Activation of ammonia dissociation by oxygen on platinum sponge studied with positron emission profiling

Article abstract of DOI:10.1016/S0021-9517(03)00191-X

Positron emission profiling (PEP) was applied to study the adsorption and dissociation of ammonia on metallic and preoxidized Pt sponge at 323-573 K. Under 1 atm, a Pt surface precovered with oxygen initiated the ammonia dissociation leading to the gaseous nitrogen-containing products. Below 423 K, the pulse experiments resulted in the formation of mainly nitrogen and a small amount of N₂O. Above 423 K, the product selectivity was changed and NO was formed. With PEP, ammonia reacted at the beginning of the catalyst bed and that a part of [¹³N]NH₃ remained at the surface as ¹³ species. The NO flush experiments indicated that NO selectivity reacted with the NH species to form N₂O. Nitrogen was preferably formed at the lower oxygen surface coverage.

Full text of DOI:10.1016/S0021-9517(03)00191-X

Journal of Catalysis 219 (2003) 156–166
www.elsevier.com/locate/jcat
Activation of ammonia dissociation by oxygen on platinum sponge
studied with positron emission profiling
a,∗
a,b
a
a
D.P. Sobczyk, A.M. de Jong, E.J.M. Hensen, and R.A. van Santen
a
Schuit Institute of Catalysis, Laboratory of Inorganic Chemistry and Catalysis, Eindhoven University of Technology,
PO Box 513, 5600 MB Eindhoven, The Netherlands
Department of Applied Physics, Centre of Plasma and Radiation Physics, Eindhoven University of Technology,
b
PO Box 513, 5600 MB Eindhoven, The Netherlands
Received 3 December 2002; revised 4 April 2003; accepted 4 April 2003
Abstract
Positron emission profiling (PEP) is applied to study the adsorption and dissociation of ammonia on metallic and preoxidized platinum
sponge in the temperature range 323–573 K. The results show that ammonia weakly interacts with platinum without dissociation. On the con-
trary, preoxidized platinum is able to dissociate ammonia. With increasing temperature more ammonia is converted into N , N O, and H O.
2
2
2
Adsorbed NO appears to be an important intermediate, while its formation strongly depends on the oxygen surface coverage. Temperature-
programmed experiments are performed to characterize the adsorbed nitrogen species. Furthermore, H , NH , and NO are used to react with
2
3
these adsorbed nitrogen species. These experiments indicate that mainly NHx species are present at the platinum surface.
2003 Elsevier Inc. All rights reserved.

13
13
Keywords: NH ; [ N]NH ; Ammonia oxidation; Ammonia dissociation; Ammonia adsorption; Platinum; Positron emission profiling
3
3
1
. Introduction
reported for chemisorbed ammonia an adsorption energy
of −108 kJ/mol based on theoretical calculations. Ammo-
nia dissociates into nitrogen and hydrogen at temperatures
above 530 K (135 kJ/mol) [3,6,9]. The ammonia dissoci-
ation can be enhanced by preadsorption of oxygen [1,3],
which is explained by the fact that atomic oxygen activates
the NHx bond cleavage [4]. The various formed fragments
are adsorbed on platinum in different configurations: NH3,
Low-temperature selective oxidation of ammonia with
oxygen, to nitrogen and water, forms a potential solution
to several ammonia spills. Theoretical calculations and ultra
high vacuum (UHV) experiments provided many insights in
the interactions of ammonia with oxygen on platinum [1–5].
However, the adsorption of ammonia and its subsequent ox-
idation have not been extensively studied at atmospheric
pressure. Our transient ammonia pulse experiments stud-
ied in situ with positron emission profiling (PEP) may offer
valuable information on the ammonia adsorption and dis-
sociation activated by oxygen on pure platinum catalysts.
These are the first elementary steps in the low-temperature
ammonia oxidation.
On clean Pt single crystals three molecular ammo-
nia desorption states are observed, at low coverage the
strongly chemisorbed ammonia desorbs around 400 K,
weakly chemisorbed at 150 K, and ammonia physisorbed in
the multilayer at 100 K [1,3,6–8]. Fahmi and van Santen [4]
OH species (1-fold); NH and NH adsorbates (2-fold); H,
2
O, and N atoms (3-fold) [4]. NO preferentially adsorbs at
the fcc hollow site at low coverages, at higher coverages
NO additionally adsorbs on the atop site in a tilted geome-
try [10,11]. Oxygen molecules occupy hollow sites [12–15]
whereas ammonia molecules occupy on-top sites [4,16]. The
preadsorption of oxygen does not block the adsorption of
ammonia [3]. The product formation of the reaction be-
tween ammonia and oxygen is dependent on the reaction
conditions, i.e., pressure, reactant ratio, temperature, and the
exposed Pt surface. The group of King proposed that NO
is the key intermediate [3,4,17] and that the reaction path-
way is dependent on the oxygen surface coverage. However,
the experiments done at atmospheric pressure by Van den
Broek [18] indicated that the reaction between the NH and
*
Corresponding author.
E-mail address: d.p.sobczyk@tue.nl (D.P. Sobczyk).
0021-9517/$ – see front matter  2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0021-9517(03)00191-X

D.P. Sobczyk et al. / Journal of Catalysis 219 (2003) 156–166
157
3
the OH species is the rate-determining step. These species
have also been detected at the platinum surfaces under UHV
conditions [1].
was then flushed with He (48 cm /min) for 20 min before
the reaction temperature was set.
A preoxidized platinum sponge is obtained by the fol-
lowing treatment: the reduced catalyst was pretreated with
1 vol% O /He flow (48 cm /min) for 1 h at 373 K followed
A suitable technique to study in situ transient and steady-
state phenomena at atmospheric pressure is positron emis-
sion profiling [19,20]. The PEP measurements offer a unique
possibility of studing the various processes at the catalyst
surface in a wide temperature range. PEP is a technique
3
2
3
by flushing with He (48 cm /min) for 1 h before the reaction
1
5
temperature was set. Additional [ O]O2 PEP experiments
indicated (not shown) that on preoxidized platinum sponge
oxygen is strongly bound and that oxygen adsorbs dissocia-
tively at this temperature, which is in line with earlier reports
−
15
in which minute (∼ 10
mol) quantities of radiolabeled
molecules emitting positrons are injected as pulses into the
feed of packed-bed reactors. The concentration distribution
of radiolabeled molecules can be measured as a function of
[
21–25].
time and position within a tubular reactor bed. In this study
we used 13N-labeled ammonia pulses.
2.2. Positron emission profiling
In this work, evidence for the activation of the ammo-
nia adsorption and dissociation promoted by oxygen on
platinum is provided by PEP pulse experiments. The con-
The 13N nucleus emits a positron upon decay. Positron
emission profiling is based on the detection of the two
1
3
version and product formation due to [ N]NH3 pulses on
preoxidized platinum have been studied as a function of
temperature. Here, we will demonstrate that the product se-
lectivity is temperature dependent and that a part of the
5
11 keV gamma photons which originate from the annihi-
lation of this positron with an electron. These two 511 keV
gamma photons are simultaneously emitted in opposite di-
rections and they travel typically a few centimeters in solid
matters. Coincidental detection of two photons by scintilla-
tion detectors (BGO) provides the position of the annihila-
tion. In practice, the tubular reactor is horizontally placed
between two arrays (upper and lower) of nine scintillation
detectors. In both arrays the detectors were tightly placed,
which results in the spatial resolution of 2.9 mm. The av-
erage concentration of all radiolabeled molecules within a
certain volume (length 2.9 mm) is measured at time inter-
vals of 0.5 s. The measurement of radiolabeled molecules
is simultaneous over the total detection length (5 cm), thus
within the 17 vol segments. In this way the concentration
distribution of the positron emitting molecules can be mea-
sured as a function of position and time [19,29].
1
3
injected [ N]NH3 remains at the preoxidized platinum sur-
face. In addition, the nature of the nitrogen species, which
remain after the ammonia pulse on preoxidized platinum
in the temperature range up to 423 K, is further investi-
gated with temperature-programmed desorption (TPD) and
temperature-programmed oxidation (TPO), and the removal
of the adsorbed species by H2, NH3, and NO reaction exper-
iments. Furthermore, we discuss the possible role of NO as
a reaction intermediate.
2
. Experimental
2
.1. Catalytic reactor
The Eindhoven 30 MeV cyclotron was used to irradi-
ate a water target with highly energetic protons of 16 MeV.
The irradiation time was 10 min and a typical beam cur-
rent of 500 nA was used. The target was a flowthrough
water target, with a total volume of 7 ml containing a dual
foil (Duratherm 600, thickness 15 µm). In this way formed
The platinum sponge was acquired from Johnson Mattey.
The sponge sample was of > 99.9% purity. The particle size
of the sponge was between 250 and 350 µm and the size of
the small non porous particles about 1.0–5.0 µm [18]. The
amount of platinum sites calculated via BET measurement
1
3
13
1
8
[ N]nitrate and [ N]nitrite were subsequently reduced to
is 3.0 × 10 sites/g and via the hydrogen chemisorption is
1
3
1
8
3
NH , using DeVarda’s alloy method [26,27]. The produc-
1
0
.3×10 sites/g. The metal surface area is determined to be
1
3
2
tion method of gaseous pulses of [ N]NH was described
3
.099 m /g. Atmospheric ammonia oxidation activity tests
elsewhere [28]. A pulse time of 10 s was used to inject a
were performed in a fixed-bed reactor setup equipped with
a quadrupole mass spectrometer (Balzers Instruments Om-
nistar GSD 3000), which was calibrated for on-line analysis
of reactants and products. A quartz tube with an internal di-
ameter of 4 mm was used as reactor. A sample of 1.8 g of
pure platinum sponge was used in a catalytic bed with a
length of 4.0 cm. The experiments were done at tempera-
tures between 323 and 673 K.
mix of 13NH3/ NH3 into the reactant stream, henceforth
14
1
3
14
−5
indicated as [ N]NH3. In this way NH3 (∼ 10 mol)
1
3
was measured with the mass spectrometer and NH3
−15
13
(
∼ 10
mol) was monitored with PEP. Since the NH3
concentration varies in each pulse, the absolute concentra-
tion of radiolabeled molecules in the various experiments
cannot be directly compared. However, the labeled species
are either introduced in a large flow of nonlabeled ammonia
or as trace amounts in ammonia-free flows. As such, these
variations will not affect the results between the various ex-
periments.
Prior to a catalytic experiment the platinum sponge was
reduced in situ by heating the sample in a 10 vol% H2/He
3
flow (40 cm /min) from 298 to 673 K. Subsequently the
sample was kept at this temperature for 2 h. The catalyst

158
D.P. Sobczyk et al. / Journal of Catalysis 219 (2003) 156–166
2
.3. PEP pulse experiments
The reduced platinum catalyst was kept under a flow
3
(
48 cm /min) of 1.0 vol% ammonia/helium or 1.0 vol%
oxygen/helium or He flow at 273–573 K. Subsequently, a
1
3
pulse of [ N]NH3 was injected in this ammonia/helium or
oxygen/helium flow or helium flow. A gas hourly space ve-
−1
13
locity (GHSV) of 5700 h is mostly used. The [ N]NH3
pulse experiments were performed over the platinum and
preoxidized platinum sponge in a similar manner. In each
PEP experiment mass spectrometry was used to simultane-
ously monitor the effluent products.
2
.4. Temperature-programmed experiments
Temperature-programmed experiments were performed
exactly 170 s (TPD) or 100 s (TPO) after the injection of
1
3
3
a pulse of [ N]NH3 in He flow of 40 cm /min (GHSV =
−1
5
700 h ) at 373 K over the preoxidized platinum catalyst.
The temperature was raised with 10 K/min (40 cm /min)
under He (TPD) or 1.0 vol% O2/He (TPO) flow.
3
2
.5. Removal of adsorbed species by H2, NH3, and NO
The reaction experiments were performed more than
1
3
3
00 s after the injection of a pulse of [ N]NH3 in He flow
3
−1
of 48 cm /min (GHSV = 5700 h ) over the preoxidized
platinum catalyst at 323–348 K. In these experiments the
He carrier flow was replaced by either 1.0 vol% NH3/He
3
or 4.0 vol% H2/He or 0.5 vol% NO/He flow (48 cm /min).
3
. Results and discussion
3
.1. Ammonia adsorption and dissociation on platinum and
preoxidized platinum
3
.1.1. Activation of ammonia adsorption by oxygen
Fig. 1A shows that an injected radiolabeled ammonia
pulse into a NH3/He flow (323 K) travels through the plat-
1
3
inum catalyst bed. The retention time of [ N]NH3 through
the catalyst bed was approximately 9 s. This is longer than
the retention time of He (less than a second). At higher tem-
peratures this ammonia retention time decreases, giving a
value of 5 s at 423 K. Judging from the absence of broad-
ening of the pulse, we expect that diffusion limitations of
ammonia in the sponge material are negligible [30]. Fig. 1A
shows that the ammonia adsorption equilibrium is fast and
indicates a weak molecular adsorption of ammonia.
The injection of [ N]NH3 into an ammonia-free He flow
over the reduced platinum sponge (Fig. 1B) clearly shows
that injected ammonia also travels through the catalyst bed
in this case. Gas-phase analysis showed that only NH3 is de-
tected at the reactor outlet, which indicates that ammonia
does not dissociate under these conditions which is in line
with earlier reports [1,2,31]. Therefore, we conclude that the
13
Fig. 1. PEP image of a pulse injection of [ N]NH at 323 K (A) in a 1 vol%
3
3
3
NH3/He flow (48 cm /min) on Pt sponge; (B) in He flow (48 cm /min)
on Pt sponge; (C) in He flow (48 cm /min) on the preoxidized Pt sponge.
The color intensity represents the concentration of NH (dark = high
3
13
3
concentration).
PEP image shows the weak binding of ammonia on plat-
inum. A relatively small amount of the labeled ammonia
species (15%) remained adsorbed at the platinum surface.
This adsorption is most probably caused by the incomplete
reduction of the platinum surface.
The in situ PEP technique allows visualization of the
promoting effect of oxygen in ammonia adsorption and dis-
sociation. In contrast to Fig. 1B, the adsorption of ammonia
1
3

D.P. Sobczyk et al. / Journal of Catalysis 219 (2003) 156–166
159
1
3
3
Fig. 2. PEP images of an pulse injection of [ N]NH in 1.0 vol% O /He flow (48 cm /min) on preoxidized Pt sponge at (A) 323, (B) 423, and (C) 573 K;
3
2
13
13
(
D) integrated N concentration as function of time for presented temperatures. The color intensity represents the concentration of NH .
3
on the preoxidized surface (Fig. 1C) is strong. The 13N
species are adsorbed already at the beginning of the catalyst
bed and do not desorb from the surface during the exper-
iment. This indicates that the presence of oxygen leads to
enhanced dissociation of ammonia on Pt. This agrees with
recent calculations performed by Fahmi and van Santen [4],
who calculated that adsorbed atomic oxygen activates the
N–H bond cleavage. Accordingly, we propose the following
mechanism for the NH3 conversion:
3.1.2. Influence of temperature on the ammonia
dissociation on preoxidized Pt
The PEP images of the radiolabeled ammonia pulses on
preoxidized platinum change significantly with increasing
1
3
temperature. Fig. 2 shows the PEP images of [ N]NH3
pulses over preoxidized platinum in an oxygen flow in the
temperature range of 323–573 K including the integrated
1
3
N concentration in the reactor at different temperatures
as a function of time. At 323 K almost no gaseous prod-
ucts are formed and the total concentration of 13N remains
at the catalyst (Figs. 2A and 2D). Below 423 K the PEP im-
ages are very similar and the 13N concentration is observed
at the beginning of the catalyst bed and the total radioac-
tivity profile (Fig. 2D) shows a sharp decrease as gaseous
products are formed. The decrease of the 13N concentration
at 423 K is due to the formation of nitrogen and some ni-
trous oxide. Fig. 3A provides evidence for the formation of
N and N O by MS analysis. These products evolve simulta-
NH3(a) + O(a) → NH2(a) + OH(a),
NH2(a) + O(a) → NH(a) + OH(a),
NH + O(a) → N(a) + OH(a).
(1)
(2)
(3)
As we did not detect dinitrogen at the relatively low tempera-
ture of 323 K, we expect that atomic nitrogen is not formed.
However, due to the high surface coverage of the various
species (OH, NH2, NH, NO, N) diffusion may be impaired,
thereby hindering reconstruction to N2.
2
2
1
4
13
neously as a response to the NH / NH pulse. The water
3
3
formation is detected with an obvious delay compared to N2
and N O evolution. This stems from the stronger interac-
2
In paragraphs 2 and 3 the nature of these adsorbed species
is investigated. The role of NO in the product formation is
mainly addressed in paragraph 3. First, we present the influ-
ence of the temperature on the ammonia dissociation on the
preoxidized platinum sponge.
tion of these water molecules with the catalytic surface. As
can be observed when comparing Figs. 3A and 3B this ad-
sorption of water is less pronounced at higher temperatures.
At a temperature of 573 K the PEP image is substantially
changed (Fig. 2C). A reaction zone is still observed at the

160
D.P. Sobczyk et al. / Journal of Catalysis 219 (2003) 156–166
Fig. 3. Formation of N , N O, NO, and H O measured by online mass
2
2
2
13
spectrometry on a response of [ N]NH injection in 1 vol% O /He flow
48 cm /min) on the preoxidized Pt sponge at (A) 423 and (B) at 573 K.
3
2
3
(
beginning of the catalyst bed, but in contrast to the obser-
vation at 423 K not all the products leave the catalyst bed
directly. The formation of NO explains the changes in the
PEP image, because NO readsorbs at the platinum surface
throughout the catalyst bed. This is not the case for N2 and
N2O, which have a very low adsorption constant. Consis-
13
Fig. 4. A [ N]NH pulse was adsorbed on the preoxidized platinum sponge
3
tently, Fig. 2D shows a relatively slow decrease of the 13
N
3
kept under He flow (40 cm /min) at 373 K followed after 170 s by a tem-
perature-programmed desorption experiment: (A) TPD spectrum; (B) PEP
image, the color intensity represents the concentration of NH ; (C) nor-
concentration from the catalyst due to the NO readsorption.
At this relatively high temperature almost all injected ammo-
nia reacts and forms gaseous products as shown in Fig. 2D.
In conclusion, the results indicate that the relatively high
oxygen coverage on platinum promotes NO(a) formation. At
low temperatures NO(a) leads to N2O via
13
3
malized concentration as function of time (T = 373–573 K, 10 K/min, He
3
flow of 40 cm /min).
3
.2. Temperature-programmed measurements
NO(a) + N(a) → N2O(g) + 2∗.
3.2.1. Temperature-programmed desorption
(4)
In this experiment the temperature is raised from 373 to
1
3
However, at temperatures higher than 573 K the NO sur-
face coverage is low, due to desorption of NO, resulting in a
lower N2O formation. The formation of nitrogen is assumed
to proceed via atomic nitrogen originating from NHx species
rather than via the dissociation of NO(a). The dissociation of
NO(a) at low temperature (< 400 K) and also in the situation
of a high adsorbate coverage is unfavorable [2,32–34].
573 K after injection of a [ N]NH pulse on the catalyst at
3
373 K (Fig. 4). The observed background of the MS sig-
nals of the products are relatively high since the catalyst
could not be sufficiently flushed with He before starting a
TPD experiment due to a relatively fast 13N decay. The de-
sorption of the inert N2 and N2O molecules is qualitatively
observed in the PEP images (Fig. 4C). The signal of the in-

D.P. Sobczyk et al. / Journal of Catalysis 219 (2003) 156–166
161
tegrated 13N concentration decreases at the same time as N2
and N2O desorb. The desorption of N2 and N2O continues
up to 480 K, because at that temperature the total radioactiv-
ity profile (Fig. 4C, from 800 s) shows no further decrease of
1
3
the N concentration. N2 and N2O have low adsorption con-
stants [35,36], meaning that were not present at first but are
formed during the TPD experiment. The formation of N2O
is assigned to reaction (4). Since radiolabeled ammonia was
injected on the preoxidized surface the oxygen coverage is
relatively high and therefore the dissociation of NO on plat-
inum is unfavorable. This leads to the assumption that N(a)
species originate from adsorbed NHx species. Another im-
portant feature that supports the presence of the NHx species
on the surface is substantial formation of water, which is
found to be accompanied by the N2 and N2O formation.
NO(a) needed for the N2O formation originates from NHx
species:
NH(a) + 2OH(a) → NO(a) + H2O(a) + H(a).
(5)
At lower temperatures NO(a) does not dissociate and is
preferentially converted into N2O.
1
3
Above 480 K about 20% of N is still adsorbed at the
platinum surface (Fig. 4C, at 800 s). In fact the 13N inte-
grated concentration profile should change to a horizontal
line, but the radioactivity profile slightly increases. This is
caused by spreading of the radioactive compound through
the catalyst bed, while the total amount of labeled species in
the reactor remains constant. When all these species are con-
centrated in the first part of the reactor and thus at the edge
of the surrounding detector, the detection efficiency of the
total amount of radioactive material is somewhat less then
when it would be concentrated in the middle. This mainly af-
fects the integrated profiles and not the PEP plots that show
the radioactive concentration for each position as function
of time. This profile of total radioactivity increase must be
taken into account only in the TPD or TPO experiments and
especially when the still high concentration of the adsorbed
1
3
N spreads from one position to subsequent positions. Still,
1
3
Fig. 5. A [ N]NH pulse was adsorbed on the preoxidized platinum sponge
3
the qualitative value of 20% is correct, because this value
is in accordance with the value found in the radioactivity
profile of the first position in the catalyst bed, which is not af-
fected by the above-described artifact. After the temperature
has reached 520 K the total radioactivity profile (Fig. 4C)
decreases for the second time. The PEP image (Fig. 4B)
3
kept under He flow (40 cm /min) at 373 K followed after 100 s by a tem-
perature-programmed oxidation experiment: (A) TPO spectrum; (B) PEP
image, the color intensity represents the concentration of NH ; (C) nor-
13
3
malized concentration as function of time (T = 373–573 K, 10 K/min,
3
1
vol% O /He flow of 40 cm /min).
2
1
3
shows that the N concentration is slowly moving through
the catalyst bed, indicating that readsorption is taking place.
The velocity of the 13N moving through the catalyst bed
is not linear because the desorption is enhanced by the in-
creasing temperature. Thus the 1 N transport through the
reactor is accelerated in time. At the moment that the ac-
tivity reaches the end of the catalyst bed (T = 520 K) only
NO is detected with MS. For that reason the desorption of
shows that NO starts to desorb (moves to subsequent posi-
tions) around 440 K.
3.2.2. Temperature-programmed oxidation
3
13
In this experiment a [ N]NH3 pulse was adsorbed on the
preoxidized platinum sponge kept under He flow at 373 K,
and then the temperature is raised from 373 to 573 K in
1 vol% O2/He flow of 40 cm /min (Fig. 5). The results
from TPO experiments are in qualitative agreement with
those of TPD experiments. In the first minutes of the TPO
3
1
3
13
N is assigned to NO. In ultrahigh vacuum experiments
on single crystals the energy for NO desorption is found
to be around 150 kJ/mol [2,37,38] and NO desorbs at tem-
peratures around 400 K. In our experiment, the PEP image
1
3
experiment (up to 423 K) N species desorb, which can
be observed in Fig. 5C as the integrated 13N concentration

162
D.P. Sobczyk et al. / Journal of Catalysis 219 (2003) 156–166
slowly decreases. This 13N desorption is assigned mainly
to nitrogen formation (Fig. 5A). The background signal of
the measurement for nitrogen and water is high and there-
fore quantitative interpretation is not possible. Hardly any
N2O is formed, which is in contrast with the TPD exper-
iment. Fig. 5C shows that the nitrogen desorption ends at
4
00 s from the start of the experiment, which corresponds
in Fig. 5A to a temperature of 420 K. At that temperature,
0% of the injected 13N is still present at the surface. As
3
noted before, the increase in radioactivity is due to a detector
artifact. Fig. 5A shows that in the presence of gas-phase oxy-
gen, NO is the major species desorbing at 540 K. The actual
desorption temperature of NO is 440 K, since at this tem-
perature the 13N concentration starts to move in the catalyst
bed. In comparison with the TPD experiment NO desorbs
at a similar temperature, which is not surprising because in
both experiments platinum was precovered with oxygen. In
correspondence with Fig. 4B, Fig. 5B shows that equilib-
rium of 13NO species is controlled by the readsorption on
the platinum surface. As already noted the NO transport is
accelerated with increasing temperature. In this TPO exper-
iment all adsorbed 1 N species are removed.
3
The increase of the temperature up to 400 K activates
the bond cleavage of NHx by atomic oxygen according to
reactions (2) and (3). Again, the formation of nitrogen is as-
sumed to proceed via recombination of atomic nitrogen. At
high oxygen surface coverage NO can be formed via
N(a) + O(a) → NO(a) + ∗,
(6)
(7)
∗
NH(a) + 2O(a) → NO(a) + OH(a) + .
Although desorption of NO is favorable above 400 K, NO
formation occurs already at lower temperatures [2]. This
can also be observed indirectly in our experiment. Nitro-
gen is largely desorbed at 420 K; however, still some 13
N
species are retained at the surface. Obviously, above 420 K
the NHx +O(a) reaction does not proceed, and nitrogen is not
produced, which suggests the absence of NHx species on the
surface. On the other hand, the formation of NO explains
the retained 13N species. NO is unreactive with respect to
oxygen and it will remain adsorbed on platinum till the de-
sorption enthalpy is overcome.
1
3
Fig. 6. A [ N]NH pulse was adsorbed on a preoxidized platinum sponge
3
3
kept under He flow (48 cm /min) followed after 380 s by a removal of
N species with hydrogen: (A) MS spectrum, shown from the moment of H
addition; (B) PEP image, the color intensity represents the concentration
2
13
of NH ; (C) normalized concentration as function of time (T = 348 K,
3
3
1
0 vol% H /He flow of 48 cm /min).
2
3
3
.3. Reaction of adsorbed N species with H2, NH3, and NO
hydrogen first reacts with oxygen forming hydroxyl groups.
OH groups do not only form water but they also react with
the adsorbed NHx species to produce N2:
.3.1. Reaction with hydrogen
1
3
A [ N]NH3 pulse was adsorbed on a preoxidized plat-
inum sponge kept under He flow at 348 K followed by a
removal of the adsorbed species with hydrogen (Fig. 6).
Fig. 6A shows that the replacement of He by H2 results
in N2O, N2, and H2O formation. It is clear from Figs. 6B
and 6C that almost all adsorbed nitrogen species are re-
moved from the platinum surface in the presence of H2.
This means that the activation energies for the surface re-
actions to form N2O and N2 are relatively low, since already
at 348 K all nitrogen species are removed. It is assumed that
NH2(a) + OH → NH + H O ,
(a)
(a)
2
(a)
(8)
(9)
NH(a) + OH(a) → N(a) + H2O(a),
∗
2
N(a) → N2 + 2 .
(10)
However, higher OH concentrations favor the formation of
NO:
NH(a) + 2OH(a) → NO(a) + H2O(a) + H(a).
(11)

D.P. Sobczyk et al. / Journal of Catalysis 219 (2003) 156–166
163
NO cannot desorb at this low temperature. Instead it reacts
with N(a) to form N2O. NO could also be an intermediate
leading to N2 formation:
NHx(a) + NO(a) → N2(a) + OHx(a)
.
(12)
However, this reaction can only occur at relatively low oxy-
gen surface coverage because the nitrogen/oxygen bond of
NO needs to be broken. For that reason this route of the ni-
trogen formation can be activated after the initial desorption
of N2O and H2O from the surface. Eventually, platinum is
reduced and the oxygen surface coverage is lowered.
3
.3.2. Reaction with ammonia
In this experiment, a [ N]NH3 pulse was adsorbed on the
1
3
preoxidized platinum sponge kept under He flow at 348 K,
followed by a removal of the adsorbed species with ammo-
nia (Fig. 7). Upon addition of ammonia to the He flow N2,
N2O, and H2O are observed at the reactor outlet (Fig. 7A).
Figs. 7B, and 7C show that all 1 N species are instanta-
neously removed from the surface. Similar to the hydrogen
experiment, OH groups are produced followed by the de-
sorption of water. These hydroxyl groups also react with
3
1
3
NHx species to form adsorbed NO and eventually dini-
trogen. The amount of N2O formed is much lower than
the amount of N2, although it agrees well with the amount
formed in the hydrogen experiment. In contrast to nitrogen
and water, N2O is only formed in the first seconds after the
replacement of He by the ammonia flow. This suggests that
in the beginning of the ammonia addition NO(a) is formed,
which is similar to the observations with the H2 flush exper-
iment. Obviously, nitrogen and water are not only produced
from the adsorbed 13N species, but mainly originate from
gas-phase ammonia. The addition of ammonia results in a
high concentration of the nitrogen-containing species on the
platinum surface. The high concentration of ammonia favors
selective N2 formation, which is also observed in UHV ex-
periments [2,6]. The production of N2 and H2O decreases
in time because less oxygen is available for the reaction.
This results in an oxygen-freesurface, which is confirmed by
1
3
measuring an additional PEP image after a [ N]NH3 pulse
in this ammonia flow (not shown). This image was identical
to Fig. 1A showing the equilibrium of ammonia on platinum.
1
3
Fig. 7. A [ N]NH pulse was adsorbed on the preoxidized platinum sponge
3
3
kept under He flow (48 cm /min) followed after 350 s by a removal
of N species with ammonia: (A) MS spectrum, shown from the moment
of NH addition; (B) PEP image, the color intensity represents the con-
centration of NH ; (C) normalized concentration as function of time
3
3
.3.3. Reaction with NO
The role of NO as an reaction intermediate is further in-
1
3
3
vestigated in an experiment, in which a [ N]NH3 pulse was
adsorbed on the preoxidized platinum sponge kept under
He flow at 323 K, followed by a removal of the adsorbed
species with nitric oxide (Fig. 8). Figs. 8B, and 8C show
13
3
(
T = 348 K, 1.0 vol% NH /He flow of 48 cm /min).
3
that upon addition of an NO flow about 50% of the 13
N
the effect of the high surface coverage most probably can be
excluded even though NO preferably adsorbs on the same
sites as oxygen (hollow sites), while ammonia adsorbs on
the atop sites. The second explanation is that NO selectively
reacts with one of the adsorbed N species. Only N2O is
formed as a result of the NO flush (Fig. 8A). High oxy-
gen surface coverage, together with high partial pressure of
species are instantaneously removed from the platinum sur-
face. This partial conversion has two possible explanations.
First, due to inaccessibility or a high surface coverage, NO
does not react with all of the adsorbed 13N species. This ex-
planation is in contrast with the experiments with hydrogen
and ammonia, where all 13N species were removed. Thus
13

164
D.P. Sobczyk et al. / Journal of Catalysis 219 (2003) 156–166
13
Fig. 9. A [ N]NH pulse was injected on the preoxidized platinum sponge
3
kept under NO/He flow at 323 K. After 400 s temperature was increased
with 15 K/min (A) MS spectrum; (B) PEP image, the color intensity repre-
13
3
sents the concentration of NH (0.5 vol% NO/He flow of 48 cm /min).
3
1
3
Injecting an additional pulse of [ N]NH3 in the NO flow,
under the same reaction conditions, 40 min after the first
pulse (Fig. 9) showed that on the platinum surface covered
with O(a) and NO(a) the adsorption of ammonia is not prohib-
1
3
ited. [ N]NH3 directly reacts at the beginning of the catalyst
bed, but in contrast to the first pulse (Fig. 8), now 80% of the
1
3
Fig. 8. A [ N]NH pulse was adsorbed on the preoxidized platinum sponge
3
3
13
kept under He flow (48 cm /min) followed after 750 s by a removal of
N species with nitric oxide: (A) MS spectrum, shown from the moment
of NO addition; (B) PEP image, the color intensity represents the con-
centration of NH ; (C) normalized concentration as function of time
T = 323 K, 0.5 vol% NO/He flow of 48 cm /min).
injected N is retained at the surface. The second difference
1
3
is the formation of N2 in addition to N2O. This [ N]NH3
pulse also leads to water formation and thus to reduction
of platinum. Since the oxygen surface coverage is lowered,
dinitrogen formation becomes more favorable, which was
also observed in the hydrogen and ammonia reaction ex-
13
3
3
(
NO, favors the formation of N2O above N2. Temperature-
programmed experiments and ammonia and hydrogen flush
experiments suggested that the adsorbed N species are
NHx. It is plausible that NO selectively reacts with one
of these species:
1
3
periments. By increasing the temperature the remaining
N
species desorb from the surface as N2 and N2O (Fig. 9A).
The desorption of nitrogen and nitrous oxide is again accom-
panied by water formation/desorption,indicating that mainly
the NHx species are present at the surface. Both desorption
peaks cease, which actually means that NO at this tempera-
ture does not decompose at the surface, which is confirmed
in another experiment, in which temperatures up to 623 K
were needed to observe production of nitrogen and oxygen
from NO. The 1 N species are already removed from the
surface at 573 K and thus 13NO does not remain at the sur-
1
3
1
3
NHx(a) + NO(a) → N2O(a) + xH(a),
H(a) + O(a) → OH(a).
The abstraction of one hydrogen from NH(a) by NO giving
N2O(a) is more probable, because NH2 would lead to H2,
which is not detected with MS.
(13)
(14)
3

D.P. Sobczyk et al. / Journal of Catalysis 219 (2003) 156–166
165
face. These NO flush experiments suggest that 13NO was not
NHx species. The formed NO also cannot desorb at these
low temperatures and therefore under these conditions N2O
is selectively formed. A known route for the formation of
N2O is the reaction between N(a) with NO(a). The reaction
experiment with NO also indicated another possibility, in
which NO selectively reacts with NH(a) to form N2O, even
at 323 K. The formation of nitrogen becomes more favorable
at lower oxygen surface coverage, which is shown in the hy-
drogen and ammonia reaction experiments. Above 440 K the
TPD and TPO experiments nicely indicate that NO desorbs
from a preoxidized platinum surface. Since at higher tem-
peratures NO desorption is possible, the selectivity toward
1
3
formed in large amounts at the surface and that the [ N]NH3
1
4
pulse in the NO flow at the preoxidized platinum surface
leads to 13NHx. Thus the formation of N2O and N2 is due to
the reaction of NHx(a) with 14NO(a), O(a), and OH(a).
3
.4. General discussion
Our PEP results clearly demonstrate that ammonia read-
ily dissociates in the presence of coadsorbed oxygen. At
low temperatures ammonia does not dissociate on the pure
platinum sponge and the adsorption of ammonia is weak.
Obviously, the presence of atomic oxygen decreases the
activation barrier for the ammonia dissociation. Mecha-
nistically, we envisage that oxygen atoms readily abstract
hydrogen from coadsorbed ammonia. Our results also sug-
gest that oxygen and ammonia occupy different adsorption
sites. Schematically,
N O decreases in favor of the NO production, as shown in
2
Fig. 3B. An interesting feature of the NO desorption from
platinum is the readsorption, which was distinctively ob-
served in the PEP images.
4
. Conclusions
adsorption of NH3 on Ptreduced → weak molecular ads.;
adsorption of NH3 on Pt–O → strong dissociative ads.
The experiments have shown that dissociation of ammo-
nia on platinum is not favorable and that the interaction
of ammonia with platinum is weak at 323 K. This study,
performed under atmospheric pressure conditions, confirms
UHV and theoretical studies proving that ammonia dissoci-
ation on platinum is activated by the presence of oxygen at
the surface. The experiments elegantly show that a platinum
surface precovered with oxygen initiates the ammonia disso-
ciation leading to the gaseous nitrogen-containing products.
Ammonia dissociation on a preoxidized platinum surface
was further investigated as a function of temperature. At
temperatures below 423 K the pulse experiments resulted
in the formation of mainly nitrogen and a small amount of
N2O. Above 423 K the product selectivity is changed and
also NO is formed. It was observed with PEP that ammo-
nia reacts at the beginning of the catalyst bed and that a part
of [ N]NH3 remained at the surface as N species. The
TPD and TPO experiments together with the reaction with
hydrogen, ammonia, and nitric oxide experiments indicated
that mainly NHx species are present at the surface. However,
NO is an important reaction intermediate. In an oxygen-rich
environment the NO formation leads to N2O and with tem-
peratures above 423 K gaseous NO is detected. The NO flush
experiments indicated that NO selectively reacts with the
NH species to form N2O. Nitrogen is preferably formed at
the lower oxygen surface coverage.
The radiolabeled PEP experiments have shown that the
preadsorbed oxygen favors the dissociation of ammonia,
which leads to production of N2, N2O, and NO. The prod-
uct selectivity strongly depends on the temperature. It has
been shown that below 423 K mainly nitrogen and ni-
trous oxide were formed and above this temperature also
NO. The PEP experiments indicate that all ammonia re-
1
3
acts at the beginning of the catalyst bed and that [ N]NH3
is partly converted into gaseous products and partly re-
mains adsorbed at the surface in some dissociated form.
Temperature-programmed experiments indicate that the re-
maining adsorbed species at the surface are mainly NHx
species because the formation of nitrogen and nitrous ox-
ide is accompanied by the production of a large amount of
water. The NO reaction experiment provides further indica-
tions that NHx species remain at the surface. Thus, upon
adsorption of ammonia hydrogen atoms are stripped off by
adsorbed oxygen species. These exothermic reactions, con-
secutively lead to the formation of atomic nitrogen and OH
groups. Fahmi and van Santen [4] showed that NH bond
scissioning is energetically more costly than that of NH2
and that the recombination of atomic nitrogen is not the
rate-determining step. Our experiments are performed at
relatively low temperature. The relatively high-surface cov-
erages also contribute to the blocking of certain surface
reaction pathways on the surface.
1
3
13
1
3
As already noted a part of [ N]NH3 reacts toward N2,
N2O, and NO. A simplified model of the reaction mecha-
nism includes formation of atomic nitrogen, which leads to
nitrogen and nitrous oxide (below 423 K). OH groups form
water and also promote the formation of NO(a) via the reac-
tion with NHx. At high oxygen surface coverage and below
Acknowledgment
The authors thank J. van Grondelle for his technical ad-
vice and support.
3
50 K the formation of nitrogen from NO dissociation is not
References
favorable [2]. Therefore, it is concluded that atomic nitro-
gen is formed from the abstraction of hydrogen from the
[1] W.D. Mieher, W. Ho, Surf. Sci. 322 (1995) 151.

166
D.P. Sobczyk et al. / Journal of Catalysis 219 (2003) 156–166
[2] J.M. Bradley, A. Hopkinson, D.A. King, J. Phys. Chem. 99 (1995)
[21] P.R. Norton, J. Catal. 36 (1975) 211.
1
7032.
[22] N. Wagner, O. Brümmer, Krist. Tech. 13 (1978) K71.
[23] J.L. Gland, Surf. Sci. 93 (1980) 487.
[24] H. He, N. Swami, B.E. Koel, J. Chem. Phys. 110 (1999) 1173.
[
[
[
3] J.M. Bradley, A. Hopkinson, D.A. King, Surf. Sci. 371 (1997) 255.
4] A. Fahmi, R.A. van Santen, Z. Phys. Chem. 197 (1996) 203.
5] M. Garcia-Hernandez, N. Lopez, I.P.R. De Moreira, J.C. Paniagua, F.
Illas, Surf. Sci. 430 (1999) 18.
6] J.L. Gland, V.N. Korchak, J. Catal. 53 (1978) 9.
7] G.B. Fisher, Chem. Phys. Lett. 79 (1981) 452.
8] 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.
9] G. Papapolymerou, V. Bontozoglou, J. Mol. Catal. A 120 (1997) 165.
[25] B. Laurßen, S. Günther, H. Marbach, M. Kiskinova, J. Janek, R. Im-
bihl, Chem. Phys. Lett. 316 (2000) 331.
[
[
[
[26] W. Vaalburg, W.A.A. Kamphuis, H.D. Beerling van der Molen, S.
Reiffers, A. Rijskamp, M.G. Woldring, Int. J. Appl. Radiat. Isot. 26
(1975) 316.
[
[
27] K. Suzuki, Y. Yoshida, Appl. Radiat. Isot. 50 (1999) 497.
28] D.P. Sobczyk, J. van Grondelle, A.M. de Jong, M.J.A. de Voigt, R.A.
van Santen, Appl. Radiat. Isot. 57 (2002) 201.
[
[
[
[
10] M. Matsumo, N. Tatsumi, K. Fukutani, T. Okano, Surf. Sci. 513 (2002)
485.
[
29] A.V.G. Mangnus, L.J. van Ijzendoorn, J.J.M. Goeij, R.H. Cunning-
ham, R.A. van Santen, M.J. de Voigt, Nucl. Instrum. Methods Phys.
Res. B 99 (1995) 649.
11] H. Aizawa, Y. Morikawa, S. Tsuneyuki, K. Fukutani, T. Ohno, Surf.
Sci. 514 (2002) 394.
12] A. Bogicevic, J. Strömquist, B.I. Lundqvist, Phys. Rev. B 57 (1998)
R4289.
13] P.J. Feibelman, Phys. Rev. B 56 (1997) 10532.
14] M. Lynch, P. Hu, Surf. Sci. 458 (2000) 1.
[30] D.M. Ruthven, Principles of Adsorption and Adsorption Processes,
Wiley, New York, 1992.
[
[
[
[
[31] J.J. Vajo, W. Tsai, W.H. Weinberg, J. Phys. Chem. 89 (1985) 3243.
[32] S.J. Lombardo, F. Esch, R. Imbihl, Surf. Sci. Lett. 271 (1992) L367.
[33] S.J. Lombardo, M. Slinko, T. Fink, T. Löher, H.H. Madden, F. Esch,
R. Imbihl, G. Ertl, Surf. Sci. 269/270 (1992) 481.
15] Ž. Šlijvan cˇ anin, B. Hammer, Surf. Sci. 515 (2002) 235.
16] F. Illas, N. López, García-Hernández, P.R. De Moreira, J. Mol. Struct.
458 (1999) 93.
[34] G. Veser, F. Esch, R. Imbihl, Catal. Lett. 13 (1992) 371.
[
[
17] M. Kim, S.J. Pratt, D.A. King, J. Am. Chem. Soc. 2000 (2000) 2409.
18] A.C.M. van den Broek, J. van Grondelle, R.A. van Santen, J. Catal. 185
[35] N.R. Avery, Surf. Sci. 131 (1983) 501.
[36] J.H. Campbell, C. Bater, W.G. Durrer, J.H. Craig Jr., Surf. Sci. 380
(1997) 17.
[37] R.J. Gorte, J.L. Gland, Surf. Sci. 102 (1981) 348.
[38] D. Burgess Jr., D.S. King, R.R. Cavangh, J. Vac. Sci. Technol. A 5
(1987) 2959.
(
1999) 297.
19] B.G. Anderson, R.A. van Santen, A.M. de Jong, Top. Catal. 8 (1999)
25.
[
[
1
20] G. Jonkers, K.A. Vonkeman, S.W.A. van der Wal, R.A. van Santen,
Nature 355 (1992) 63.