Determination of surface coverage of catalysts: Temperature programmed experiments on platinum and iridium sponge catalysts after low temperature ammonia oxidation

Article abstract of DOI:10.1006/jcat.1999.2506

The activity of iridium and platinum sponge catalysts was studied in the low temperature gas phase oxidation of ammonia with oxygen. Under the reaction conditions used, iridium was found to be more active and more selective to nitrogen than platinum. Furthermore it was established from activity measurements that both catalysts lose activity as a function of time on stream due to inhibition of the surface by reaction intermediates. The used catalysts were studied with XPS and temperature programmed techniques. It was found that the surface of the catalysts had a high coverage of NH and OH and some additional NH₂. It seems most likely that the reaction mechanism proceeds through a stepwise dehydrogenation of the ammonia molecule. It appears that the last dehydrogenation step (NH by OH to N and water) is the rate determining step. The high selectivity of iridium to nitrogen can be explained by the higher activity of iridium in dissociating NO.

Full text of DOI:10.1006/jcat.1999.2506

Journal of Catalysis 185, 297–306 (1999)
Article ID jcat.1999.2506, available online at http://www.idealibrary.com on
Determination of Surface Coverage of Catalysts: Temperature
Programmed Experiments on Platinum and Iridium Sponge
Catalysts after Low Temperature Ammonia Oxidation
A. C. M. van den Broek, J. van Grondelle, and R. A. van Santen
Schuit Institute of Catalysis, Laboratory of Inorganic Chemistry and Catalysis, Eindhoven University of Technology,
P.O. Box 513, 5600 MB Eindhoven, the Netherlands
Received September 10, 1998; revised March 24, 1999; accepted March 29, 1999
Various catalysts of different types have been tested for
The activity of iridium and platinum sponge catalysts was stud- the ammonia oxidation reaction: biological catalysts (3),
ied in the low temperature gas phase oxidation of ammonia with metal oxide catalysts (4), ion-exchanged zeolites (4–6), and
oxygen. Under the reaction conditions used, iridium was found to
be more active and more selective to nitrogen than platinum. Fur-
metallic catalysts (1, 2, 7, 8). At a low reaction temperature
(<473 K) metallic catalysts are the only catalysts that com-
thermore it was established from activity measurements that both
catalysts lose activity as a function of time on stream due to inhi-
bition of the surface by reaction intermediates. The used catalysts
were studied with XPS and temperature programmed techniques.
It was found that the surface of the catalysts had a high coverage of
NH and OH and some additional NH2. It seems most likely that the
reaction mechanism proceeds through a stepwise dehydrogenation
bine a high activity and selectivity to nitrogen (1, 2, 7–9).
Of the various metals, platinum (7–9) and iridium (9) are
found to be the most active catalysts. Furthermore it was es-
tablished (10, 11) that larger metal particles are more active
in the oxidation of ammonia than smaller metal particles.
Also it was found that the metal catalysts lose a large por-
of the ammonia molecule. It appears that the last dehydrogenation tion of their activity as a function of time on stream (10–14).
step (NH by OH to N and water) is the rate determining step. The This loss of activity of the platinum catalysts was attributed
high selectivity of iridium to nitrogen can be explained by the higher to oxidation of the platinum surface (11) or to block-
activity of iridium in dissociating NO.
�c 1999 Academic Press
ing of the surface with nitrogen containing intermediates
(
13, 14).
Information on mechanistic aspects of the ammonia oxi-
INTRODUCTION
dation reaction is available for platinum catalysts from sur-
face science (15–18), electrochemical (13, 14), and theoret-
ical studies (19). The reaction mechanism was described
in studies on the Pt(100) surface (15) or the Pt(111) sur-
face (17, 18) or platinized platinum electrode (13, 14). The
mechanism starts with the adsorption of the reactants on
different sites (denoted s for adsorption through a N atom
Selective oxidation of ammonia with oxygen to nitrogen
and water could be a solution to severalammonia emissions,
caused by various sources like the selective catalytic reduc-
tion of NOx (SCR process), soda production, or agricul-
ture (1). The temperature at which the catalytic oxidation
of the ammonia has to take place depends on the process
that causes the ammonia emission (2). At low temperatures
�
and for adsorption through an O atom) on the surface:
(
<800 K) nitrogen is the favored product of the reaction for
NH3 + s → NH3(a)
[4]
[5]
most catalyst systems:
�
O2 + 2 → 2 O(a).
1
2
NH3 + 1 O2 → N2 + 3 H2O.
[1]
2
After adsorption the NH3 molecule can react with O, as in
Eqs. [6]–[8] or with OH (after it has been formed) as in
Eqs. [9]–[11]:
However, the formation of N2O is often seen as a byproduct
at these temperatures:
2
NH3 + 2 O2 → N2O + 3 H2O.
[2]
NH3(a) + O(a) → NH2(a) + OH(a)
NH2(a) + O(a) → NH(a) + OH(a)
NH(a) + O(a) → N(a) + OH(a)
[6]
[7]
[8]
At higher temperatures NO can be produced:
1
2
2
NH3 + 2 O2 → 2 NO + 3 H2O.
[3]
297
0021-9517/99 $30.00
Copyright �c 1999 by Academic Press
All rights of reproduction in any form reserved.

2
98
VAN DEN BROEK, VAN GRONDELLE, AND VAN SANTEN
NH3(a) + OH(a) → NH2(a) + H2O(a)
[9] Electron Microscopy Measurements
NH2(a) + OH(a) → NH(a) + H2O(a)
NH(a) + OH(a) → N(a) + H2O(a).
[10]
[11]
Scanning electron microscopy (SEM) and energy disper-
sive spectroscopy(EDS) measurementswere done on a Jeol
8
40 electron microscope. Both reduced and used platinum
and iridium sponge samples were measured.
Thereafter N2 can be formed (15, 17):
2
N(a) → N2(g) + 2s.
[12]
Krypton BET Measurements
Alternatively (13, 14), N2 can be formed from NHx(a) (x = 1
or 2) via a N2Hx (x = 2 to 4):
Krypton BET measurementswere done on a Micromerit-
ics ASAP 2000 setup, using a standard procedure without
any pretreatment of the sample. Krypton was used as an ad-
sorbent because of the low surface area of the metal sponge
NHx (a) + NH y(a) → N2Hx+y(a)
[13a]
N2Hx+y(a) + (x + y) OH(a) → N2(g) + (x + y) H2O. [13b] samples. The average metal particle size was determined
from the surface area, assuming cubic close packed sym-
Different mechanisms are presented in the literature on
metry for both metals.
the formation of NO at low temperature. Two reactions are
reported (15, 17):
Experimental Setup
�
NH(a) + 2 O(a) → NO(a) + OH(a) +
[14]
Catalytic and temperature programmed tests were done
�
N(a) + O(a) → NO(a) + .
[15] in a fixed bed reaction setup equipped with a gas chromato-
graph (ATI UNICAM 610, spec 2790) and a quadrupole
mass spectrometer (Balzers QMG-420) for on-line analysis
The onset of NO desorption from the platinum surface,
of the reactants and products. The gas chromatograph was
capable of analyzing NH3, O2, H2O, N2, and N2O quantita-
NO(a) → NO(g) + s,
[16]
tively. The mass spectrometer was used to check the results
of the gas chromatograph and for the detection of NO and
NO2. A quartz tube with an internal diameter of 4 mm was
used as the reactor.
was observed in various studies starting at a temperature
of 300 K (17), 400 K (15), or 550 K (18).
Also, on the Pt(100) surface it is possible at temperatures
higher than 350 K to dissociate NO (15):
NO(a) + � → N(a) + O(a).
[17]
Steady State Catalytic Tests
On the Pt(111) or Pt(110) surface it is more difficult to
dissociate NO (20).
For the steady state catalytic tests the reactor was filled
with the metal sponge (12.2 mg for Ir, 92.8 mg for Pt) diluted
with about 130 mg of � -alumina. The reactor was assumed
to be differential at low conversion, using calculated estima-
tions of diffusion and heat transfer. The catalyst was diluted
with � -alumina in order to satisfy plug flow requirements.
It was determined experimentally that � -alumina is not ac-
tive for oxidation of ammonia at the conditions used in this
study. Catalytic tests were done with a reaction mixture con-
sisting of 2.0 vol% of ammonia and 1.5 vol% of oxygen in
In this study platinum and iridium sponge catalysts were
tested in the low temperature (<373 K) gas phase ammonia
oxidation reaction. In order to determine the cause of the
time on stream behavior, used catalysts were studied with
XPS and temperature programmed experiments.
METHODS
Sample Preparation
3
helium with a total flow of 48.8 cm /min (STP). The reaction
The platinum and iridium sponge samples were ac- temperature was adjusted to keep conversion below 10% .
quired from Johnson Matthey. The sponge samples were Steady state experiments were done using the gas chro-
of >99.9% purity. The size of the sponges was between 250 matograph after at least 12 h on stream. Reaction orders
and 350 �m for the platinum sample and <841 �m for the were determined by lowering the concentration of one of
iridium sample. In the reaction setup particles were used the reactants at a fixed concentration of the other reactant.
with a particle size between 125 and 425 �m; if necessary, For the platinum catalysts reaction orders were determined
samples of the catalyst were pelleted, crushed, and sieved to at a temperature of 433 K. For the iridium catalysts reaction
acquire the appropriate particle size. A diluted sample was orders were determined at a temperature of 403 K. Activa-
made by mixing a sample of the sponge with � -alumina par- tion energies were determined by varying the temperature
ticles of the same particle size. All catalysts were reduced in (for the platinum catalysts in the range 413–443 K, for the
the reaction setup in a dried hydrogen flow to a temperature iridium catalysts in the range 413–383 K). Used catalysts
of 573 K.
were regenerated by reduction in a dried hydrogen flow at

SURFACE COVERAGE OF Pt AND Ir CATALYST AFTER AMMONIA OXIDATION
299
5
73 K. The regenerated catalyst showed exactly the same this is the isothermal part of the TPR. After 1 h the temper-
behavior as the fresh catalysts.
ature programmed part of the TPR was started, in which the
temperature was raised at 10 K per minute to 773 K. After
30 min at 773 K the temperature was lowered to the reac-
tion temperature and the flow was switched to pure helium.
Like in the TPD and reduction experiment, the catalyst was
found to be regenerated by the TPR experiment and was
used in the next experiment. Both the TPD and reduction
experiment and the TPR experiment were repeated using
D2 instead of H2. The experiments with H2 made it possi-
ble to distinguish between NH3 and H2O (mass 17 and 18,
respectively). The results of the D2 experiments gave in-
formation about the occurrence of the various deuterated
forms: ND3 and D2O (mass 20), ND2H and HDO (mass 19),
NDH2 and H2O (mass 18), and NH3 (mass 17). However,
in the calculation of concentrations from these data the ex-
perimental error was enhanced largely by interference of
fragmentation ions of the different deuterated species. As
Time on Stream (TOS) Experiments
For the time on stream (TOS) and the temperature pro-
grammed experiments a sample of 1.0 g of pure sponge
was used as the catalytic bed. In the TOS experiments a
flow with the reactants was switched to a reduced catalyst
at time zero. The concentrations of the reactants and prod-
ucts were measured using the mass spectrometer for at least
2
h on stream. A flow consisting of 2.0 vol% of ammo-
nia and 1.5 vol% of oxygen in helium with a total flow of
3
4
8.8 cm /min (STP) was used. The experiment was done at
a temperature of 323 K for the iridium sponge catalyst and
at 373 K for the platinum sponge catalyst.
XPS Experiments
XPS experiments were done on a VG Scientific XP spec- a consequence it was not possible to distinguish between
trometer, Al K� photon radiation was used. The measure- the various deuterated species.
ments were corrected for charging by using the Ir 4f signal
at 60.9 eV or the Pt 4f signal at 71.2 eV as a reference. XP
spectra were taken of freshly reduced platinum and irid-
RESULTS
ium sponge samples as well as of samples that were used
in a TOS experiment. The samples were taken out of the
Particle Size Determination
The results of the particle size determination are given
in Table 1. The particle sizes determined with krypton BET
agree with the particle sizes that were determined with
SEM. From SEM it can be deduced that the sponges consist
of small nonporous particles that are clustered together in
much larger sponge-like structures. From EDS it was found
that no significant impurities are present in the sponges.
reactor and transported in air to be fitted on an iron stub
carrying an indium foil. Because of interference of an InO2
layer on the indium foil it was not possible to get a clear O
1
s signal for oxygen on platinum or iridium.
Temperature Programmed Experiments
Two different experiments were done to examine the ad-
sorbed species on the catalyst: temperature programmed
desorption (TPD) followed by reduction and temperature
programmed reduction (TPR). In the TPD and reduction
experiment, a TOS experiment is done first with the cata-
lyst for 2 h. Then the catalyst was flushed with a helium flow
Steady State Catalytic Tests
The results of the steady state catalytic tests are summa-
rized in Table 2. The iridium catalyst was found to be more
active than the platinum catalyst. Furthermore the selectiv-
ity to nitrogen was also higher for the iridium catalyst. The
reaction orders that were determined are all low, the order
in O2 is lower for iridium than for platinum, whereas the
order in NH3 is higher for iridium than for platinum. The
activation energy was found to be somewhat higher for the
iridium catalyst.
3
of 10.2 cm /min (STP) for 1 h at the reaction temperature.
Subsequently, in the TPD experiment, the temperature was
raised at 10 K per minute to a temperature of 773 K. Af-
ter 30 min at 773 K, the reduction experiment was carried
out by switching the helium flow to a 20 vol% H2 in he-
3
lium mixture, with a flow rate of 10.2 cm /min (STP). After
1
h at 773 K the temperature was lowered to the reaction
temperature and the flow was switched back to pure he-
lium. The catalyst was found to be regenerated by the TPD
and reduction experiment and was used in the next exper-
iment. In the TPR experiment a TOS experiment is first
done with the catalyst for 2 h. Then the catalyst was flushed
TABLE 1
Metal Particle Size Determination
BET
SEM
size (�m)
3
2
a
with a helium flow of 10.2 cm /min (STP) for 1 h at the
Catalyst
Area (m /g)
Size (�m)
reaction temperature. Subsequently, the TPR experiment
was started. The reactor flow was switched to a 20 vol% H2
in helium mixture with a flow rate of 10.2 cm /min (STP).
Ir sponge
Pt sponge
0.567
0.099
0.50
3.0
0.2–3.0
1.0–5.0
3
a
First the temperature is kept at the reaction temperature;
Size of the small nonporous particles.

3
00
VAN DEN BROEK, VAN GRONDELLE, AND VAN SANTEN
TABLE 2
Steady State Activity, Selectivity, Reaction Orders,
and Activation Energy
Selectivity
to N2
Order
in O2
Order
in NH3
Eact
(kJ/mol)
Catalyst TOF (h 1)
�
a
97a
80
b
d
b
c
Ir sponge
Pt sponge
640 � 9
72 � 7
� 0.3 � 0.1
0.3 � 0.1
0.24 � 0.05 126 � 3
a
a
d
e
� 0.1 � 0.2
108 � 3
Note. Error margins are 95% confidence intervals.
a
Determined at 413 K.
Determined at 403 K.
Determined from 383 to 413 K.
Determined at 433 K.
b
c
FIG. 2. Initiation of reaction with a stoichiometric reaction mixture
with a platinum sponge catalyst. ᭡, N2; ᭺, N2O; ammonia concentration,
d
e
Determined from 413 to 443 K.
2.0 vol% ; oxygen concentration, 1.5 vol% ; GHSV, 3000.
Time on Stream (TOS) Experiments
Fig. 2, immediately at the start of the reaction mostly N2O
is produced, with the production decreasing rapidly after
As can be seen in Figs. 1 and 2, both iridium and plat-
inum catalysts showed a severe loss of activity after the
catalyst was exposed to the reaction mixture. The resultant
steady state activity per surface site was the same order of
magnitude for both catalysts, but it must be noted that the
iridium catalyst was tested at 323 K, whereas the platinum
catalyst was tested at 373 K. The iridium sponge catalyst
first becomes more activate at the start of the reaction and
8
s on stream. Thereafter a large increase in N2 production
is seen, with the production rapidly decreasing after 12 s on
stream. At steady state the selectivity to N2 is constant at
7
5% , with the remainder being N2O.
XPS Experiments
The XP spectra around the binding energy of N 1s of
then becomes less active. Figure 1 clearly shows that max- both reduced and deactivated iridium sponge are shown in
imum activity was reached in 20 s; thereafter the activity Fig. 3. The broad N 1s peak centered around 399 eV can be
decreased to 10% of the maximum activity in 60 s. Finally observed only in the used sample. From literature (21, 22) it
after 1200 s (not shown) the activity is stable at 2% of the is known that the N 1s signal of NH (x = 0–3) has a binding
x
maximum activity. The selectivity to N2 is 100% at the start energybelow400eV, whereasthe N 1ssignalsofNO , N H ,
x
2
x
+
of the reaction, dropping to 95% during some TOS, the re- and NH4 have a binding energy above 400 eV. Therefore,
mainder being N2O. The platinum sponge catalyst is also the measured N 1s signal must be due to an adsorbed NHx
found to become more active at the start and subsequently (x = 0–2) species, excluding NH since it would already be
3
become less active. As shown in Fig. 2, the activity increases desorbed at the temperature used in this experiment. The
during the first 12 s on stream, and then it decreases. The broad nature of the peak suggests that a mixture of NHx
activity decreased to 10% of the initial activity in just 60 s. (d xi u=m0 c– a2 t) a sl yp se t c. ies are present on the surface of the used iri-
After 1200 s on stream (not shown) the activity is stable
at 2% of the maximum activity. As can be clearly seen in
The XP spectra around the binding energy of N 1s of
both reduced and deactivated platinum sponge are shown
in Fig. 4. The N 1s peak centered around 398 eV can be
FIG. 1. Initiation of reaction with a stoichiometric reaction mixture
with an iridium sponge catalyst. ᭡, N2; ᭺, N2O; ammonia concentration,
2
.0 vol% ; oxygen concentration, 1.5 vol% ; GHSV, 3000.
FIG. 3. XP-spectra for N 1s of used (a) or reduced (b) iridium sponge.

SURFACE COVERAGE OF Pt AND Ir CATALYST AFTER AMMONIA OXIDATION
301
TABLE 3
Amounts of Desorbed Species in the TPD and Reduction Experi-
ments for Iridium Sponge, Variation of Reactant Concentration
Desorbed species per surface site
Reactant
concentration
TPD
Reduction
NH3
O2
N2
N2O
NH3
H2O
N2
NH3
H2O
a
a
a
b
a
a
b
b
b
2
2
1
.0
.0
.0
0.8 0.128 0.010 0.025 0.24 0.003 0.010 0.00
a
a
a
b
a
a
1.5 0.122 0.011 0.021 0.24 0.003 0.001 0.01
a
a
a
b
a
a
1.5 0.104 0.011 0.019 0.23 0.003 0.002 0.02
a
FIG. 4. XP-spectra for N 1s of used (a) or reduced (b) platinum
sponge.
95% Confidence interval is � 0.003 for N , N O, and NH .
2
2
3
b
95% Confidence interval is � 0.01 for H O.
2
observed only in the used sample. Since the peak is similar
to the that of the iridium sample it must also be due to
adsorbed NHx (x = 0–2) species. This is in agreement with
an XPS study (23) which showed that on Pt(111) NH3 and
NHx (x = 1 or 2) give rise to N 1s peaks at 399.8 and 397.5 eV,
respectively. Different values are reported in literature for
the peak position of Nads on Pt. In a study of NO adsorption
on Pt(100) and Pt(310) (20) it was found that Nads gives
rise to a N 1s peak and 399.5 eV, whereas in a study of
ammonia oxidation on Pt(111) (24) the N 1s peak was found
at 397.6 eV. Since the N 1s measured in this study is very
broad we cannot exclude the contribution of Nads on the
basis of any of the literature values.
two shoulders at 450 and 510 K, respectively. In the subse-
quent reduction with H2, also listed in Table 3, only small
amounts of H2O, N2, and NH3 were observed. In the D2
reduction (not shown) no signal for mass 20 (D2O or ND3)
was detected. The results of the TPR experiment with H2
are shown in Fig. 6 (isothermal part) and Fig. 7 (tempera-
ture programmed part). In the isothermal part of the TPR
experiment the immediate formation of N2, followed by
the subsequent production of NH3 and H2O can be clearly
seen. In the temperature programmed part NH3 and H2O
are observed in two peaks at 410 and 480 K, respectively.
The first peak consist of mainly H2O, whereas the latter is
observed in both H2O and NH3. In the TPR experiment
with D2 (not shown) no signal for mass 20 (D2O or ND3)
was detected.
Temperature Programmed Experiments on Iridium Sponge
The result of the TPD experiment on iridium sponge is
shown in Fig. 5 and summarized in Table 3. In Fig. 5, the
desorption of N2 and H2O can be seen clearly together with
smaller amounts of N2O and NH3. The peaks of N2 and N2O
are relatively sharp and are centered around a temperature
of 410 K, whereas NH3 and H2O show broad peaks centered
around a temperature of 430 K. This difference is probably
caused by a stronger adsorption of NH3 and H2O in the re-
action system. The peaks of N2 and NH3 also appear to have
The results of the TPD experiment after ammonia
oxidation with various reactant concentrations are shown
in Fig. 8 and also summarized in Table 3. From the results
it appears that when the catalyst is exposed to a lower
concentration of NH3, less N2 and H2O are formed in
the TPD experiment. When the catalyst is exposed to a
lower concentration of O2, the amounts of N2 and H2O
formed in the TPD are about the same as after exposure to
FIG. 5. TPD experiment on iridium sponge after reaction at 323 K.
, N2; ᭿, H2O; ᭺, N2O; +, NH3.
FIG. 6. Isothermal part of TPR experiment on iridium sponge after
reaction at 323 K. ᭡, N2; ᭿, H2O, +, NH3.
᭡

3
02
VAN DEN BROEK, VAN GRONDELLE, AND VAN SANTEN
FIG. 9. TPD experiment on platinum sponge after reaction at 373 K.
, N2; ᭿, H2O; ᭺, N2O; +, NH3.
FIG. 7. Temperature programmed part ofTPR experiment on iridium
sponge after reaction at 323 K. ᭡, N2; ᭿, H2O, +, NH3.
᭡
a lower amount of NH3, more H2O was produced in the
reduction. In the D2 reduction (not shown) no signal for
mass 20 (D2O or ND3) was detected.
stoichiometric concentrations. However, after exposure to
a lower concentration of O2 the first peak in the TPD is
lower and the second peak in the TPD is higher than in
the TPD after exposure to stoichiometric concentrations.
The subsequent reduction with H2 is also listed in Table 3.
It was found that after ammonia oxidation with a lower
amount of O2, less H2O and more NH3 were produced
Temperature Programmed Experiments
on Platinum Sponge
The result of the TPD experiment on platinum sponge is
in the reduction, whereas after ammonia oxidation with shown in Fig. 9 and summarized in Table 4. In Fig. 9, the
desorption of N2, H2O, N2O, and NH3 can be seen clearly;
all signals show broad peaks. The peaks of N2 and N2O
are centered around a temperature of 490 K, whereas the
NH3 and H2O peaks are centered around a temperature of
5
10 K. Similar to the case of the iridium experiments, this
difference in temperature is probably caused by a stronger
adsorption of NH3 and H2O in the reaction system. In the
subsequent reduction with H2, also listed in Table 4, a signif-
icant amount of H2O together with small amounts of N2 and
NH3 were observed. In the D2 reduction (not shown) no sig-
nal for mass 20 (D2O or ND3) was detected. The results of
the TPR experiment with H2 are shown in Figs. 10 (isother-
mal part) and 11 (temperature programmed part). In the
isothermal part of the TPR experiment only the produc-
tion of H2O can be seen. In the temperature programmed
TABLE 4
Amounts of Desorbed Species in the TPD and Reduction Experi-
ments for Platinum Sponge, Variation of Reactant Concentration
Desorbed species per surface site
Reactant
concentration
TPD
Reduction
NH3
O2
N2
N2O
NH3
H2O
N2
NH3
H2O
a
a
a
b
b
c
a
a
a
b
b
b
0.06c
0.01c
0.14c
2
2
1
.0
.0
.0
0.8 0.14
1.5 0.09
1.5 0.15
0.01
0.03
0.01
0.05
0.26
0.24
0.27
0.01
0.01
0.01
0.02
0.00
0.00
b
b
c
c
0.06
b
b
0.04
a
b
c
FIG. 8. Experiments after reaction of iridium sponge with various
reactant concentrations. (a) N2 signal of TPD; (b) H2O signal of TPD.
Reaction conditions: ᭡, excess NH3; ᭿, stoichiometric; ᭹, excess O2.
95% Confidence interval is � 0.01 for N2.
95% Confidence interval is � 0.02 for N2O and NH3.
95% Confidence interval is � 0.03 for H2O.

SURFACE COVERAGE OF Pt AND Ir CATALYST AFTER AMMONIA OXIDATION
303
FIG. 10. Isothermal part of TPR experiment on platinum sponge after
reaction at 373 K. ᭡, N2; ᭿, H2O; + , NH3.
part a small amount of NH3 can be seen at temperatures
above 480 K together with a larger amount of H2O peak-
ing at the same temperature. In the TPR experiment with
D2 (not shown) no signal for mass 20 (D2O or ND3) was
detected.
The results of the TPD experiment after reaction with
various reactant concentrations are shown in Fig. 12 and
also summarized in Table 4. From the results it appears
that when the catalyst is exposed to reactant concentra-
tions other than stoichiometric, more N2 and H2O and less
N2O and NH3 are formed. The subsequent reduction with
H2 is also listed in Table 4. It was found that after ammonia
oxidation with a lower amount of O2, less H2O and more
NH3 were produced in the reduction. Whereas after ammo-
nia oxidation with a lower amount of NH3, more H2O was
FIG. 12. Experiments after reaction of platinum sponge with various
reactant concentrations. (a) N2 signal of TPD; (b) H2O signal of TPD.
Reaction conditions: ᭡, excess NH3; ᭿, stoichiometric; ᭹, excess O2.
produced in the reduction. In the D2 reduction (not shown) tial increase in activity. In the steady state measurements
no signal for mass 20 (D2O or ND3) was detected.
shown in Table 2 it was found that the reaction orders were
low, which is an indication of a highly covered catalyst sur-
face. Therefore it is likely that the loss of activity is caused
by intermediate species adsorbed on the catalyst surface.
In order to determine the nature of the surface species
on the deactivated catalyst it must be first noted that the
catalyst is flushed with helium at the reaction temperature
DISCUSSION
Iridium Sponge
In the TOS experiment shown in Fig. 1 it appeared that
the iridium sponge catalyst becomes less active after an ini-
(
323 K) before the start of the TPD experiment; therefore
any reaction product on the surface will desorb at this stage
of the experiment. This is consistent with literature findings
on the desorption of molecularly adsorbed N2 (25), N2O
(
25), and H2O (26) on iridium surfaces. Molecularly ad-
sorbed O2, however, can be expected to remain adsorbed
on the surface at this temperature (25), whereas molecu-
larly adsorbed NH3 will desorb (27). From the results of
the XPS study on the deactivated catalyst sample it was
concluded that the only N containing species on the surface
can be N, NH, and NH2. From the results of the reduction
and TPR experiment with deuterium it was found that no
ND3 or D2O was produced. From this result it can be con-
cluded that there is no Oads or O2,ads on the surface, since
D2O would be produced if Oads or O2,ads would react with
Dads. Furthermore it can be deduced from this result that
FIG. 11. Temperature programmed part of TPR experiment on plat-
inum sponge after reaction at 373 K. ᭡, N2; ᭿, H2O; +, NH3.

3
04
VAN DEN BROEK, VAN GRONDELLE, AND VAN SANTEN
TABLE 5
of OH and some NH remain on the surface. Therefore, in
x
the subsequent reduction, H2O together with some NH3 is
produced.
Calculated Surface Composition of Deactivated Catalysts
Reactant
concentration
Platinum Sponge
Iridium sponge
Platinum sponge
In the TOS experiment shown in Fig. 2 it appeared that
activity of the platinum sponge catalyst decreases strongly
after an initial increase in activity. Furthermore the selec-
tivity to nitrogen and nitrous oxide changes strongly in the
initial period. In the steady state measurements shown in
Table 2 it was found that the reaction orders were low, which
is an indication of a highly covered catalyst surface. Similar
to the case of iridium, it is therefore likely that the loss of
NH3
O2 2 NH 2 NH2 2 OH 2 NH 2 NH2 2 OH
0.32a
0.29a
0.24a
0.00a
0.00a
0.02a
0.25a
0.26a
0.26a
0.35
0.25
0.35
b
b
b
0.03
0.07
0.01
b
b
b
0.33b
0.038
2
2
1
.0
.0
.0
0.8
1.5
1.5
b
b
0.43
a
9
9
5% Confidence interval is � 0.02 for iridium.
5% Confidence interval is � 0.06 for platinum.
b
there is no Nads on the surface. In the TPR experiment with activity is caused by intermediate species adsorbed on the
H2 the production of NH3 is observed. Furthermore, it is catalyst surface. Furthermore the change in selectivity is a
known that the hydrogenation of NH2 is the rate deter- strong indication of a change in surface coverage; the re-
mining step in the production of NH3 over Rh(111) (28). actions forming N2 and N2O are apparently dependent on
Therefore, there should be some ND3 formed in the TPR surface coverage in the case of platinum.
experiment if there is Nads present on the surface. Since no
Similar to the iridium case discussed above, the cata-
ND3 is observed it can be concluded that there is no Nads lyst was flushed with helium at the reaction temperature
on the surface. The only surface species on the deactivated (373 K) before the start of the TPD experiment; therefore
iridium sponge catalyst can therefore only be OH, NH, and any molecularly adsorbed reactant and reaction product on
NH2. The composition of the iridium surface after the TOS the surface would desorb at this stage of the experiment.
experiment can now be calculated from the results of the This is consistent with literature findings on the desorption
TPD and reduction experiments. The results are shown in of molecularly adsorbed N2 (29), N2O (30), H2O (31), O2
Table 5, together with the results for the platinum sponge (32), and NH3 (23, 33) on platinum surfaces. From the re-
catalyst. The values for NH and NH2, however, must be sults of the XPS study on the used catalyst sample it was
treated with some care. Since the desorption of H2O and concluded that the only N containing species on the surface
NH3 is broad it is possible that some signal is missed which can be N, NH, and NH2. In the results of the reduction and
would result in a larger value for NH2 and a lower value TPR experiment with deuterium it was found that no ND3
for NH. However, obviously there is more NH than NH2 or D2O was produced. As discussed above for the iridium
on the surface.
catalyst it can be concluded from this result that there is no
The results in Table 3 also show that in the TPD after Oads or Nads on the surface. The only surface species on the
reaction with a stoichiometric reaction mixture almost all deactivated platinum sponge catalyst can therefore only be
NHx and OH reacts to form products. Only a small amount OH, NH, and NH2. Therefore in contrast with electrochem-
of OH and NHx remains on the surface and reacts with H2 ical studies on platinum (13, 14) it can be excluded in this
in the subsequent reduction experiment. Furthermore, the study that the loss of activity is due to Nads. The composi-
results show that variation of the reactant concentrations tion of the platinum surface the TOS experiment can now
varies the surface composition of the used catalyst. Com- be calculated from the results of the TPD and reduction ex-
pared to reaction with a stoichiometric reaction mixture, it periments. The results are shown in Table 5, together with
would be expected that reaction with a lower concentration the results for the iridium sponge catalyst. The values for
of oxygen gives rise to a larger amount of NHx and a smaller NH and NH2, however, must be treated with some care.
amount of OH on the surface. This is indeed found in the Since the desorption of H2O and NH3 is broad it is possible
combined TPD and reduction experiments. In the TPD ex- that some signal is missed which would result in a larger
periment more N2 and NH3 and less H2O are produced. value for NH2 and a lower value for NH. However, obvi-
After the TPD experiment only some NHx remains on the ously there is more NH than NH2 on the surface.
surface. Therefore, in the subsequent reduction, only some
The results in Table 4 also show that in the TPD almost
NH3 is produced. Compared to reaction with a stoichio- all NHx reacts to form products, whereas some OH remains
metric reaction mixture, it would be expected that reaction on the surface. This OH is reduced with H2 in the subse-
with a lower concentration of ammonia gives rise to a larger quent reduction experiment. Furthermore the results show
amount of OH and a smaller amount of NHx on the surface. that variation of the reactant concentrations in the reaction
This is indeed found in the combined TPD and reduction varies the surface composition. Compared to reaction with
experiments. In the TPD experiment less N2, NH3, and H2O a stoichiometric reaction mixture, it would be expected that
are produced. After the TPD experiment a larger amount reaction with a lower concentration of oxygen gives rise to

SURFACE COVERAGE OF Pt AND Ir CATALYST AFTER AMMONIA OXIDATION
305
a larger amount of NHx and a smaller amount of OH on much more active in than platinum in the low temperature
the surface. This is indeed found in the combined TPD and ammonia oxidation reaction it appears that bond breaking
reduction experiments. In the TPD experiment, however, a will be the rate determining step in this reaction. Therefore
larger amount of H2O is produced. This indicates that the the reaction producing N2 will not proceed through a N2Hx
reactions forming N2 and N2O are coupled to the reaction intermediate (Eq. [13]). In that case platinum would be ex-
forming H2O. Because of the larger amount of NHx on the pected to be the most active catalyst. It was shown that there
surface, besides N2 and N2O also a larger amount of H2O is no N2Hx intermediate on the surface. Furthermore no hy-
is produced in the TPD experiment. After the TPD exper- drazine is seen as a product of the reaction. Therefore the
iment less OH remains on the surface together with some formation of the N2Hx intermediate would have to be the
NHx. Therefore, in the subsequent reduction, less H2O and rate determining step. Since this is a N–N bond forming step
even some NH3 is produced. Compared to reaction with a it would be expected that platinum would be more active
stoichiometric reaction mixture, it would be expected that than iridium. The other route for forming N2 is the associa-
reaction with a lower concentration of ammonia gives rise tive desorption of Nads (Eq. [12]). This reaction is known in
to a larger amount of OH and a smaller amount of NHx on surface science to occur at higher temperatures (400 K for
the surface. However, this is not found. From the combined Pt (40), 500 K for Ir (35)) than the reaction temperatures
TPD and reduction experiments both a higher NHx as well used in this study. However, nitrogen is seen as a reaction
as a higher OH coverage is found. Apparently the surface product. Since there is no Nads found on the surface all Nads
with a higher OH coverage can contain more NHx, while that was on the surface must be desorbed during the flush-
the value of x is lower.
ing with helium. Possibly the high surface coverage during
reaction makes the desorption of N2 possible at this low
temperature. This could also be the explanation of the in-
creasing activity at the start of the reaction; the desorption
of nitrogen is only possible at a high surface coverage. Since
the associative desorption of N2 would be expected to be
faster on platinum this step cannot be the rate determin-
ing step in a steady state reaction. Furthermore the surface
would be covered with Nads if this step would be rate de-
termining. Therefore it seems more likely that the stepwise
dehydrogenation of NH3 (Eqs. [6]–[8]) is rate determining.
In that case iridium would be expected to be more active
than platinum, since N–H bonds are broken in these steps.
Since the surfaces of both platinum and iridium are cov-
ered with mainly NH and OH it appears that the reactions
using this NH are the slowest step (Eqs. [8] and [11]). Since
there is no O on the surface the main reaction therefore
is the reaction of NH with OH (Eq. [11]). From compu-
tational chemistry it is known that this surface reaction is
endothermic (19). Therefore this path will be much slower
than the path via the reaction of NH with O (Eq. [8]). The
loss of activity is therefore caused by inhibition of active
sites with NH and OH. The TPD experiments provide ev-
Comparison of Iridium and Platinum Sponges
At steady state, after the initial loss of activity the iridium
catalyst showed to be almost 10 times more active than the
platinum catalyst. Furthermore the selectivity to nitrogen
is significantly enhanced on the iridium catalyst. Therefore
iridium is a more suitable catalyst than platinum for the
low temperature selective oxidation of ammonia under the
conditions used in this study. Since the reaction orders that
were found are different for iridium and platinum, other
reaction conditions may change the difference in perfor-
mance. Both the used platinum and the iridium have sur-
faces covered with NH, NH2, and OH species. It appears
that there is somewhat more NH2 and OH on the platinum
surface than on the iridium surface. The largest difference
in behavior between iridium and platinum can be seen in
the TPD experiments: the products desorb from the sur-
face at much lower temperatures in the case of the iridium
catalyst.
Time on Stream (TOS) Behavior and Reaction Mechanism
Iridium is known to form stronger interactions with ad- idence for this mechanism. The production of water from
sorbed species than platinum, for example with oxygen and hydroxyl groups on platinum or iridium should occur at
nitrogen adatoms (34, 35). Therefore iridium catalysts are temperatures below the reaction temperature used in this
more active than platinum catalysts in reactions in which study (26, 31). However, during the TPD production of wa-
bond breaking is rate determining. This is the reason that ter is observed. This water therefore must be produced by
iridium is more active than platinum in the decomposition the reaction of NHx with OH (Eqs. [10] and [11]).
of ammonia (36), but not in the decomposition of hydrazine
The reaction forming N2O must proceed via a NO inter-
(
37, 38). Iridium catalyst are less active than platinum in mediate. Since no NO is seen as a reaction product or on
reactions in which desorption of the products is the rate de- the surface of the deactivated catalyst the reaction forming
termining step. Because of the stronger interaction with the N2O must be fast. The lower production of N2O on iridium
adsorbed oxygen atoms the reaction rate of iridium is lower can be explained by the higher activity of iridium to disso-
than that of platinum in the hydrogen and carbon monoxide ciate NO (35), therefore reducing the lifetime of NO on the
oxidation (39). Since in this study it was found that iridium is iridium surface to react with N to form N2O.

3
06
VAN DEN BROEK, VAN GRONDELLE, AND VAN SANTEN
CONCLUSIONS 10. Ostermaier, J. J., Katzer, J. R., and Manogue, W. H., J. Catal. 33, 457
In this study it was shown that, under the reaction con-
(
1974).
1976).
1
1
1. Ostermaier, J. J., Katzer, J. R., and Manogue, W. H., J. Catal. 41, 277
(
dition used, iridium is a more active and selective catalyst
than platinum for the low temperature (<373 K) oxidation
of ammonia with oxygen to nitrogen. Furthermore it was
2. DeLaney, J. E., and Manogue, W. H., in “Proceedings, 5th Interna-
tional Congress on Catalysis, Palm Beach, 1972” (J. W. Hightower,
Ed.), pp. 13–267. North-Holland, Amsterdam, 1973.
established that both catalysts lose activity with time on 13. Gerischer, H., and Mauerer, A., J. Electroanal. Chem. 25, 421 (1970).
1
1
1
4. Gootzen, J. F. E., Wonders, A. H., Visscher, W., van Santen, R. A., and
van Veen, J. A. R., Electrochim. Acta 43, 1851 (1998).
5. Bradley, J. M., Hopkinson, A., and King, D. A., J. Phys. Chem. 99,
stream due to inhibition of the surface with reaction inter-
mediates. The used catalysts were found to contain a high
coverage with NH and OH and some additional NH2. Since
iridium is more active than platinum the reaction mecha-
17032 (1995).
6. Gland, J. L., and Korchak, V. N., J. Catal. 53, 9 (1978).
nism must proceed through a stepwise dehydrogenation of 17. Mieher, W. D., and Ho, W., Surf. Sci. 322, 151 (1995).
1
8. Asscher, M., Guthrie, W. L., Lin, T.-H., and Samorjai, G. A., J. Phys.
Chem. 88, 3233 (1984).
the ammonia molecule. Since most of the nitrogen on the
surface is NH it appears that the last dehydrogenation step
1
9. Fahmi, A., and van Santen, R. A., Z. Phys. Chem. 197, 203 (1996).
0. Sugai, S., Takeuchi, K., Ban, T., Miki, H., Kawasaki, K., and Kioka, T.,
Surf. Sci. 282, 67 (1993).
(
NH with OH to N and water) is the rate determining step.
2
The high selectivity of iridium to nitrogen can be explained
by the stronger interaction of iridium with reaction inter- 21. Moulder, J. F., Stickle, W. F., Sobol, P. E., and Bomben, K. D., in
Handbook of X-Ray Photoelectron Spectroscopy” (J. Chastain, Ed.).
“
mediates.
Perkin-Elmer Corporation, Eden Prairie, MN, 1992.
2
2
2
2. Alberas, D. J., Kiss, J., Liu, Z.-M., and White, J. M., Surf. Sci. 278, 51
ACKNOWLEDGMENTS
(
1992).
3. Sun, Y.-M., Sloan, D., Ihm, H., and White, J. M., J. Vac. Sci. Technol.
A 14, 1516 (1996).
4. Amorelli, T. S., Carley, A. F., Rajumon, M. K., Roberts, M. W., and
Wells, P. B., Surf. Sci. 315, L990 (1994).
The authors acknowledge M. M. R. M. Hendrix of the department
of solid state chemistry and material science of Eindhoven University of
Technology, Eindhoven, The Netherlands, for the SEM and EDX mea-
surements. This work has been performed under the auspices of NIOK
2
2
5. Cornish, J. C. L., and Avery, N. R., Surf. Sci. 235, 209 (1990).
6. Wittrig, T. S., Ibbotson, D. E., and Weinberg, W. H., Surf. Sci. 102, 506
(
Netherlands Institute for Research in Catalysis).
(
1981).
REFERENCES
27. Purtell, R. J., Merrill, R. P., Seabury, C. W., and Rhodin, T. N., Phys.
Rev. Lett. 44, 1279 (1980).
1
. Bischoff, M., Strauss, G., and Schultz, E., Offenlegungsschrift
Bundesrepublik Deutschland, DE 40 20 914 A1, 1992.
. Brodkrob, G., Falke, H., Warmbold, A., Schmidt, T., Schultz, E., and
Hoffmeister, M., European patent 0 638 350 A2, 1995.
. Cole, J., Trends Biotechnol. 11, 368 (1993).
28. van Hardeveld, R. M., van Santen, R. A., and Niemantsverdriet, J. W.,
J. Phys. Chem. 101, 998 (1997).
29. Campbell, J. H., Bater, C., Durrer, W. G., and Craig, J. H., Jr., Surf.
Sci. 380, 17 (1997).
2
3
4
30. Avery, N. R., Surf. Sci. 131, 501 (1983).
. Sazonova, N. N., Simakov, A. V., Nikoro, T. A., Barannik, G. B.,
Lyakhova, V. F., Zheivot, V. I., Ismagilov, Z. R., and Veringa, H., React.
Kinet. Catal. Lett. 57, 71 (1996).
31. Wagner, F. T., and Moylan, T. E., Surf. Sci. 191, 121 (1987).
32. Gland, J. L., Sexton, B. A., and Fisher, G. B., Surf. Sci. 95, 587 (1980).
33. Bradley, J. M., Hopkinson, A., and King, D. A., Surf. Sci. 371, 255
(1997).
34. Miyazaki, E., and Yasumori, I., Surf. Sci. 57, 755 (1976).
35. Savkin, V. V., and Kislyuk, M. U., Kinet. Catal. 37, 555 (1996).
36. Papapolymerou, G., and Bontozoglou, V., J. Mol. Catal. A 120, 165
(1997).
5
. Gopalakrishnan, R., Davidson, J., Stafford, P., Hecker, W. C., and
Bartholomew, C. H., in “Environmental Catalysis” (J. N. Armor, Ed.),
p. 74. ACS, Washington, 1994.
. Van den Broek, A. C. M., van Grondelle, J., and van Santen, R. A.,
Catal. Lett. 55, 79 (1998).
6
7
. Il’chenko, N. I., Golodets, G. I., and Avilova, I. M., Kinet. Catal. 16,
37. Papapolymerou, G. A., Botis, A. G., Papargyris, A. D., Spiliotis, X. D.,
and Kasidakis, D., React. Kinet. Catal. Lett. 49, 339 (1993).
38. Papapolymerou, G. A., and Schmidt, L. D., Langmuir 3, 1098 (1987).
39. Nieuwenhuys, B. E., Surf. Sci. 126, 307 (1983).
40. Schwaha, K., and Bechtold, E., Surf. Sci. 66, 383 (1977).
1
264 (1975).
8
9
. Il’chenko, N. I., Russ. Chem. Rev. 45, 1119 (1976).
. Takao, Y., Miyazaki, K., Shimizu, Y., and Egashira, M., J. Electrochem.
Soc. 141, 1028 (1994).