Selective NH3 oxidation on (110) and (111) iridium surfaces

Article abstract of DOI:10.1016/j.jcat.2005.07.022

Ammonia oxidation was studied in situ on Ir(110) and Ir(111) under low-pressure (～10-7mbar) conditions. NH₃ does not dissociate on a flat Ir(111) surface, but surface defects and O_ad facilitate NH _3ad decomposition. High-energy resolution fast XPS measurements were used to monitor the surface coverage during reaction on Ir(110), and temperature-programmed reaction measurements were applied to reveal the gas phase reaction products during reaction. The steady-state NH₃ oxidation reaction starts between 350 and 500 K on both surfaces, which also show similar selectivity. Below 600 K, N₂ and H₂O are the principal reaction products. Above 600 K, the selectivity changes to NO and H₂O, with the exact temperature depending on the NH ₃:O₂ pressure ratio. The surface population changes from NH_ad/N_ad to O_ad around 500 K, about 200 K lower than the selectivity change from N₂ to NO observed in the gas phase. This behavior can be explained by considering the activation energies for N₂ and NO_ad formation. We present a model to explain why Ir is more selective toward N₂ than Pt.

Full text of DOI:10.1016/j.jcat.2005.07.022

Journal of Catalysis 235 (2005) 92–102
www.elsevier.com/locate/jcat
Selective NH3 oxidation on (110) and (111) iridium surfaces
C.J. Weststrate a,∗, J.W. Bakker a, E.D.L. Rienks a, J.R. Martinez a, C.P. Vinod a,b, S. Lizzit c,
L. Petaccia c, A. Baraldi d,e, B.E. Nieuwenhuys a,b
a
Leids Instituut voor chemisch onderzoek, Universiteit Leiden, PO Box 9502, Einsteinweg 55, 2333 CC Leiden, The Netherlands
b
Technische Universiteit Eindhoven, Schuit Institute of Catalysis, PO Box 5600 MB, Eindhoven, The Netherlands
c
Sincrotrone Trieste, Strada Statale 14, km 163.5, I-34012 Basovizza, Italy
d
Physics Department and Center of Excellence for Nanostructured Materials, Trieste University, Via Valerio 2, I-34127 Trieste, Italy
e
Laboratorio TASC-INFM, S.S. 14, km 163.5, I-34012 Trieste, Italy
Received 10 June 2005; revised 22 July 2005; accepted 25 July 2005
Available online 22 August 2005
Abstract
−7
Ammonia oxidation was studied in situ on Ir(110) and Ir(111) under low-pressure (∼10 mbar) conditions. NH does not dissociate
3
on a flat Ir(111) surface, but surface defects and O facilitate NH
decomposition. High-energy resolution fast XPS measurements were
ad
3 ad
used to monitor the surface coverage during reaction on Ir(110), and temperature-programmed reaction measurements were applied to reveal
the gas phase reaction products during reaction. The steady-state NH oxidation reaction starts between 350 and 500 K on both surfaces,
3
which also show similar selectivity. Below 600 K, N and H O are the principal reaction products. Above 600 K, the selectivity changes to
2
2
NO and H O, with the exact temperature depending on the NH :O pressure ratio. The surface population changes from NH /N to O
2
3
2
ad ad
ad
around 500 K, about 200 K lower than the selectivity change from N to NO observed in the gas phase. This behavior can be explained by
2
considering the activation energies for N and NO formation. We present a model to explain why Ir is more selective toward N than Pt.
2
ad
2
2005 Elsevier Inc. All rights reserved.
PACS: 82.20.-w; 82.30.-b; 82.45.Jn; 82.65.+r; 82.80.Pv
Keywords: NH ; Ammonia oxidation; Ir(110); Ir(111); Iridium; X-Ray photoelectron spectroscopy; Selective oxidation; Catalytic surface science
3
1. Introduction
into N2 and H2O. Ir catalysts also show a high selectivity
toward N2 rather than NOx. Carabineiro and Nieuwenhuys
[9,10] investigated the oxidation of NH3 on Ir(110) and
Ir(510) using temperature-programmed desorption (TPD)
and temperature-programmed reaction (TPR) between 300
and 800 K. They found that the Ir(110) surface showed se-
lectivity toward N2, but a small amount of N2O was also
observed. They concluded that Ir(510) is more active and
selective (less N2O) than Ir(110). In contrast to what was re-
ported for Ir(110), Ir(510) and Ir(100) [9,10,12,34,39], NH3
does not dissociate on Ir(111) [27,32].
In recent work, we discussed some fundamental aspects
of adsorption and decomposition of NH3 on Ir(110), as well
as the effect of oxygen on ammonia adsorption and decom-
position [39,40]. In these experiments we adsorbed ammonia
at a low surface temperature and subsequently heated the
Since the advent of strict pollution control regulations,
clean processes and waste gas stream cleaning have become
important issues in the chemical industry. Ammonia is one
of the pollutants that must be removed from waste streams.
In the environment, NH3 is converted to nitrite and nitrate.
Nitrate contributes to acid rain and can cause unwanted fer-
tilization of surface waters and delicate ecosystems. Ammo-
nia can be removed from a waste stream via oxidation toward
N2 and H2O, using a heterogeneous catalyst.
Van den Broek, Grondelle, and van Santen [38] have
shown that Ir is an active catalyst for the oxidation of NH3
*
Corresponding author.
E-mail address: c.westrate@chem.leidenuniv.nl (C.J. Weststrate).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2005.07.022
2005 Elsevier Inc. All rights reserved.

C.J. Weststrate et al. / Journal of Catalysis 235 (2005) 92–102
93
surface in vacuum. The results demonstrated that NH3 de-
composition takes place on Ir(110) in both the presence and
the absence of Oad. Nitrogen formed via NH3 ad decompo-
sition desorbs between 500 and 700 K in the absence of
Oad. When oxygen is adsorbed before dosing the ammonia,
a larger fraction of the NH3 ad decomposes. Oad especially
enhances NHad dissociation, which would otherwise inhibit
NH3 ad dissociation below the NH3 desorption temperature
(400 K). In the presence of Oad we observed N2 formation at
350 K, about 200 K lower than in the absence of Oad. This ef-
fect of oxygen is assigned to repulsive interactions between
Nad and Oad [14,28,40]. NOad/NO (g) was observed in an
experiment where a mixed Oad/NH3 ad layer was heated in
the presence of O2 (∼1 × 10−7 mbar). NO forms when both
Nad and Oad are present on the surface and above 450 K.
Several steps in the mechanism responsible for ammonia
oxidation on Ir catalysts remain topics of debate. Two mech-
anisms have been proposed for N2 formation on Ir and Pt:
Nad combination [8,30,39] [Eq. (1)] and NHx ad combina-
tion [17,19,20] [Eq. (2)]
XPS and TPR. XPS allows us to measure the nature and con-
centration of different surface species (NHx ad, Oad, NOad)
in situ during reaction, for different temperatures. This ap-
proach, combining information about surface coverage and
gas phase products, is a powerful method for investigating
the reaction mechanism.
2. Experimental
The TPD and TPR measurements were performed at
the Leiden Institute of Chemistry using a vacuum system
equipped with a differentially pumped, shielded mass spec-
trometer, to reduce the contribution from the heating wires
and the edges of the crystal. The sample was placed in front
of a 2-mm-wide hole, at a distance of ∼2 mm. The system is
equipped with a sputter gun for sample cleaning, and LEED
optics. The base pressure of the system is <5 × 10−10 mbar.
A heating rate of 0.5 K/s was used for all TPR experiments.
A linear background was subtracted from the TPR results, to
correct for the slow increase in the partial pressures, espe-
cially for m/e = 28 (N2 and CO) and m/e = 18 (H2O).
For the TPD experiments on Ir(111), a heating rate of
5 K/s was used. In some experiments 15NH3 was used to
distinguish between N2 and CO.
High-energy resolution fast XPS measurements [4,6]
were performed at the SuperESCA beamline of ELETTRA,
the synchrotron radiation facility in Trieste, Italy. The vac-
uum system, with a base pressure of ∼1 × 10−10 mbar, is
equipped with a double-pass hemispherical electron energy
analyzer [7], a sputter gun for sample cleaning, a mass spec-
trometer, and LEED optics.
2Nad → N2(g),
(1)
(2)
2NHx ad → N2(g) + xH2(g).
Similar mechanisms are proposed for NO formation
Nad + Oad → NOad,
(3)
(4)
NHx ad + (x + 1)Oad → NOad + xOHad.
Another issue under discussion is the difference in selec-
tivity between Ir and Pt. Bradley, Hopkinson and King [8,26]
proposed a reaction model for Pt(100), which is summarized
in the following set of equations:
Both Ir(110) and Ir(111) surfaces were cleaned with Ar+
sputtering and annealing cycles (∼1200 K), followed by
oxygen treatments at 700–1000 K and a final hydrogen treat-
ment to remove oxygen. Residual hydrogen was removed
by a flash to 700 K. O1s and C1s core level regions did
not show oxygen or carbon contamination after cleaning on
Ir(110). For Ir(111), LEED, AES, and oxygen TPD [24,36]
were used to check for surface contaminants. Thermal des-
orption of an oxygen-covered surface is especially sensitive
for surface carbon. Carbon present on the surface forms CO
or CO2 in the presence of Oad, which is then detected during
the heating ramp.
N1s spectra were measured with a photon energy of
496 eV; O1s spectra, with a photon energy of 650 eV. For the
temperature-programmed XPS (TP-XPS) [5] measurements,
a heating rate of 0.3 K/s was used. The XPS measurements
were done only for Ir(110).
2NH3 ad + 3Oad → 2Nad + 3H2O(g),
Nad + Oad → NOad,
NOad → Nad + Oad (>350 K),
2Nad → N2(g),
(5)
(6)
(7)
(8)
(9)
NOad → NO(g).
Eq. (5) describes Oad-induced NH3 ad dehydrogenation. Nad
formed in this step can desorb as N2 [Eq. (8)], but the major
part reacts with Oad to form NOad [Eq. (6)]. NO dissociation
[Eq. (7)] occurs only above 350 K, and is inhibited when the
Oad coverage is larger than 0.2 l. Above 350 K, Nad (formed
via NOad dissociation) desorbs as N2(g) via Eq. (8). Above
400 K, NOad desorbs [Eq. (9)], a process favored over dis-
sociation. In this way the selectivity shifts toward NO(g).
Van den Broek et al. [38] extended this model to Ir cat-
alysts and attributed the difference in selectivity between Pt
and Ir catalysts to a difference in NO dissociation activity.
NO dissociates more easily on Ir than on Pt; in the BHK
model this means that the reaction rate of Eq. (7) is larger,
and thus N2 is the principal product.
The XPS data were evaluated by fitting the spectra with
˘
Doniach–Sunjic´ functions [25], convoluted with a Gaussian
function and superimposed on a linear background. Core-
level binding energies of the different species were measured
with respect to the Fermi level.
We report new experimental results for ammonia oxida-
tion on Ir(110), obtained with high-energy resolution fast
According to Ibbotson et al. [36] a surface saturated with
O2 at 200 K results in a coverage of 1 monolayer (ML). This

94
C.J. Weststrate et al. / Journal of Catalysis 235 (2005) 92–102
was used to calibrate the O1s signal. Because we do not
have any reference structure for nitrogen we have arbitrarily
normalized the sum of all N1s species at 300 K to 1.
The ammonia pressure was 5 × 10−8 mbar during the
TP-XPS and 1 × 10−7 mbar during the TPR measurements,
respectively, and the O2 partial pressure was adjusted ac-
cording to the NH3:O2 pressure ratio. A different pressure
was used in the TPR experiments to compensate for the
shielded MS. Because of the close proximity of the sam-
ple surface to the entrance of the differentially pumped MS,
the reactant pressure is expected to be slightly lower at the
sample surface than in the rest of the vacuum chamber.
3. Results and discussion
3.1. NH3 decomposition on Ir(111): TPD experiments
TPD and TPR experiments (Fig. 1) were performed
to characterize NH3 ad adsorption and decomposition on
Ir(111). Panel (a) shows NH3 desorption (TPD, 5 K/s) after
different exposures to ammonia at 100 K. In these exper-
iments only molecular ammonia desorption was observed,
without the formation of N2(g) and H2(g). This result con-
firms the finding of Purtell et al. [32] that NH3 ad does not
decompose on Ir(111).
Panel (b) shows the effect of preadsorbed oxygen on the
stability of NH3 ad. The desorption spectra were obtained af-
ter the surface was exposed to O2(10 L at 200 K, saturation)
and subsequently exposed to 15NH3 (5 L, saturation). Both
N2 and H2O desorption are observed. In contrast to what
was found for the initially clean surface, NH3 ad dissociated
in the presence of Oad, similar to what was observed for sev-
eral other transition metals, such as Pt, Ni, Cu, and Ag [1,
11,16,18,21,22,35,37]. 15N2 desorption occurred at around
500 K, a temperature comparable to the N2 desorption peak
measured after NH3 decomposition from Ir(100), but unlike
the N2 desorption observed from Ir(110) in either the ab-
sence (620 K [9,39,40]) or the presence (350 K [40]) of Oad.
Water formation occurred between 200 and 400 K, similar
to what was observed for Ir(110) [40]. Nitrogen desorption
occurred around 500 K, above the H2O formation temper-
ature. This shows that Nad combination rather than NH3 ad
dehydrogenation is the rate-determining step for N2 forma-
tion. Comparing the 15N2 desorption of panel (b) with panel
(c), e.g. N2 formation in the absence of Oad shows that the
presence of Oad exerts only a small influence on the N2 des-
orption temperature, not as large as that observed for Ir(110)
(repulsive interaction [14,28,40]). An explanation for this
observation could be that all Oad is consumed during the
NH3 ad dehydrogenation and there is no oxygen left at the
temperature where Nad desorption occurs.
Fig. 1. NH decomposition on Ir(111). (a) Molecular NH desorption after
3
3
15
uptake at 100 K (no decomposition observed, heating rate 5 K/s); (b)
N
2
15
15
desorption after exposure of an O covered surface to NH (5 L NH ,
ad
3
3
15
5 K/s); and (c) steady state NH decomposition on sputtered surfaces
NH pressure 1 × 10 mbar, heating rate 0.5 K/s).
3
15
−7
(
3
tion process. Mortensen et al. [31] reported that the dissocia-
tive sticking coefficient of NH3 on Ru(0001), a close-packed
structure, is enhanced when the surface is slightly sputtered.
The present study found similar behavior for NH3 dissocia-
tion on Ir(111). Panel (c) shows that 15NH3 decomposition
occurs as a steady-state reaction (1 × 10−7 mbar 15NH3,
heating rate 0.5 K/s) after the surface has been sputtered
(energy, 2 keV; sample current, ∼5 µA, time 0–180 s). The
reaction rate reaches a maximum at 580 K and decreases
above this temperature. This behavior is very similar to that
of steady-state NH3 decomposition on Ir(110), shown in
The fact that NH3 dissociates spontaneously on the less
densely packed (110) and (100) surface structures Ir(110),
but not on Ir(111), suggests that atoms with a low coordina-
tion number play an important role in the NH3 decomposi-

C.J. Weststrate et al. / Journal of Catalysis 235 (2005) 92–102
95
Ref. [39], in which a similar maximum was observed at a
similar temperature. The fact that the smooth surface (0 min
of sputtering) also shows some activity is attributed to de-
fects on the annealed surface. Some authors reported that
NH3 dissociates on Pt surfaces above its desorption tem-
perature [23,30], and this might contribute to the observed
activity of the untreated surface as well. For the sputtered
surfaces, the heating and cooling branches show significant
differences in activity. The maximum temperature reached
during the experiments was 1100 K. At this temperature,
the surface partially reorders, and thus reactivity is lower
in the cooling branch. Our observations support the model
proposed by Mortensen et al. [31] in which defects, such
as steps and kinks, are responsible for NH3 dissociation on
smooth surfaces like Ru(0001) and Ir(111).
3.2. TPR and XPS measurements during the NH3/O2
reaction on Ir(110)
Fig. 2 shows the different N1s surface species probed
with the high-energy resolution fast XPS measurements dur-
ing TP-XPS of a coadsorbed layer in NH3 and O2 (5 ×
10−8 mbar NH3, 5 × 10−7 mbar O2). Panel (a) shows two
averaged spectra (average of three spectra, at 400 and 580 K)
in which the fitting components are shown. Panel (b) shows
some of the actual spectra, in which the thermal evolu-
tion of the different peaks can be seen. We have previously
shown [39,40] that the observed peaks can be assigned to
NOad (400.0 eV), NH3 ad (399.0 eV), NHad (397.5 eV) and
Fig. 2. (a) Two (400 and 580 K, average of three spectra) selected spec-
tra that show the fitting components used to evaluate XP spectra. (b) The
actual spectra obtained during a TP-XPS experiment (first heating in a
heat–cool–heat sequence, NH /O = 1:10, heating rate 0.3 K/s).
N
ad (396.6 eV) respectively. In one of these publications [39]
3
2
we suggested that NH2 ad is a very unstable intermediate
that decomposes immediately after formation toward NHad.
Thus, it is not observed in our experiments. In the O1s spec-
tra (not shown) obtained during the in situ experiments, Oad
(530.5 eV) and NOad (533.0 eV) were found. Any peaks re-
lated to OH or H2O were not observed above 200 K [40]
so the surface concentration of these species was very low
under reaction conditions.
Fig. 3 shows the gas phase products (Figs. 3a and c)
and the relative concentrations of different surface species
measured with XPS (Figs. 3b and d) during subsequent
heating-cooling cycles in a reaction mixture of NH3 and O2
(1 × 10−7 mbar NH3, ratio 1:1). Before the surface was ex-
posed to the reaction mixture, it was saturated with Oad (5 L
O2 at 200 K).
A small N2 desorption peak was observed between 400
and 500 K and it was assigned to N2 desorption due to repul-
sive interaction [13,40] between Nad and Oad. The steady-
state reaction started at 450 K, the usual temperature for N2
desorption in the absence of Oad [39,40]. The XPS results
(Fig. 3b) show that the concentration of Oad was very low at
this point, resulting in low H2 production next to H2O and
N2. The maximum observed in the N2 formation between
500 and 600 K was related to the desorption of the Nad con-
centration built up below 450 K. In an experiment with a
lower heating rate (0.1 K/s) this peak was absent, indicating
that the observed signal was a convolution of the steady-state
reaction and a non-steady-state desorption peak.
Above 600 K, the surface composition changed from
NHx ad- to Oad-covered. Intuitively, one would expect that
the N-selectivity would also change at the same temperature,
that is from N2 to NO. In earlier work [40] using an exper-
iment in which a mixed Oad/NH3 ad layer was heated in the
presence of O2, we showed that NOad formed above 450 K
when both Nad and Oad were present on the surface. Under
steady-state reaction conditions, we did not observe NO(g)
formation below 900 K. We discuss this apparent discrep-
ancy in more detail in Section 3.5.
The decrease of the Oad signal and the increase of the
Nad signal observed on XPS show (Fig. 3b) that the reac-
tion between NH3 ad/NHad and Oad occurred between 250–
400 K [40]. This resulted in a high Nad concentration at
400 K, with almost all the Oad removed from the sur-
face and formation of both H2O and NHx ad species. The
NHx ad species block further O2 adsorption and/or dissocia-
tion. Fig. 3a indeed shows H2O desorption between 200 and
400 K as a result of the reaction of NH3 ad/NHad and Oad. No
other desorption products were observed below 400 K.

96
C.J. Weststrate et al. / Journal of Catalysis 235 (2005) 92–102
−7
Fig. 3. In situ measurements during the NH /O reaction with a pressure ratio 1:1. (a) TPR during first heating (0.5 K/s, 1 × 10 mbar NH ), (b) TP-XPS
3
2
3
−8
−7
during first heating (0.3 K/s, 5 × 10
−8
mbar NH ), (c) TPR during cooling (0.5 K/s, 1 × 10
mbar NH ), and (c) TP-XPS during cooling (0.3 K/s,
3
3
5 × 10 mbar NH ).
3

C.J. Weststrate et al. / Journal of Catalysis 235 (2005) 92–102
97
The observed reaction rate (−d[NH3]/dt ≡ d[H2O]/dt)
decreased slightly with increasing temperature. It probably
was limited by the availability of reactants at high tempera-
ture; that is, NH3 ad desorption was competing with NH3 ad
dissociation [39].
In the cooling branch (Figs. 3c and d), N-selectivity
changed back to N2 at ∼900 K. The surface changed from
Oad- to Nad-covered at about 600 K, and the surface oxy-
gen concentration was close to zero below 500 K. The N2
formation rate was low at 500 K, and the reaction stopped
completely at 400 K, due to inhibition by NHx ad, which
blocks O2 adsorption.
In the cooling branch no H2 formation was observed, and
the only reaction product containing hydrogen was water.
This finding can be explained by hysteresis observed in the
Oad coverage. During the heating branch, the Oad concen-
tration was negligible at 400 K; it started to grow at 600 K,
so not enough Oad was available to react with all of the Had
produced between 400 K and 600 K. In the cooling branch,
Oad was present down to 500 K, allowing H2O formation at
this temperature.
We also observed a slow decrease in the Nad concen-
tration in the cooling branch, even below 400 K. This de-
crease cannot be due to N2 desorption, because N2 desorp-
tion was not observed below 400 K in the cooling branch
(see Fig. 3c). Therefore, we assigned the decrease in the Nad
coverage (Fig. 3d) to rehydrogenation of Nad to NHad [39].
pressure was increased. For a ratio 1:5, NO(g) formation was
observed above ∼800 K (900 K for 1:1), whereas for the ra-
tio 1:10, NO(g) formation was observed above 600 K. For
both ratios, NO was the only N-containing gas phase prod-
uct above 850 K. We also observed NOad on the surface,
which formed above 500 K (when the surface was Oad cov-
ered) and was present on the surface up to 600 K, when it
desorbed.
Fig. 5 shows the NOad concentrations for the ratios 1:5
and 1:10 (NOad was not observed for the ratio 1:1). In the
heating branch NOad formed above 500 K. The NOad con-
centration started to decrease above 550 K. The NOad con-
centration slowly increased in the cooling branch between
600 and 450 K, reaching a constant value below 450 K. At
these temperatures dissociation did not occur, and NOad was
present on the surface as a spectator species.
The rate of H2O formation (which is indicative of the
overall reaction rate, −d[NH3]/dt ≡ d[H2O]/dt) is not in-
fluenced by the O2 partial pressure. This means that the
reaction is almost zeroth order in O2 and oxygen adsorp-
tion and dissociation are not the rate-determining steps in
the overall reaction.
A small desorption peak was observed for m/e = 44
around the temperature at which the surface composition
changed to Oad-covered and where Nad and NOad coexisted
on the surface. This could be due to N2O, but the formation
of CO2 via reaction of Oad and CO from the background
cannot be excluded. Baerns et al. [2] recently published an
extensive study on ammonia oxidation on Pt catalysts. They
studied the reaction in different pressure regimes, ranging
from 10−5 to 103 mbar. They did not observe N2O forma-
tion in the UHV experiments; they observed N2O only for
pressures higher than ∼1 × 10−3 mbar. Their interpretation
of their findings was that N2O forms only when the concen-
tration of both NOad and Nad are sufficiently high. This is in
line with our observations, because we saw N2O formation
only when both NOad and Nad were present on the surface
in measurable amounts. In our case the Nad concentration
dropped very fast above 500 K, with the Nad availability pos-
sibly limiting N2O formation.
3.3. The influence of the NH3:O2 pressure ratio on the
activity and selectivity
Fig. 4 shows the effect of the NH3/O2 pressure ratio on
the observed reaction products and on the surface composi-
tion during reaction. The heating branch is shown for two
different NH3/O2 ratios: 1:5 and 1:10.
An increase in O2 partial pressure had little effect on
the surface composition below 400 K. This finding is not
surprising, because the surface was covered with Oad and
NHx ad and O2 adsorption was inhibited by NHx ad. Fig. 4
shows only the sum of NH3 ad and NHad, not the individ-
ual traces, to simplify the figure. The greatest effect of the
O2 partial pressure was observed in the temperature range
at which the surface changed from Nad-covered to Oad-
covered. For a ratio of 1:5 (Figs. 4a and b), the change
occurred at around 520 K (620 K for 1:1), whereas for a
ratio of 1:10 (Fig. 4c and d), it occurred around 480 K. We
did not observe H2 formation for higher O2 partial pressures,
because Oad in those cases was already available at 500 K,
and H2O could easily form.
A small influence of the oxygen partial pressure was ob-
served in the gas phase below 400 K. The low-temperature
N2 desorption peak (300–500 K) was larger for higher O2
partial pressure. At higher O2 partial pressures, more O2
reached the surface, the Oad coverage was slightly higher
and repulsive interactions are more pronounced. NO(g) for-
mation shifted to lower temperature when the O2 partial
3.4. NH3 oxidation on Ir(111)
Fig. 6 shows the gas phase reaction products in the cool-
ing branch of the NH3 + O2 reaction on Ir(111) for different
reactant ratios. The first and second heating were very simi-
lar to the cooling branch, and hysteresis was not observed.
In the heating branch, the steady-state reaction started at
∼500 K (ratio 1:1); in the cooling branch, it stopped at the
same temperature. This temperature decreased slightly when
the O2 partial pressure was increased. For an NH3:O2 ra-
tio of 1:10, the steady-state reaction started at ∼400 K in
the heating branch and stopped at the same temperature in
the subsequent cooling branch. Below 500 K, the reaction
products were N2 and H2O. For the ratio 1:1, some H2 was
observed around 500 K as well, but the main H-containing

98
C.J. Weststrate et al. / Journal of Catalysis 235 (2005) 92–102
−7
Fig. 4. The effect of the NH /O ratio on the selectivity of the surface and on the nature of the reaction products (heating 1). (a) TPR (0.5 K/s, 1 × 10 mbar
3
2
−8
−7
NH ) and (b) TP-XPS (0.3 K/s, 5 × 10 mbar NH ) for the NH /O ratio 1:5; and (c) TPR (0.5 K/s, 1 × 10 mbar NH ) and (d) TP-XPS (0.3 K/s,
3
3
3
2
3
−8
5 × 10 mbar NH ) for the ratio 1:10.
3

C.J. Weststrate et al. / Journal of Catalysis 235 (2005) 92–102
99
via Nad + Nad rather than via NHx ad + NHx ad. In the pres-
ence of Oad on the surface, the NHad concentration is lower
than in the absence of Oad [40], and a mechanism involving
NHx ad species is not likely. N2 formation via a reaction be-
tween NOad and NH3 ad is also not very likely, because NOad
is observed only above 450 K, where the NH3 ad concentra-
tion is very low.
In earlier work we showed [40] that NOad formation oc-
curs via Nad + Oad rather than via NHx ad + Oad, because
the NHx ad concentration is negligible when the NOad for-
mation starts. Fig. 4 shows the same thing: NO formation
starts when the NHx ad concentration is zero.
The difference in selectivity between Pt and Ir catalysts
is not yet completely understood. In the model proposed by
Bradley, Hopkinson, and King (BHK) [8] for Pt(100) NO
formation at low temperature and NO dissociation [3,29,
33] at higher temperature play a crucial role [Eqs. (5)–(9)].
NOad forms around 250 K on Pt(100) and dissociation starts
at 350 K. TP-RAIRS experiments done by Kim, Pratt and
King [26] show ed that NOad was present on Pt(100) be-
tween 250 and 500 K, in line with the BHK model.
In our measurements we monitored in situ all the surface
species during the reaction (NOad, NH3 ad, NHad, Nad and
Oad). Our results suggest that the BHK model cannot be ap-
plied to Ir(110) or Ir(111). In their model NOad is observed
between 250 and 500 K, whereas in our measurements NOad
formation is observed only above 500 K. NHad, Nad, and Oad
coexist on the surface between 200 and 400 K and hence, in
principle, NOad could be formed. This suggests that there
is a kinetic limitation for the formation of NOad. The pres-
ence of NOad in the cooling branch, without dissociation,
indicates that NO dissociation does not occur below 400 K.
In our experiment the surface was always covered with Oad
when NOad was present. Because Oad blocks NO dissocia-
tion [8,13,14] it is also very unlikely that NO dissociates in
this temperature regime under reaction conditions.
Fig. 5. TP-XPS data showing NO (400.0 eV) formation during NH ox-
ad
3
idation for two different NH :O pressure ratios (1:5 and 1:10, 0.3 K/s,
3
2
−8
5 × 10 mbar NH ).
3
product was always H2O. NO was formed at higher tempera-
tures, depending on the NH3/O2 ratio. For a ratio of 1:1, NO
formed above 800 K, whereas for a ratio of 1:10, it formed
above 500 K and for a ratio of 1:5 it formed above 600 K.
The results are similar to the results obtained for Ir(110).
The temperature range where the reaction started and
stopped was between 400 and 500 K for both surfaces.
Moreover, a change in N-selectivity also occurred at simi-
lar temperatures.
A significant difference between Ir(111) and Ir(110) was
that the reaction rate on Ir(111) increased much more with
increasing O2 partial pressure. On Ir(111), the reaction rate
was almost first order in O2 pressure, whereas on Ir(110),
the reaction rate did not change significantly with increasing
O2 pressure. This indicates that O2 adsorption and disso-
ciation is a very important step on Ir(111), in contrast to
what we found for Ir(110). This is in line with the results
of TPD experiments presented in Section 3.1, which showed
that chemisorbed oxygen is needed for NH3 ad dissociation.
We assume that the mechanism responsible for the N-
selectivity is similar on both Ir surfaces, because the se-
lectivity of NH3 oxidation on both surfaces showed similar
behavior for different temperatures and different NH3:O2
pressure ratios. The mechanism responsible for NH3 dissoci-
ation might be different because of the different reactivity of
the surfaces. On Ir(111), we needed Oad for the initial NH3 ad
dissociation step, whereas on Ir(110), it proceeded also in
the absence of oxygen. On Ir(111), the reaction rate also
strongly depended on the O2 pressure and not on Ir(110).
During the reaction, Ir(110) surface composition changes
at a certain temperature from NHx ad to Oad (Fig. 4). The ob-
served selectivity in the gas phase also changes from N2 to
NO, but the temperature at which this change takes place is
200 K higher than the observed change of surface compo-
sition. To explain this large difference, we need to look at
some simple kinetic equations
d[N2]
= kN θN2 ,
(10)
(11)
2
dt
d[NO]
= kNOθNθO.
dt
3.5. The selectivity of the NH3 oxidation on Ir(110) and
Ir(111)
If kN and kNO would have the same value, then (i) the
2
selectivity of the reaction (above the NO and N2 desorp-
tion temperature, i.e., >600 K) would depend only on the
Nad and Oad (θN and θO) coverage, and (ii) the selectivity
change toward NO would occur at the same temperature as
the change of the surface composition. Our results show that
this is not the case, so differences in the k values instead
of the coverages of the different species must be responsi-
In Section 1 several issues were mentioned that are still
under debate. One of these was the mechanism of N2 and
NOad formation. For both reactions, two possible mecha-
nisms have been proposed, one involving only Nad and Oad
and the other involving NHx ad and Oad. In our previous
study in the absence of Oad, we showed [39] that N2 forms

100
C.J. Weststrate et al. / Journal of Catalysis 235 (2005) 92–102
Fig. 6. TPR results for NH oxidation on Ir(111). The cooling branch for different NH /O pressure ratios (1:1, 1:5, 1:10) is shown (heating rate 0.5 K/s,
3
3
2
−7
1 × 10 mbar NH ).
3
ble for the observed difference between surface composition
and product selectivity. The experimental results can be re-
ences in activation energies. This leads to the conclusion
that the Ea is greater for NO formation than for N2 forma-
tion.
produced when we assume that kN is larger than kNO, that
2
is, N2 formation is easier than NO formation. The tempera-
Our recently published study concerning the effect of Oad
on the NH3 ad chemistry on Ir(110) [40] gave more informa-
tion on the activation energies for NO and N2 formation. N2
formation was observed around 600 K in TPD experiments.
Heating of a mixed NH3/Oad layer in the presence of O2(g)
resulted in the formation of NOad, starting around 450 K.
These two observations suggest that NOad forms more eas-
ily than N2. We also found that N2 formation is influenced
ture dependence of the k values is described by the Arrhe-
nius equation, k = A0e−E /(RT ), where k is a prefactor and
a
Ea is the activation energy. Because both N2 and NO for-
mation proceed via simple recombination reactions of Nad
+ Nad or Nad + Oad [39,40] the prefactors are expected
to be rather similar for both NOad and N2 formation [15].
In that case different k values should be related to differ-

C.J. Weststrate et al. / Journal of Catalysis 235 (2005) 92–102
101
by the presence of O2. In the presence of Oad, N2 forma-
References
tion occurs around 350 K, so in that case it is more easy than
NOad formation.
[1] B. Afsin, P.R. Davies, A. Paskusky, M.W. Roberts, D. Vincent, Surf.
Sci. 284 (1993) 109–120.
In summary, we suggest that the difference between Pt
and Ir surfaces during catalytic ammonia oxidation can be
found in the lower activation energy for NO formation on Pt
with respect to Ir. This contradicts earlier literature reports
in which the difference was explained in terms of NO disso-
ciation [38].
[2] M. Baerns, R. Imbihl, V.A. Kondratenko, R. Kraehnert, W.K. Offer-
mans, R.A. van Santen, A. Scheibe, J. Catal. 232 (2005) 226–238.
[3] W.F. Banholzer, R.I. Masel, Surf. Sci. 137 (1984) 339–360.
[4] A. Baraldi, M. Barnaba, B. Brena, D. Cocco, G. Comelli, S. Lizzit, G.
Paolucci, R. Rosei, J. Electron. Spectrosc. Relat. Phenom. 76 (1995)
145–149.
[5] A. Baraldi, G. Comelli, S. Lizzit, D. Cocco, G. Paolucci, R. Rosei,
Surf. Sci. 367 (3) (1995) L67–L72.
[6] A. Baraldi, G. Comelli, S. Lizzit, M. Kiskinova, G. Paolucci, Surf. Sci.
Rep. 49 (6–8) (2003) 169–224.
4. Summary and conclusions
[7] A. Baraldi, V.R. Dhanak, J. Electron Spectrosc. Relat. Phenom. 67 (1)
(1994) 211–220.
[8] J.M. Bradley, A. Hopkinson, D.A. King, J. Phys. Chem. 99 (1995)
17032–17042.
[9] S.A.C. Carabineiro, B.E. Nieuwenhuys, Surf. Sci. 505 (2002) 163–
170.
[10] S.A.C. Carabineiro, B.E. Nieuwenhuys, Surf. Sci. 532 (2003) 87–95.
[11] A.F. Carley, P.R. Davies, K.R. Harikumar, R.V. Jones, G.U. Kulkarni,
M.W. Roberts, Top. Catal. 12 (1–4) (2001) 101–109.
[12] T.V. Choudhary, A.K. Santra, C. Sivadinaranaya, B.K. Min, C.-W. Yi,
K. Davis, D.W. Goodman, Catal. Lett. 77 (1–3) (2001) 1–5.
[13] C.A. de Wolf, J.W. Bakker, P.T. Wouda, B.E. Nieuwenhuys, A.
Baraldi, S. Lizzit, M. Kiskinova, J. Chem. Phys. 113 (2000) 10717–
10722.
[14] C.A. de Wolf, J.W. Bakker, P.T. Wouda, B.E. Nieuwenhuys, A.
Baraldi, S. Lizzit, M. Kiskinova, J. Phys. Chem. B. 105 (2001) 4254–
4262.
[15] J.A. Dumesic, D.F. Rudd, L.M. Aparicio, J.E. Rekoske, A.A. Treviño,
The Microkinetics of Heterogeneous Catalysis, American Chemical
Society, 1993.
[16] A. Fahmi, R.A. van Santen, Z. Phys. Chem. 197 (1996) 203–217.
[17] H. Gerischer, A. Mauerer, J. Electroanal. Chem. 25 (1970) 421.
[18] J.M. Gohndrone, C.W. Olsen, A.L. Backman, T.R. Gow, E. Yagasaki,
R.I. Masel, J. Vac. Sci. Technol. A 7 (3) (1988) 1986–1990.
[19] J.F. Gootzen, A.H. Wonders, W. Visscher, R.A. van Santen, J.A.R. van
Veen, Electrochim. Acta 43 (1998) 1851.
[20] V.V. Gorodetskii, V.A. Sobyanin, in: T. Seiyama, K. Tanabe (Eds.),
Proceedings of 7th International Congress on Catalysis, Elsevier and
Kodansha LTD, Tokyo, 1980, pp. 566–577.
[21] H. Guo, D. Chrysostomou, J. Flowers, F. Zaera, J. Phys. Chem. B 107
(2003) 502–511.
[22] H. Guo, F. Zaera, Surf. Sci. 524 (2003) 1–14.
[23] W.L. Guthrie, J.D. Sokol, G.A. Somorjai, Surf. Sci. 109 (1981) 390–
418.
We studied the adsorption and dissociation of NH3 on
Ir(111). Ammonia does not dissociate on this surface, but
defects created by sputtering can facilitate NH3 ad dissocia-
tion on this surface. Oad also facilitates NH3 ad dissociation
and NH3 adsorption on a sputtered surface and an oxygen-
covered surface results in N2 formation around 500 K.
We also studied the steady-state NH3 oxidation reaction
on Ir(110) and Ir(111) under low-pressure conditions. On
both surfaces, the steady-state NH3 oxidation reaction starts
around 400 K (with the exact temperature depending on the
NH3/O2 pressure ratio), the initial products being N2 and
H2O. The surface composition changes around 500 K from
NHx ad to Oad, but the selectivity of the products observed in
the gas phase changes at a significantly higher temperatures.
An increase in O2 partial pressure greatly enhances the
reaction rate on Ir(111)and enhances it by much less on
Ir(110). This difference is explained by the fact that NH3
dissociates on Ir(110) even in the absence of Oad, whereas
on Ir(111) Oad is needed for significant NH3 dissociation.
The different behavior of Ir and Pt catalysts during cat-
alytic ammonia oxidation is explained in terms of the dif-
ferent activity of these metals in NO formation. NOad forms
readily on Pt surfaces (above 300 K [26]), whereas on Ir,
NOad formation occurs only above 450 K, in a large excess
of Oad. N2 is more easily formed under these conditions (Oad
lowers the activation energy of N2 formation) than is NO on
Ir(110), resulting in preferential N2(g) formation even when
the surface is Oad-covered.
[24] D.I. Hagen, B.E. Nieuwenhuys, G. Rovida, G.A. Somorjai, Surf.
Sci. 57 (2) (1976) 632–650.
Acknowledgments
[25] J.J. Joyce, M. del Guidice, J.H. Weaver, J. Electron Spectrosc. Relat.
Phenom. 49 (1) (1989) 31–45.
[26] M. Kim, S.J. Pratt, D.A. King, J. Am. Chem. Soc. 122 (2000) 2409–
2410.
[27] W.P. Krekelberg, J. Greeley, M. Mavrikakis, J. Phys. Chem. B 108
(2004) 987–994.
[28] E.S. Kurkina, N.L. Semendyaeva, A.I. Boronin, Kinet. Catal. 42 (5)
(2001) 773–789.
[29] R.I. Masel, Catal. Rev. Sci. Eng. 28 (1986) 335–369.
[30] W.D. Mieher, W. Ho, Surf. Sci. 322 (1995) 151.
[31] H. Mortensen, L. Diekhöner, A. Baurichter, E. Jensen, A.C. Luntz,
J. Chem. Phys. 113 (16) (2000) 6882–6887.
The authors thank ELETTRA and the E.U. for financial
support for the measurements at the SuperESCA beamline
of ELETTRA. They also acknowledge R.C.V. van Schie
for providing technical support. C.J.W. acknowledges the
Technology Foundation STW, Applied Science Division of
NWO and the Technology Program of the Ministry of Eco-
nomic Affairs for financial support under project number
UPC.5037. J.R.M. acknowledges the Socrates/Erasmus pro-
gram for the opportunity to come to Leiden University. A.B.
acknowledges financial support from the MIUR under pro-
gram PRIN2003.
[32] R.J. Purtell, P. Merrill, Phys. Rev. Lett. 44 (19) (1980) 1279–1281.
[33] E.D.L. Rienks, J.W. Bakker, A. Baraldi, S.A.C. Carabineiro, S. Lizzit,
C.J. Weststrate, B.E. Nieuwenhuys, Surf. Sci. 516 (2002) 109–117.

102
C.J. Weststrate et al. / Journal of Catalysis 235 (2005) 92–102
[34] A.K. Santra, B.K. Min, C.W. Yi, K. Luo, T.V. Choudhary, D.W. Good-
man, J. Phys. Chem. B 106 (2001) 340–344.
[35] D.P. Sobczyk, A.M. de Jong, E.J.M. Hensen, R.A. van Santen,
J. Catal. 219 (2003) 156–166.
[36] J.L. Taylor, D.E. Ibbotson, W.H. Weinberg, Surf. Sci. 79 (2) (1979)
349–384.
[38] A.C.M. van den Broek, J. van Grondelle, R.A. van Santen, J. Catal. 185
(1999) 297–306.
[39] C.J. Weststrate, J.W. Bakker, E.D.L. Rienks, S. Lizzit, L. Petaccia,
A. Baraldi, C.P. Vinod, B.E. Nieuwenhuys, J. Chem. Phys. 122 (18)
(2005) 184705.
[40] C.J. Weststrate, J.W. Bakker, E.D.L. Rienks, S. Lizzit, L. Petac-
cia, A. Baraldi, C.P. Vinod, B.E. Nieuwenhuys, Phys. Chem. Chem.
Phys. 7 (13) (2005) 2629–2634.
[37] D.M. Thornburg, R.J. Madix, Surf. Sci. 220 (2–3) (1989) 268–
294.