known to be highly structure sensitive on Pt
surfaces [21–24]. For this reason one would expect
that the step orientation, the step density, and a
change of the step structure by the reaction has a
profound influence on the activity and selectivity
of the catalyst. In catalytic CO oxidation on
Pt(110) it was demonstrated that even very mild
2. Experimental
ꢀ
4
reaction conditions in the 10 mbar range can
cause a roughening or (micro)facetting of the sur-
face [25,26].
The reaction was studied here in a standard
UHV system equipped with LEED, a retarding
field analyzer for Auger electron spectroscopy, a
scanning tunneling microscope (DME), and a dif-
ferentially pumped quadrupole mass spectrometer
(QMS) for rate measurements. The base pressure
Single crystal studies of the NH + O reaction
3
2
have been conducted with the orientations
Pt(100) [13,14], Pt(111) [10–12], and Pt(s)-
1
2(111) · (111) [9]. From the study of the stepped
ꢀ
10
Pt(111) surface Gland and Korchak concluded
that steps play a dominant role in determining
the reactivity of the surface and that the availabil-
in the UHV system was 2 · 10
mbar. The sam-
ples were heated indirectly by a filament behind
the backside of the crystal either via radiation or

A. Scheibe et al. / Surface Science 577 (2005) 1–14
3
via electron bombardment. They were both pre-
pared by repeated cycles of Ar-ion-sputtering,
heating in oxygen (pO2 ¼ 1 Â 10ꢀ mbar), followed
by annealing to 1300 K. Gases of purity 5.0 for
oxygen and 2.5 for ammonia were used. All pres-
sures given here are uncorrected.
background was taken from rate measurements
at 300 K assuming that the catalytic activity at
this temperature is negligible. In order to deter-
mine the absolute reaction rates the QMS was cal-
ibrated against the ionization gauge in the main
chamber.
6
During rate measurements the cone (opening
mm), which connected the differentially pumped
2
QMS with the main chamber, was brought 1 mm
in front of the surface. In this way reaction prod-
ucts from the backside of the sample and from
filaments were excluded. NO formation was fol-
lowed in QMS via m/e = 30, N2 via m/e = 2 8.
The CO contribution to the mass 28 signal was
determined from the m/e = 12signal originating
from CO fragmentation and then subtracted. The
3. Results
3.1. Temperature dependence of the kinetics
Structural models of the Pt(533) and Pt(443)
surfaces are displayed in Fig. 1, together with a
characterization of the surfaces by LEED and
scanning tunneling microscopy. The temperature
Fig. 1. Structural models of the stepped Pt(111) surfaces used in this investigation together with a characterization by scanning
tunneling microscopy (STM) and LEED. (a) Structural model of Pt(533) surface, (b) LEED pattern of Pt(533), (c) STM image of
Pt(533), (d) model of Pt(443) surface, (e) LEED pattern of Pt(443), (f) STM image of Pt(443).

4
A. Scheibe et al. / Surface Science 577 (2005) 1–14
Fig. 1 (continued)
dependence of the NO and N production rates of
2
ally small and they are probably transients, i.e.
they tend to vanish if the heating/cooling rate is
chosen sufficiently slow. As was demonstrated by
LEED and STM, the large hystereses on Pt(533)
have their origin in a reaction-induced doubling
of the step height and terrace width, which occurs
Pt(533) and Pt(443) are reproduced in Figs. 2and
for different partial pressure ratios p =p
3
of
NH3
O2
oxygen to ammonia. The data were obtained by
slowly cycling the temperature with fixed pO2 and
pNH3 , starting always with a freshly prepared sur-
face. The qualitative picture is similar irrespective
on the heating branch of a cycle [15]. The N rate
2
of orientation and partial pressure ratio. N pro-
2
curve exhibits a dip when the surface restructures.
The hysteresis before this dip appears not to be a
true hysteresis because it vanishes when the heat-
ing/cooling is carried out sufficiently slow. Under
reaction conditions this doubling of the step height
is reversible and above ꢁ650–750 K single atomic
steps are restored as indicated in the bars on top
duction is the preferred reaction channel at low
temperature but, with increasing temperature,
NO formation is favored, while N2 production
drops to a low level.
On Pt(533) quite large hysteresis effects are ob-
served while on Pt(443) the hystereses are gener-

A. Scheibe et al. / Surface Science 577 (2005) 1–14
5
Pt(533)
Pt(443)
(
a)
single step double step
single step
(
a)
3.0
2.0
N2
NO
single step
1
0:1
1
0
0
0
.2
.8
.4
.0
N2
NO
3:1
1.0
0.0
600
800
1000
6
00
800
1000
(
b)
N2
NO
1
.2
.8
(
b)
single step double step single step
single step
1
:1
0
3
:1
2.0
N2
NO
0.4
0.0
1
1
0
0
.5
.0
.5
.0
600
800
1000
(
c)
0
.6
.4
N2
NO
6
00
800
1000
1
:3.3
0
single step double step single step
single step
(
c)
0.2
0.0
2
2
1
1
0
0
.5
.0
.5
.0
.5
.0
N2
NO
1
:1
600
800
T [K]
1000
Fig. 3. Temperature dependence of the N
2
and NO production
rates on Pt(443) at varying mixing ratios p =pNH3 with fixed
O2
ꢀ4
pNH3 ¼ 1 Â 10 mbar.
A heating rate of 100 K/min was
applied. Error bars, cf. caption of Fig. 2.
6
00
800
1000
T [K]
Fig. 2. Temperature dependence of the N
2
and NO production
towards enhanced NO production [15,29]. The
reaction-induced doubling of the step height was
rates on Pt(533) at varying mixing ratios p =pNH3 with fixed
O2
ꢀ4
pO2 ¼ 1 Â 10 mbar. The arrows indicate that the rates were
measured in heating/cooling cycles with 50 K/min starting in
each case with a freshly prepared surface. The error bars are
indicated by the size of the symbols used for representing the
data points. This applies also to all other figures unless the error
bars are explicitly given. The bar on top of each diagram shows
the structural changes of the surface as determined by LEED.
ꢀ
5
not observed at p < 10 mbar, indicating that a
certain threshold in pressure exists for such struc-
tural changes to occur under reaction conditions.
This is demonstrated in Fig. 4, where a large hys-
teresis is observed at a total pressure of 2 ·
ꢀ4
10
mbar. The hysteresis, however, becomes
much smaller when the total pressure is reduced
5
ꢀ
of the diagrams in Fig. 2. On the cooling branch
the steps remain single atomic. This doubling in
step height, which, at higher temperature T P
to 6 · 10 mbar. On Pt(443) no reaction-induced
structural changes were detectable in the pressure
ꢀ6
ꢀ3
range from 10 to 10 mbar with LEED.
We note in Fig. 2 that the rate maxima for both,
1
100 K can also be induced by oxygen alone, is
associated with a significant change in selectivity
N and NO, shift with increasing ratio p =p to
NH3
2
O
2

6
A. Scheibe et al. / Surface Science 577 (2005) 1–14
Pt(533)
ture on Pt(443) as compared to Pt(533) by
approximately 100–140 K in the case of N pro-
duction and by 50–80 K in the case of NO produc-
tion. We also note that similar to Pt(533) all rate
2
single step double step
single step
(a)
single step
-5
maxima shift with decreasing ratio p =p
higher temperature but the shift is much smaller,
reaching only 30 K in the case of nitrogen and
50 K for NO production.
to
NH3
1.2
0.8
0.4
0.0
6 x 10 mbar
N2
NO
O2
3.2. Reactive sticking coefficients
6
00
800
1000
In our experimental set-up the differentially
pumped QMS is behind a cone whose tip is about
1–2mm Õs away from the sample surface. Due to
this geometric arrangement to a good approxima-
tion only molecules reflected from the surface can
enter the cone to be detected by the QMS: We can
thus determine the reactive sticking coefficient sreac
in situ from the measured variation of the reactant
partial pressures, i.e. of O or NH . Denoting the
single step double step single step
single step
(
b)
-4
2
.5
2x10 mbar
N2
NO
2
1
1
0
0
.0
.5
.0
.5
.0
2
3
signal of a gas without reaction by I and during
0
reaction with I we calculate the reactive sticking
coefficient sreac as
6
00
800
1000
T [K]
I
0
ꢀ I
sreac
¼
Fig. 4. Influence of the total pressure on the width and the
amplitude of the hysteresis that arises during a heating/cooling
cycle on Pt(533) with a mixing ratios pO2 =pNH3 of 1:1. The bars
on top of each figure demonstrate the structural changes as
verified by LEED. A heating rate of 100 K/min was used. Error
I
0
In our case we take the signal at 300 K for I0,
assuming a negligible reaction rate at this
temperature.
ꢀ5
bars, cf. caption of Fig. 2. (a) ptotal = 6 · 10 mbar, (b)
ꢀ
4
Fig. 5 displays a comparison of the reactive
sticking coefficients for both surfaces and both
gases measured under the steady state conditions
of a temperature cycling experiment. Under steady
state conditions the number of atoms which ad-
sorb and react, has to be equal to the sum of the
atoms in the products which leave the surface.
The number of nitrogen atoms adsorbing has to
be equal to the number of N atoms in the products
leaving the surface and the corresponding amount
of oxygen depends on the relative contributions of
the two pathways given by (R1) and (R2). We have
p
total = 2 · 10 mbar.
lower temperature, i.e. in the case of nitrogen from
about 670 K for a 1:1 ratio to 530 K as the ratio is
increased to 10:1. For NO production the corre-
sponding shift is nearly the same. Furthermore,
the width of the N peak narrows considerably
2
with increasing excess of oxygen from about
4
00 K at a 1:1 ratio to roughly 250 K at 10:1. With
increasing p =p also the region of reaction-
induced restructuring shifts to lower temperature
O2
NH3
as reflected by the shift in the dip of the N rate
2
4
s
reacðNH
3
ÞkNH3 pNH3 ¼ 4rNO þ 2rN2
curves.
The comparison of the absolute rates shows
that, depending on the ratio p =pNH3 , the
Pt(533) orientation is catalytically more active
¼
PsreacðO ÞkO2 pO2
2
O2
where k is the impingement rate of the gas i per
i
2
1 cm surface area according to kinetic gas theory,
i.e.
than Pt(443) by a factor 2to 4. The rate maxima
of N /NO production appear at higher tempera-
2

A. Scheibe et al. / Surface Science 577 (2005) 1–14
7
0
0
0
0
0
.20
NH3 / Pt(533)
O2 / Pt(533)
NH3 / Pt(443)
O2 / Pt(443)
(
a)
.15
.10
.05
.00
3
:1
500
600
700
800
900
1000
1100
(
b)
0.20
NH3 / Pt(533)
O2 / Pt(533)
NH3 / Pt(443)
O2 / Pt(443)
0.15
0.10
0.05
0.00
1
:1
500
600
700
800
900
1000
1100
T [K]
Fig. 5. Comparison of the reactive sticking coefficients, sreac, of the reacting gases on Pt(533) and Pt(443) for different mixing ratios.
ꢀ5
The ammonia partial pressure was kept fixed at pNH3 ¼ 3 Â 10 mbar in all experiments. The data shown were obtained from the
heating part of a heating/cooling cycle with a heating rate of 100 K/min. Error bars, cf. caption of Fig. 2. (a) p ¼ 1 Â 10ꢀ mbar,
4
O2
b) pO2 ¼ 3 Â 10ꢀ mbar.
5
(
p
i
as compared to the 1/7 of Pt(443) (cf. Fig. 1),
one would expect a higher oxygen sticking coeffi-
cient on Pt(533). This is what we see but the ratio
of 2.4 is larger than the ratio of the step densities,
which only amounts to 7/4 = 1.75. We should keep
in mind that the reactive sticking coefficient is only
a lower boundary for the total sticking coefficient
s, i.e. we always have s > sreac since particles
adsorbing with the sticking probability s may
either react or desorb without reaction.
k
i
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
i
pm kT
and P denotes a stoichiometric factor according to
the different O /NH ratios in the two reaction
channels, described by reaction equations (R1)
and (R2). P is given by
2
3
2
ð5rNO þ 3rN2 Þ
P ¼
r
NO þ r
N
2
With the above expression we can check
whether mass conservation is fulfilled and this is
the case for the data in Fig. 5, together with the
corresponding reaction rates. The data demon-
strate that the reactive sticking coefficient for
ammonia can be quite high if oxygen is in excess,
reaching nearly 0.2in the N 2 rate maximum of
Pt(533). The reactive sticking coefficient for oxy-
gen is highest when ammonia is in excess going
up to 0.12on Pt(533) and 0.05 on Pt(443). Due
to a higher step density on Pt(533), which is 1/4
With increasing temperature the reactivity of
the surface decreases beyond 600–700 K, as re-
flected by a decreasing sreac in Fig. 5. The surface
residence time of ammonia becoming too small be-
fore reaction can occur will be one of the factors
responsible for this decrease. The other factor is
that on Pt(111) surfaces the oxygen sticking for
dissociative chemisorption decreases with rising
temperature as a consequence of the smaller popu-
lation of the molecular precursor state for dissoci-
ation [22].

8
A. Scheibe et al. / Surface Science 577 (2005) 1–14
3
.3. Dependence on partial pressure
7. The data in Fig. 6 for T = 550 K and
ꢀ
5
pNH3 ¼ 1 Â 10 mbar show that at low p , N
O2
2
The dependence of the kinetics of NO and N2
formation is the preferred reaction channel. With
increasing pO2 N formation starts to level off while
production on the oxygen partial pressure at two
temperatures, T = 550 K and T = 625 K, is dis-
played in Figs. 6–9 for Pt(533) and Pt(443),
respectively. All data shown here were measured
2
NO formation continues to rise. The saturation of
the N production with increasing p can only be
2
O2
ꢀ
5
seen up to pNH3 ¼ 3 Â 10 mbar (not shown here),
ꢀ
4
by stepwise increasing p , starting from zero.
whereas in the 10 mbar range the required pO2
lies beyond the experimentally accessible parame-
O2
Hysteresis effects, which arise when the pressure
is cycled due to reaction-induced structural
changes, are thus absent in the diagrams [15]. We
first discuss the results for Pt(533) in Figs. 6 and
ꢀ
5
ter range. At pNH3 ¼ 1 Â 10 mbar in Fig. 6a the
reaction order (RO) for N and NO production
2
Pt(533) T = 625K
Pt(533) T = 550K
(
a)
N2
NO
(a)
(b)
(c)
N2
NO
1
1
0
.1
0
.1
0
.01
0
.01
1
-
7
-6
-5
-4
1
0
10
10
10
10
10
10
10
-
7
7
7
-6
-6
-6
-5
-5
-5
-4
-4
-4
0
10
10
10
10
10
10
10
p(O ) [mbar]
2
(
b)
N2
NO
N2
NO
1
.1
.01
1
1
.1
0
0
0
.01
0
-7
-6
-5
10
-4
-
1
0
0
10
p(O ) [mbar]
2
N2
NO
N
2
NO
(c)
1
.1
.01
1
0.1
.01
0
0
0
-
-7
-6
-5
-4
1
0
10
10
10
p(O ) [mbar]
p(O ) [mbar]
2
2
Fig. 6. Dependence of the N
2
and NO production rates of
2
Fig. 7. Dependence of the N and NO production rates of
Pt(533) on pO2 at T = 550 K. The data were measured by
stepwise increasing pO2 starting from zero. The lines connecting
the experimental data points have been drawn to guide the eye.
Pt(533) on pO2 at T = 625 K. Data were measured by stepwise
increasing pO2 . The lines connecting the experimental data
points have been drawn to guide the eye. (a) p
10 mbar, (b) pNH3 ¼ 1 Â 10 mbar, (c) pNH3 ¼ 3 Â 10
¼ 1Â
NH3
(
a) p
¼ 1 Â 10ꢀ mbar, (b) pNH3 ¼ 1 Â 10ꢀ4 mbar, (c) pNH3
5
¼
ꢀ5
ꢀ4
ꢀ4
NH3
3
 10ꢀ mbar.
4
mbar.

A. Scheibe et al. / Surface Science 577 (2005) 1–14
9
Pt(443) T = 550K
Pt(443) T = 625K
(a)
N2
NO
(a)
N
2
NO
1
1
0
.1
0.1
0.01
0.01
-
6
-5
-4
-3
-6
-5
-4
-3
10
10
10
10
10
10
10
10
(
b)
(
b)
N2
NO
N2
1
1
.1
.01
NO
0
.1
0
0
.01
0
-
6
-5
-4
-3
-6
-5
-4
-3
10
10
10
10
1
0
10
10
10
p(O ) [mbar]
p(O2) [mbar]
2
Fig. 8. Dependence of the N
2
and NO production rates of
2
Fig. 9. Dependence of the N and NO production rates of
Pt(443) on pO2 at T = 550 K. Data were measured by stepwise
Pt(443) on pO2 at T = 625 K. Data were measured by stepwise
increasing pO2 . The lines connecting the experimental data
points have been drawn to guide the eye. (a) pNH3 ¼ 1 Â 10ꢀ5
increasing pO2 . The lines connecting the experimental data
5
points have been drawn to guide the eye. (a) p
mbar, (b) pNH3 ¼ 1 Â 10 mbar.
¼ 1 Â 10ꢀ
NH3
ꢀ4
ꢀ4
mbar, (b) pNH3 ¼ 1 Â 10 mbar.
is 0.8 and 0.7, respectively. Increasing p
to
NH3
spective of pNH3 . We can attribute the low RO
ꢀ
4
1
that both products have the same RO of 0.7. A
· 10 mbar decreases the RO for N to 0.7 so
2
order for N formation below p =p
= 1:10 to
the presence of a large ammonia coverage on the
2
O
NH3
2
ꢀ
4
further increase of pNH3 to 3 · 10 mbar leaves
the RO for NO unchanged but the RO for N2
drops to 0.3, as shown by Fig. 6c.
surface inhibiting O adsorption. The inhibition
2
restricts the reaction to a few reactive centers, i.e.
surface defects, where ammonia cannot block
adsorption sites.
The corresponding data for Pt(443) measured
at the same temperatures of 550 K and 625 K are
displayed in Figs. 8 and 9. With respect to the po-
sition relative to the production maxima the condi-
tions are different from Pt(533), as shown by Fig.
At T = 625 K with oxygen being in excess, one
has reaction conditions close to the N rate maxi-
2
mum in Fig. 2. The corresponding dependence of
is displayed in Fig. 7. With
the rates on p
O2
ꢀ
5
pNH3 ¼ 1 Â 10 mbar the NO and N production
2
increase both initially with a RO close to unity,
as demonstrated by Fig. 7a. Beyond p ¼ 2Â
O2
3. On Pt(443) the rate maximum for N produc-
2
ꢀ
5
1
0
mbar the rates level off and increase only
ꢀ
tion lies about 100 K higher in temperature than
on Pt(533). We therefore expect no saturation of
N production upon increasing p and the NO
4
slightly with p . Increasing pNH3 in the 10 mbar
range we can distinguish two different regimes in
O2
2
O2
N production, as evidenced by Fig. 7b. Below
2
production should be quite low, unless we go to
very high p . This is more or less what we observe
p =p
excess, the N production increases weakly with
= 1:10, with ammonia being strongly in
O2
NH3
O2
ꢀ
5
2
in Figs. 8 and 9. At pNH3 ¼ 1 Â 10 mbar in Fig.
pO2 , yielding a RO of 0.2. Above the 1:10 ratio
the RO jumps to one. The RO order for NO for-
mation is always close to one at T = 625 K irre-
8a the N production exhibits a RO of 0.4 with
2
respect to oxygen. NO formation sets in notably
= 1:10, increasing then
NH3
only beyond p =p
O2

10
A. Scheibe et al. / Surface Science 577 (2005) 1–14
ꢀ4
steeply with RO ꢁ 1. Raising p
to 1 · 10
mbar increases the RO for N to 0.6, whereas the
Pt(443) T = 550K
NH3
2
RO for NO formation decreases to 0.25, as dem-
onstrated by Fig. 8b.
At higher temperature, at T = 625 K, the N2
4
2
N2
NO
0.1
formation saturates with increasing p , as shown
O2
4
2
ꢀ
4
in Fig. 9a for pNH3 ¼ 1 Â 10 mbar. NO produc-
tion becomes significant only beyond a 10:1 ratio
exhibiting 1st order kinetics with respect to oxy-
0.01
4
gen. As shown by Fig. 9b, raising p
to
-7
-6
-5
-4
-3
NH3
1
0
10
10
10
10
ꢀ
4
1
· 10 mbar changes the RO for N production
2
p(NH3) [mbar]
to 0.7 in the investigated parameter range. NO
production obeys a kinetics with RO = 0.3 under
conditions of ammonia being in excess.
The dependence of the reaction kinetics of
Pt(533) on the ammonia partial pressure is dis-
played in Fig. 10 for T = 525 K and T = 62 5 K.
Experimentally it is difficult to extend the parame-
ter range to very low pNH3 due to saturation of the
2
Fig. 11. Dependence of the N and NO production rates of
ꢀ
5
Pt(443) at T = 550 K on pNH3 at fixed pO2 ¼ 1 Â 10 mbar.
Data measured by stepwise increasing pNH3 . The lines connect-
ing the experimental data points have been drawn to guide the
eye.
walls with NH . At T = 625 K in Fig. 10a the N
3
2
production saturates quite early below a 10:1 ratio
p =p . The NO kinetics exhibits a quite unusual
O2
NH3
ꢀ
5
behavior because above pNH3 ¼ 1 Â 10 mbar the
rate of NO formation increases with pNH3 , whereas
the opposite tendency is expected. The effect is
reproducible and could so far not been traced back
to an artefact like reaction at hot filaments. At
Pt(533)
(a)
N2
NO
1
T = 625 K the formation of N and NO increases
2
0
.1
initially with pNH3 and then both rates saturate
quite early.
0
.01
T = 525 K
On Pt(443) we obtain a quite similar behavior
as on Pt(533), as demonstrated in Fig. 11 for
T = 550 K. Over a broad parameter range both
production rates exhibit a zeroth order kinetics
with respect to pNH3 , except that at high pNH3 the
NO production starts to increase with pNH3 . Just
the opposite effect should be expected from the
surface chemistry but the effect turned out to be
reproducible.
6
8
2
4
6 8
2
4
6 8
2
4
-
6
-5
-4
1
0
10
p(NH ) [mbar]
10
3
(
b)
N2
1
.1
.01
NO
0
0
T = 625 K
4. Discussion
6
8
2
4
6 8
2
4
6 8
2
4
-6
-5
-4
1
0
10
p(NH ) [mbar]
10
While operating conditions in the Ostwald pro-
3
cess have been optimized a long time ago achieving
high conversions, comparatively few kinetic and
mechanistic studies with the NH + O reaction
2
Fig. 10. Dependence of the N and NO production rates of
Pt(533) on pNH3 at fixed pO2 ¼ 3 Â 10ꢀ mbar. Data were
measured by stepwise increasing pNH3 . The lines connecting the
experimental data points have been drawn to guide the eye.
5
3
2
on platinum have been published. The kinetics of
the NH + O reaction on polycrystalline Pt, Rh,
(a) T = 525 K, (b) T = 62 5 K.
3
2

A. Scheibe et al. / Surface Science 577 (2005) 1–14
11
ðR9Þ
and Ir wires have been measured in the p range
.1–1 mbar by Pignet and Schmidt [4]. On sup-
NH + O () N + OH
ad
ad
ad
ad
0
ported catalysts the variation of the catalytic activ-
ity with Pt loading was attributed to particle size
effects [7]. The kinetics of the reaction in a microre-
actor with supported Pt as catalyst have been
investigated by Rebrov et al. [8]. With positron
emission profiling Sobczyk et al. succeeded in fol-
lowing the distribution of isotope labeled species in
a reactor with a resolution of 3 mm [30].
NH3;ad +OH !NH +H O+*
ðR10Þ
ðR11Þ
ðR12Þ
ðR13Þ
ðR14Þ
ðR15Þ
ðR16Þ
ðR17Þ
ðR18Þ
ad
2;ad
2
NH2;ad + OH ! NH + H O + *
ad
ad
2
NH + OH ! N + H O + *
ad
ad
ad
2
2N ! N + 2 *
ad
2
The dynamics of the NH + O reaction on Pt
3
N + O () NO + *
2
ad
ad
ad
have been investigated by several groups in the
mbar—1 atm pressure range. Rate oscillations
and stationary and non-stationary T-patterns on
electrically heated Pt wires and ribbons have been
reported [31–34]. Hysteresis effects, in connection
with ignition/extinction phenomena, were demon-
strated by Barelko and Volodin [5]. All these phe-
nomena appear to be limited to the use of high
pressures (p > 1 mbar) because so far no such phe-
nomena have been reported in low pressure single
crystal studies.
NO ! NO + *
ad
O + H () OHad
ad
ad
OH + H ! H O + 2 *
ad
ad
2
2OHad () O + H O
ad
2
For N O formation, which was not observed at
2
ꢀ
3
low p (p < 10 mbar), two reaction pathways
can be considered for N O, which should immedi-
2
At higher pressure mass transport and not sur-
face chemistry will limit the conversion rates
because of the high catalytic activity of the Pt sur-
face. The high activity is also corroborated by our
results, showing that the reactive sticking coeffi-
cient for ammonia can reach 0.2(cf. Fig. 5). In
low pressure single crystal studies such transport
limitations are absent.
ately desorb at T > 300 K:
NOad + Oad ! N
2
O + 2 *
ðR19Þ
ðR20Þ
2
NO ! N O + O + *
ad
2
ad
Quantum chemical calculations support the latter
mechanism on Pt surfaces since it exhibits a lower
activation barrier than step (R19) [35]. The reverse
reaction to (R20) was shown to be dominant in
According to recent studies the dissociation of
ammonia according to
N O decomposition over Pt [36].
2
NH3;ad ! Nad + 3Had
ðR4Þ
The different regimes of the reaction can easily
be summarized. The main factors controlling the
NO formation are the availability of atomic oxy-
gen and nitrogen and a high enough temperature
so that NO can desorb. At lower temperature with
a small oxygen coverage and a large amount
NH (X = 1–3) species N formation will be domi-
contributes little to the overall activity [13]. Direct
hydrogen abstraction by chemisorbed oxygen or
OH species is the dominant reaction pathway for
product formation. With the asterisk * denoting
a vacant adsorption site the reaction mechanism
in simplified form can be formulated as
X
2
nant. At too low temperatures the reactivity of the
surface tends towards zero, apparently as a conse-
quence of a high adsorbate coverage inhibiting
oxygen adsorption. At too high temperature the
reactivity of the surface also becomes very small
O
2
+ 2 *() 2Oad
ðR5Þ
ðR6Þ
ðR7Þ
ðR8Þ
NH
3
+ * () NH3;ad
NH3;ad + Oad () NH2;ad + OHad
due to the residence time of NH becoming too
3
small for reaction and as a consequence of an
oxygen sticking coefficient which decreases with
NH2;ad + Oad () NHad + OHad

12
A. Scheibe et al. / Surface Science 577 (2005) 1–14
temperature [22]. Simulations with a realistic
mathematical model based on the above scheme
could reproduce the behavior of the single crystal
qualitatively correct and to a large degree even
quantitatively [28,37].
aside the possibility that this factor might be with-
in the experimental uncertainty. Both, dissociative
oxygen chemisorption as well as NO decomposi-
tion are highly structure sensitive as is known from
numerous studies with Pt single crystal surfaces
[21–24]. In addition, it is also demonstrated that
ammonia adsorption on platinum surfaces is
structure sensitive with the sticking coefficient
increasing on rough surfaces [39]. The sticking
coefficient of ammonia on Pt(210) is about ten
times higher on Pt(210) than on Pt(111) and for
the dissociation probability the difference is even
larger.
A molecular beam study of the NH + O reac-
3
2
tion on Pt(100) yielded rather similar results as
found here on Pt(533) [13]. With a NH /O mix-
ture containing 90% O the rate maximum of N
3
2
2
2
production is found at 500 K while we observe
the maximum at 520 K with a ratio 10:1 (cf. Fig.
2
). For NO production we note a significant differ-
ence because the rate maximum on Pt(100) for a
mixture with 90% O2 lies around 600–700 K,
whereas we find a rate maximum at 770 K for a
For oxygen adsorption the differences in the ini-
tial sticking coefficients are not known for the two
surfaces since only data for Pt(533)/O can be
1
0:1 ratio (cf. Fig. 2). When we compare our data
2
with those of Pignet and Schmidt for Pt wires we
realize that their rate maxima lie at higher temper-
ature as a consequence of the higher pressure of
p ꢁ 1 mbar they use [4]. With a gas mixture with
found in the literature [40]. If we assume that the
reactive sticking coefficients displayed in Fig. 5 re-
flects this difference then we have roughly the same
factor of 2we already have from the step density.
In temperature programmed desorption (TPD)
experiments it was shown that about 60% of the
initially molecularly adsorbed NO decomposes
on Pt(100), whereas Pt(111) and Pt(110) exhibit
a rather low dissociation probability [23,24,41].
The highest activity for NO decomposition was
found on Pt(410). According to the principles of
microscopic reversibility one would expect to find
a similar trend also for the reverse process, i.e.
NO formation via (R14). If we assume that
(100) steps exhibit a particularly high activity in
NO dissociation and hence NO association, this
would explain the higher activity of Pt(533).
4
0% NH the rate maximum for N lies at 620 K
3 2
but for a mixture with 90% NH the maximum
3
shifts strongly to ꢁ1000 K. The corresponding
numbers for NO production are 1050 K and
ꢁ
1300 K, respectively, for the two mixtures. The
trend that the rate maxima with decreasing oxygen
content of the gas mixture shift to higher temper-
ature is the same as in our experiments (cf. Fig. 2).
The concept of using stepped surfaces as a
model catalyst for structure sensitive reactions
has been applied in quite a number of studies since
the first investigations by Somorjai and coworkers
[
38]. The choice to use the two orientations
Pt(533) and Pt(443) as model catalysts for the
NH + O reaction on Pt was motivated by two
Our data in fact support a particular role of
(100) steps because we observe a drastic change
in selectivity from preferential N formation to
3
2
facts. On one hand in the absence of reaction-in-
duced restructuring the (111) orientation will be
dominant on polycrystalline Pt for thermodynamic
reasons, while on the other hand a flat Pt(111) sur-
face is fairly unreactive due to the low sticking
coefficient of oxygen [21,22].
Compared to Pt(443) the Pt(533) orientation is
more active by a factor of 2–4, as judged from the
height of the rate maxima. The higher step density
of Pt(533) compared to Pt(443) already accounts
for a factor of 7/4 = 1.75 but the remaining factor
of 1.1–2.2 might be due to a higher catalytic activ-
ity of (100) steps compared to (111) steps, leaving
2
NO production as the single atomic steps on
Pt(533) undergo a doubling in step height. Quan-
tum chemical calculations show that the transition
state for NO dissociation on Pt(100)-1 · 1 is a flat-
lying NO molecule with the N atom and the O-
atom bonded each to two Pt atoms of the four
Pt atoms, forming the square of the (1 · 1) unit cell
of the substrate [35]. One might speculate that this
transition state on the (100) double steps of
Pt(533) is energetically less affected by space inter-
actions with surrounding adsorbate molecules
than on the single atomic steps of the same surface.

A. Scheibe et al. / Surface Science 577 (2005) 1–14
13
The energetically more favorable transition state
Acknowledgment
might explain why the restructuring leads to a
preference towards NO formation.
On Pt(533) the doubling of the step height was
This work was supported by the DFG under the
priority program 1091 ‘‘Bridging the gap between
ideal and real systems in heterogeneous catalysis’’.
ꢀ
5
shown to occur only at p > 10 mbar in a temper-
ature window of roughly 500–750 K [15,39]. On
Pt(443) no structural changes were detectable in
LEED after exposure to reaction conditions in
References
ꢀ
5
ꢀ4
the 10 and 10 mbar range [37]. With STM
however, it was demonstrated that even adsorp-
tion of ammonia alone at 300 K leads to a step
restructuring [37]. The originally straight step
edges start to meander in the presence of adsorbed
ammonia. Because the average step separation
does not change the meandering does not show
up in LEED by a different spot splitting. A step
meandering under reaction conditions can thus
not be excluded for Pt(443).
[
1] T.H. Chilton, The manufacture of nitric acid by the
oxidation of ammoniaChemical Engineering Progress
Monograph Series No. 3, vol. 56, American Institute of
Chemical Engineers, New York, 1960.
[2] C.W. Nutt, S.W. Kapur, Nature 220 (1968) 697;
C.W. Nutt, S.W. Kapur, Nature 224 (1969) 169.
[
3] Y.M. Fogel, B.T. Nadykto, V.G. Rybalko, V.I. Shvachko,
I.E. Korebehanskaya, Kinet. Catal. 5 (1964) 169.
4] T. Pignet, L.D. Schmidt, J. Catal. 40 (1975) 212.
5] V.V. Barelko, Yu.E. Volodin, Dokl. Akad. Nauk SSSR
211 (1973) 1373.
[
[
[
[
6] B.A. Morrow, I.A. Cody, J. Catal. 45 (1976) 151.
7] J.J. Ostermaier, J.R. Katzer, W.H. Manogue, J. Catal. 33
The morphological changes Pt undergoes in the
NH + O reaction were studied systematically by
3
2
(1974) 457.
8] E.V. Rebrov, M.H.J.M. de Croon, J.C. Schouten, Chem.
Eng. J. 90 (2002) 61.
Schmidt et al. using small single crystalline Pt
spheres (<0.5 mm) as catalysts [16–18]. The condi-
tions they used—atmospheric pressure and tem-
peratures up to 1100 K—are, however, quite
drastic compared to ours. Under these conditions,
which are comparable to those of the Ostwald pro-
cess, different mechanisms will be dominant in cat-
alytic etching, like the formation of volatile Pt
oxides.
[
[9] J.L. Gland, V.N. Korchak, J. Catal. 53 (1978) 9.
10] M. Asscher, W.L. Guthrie, T.-H. Lin, G.A. Somorjai, J.
Phys. Chem. 88 (1984) 3233.
[
[
[
11] D.S.Y. Hsu, D.W. Squire, M.C. Liu, J. Chem. Phys. 89
(
12] W.D. Michev, W. Ho, Surf. Sci. 322 (1995) 151.
1988) 2861.
[13] J.M. Bradley, A. Hopkinson, D.A. King, J. Phys. Chem.
9 (1995) 17032.
14] M. Kim, S.J. Pratt, D.A. King, J. Am. Chem. Soc. 122
2000) 2409.
15] A. Scheibe, S. G u¨ nther, R. Imbihl, Catal. Lett. 86 (2003)
3.
9
[
[
[
[
[
(
5
. Conclusions
3
16] M. Flytzani-Stephanopoulos, S. Wong, L.D. Schmidt, J.
Catal. 49 (1977) 51.
It was shown that both stepped Pt(111) sur-
faces, Pt(533) and Pt(443), are catalytically very
active in the NH + O reaction, as evidenced by
17] M. Flytzani-Stephanopoulos, L.D. Schmidt, Prog. Surf.
Sci. 9 (1979) 83.
18] R. McCabe, T. Pignet, L.D. Schmidt, J. Catal. 32(1974)
3
2
the high values of the reactive sticking coefficients.
Pt(533) is more active than Pt(443) by a factor of
1
[19] M.R. Lyubovsky, V.V. Barelko, J. Catal. 149 (1994) 23.
14.
[
[
20] T.C. Wei, J. Phillips, Adv. Catal. 41 (1992) 359.
21] A. Winkler, X. Guo, H.R. Siddiqui, P.L. Hagans, J.T.
Yates, Surf. Sci. 201 (1988) 419.
2
–4 what can be attributed to (i) the higher step
density of Pt(533) and (ii) possibly to a higher
activity of (100) step sites as compared to (111)
step sites. The apparent reaction order of product
formation varies strongly with the reaction condi-
tions. What this study also demonstrates is that
reaction-induced structural changes are also occur-
ring under quite mild reaction conditions and that
these changes affect the activity and selectivity of
the NH + O reaction.
[
22] C.T. Campbell, G. Ertl, H. Kuipers, J. Segner, J. Chem.
Phys. 73 (1980) 5862.
[23] R.J. Gorte, L.D. Schmidt, J.L. Gland, Surf. Sci. 109 (1981)
67.
3
[
[
[
24] J.M. Gohndrone, R.I. Masel, Surf. Sci. 209 (1989) 44.
25] R. Imbihl, Mod. Phys. Lett. B 6 (1992) 493.
26] S. Ladas, R. Imbihl, G. Ertl, Surf. Sci. 197 (1988) 153;
J. Falta, R. Imbihl, M. Henzler, Phys. Rev. Lett. 64 (1990)
1409.
3
2

14
A. Scheibe et al. / Surface Science 577 (2005) 1–14
[
[
27] A. Fahmi, R.A. van Santen, Z. Phys. Chem. 197 (1996)
03;
[35] Q. Ge, M. Neurock, JACS 126 (2004) 1551.
[36] V.A. Kondratenko, M. Baerns, J. Catal., submitted for
publication.
2
W. Offermans, R.H. van Santen, in preparation.
28] A. Scheibe, M. Hinz, R. Imbihl, Surf. Sci., submitted for
publication.
[37] A. Scheibe, Thesis, Hannover, 2003.
[38] S.M. Davis, F. Zaera, G.A. Somorjai, J. Catal. 77 (1982)
439;
[
[
29] D.W. Blakely, G.A. Somorjai, Surf. Sci. 65 (1977) 419.
30] D.P. Sobczyk, A.M. de Jong, E.J.M. Hansen, R.A. van
Santen, J. Catal. 219 (2003) 156.
S.M. Davis, F. Zaera, G.A. Somorjai, J. Catal. 85 (1984)
206.
[
31] M. Flytzani-Stephanopoulos, L.D. Schmidt, R. Caretta, J.
Catal. 64 (1980) 346.
[39] J.M. Gohndrone, C.W. Olsen, A.L. Backman, T.R. Gow,
E. Yagasaki, R.I. Masel, J. Vac. Sci. Technol. A 7 (1989)
1986.
[
[
[
32] C.G. Takoudis, L.D. Schmidt, J. Catal. 84 (1983) 235.
33] M. Sheintuch, J. Schmidt, J. Phys. Chem. 92(1988) 3404.
34] L. Lobban, G. Philippou, D. Luss, J. Phys. Chem. 93
[40] A.T. Gee, B.E. Hayden, J. Chem. Phys. 113 (2000) 10333.
[41] T. Fink, J.-P. Dath, R. Imbihl, G. Ertl, J. Chem. Phys. 95
(1991) 2109.
(1989) 733.