Selectivity-directing factors of ammonia oxidation over PGM gauzes in the temporal analysis of products reactor: Primary interactions of NH3 and O2

Article abstract of DOI:10.1016/j.jcat.2004.06.023

The temporal analysis of products reactor was applied to study selectivity-directing factors of the high-temperature NH₃ oxidation to NO, N₂O, and N₂ over commercial knitted Pt and woven Pt-Rh alloy gauzes at 973-1173 K. T

Full text of DOI:10.1016/j.jcat.2004.06.023

Journal of Catalysis 227 (2004) 90–100
www.elsevier.com/locate/jcat
Selectivity-directing factors of ammonia oxidation over PGM gauzes
in the Temporal Analysis of Products reactor:
Primary interactions of NH3 and O2
J. Pérez-Ramírez a,∗, E.V. Kondratenko b,∗, V.A. Kondratenko b, M. Baerns b
a
Yara Technology Centre Porsgrunn, Catalysis and Nitric Acid Technology, PO Box 2560, N-3908, Porsgrunn, Norway
b
Institute for Applied Chemistry Berlin-Adlershof, Richard-Willstätter-Str.12, D-12489, Berlin, Germany
Received 4 April 2004; revised 20 June 2004; accepted 26 June 2004
Available online 31 July 2004
Abstract
The Temporal Analysis of Products (TAP) reactor has been applied to study selectivity-directing factors of the high-temperature NH
3
oxidation to NO, N O, and N over commercial knitted Pt and woven Pt–Rh alloy gauzes at 973–1173 K. The unique features of the TAP
2
2
technique enable investigation of the mechanism of this highly exothermic process under isothermal conditions over catalysts of industrial
relevance, and the applicaton of much higher peak pressures as compared to surface science techniques in ultrahigh vacuum. This article
focuses on the investigation of primary interactions of NH and O . The overall reaction mechanism was found to be very similar over
3
2
both Pt and Pt–Rh gauzes. NH activation is favored over O -pretreated gauzes, while the as-received gauzes are virtually inactive for NH
3
2
3
decomposition to N . The selectivity to NO primarily depends on the concentration of adsorbed oxygen species. A high ratio of adsorbed
2
O/NH species favors NO formation, confirming that undesirable secondary reaction paths are minimized at a high O coverage. Besides, our
x
results suggest that the nature of oxygen species influences the distribution of NO and N in the product. It is put forward that weakly bounded
2
oxygen species lead to a high NO selectivity, while strongly bounded oxidize NH into N . The interaction of NH and NO also contributes to
3
2
3
15
N formation, while direct NO decomposition is practically suppressed over the oxidized gauzes. Application of isotopically labeled NH
2
3
3
and high peak pressures were essential for detecting N O formation during ammonia oxidation. Analysis of secondary interactions of NH
2
and NO in the authors’ next project is required to further unravel the origins of reaction by-products like N O and N .
2
2
2004 Elsevier Inc. All rights reserved.
Keywords: NH oxidation; Platinum; Rhodium; Gauze; Selectivity; NO; N ; N O; Mechanism; Transient experiments; TAP reactor; Isotopes
3
2
2
1. Introduction
of the reaction mechanism and kinetics of product forma-
tion. However, despite numerous studies over Pt wires [2–4]
and Pt single crystals [5–10], a generally accepted mecha-
nistic description of the process at a molecular level has not
yet been achieved. Early theories on the mechanism of am-
monia oxidation differ in the nature of the formed interme-
diates, which determine a certain reaction product. Nitroxyl
(HNO) [11], hydroxylamine(NH2OH) [12], and imide (NH)
[13] were postulated as possible intermediate species. Later,
application of modern surface science techniques in ultra-
high vacuum (UHV) provided new insights into the reaction
mechanism over single Pt crystals in a broad temperature
range (300–1700 K) [5,8,9,14]. Briefly, these studies con-
The catalytic oxidation of ammonia with air over PGM
(platinum group metals) gauzes is one of the highest tem-
perature (1073–1173 K) and shortest contact time (10−3
–
10−4 s) processes in the chemical industry. The reaction
is highly exothermic, yielding NO with a high selectivity
(95–97%) and N2 and N2O as undesired by-products [1].
Academic interest lies in the challenge of elucidating details
*
Corresponding authors.
E-mail addresses: javier.perez.ramirez@yara.com (J. Pérez-Ramírez),
evgenii@aca-berlin.de (E.V. Kondratenko).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2004.06.023
2004 Elsevier Inc. All rights reserved.

J. Pérez-Ramírez et al. / Journal of Catalysis 227 (2004) 90–100
91
Fig. 1. Schematic impression of proposed mechanisms of product formation during ammonia oxidation on a platinum surface as derived from surface science
studies in UHV (see references in text).
cluded that the oxidation of NH3 is initiated by the formation
of reactive intermediate NHx species. Further oxidation of
these ammonia fragments ultimately results in NO forma-
tion, while N2 is formed via recombination of two nitrogen
atoms or by decomposition of NO, as schematically shown
in Fig. 1. Based on experiments revealing that the adsorp-
tion of ammonia is not blocked by preadsorbed oxygen, two
different sites on the platinum surface have been suggested
for ammonia and oxygen adsorption, on top and hollow po-
sitions, respectively [9,10,15].
Despite the valuable mechanistic information derived
from UHV studies, these are carried out over model surfaces
(single crystals) at extremely low partial pressures (typically
10−5–10−7 Pa), representing a considerable gap of materials
and pressure with respect to realistic industrial conditions.
The robustness of this approach for achieving an accurate
description of the overall mechanism can be questioned,
since N2O was never detected as a product in UHV stud-
ies and thus its formation was not discussed. Accordingly,
the reaction has been considered as biphasic, yielding NO
and N2 [9]. Unquestionably, processes involving oxidation
of ammonia produce N2O as a by-product, and nitric acid
production is the largest source of N2O in the chemical in-
dustry (400 kton N2O per year) [1]. On this basis, a proper
understanding of the origin(s) of N2O would be of great fun-
damental and practical relevance.
experiments were limited to 700 K and extrapolation to rel-
evant burner temperatures is not straightforward.
In summary, two practical aspects can largely contribute
to an improved mechanistic and kinetic understanding of the
high-temperature ammonia oxidation: (i) isothermal condi-
tions at typical burner temperatures and (ii) application of
industrially relevant catalysts. This has been achieved by
the application of a transient pulse isotopic technique, the
Temporal Analysis of Products (TAP) reactor. To the best of
our knowledge, no TAP studies on ammonia oxidation have
been previously reported. This paper introduces the unique
features of the TAP technique for the above purposes and
focuses on selectivity-directing factors derived from the pri-
mary interactions between O2 and NH3 over commercial Pt
and Pt–Rh alloy gauzes at 973–1173 K. This study is com-
plemented by the analysis of secondary interactions of NH3
and NO in an upcoming manuscript. Our results lead to an
improved description of the mechanism(s) of product forma-
tion in ammonia burners.
2. TAP technique in NH3 oxidation
2.1. Temperature control
Transient experiments were performed in the Temporal
Analysis of Products reactor [24,25]. The gas transport in the
catalytic reactor is determined by the pulse size, which can
be varied in the range of 1013–1017 molecules per pulse. In
the Knudsen diffusion regime (pulse size < 1015 molecules),
purely heterogeneous steps of activation of molecules on the
catalyst surface are taken into account, since any collision
of molecules in the gas phase is minimized. The eventual
influence of gas-phase interactions or peak pressure on a
particular process can be assessed in a molecular diffusion
regime (pulse size > 1015 molecules).
The difficulties for assessing the mechanism of the high-
temperature ammonia oxidation with traditional experimen-
tal approaches at ambient pressure are principally caused
by ignition of the reaction as well as by the structural and
morphological changes of the metal alloy surface occurring
under reaction conditions [16–18]. Kinetic instabilities have
been observed over Pt wires at total pressures above 10−1 Pa
[19–21]. Application of structured microreactors coated with
Pt/Al2O3 has enabled kinetic data of the low-temperature
ammonia oxidation to be derived [22,23]. However, these

92
J. Pérez-Ramírez et al. / Journal of Catalysis 227 (2004) 90–100
A key advantage of the TAP technique over standard
the microreactor is described by Ne diffusion. The highest
pressure is situated at the reactor inlet for the diffusive flow.
The additional peak pressure observed in the catalyst posi-
tion is caused by the axial temperature gradients along the
reactor. For example, if the catalyst temperature is 1073 K,
the inlet and outlet temperatures of the microreactor are 340
and 380 K, respectively. Thus, the pressure gradient depends
on diffusive flow and the temperature profile in the reactor.
As derived from Fig. 2, the pressure in the catalyst (gauze)
position increases up to 160 Pa. For a typical UHV investiga-
tion at pressures of 10−7 Pa, the ratio of PUHV/Pambient is ca.
10−12, while in the TAP reactor the ratio of PTAP /Pambient is
increased up to 10−3 (using a pulse size of 1016 molecules).
Accordingly, the pressure gap between the TAP and the am-
bient pressure studies is largely reduced with respect to UHV
studies.
steady-state and unsteady-state techniques is the excel-
lent temperature control due to the very low heat pro-
duction (or consumption) associated to typical pulse sizes
(0.1–10 nmol). Therefore, highly exothermic or endother-
mic reactions can be studied under isothermal conditions.
The following estimation illustrates this unique feature for
high-temperature ammonia oxidation studies. For the ox-
idation of 10 nmol of NH3 to N2 (100% conversion and
selectivity) at 1073 K, an adiabatic temperature rise of ca.
2.8 K was estimated using the following expression [26]:
N2
1073 K
nꢀH
ꢀTad
=
.
(1)
mcatCp(Pt)
In this equation, n is the amount of NH3 per pulse (10 nmol),
ꢀHN10273 K is the reaction enthalpy of N2 formation at 1073 K
(−635 kJ mol−1), mcat is the catalyst mass (20 mg), and
Cp(Pt) is specific heat capacity of Pt (0.133 J g−1 K−1). The
selective oxidation of ammonia to N2O and NO would lead
to a lower ꢀT ad due to the lower reaction enthalpies (−552
and −454 kJ mol−1, respectively, at 1073 K) as compared
to the oxidation to N2. In practice, the calculated tempera-
ture rise is overestimated due to heat conduction away from
the gauze. Besides, the temperature within the gauze volume
is expected to be uniform due to the good heat conducting
properties of platinum metal.
2.3. Mean residence time
For a three-zone reactor model, the mean residence time
in the TAP reactor can be determined according to [28],
ꢀL · LII
dif
res,cat
τ
≈
,
(2)
D
where D is the gas diffusion coefficient (m2 s−1), ꢀL is
the gauze thickness (m), LII is the length of the second in-
dif
2.2. Peak pressure
ert zone (m), and τ
is the mean residence time (s).
res,cat
For a pulse experiment at 1073 K with a single gauze of
0.5 mm thickness, the mean residence time was estimated
to be ca. 10−3 s, which is comparable to typical values in
industrial burners (10−3–10−4 s).
Peak pressures in the TAP reactor can increase up to sev-
eral Pa. Fig. 2 shows a simulated pressure profile during sin-
gle pulsing of an inert gas like neon at 1073 K. A three-zone
model considering a temperature profile in the microreactor
was applied for the computation [27]. The mass transport in
3. Experimental
3.1. Gauze catalysts
Transient experiments were performed over two commer-
cial PGM gauzes: (i) pure Pt, knitted pattern, Multinit type 4,
supplied by Degussa, and (ii) Pt–Rh alloy, 95 wt% Pt and
5 wt% Rh, woven pattern, 1024 mesh per cm2, supplied by
K.A. Rasmussen. The wire diameter was 76 µm in both sam-
ples, as shown in Fig. 3. The SEM pictures evidence the
smooth surface of the as-received samples, as expected for
fresh noble metal gauzes. Due to the small amount of gases
pulsed in the TAP reactor and the duration of the experi-
ments, any surface reconstruction can be excluded in our ex-
periments. This was confirmed by microscopy studies over
the used gauzes. The typical phenomenon of sprouting or
cauliflower formation in the gauze during the first days on
stream in industrial burners causes dramatic increase of the
surface area of the gauze, as well as changes of the surface
composition [29].
Fig. 2. Contour plot of simulated pressure profiles in the TAP microreactor
−2
2
−1
−1
during Ne pulsing at 1073 K (D = 1.310 m s , M = 20 gmol
,
w
Ne
16
pulse size ∼ 10 molecules). The distance between two isobars is 20 Pa.

J. Pérez-Ramírez et al. / Journal of Catalysis 227 (2004) 90–100
93
ambient pressure and 1273 K for 2 h. After evacua-
tion, mixtures of O2:Ne = 1:1 and 14NH3:Xe = 1:1 (or
15NH3:Xe = 1:1) were sequentially pulsed at reaction
temperatures of 1023 and 1073 K using time delays of
ꢀt = 0.1–2 s.
– Single pulsing of NO over fresh, H2-pretreated, and
O2-pretreated gauzes. The gauzes were pretreated in
flowing pure O2 or H2 at ambient pressure and 1273 K
for 2 h. After evacuation, a mixture of NO:Ne = 1:1 was
pulsed over the catalyst at 973–1173 K.
In the experiments, Ne (4.5), Xe (4.0), O2 (4.5), NO (2.5),
14NH3 (2.5), and 15NH3 (99.9% atoms of 15N) were used
without additional purification. Isotopically labeled ammo-
nia was purchased from ISOTEC. Transient responses were
monitored at atomic mass units (AMUs) related to reactants,
reaction products, and inert gases at the reactor outlet using
a quadruple mass spectrometer (Hiden Analytical). The fol-
lowing AMUs were analyzed: 132 (Xe), 46 (NO2, 15N2O),
45 (15N14NO), 44 (N2O, CO2), 32 (O2), 31 (15NO, HNO),
30 (N2O, NO, 15N2), 29 (15N14N), 28 (N2O, N2), 20 (Ne),
18 (H2O, 15NH3), 17 (NH3, OH), 16 (O2, NH3), 15 (NH3),
and 2 (H2). For each AMU, pulses were repeated 10 times
and averaged to improve the signal-to-noise ratio. The con-
centration of feed components and reaction products was
determined from the respective AMUs using standard frag-
mentation patterns and sensitivity factors. The transient re-
sponses were typically normalized for a better comparison
of pulse shapes.
Fig. 3. Scanning electron micrographs of the knitted Pt gauze and woven
Pt–Rh gauze used in this study.
3.2. TAP experiments
TAP experiments were carried out in a quartz microreac-
tor (40 mm length and 6 mm i.d.), containing either one sin-
gle piece of woven Pt–Rh gauze (25 mg) or knitted Pt gauze
(110 mg), which is placed in the isothermal zone between
two layers of quartz particles (sieve fraction 250–350 µm).
During pulse experiments the catalyst is under vacuum con-
ditions (10−5 Pa). The pulse size was varied from ca. 1014 to
1016 molecules in order to perform experiments in Knudsen
and molecular diffusion regimes, respectively. The follow-
ing transient experiments were performed:
4. Results
4.1. Single pulsing of NH3 in Knudsen diffusion regime
The interaction of ammonia with the as-received and
O2-pretretated Pt and Pt–Rh gauzes was studied by single
pulsing of ammonia in a Knudsen diffusion regime at dif-
ferent temperatures. No difference in ammonia conversion
and products concentration was observed with the number of
pulses, indicating that the gauze performance did not change
during the TAP experiments. As shown in Fig. 4, the degree
of ammonia conversion is very low upon NH3 pulsing over
the as-received Pt and Pt–Rh gauzes, revealing the inactiv-
ity of essentially reduced noble metal surfaces for ammonia
decomposition at the short residence times in the TAP reac-
tor. Contrarily, significant amounts of N2, H2, and H2O were
produced when ammonia was pulsed over the O2-pretreated
gauzes (Table 1). As shown in Fig. 4, the NH3 conversion is
in the range of 30–40%. The observed decrease in N2 for-
mation with an increased temperature can be associated to
a decreased oxygen coverage. Our results suggest that the
O2 pretreatment results in the formation of oxygen species
that are essential for ammonia activation. Desorption of such
oxygen species can be assumed to increase with an increase
in temperature. It should be stressed that in these experi-
– Single pulsing of NH3 over as-received and O2-pretreat-
ed gauzes. A mixture of 14NH3:Ne = 1:1 was pulsed
over the as-received gauzes in the range of 973–1173 K.
The as-received gauzes were pretreated by pulsing
O2 at the corresponding reaction temperature for 2 h
(∼ 20 nmol of O2 per pulse), followed by NH3 pulsing
ca. 20 s after such a treatment.
– Single pulsing of NH3–O2 mixtures over O2-pretreated
gauzes. The gauzes were pretreated in flowing O2 at am-
bient pressure and 1273 K for 2 h. After evacuation,
mixtures of 14NH3:O2:Ne = 1:2:1 or 15NH3:O2:Ne =
1:2:1 were pulsed in the temperature range of 973–
1173 K.
– Sequential pulsing of O2 and NH3 over O2-pretreated
gauzes. The gauzes were pretreated in flowing O2 at

94
J. Pérez-Ramírez et al. / Journal of Catalysis 227 (2004) 90–100
4.2. Interaction of O2 and NH3 in the Knudsen diffusion
regime
In order to derive mechanistic insights into product for-
mation of ammonia oxidation over Pt and Pt–Rh gauzes,
primary interactions of O2 and NH3 were investigated in se-
quential pulse experiments at time delays in the range of
0–2 s. This contrasts with the longer period of 20 s be-
tween the last oxygen pulse and the first ammonia pulse
in the experiments described in Section 4.1, and enables an
analysis of the effect of the concentration and the nature of
O-adsorbed species formed during the first O2 pulse (pump)
on NH3 activation in the second pulse (probe). Fig. 5 shows
the transient responses resulting from sequential pulsing of
oxygen (O2:Ne = 1:1) and ammonia (14NH3:Xe = 1:1) at
ꢀt = 0.2 s for Pt gauze and ꢀt = 0.5 s for Pt–Rh gauze.
N2, NO, H2O, and H2 were the only reaction products de-
tected in ammonia pulse, while no products were observed
in the oxygen pulse. Accordingly, no N-containing species,
which can be eventually oxidized in the O2 pulse, were sta-
bilized on the catalyst surface. The transient responses of
the different species over Pt and Pt–Rh gauzes were very
similar, suggesting a minor influence of the gauze composi-
tion on the intrinsic mechanism of primary NH3–O2 interac-
tions.
Fig. 4. Influence of catalyst pretreatment on the NH conversion dur-
3
ing single pulsing of NH :Ne = 1:1 over Pt (circles) and Pt–Rh (square)
3
gauzes. Open symbols: as-received (nonpretreated) gauzes; solid symbols:
14
O -pretreated gauzes. Pulse size of NH ∼ 10 molecules.
2
3
Table 1
Performance of oxygen-pretreated noble metal gauzes in ammonia decom-
position in the TAP reactor
Sample T (K) Y(N ) (%) Y(NO) (%) Y(H ) (%) Y(H O) (%)
2
2
2
Pt
973
33
30
–
–
33
25
5.8
The O2 transient response clearly shows that the concen-
tration of gas-phase oxygen sharply decreases upon NH3
pulsing (Fig. 5). This sharp decrease in the oxygen signal
indicates that adsorbed oxygen species rapidly react with
NH3. For both Pt and Pt–Rh gauzes, NO appears directly
at the time of the ammonia pulse, indicating its forma-
tion directly from ammonia. The transient response of NO
is sharper and appears at shorter times than the transient
response of N2 (Fig. 5). This suggests that the primarily
formed NO is transformed into N2 via a secondary reac-
tion with NH3. This aspect will be further analyzed upon
investigation of NO and NH3 interactions in an upcoming
manuscript.
The NO yield and NH3 conversion vs. time delay in O2–
NH3 sequential pulse experiments at 1073 K are shown
in Fig. 6. NO yields close to 70% with a degree of NH3
conversion above 80% were obtained when a mixture of
O2:NH3:Ne = 2:1:1 was pulsed (equivalent to ꢀt = 0 s)
over the gauzes. The NO yield over Pt and Pt–Rh gauzes
strongly decreased upon increasing the time delay, being ca.
1% at ꢀt = 0.1 s. However, the NH3 conversion remained
practically unchanged in the range ꢀt = 0–2 s, indicating a
complete selectivity towards N2, as observed during single
pulsing of NH3 over the O2-pretreated gauzes at a somewhat
lower degree of ammonia conversion (ca. 40%, see Fig. 4).
The result in Fig. 6 strongly suggests the importance not
only of the oxygen coverage but also of the nature of the
adsorbed oxygen species on the selectivity of NH3 oxidation
to NO.
1073
4
Pt–Rh
1073
1173
47
33
–
0.02
21
21
25
14
14
Conditions: single pulsing of NH :Ne = 1:1; NH pulse size ∼ 10 mole-
3
3
cules.
ments, NH3 pulses were introduced in the TAP reactor 20 s
after the O2 pretreatment was finalized. This relatively long
time suggests that stable surface oxygen species deposited
by pulsing O2 over the as-received gauzes are responsible
for NH3 conversion into N2, H2, and H2O formation. As
shown in Table 1, hydrogen is obtained in excess water over
Pt gauze (H2/H2O ∼ 6), while a ratio of H2/H2O ∼ 1 was
achieved over Pt–Rh gauze. This can be attributed to the
presence of rhodium in the latter catalyst, which has a higher
affinity for oxygen than platinum. The higher degree of oxi-
dation of Pt–Rh gauze as compared to Pt gauze may induce
a more extended oxidation of H-containing species on the
catalyst surface leading to H2O. Only traces of NO were ob-
served over Pt–Rh gauze at 1173 K (Table 1). The insignif-
icant amount of NO produced in these experiments can be
attributed to the low concentration of oxygen species on the
catalyst surface (O coverage). However, it cannot be totally
excluded that long-living, strongly bounded oxygen species
are able to dehydrogenate ammonia but not transferable into
the corresponding ammonia intermediate to yield NO, with
the subsequent N2 production. A more detailed analysis on
the effect of concentration and nature of O species on the
products distribution is provided in Sections 4.2 and 5.1.
Finally, a very weak and broad signal at AMU 44 was
observed during sequential pulse experiments over Pt and

J. Pérez-Ramírez et al. / Journal of Catalysis 227 (2004) 90–100
95
Table 2
Performance of noble metal gauzes and SiO spheres in NH oxidation in
2
3
the TAP reactor
15
15
Sample
T (K) Diffusion regime X( NH ) (%) Y( NO) (%)
3
Pt gauze
1023 Knudsen
Molecular
1073 Knudsen
Molecular
92
91
90
92
55
61
56
62
Pt–Rh gauze 1023 Knudsen
Molecular
87
87
90
86
57
59
61
58
1073 Knudsen
Molecular
SiO spheres 1023 Molecular
2
7
9
1.6
2.2
1073 Molecular
15
15
Conditions: single pulsing of a mixture NH :O :Ne = 1:2:1; NH
3
2
3
14
16
pulse size ∼ 10 molecules (Knudsen diffusion regime) and ∼ 10 mole-
cules (molecular diffusion regime).
Fig. 5. Transient responses during sequential pulsing of O :Ne = 1:1 and
2
NH :Xe = 1:1 over Pt and Pt–Rh gauzes at 1073 K. Pulse sizes of O and
3
2
14
NH ∼ 10 molecules.
3
Fig. 6. NO yield (triangles) and NH conversion (circles) over Pt (open
3
Fig. 7. Transient responses during single pulsing of an ammonia–oxygen
15
symbols) and Pt–Rh (solid symbols) gauzes vs. time delay between O
mixture ( NH :O :Ne = 1:2:1) over Pt and Pt–Rh gauzes at 1073 K. Pulse
2
3
2
15
16
and NH in sequential pulse experiments at 1073 K. Pulse sizes of O and
size of NH ∼ 10 molecules.
3
2
3
14
NH ∼ 10 molecules.
3
during O2 pulsing in the TAP reactor at the high tempera-
tures investigated. Discrimination between N2O and CO2 in
mass spectrometry has been achieved by application of iso-
topically labeled 15NH3 instead of the nonlabeled 14NH3, as
described in Section 4.3.
Pt–Rh gauzes (not shown in Fig. 5). This cannot be unam-
biguously attributed to N2O, since the mass also corresponds
to CO2. It is well known that the surface of commercial
noble metal gauzes is covered by substantial amounts of car-
bon [30]. These carbon deposits can be oxidized to CO2

96
J. Pérez-Ramírez et al. / Journal of Catalysis 227 (2004) 90–100
4.3. Interaction of O2 and NH3 in the molecular diffusion
regime
15N2O and the signal with AMU 44 (mainly CO2) when
a mixture of 15NH3:O2:Ne = 1:2:1 was pulsed over the Pt
and Pt–Rh gauzes at 1073 K in Knudsen and molecular dif-
fusion regimes. The formation of nitrous oxide during am-
monia oxidation is a function of the peak pressure in the
TAP reactor. 15N2O was never detected when small pulses
of the ammonia–oxygen mixture were applied (in the Knud-
sen diffusion regime), while a weak signal at AMU 44 can
be observed. Contrarily, 15N2O is clearly observed when the
pulse size was increased in the molecular diffusion regime.
For both samples, the transient response of CO2 with AMU
44 is more intense and broader. Accordingly, the amount
of CO2 derived from carbon impurities in the commercial
gauzes (even after pretreatment in pure O2 at 1273 K) is suf-
ficient to mask the signal corresponding to N2O, although
both CO2 and N2O can be nicely uncoupled in mass spec-
trometry if isotopically labeled ammonia is applied.
Blank experiments in the TAP microreactor filled with
quartz particles were carried out in order to exclude N2O
formation as a consequence of a homogeneous process or
related to the reactor walls. The amount of N2O formed
in these experiments was 5 times lower N2O as compared
to experiments with the gauzes in Fig. 7. Accordingly, it
can concluded that the formation of N2O is a consequence
15NO was the main N product upon pulsing of a mix-
ture of 15NH3:O2:Ne = 1:2:1 over the Pt and Pt–Rh gauzes,
with a yield > 55 % (Table 2 and Fig. 7). As shown in Ta-
ble 2, the activity of inert SiO2 spheres is very low, revealing
the negligible contribution of gas-phase reactions under the
applied conditions in the TAP reactor. No dependence was
found between the yield of NO and the diffusion regime, i.e.,
Knudsen or molecular diffusion. This further supports that
NO formation purely stems from heterogeneous processes
occurring at the catalyst surface.
A relevant aspect of the experiments with 15NH3 under
molecular diffusion regime is the detection of nitrous oxide
as a reaction product (15N2O in Fig. 7). Isotopically labeled
15NH3 was used to discriminate between eventual formation
of 15N2O (AMU 46) and the always present background sig-
nal of CO2 (AMU 44). Fig. 8 shows the vital importance of
the pulse size and the application of the N-labeled ammo-
nia isotope in the TAP experiments to obtain, for the first
time, relevant mechanistic information about N2O forma-
tion during the high-temperature ammonia oxidation over
PGM gauzes. The figure shows the transient responses of
15
15
Fig. 8. Transient responses of N O and the signal with AMU 44 during single pulsing of an ammonia–oxygen mixture ( NH :O :Ne = 1:2:1) over Pt and
2
3
2
15
14
16
Pt–Rh gauzes at 1073 K. Pulse sizes of NH were ∼ 10 and ∼ 10 molecules in Knudsen and molecular diffusion regimes, respectively.
3

J. Pérez-Ramírez et al. / Journal of Catalysis 227 (2004) 90–100
97
5. Discussion
5.1. Primary interactions of ammonia and oxygen
The low activity of the as-received Pt and Pt–Rh gauzes
for ammonia decomposition at high temperatures (see Ta-
ble 1 and Fig. 1) is in good agreement with a previous
study by Bradley et al. [10] using molecular beams un-
der UHV at temperatures up to 400 K. These authors con-
cluded that NH3 did not dissociate over a hex-R Pt surface,
while the reaction proceeds very slowly over a clean (1 × 1)
Pt surface. In contrast with these results, Schmidt and co-
workers [32–34] determined the kinetics of ammonia de-
composition over different Pt specimens in the temperature
range of 500–1400 K, obtaining the following activity or-
der: polycrystalline > (210) > (110) > (100)-hex > (111).
The ability of polycrystalline Pt to dissociate ammonia was
found to be one order of magnitude higher than that of
Pt(111). This apparent controversy can be explained if tak-
ing into account that the works by Schmidt and co-workers
involved pretreatment of the Pt catalysts in an oxygen flow.
As we have demonstrated, the oxygen species formed during
O2 pretreatement are responsible for ammonia decomposi-
tion. In addition, the residence time in the TAP reactor was
104–105 times shorters than that applied in [34]. Accord-
ingly, a very slow ammonia activation process on essentially
reduced metal sites cannot be detected under TAP condi-
tions. Single pulsing of NH3 over the O2-pretreated Pt and
Pt–Rh gauzes leads to considerable amounts of N2, H2, and
H2O. This is also consistent with UHV studies on Pt(100)
[10], further supporting that adsorbed oxygen species dra-
matically enhance NH3 activation.
Fig. 9. N yield upon single pulsing of NO:Ne = 1:1 over as-received and
2
14
pretreated Pt and Pt–Rh gauzes. Pulse size of NO ∼ 10 molecules.
The formation of H2O and N2 in our experiments can be
explained by the simplified reaction scheme in Eqs. (3)–(8).
First, gas-phase oxygen and ammonia adsorb over an ac-
tive site, denoted as “s” [Eqs. (3) and (4)]. Based on low-
temperature UHV studies, a dual-site model has been pos-
tulated for NH3 oxidation over platinum, where ammonia
and oxygen adsorb on on-top and hollow sites, respec-
tively [9,10,15,23]. Since, no experimental evidence to sup-
port this model can be derived from our experiments, one
common adsorption site for both species has been consid-
ered. The primary step of ammonia decomposition is H strip-
ping from NH3 by adsorbed atomic oxygen species leading
to NH2 and OH fragments [Eq. (5)]. NH2 fragments can be
further dehydrogenated by adsorbed oxygen species (or hy-
droxyl groups) according to Eqs. (5)–(9). Recombination of
two surface nitrogen atoms [Eq. (10)] will ultimately result
in N2 formation [8–10,14].
of a heterogeneous process and the no observation in the
Knudsen regime can be related to the detection limit of
the analytical unit in relation to the low amount of N2O
formed.
4.4. Single pulsing of NO
Contrary to the results of single NH3 pulse experiments in
Section 4.1, the as-received or H2-pretreated gauzes show a
significant activity toward direct NO decomposition to N2
and O2 at 973–1173 K (Fig. 9). The N2 yield is signifi-
cantly higher over the H2-pretreated gauze, indicating the
affinity of NO for the reduced noble metal surface, in agree-
ment with previous NO dissociation studies over a clean
Pt(100) single crystal [9,31]. N2O was not observed as a
reaction product in the temperature range investigated. NO
decomposition over Pt and Pt–Rh gauzes was nearly sup-
pressed in the presence of adsorbed oxygen species in view
of the very low NO conversion over the O2-pretreated gauzes
(Fig. 9). Thus, the gauze activity for direct NO decomposi-
tion is a function of the degree of oxidation of the catalyst
surface.
NH3 + s −→ s–NH3,
(3)
(4)
(5)
(6)
(7)
O2 + 2s −→ 2s–O,
s–NH3 + s–O −→ s–NH2 + s–OH,
s–NH2 + s–O −→ s–NH + s–OH,
s–NH + s–O −→ s–N + s–OH,

98
J. Pérez-Ramírez et al. / Journal of Catalysis 227 (2004) 90–100
s–NH2 + s–OH −→ s–NH + s + H2O,
s–NH + s–OH −→ s–N + s + H2O,
s–N + s–N −→ N2 + 2s.
The formation of hydrogen during single pulsing of NH3
(Sections 4.1 and 4.2) suggest a relatively low oxygen cov-
erage of the surface as compared to the amount of ammonia
pulsed. Based on this, several additional reaction routes for
N2 and H2 formation can be suggested in Eqs. (11)–(14),
which should be favored in the presence of excess ammonia.
(8)
(9)
Nonetheless, from the results of single pulsing of NH3
over the O2-pretreated gauzes and sequential pulsing of O2
and NH3 at different time delays, it can be put forward that
not only the O coverage is essential for a high NO selec-
tivity, but also the nature of the adsorbed oxygen species.
In single ammonia pulsing over the oxidized gauzes, no NO
was detected, while the degree of NH3 conversion to N2 was
substantial (ca. 40%). A certain NO yield was observed in
sequential pulsing of oxygen and ammonia with ꢀt = 0.2
or 0.5 s, but not comparable to that obtained upon puls-
ing NH3–O2 mixtures (e.g., 1% vs 70% over Pt–Rh gauze
at 1073 K, see Fig. 6). Still, the degree of NH3 conversion
at different time delays in the range ꢀt = 0–2 s was very
similar (ca. 80%). Based on these data, we tentatively pro-
pose that strongly bounded oxygen species, which stay on
the catalyst surface after the O2 pretreatment, catalyze NH3
activation. However in view of the high N2 selectivity, the
stability of the O species on the metal should be such that the
recombination of dehydrogenated NHx fragments is much
more favorable that the O transfer into NHx to form NO. We
can exclude direct NO decomposition as a possible reaction
route for the high N2 production, since this process is sup-
pressed over O2-pretreated Pt surfaces (see Section 4.4).
In a simplistic way, Fig. 10 illustrates the proposed effect
of the coverage and nature of adsorbed O species on the NO
and N2 selectivity during the high-temperature NH3 oxida-
tion over PGM gauzes. Two different pools of adsorbed oxy-
gen species can be considered, which are characterized by
strong (O) or weak (O*) binding to the metal on the catalyst
surface. Strongly bounded O species are active for NH3 acti-
vation [Eqs. (5)–(7)], leading to N2, H2O, and H2. Based on
these results, water formation [Eqs. (8) and (9)] is believed to
be energetically favorable in comparison with removal of the
strongly boundedoxygen by ammonia fragments resulting in
NO formation. Therefore, a high NO selectivity requires not
only an optimal (high) ratio of O/NHx on the catalyst sur-
face, but also the presence of short-living and highly mobile
oxygen species.
(10)
s–NHx + s–NHx −→ N2 + xH2,
s–NH2 + s −→ s–NH + s–H,
s–NH + s −→ s–N + s–H,
s–H + s–H −→ H2 + 2s.
However, hydrogen was not observed upon pulsing of
a mixture of oxygen and ammonia over the Pt and Pt–Rh
gauzes, water being the only H-containing reaction prod-
uct. In these experiments, a high yield of NO was achieved
(Fig. 6). From these results, it can be concluded that the se-
lectivity of NH3 oxidation toward NO formation is increased
when the concentration of surface oxygen species (O cover-
age) is sufficiently high for striping all H atoms from NH3
and for further oxidation of the resulting NHx (x < 3) in-
termediates to NO [Eq. (15)]. The structure of the preferred
NHx intermediate has been subject of controversy and re-
mains unsolved.
(11)
(12)
(13)
(14)
s–NHx + ns–O −→ NO + s–HxO + (n − 1)s.
(15)
At low O coverages, the relative concentration of surface
NHx species should be obviously higher than the concen-
tration of surface O species. Therefore, recombination of
two highly reactive NHx fragments will prevail over the ox-
idation process in Eq. (15). Contrarily, intermediate NHx
species will be entrapped by oxygen species at high O cov-
erages, favoring the pathway toward NO formation.
Fig. 10. Influence of the oxygen coverage and the nature of adsorbed oxygen species on the product distribution during NH oxidation. O and O* denote
3
strongly bounded and weakly bounded oxygen species, respectively.

J. Pérez-Ramírez et al. / Journal of Catalysis 227 (2004) 90–100
99
5.2. Products formation in ammonia oxidation
expected that the rate of N2O formation is a stronger func-
tion of NH3 and NO partial pressures as compared to those
of NO and N2 formation. As a consequence, the extremely
low reactant peak pressures (1–5 Pa) in a Knudsen diffusion
regime are apparently suitable for a proper formation of NO
and N2 but insufficient for N2O. At the investigated temper-
atures in the TAP reactor, N2O is the N-containing product
formed in the lowest concentration, which is also the case in
industrial ammonia burners [1]. This reasoning can also ex-
plain the no identification of N2O as a reaction product in
surface science studies under ultrahigh vacuum conditions.
As noted in Section 2.2, the peak pressure in the TAP reactor
for a pulse size of 1016 molecules is considerably higher (ca.
109 times) than in UHV. Furthermore, these investigations
have been carried out with nonisotopically labeled ammo-
nia, which makes it impossible to accurately determine the
small amounts of N2O formed.
The shape of the transient responses of NO during O2–
NH3 interactions in sequential pulse experiments (Fig. 5)
clearly indicates that NO is a primary product of the high-
temperature reaction between surface NHx and O species
over Pt and Pt–Rh gauzes. The yield of NO was very sim-
ilar in Knudsen and molecular diffusion regimes (Table 2),
which further supports that gas-phase reactions play no role
in NO formation under the transient vacuum conditions of
the TAP reactor.
With respect to N2 formation, several reaction pathways
can be discussed. Apart from the recombination of NHx frag-
ments [Eq. (11)] at low O/NHx ratios (low O coverage or
excess of NH3) or in the presence of strongly bounded O
species (see Fig. 10), two other processes should be con-
sidered: (i) NO decomposition and (ii) a secondary process
involving NH3 and NO. Single pulsing of NO in Fig. 9 and
previous studies [9,31] have shown the high affinity of re-
duced noble metal surfaces for NO decomposition to N2
[Eq. (16)], with no formation of N2O. However, this pathway
is practically suppressed over oxidized noble metal surfaces,
thus having an insignificant contribution in NH3 oxidation
under excess of O2.
6. Conclusions
The TAP technique offers unique features for investigat-
ing the mechanism and kinetics of the high-temperature am-
monia oxidation:
Sequential pulsing experiments of O2 and NH3 in Fig. 5
have shown a very fast reaction of the formed NO at the time
of the NH3 pulse. As a result, the NO pulse strongly de-
creases and a broad N2 transient response is obtained. These
results strongly suggest that the reduction of NO with ad-
sorbed NHx species [Eq. (17)] is a major pathway to N2 in
ammonia burners, as concluded from UHV studies [9,10].
Selectivity-directing factors toward N2 formation will be fur-
ther assessed in an upcoming manuscript.
– With respect to steady-state techniques, the excellent
control of temperature due to the low amount of gases
pulsed (0.1–10 nmol), as well as the minor influence of
homogeneous processes and wall effects;
– With respect to surface science techniques in UHV, the
use of catalytic surfaces of industrial relevance and the
operation at a much higher peak pressure.
The following mechanistic aspects have been elucidated
from the investigation of primary NH3–O2 interactions over
commercial Pt and Pt–Rh gauzes at 973–1173 K:
2NO + 2s −→ N2 + s–O + s,
s–NHx + NO −→ N2 + s–HxO.
(16)
(17)
Finally, the formation of N2O during NH3 oxidation
should be discussed. As noted in the Introduction, N2O was
never detected as a product during NH3 oxidation over Pt
single crystals in surface science studies under UHV condi-
tions. Overcoming this limitation, our TAP study has demon-
strated N2O formation when a mixture of oxygen and am-
monia was pulsed over Pt and Pt–Rh gauzes in a molecular
diffusion regime. The amount of N2O formed is very small,
not being detected during ammonia oxidation in a Knudsen
diffusion regime, i.e., at low peak pressures. Associated with
the low N2O concentrations, it should be stressed that the
application of isotopically labeled ammonia (15NH3) was
essential in order to uncouple the analysis of N2O and CO2
(due to carbon impurities on the gauze surface) in mass spec-
trometry. The small amounts of N2O formed suggest that the
contribution of NH3 oxidation toward this product is negli-
gible. In fact, a recent TAP study has proven N2O results
from the reaction between adsorbed ammonia intermediates
and nitric oxide over Pt–Rh gauze [35]. Therefore it can be
– Overall, the mechanism of NH3 oxidation was found to
be very similar over gauzes having different composi-
tion (Pt or Pt–Rh) and geometry (knitted or woven).
– NH3 activation requires the presence of adsorbed oxy-
gen species on the catalyst surface. Reduced metal sur-
faces show no activity for ammonia decomposition.
– NO is a primary product of NH3 oxidation and its se-
lectivity is favored at high O coverage. Nonetheless,
the nature of oxygen species is also suggested to influ-
ence the product distribution of NO and N2. Strongly
bounded oxygen species activate ammonia and lead to
a high N2 selectivity by recombination of NHx species.
On the contrary, weakly bounded oxygen species are ef-
fectively transferred into the NHx intermediates yielding
NO. Accordingly, the highest NO yield is achieved when
NH3 and O2 are simultaneously pulsed over the Pt and
Pt–Rh gauzes.
– The selectivity toward NO does not strongly depend on
the diffusion regime (Knudsen or molecular), indicating

100
J. Pérez-Ramírez et al. / Journal of Catalysis 227 (2004) 90–100
that the desired product in the high-temperature am-
monia oxidation originates from purely heterogeneous
processes.
– Apart from the recombination of NHx fragments at low
surface O/NHx ratios as the source of N2, it can be
concluded that N2 also originates from the secondary
reaction of NH3 and NO.
[3] C.W. Nutt, S.W. Karup, Nature 224 (1969) 169.
[4] T. Pignet, L.D. Schmidt, J. Catal. 40 (1975) 212.
[5] J.L. Gland, V.N. Korchak, J. Catal. 53 (1978) 9.
[6] J.L. Gland, G.C. Woodward, J. Catal. 61 (1980) 543.
[7] M. Asscher, W.L. Guthrie, T.-H. Lin, G.A. Somorjai, J. Phys. Chem.
88 (1984) 3233.
[8] W.D. Mieher, W. Ho, Surf. Sci. 322 (1995) 151.
[9] J.M. Bradley, A. Hopkinson, D.A. King, J. Phys. Chem. 99 (1995)
17032.
– Direct decomposition of NO into N2 and O2 effectively
occurs over the reduced gauzes, but is practically sup-
pressed in the presence of adsorbed oxygen species.
– Small amounts of N2O were detected upon pulsing of
ammonia–oxygen mixtures over Pt and Pt–Rh gauzes.
To this end, the use of isotopically labeled 15NH3 and
high peak pressures (molecular diffusion regime) are
required. Blank experiments confirmed that N2O forma-
tion stems from a heterogeneous process. The no ob-
servation of N2O during ammonia oxidation under the
Knudsen diffusion regime and in previous UHV studies
is related to the detection limit of the analysis.
[10] J.M. Bradley, A. Hopkinson, D.A. King, Surf. Sci. 371 (1997) 255.
[11] L. Andrussow, Z. Angew. Chem. 39 (1926) 321.
[12] M. Bodenstein, Z. Elektrochem. 41 (1935) 466.
[13] F. Raschig, Z. Angew. Chem. 40 (1927) 1183.
[14] M. Kim, S.J. Pratt, D.A. King, J. Am. Chem. Soc. 122 (2000) 2409.
[15] N. Materer, U. Starke, A. Barbieri, R. Döll, K. Heinz, M.A. Van Hove,
G.A. Somorjai, Surf. Sci. 325 (1995) 207.
[16] R.W. McCabe, T. Pignet, L.D. Schmidt, J. Catal. 32 (1974) 114.
[17] M. Flytzani-Stephanopoulos, S. Wong, L.D. Schmidt, J. Catal. 49
(1977) 51.
[18] M.R. Luibovskii, V.V. Barelko, Kinet. Katal. 35 (1994) 412.
[19] M. Sheituch, L.D. Schmidt, J. Phys. Chem. 92 (1988) 3404.
[20] L. Lobban, G. Philippou, D. Luss, J. Phys. Chem. 93 (1989) 733.
[21] G. Philippou, D. Luss, Chem. Eng. Sci. 48 (1993) 2313.
[22] E.V. Rebrov, M.H.J.M. de Croon, J.C. Schouten, Catal. Today 69
(2001) 183.
Acknowledgments
[23] E.V. Rebrov, M.H.J.M. de Croon, J.C. Schouten, Chem. Eng. J. 90
(2002) 61.
[24] J.T. Gleaves, J.R. Ebner, T.C. Kuechler, Catal. Rev.-Sci. Eng. 30
(1988) 49.
[25] J.T. Gleaves, G.S. Yablonsky, P. Phanawadee, Y. Schuurman, Appl.
Catal. A 160 (1997) 55.
[26] O. Levenspiel, in: Chemical Reaction Engineering, Wiley, New York,
1999, p. 668.
[27] M. Soick, D. Wolf, M. Baerns, Chem. Eng. Sci. 55 (2000) 2875.
[28] S.O. Shekhtman, G.S. Yablonsky, S. Chen, J.T. Gleaves, Chem. Eng.
Sci. 54 (1999) 4371.
[29] R.J. Farrauto, C.H. Bartholomew, in: Fundamentals of Industrial Cat-
alytic Processes, Chapman & Hall, London, 1997, p. 481.
[30] P.A. Kozub, G.I. Gryn, I.I. Goncharov, Platinum Metals Rev. 44 (2000)
77.
Yara International ASA is acknowledged for financial
support and for permission to publish the results with Pt–Rh
gauze. V.A.K. and M.B. are thankful for financial support
from the Deutsche Forschungsgemeinschaft (DFG) in the
frame of the competence network “Bridging the gap between
real and ideal systems in heterogeneous catalysis SPP 1091.”
The authors are indebted to L. Mader for technical assistance
in the TAP experiments.
References
[31] R.J. Gorte, L.D. Schmidt, J.L. Gland, Surf. Sci. 109 (1981) 367.
[32] D.G. Löffler, L.D. Schmidt, J. Catal. 41 (1976) 440.
[33] D.G. Löffler, L.D. Schmidt, Surf. Sci. 59 (1976) 195.
[34] G.A. Papapolymerou, L.D. Schmidt, Langmuir 1 (1985) 488.
[35] J. Pérez-Ramírez, E.V. Kondratenko, Chem. Commun. (2004) 376.
[1] J. Pérez-Ramírez, F. Kapteijn, K. Schöffel, J.A. Moulijn, Appl. Catal.
B 44 (2003) 117.
[2] Y.M. Fogel, B.T. Nadytko, V.F. Rybalko, V.I. Shvachko, I.E. Ko-
robchanskaya, Kin. Katal. 5 (1964) 496.