Oscillatory behaviour in the NO-H2 reaction over Ir(5 1 0)

Article abstract of DOI:10.1016/S0039-6028(00)00900-6

The non-linear behaviour in the NO-H₂ reaction over the stepped Ir(5 1 0) was studied using mass spectrometry, LEED and AES with NO pressures between 7.7×10^-8 and 7.7×10^-7 mbar and H₂/NO ratios ranging from 1 to 54. In addition, TDS experiments were carried out to study the interaction between the species present on the surface during the reaction. During temperature programmed heating of a NO saturated surface in the presence of a hydrogen flow an explosive formation of N₂, and to a lesser extent H₂O, was observed, indicative of an autocatalytic reaction. Furthermore, TDS experiments on a NO saturated surface that was precovered with O, showed that the presence of O_ads lowers the NO activation energy of desorption by 12 kJ mol^-1. During a heating-cooling cycle an adsorbate induced lifting of the surface reconstruction was observed under similar reaction conditions at which oscillations in rate were observed. The rates of N₂ and H₂O formation are in-phase during these oscillations. No other products were observed under oscillatory conditions. Possible mechanisms of the oscillations are discussed. Finally, a comparison between the non-linear behaviour over Ir(5 1 0), Pt(1 0 0) and other Ir and Rh surfaces was made.

Full text of DOI:10.1016/S0039-6028(00)00900-6

Surface Science 471 (2001) 114±124
www.elsevier.nl/locate/susc
Oscillatory behaviour in the NO±H2 reaction over Ir(5 1 0)
C.A. de Wolf *, M.O. Hattink, B.E. Nieuwenhuys
Leiden Institute of Chemistry, Leiden University, P.O. Box 9502, 2300 RA Leiden, Netherlands
Received 23 June 2000; accepted for publication 3 October 2000
Abstract
The non-linear behaviour in the NO±H2 reaction over the stepped Ir(5 1 0) was studied using mass spectrometry,
8
7
LEED and AES with NO pressures between 7:7 Â 10 and 7:7 Â 10 mbar and H2/NO ratios ranging from 1 to 54. In
addition, TDS experiments were carried out to study the interaction between the species present on the surface during
the reaction. During temperature programmed heating of a NO saturated surface in the presence of a hydrogen ¯ow an
explosive formation of N2, and to a lesser extent H2O, was observed, indicative of an autocatalytic reaction. Fur-
thermore, TDS experiments on a NO saturated surface that was precovered with O, showed that the presence of Oads
1
lowers the NO activation energy of desorption by 12 kJ mol
.
During a heating±cooling cycle an adsorbate induced lifting of the surface reconstruction was observed under similar
reaction conditions at which oscillations in rate were observed. The rates of N2 and H2O formation are in-phase during
these oscillations. No other products were observed under oscillatory conditions. Possible mechanisms of the oscilla-
tions are discussed. Finally, a comparison between the non-linear behaviour over Ir(5 1 0), Pt(1 0 0) and other Ir and Rh
surfaces was made. Ó 2001 Elsevier Science B.V. All rights reserved.
Keywords: Models of non-linear phenomena; Catalysis; Iridium; Hydrogen molecule; Nitrogen oxides; Stepped single crystal surfaces
1. Introduction
during the NO±H2 reaction both the activity and
the selectivity can oscillate in time. For example,
over Pt(1 0 0) the rates of N2 and NH3 formation
oscillate in-phase while the minor formation of
N2O oscillates out of phase with these products [8±
10]. In contrast, over stepped rhodium surfaces
consisting of (1 1 1) terraces, Ir(1 1 0) and Ir(2 1 0)
the rates of N2 and NH3 formation have an out of
phase relationship and no N2O is formed under
these conditions [11±15].
Especially for stepped surfaces, the non-linear
behaviour was ®rst investigated by using hemi-
spherically shaped single crystal tips, consisting of
numerous crystal planes, with ®eld emission mi-
croscopy (FEM) [5]. It was observed that waves
moving over a rhodium tip in a oscillatory way
Since the discovery of non-linear behaviour of
catalytic reactions over single crystal surfaces a
number of reactions and surfaces have been ex-
amined [1±7]. The NO±H2 reaction over the Pt
group metals received a lot of attention in this
context [5±7]. The reaction is not only of interest
because of its role in automotive catalysis, but also
because of the various reaction products that can
be formed, viz., N2, NH3 and N2O. In addition,
* Corresponding author. Tel.: +31-71-527-4484; fax: +31-71-
527-4451.
E-mail address: c.wolf@chem.leidenuniv.nl (C.A. de Wolf).
0039-6028/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved.
PII: S0039-6028(00)00900-6

C.A. de Wolf et al. / Surface Science 471 (2001) 114±124
115
were seeded on the (5 3 3) and (3 2 1) planes and,
possibly, on the edges of the (1 1 1) plane [16]. A
study of the extended Rh(5 3 3) and Rh(1 1 1)
planes showed that oscillations could also be ob-
served on the macroscopic scale under comparable
conditions, by measuring the formation rate of the
products with the mass spectrometer [11±13]. This
shows that the observation of non-linear behav-
iour with FEM can give a good indication of the
crystal planes and reaction conditions for which
macroscopic rate oscillations might be expected.
More recently, the NO±H2 reaction over an Ir
®eld emitter was studied. This work showed the
occurrence of oscillatory waves that are seeded on
the (5 1 0) plane and spread to the (1 0 0) plane
[5,17]. A short study of the extended Ir(1 0 0) sur-
face already revealed a hysteresis in the product
formation during a heating±cooling cycle and
damped oscillations in the N2 and H2O production.
Under the conditions at which the oscillations were
observed no NH3 was formed and the rates of H2O
and N2 formation oscillated in-phase. Since
molecular NO could still be present under the
relatively low temperature conditions, a vacancy
mechanism comparable to the one that was pro-
posed for the NO±H2 reaction over Pt(1 0 0), was
suggested [8,17]. The absence of NH3 formation,
which is a major product over Pt(1 0 0), was ex-
plained by the formation of a NHx; ads species, that
might act as a bu€er. Recent ®eld ion microscopy
(FIM) and FEM results support this idea [7].
The study described in the present paper was
based on earlier FEM work of the NO±H2 reac-
tion on a Ir tip and focuses on the non-linear be-
haviour of the reaction over an extended Ir(5 1 0)
single crystal surface. The (5 1 0) surface is com-
posed of ®ve atom wide (1 0 0) terraces separated
by (1 1 0) steps. The results will be compared with
oscillations in rate and selectivity of this reaction
over Pt(1 0 0), Ir(1 0 0), Ir(1 1 0) and the above-
mentioned Rh surfaces.
mass spectrometer, LEED optics and a CMA for
AES. The base pressure was always better than
10
2 Â 10
mbar.
The crystal was cut by spark erosion to within
0.5° of the desired direction and polished down to
a grain size of 0.1 lm. The sample was cleaned by
‡
standard techniques like Ar sputtering, repetitive
heating in O2 or H2 and annealing at 1200 K. The
surface structure and cleanliness were checked
before each series of experiments by LEED and
AES.
The reactions were performed in the ¯ow mode
using a turbo molecular pump (170 l s 1). During
the reactions the pumping rate was ꢀ50 l s 1. The
crystal was turned in front of a small hole which
gave access to the mass spectrometer chamber. In
this way the contributions of the edges of the
crystal were minimised.
For NH3 amu 16 was followed instead of amu
17 to avoid interference with OH, a cracking
product of water. Gases of high purity (Messer
Griesheim, commercial purity: 99.5±99.999%) were
dosed by means of two leak valves from the dosing
system into the main chamber. The pressure
readings of the ion gauge were corrected using
relative sensitivities for NO and H2 to N2 of 1.3
and 0.46, respectively.
Before starting the heating±cooling cycles the
NO pressure was stabilised for at least half an hour
and after adjusting the H2 pressure the second
cycle was recorded which proved to be perfectly
reproducible.
Due to the high pumping rate of the system and
the low pressures of the reactants the oscillations
take place under ®xed external conditions.
3. Results
In a ®rst approach to ®nd oscillatory behaviour
in the NO±H2 reaction over the extended Ir(5 1 0)
surface, the product formation during heating±
cooling cycles in the reaction mixture was followed
using the mass spectrometer. The existence of a
hysteresis in the N2 formation during such a cycle
is considered to be strongly related to the non-
linear behaviour that leads to oscillations. There-
fore, the NO±H2 reaction over Ir(5 1 0) has been
2. Experimental
The measurements were performed in a UHV
system with a volume of approximately 30 l,
equipped with a di€erentially pumped quadrupole

116
C.A. de Wolf et al. / Surface Science 471 (2001) 114±124
studied in the temperature range between 300 and
the temperature the rꢁN2† and the NO conversion
start to decrease around 500 K accompanied by an
increase in the rate of NH3 formation. Depending
on R only a small part of the Nads reacts to give
NH3, whereas the major part of the NO is still
converted to N2, or, for higher H2/NO ratios,
the selectivity switches completely resulting in the
formation of only NH3. The increase in the
rꢁNH3† goes along with a decrease in rꢁH2O†, an
e€ect which becomes more pronounced at higher
H2/NO ratios, i.e. when more NH3 is produced. A
further decrease of the temperature changes the
selectivity back to N2. Remarkably, the local
maximum in rꢁN2† is accompanied by a local
maximum in the NO conversion and the H2O
formation. These maxima, and thus the hystereses,
shift to lower temperatures for increasing H2/NO
ratios.
800 K with NO pressures ꢁpꢁNO†† ranging from
8
7
7:7 Â 10 to 7:7 Â 10 mbar and H2/NO ratios,
R, between 1 and 54. Under these conditions N2,
NH3 and H2O were formed but no N2O formation
was observed.
Fig. 1 shows the hysteresis phenomena that
were observed during a heating±cooling cycle with
7
pꢁNO† ˆ 7:7  10 mbar and various H2/NO ra-
tios. As can be seen from this ®gure, the conver-
sion of NO starts around 450 K accompanied by
an increase in the rate of H2O formation and, for
the lower H2/NO ratios, by an increase in the rate
of N2 formation ꢁrꢁN2††. At R ˆ 10:2 the forma-
tion of NH3 is favoured over the formation of N2
at this temperature and only a small local maxi-
mum in the rꢁN2† is observed. Around 500 K a
plateau is reached in the NO conversion and
rꢁN2†, indicating that the maximum reaction rate
is reached. Interestingly, at higher H2/NO ratios
the maximum NO conversion becomes lower. The
H2O pressure is still increasing, but this is caused
by the low pumping rate of H2O. Upon decreasing
Oscillatory behaviour could be obtained in the
same NO pressure regime for H2/NO ratios vary-
ing between 1.1 and 7.3 and for temperatures be-
tween 450 and 403 K, respectively. The oscillations
could be triggered by the following procedure:
7
Fig. 1. The rates of NO conversion and NH3, H2O and N2 formation during a heating±cooling cycle at a NO pressure of 7:7 Â 10
mbar. The partial pressures of NH3, H2O, N2 and NO are represented by the solid, dotted, thick and dashed line, respectively. The
H2/NO ratios vary from 2.8 to 10.2. The heating rate was 0.5 K s 1. The arrow in the bottom left corner of the right panel points to the
low temperature maximum in the NO conversion.

C.A. de Wolf et al. / Surface Science 471 (2001) 114±124
117
7
®rst, the NO pressure was stabilised at 7:7 Â 10
mbar, then hydrogen was introduced until a H2/
NO ratio between 8 and 9 was reached. Subse-
quently, the crystal was slowly heated in the re-
action mixture till 650 K and cooled to a
temperature between 450 and 403 K. Finally, the
hydrogen pressure was reduced until the oscilla-
tions started.
Surprisingly, oscillations could only be ob-
served for strict combinations of the H2/NO ratio
and temperature. In other words, the oscillations
faded away immediately when either the temper-
ature or the hydrogen pressure was altered. A close
inspection of the temperature needed to obtain
oscillations at a certain H2/NO ratio revealed that
it almost exactly coincides with the sharp low
temperature maximum in the N2 formation rate on
the cooling branch of the heating±cooling cycle,
con®rming the strong relation between the hys-
teresis phenomena and the oscillations. This is il-
lustrated in Fig. 2.
Fig. 3. Oscillations in the partial pressures of H2O, N2 and NO
7
above the Ir(5 1 0) surface at a NO pressure of 7:7 Â 10 mbar
and a H2/NO ratio of 4.5. The temperature was 413 K.
Examples of the oscillatory behaviour in the
NO±H2 reaction over Ir(5 1 0) are shown in Figs. 3
and 4. These ®gures clearly demonstrate the in-
phase relationship between the formation rates of
Fig. 4. Oscillations in the partial pressures of H2O, N2 and NO
7
above the Ir(5 1 0) surface at a NO pressure of 7:7 Â 10 mbar
and a H2/NO ratio of 1.1. The temperature was 450 K.
N2 and H2O and the conversion of NO. NH3 is not
formed under these conditions. The period of the
oscillations is rather irregular and varies between
34 and 37 min for R ˆ 4:5 and between 21 and 27
min for R ˆ 1:1. However, as can be seen in Fig. 4
there are also time intervals with a more regular
pattern. In Fig. 3 some kind of superimposed ®ne
structure can be observed that becomes more
pronounced when the reaction has been oscillating
for more than 5 h. The major part of this ``period
doubling'' is caused by slight changes in the H2
Fig. 2. The occurrence of oscillations in rate of the NO±H2
reaction over Ir(5 1 0). The large solid circles indicate the con-
ditions for which oscillations were found at a NO pressure of
7
7:7 Â 10 mbar. The solid line represents the temperature
evaluation of the low temperature maximum in the rate of N2
formation on the cooling branch of the heating±cooling cycle
with the H2/NO ratio.

118
C.A. de Wolf et al. / Surface Science 471 (2001) 114±124
Fig. 5. The LEED patterns of the Ir(5 1 0) surface: (A) a clean reconstructed surface at 76 eV, (B) a (1 Â 1) phase observed on the
8
heating branch of a heating±cooling cycle with pꢁNO† ˆ 7:7  10 mbar and H2/NO ˆ 5.7 at 75 eV, and (C) a (1  5) phase observed
on the cooling branch between 730 and 360 K of the aforementioned heating±cooling cycle at 76 eV.
pressure due to the consumption of hydrogen in
the gas inlet system. Because of this uncontrollable
decline of the H2 pressure and the long period of
the oscillatory cycle no attempt was made to ex-
amine the possible route to deterministic chaos via
period doublings.
heating branch that corresponds with the start of
the H2O formation. On the cooling branch the
O510 eV=Ir229 eV signal increases again when the NO
conversion has reached the local low tempera-
ture minimum. The O510 eV peak can be attributed
to atomic O. The N381
peak overlaps with the
eV
In order to obtain a better insight into the
processes that take place on the surface during the
oscillations the heating±cooling cycles were stud-
ied by AES and LEED. Unfortunately, these
techniques could not be used under the high NO
pressure conditions that were needed to ®nd os-
cillations. However, since the hysteresis phenom-
ena at lower NO pressure are similar to the high
pressure ones, still a lot of information can be
obtained by studying the hysteresis at pꢁNO† ˆ
Ir380 peak and, therefore, could not give any
eV
conclusive information.
8
7:7 Â 10 mbar.
Fig. 5B and C show the LEED patterns that
were acquired during a heating±cooling cycle with
a H2/NO ratio of 5.7. For comparison, the LEED
pattern of the clean Ir(5 1 0) surface is shown in
Fig. 5A.
The pattern in Fig. 5B was observed on the
heating branch (Tmax ˆ 800 K) and on the cooling
branch until 730 K. Further cooling resulted in the
LEED pattern shown in Fig. 5C, which was re-
placed by the pattern of Fig. 5B around 360 K.
Interestingly, the switch from the pattern in Fig.
5C to the one in Fig. 5B coincides with the local
minimum in the NO conversion that precedes the
local maxima in the N2 formation rate and the NO
conversion on the cooling branch of the heating±
cooling cycle (see also Fig. 1).
Fig. 6. The O510 eV=Ir229 AES signal ratios in addition to the
eV
H2O and NO partial pressures during a heating±cooling cycle at
8
a NO pressure of 7:7 Â 10 mbar and a H2/NO ratio of 10. The
The AES measurements, shown in Fig. 6, show
a decrease in the O510 eV=Ir229 eV signal on the
up triangles represent the heating branch and the down trian-
gles represent the cooling branch.

C.A. de Wolf et al. / Surface Science 471 (2001) 114±124
119
Finally, TDS measurements were performed to
study the interaction between the species that
might be present on the surface during the reac-
tion. A selection of the results is shown in Fig. 7.
Fig. 7A presents the formation rates of the prod-
ucts that were formed during the temperature
programmed desorption in the presence of a hy-
drogen ¯ow from a NO saturated surface. No
correction was made for the di€erent MS sensi-
tivities of the gases. Therefore, this ®gure will only
be used for qualitative considerations. The ®gure
shows that the formation of N2, H2O and NH3
start simultaneously around 350 K. However, the
formation of H2O and especially N2 show a very
sharp trailing edge at ꢀ375, whereas the formation
of NH3 continues until a temperature of more than
450 K is reached. No other products were ob-
served.
The AES measurements showed the presence of
Oads in the temperature region where oscillations
can take place. Therefore, NO was dosed onto a
surface that was predosed with O2 at 350 K. At
this temperature O2 adsorbs dissociatively on
Ir(1 0 0) [18].
Fig. 7B shows the rate of NO desorption from a
NO saturated surface in addition to the rates of
NO formation from a NO saturated surface that
was predosed with increasing amounts of oxygen.
The NO desorption can be roughly divided into
three peaks labelled a1 to a3, of which the a1 peak
is only present when the surface is precovered with
Oads. Upon increasing the amount of predosed
oxygen the a2 peak becomes more pronounced,
while the size of the a3 peak is less in¯uenced by
the Oads precoverage. Furthermore, the position of
the peaks is not in¯uenced by the amount of Oads
.
The N2 desorption can also be divided into
three peaks, labelled b1 to b3. As can be seen in
Fig. 7C the b1 peak shifts to slightly higher tem-
peratures, the b2 peak disappears and the b3 peak
remains at the same temperature upon increasing
the amount of Oads. The total amount of N2 for-
mation decreases with an increasing amount of
predosed oxygen.
4. Discussion
The existence of non-linear e€ects in the
NO±H2 reaction over the extended Ir(5 1 0) surface
has been clearly demonstrated in Fig. 1 and in
Figs. 3 and 4, showing hysteresis phenomena and
kinetic oscillations, respectively. The oscillatory
conditions di€er from the conditions needed to
obtain oscillations over the Ir ®eld emitter, in that
the NO pressure and the H2/NO ratios are lower in
the present study [5,17].
Fig. 7. TD spectra of a NO saturated Ir(5 1 0) surface with (A)
6
2 Â 10 mbar H2 in the ¯ow during heating, (B and C) a
We will start the discussion by comparing the
hysteresis phenomena in the present system with
the more extensively studied phenomena over
precoverage of Oads. The predose of O2 increases in the direc-
tion of the arrows from 0.45 to 7 L. The heating rate was 2
1
K s and the adsorption temperature was 300 K.

120
C.A. de Wolf et al. / Surface Science 471 (2001) 114±124
other Pt group metals followed by a discussion of
the oscillatory behaviour.
its starting position. This short increase in the
activity was not observed for Ir(1 1 0) or the Rh-
(1 1 1)-like surfaces.
The hysteresis phenomena over Ir(5 1 0) have a
lot of features in common with the hystereses ob-
served for the NO±H2 reaction over Ir(1 1 0) and
Rh surfaces with (1 1 1) terraces [11±15]. Over all
these surfaces the various reaction products are
formed in similar order, suggesting an analogous
mechanism. Fast XPS studies of the NO±H2 re-
action over Ir(1 1 0) and Rh(5 3 3) during a heat-
ing±cooling cycle revealed that the inactive surface
is covered with atomic O and molecular NO,
whilst the active surface is either covered with N
(or NHx) or adsorbate free. The dissociative ad-
sorption can be blocked by a critical N coverage or
by the combination of NOads and Oads. Further-
more, the NO uptake curves on Ir(1 1 0) showed
that the presence of Oads has a strongly destabil-
ising e€ect on Nads, which was considered to be a
crucial factor in the non-linear behaviour [19±21].
However, the TDS results shown in Fig. 7C shows
that Oads has not such a strong e€ect on Nads on
Ir(5 1 0). Instead of a lowering of the desorption
temperature of N2, as would be expected if Oads
was to destabilise Nads, the desorption temperature
of the b1 peak is increased slightly. Furthermore,
upon increasing the Oads precoverage the relative
contribution of the high temperature b3 peak is
even increasing.
In addition, there are some major di€erences
between the hysteresis phenomena described in
this study and the ones recorded for the Ir(1 1 0)
and the Rh surfaces. First of all, in the case of
Ir(5 1 0) the rꢁH2O† is strongly in¯uenced by the
formation of NH3, indicating that the reaction
between N and H proceeds faster than the reaction
between O and H in the temperature regime be-
tween 400 and 500 K. Over the Rh surfaces and
Ir(1 1 0) the rꢁH2O† is not in¯uenced by the NH3
formation even at low H2/NO ratios. This might
suggest that the NO conversion over Ir(5 1 0) ex-
ceeds the NO conversion over the other surfaces
and, hence, that there is no competition for the H
atoms between Oads and Nads over the Ir(1 1 0) and
Rh surfaces.
By comparing the hysteresis phenomena of the
present system with the NO±H2/Pt(1 0 0) system
two essentially similar features should be men-
tioned: the reconstruction behaviour and the sur-
face explosion.
During the NO±H2 reaction over Ir(5 1 0) the
surface switches from an almost complete (1 Â 1)
structure on the heating branch at room temper-
ature (Fig. 5B) to a reconstructed (1 Â 5) surface
with at least two domains on the cooling branch at
730 K (Fig. 5C). The (1 Â 5) reconstruction is
8
lifted again around 360 K for pꢁNO† ˆ 7:7  10
mbar, viz. at the same temperature at which the
NO conversion reaches the local minimum. The
extra spots near the (1 Â 1) substrate spots are
caused by the steps and show that the recon-
struction occurs in a diagonal direction with re-
spect to the step edges.
The (5 1 0) surface consists of ®ve atom wide
(1 0 0) terraces, therefore, it might be useful to
compare the reconstruction behaviour of Ir(5 1 0)
with Ir(1 0 0) before we continue the comparison
with Pt(1 0 0). The surface structure of a clean
Ir(1 0 0) crystal has been the subject of numerous
studies [22±35]. The main conclusion is that the
clean Ir(1 0 0) surface reconstructs to a perfect
(1 Â 5) structure, comparable to Fig. 5A, which
can be assigned to a buckled hexagonal recon-
structed top layer. In addition, a clean (1 Â 1)
surface can be prepared by a special procedure,
consisting of oxidation±reduction treatments with
O2 and H2 or CO. The unreconstructed surface is
stable up to 800 K. The (1 Â 5) reconstruction can
be lifted by the adsorption of, e.g., H2 [36,37], O2
[18,25,26,30] and NO [38,39], although the (1 Â 5)
spots never completely disappear from the LEED
pattern. Furthermore, a H2 induced (1 Â 3) re-
construction of the Ir(1 0 0) surface is mentioned
[37]. The (1 Â 1) surface is generally believed to
have a higher sticking coecient for the adsor-
bates mentioned than the (1 Â 5) reconstructed
surface. In addition, the (1 Â 1) surface actively
promotes NO dissociation at T > 200 K, whilst
the NO adsorption on the (1 Â 5) reconstructed
surface is entirely molecular at 300 K. Before NO
Finally, over Ir(5 1 0) the partial NO pressure
shows a local minimum around 400 K on the
cooling branch, before the NO pressure returns to

C.A. de Wolf et al. / Surface Science 471 (2001) 114±124
121
dissociation can take place on the reconstructed
surface the reconstruction has to be (partially)
lifted by the adsorption of NO [39]. According to
another model the NO adsorption is initially mo-
lecular, followed by dissociative adsorption and
then further molecular adsorption up to saturation
[38].
formation of both N2 and H2O is observed on the
heating branch. This indicates that at the higher
NO pressure the adsorption and dissociation of
NO are faster than the removal of Oads by H2,
resulting in the recombination of Nads and Oads
because no build-up of Oads was observed by AES.
,
The atom recombination ‰Nads ‡ Oads ! NOgas
Š
Related reconstruction behaviour has been re-
ported for Pt(1 0 0). The clean Pt(1 0 0) recon-
structs to a (5 Â 20) surface structure, usually
referred to as the (hex) phase [22,40]. Similar to the
Ir(1 0 0) surface, a ꢁhex† ! ꢁ1  1† transition is
induced on Pt(1 0 0) by adsorption of, amongst
others, H2 [41±45], O2 [46] and NO [47±50]. The
observed hysteresis in the rꢁN2† for the NO±H2
reaction over Pt(1 0 0) was explained by the dif-
ference in activity between the (hex) and the
(1 Â 1) phase in the dissociation of NO. Upon
heating the (1 Â 1) phase actively catalyses the
conversion of NO. At high temperatures the sur-
face coverage drops below the critical value needed
to maintain the (1 Â 1) surface structure and a
transition to the inactive (hex) phase takes place.
As a consequence, less NO is converted on the
cooling branch until the surface coverage reaches
the critical value again that lifts the reconstruction
to the more active (1 Â 1) phase [10,51].
has been reported before on the Ir(1 0 0)-(1 Â 1)
surface [56].
On the basis of the discussed similarities be-
tween Ir(5 1 0) and the Ir(1 1 0) and the Rh surfaces
on one hand, and the Pt(1 0 0) surface on the other
hand, we propose the following model for the
hysteresis phenomena in the NO±H2 reaction over
Ir(5 1 0).
At room temperature the surface is covered
with NO and O and, as a consequence, the surface
is not reconstructed. Heating the crystal leads to
the desorption of NO, which is promoted by the
presence of Oads. At low NO pressures and H2/
NO > 20 the NO desorption leads to an autocat-
alytic clearing of the surface, whereas at higher
NO pressures part of the Nads recombines with
Oads to NOgas, resulting in a more gradual increase
of the reaction rate. Subsequently, NH3 is formed
in the temperature regime between 400 and 500 K
for H2/NO > 5, in addition to N2 and H2O. At
temperatures >500 K only N2 and H2O are
formed and the reaction rate reaches its maximum,
resulting in a nearly adsorbate free surface. Con-
sequently, the surface reconstructs to the (1 Â 5)
phase. In contrast to the NO±H2/Pt(1 0 0) system,
over Ir(5 1 0) this does not directly lead to a re-
duction of the activity, probably because the
reconstruction is not complete. According to
Gardner et al. a partially reconstructed Ir(1 0 0)
surface still contains active (1 Â 1) islands [39]. The
decrease in the reaction rate with the H2/NO ratio
is probably related to the more complete surface
reconstruction at lower Oads coverages.
A so-called surface explosion in the NO±H2
reaction over Pt(1 0 0) takes place upon heating the
surface in the reaction mixture or during NOads
and Hads coadsorption experiments [10,52,53]. This
autocatalytic formation of N2 and H2O can take
place because the removal of Oads is much faster
than the recombination of Nads and Oads and be-
cause of the strongly non-linear dependence of the
NOads dissociation rate upon the concentration of
vacancies [54,55]. A comparable feature is ob-
served in Fig. 7A for the temperature programmed
desorption of N2 and H2O from a NO saturated
Ir(5 1 0) surface in the presence of a H2 ¯ow. The
very narrow temperature regime of the N2 and
H2O formation points to an autocatalytic NO
dissociation, which probably has the same origin
as the one observed over Pt(1 0 0). In contrast with
Pt(1 0 0) the surface explosion is not observed for
the NO±H2 reaction during the heating±cooling
On the cooling branch, the reaction rate starts
to decrease around 500 K. At this temperature the
N ‡ 3H ! NH3 reaction proceeds faster than
the O ‡ 2H ! H2O reaction. Probably, Oads and
possibly NHx build up on the surface, as was ob-
served over Ir(1 1 0) and was suggested for Ir(1 0 0)
and Rh(5 3 3). As a result, the NO dissociation is
hindered and the dissociative hydrogen adsorption
7
cycle at pꢁNO† ˆ 7:7  10 mbar (Fig. 1). How-
ever, at a 10 times lower NO pressure an explosive

122
C.A. de Wolf et al. / Surface Science 471 (2001) 114±124
is blocked [7,19,21]. As a consequence the selec-
tivity switches back to N2. Due to the build-up of
the adsorbates the surface reconstruction is lifted
resulting in the more active (1 Â 1) phase. The
(1 Â 1) phase promotes the NO dissociation lead-
ing to the low temperature maximum in the NO
For the non-linear behaviour in the NO±H2
reaction over the related Ir(1 0 0) surface a Pt-like
model has been proposed as well. In this model
NOads (and/or possibly Oads) causes a decrease in
the desorption energy of NO. The desorption of
NO from a NO (and/or O) covered surface leads to
an autocatalytic clearing of the surface [7]. The
results of the TDS experiments of a saturated NO
surface that has been precovered by Oads, shown in
Fig. 7B and C, suggest that this model can be valid
for the Ir(5 1 0) surface. From Fig. 7B it is clear
that by increasing the precoverage of Oads a new
NO desorption channel is opened, labelled a1. For
this a1 peak a NO activation energy of desorption
conversion. The formed Nads recombine to N2; gas
,
but the removal of Oads is not fast enough and an
O layer is quickly built up, which blocks further
reaction.
The oscillations were observed on the cooling
branch near the low temperature maximum in the
NO conversion, viz. in the same temperature re-
gime where the surface reconstruction takes place.
This might be an indication that the surface re-
construction may play a role in the oscillatory
mechanism. However, this is not necessarily true.
For the oscillations in the NO±H2 reaction over
Pt(1 0 0) two mechanisms have been proposed.
According to one model the oscillations were
caused by the repetitive transition between the
active (1 Â 1) surface and the inactive (hex) phase,
although the phase transition has never been ex-
perimentally observed [10,55,57]. According to the
other model the surface phase transition is not the
driving force behind the oscillations. Instead, a
vacancy model was proposed in which the auto-
catalytic increase in the number of vacant sites for
NO dissociation was considered to be the most
important step [8,51,54]. Makeev et al. showed
that the introduction of the surface phase transi-
tion is only necessary for the modelling of the
hysteresis phenomena [51,54].
The surface phase transition model and the
vacancy model can also be applied for the ob-
served oscillations over Ir(5 1 0). Unfortunately,
the high pressure conditions did not allow us to
follow the oscillations with LEED, therefore, we
cannot exclude the occurrence of a surface recon-
struction during the oscillations. However, even if
a surface phase transition were to occur during the
oscillations it need not be the driving force behind
the oscillations. It can also be a consequence of the
periodic changes of the adsorbate coverages. In
addition, the hysteresis phenomena show that the
in¯uence of the surface phase transition on the NO
conversion is negligible on Ir(5 1 0). Consequently,
a vacancy model is considered to be more likely.
1
1
of 92 kJ mol was found which is 12 kJ mol
lower than the one that was found for the a2 peak,
viz. the peak with the lowest NO desorption en-
ergy from a clean surface.
The high sensitivity of the oscillatory regime
towards the hydrogen pressure indicates a strong
in¯uence of the hydrogen pressure on the crucial
step of the mechanism. The formation of NHx; ads
during the oscillations in the NO±H2 reaction has
been suggested before on the basis of FEM and
FIM results on a Ir-tip [7]. The NHx; ads species
was thought to play the role of a bu€er species
[3,7].
On the basis of these considerations we pro-
pose the following model for the oscillations in the
NO±H2 reaction over the extended Ir(5 1 0) sur-
face:
(1) The surface is covered with NO and O. The
amount of vacancies is small and, hence, the re-
action rate is low.
(2) Due to the presence of Oads the desorption
energy of NO is lowered, leading to the removal of
NO from the surface and the creation of vacant
sites. The increasing amount of vacancies allows
more and more NOads to dissociate. The formed
Nads reacts with Nads or Oads, resulting in a gradual
increase of the total reaction rate.
(3) Oads starts to build up, either because the
Oads ‡ Hads reaction is not very e€ective due to the
low hydrogen concentration on the surface, or
because the hydrogen is consumed by Nads to form
NHx; ads (because no NH3 leaves the surface during
the oscillations).

C.A. de Wolf et al. / Surface Science 471 (2001) 114±124
123
to be the most likely mechanism. The formation of
NHx; ads might play a crucial role in the feedback
mechanism.
Acknowledgements
The Netherlands organisation for Scienti®c re-
search (NWO) is gratefully acknowledged for ®-
nancial support. Part of the work is supported by
INTAS, reference number 99-01882. This study
has been performed under the auspices of NIOK,
the Netherlands Institute for Catalysis, lab. report
no. UL 00-2-06.
Fig. 8. A schematic representation of one oscillatory cycle in
NO±H2 reaction over Ir(5 1 0). The four distinct stages are in-
dicated.
References
(4) The NO dissociation becomes suppressed by
the O layer, leading again the a surface covered
with NO and O.
[1] G. Ertl, Adv. Catal. 37 (1990) 213.
[2] R. Imbihl, Prog. Surf. Sci. 44 (1993) 185.

[3] F. Schuth, B.E. Henry, L.D. Schmidt, Adv. Catal. 39
(1993) 51.
[4] R. Imbihl, G. Ertl, Chem. Rev. 95 (1995) 697.
[5] P.D. Cobden, N.M.H. Janssen, Y. van Breugel, B.E.
Nieuwenhuys, Faraday Discuss. 105 (1996) 57.
[6] N.M.H. Janssen, P.D. Cobden, B.E. Nieuwenhuys, J.
Phys. Cond. Mat. 9 (1997) 1889.
The four stages of the oscillatory cycle are indi-
cated in Fig. 8.
[7] P.D. Cobden, C.A. de Wolf, M.Yu. Smirnov, A. Makeev,
B.E. Nieuwenhuys, J. Mol. Catal. 158 (2000) 115.
[8] J. Siera, P.D. Cobden, K. Tanaka, B.E. Nieuwenhuys,
Catal. Lett. 10 (1991) 335.
5. Conclusions
Non-linear behaviour in the NO±H2 reaction
over an extended Ir(5 1 0) surface was observed. In
addition to oscillations and hysteresis phenomena
also surface explosions were observed during TDS
experiments of a saturated NO surface in a hy-
drogen ¯ow.
The clean Ir(5 1 0) surface reconstructs to a
(1 Â 5) phase. During a heating±cooling cycle a
adsorbate induced lifting of the surface recon-
struction was observed near the starting conditions
for the oscillations.
The rates of N2 and H2O formation oscillate in-
phase. No other products were formed under the
conditions where oscillations could be observed.
Both a surface phase reconstruction model and
a vacancy model were mentioned to explain the
mechanism of the oscillations. A vacancy model in
which the desorption energy of NO depends non-
linearly on the O or NO coverage was considered
[9] P.D. Cobden, J. Siera, B.E. Nieuwenhuys, J. Vac. Sci.
Technol. A 10 (1992) 2487.

[10] M. Slinko, T. Fink, T. Loher, H.H. Madden, S.J.
Lombardo, R. Imbihl, G. Ertl, Surf. Sci. 264 (1992) 157.
[11] N.M.H. Janssen, B.E. Nieuwenhuys, M. Ikai, K. Tanaka,
A.R. Cholach, Surf. Sci. 319 (1994) L29.
[12] N.M.H. Janssen, P.D. Cobden, B.E. Nieuwenhuys, M.
Ikai, K. Mukai, K. Tanaka, Catal. Lett. 35 (1995) 155.
[13] P.D. Cobden, N.M.H. Janssen, Y. van Breugel, B.E.
Nieuwenhuys, Surf. Sci. 366 (1996) 432.
[14] C.A. de Wolf, B.E. Nieuwenhuys, A. Sasahara, K. Tanaka,
M.M. Slinko, M.Yu. Smirnov, Surf. Sci. 411 (1998) L904.
[15] C.A. de Wolf, B.E. Nieuwenhuys, M.M. Slinko, M.Yu.
Smirnov, Surf. Sci. 433±435 (1999) 63.
[16] M.F.H. van Tol, A. Gielbert, B.E. Nieuwenhuys, Catal.
Lett. 16 (1992) 297.
[17] P.D. Cobden, Y. van Breugel, B.E. Nieuwenhuys, Surf.
Sci. 402±404 (1998) 155.

[18] T. Ali, B. Klotzer, A.V. Walker, Q. Ge, D.A. King,
J. Chem. Phys. 109 (1998) 9967.
[19] P.D. Cobden, B.E. Nieuwenhuys, F. Esch, A. Baraldi, G.
Comelli, S. Lizzit, M. Kiskinova, Surf. Sci. 416 (1998) 264.

124
C.A. de Wolf et al. / Surface Science 471 (2001) 114±124
[20] C.A. de Wolf, J.W. Bakker, P.T. Wouda, B.E. Nie-
uwenhuys, A. Baraldi, S. Lizzit, M. Kiskinova, J. Chem.
Phys., in press.
[39] P. Gardner, R. Martin, R. Nalezinski, C.L.A. Lamont,
M.J. Weaver, A.M. Bradshaw, J. Chem. Soc. Faraday
Trans. 91 (1995) 3575.
[21] C.A. de Wolf, J.W. Bakker, P.T. Wouda, B.E. Nie-
uwenhuys, A. Baraldi, S. Lizzit, M. Kiskinova, in press.
[40] S. Hagstrom, H.B. Lyon, G.A. Somorjai, Phys. Rev. Lett.
15 (1965) 491.
ꢀ
[22] M.A. van Hove, R.J. Koestner, P.C. Stair, J.P. Biberian,
L.L. Kesmodel, I. Bartos, G.A. Somorjai, Surf. Sci. 103
[41] P.R. Norton, J.A. Davies, D.K. Creber, C.W. Sitter, T.E.
Jackman, Surf. Sci. 108 (1981) 205.
ꢁ
(1981) 189.
[23] J.T. Grant, Surf. Sci. 18 (1969) 228.
[42] B. Penneman, K. Oster, K. Wandelt, Surf. Sci. 249 (1991)
35.
[24] A. Ignatiev, A.V. Lones, T.N. Rhodin, Surf. Sci. 30 (1972)
573.
[43] J. Radnik, F. Gitmans, B. Penneman, K. Oster, K.
Wandelt, Surf. Sci. 287±288 (1993) 330.
ꢀ
[25] T.N. Rhodin, G. Broden, Surf. Sci. 60 (1976) 466.
[26] J. Kuppers, H. Michel, Appl. Surf. Sci. 3 (1979) 179.

[44] B. Klotzer, E. Bechtold, Surf. Sci. 295 (1993) 374.

[45] A.T. Pasteur, St. J. Dixon-Warren, D.A. King, J. Chem.
Phys. 103 (1995) 2251.
ꢀ
[27] M.A. van Hove, R.J. Koestner, P.C. Stair, J.P. Biberian,
L.L. Kesmodel, I. Bartos, G.A. Somorjai, Surf. Sci. 103
ꢁ
[46] K. Griths, T.E. Jackman, T.A. Davies, P.R. Norton,
Surf. Sci. 138 (1984) 113.
[47] H.P. Bonzel, G. Pirurg, Surf. Sci. 62 (1977) 45.
(1981) 218.
[28] E. Lang, K. Muller, K. Heinz, M.A. van Hove, R.J.

Koestner, G.A. Somorjai, Surf. Sci. 127 (1983) 347.

[48] E. Ritter, R.J. Behm, G. Potschke, J. Winterlin, Surf. Sci.
ꢀ
[29] D. Tomanek, K.H. Bennemann, Surf. Sci. 163 (1985) 503.
[30] K. Heinz, G. Schmidt, L. Hammer, K. Muller, Phys. Rev.
181 (1987) 403.
[49] P. Gardner, M. Tushaus, R. Martin, A.M. Bradshaw, Surf.
Sci. 240 (1990) 112.

B 32 (1985) 6214.
[31] N. Bickel, K. Heinz, Surf. Sci. 163 (1985) 435.
[32] K. Watanabe, D.J.W. Geldart, H.J. Kreuzer, Surf. Sci. 192
(1987) 344.
[50] K. Mase, Y. Murata, Surf. Sci. 242 (1991) 132.
[51] A.G. Makeev, B.E. Nieuwenhuys, Surf. Sci. 418 (1998)
432.
[33] A. Filippetti, V. Fiorentini, Surf. Sci. 377±379 (1997) 112.
[34] Q. Ge, D.A. King, N. Marzari, M.C. Payne, Surf. Sci. 418
(1998) 529.
[52] M.W. Lesley, L.D. Schmidt, Surf. Sci. 155 (1985) 215.
[53] D.Yu. Zemlyanov, M.Yu. Smirnov, V.V. Gorodetski,
Catal. Lett. 28 (1994) 153.
[35] V. Jahns, D.M. Zehner, G.M. Watson, D. Gibbs, Surf. Sci.
430 (1999) 55.
[54] A.G. Makeev, B.E. Nieuwenhuys, J. Chem. Phys. 108
(1998) 3740.

[36] T. Ali, A.V. Walker, B. Klotzer, D.A. King, Surf. Sci. 414
(1998) 304.
[55] M. Gruyters, A.T. Pasteur, D.A. King, J. Chem. Soc.
Faraday Trans. 92 (1996) 2941.

[37] K. Moritani, M. Okada, T. Kasai, Y. Murata, Surf. Sci.
445 (2000) 315.
[38] J. Kanski, T.N. Rhodin, Surf. Sci. 65 (1977) 63.
[56] J. Kuppers, H. Michel, Surf. Sci. 85 (1979) 201.
[57] S.J. Lombardo, T. Fink, R. Imbihl, J. Chem. Phys. 98
(1993) 5526.